1 2017-09-26 Richard Biener <rguenther@suse.de>
3 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
5 (scop_detection::build_scop_breadth): ... this. Removed.
6 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
7 (scop_detection::harmful_stmt_in_bb): Likewise.
8 (scop_detection::graphite_can_represent_stmt): Likewise.
9 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
10 (scop_detection::can_represent_loop): Remove recursion, fold in ...
11 (scop_detection::can_represent_loop_1): ... this. Removed.
12 (scop_detection::harmful_loop_in_region): Simplify after inlining
13 the above and remove more quadraticness.
14 (build_scops): Adjust.
15 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
18 2017-09-26 Jakub Jelinek <jakub@redhat.com>
21 * config/i386/i386.c (ix86_print_operand_address_as): Only test
22 REGNO (base) == SP_REG if base is a REG.
25 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
26 if it is different SSA_NAME.
27 (optimize_range_tests_cmp_bitwise): New function.
28 (optimize_range_tests): Call it.
30 2017-09-26 Richard Biener <rguenther@suse.de>
32 PR tree-optimization/82321
33 * graphite.c (canonicalize_loop_closed_ssa): Properly check
34 for the def being inside the loop.
36 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
40 * config/s390/s390-builtins.def: Fix constraint on op4.
42 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
44 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
45 independent expanders.
46 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
47 ("vec_ordered", "vec_unordered"): New expanders.
49 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
51 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
54 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
56 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
58 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
60 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
62 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
63 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
64 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
66 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
68 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
70 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
72 ("*vec_slb<mode>"): New insn pattern.
73 ("vec_shr_<mode>"): New expander.
74 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
75 and force the shift count operand to V16QImode.
76 ("vec_srb<mode>"): Set shift count mode to V16QI.
78 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
80 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
81 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
82 ("vec_widen_smult_hi_<mode>"): New expander definitions.
84 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
87 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
89 2017-09-26 Richard Biener <rguenther@suse.de>
91 PR tree-optimization/82320
92 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
95 2017-09-25 Jeff Law <law@redhat.com>
97 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
98 prototype for new argument.
99 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
100 mostly extracted from rs6000_emit_allocate_stack.
101 (rs6000_emit_probe_stack_range_stack_clash): New function.
102 (rs6000_emit_allocate_stack): Call
103 rs6000_emit_probe_stack_range_stack_clash as needed.
104 (rs6000_emit_probe_stack_range): Add additional argument
105 to call to gen_probe_stack_range{si,di}.
106 (output_probe_stack_range): New.
107 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
108 (output_probe_stack_range_stack_clash): New.
109 (rs6000_emit_prologue): Emit notes into dump file as requested.
110 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
111 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
112 Add additional operand and pass it to output_probe_stack_range.
114 2017-09-25 Bin Cheng <bin.cheng@arm.com>
116 PR tree-optimization/82163
117 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
118 (checking_verify_loop_closed_ssa): New parameter.
119 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
120 (check_loop_closed_ssa_stmt): Delete.
121 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
122 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
123 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
126 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
128 * brig-builtins.def: Treat HSAIL barrier builtins as
129 setjmp/longjump style functions.
131 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
133 * target.def (constant_alignment): New hook.
134 * defaults.h (CONSTANT_ALIGNMENT): Delete.
135 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
136 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
137 * doc/tm.texi: Regenerate.
138 * targhooks.h (default_constant_alignment): Declare.
139 (constant_alignment_word_strings): Likewise.
140 * targhooks.c (default_constant_alignment): New function.
141 (constant_alignment_word_strings): Likewise.
142 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
143 instead of CONSTANT_ALIGNMENT.
144 * varasm.c (align_variable, get_variable_align, build_constant_desc)
145 (force_const_mem): Likewise.
146 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
147 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
148 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
149 (TARGET_CONSTANT_ALIGNMENT): Redefine.
150 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
152 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
153 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
154 constant_alignment_word_strings.
155 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
156 (CONSTANT_ALIGNMENT): Likewise.
157 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
158 (arm_constant_alignment): New function.
159 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
160 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
161 constant_alignment_word_strings.
162 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
163 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
164 constant_alignment_word_strings.
165 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
166 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
167 (cris_constant_alignment): New function.
168 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
169 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
170 (epiphany_constant_alignment): New function.
171 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
172 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
173 constant_alignment_word_strings.
174 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
175 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
176 constant_alignment_word_strings.
177 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
178 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
179 constant_alignment_word_strings.
180 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
181 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
182 * config/i386/i386.c (ix86_constant_alignment): Make static.
183 Use the same interface as the target hook.
184 (TARGET_CONSTANT_ALIGNMENT): Redefine.
185 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
186 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
187 constant_alignment_word_strings.
188 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
189 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
190 (TARGET_CONSTANT_ALIGNMENT): Redefine.
191 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
192 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
193 constant_alignment_word_strings.
194 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
195 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
196 constant_alignment_word_strings.
197 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
198 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
199 constant_alignment_word_strings.
200 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
201 * config/microblaze/microblaze.c (microblaze_constant_alignment):
203 (TARGET_CONSTANT_ALIGNMENT): Redefine.
204 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
205 * config/mips/mips.c (mips_constant_alignment): New function.
206 (TARGET_CONSTANT_ALIGNMENT): Redefine.
207 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
208 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
209 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
210 (mmix_constant_alignment): Make static. Use the same interface
212 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
213 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
214 constant_alignment_word_strings.
215 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
216 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
217 constant_alignment_word_strings.
218 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
219 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
220 constant_alignment_word_strings.
221 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
222 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
223 (rs6000_constant_alignment): New function.
224 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
225 * config/riscv/riscv.c (riscv_constant_alignment): New function.
226 (TARGET_CONSTANT_ALIGNMENT): Redefine.
227 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
228 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
229 (rs6000_constant_alignment): New function.
230 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
231 * config/s390/s390.c (s390_constant_alignment): New function.
232 (TARGET_CONSTANT_ALIGNMENT): Redefine.
233 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
234 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
235 constant_alignment_word_strings.
236 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
237 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
238 (sparc_constant_alignment): New function.
239 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
240 * config/spu/spu.c (spu_constant_alignment): New function.
241 (TARGET_CONSTANT_ALIGNMENT): Redefine.
242 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
243 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
244 constant_alignment_word_strings.
245 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
246 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
247 constant_alignment_word_strings.
248 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
249 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
250 constant_alignment_word_strings.
251 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
252 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
253 (visium_constant_alignment): New function.
254 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
255 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
256 (xtensa_constant_alignment): New function.
257 * system.h (CONSTANT_ALIGNMENT): Poison.
259 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
261 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
262 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
263 (rs6000_builtin_valid_without_lhs): New helper function.
264 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
265 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
267 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
269 * target.h (vec_perm_indices): Use unsigned short rather than
271 (auto_vec_perm_indices): Likewise.
272 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
273 Use unsigned int rather than unsigned char.
274 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
276 2017-09-25 Richard Biener <rguenther@suse.de>
278 * cfgloop.h (sort_sibling_loops): Declare.
279 * cfgloop.c (sort_sibling_loops_cmp): New helper.
280 (sort_sibling_loops): New function sorting the sibling loop list
282 * graphite.c (graphite_transform_loops): Sort sibling loops.
284 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
286 * target.def (vec_perm_const_ok): Change sel parameter to
288 * optabs-query.c (can_vec_perm_p): Update accordingly.
289 * doc/tm.texi: Regenerate.
290 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
291 auto_vec_perm_indices and remove separate nelt field.
292 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
293 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
294 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
295 (aarch64_expand_vec_perm_const): Update accordingly.
296 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
298 * config/arm/arm.c (expand_vec_perm_d): Change perm to
299 auto_vec_perm_indices and remove separate nelt field.
300 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
301 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
302 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
304 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
306 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
307 sel to vec_perm_indices.
308 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
309 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
310 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
312 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
315 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
318 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
319 on the FUNCTION_DECL function context if it has a DIE that is a
322 2017-09-25 Richard Biener <rguenther@suse.de>
324 PR tree-optimization/82285
325 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
328 2017-09-25 Tom de Vries <tom@codesourcery.com>
332 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
335 2017-09-25 Richard Biener <rguenther@suse.de>
337 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
338 ISL errors other than isl_error_quota happen. Dump if the
339 schedule is the same.
340 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
341 errors instead of aborting inside ISL.
343 2017-09-25 Iain Sandoe <iain@codesourcery.com>
346 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
347 of libgcc_eh for m64.
348 * config/i386/darwin64.h: Likewise.
350 2017-09-25 Richard Biener <rguenther@suse.de>
353 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
354 attribute for incomplete types nor twice for complete ones.
356 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
359 * config/i386/i386.c (ix86_print_operand_address_as): Encode
360 %esp as %rsp to avoid 0x67 prefix if there is no index or base
363 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
366 * config/i386/i386.opt (mprefer-avx256): Use
367 ix86_target_flags variable.
368 * config/i386/i386.c (ix86_target_string): Move
369 -mprefer-avx256 to flag2_opts.
371 2017-09-22 Jakub Jelinek <jakub@redhat.com>
374 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
375 and x != -1 | y != -1 into (x & y) != -1.
377 2017-09-22 Steve Ellcey <sellcey@cavium.com>
379 * config.gcc: Add new case statement to set
380 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
381 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
382 s390x-*-linux* case statements. Added aarch64 to the list of
383 supported architectures.
385 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
387 PR tree-optimization/82289
388 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
389 STMT_VINFO_RELEVANT_P.
391 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
392 Alan Hayward <alan.hayward@arm.com>
393 David Sherwood <david.sherwood@arm.com>
395 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
396 for VR_RANGE only; don't allow VR_ANTI_RANGE.
397 (extract_range_from_binary_expr_1): Don't call
398 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
400 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
401 Alan Hayward <alan.hayward@arm.com>
402 David Sherwood <david.sherwood@arm.com>
404 * target.def (preferred_vector_alignment): New hook.
405 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
407 * doc/tm.texi: Regenerate.
408 * targhooks.h (default_preferred_vector_alignment): Declare.
409 * targhooks.c (default_preferred_vector_alignment): New function.
410 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
412 (DR_TARGET_ALIGNMENT): New macro.
413 (aligned_access_p): Update commentary.
414 (vect_known_alignment_in_bytes): New function.
415 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
417 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
418 Calculate the misalignment based on the target alignment rather than
420 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
421 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
422 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
423 the target alignment, rather than masking the element misalignment
424 with the number of elements in a vector. Also use the target
425 alignment when calculating the maximum number of peels.
426 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
427 instead of TYPE_ALIGN_UNIT.
428 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
429 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
430 (vect_create_addr_base_for_vector_ref): Update call accordingly.
431 (vect_create_data_ref_ptr): Likewise.
432 (vect_setup_realignment): Realign by ANDing with
433 -DR_TARGET_MISALIGNMENT.
434 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
435 the number of peels based on DR_TARGET_ALIGNMENT.
436 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
437 with the guaranteed alignment boundary when deciding whether
439 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
440 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
441 (ensure_base_align): Remove stmt_info parameter. Get the
442 target base alignment from DR_TARGET_ALIGNMENT.
443 (vectorizable_store): Update call accordingly. Interpret
444 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
446 (vectorizable_load): Likewise.
448 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
449 Alan Hayward <alan.hayward@arm.com>
450 David Sherwood <david.sherwood@arm.com>
452 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
453 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
454 (vect_enhance_data_refs_alignment): Likewise.
456 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
458 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
459 error. Only quit immediately if parsing is complete.
460 (BEGIN): Initialize fatal_err and parse_done.
461 (begin fpu, end fpu): Check number of arguments.
462 (begin arch, end arch): Likewise.
463 (begin cpu, end cpu): Likewise.
464 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
465 (optalias): Likewise.
467 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
469 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
470 * config/arm/arm-isa.h: Delete. Move definitions to ...
471 * arm-cpus.in: ... here. Use new feature and fgroup values.
472 * config/arm/arm.c (arm_option_override): Use lower case for feature
474 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
475 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
476 * config/arm/parsecpu.awk (END): Add new command 'isa'.
478 (print_isa_bits_for): New function.
479 (gen_isa): New function.
480 (gen_comm_data): Use print_isa_bits_for.
481 (define feature): New keyword.
482 (define fgroup): New keyword.
483 * config/arm/t-arm (TM_H): Remove.
484 (GTM_H): Add arm-isa.h.
485 (arm-isa.h): Add rule to generate file.
486 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
487 case for feature bit names.
489 2017-09-22 Richard Biener <rguenther@suse.de>
491 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
493 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
494 print of no dependency loops ...
495 * graphite.c (graphite_transform_loops): ... here.
496 (canonicalize_loop_closed_ssa_form): Work from inner to outer
498 (same_close_phi_node, remove_duplicate_close_phi,
499 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
500 (canonicalize_loop_closed_ssa): ... here and simplify.
501 * graphite-optimize-isl.c: Include tree-vectorizer.h.
502 (optimize_isl): Use dump_printf_loc to tell when we stopped
503 optimizing because of an ISL timeout.
505 2017-09-22 Richard Biener <rguenther@suse.de>
507 PR tree-optimization/82291
508 * tree-if-conv.c (predicate_mem_writes): Make sure to
509 remove writes in blocks predicated with false.
511 2017-09-22 Richard Biener <rguenther@suse.de>
513 * sese.c: Include cfganal.h.
514 (if_region_set_false_region): Remove.
515 (create_if_region_on_edge): Likewise.
516 (move_sese_in_condition): Re-implement without destroying
519 2017-09-22 Richard Biener <rguenther@suse.de>
521 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
522 Verify both BBs contain loop PHI nodes before dispatching to
524 (graphite_regenerate_ast_isl): Do not recompute dominators,
525 do not verify three times. Restructure for clarity.
526 * graphite-scop-detection.c (same_close_phi_node,
527 remove_duplicate_close_phi, make_close_phi_nodes_unique,
528 defined_in_loop_p, canonicalize_loop_closed_ssa,
529 canonicalize_loop_closed_ssa_form): Simplify, remove excess
530 checking and SSA rewrite, move to ...
531 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
532 (graphite_initialize): Do not pass in ctx, do not reset the
533 SCEV cache, compute only dominators.
534 (graphite_transform_loops): Allocate ISL ctx after
535 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
536 Maintain post-dominators only around build_scops.
537 * sese.c (if_region_set_false_region): Make static. Free
538 and recompute dominators.
539 (move_sese_in_condition): Assert we don't get called with
540 post-dominators computed.
541 * sese.h (if_region_set_false_region): Remove.
543 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
545 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
546 mode attribute for TARGET_AVX512VL.
548 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
550 * config/i386/i386.opt (mprefer-avx256): New option.
551 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
553 (ix86_preferred_simd_mode): Return 256-bit AVX modes
554 for TARGET_PREFER_AVX256.
555 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
557 2017-09-21 Jeff Law <law@redhat.com>
559 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
560 Fix dump output if the only stack space is for pushed registers.
562 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
564 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
567 2017-09-21 Martin Sebor <msebor@redhat.com>
570 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
571 code (in C++) or code that triggers warnings.
573 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
575 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
577 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
579 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
580 * sched-rgn.c: Ditto.
581 * sel-sched-ir.c: Ditto.
583 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
585 * toplev.h (set_random_seed): Adjust return type.
586 * toplev.c (init_local_tick): Move eager initialization of random_seed
587 to get_random_seed. Adjust comment.
588 (init_random_seed): Inline to get_random_seed, delete.
589 (get_random_seed): Initialize random_seed lazily.
590 (set_random_seed): Do not return previous value.
591 (print_switch_value): Do not call get_random_seed.
593 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
595 * cgraph.c (delete_function_version): New, broken out from...
596 (cgraph_node::delete_function_version): ...here. Rename to
597 cgraph_node::delete_function_version_by_decl. Update all uses.
598 (cgraph_node::remove): Call delete_function_version.
600 2017-09-21 Jakub Jelinek <jakub@redhat.com>
603 * tree-inline.c (expand_call_inline): Emit clobber stmts for
604 VAR_DECLs to which addressable non-volatile parameters are mapped
605 and for id->retvar after the return value assignment. Clear
606 id->retval and id->retbnd after inlining.
608 2017-09-21 Richard Biener <rguenther@suse.de>
610 PR tree-optimization/82276
611 PR tree-optimization/82244
612 * tree-vrp.c (build_assert_expr_for): Set
613 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
615 (remove_range_assertions): Revert earlier change.
617 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
620 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
622 2017-09-21 Richard Biener <rguenther@suse.de>
624 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
625 Restore valid IL after code generation errors.
626 * graphite.c (graphite_transform_loops): Diagnose code
627 generation issues as MSG_MISSED_OPTIMIZATION and continue
628 with processing SCOPs.
630 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
631 Alan Hayward <alan.hayward@arm.com>
632 David Sherwood <david.sherwood@arm.com>
634 * calls.c (compute_argument_addresses): Use simplify_gen_binary
635 rather than choosing between plus_constant and gen_rtx_<CODE>.
636 * expr.c (emit_push_insn): Likewise.
637 (expand_expr_real_2): Likewise.
639 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
640 Alan Hayward <alan.hayward@arm.com>
641 David Sherwood <david.sherwood@arm.com>
643 * loop-unroll.c (split_iv): Call copy_rtx on the step.
645 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
646 Alan Hayward <alan.hayward@arm.com>
647 David Sherwood <david.sherwood@arm.com>
649 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
650 calling tree_to_uhwi.
652 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
653 Alan Hayward <alan.hayward@arm.com>
654 David Sherwood <david.sherwood@arm.com>
656 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
657 INTEGER_CST rather than a negative test for ADDR_EXPR.
659 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
660 Alan Hayward <alan.hayward@arm.com>
661 David Sherwood <david.sherwood@arm.com>
663 * tree-vrp.c (extract_range_from_binary_expr_1): Check
664 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
666 2017-09-21 Richard Biener <rguenther@suse.de>
668 PR tree-optimization/71351
669 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
670 graphite_create_new_loop_guard): Remove, fold remaining parts
672 (translate_isl_ast_node_for): ... here and simplify.
674 2017-09-21 Jakub Jelinek <jakub@redhat.com>
677 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
678 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
679 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
680 alternative always use QI mode, for -Os imov (=R,R) alternative
681 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
684 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
685 Jeff Law <law@redhat.com>
687 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
688 (allocate_stack_space): New function, partially extracted from
690 (s390_emit_prologue): Track offset to most recent stack probe.
691 Code to allocate space moved into allocate_stack_space.
692 Dump actions when no stack is allocated.
693 (s390_prologue_plus_offset): New function.
694 (s390_emit_stack_probe): Likewise.
696 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
698 * common.opt (Wa, Wl, Wp, g, gz=): Add
700 (gno-column-info): Remove.
701 (gcolumn-info): Drop RejectNegative.
703 (gno-record-gcc-switches): Remove.
704 (grecord-gcc-switches): Drop RejectNegative.
705 (gno-split-dwarf): Remove.
706 (gsplit-dwarf): Drop RejectNegative.
707 (gno-strict-dwarf): Remove.
708 (gstrict-dwarf): Drop RejectNegative.
709 * config/darwin.opt (gfull, gused): Add RejectNegative.
710 * dwarf2out.c (gen_producer_string): Drop
711 gno-record-gcc-switches handler.
712 * optc-gen.awk: Add g to prefixes with negative forms.
713 * opts-common.c (remapping_prefix_p): New.
714 (find_opt): Check it.
715 (generate_canonical_option): Test g prefix.
716 (option_map): Add -gno- mapping.
717 (add_misspelling_candidates): Check remapping_prefix_p.
719 2017-09-20 Jeff Law <law@redhat.com>
721 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
722 thinko in stack clash protection support.
724 * explow.c (compute_stack_clash_protection_loop_data): Use
725 CONST_INT_P instead of explicit test. Verify object is a
726 CONST_INT_P before looking at INTVAL.
727 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
728 instead of explicit test.
730 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
733 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
734 address instead of to r1 and r11.
736 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
738 * config.gcc: Support "knm".
739 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
740 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
742 * config/i386/i386.c (m_KNM): Define.
743 (processor_target_table): Add "knm".
745 (ix86_option_override_internal): Add "knm".
746 (ix86_issue_rate): Add PROCESSOR_KNM.
747 (ix86_adjust_cost): Ditto.
748 (ia32_multipass_dfa_lookahead): Ditto.
749 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
750 (fold_builtin_cpu): Add M_INTEL_KNM.
751 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
752 (processor_type): Add PROCESSOR_KNM.
753 * config/i386/x86-tune.def: Add m_KNM.
754 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
756 2017-09-20 Richard Biener <rguenther@suse.de>
758 PR tree-optimization/80213
759 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
760 are allowed in empty BBs as well.
761 (canonicalize_loop_closed_ssa): Also look for other complex
763 (scop_detection::get_sese): Include the loop-closed PHI block
765 (scop_detection::merge_sese): Remove code adding extra blocks.
766 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
767 (build_scops): Assert the final returned scop is invalid.
769 2017-09-20 Richard Biener <rguenther@suse.de>
771 PR tree-optimization/82264
772 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
774 (vn_phi_lookup): Likewise.
775 (vn_phi_insert): Likewise.
777 2017-09-20 Jakub Jelinek <jakub@redhat.com>
779 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
780 that fits into uhwi or shwi, add DW_AT_const_value regardless
781 of early_dwarf without going through RTL, using add_AT_unsigned
784 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
785 (DEBUG_LTO_ABBREV_SECTION): Likewise.
786 (DEBUG_LTO_MACINFO_SECTION): Likewise.
787 (DEBUG_MACRO_SECTION): Likewise.
788 (DEBUG_LTO_MACRO_SECTION): Likewise.
789 (DEBUG_STR_DWO_SECTION): Likewise.
790 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
791 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
792 (DEBUG_LTO_DWO_LINE_SECTION): Define.
793 (DEBUG_LTO_LINE_STR_SECTION): Define.
794 (init_sections_and_labels): Initialize debug_line_str_section
795 variable. Initialize debug_loc_section for -gdwarf-5 to
796 DEBUG_LOCLISTS_SECTION. Formatting fixes.
798 2017-09-20 Richard Biener <rguenther@suse.de>
800 * graphite-sese-to-poly.c (extract_affine): Properly handle
801 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
803 2017-09-20 Richard Biener <rguenther@suse.de>
805 PR tree-optimization/81373
806 * graphite-scop-detection.c (build_cross_bb_scalars_def):
807 Force SESE live-out defs to be handled even if they are
810 2017-09-19 Jeff Law <law@redhat.com>
812 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
813 nothing for stack adjustments with REG_STACK_CHECK.
814 * sched-deps.c (parse_add_or_inc): Reject insns with
815 REG_STACK_CHECK from dependency breaking.
816 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
817 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
818 * reg-notes.def (STACK_CHECK): New note.
820 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
821 (ix86_expand_prologue): Dump stack clash info as needed.
822 Call ix86_adjust_stack_and_probe_stack_clash as needed.
824 * function.c (dump_stack_clash_frame_info): New function.
825 * function.h (dump_stack_clash_frame_info): Prototype.
826 (enum stack_clash_probes): New enum.
828 * config/alpha/alpha.c (alpha_expand_prologue): Also check
829 flag_stack_clash_protection.
830 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
831 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
832 (arm_frame_pointer_required): Likewise.
833 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
834 (ia64_expand_prologue): Likewise.
835 * config/mips/mips.c (mips_expand_prologue): Likewise.
836 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
837 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
838 (sparc_flat_expand_prologue): Likewise.
839 * config/spu/spu.c (spu_expand_prologue): Likewise.
841 * explow.c: Include "params.h".
842 (anti_adjust_stack_and_probe_stack_clash): New function.
843 (get_stack_check_protect): Likewise.
844 (compute_stack_clash_protection_loop_data): Likewise.
845 (emit_stack_clash_protection_loop_start): Likewise.
846 (emit_stack_clash_protection_loop_end): Likewise.
847 (allocate_dynamic_stack_space): Use get_stack_check_protect.
848 Use anti_adjust_stack_and_probe_stack_clash.
849 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
850 (emit_stack_clash_protection_loop_start): Likewise.
851 (emit_stack_clash_protection_loop_end): Likewise.
852 * rtl.h (get_stack_check_protect): Prototype.
853 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
854 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
855 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
857 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
859 * doc/tm.texi: Rebuilt.
860 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
861 get_stack_check_protect.
862 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
863 * config/arm/arm.c (arm_expand_prologue): Likewise.
864 (arm_frame_pointer_required): Likewise.
865 * config/i386/i386.c (ix86_expand_prologue): Likewise.
866 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
867 * config/mips/mips.c (mips_expand_prologue): Likewise.
868 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
869 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
870 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
871 (sparc_flat_expand_prologue): Likewise.
873 * common.opt (-fstack-clash-protection): New option.
874 * flag-types.h (enum stack_check_type): Note difference between
875 -fstack-check= and -fstack-clash-protection.
876 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
877 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
878 * toplev.c (process_options): Issue warnings/errors for cases
879 not handled with -fstack-clash-protection.
880 * doc/invoke.texi (-fstack-clash-protection): Document new option.
881 (-fstack-check): Note additional problem with -fstack-check=generic.
882 Note that -fstack-check is primarily for Ada and refer users
883 to -fstack-clash-protection for stack-clash-protection.
884 Document new params for stack clash protection.
886 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
888 * config/i386/i386.c (ix86_split_long_move): Do not handle
889 address used for LEA in a special way.
891 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
893 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
895 2017-09-19 Martin Sebor <msebor@redhat.com>
898 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
899 of incompatible types.
901 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
903 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
904 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
905 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
906 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
908 2017-09-19 Richard Biener <rguenther@suse.de>
910 PR tree-optimization/82244
911 * tree-vrp.c (remove_range_assertions): Do not propagate
912 a constant to abnormals but replace the assert with a copy.
914 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
916 PR rtl-optimization/57878
917 PR rtl-optimization/68988
918 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
919 avoidance test involving non_reload_pseudos. Move frequency test
920 below the general fragmentation avoidance test.
922 2017-09-19 Richard Biener <rguenther@suse.de>
924 PR tree-optimization/69728
925 * graphite-sese-to-poly.c (schedule_error): New global.
926 (add_loop_schedule): Handle empty domain by failing the
928 (build_original_schedule): Handle schedule_error.
930 2017-09-19 Richard Biener <rguenther@suse.de>
932 * graphite-scop-detection.c (scop_detection::can_represent_loop):
933 Do not iterate to sibling loops but only to siblings of inner
936 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
939 * config/m68k/m68k.md (moveq feeding equality comparison): Check
940 that the registers are different.
942 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
944 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
945 to processor_model and "amdfam17h" to arch_names_table.
946 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
948 2017-09-18 Jakub Jelinek <jakub@redhat.com>
951 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
953 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
954 Alan Hayward <alan.hayward@arm.com>
955 David Sherwood <david.sherwood@arm.com>
957 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
959 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
961 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
962 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
963 vect_schedule_slp_instance.
964 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
965 Update call to vect_slp_analyze_node_operations. Simplify return
967 (vect_slp_analyze_bb_1): Update call accordingly.
968 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
969 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
970 (vect_schedule_slp): Update call accordingly.
972 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
973 Alan Hayward <alan.hayward@arm.com>
974 David Sherwood <david.sherwood@arm.com>
976 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
977 with types that aren't in fact scalar.
979 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
981 * tree-vect-slp.c (vect_record_max_nunits): New function,
983 (vect_build_slp_tree_1): ...here.
984 (vect_build_slp_tree_2): Call it for phis too.
986 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
988 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
989 to vect_get_vec_def_for_operand when getting the mask operand.
991 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
992 Alan Hayward <alan.hayward@arm.com>
993 David Sherwood <david.sherwood@arm.com>
995 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
998 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
999 Alan Hayward <alan.hayward@arm.com>
1000 David Sherwood <david.sherwood@arm.com>
1002 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
1003 calculation for vector booleans.
1005 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1006 Alan Hayward <alan.hayward@arm.com>
1007 David Sherwood <david.sherwood@arm.com>
1009 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
1011 (vect_transform_stmt): ...here.
1012 (vect_analyze_stmt): Use it instead of calling
1013 vectorizable_live_operation directly.
1015 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
1017 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
1018 non-SIMT targets in acc vector loops.
1020 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
1022 * configure.ac: Add arc and check if assembler supports gdwarf2.
1023 * configure: Regenerate.
1025 2017-09-18 Richard Biener <rguenther@suse.de>
1027 PR tree-optimization/82220
1028 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
1029 epilogue niters from the min_profitable_iters compute.
1031 2017-09-18 Jakub Jelinek <jakub@redhat.com>
1034 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
1035 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
1036 (ix86_init_pic_reg): Revert 2017-09-01 changes.
1038 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
1041 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
1042 switching to a new text section.
1044 2017-09-18 Richard Biener <rguenther@suse.de>
1046 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
1048 (build_alias_set): Reject aliases with no access function.
1050 2017-09-18 Richard Biener <rguenther@suse.de>
1052 PR tree-optimization/79622
1053 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
1055 (build_cross_bb_scalars_use): Likewise.
1057 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
1059 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
1061 2017-09-18 Alan Modra <amodra@gmail.com>
1064 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
1065 stack_pointer_rtx for count 0. Update comments. Break up
1066 large rtl expression.
1068 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
1070 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
1071 Increase to 20 bytes.
1072 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
1073 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
1074 or avx version of the stub.
1076 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
1079 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
1080 compute the minimum stack alignment. Also update preferred stack
1081 boundary for leaf functions.
1083 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
1085 PR tree-optimization/82228
1086 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
1089 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
1091 * common/config/nds32/nds32-common.c
1092 (nds32_option_optimization_table): Refine formatting.
1093 (nds32_option_optimization_table): Use -fsched-pressure and
1094 -fomit-frame-pointer for specific optimization level.
1096 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
1098 * config/nds32/nds32.c: Refine formatting and comments.
1099 * config/nds32/nds32.h: Likewise.
1100 * config/nds32/nds32.md: Likewise.
1101 * config/nds32/nds32-cost.c: Likewise.
1102 * config/nds32/nds32-isr.c: Likewise.
1103 * config/nds32/nds32-md-auxiliary.c: Likewise.
1104 * config/nds32/nds32-multiple.md: Likewise.
1105 * config/nds32/nds32-predicates.c: Likewise.
1107 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
1108 Jakub Jelinek <jakub@redhat.com>
1110 Add support for -std=c++2a.
1111 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
1113 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
1115 2017-09-15 Steve Ellcey <sellcey@cavium.com>
1118 * doc/extend.texi (Common Function Attributes): Add
1119 references to ARM, AArch64, and S/390 specific attributes.
1120 (Function Specific Option Pragmas): Add AArch64 and S/390
1121 to list of back ends that support the target pragma.
1123 2017-09-15 Nathan Sidwell <nathan@acm.org>
1125 * doc/standards.texi: Fix C++17 description. Update URLs for
1128 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
1130 * common.opt (Wcast-align=strict): New warning option.
1131 * doc/invoke.texi: Document -Wcast-align=strict.
1133 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
1135 * cgraph.h (cgraph_thunk_info): Add comments.
1136 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
1137 assert for VIRTUAL_* arguments stricter.
1139 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
1141 PR tree-optimization/71026
1142 * match.pd: Move RDIV patterns from fold-const.c
1143 * fold-const.c (distribute_real_division): Removed.
1144 (fold_binary_loc): Remove calls to distribute_real_divison.
1146 2017-09-15 Jakub Jelinek <jakub@redhat.com>
1148 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
1149 c++1z and gnu++1z as deprecated. Change other references to
1150 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
1151 Change -Wc++1z-compat to -Wc++17-compat.
1152 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
1153 * dwarf2out.c (highest_c_language): Handle C++17.
1154 (gen_compile_unit_die): Likewise.
1156 2017-09-15 Jakub Jelinek <jakub@redhat.com>
1158 PR rtl-optimization/82192
1159 * combine.c (make_extraction): Don't look through non-paradoxical
1160 SUBREGs or TRUNCATE if pos + len is or might be bigger than
1163 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
1164 Alan Hayward <alan.hayward@arm.com>
1165 David Sherwood <david.sherwood@arm.com>
1167 * target.def (function_arg_offset): New hook.
1168 * targhooks.h (default_function_arg_offset): Declare.
1169 * targhooks.c (default_function_arg_offset): New function.
1170 * function.c (locate_and_pad_parm): Use
1171 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
1172 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
1173 (TARGET_FUNCTION_ARG_OFFSET): ...this.
1174 * doc/tm.texi: Regenerate.
1175 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
1176 * config/spu/spu.c (spu_function_arg_offset): New function.
1177 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
1178 * system.h (FUNCTION_ARG_OFFSET): Poison.
1180 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
1181 Alan Hayard <alan.hayward@arm.com>
1182 David Sherwood <david.sherwood@arm.com>
1184 * target.def (truly_noop_truncation): New hook.
1185 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
1186 than TRULY_NOOP_TRUNCATION.
1187 * hooks.h (hook_bool_uint_uint_true): Declare.
1188 * hooks.c (hook_bool_uint_uint_true): New function.
1189 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
1190 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
1191 * doc/tm.texi: Regenerate.
1192 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
1193 rather than TRULY_NOOP_TRUNCATION in comments.
1194 (simplify_comparison): Likewise.
1195 (record_truncated_value): Likewise.
1196 * expmed.c (extract_bit_field_1): Likewise.
1197 (extract_split_bit_field): Likewise.
1198 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
1199 instead of TRULY_NOOP_TRUNCATION.
1200 * function.c (assign_parm_setup_block): Likewise.
1201 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
1202 * rtlhooks.c: Include target.h.
1203 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
1204 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
1205 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
1206 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
1207 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
1208 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
1209 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
1210 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
1211 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
1212 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
1213 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
1214 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
1215 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
1216 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
1217 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
1218 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
1219 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
1220 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
1221 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
1222 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
1223 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
1224 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
1225 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
1226 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
1227 * config/mips/mips.c (mips_truly_noop_truncation): New function.
1228 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
1229 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
1230 rather than TRULY_NOOP_TRUNCATION in comments.
1231 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
1232 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
1233 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
1234 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
1235 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
1236 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
1237 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
1238 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
1239 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
1240 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
1241 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
1242 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
1243 rather than TRULY_NOOP_TRUNCATION in comments.
1244 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
1245 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
1246 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
1247 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
1248 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
1249 TRULY_NOOP_TRUNCATION condition.
1250 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
1251 (TRULY_NOOP_TRUNCATION): Delete.
1252 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
1253 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
1254 * config/spu/spu.c (spu_truly_noop_truncation): New function.
1255 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
1256 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
1257 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
1258 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
1259 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
1260 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
1261 rather than TRULY_NOOP_TRUNCATION in comments.
1262 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
1263 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
1264 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
1265 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
1266 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
1267 * system.h (TRULY_NOOP_TRUNCATION): Poison.
1269 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
1272 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
1273 (*cmp_ior): Likewise.
1274 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
1275 (*ior_scc_scc_cmp): Likewise.
1276 (*and_scc_scc): Likewise.
1277 (*and_scc_scc_cmp): Likewise.
1279 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
1280 Alan Hayard <alan.hayward@arm.com>
1281 David Sherwood <david.sherwood@arm.com>
1283 * target.def (can_change_mode_class): New hook.
1284 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
1285 (hard_regno_nregs): Likewise.
1286 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
1287 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
1288 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
1289 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
1290 (LOAD_EXTEND_OP): Update accordingly.
1291 * doc/tm.texi: Regenerate.
1292 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
1293 CANNOT_CHANGE_MODE_CLASS.
1294 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
1295 (REG_CAN_CHANGE_MODE_P): ...this new macro.
1296 * combine.c (simplify_set): Update accordingly.
1297 * emit-rtl.c (validate_subreg): Likewise.
1298 * recog.c (general_operand): Likewise.
1299 * regcprop.c (mode_change_ok): Likewise.
1300 * reload1.c (choose_reload_regs): Likewise.
1301 (inherit_piecemeal_p): Likewise.
1302 * rtlanal.c (simplify_subreg_regno): Likewise.
1303 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
1304 instead of CANNOT_CHANGE_MODE_CLASS.
1305 (reload_cse_simplify_operands): Likewise.
1306 * reload.c (push_reload): Use targetm.can_change_mode_class
1307 instead of CANNOT_CHANGE_MODE_CLASS.
1308 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
1309 REG_CANNOT_CHANGE_MODE_P.
1310 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1311 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
1312 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1313 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1314 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1315 (arm_can_change_mode_class): New function.
1316 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
1317 than CANNOT_CHANGE_MODE_CLASS in comments.
1318 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1319 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
1320 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
1321 (ix86_can_change_mode_class): ...this new function, inverting the
1322 sense of the return value.
1323 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1324 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1325 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1326 (ia64_can_change_mode_class): New function.
1327 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1328 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
1329 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
1330 (m32c_can_change_mode_class): ...this new function, inverting the
1331 sense of the return value.
1332 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1333 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1334 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
1335 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
1336 (mips_can_change_mode_class): ...this new function, inverting the
1337 sense of the return value.
1338 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1339 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1340 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1341 (msp430_can_change_mode_class): New function.
1342 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1343 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
1344 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1345 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1346 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1347 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
1348 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1349 (pa_cannot_change_mode_class): Replace with...
1350 (pa_can_change_mode_class): ...this new function, inverting the
1351 sense of the return value.
1352 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
1353 than CANNOT_CHANGE_MODE_CLASS in comments.
1354 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1355 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
1356 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1357 (pdp11_cannot_change_mode_class): Replace with...
1358 (pdp11_can_change_mode_class): ...this new function, inverting the
1359 sense of the return value.
1360 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1361 * config/powerpcspe/powerpcspe-protos.h
1362 (rs6000_cannot_change_mode_class_ptr): Delete.
1363 * config/powerpcspe/powerpcspe.c
1364 (rs6000_cannot_change_mode_class_ptr): Delete.
1365 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1366 (rs6000_option_override_internal): Assign to
1367 targetm.can_change_mode_class instead of
1368 rs6000_cannot_change_mode_class_ptr.
1369 (rs6000_cannot_change_mode_class): Replace with...
1370 (rs6000_can_change_mode_class): ...this new function, inverting the
1371 sense of the return value.
1372 (rs6000_debug_cannot_change_mode_class): Replace with...
1373 (rs6000_debug_can_change_mode_class): ...this new function.
1374 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1375 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
1376 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1377 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1378 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
1380 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
1381 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1382 (rs6000_option_override_internal): Assign to
1383 targetm.can_change_mode_class instead of
1384 rs6000_cannot_change_mode_class_ptr.
1385 (rs6000_cannot_change_mode_class): Replace with...
1386 (rs6000_can_change_mode_class): ...this new function, inverting the
1387 sense of the return value.
1388 (rs6000_debug_cannot_change_mode_class): Replace with...
1389 (rs6000_debug_can_change_mode_class): ...this new function.
1390 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1391 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
1392 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
1393 (s390_can_change_mode_class): ...this new function, inverting the
1394 sense of the return value.
1395 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1396 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1397 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
1398 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1399 (sh_cannot_change_mode_class): Replace with...
1400 (sh_can_change_mode_class): ...this new function, inverting the
1401 sense of the return value.
1402 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1403 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1404 (sparc_can_change_mode_class): New function.
1405 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1406 * config/spu/spu.c (spu_can_change_mode_class): New function.
1407 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1408 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1409 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1410 (visium_can_change_mode_class): New function.
1411 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
1413 2017-09-15 Richard Biener <rguenther@suse.de>
1415 PR tree-optimization/82217
1416 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
1417 but not undefined case.
1419 2017-09-15 Jakub Jelinek <jakub@redhat.com>
1422 * postreload.c (reload_cse_simplify_operands): Skip
1423 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
1425 2017-09-15 Richard Biener <rguenther@suse.de>
1427 PR tree-optimization/68823
1428 * graphite-scop-detection.c (build_alias_set): If we have a
1429 possible dependence check whether we can handle them by just
1430 looking at the DRs DR_ACCESS_FNs.
1431 (build_scops): If build_alias_set fails, fail the SCOP.
1433 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
1435 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
1436 to support float128 built-in functions that require the ISA 3.0
1438 (BU_FLOAT128_3_HW): Likewise.
1439 (SQRTF128): Add support for the IEEE 128-bit square root and fma
1441 (FMAF128): Likewise.
1443 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
1444 support for built-in functions that need the ISA 3.0 IEEE 128-bit
1445 floating point instructions.
1446 (rs6000_invalid_builtin): Likewise.
1447 (rs6000_builtin_mask_names): Likewise.
1448 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
1449 (RS6000_BTM_FLOAT128_HW): Likewise.
1450 (RS6000_BTM_COMMON): Likewise.
1451 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
1453 * doc/extend.texi (RS/6000 built-in functions): Document the
1454 IEEE 128-bit floating point square root and fused multiply-add
1457 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
1459 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
1460 reg (r2) isn't in the set of registers defined in the prologue.
1462 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
1463 Alan Hayward <alan.hayward@arm.com>
1464 David Sherwood <david.sherwood@arm.com>
1466 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
1467 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
1468 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
1469 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
1470 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
1472 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
1473 max_vectorization_factor.
1474 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
1476 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
1477 Alan Hayward <alan.hayward@arm.com>
1478 David Sherwood <david.sherwood@arm.com>
1480 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
1481 (vect_worthwhile_without_simd_p): Declare.
1482 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
1483 (vectorizable_reduction): Use it.
1484 * tree-vect-stmts.c (vectorizable_shift): Likewise.
1485 (vectorizable_operation): Likewise.
1487 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
1488 Alan Hayward <alan.hayward@arm.com>
1489 David Sherwood <david.sherwood@arm.com>
1491 * tree-vectorizer.h (vect_get_num_copies): New function.
1492 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
1493 * tree-vect-loop.c (vectorizable_reduction): Likewise.
1494 (vectorizable_induction): Likewise.
1495 (vectorizable_live_operation): Likewise.
1496 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
1497 (vectorizable_bswap): Likewise.
1498 (vectorizable_call): Likewise.
1499 (vectorizable_conversion): Likewise.
1500 (vectorizable_assignment): Likewise.
1501 (vectorizable_shift): Likewise.
1502 (vectorizable_operation): Likewise.
1503 (vectorizable_store): Likewise.
1504 (vectorizable_load): Likewise.
1505 (vectorizable_condition): Likewise.
1506 (vectorizable_comparison): Likewise.
1507 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
1509 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
1510 Alan Hayward <alan.hayward@arm.com>
1511 David Sherwood <david.sherwood@arm.com>
1513 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
1514 of vect_init_vector.
1516 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
1517 Alan Hayward <alan.hayward@arm.com>
1518 David Sherwood <david.sherwood@arm.com>
1520 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
1521 an inline wrapper that provides a location.
1522 (gimple_build_vector): Likewise.
1523 * gimple-fold.c (gimple_build_vector_from_val): New function.
1524 (gimple_build_vector): Likewise.
1525 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
1526 functions to build the initial value. Always return a gimple value.
1527 (get_initial_defs_for_reduction): Likewise. Only compute
1529 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
1530 vect_init_vector on the results from get_initial_def(s)_for_reduction.
1531 (vectorizable_induction): Use gimple_build_vector rather than
1534 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
1535 Alan Hayward <alan.hayward@arm.com>
1536 David Sherwood <david.sherwood@arm.com>
1538 * target.h (vec_perm_indices): New typedef.
1539 (auto_vec_perm_indices): Likewise.
1540 * optabs-query.h: Include target.h
1541 (can_vec_perm_p): Take a vec_perm_indices *.
1542 * optabs-query.c (can_vec_perm_p): Likewise.
1543 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
1544 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
1545 * tree-vect-generic.c (lower_vec_perm): Likewise.
1546 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
1547 (vect_grouped_load_supported): Likewise.
1548 (vect_shift_permute_load_chain): Likewise.
1549 (vect_permute_store_chain): Use auto_vec_perm_indices.
1550 (vect_permute_load_chain): Likewise.
1551 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
1552 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
1553 Update uses of can_vec_perm_p.
1554 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
1555 mode with a number of elements. Take a vec_perm_indices *.
1556 (vect_create_epilog_for_reduction): Update accordingly.
1557 Use auto_vec_perm_indices.
1558 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
1559 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
1560 (vect_transform_slp_perm_load): Likewise.
1561 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
1562 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
1563 (vect_gen_perm_mask_checked): Likewise.
1564 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
1565 (vect_gen_perm_mask_checked): Likewise.
1566 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
1567 (vectorizable_store): Likewise.
1568 (vectorizable_load): Likewise.
1569 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
1570 (vectorizable_bswap): Likewise.
1572 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
1573 Alan Hayward <alan.hayward@arm.com>
1574 David Sherwood <david.sherwood@arm.com>
1576 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
1577 * tree.c (build_vector): Likewise.
1578 (build_vector_from_ctor): Update accordingly.
1579 (build_vector_from_val): Likewise.
1580 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
1581 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
1582 * tree-vect-generic.c (add_rshift): Likewise.
1583 (expand_vector_divmod): Likewise.
1584 (optimize_vector_constructor): Likewise.
1585 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
1586 (vect_transform_slp_perm_load): Likewise.
1587 (vect_schedule_slp_instance): Likewise.
1588 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
1589 (vectorizable_call): Likewise.
1590 (vect_gen_perm_mask_any): Likewise. Add elements in order.
1591 * expmed.c (make_tree): Likewise.
1592 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
1593 a vector passed to build_vector.
1594 (fold_convert_const): Likewise.
1595 (exact_inverse): Likewise.
1596 (fold_ternary_loc): Likewise.
1597 (fold_relational_const): Likewise.
1598 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
1599 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
1600 (const_unop): Likewise. Store the reduction accumulator in a
1601 variable rather than an array.
1602 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
1603 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
1604 the new vector, rather than constructing it after the input arrays.
1605 (native_interpret_vector): Use auto_vec<tree> when building
1606 a vector passed to build_vector. Add elements in order.
1607 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
1608 auto_vec<tree> when building a vector passed to build_vector.
1609 (vect_create_epilog_for_reduction): Likewise.
1610 (vectorizable_induction): Likewise.
1611 (get_initial_def_for_reduction): Likewise. Fix indentation of
1613 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
1615 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
1616 passed to build_vector.
1618 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
1619 Alan Hayward <alan.hayward@arm.com>
1620 David Sherwood <david.sherwood@arm.com>
1622 * tree-core.h (tree_base::u): Add an "nelts" field.
1623 (tree_vector): Use VECTOR_CST_NELTS as the length.
1624 * tree.c (tree_size): Likewise.
1625 (make_vector): Initialize VECTOR_CST_NELTS.
1626 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
1627 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
1628 TYPE_VECTOR_SUBPARTS.
1629 * expr.c (const_vector_mask_from_tree): Consistently use "units"
1630 as the number of units, setting it from VECTOR_CST_NELTS.
1631 (const_vector_from_tree): Likewise.
1632 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
1633 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
1634 (fold_negate_expr_1): Likewise.
1635 (fold_convert_const): Likewise.
1636 (const_binop): Likewise. Differentiate the number of output and
1638 (const_unop): Likewise.
1639 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
1640 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
1641 in cases that did the opposite.
1643 2017-09-14 Richard Biener <rguenther@suse.de>
1645 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
1648 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
1650 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
1652 2017-09-14 Jakub Jelinek <jakub@redhat.com>
1655 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
1656 if and where to split a bb, except for splitting before debug insn
1657 sequences followed by non-label real insn. Delete debug insns
1658 in between basic blocks.
1660 * combine.c (make_compound_operation_int): Formatting fixes.
1662 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
1663 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
1664 * config/netbsd.h (LINK_EH_SPEC): Likewise.
1665 * config/sol2.h (LINK_EH_SPEC): Likewise.
1666 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
1667 * config/s390/linux.h (LINK_SPEC): Likewise.
1668 * config/freebsd.h (LINK_EH_SPEC): Likewise.
1669 * config/openbsd.h (LINK_EH_SPEC): Likewise.
1670 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
1671 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
1672 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
1673 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
1674 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
1675 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
1677 2017-09-13 Jakub Jelinek <jakub@redhat.com>
1679 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
1681 (ENDFILE_LINUX_SPEC): Likewise.
1682 (LINK_EH_SPEC): Likewise.
1683 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
1684 (LINK_OS_LINUX_SPEC32): Likewise.
1685 (LINK_OS_LINUX_SPEC64): Likewise.
1686 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
1687 (LINK_OS_LINUX_SPEC): Likewise.
1689 2017-09-13 Martin Liska <mliska@suse.cz>
1692 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
1693 CASE_HIGH is NULL_TREE.
1695 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
1696 Alan Hayward <alan.hayward@arm.com>
1697 David Sherwood <david.sherwood@arm.com>
1699 * target.def (secondary_memory_needed): New hook.
1700 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
1701 instead of SECONDARY_MEMORY_NEEDED.
1702 (secondary_memory_needed_mode): Likewise.
1703 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
1704 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
1705 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
1706 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
1707 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
1708 * doc/tm.texi: Regenerate.
1709 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
1710 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
1711 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1712 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
1713 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
1714 * config/i386/i386.c (inline_secondary_memory_needed): Put the
1715 mode argument first and change the reg_class arguments to reg_class_t.
1716 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
1717 Make static. Update the call to inline_secondary_memory_needed.
1718 (ix86_register_move_cost): Update the call to
1719 inline_secondary_memory_needed.
1720 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1721 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
1723 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
1724 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
1726 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
1727 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
1728 * config/mips/mips.c (mips_secondary_memory_needed): Make static
1729 and match hook interface. Add comment from mips.h.
1730 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1731 * config/mmix/mmix.md (truncdfsf2): Refer to
1732 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
1734 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
1735 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
1736 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1737 (pa_secondary_memory_needed): New function.
1738 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
1739 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
1740 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1741 (pdp11_secondary_memory_needed): Make static and match hook interface.
1742 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
1743 * config/powerpcspe/powerpcspe-protos.h
1744 (rs6000_secondary_memory_needed_ptr): Delete.
1745 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
1747 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1748 (rs6000_option_override_internal): Assign to
1749 targetm.secondary_memory_needed rather than
1750 rs6000_secondary_memory_needed_ptr.
1751 (rs6000_secondary_memory_needed): Match hook interface.
1752 (rs6000_debug_secondary_memory_needed): Likewise.
1753 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
1754 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
1755 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
1756 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1757 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
1758 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
1760 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
1761 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1762 (rs6000_option_override_internal): Assign to
1763 targetm.secondary_memory_needed rather than
1764 rs6000_secondary_memory_needed_ptr.
1765 (rs6000_secondary_memory_needed): Match hook interface.
1766 (rs6000_debug_secondary_memory_needed): Likewise.
1767 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
1768 * config/s390/s390.c (s390_secondary_memory_needed): New function.
1769 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1770 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
1771 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1772 (sparc_secondary_memory_needed): New function.
1773 * lra-constraints.c (check_and_process_move): Refer to
1774 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
1776 (curr_insn_transform): Likewise.
1777 (process_alt_operands): Use targetm.secondary_memory_needed
1778 instead of TARGET_SECONDARY_MEMORY_NEEDED.
1779 (check_secondary_memory_needed_p): Likewise.
1780 (choose_split_class): Likewise.
1781 * reload.c: Unconditionally include code that was previously
1782 conditional on SECONDARY_MEMORY_NEEDED.
1783 (push_secondary_reload): Use targetm.secondary_memory_needed
1784 instead of TARGET_SECONDARY_MEMORY_NEEDED.
1785 (push_reload): Likewise.
1786 * reload1.c: Unconditionally include code that was previously
1787 conditional on SECONDARY_MEMORY_NEEDED.
1788 (choose_reload_regs): Use targetm.secondary_memory_needed
1789 instead of TARGET_SECONDARY_MEMORY_NEEDED.
1790 (gen_reload): Likewise.
1791 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
1793 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
1794 Alan Hayward <alan.hayward@arm.com>
1795 David Sherwood <david.sherwood@arm.com>
1797 * target.def (secondary_memory_needed_mode): New hook:
1798 * targhooks.c (default_secondary_memory_needed_mode): Declare.
1799 * targhooks.h (default_secondary_memory_needed_mode): New function.
1800 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
1801 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
1802 * doc/tm.texi: Regenerate.
1803 * lra-constraints.c (check_and_process_move): Use
1804 targetm.secondary_memory_needed_mode instead of
1805 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
1806 (curr_insn_transform): Likewise.
1807 * reload.c (get_secondary_mem): Likewise.
1808 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
1809 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
1811 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
1812 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
1813 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
1814 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
1815 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
1817 * config/powerpcspe/powerpcspe-protos.h
1818 (rs6000_secondary_memory_needed_mode): Delete.
1819 * config/powerpcspe/powerpcspe.c
1820 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
1821 (rs6000_secondary_memory_needed_mode): Make static.
1822 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
1823 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
1825 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
1827 (rs6000_secondary_memory_needed_mode): Make static.
1828 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
1829 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
1830 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
1831 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
1832 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
1834 (sparc_secondary_memory_needed_mode): New function.
1835 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
1837 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
1839 * config/aarch64/constraints.md (Umq): New constraint.
1840 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
1842 (mov<mode>): Update condition.
1844 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1846 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
1847 when bitposition is the same.
1849 2017-09-13 Richard Biener <rguenther@suse.de>
1851 * dwarf2out.c (output_die_symbol): Remove.
1852 (output_die): Do not output a DIEs symbol.
1854 2017-09-13 Richard Biener <rguenther@suse.de>
1857 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
1858 default-def to avoid breaking iterator update with the weird
1859 interaction with cgraph_update_edges_for_call_stmt_node.
1861 2017-09-13 Richard Biener <rguenther@suse.de>
1863 * tree-cfg.c (verify_gimple_assign_binary): Add verification
1864 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
1865 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
1866 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
1868 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
1870 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
1871 Disable pc relative literal load irrespective of
1872 TARGET_FIX_ERR_A53_84341 for default.
1874 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
1876 * config/sparc/sparc.c (output_return): Output the source location of
1877 the insn in the delay slot, if any.
1878 (output_sibcall): Likewise.
1880 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
1883 * common.opt (-static-pie): New alias.
1884 (shared): Negate static-pie.
1885 (-no-pie): Update help text.
1887 (static-pie): New option.
1888 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
1889 -static-pie support.
1890 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
1891 (LINK_EH_SPEC): Likewise.
1892 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
1893 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
1894 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
1895 * gcc.c (LINK_COMMAND_SPEC): Likewise.
1896 (init_gcc_specs): Likewise.
1897 (init_spec): Likewise.
1898 (display_help): Update help message for -pie.
1899 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
1902 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
1904 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
1905 (movdi_aarch64): Likewise.
1906 (movti_aarch64): Likewise.
1908 2017-09-12 Simon Wright <simon@pushface.org>
1911 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
1912 calculation of the minor version, always output as 0.
1914 2017-09-12 Jakub Jelinek <jakub@redhat.com>
1917 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
1918 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
1919 on it early, rather than manual conversion late. For
1920 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
1921 instead of performing manual conversion.
1923 2017-09-12 Carl Love <cel@us.ibm.com>
1925 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
1926 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
1928 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
1929 VMULOSW): Add definitions.
1930 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
1931 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
1932 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
1933 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
1934 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
1936 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
1938 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
1940 (movti_aarch64): Likewise.
1941 (movdf_aarch64): Likewise.
1942 (movtf_aarch64): Likewise.
1943 (load_pairdi): Likewise.
1944 (store_pairdi): Likewise.
1945 (load_pairdf): Likewise.
1946 (store_pairdf): Likewise.
1947 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
1948 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
1949 (ldr_got_small_<mode>): Likewise.
1950 (ldr_got_small_28k_<mode>): Likewise.
1951 (ldr_got_tiny): Likewise.
1952 * config/aarch64/iterators.md (ldst_sz): New.
1953 (ldpstp_sz): Likewise.
1954 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
1956 (thunderx_load): Split load_8 to load_16.
1957 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
1959 (thunderx2t99_storepair_basic): Split store_8 to store_16.
1960 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
1961 (xgene1_store_pair): Split store_8 to store_16.
1962 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
1963 (falkor_st_0_st_sd): Split store_8 to store_16.
1965 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
1967 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
1968 and store1/2/3/4 to store_4/8/12/16.
1969 * config/aarch64/aarch64.md: Update for rename.
1970 * config/arm/arm.md: Likewise.: Likewise.
1971 * config/arm/arm.c: Likewise.
1972 * config/arm/thumb1.md: Likewise.
1973 * config/arm/thumb2.md: Likewise.
1974 * config/arm/vfp.md: Likewise.
1975 * config/arm/arm-generic.md: Likewise.
1976 * config/arm/arm1020e.md: Likewise.
1977 * config/arm/arm1026ejs.md: Likewise.
1978 * config/arm/arm1136jfs.md: Likewise.
1979 * config/arm/arm926ejs.md: Likewise.
1980 * config/arm/cortex-a15.md: Likewise.
1981 * config/arm/cortex-a17.md: Likewise.
1982 * config/arm/cortex-a5.md: Likewise.
1983 * config/arm/cortex-a53.md: Likewise.
1984 * config/arm/cortex-a57.md: Likewise.
1985 * config/arm/cortex-a7.md: Likewise.
1986 * config/arm/cortex-a8.md: Likewise.
1987 * config/arm/cortex-a9.md: Likewise.
1988 * config/arm/cortex-m4.md: Likewise.
1989 * config/arm/cortex-m7.md: Likewise.
1990 * config/arm/cortex-r4.md: Likewise.
1991 * config/arm/exynos-m1.md: Likewise.
1992 * config/arm/fa526.md: Likewise.
1993 * config/arm/fa606te.md: Likewise.
1994 * config/arm/fa626te.md: Likewise.
1995 * config/arm/fa726te.md: Likewise.
1996 * config/arm/fmp626.md: Likewise.
1997 * config/arm/iwmmxt.md: Likewise.
1998 * config/arm/ldmstm.md: Likewise.
1999 * config/arm/marvell-pj4.md: Likewise.
2000 * config/arm/xgene1.md: Likewise.
2001 * config/aarch64/thunderx.md: Likewise.
2002 * config/aarch64/thunderx2t99.md: Likewise.
2003 * config/aarch64/falkor.md: Likewise.
2005 2017-09-12 Martin Liska <mliska@suse.cz>
2007 * attribs.c (private_lookup_attribute): New function.
2008 * attribs.h (private_lookup_attribute): Declared here.
2009 (lookup_attribute): Called from this place.
2011 2017-09-12 Richard Biener <rguenther@suse.de>
2013 PR tree-optimization/82157
2014 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
2015 stmts with side-effects.
2017 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2018 Alan Hayward <alan.hayward@arm.com>
2019 David Sherwood <david.sherwood@arm.com>
2021 * target.def (hard_regno_nregs): New hook.
2022 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
2023 * targhooks.h (default_hard_regno_nregs): Declare.
2024 * targhooks.c (default_hard_regno_nregs): New function.
2025 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
2026 (TARGET_HARD_REGNO_NREGS): ...this hook.
2027 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
2028 (CLASS_MAX_NREGS): Likewise.
2029 * doc/tm.texi: Regenerate.
2030 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
2031 instead of HARD_REGNO_NREGS.
2032 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
2033 HARD_REGNO_NREGS in the comment.
2034 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
2035 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
2036 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
2037 Return an unsigned int.
2038 (TARGET_HARD_REGNO_NREGS): Redefine.
2039 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
2040 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
2041 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
2042 (arc_hard_regno_nregs): New function.
2043 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
2044 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
2045 (arm_hard_regno_nregs): New function.
2046 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
2047 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
2048 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
2049 (TARGET_HARD_REGNO_NREGS): Redefine.
2050 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
2051 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
2052 (HARD_REGNO_NREGS): Delete.
2053 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
2054 (cr16_hard_regno_nregs): New function.
2055 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
2056 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
2057 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
2058 (cris_hard_regno_nregs): New function.
2059 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
2060 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
2061 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
2062 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
2063 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
2064 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
2065 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
2066 (frv_hard_regno_nregs): Make static. Take and return an
2068 (frv_class_max_nregs): Remove outdated copy of documentation.
2069 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
2070 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
2071 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
2072 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
2073 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
2074 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
2075 (TARGET_HARD_REGNO_NREGS): Redefine.
2076 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
2077 (CLASS_MAX_NREGS): Update comment.
2078 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
2079 (ia64_hard_regno_nregs): New function.
2080 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
2081 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
2082 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
2083 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
2084 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
2086 (m32c_hard_regno_nregs): Likewise. Make static.
2087 (TARGET_HARD_REGNO_NREGS): Redefine.
2088 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
2089 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
2090 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
2091 (m68k_hard_regno_nregs): New function.
2092 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
2093 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
2094 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
2095 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
2096 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
2097 Take and return an unsigned int.
2098 (TARGET_HARD_REGNO_NREGS): Redefine.
2099 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
2100 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
2101 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
2102 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
2103 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
2104 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
2105 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
2106 (msp430_hard_regno_nregs): Make static. Take and return an
2108 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
2109 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
2110 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
2111 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
2112 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
2113 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
2114 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
2115 (TARGET_HARD_REGNO_NREGS): Redefine.
2116 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
2117 (PA_HARD_REGNO_NREGS): ...this.
2118 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
2119 (PA_HARD_REGNO_NREGS): ...this.
2120 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
2121 (pa_hard_regno_nregs): New function.
2122 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
2123 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
2124 (pdp11_hard_regno_nregs): New function.
2125 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
2126 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
2127 (rs6000_hard_regno_nregs_hook): New function.
2128 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
2129 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
2130 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
2131 Take and return an unsigned int. Move earlier in file.
2132 (TARGET_HARD_REGNO_NREGS): Redefine.
2133 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
2134 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
2135 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
2136 (rl78_hard_regno_nregs): Make static. Take and return an
2138 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
2139 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
2140 (rs6000_hard_regno_nregs_hook): New function.
2141 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
2142 * config/rx/rx.c (rx_hard_regno_nregs): New function.
2143 (TARGET_HARD_REGNO_NREGS): Redefine.
2144 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
2145 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
2146 instead of HARD_REGNO_NREGS.
2147 (s390_hard_regno_nregs): New function.
2148 (s390_hard_regno_mode_ok): Add comment from s390.h.
2149 (TARGET_HARD_REGNO_NREGS): Redefine.
2150 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
2151 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
2152 (sh_hard_regno_nregs): New function.
2153 (sh_pass_in_reg_p): Use it.
2154 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
2155 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
2156 (sparc_hard_regno_nregs): New function.
2157 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
2158 * config/spu/spu.c (spu_hard_regno_nregs): New function.
2159 (spu_function_arg_advance): Use it, supplying a valid register number.
2160 (TARGET_HARD_REGNO_NREGS): Redefine.
2161 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
2162 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
2163 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
2164 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
2165 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
2166 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
2167 (CLASS_MAX_NREGS): Remove copy of old documentation.
2168 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
2169 (visium_hard_regno_nregs): New function.
2170 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
2171 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
2172 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
2173 xtensa_hard_regno_nregs): New function.
2174 * system.h (HARD_REGNO_NREGS): Poison.
2176 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2178 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
2179 hard_regno_nregs instead of HARD_REGNO_NREGS.
2180 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
2181 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
2182 (c6x_expand_epilogue): Likewise.
2183 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
2184 (frv_read_iacc_argument): Likewise.
2185 * config/sh/sh.c: Include regs.h.
2186 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
2187 (regs_used): Likewise.
2188 (output_stack_adjust): Likewise.
2189 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
2190 * expmed.c: Include regs.h.
2191 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
2192 * ree.c: Include regs.h.
2193 (combine_reaching_defs): Use hard_regno_nregs instead of
2195 (add_removable_extension): Likewise.
2197 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2199 * regs.h (hard_regno_nregs): Turn into a function.
2200 (end_hard_regno): Update accordingly.
2201 * caller-save.c (setup_save_areas): Likewise.
2202 (save_call_clobbered_regs): Likewise.
2203 (replace_reg_with_saved_mem): Likewise.
2204 (insert_restore): Likewise.
2205 (insert_save): Likewise.
2206 * combine.c (can_change_dest_mode): Likewise.
2207 (move_deaths): Likewise.
2208 (distribute_notes): Likewise.
2209 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
2210 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
2211 (rs6000_split_multireg_move): Likewise.
2212 (rs6000_register_move_cost): Likewise.
2213 (rs6000_memory_move_cost): Likewise.
2214 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
2215 (rs6000_split_multireg_move): Likewise.
2216 (rs6000_register_move_cost): Likewise.
2217 (rs6000_memory_move_cost): Likewise.
2218 * cselib.c (cselib_reset_table): Likewise.
2219 (cselib_lookup_1): Likewise.
2220 * emit-rtl.c (set_mode_and_regno): Likewise.
2221 * function.c (aggregate_value_p): Likewise.
2222 * ira-color.c (setup_profitable_hard_regs): Likewise.
2223 (check_hard_reg_p): Likewise.
2224 (calculate_saved_nregs): Likewise.
2225 (assign_hard_reg): Likewise.
2226 (improve_allocation): Likewise.
2227 (calculate_spill_cost): Likewise.
2228 * ira-emit.c (modify_move_list): Likewise.
2229 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
2230 (ira_hard_reg_in_set_p): Likewise.
2231 * ira.c (setup_reg_mode_hard_regset): Likewise.
2232 (clarify_prohibited_class_mode_regs): Likewise.
2233 (check_allocation): Likewise.
2234 * lra-assigns.c (find_hard_regno_for_1): Likewise.
2235 (lra_setup_reg_renumber): Likewise.
2236 (setup_try_hard_regno_pseudos): Likewise.
2237 (spill_for): Likewise.
2238 (assign_hard_regno): Likewise.
2239 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
2240 * lra-constraints.c (in_class_p): Likewise.
2241 (lra_constraint_offset): Likewise.
2242 (simplify_operand_subreg): Likewise.
2243 (lra_constraints): Likewise.
2244 (split_reg): Likewise.
2245 (split_if_necessary): Likewise.
2246 (invariant_p): Likewise.
2247 (inherit_in_ebb): Likewise.
2248 * lra-lives.c (process_bb_lives): Likewise.
2249 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
2250 (get_hard_regs): Likewise.
2251 (do_remat): Likewise.
2252 * lra-spills.c (assign_spill_hard_regs): Likewise.
2253 * mode-switching.c (create_pre_exit): Likewise.
2254 * postreload.c (reload_combine_recognize_pattern): Likewise.
2255 * recog.c (peep2_find_free_register): Likewise.
2256 * regcprop.c (kill_value_regno): Likewise.
2257 (set_value_regno): Likewise.
2258 (copy_value): Likewise.
2259 (maybe_mode_change): Likewise.
2260 (find_oldest_value_reg): Likewise.
2261 (copyprop_hardreg_forward_1): Likewise.
2262 * regrename.c (check_new_reg_p): Likewise.
2263 (regrename_do_replace): Likewise.
2264 * reload.c (push_reload): Likewise.
2265 (combine_reloads): Likewise.
2266 (find_dummy_reload): Likewise.
2267 (operands_match_p): Likewise.
2268 (find_reloads): Likewise.
2269 (find_equiv_reg): Likewise.
2270 (reload_adjust_reg_for_mode): Likewise.
2271 * reload1.c (count_pseudo): Likewise.
2272 (count_spilled_pseudo): Likewise.
2273 (find_reg): Likewise.
2274 (clear_reload_reg_in_use): Likewise.
2275 (free_for_value_p): Likewise.
2276 (allocate_reload_reg): Likewise.
2277 (choose_reload_regs): Likewise.
2278 (reload_adjust_reg_for_temp): Likewise.
2279 (emit_reload_insns): Likewise.
2280 (delete_output_reload): Likewise.
2281 * rtlanal.c (subreg_get_info): Likewise.
2282 * sched-deps.c (sched_analyze_reg): Likewise.
2283 * sel-sched.c (init_regs_for_mode): Likewise.
2284 (mark_unavailable_hard_regs): Likewise.
2285 (choose_best_reg_1): Likewise.
2286 (verify_target_availability): Likewise.
2287 * valtrack.c (dead_debug_insert_temp): Likewise.
2288 * var-tracking.c (track_loc_p): Likewise.
2289 (emit_note_insn_var_location): Likewise.
2290 * varasm.c (make_decl_rtl): Likewise.
2291 * reginfo.c (choose_hard_reg_mode): Likewise.
2292 (init_reg_modes_target): Refer directly to
2293 this_target_regs->x_hard_regno_nregs.
2295 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2297 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
2298 instead of hard_regno_nregs.
2300 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2302 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
2303 end_hard_regno instead of hard_regno_nregs.
2304 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
2305 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
2306 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
2307 * ira-color.c (improve_allocation): Likewise.
2308 * lra-assigns.c (find_hard_regno_for_1): Likewise.
2309 * lra-lives.c (mark_regno_live): Likewise.
2310 (mark_regno_dead): Likewise.
2311 * lra-remat.c (operand_to_remat): Likewise.
2312 * lra.c (collect_non_operand_hard_regs): Likewise.
2313 * postreload.c (reload_combine_note_store): Likewise.
2314 (move2add_valid_value_p): Likewise.
2315 * reload.c (regno_clobbered_p): Likewise.
2317 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2319 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
2321 * config/v850/v850.c (v850_reorg): Likewise.
2322 * reload.c (refers_to_regno_for_reload_p): Likewise.
2323 (find_equiv_reg): Likewise.
2324 * reload1.c (reload_reg_reaches_end_p): Likewise.
2326 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2328 * caller-save.c (add_used_regs): Use REG_NREGS instead of
2330 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
2331 * config/arm/arm.c (output_move_neon): Likewise.
2332 (arm_attr_length_move_neon): Likewise.
2333 (neon_split_vcombine): Likewise.
2334 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
2335 (c6x_mark_reg_written): Likewise.
2336 (c6x_dwarf_register_span): Likewise.
2337 * config/i386/i386.c (ix86_save_reg): Likewise.
2338 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
2339 (rws_access_reg): Likewise.
2340 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
2341 * mode-switching.c (create_pre_exit): Likewise.
2342 * ree.c (combine_reaching_defs): Likewise.
2343 (add_removable_extension): Likewise.
2344 * regcprop.c (find_oldest_value_reg): Likewise.
2345 (copyprop_hardreg_forward_1): Likewise.
2346 * reload.c (reload_inner_reg_of_subreg): Likewise.
2347 (push_reload): Likewise.
2348 (combine_reloads): Likewise.
2349 (find_dummy_reload): Likewise.
2350 (reload_adjust_reg_for_mode): Likewise.
2351 * reload1.c (find_reload_regs): Likewise.
2352 (forget_old_reloads_1): Likewise.
2353 (reload_reg_free_for_value_p): Likewise.
2354 (reload_adjust_reg_for_temp): Likewise.
2355 (emit_reload_insns): Likewise.
2356 (delete_output_reload): Likewise.
2357 * sel-sched.c (choose_best_reg_1): Likewise.
2358 (choose_best_pseudo_reg): Likewise.
2360 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2361 Alan Hayward <alan.hayward@arm.com>
2362 David Sherwood <david.sherwood@arm.com>
2364 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
2365 * target.def (slow_unaligned_access): New hook.
2366 * targhooks.h (default_slow_unaligned_access): Declare.
2367 * targhooks.c (default_slow_unaligned_access): New function.
2368 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
2369 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
2370 * doc/tm.texi: Regenerate.
2371 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
2372 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
2373 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
2375 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
2376 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
2378 (rs6000_slow_unaligned_access): New function.
2379 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
2380 (expand_block_compare): Likewise.
2381 (expand_strn_compare): Likewise.
2382 (rs6000_rtx_costs): Likewise.
2383 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
2384 (riscv_slow_unaligned_access): Likewise.
2385 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
2386 (riscv_slow_unaligned_access_p): ...this and make static.
2387 (riscv_option_override): Update accordingly.
2388 (riscv_slow_unaligned_access): New function.
2389 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
2390 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
2391 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
2392 (rs6000_slow_unaligned_access): New function.
2393 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
2394 (rs6000_rtx_costs): Likewise.
2395 * config/rs6000/rs6000-string.c (expand_block_compare)
2396 (expand_strn_compare): Use targetm.slow_unaligned_access instead
2397 of SLOW_UNALIGNED_ACCESS.
2398 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
2399 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
2400 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
2401 of SLOW_UNALIGNED_ACCESS.
2402 * expmed.c (simple_mem_bitfield_p): Likewise.
2403 * expr.c (alignment_for_piecewise_move): Likewise.
2404 (emit_group_load_1): Likewise.
2405 (emit_group_store): Likewise.
2406 (copy_blkmode_from_reg): Likewise.
2407 (emit_push_insn): Likewise.
2408 (expand_assignment): Likewise.
2409 (store_field): Likewise.
2410 (expand_expr_real_1): Likewise.
2411 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
2412 * lra-constraints.c (simplify_operand_subreg): Likewise.
2413 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
2414 * gimple-ssa-store-merging.c: Likewise in block comment at start
2416 * tree-ssa-strlen.c: Include target.h.
2417 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
2418 of SLOW_UNALIGNED_ACCESS.
2419 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
2421 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2423 PR rtl-optimization/82185
2424 * expmed.c (emit_store_flag_int): Only test tem if it has been
2427 2017-09-12 Richard Biener <rguenther@suse.de>
2430 * match.pd ((FTYPE) N CMP CST): Fix typo.
2432 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
2434 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
2436 (mips_near_type_p): Add 'short_call' attribute as a synonym
2438 * doc/extend.texi (short_call): Document new function attribute.
2440 2017-09-12 Jakub Jelinek <jakub@redhat.com>
2443 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
2444 assertion check that in the condition.
2445 (get_atomic_generic_size): Likewise. Before testing if parameter
2446 has pointer type, if it has array type, call for C++
2447 default_conversion to perform array-to-pointer conversion.
2449 2017-09-12 Richard Biener <rguenther@suse.de>
2451 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
2452 for operations we cannot scalarize.
2454 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
2456 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
2457 vectors heap vectors. Clean up comments.
2458 Make visited_bbs a reference.
2459 (profitable_jump_thread_path): Make GC
2460 vectors heap vectors. Clean up comments.
2462 (convert_and_register_jump_thread_path): Make GC vectors heap
2464 (check_subpath_and_update_thread_path): Same. Clean up comments.
2465 Make visited_bbs a reference.
2466 (handle_phi): Abstract common code to to
2467 register_jump_thread_path_if_profitable.
2468 Rename VAR_BB to DEF_BB.
2470 Make GC vectors heap vectors.
2471 Make visited_bbs a reference.
2472 (handle_assignment): Same.
2473 (register_jump_thread_path_if_profitable): New.
2474 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
2476 Make GC vectors heap vectors. Clean up comments.
2477 Make visited_bbs a reference.
2478 (find_jump_threads_backwards): Make visited_bbs live in the stack.
2479 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
2482 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
2485 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
2486 words of E_DImode object are reachable by xtensa_uimm8x4 access.
2488 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
2490 Revert r251800 and r251799.
2492 2017-09-11 Martin Jambor <mjambor@suse.cz>
2495 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
2496 arguments in advance.
2497 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
2498 use it to find predecessor edges.
2499 (naive_outof_ssa): Collect vector of predecessors.
2501 2017-09-08 Jason Merrill <jason@redhat.com>
2503 PR c++/70029 - ICE with ref-qualifier and -flto
2504 * langhooks.h (struct lang_hooks_for_types): Add
2505 copy_lang_qualifiers.
2506 * attribs.c (build_type_attribute_qual_variant): Use it.
2507 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
2509 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
2510 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
2512 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
2515 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
2516 (*mulsi3_sp64): New instruction.
2517 (mulsi3): New expander.
2519 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
2521 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
2523 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2525 * sancov.c: Include memmodel.h.
2527 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
2530 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
2533 2017-09-07 Carl Love <cel@us.ibm.com>
2535 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
2536 the sldi instruction.
2538 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
2540 * sancov.c: Include tm_p.h.
2542 2017-09-07 Jakub Jelinek <jakub@redhat.com>
2545 * output.h (switch_to_other_text_partition): New declaration.
2546 * varasm.c (switch_to_other_text_partition): New function.
2547 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
2548 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
2549 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
2550 to the other text partition before emitting LCL label and switch back
2551 after emitting the word after it.
2553 2017-09-07 Richard Biener <rguenther@suse.de>
2555 * passes.def (pass_split_crit_edges): Remove instance before PRE.
2556 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
2557 critical edges here, after loop init.
2558 (pass_data_pre): Remove PROP_no_crit_edges flags.
2559 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
2560 for valueization of call args to avoid leaking VN_TOP.
2561 (visit_use): Assert we do not visit default defs.
2562 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
2563 Use error_mark_node to more easily detect leaking VN_TOP.
2564 All default-defs are varying, not VN_TOP. Mark them visited.
2565 (run_scc_vn): Make code match comment.
2567 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
2569 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
2570 OPTION_MASK_FLOAT128_KEYWORD.
2571 (POWERPC_MASKS): Likewise.
2572 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
2573 support for the -mfloat128-type option, and make -mfloat128
2574 default on PowerPC Linux systems. Define or undefine
2575 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
2576 Define __float128 to be __ieee128 if IEEE 128-bit support is
2577 enabled, or undefine it.
2578 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
2579 Delete defining __FLOAT128_TYPE__.
2580 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
2581 -mfloat128-type option and make -mfloat128 default on PowerPC
2583 (TARGET_FLOAT128_TYPE): Likewise.
2584 (-mfloat128-type): Likewise.
2585 * config/rs6000/rs6000.c (rs6000_option_override_internal):
2586 Delete the -mfloat128-type option and make -mfloat128 default on
2587 PowerPC Linux systems. Always use __ieee128 to be the keyword for
2588 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
2589 128-bit floating point is enabled. Change tests from using
2590 -mfloat128-type to -mfloat128.
2591 (rs6000_mangle_type): Use the correct mangling for the __float128
2592 type even if normal long double is restricted to 64-bits.
2593 (floatn_mode): Enable the _Float128 type by default on VSX Linux
2595 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
2596 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
2597 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
2599 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
2600 documentation for -mfloat128.
2602 2017-09-06 Olivier Hainque <hainque@adacore.com>
2604 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
2606 2017-09-06 Wish Wu <wishwu007@gmail.com>
2607 Jakub Jelinek <jakub@redhat.com>
2609 * asan.c (initialize_sanitizer_builtins): Add
2610 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
2611 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
2612 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
2613 BT_FN_VOID_UINT64_PTR variables.
2614 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
2615 (BT_FN_VOID_UINT16_UINT16): Likewise.
2616 (BT_FN_VOID_UINT32_UINT32): Likewise.
2617 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
2618 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
2619 (BT_FN_VOID_UINT64_PTR): Likewise.
2620 * common.opt (flag_sanitize_coverage): New variable.
2621 (fsanitize-coverage=trace-pc): Remove.
2622 (fsanitize-coverage=): Add.
2623 * flag-types.h (enum sanitize_coverage_code): New enum.
2624 * fold-const.c (fold_range_test): Disable non-short-circuit
2625 optimization if flag_sanitize_coverage.
2626 (fold_truth_andor): Likewise.
2627 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
2628 * opts.c (COVERAGE_SANITIZER_OPT): Define.
2629 (coverage_sanitizer_opts): New array.
2630 (get_closest_sanitizer_option): Add OPTS argument, handle also
2631 OPT_fsanitize_coverage_.
2632 (parse_sanitizer_options): Adjusted to also handle
2633 OPT_fsanitize_coverage_.
2634 (common_handle_option): Add OPT_fsanitize_coverage_.
2635 * sancov.c (instrument_comparison, instrument_switch): New function.
2636 (sancov_pass): Add trace-cmp support.
2637 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
2638 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
2639 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
2640 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
2641 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
2642 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
2643 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
2644 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
2645 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
2646 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
2648 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
2650 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
2651 error. Only quit immediately if parsing is complete.
2652 (BEGIN): Initialize fatal_err and parse_done.
2653 (begin fpu, end fpu): Check number of arguments.
2654 (begin arch, end arch): Likewise.
2655 (begin cpu, end cpu): Likewise.
2656 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
2657 (optalias): Likewise.
2659 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
2661 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
2662 * config/arm/arm-isa.h: Delete. Move definitions to ...
2663 * arm-cpus.in: ... here. Use new feature and fgroup values.
2664 * config/arm/arm.c (arm_option_override): Use lower case for feature
2666 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
2667 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
2668 * config/arm/parsecpu.awk (END): Add new command 'isa'.
2670 (print_isa_bits_for): New function.
2671 (gen_isa): New function.
2672 (gen_comm_data): Use print_isa_bits_for.
2673 (define feature): New keyword.
2674 (define fgroup): New keyword.
2675 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
2676 (arm-isa.h): Add rule to generate file.
2677 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
2678 case for feature bit names.
2680 2017-09-06 Richard Biener <rguenther@suse.de>
2682 * tree-ssa-pre.c (NECESSARY): Remove.
2683 (create_expression_by_pieces): Do not touch pass-local flags.
2684 (insert_into_preds_of_block): Likewise.
2685 (do_pre_regular_insertion): Likewise.
2686 (eliminate_insert): Likewise.
2687 (eliminate_dom_walker::before_dom_children): Likewise.
2688 (fini_eliminate): Do not look at inserted_exprs.
2689 (mark_operand_necessary): Remove.
2690 (remove_dead_inserted_code): Replace with simple work-list
2691 algorithm based on inserted_exprs and SSA uses.
2692 (pass_pre::execute): Re-order fini_eliminate and
2693 remove_dead_inserted_code.
2695 2017-09-06 Olivier Hainque <hainque@adacore.com>
2697 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
2698 for VxWorks 7. Adjust surrounding comments.
2700 2017-09-06 Richard Biener <rguenther@suse.de>
2702 * gimple-ssa-strength-reduction.c
2703 (find_candidates_dom_walker::before_dom_children): Also allow
2706 2017-09-06 Richard Biener <rguenther@suse.de>
2708 PR tree-optimization/82108
2709 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
2710 for gap in the non-permutation SLP case.
2712 2017-09-06 Martin Jambor <mjambor@suse.cz>
2714 PR tree-optimization/82078
2715 * tree-sra.c (sort_and_splice_var_accesses): Move call to
2716 add_access_to_work_queue...
2717 (build_accesses_from_assign): ...here.
2718 (propagate_all_subaccesses): Make sure racc is the group
2719 representative, if there is one.
2721 2017-09-06 Jakub Jelinek <jakub@redhat.com>
2724 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
2727 2017-09-06 Richard Biener <rguenther@suse.de>
2729 * gimple-ssa-strength-reduction.c
2730 (find_candidates_dom_walker::before_doom_children): Use a
2731 type and not a mode check.
2733 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
2736 * config/arm/predicates.md (arm_general_adddi_operand): Create new
2738 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
2740 2017-09-05 Jeff Law <law@redhat.com>
2742 PR tree-optimization/64910
2743 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
2744 cases where we have 3 or more operands.
2746 2017-09-05 Jakub Jelinek <jakub@redhat.com>
2749 * omp-low.c (lower_omp_for): Recompute tree invariant if
2750 gimple_omp_for_initial/final is ADDR_EXPR.
2753 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
2754 into gimple val before gimplification fo the COND_EXPR.
2756 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
2758 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
2759 REGION_COPY argument.
2760 (thread_through_all_blocks): Remove unused argument to
2761 duplicate_thread_path.
2763 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2764 Alan Hayward <alan.hayward@arm.com>
2765 David Sherwood <david.sherwood@arm.com>
2767 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
2768 Take a scalar_mode rather than a machine_mode.
2769 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
2770 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
2771 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
2772 (aarch64_gen_adjusted_ldpstp): Likewise.
2773 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
2775 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2776 Alan Hayward <alan.hayward@arm.com>
2777 David Sherwood <david.sherwood@arm.com>
2779 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
2780 Take a scalar_int_mode instead of a machine_mode.
2781 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
2782 (aarch64_output_scalar_simd_mov_immediate): Likewise.
2783 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
2784 (aarch64_simd_attr_length_rglist): Delete.
2785 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
2786 a scalar_int_mode instead of a machine_mode.
2787 (aarch64_add_offset): Likewise.
2788 (aarch64_internal_mov_immediate): Likewise
2789 (aarch64_add_constant_internal): Likewise.
2790 (aarch64_add_constant): Likewise.
2791 (aarch64_movw_imm): Likewise.
2792 (aarch64_rtx_arith_op_extract_p): Likewise.
2793 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
2794 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
2795 Remove assert that the mode isn't a vector.
2796 (aarch64_output_scalar_simd_mov_immediate): Likewise.
2797 (aarch64_expand_mov_immediate): Update calls after above changes.
2798 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
2799 (aarch64_and_bitmask_imm): Check for scalar integer modes.
2800 (aarch64_move_imm): Likewise.
2801 (aarch64_can_const_movi_rtx_p): Likewise.
2802 (aarch64_strip_extend): Likewise.
2803 (aarch64_extr_rtx_p): Likewise.
2804 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
2805 a CONST_INT when the mode parameter is VOIDmode.
2806 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
2808 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2810 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
2811 * stor-layout.c (bitwise_mode_for_mode): Likewise.
2812 (bitwise_type_for_mode): Update accordingly.
2814 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2816 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
2817 * stor-layout.c (mode_for_size_tree): Likewise.
2818 (mode_for_array): Update accordingly.
2819 (layout_decl): Likewise.
2820 (compute_record_mode): Likewise. Only set the mode once.
2822 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2824 * target.def (get_mask_mode): Change return type to opt_mode.
2826 * doc/tm.texi: Regenerate.
2827 * targhooks.h (default_get_mask_mode): Return an opt_mode.
2828 * targhooks.c (default_get_mask_mode): Likewise.
2829 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
2830 * optabs-query.c (can_vec_mask_load_store_p): Update use of
2831 targetm.get_mask_mode.
2832 * tree.c (build_truth_vector_type): Likewise.
2834 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2836 * machmode.h (mode_for_vector): Return an opt_mode.
2837 * stor-layout.c (mode_for_vector): Likewise.
2838 (mode_for_int_vector): Update accordingly.
2839 (layout_type): Likewise.
2840 * config/i386/i386.c (emit_memmov): Likewise.
2841 (ix86_expand_set_or_movmem): Likewise.
2842 (ix86_expand_vector_init): Likewise.
2843 (ix86_get_mask_mode): Likewise.
2844 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
2846 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
2847 * expmed.c (extract_bit_field_1): Likewise.
2848 * expr.c (expand_expr_real_2): Likewise.
2849 * optabs-query.c (can_vec_perm_p): Likewise.
2850 (can_vec_mask_load_store_p): Likewise.
2851 * optabs.c (expand_vec_perm): Likewise.
2852 * targhooks.c (default_get_mask_mode): Likewise.
2853 * tree-vect-stmts.c (vectorizable_store): Likewise.
2854 (vectorizable_load): Likewise.
2855 (get_vectype_for_scalar_type_and_size): Likewise.
2857 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2859 * machmode.h (mode_for_int_vector): New function.
2860 * stor-layout.c (mode_for_int_vector): Likewise.
2861 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
2862 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
2863 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
2864 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
2865 (s390_expand_vcond): Likewise.
2867 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2869 * machmode.h (opt_machine_mode): New type.
2870 (opt_mode<T>): Allow construction from anything that can be
2872 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
2873 (mode_for_size): Return an opt_machine_mode.
2874 * stor-layout.c (mode_for_size): Likewise.
2875 (mode_for_size_tree): Update call accordingly.
2876 (bitwise_mode_for_mode): Likewise.
2877 (make_fract_type): Likewise.
2878 (make_accum_type): Likewise.
2879 * caller-save.c (replace_reg_with_saved_mem): Update call
2881 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
2882 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
2883 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
2884 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
2885 * expmed.c (extract_bit_field_1): Likewise.
2886 * reload.c (get_secondary_mem): Likewise.
2887 * varasm.c (assemble_integer): Likewise.
2888 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
2891 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2893 * machmode.h (decimal_float_mode_for_size): New function.
2894 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
2895 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
2896 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
2897 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
2898 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
2899 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
2901 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2903 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
2904 (get_builtin_sync_mode): Likewise.
2905 (expand_ifn_atomic_compare_exchange): Likewise.
2906 (expand_builtin_atomic_clear): Likewise.
2907 (expand_builtin_atomic_test_and_set): Likewise.
2908 (fold_builtin_atomic_always_lock_free): Likewise.
2909 * calls.c (compute_argument_addresses): Likewise.
2910 (emit_library_call_value_1): Likewise.
2911 (store_one_arg): Likewise.
2912 * combine.c (combine_instructions): Likewise.
2913 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
2914 * config/arm/arm.c (arm_function_value): Likewise.
2915 (aapcs_allocate_return_reg): Likewise.
2916 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
2917 * config/i386/i386.c (construct_container): Likewise.
2918 (ix86_gimplify_va_arg): Likewise.
2919 (ix86_expand_sse_cmp): Likewise.
2920 (emit_memmov): Likewise.
2921 (emit_memset): Likewise.
2922 (expand_small_movmem_or_setmem): Likewise.
2923 (ix86_expand_pextr): Likewise.
2924 (ix86_expand_pinsr): Likewise.
2925 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
2926 * config/microblaze/microblaze.c (microblaze_block_move_straight):
2928 * config/mips/mips.c (mips_function_value_1) Likewise.
2929 (mips_block_move_straight): Likewise.
2930 (mips_expand_ins_as_unaligned_store): Likewise.
2931 * config/powerpcspe/powerpcspe.c
2932 (rs6000_darwin64_record_arg_advance_flush): Likewise.
2933 (rs6000_darwin64_record_arg_flush): Likewise.
2934 * config/rs6000/rs6000.c
2935 (rs6000_darwin64_record_arg_advance_flush): Likewise.
2936 (rs6000_darwin64_record_arg_flush): Likewise.
2937 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
2938 (sparc_function_value_1): Likewise.
2939 * config/spu/spu.c (adjust_operand): Likewise.
2940 (spu_emit_branch_or_set): Likewise.
2941 (arith_immediate_p): Likewise.
2942 * emit-rtl.c (gen_lowpart_common): Likewise.
2943 * expr.c (expand_expr_real_1): Likewise.
2944 * function.c (assign_parm_setup_block): Likewise.
2945 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
2946 * reload1.c (alter_reg): Likewise.
2947 * stor-layout.c (mode_for_vector): Likewise.
2948 (layout_type): Likewise.
2950 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2952 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
2953 (spu_convert_move): Likewise.
2954 * lower-subreg.c (resolve_simple_move): Likewise.
2956 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2959 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
2960 define_insn to a define_expand.
2961 (altivec_vsum2sws_direct): New define_insn.
2962 (altivec_vsumsws): Convert from a define_insn to a define_expand.
2964 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
2966 * config/arm/arm.c (arm_option_params_internal): Improve setting of
2969 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
2974 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
2975 realign stack if stack alignment needed is less than incoming
2978 2017-09-05 Marek Polacek <polacek@redhat.com>
2981 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
2984 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
2986 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
2988 2017-09-05 Richard Biener <rguenther@suse.de>
2990 PR tree-optimization/82084
2991 * fold-const.c (can_native_encode_string_p): Handle wide characters.
2993 2017-09-05 Richard Biener <rguenther@suse.de>
2995 PR tree-optimization/82102
2996 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
2998 2017-09-05 Martin Liska <mliska@suse.cz>
3000 PR tree-optimization/82032
3001 * tree-cfg.c (generate_range_test): New function.
3002 * tree-cfg.h (generate_range_test): Declared here.
3003 * tree-cfgcleanup.c (convert_single_case_switch): New function.
3004 (cleanup_control_expr_graph): Use it.
3005 * tree-switch-conversion.c (try_switch_expansion): Remove
3007 (emit_case_nodes): Use generate_range_test.
3009 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
3012 * config/i386/i386.md (*<btsc><mode>_mask): Add
3013 TARGET_USE_BT to insn constraint.
3014 (*btr<mode>_mask): Ditto.
3016 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
3018 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
3019 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
3021 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
3024 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
3025 TARGET_NEON and TARGET_IWMMXT.
3026 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
3027 TARGET_NEON and TARGET_IWMMXT.
3028 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
3030 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
3032 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
3033 (ix86_rewrite_tls_address): Ditto.
3034 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
3035 (ix86_rewrite_tls_address_1): Ditto.
3036 (ix86_rewrite_tls_address): Ditto.
3037 * config/i386/predicates.md (tls_address_pattern): New predicate.
3038 * config/i386/i386.md (TLS address splitter): New splitter.
3040 2017-09-04 Richard Biener <rguenther@suse.de>
3042 PR tree-optimization/82084
3043 * fold-const.h (can_native_encode_string_p): Declare.
3044 * fold-const.c (can_native_encode_string_p): Factor out from ...
3045 (native_encode_string): ... here.
3046 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
3047 vectorizing stores from constants we later cannot handle.
3049 2017-09-04 Marek Polacek <polacek@redhat.com>
3052 * doc/invoke.texi: Update -Wtautological-compare documentation.
3054 2017-09-04 Jeff Law <law@redhat.com>
3056 PR tree-optimization/64910
3057 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
3058 swap the first and last operand if the last is a constant.
3060 2017-09-04 Marek Polacek <polacek@redhat.com>
3063 * convert.c (do_narrow): When sanitizing signed integer overflows,
3064 bail out for signed types.
3065 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
3067 2017-09-04 Richard Biener <rguenther@suse.de>
3069 PR tree-optimization/82060
3070 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
3071 Move devirtualization after stmt folding and before EH/AB/noreturn
3072 cleanup to get the stmt refs canonicalized. Use a bool instead
3073 of gimple_modified_p since that doesn't work for NOPs. Schedule
3074 NOPs generated by folding for removal.
3076 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
3077 Alan Hayward <alan.hayward@arm.com>
3078 David Sherwood <david.sherwood@arm.com>
3080 * coretypes.h (pad_direction): New enum.
3081 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
3082 (FUNCTION_ARG_PADDING): Likewise.
3083 * target.def (function_arg_padding): New hook.
3084 * targhooks.h (default_function_arg_padding): Declare.
3085 * targhooks.c (default_function_arg_padding): New function.
3086 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
3087 (TARGET_FUNCTION_ARG_PADDING): ...this.
3088 * doc/tm.texi: Regenerate.
3089 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
3090 instead of direction.
3091 (compute_argument_addresses): Likewise.
3092 (load_register_parameters): Likewise.
3093 (emit_library_call_value_1): Likewise.
3094 (store_one_arg): Use targetm.calls.function_arg_padding instead
3095 of FUNCTION_ARG_PADDING.
3096 (must_pass_in_stack_var_size_or_pad): Likewise.
3097 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
3098 (emit_group_store): Likewise.
3099 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
3100 instead of FUNCTION_ARG_PADDING.
3101 (emit_push_insn): Likewise, and propagate enum change throughout
3103 * function.h (direction): Delete.
3104 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
3106 * function.c (assign_parm_find_stack_rtl): Likewise.
3107 (assign_parm_setup_block_p): Likewise.
3108 (assign_parm_setup_block): Likewise.
3109 (gimplify_parameters): Likewise.
3110 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
3111 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
3113 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
3114 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
3115 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
3116 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
3117 (aarch64_function_arg_padding): ...this new function.
3118 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
3119 (TARGET_FUNCTION_ARG_PADDING): Redefine.
3120 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
3121 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
3122 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
3123 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3124 (arm_pad_arg_upward): Replace with...
3125 (arm_function_arg_padding): ...this new function.
3126 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
3128 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
3129 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
3130 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3131 (ia64_hpux_function_arg_padding): Replace with...
3132 (ia64_function_arg_padding): ...this new function. Use pad_direction
3133 instead of direction. Check for TARGET_HPUX.
3134 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
3135 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3136 (iq2000_function_arg_padding): New function.
3137 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
3138 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
3139 (mips_function_arg_padding): ...this new function.
3140 (mips_pad_reg_upward): Update accordingly.
3141 (TARGET_FUNCTION_ARG_PADDING): Redefine.
3142 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
3143 targetm.calls.function_arg_padding.
3144 (FUNCTION_ARG_PADDING): Delete.
3145 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
3146 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
3147 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
3148 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
3149 (nios2_block_reg_padding): Return pad_direction instead of direction.
3150 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
3151 instead of direction.
3152 (nios2_function_arg_padding): Likewise. Make static.
3153 (TARGET_FUNCTION_ARG_PADDING): Redefine.
3154 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
3155 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
3156 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
3157 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3158 (pa_function_arg_padding): Make static. Return pad_direction instead
3160 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
3161 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
3162 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
3163 instead of direction. Use targetm.calls.function_arg_padding.
3164 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
3165 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
3166 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
3167 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
3168 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
3170 (function_arg_padding): Rename to...
3171 (rs6000_function_arg_padding): ...this. Make static. Return
3172 pad_direction instead of direction.
3173 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
3174 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
3175 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
3176 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
3177 instead of direction. Use targetm.calls.function_arg_padding.
3178 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
3179 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
3180 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
3181 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
3182 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3183 (function_arg_padding): Rename to...
3184 (rs6000_function_arg_padding): ...this. Make static. Return
3185 pad_direction instead of direction.
3186 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
3187 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
3188 * config/s390/s390.c (s390_function_arg_padding): New function.
3189 (TARGET_FUNCTION_ARG_PADDING): Redefine.
3190 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
3191 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
3192 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3193 (function_arg_padding): Rename to...
3194 (sparc_function_arg_padding): ...this. Make static. Return
3195 pad_direction instead of direction.
3196 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
3197 * config/spu/spu.c (spu_function_arg_padding): New function.
3198 (TARGET_FUNCTION_ARG_PADDING): Redefine.
3199 * system.h (FUNCTION_ARG_PADDING): Poison.
3201 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
3202 Alan Hayward <alan.hayward@arm.com>
3203 David Sherwood <david.sherwood@arm.com>
3205 * target.def (modes_tieable_p): New hook.
3206 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
3207 (TARGET_MODES_TIEABLE_P): ...this.
3208 * doc/tm.texi.in: Regenerate.
3209 * hooks.h (hook_bool_mode_mode_true): Declare.
3210 * hooks.c (hook_bool_mode_mode_true): New function.
3211 * combine.c (subst): Use targetm.modes_tieable_p instead of
3213 * dse.c (find_shift_sequence): Likewise.
3214 * expmed.c (extract_low_bits): Likewise.
3215 * lower-subreg.c: Include target.h.
3216 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
3218 * rtlanal.c (rtx_cost): Likewise.
3219 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
3220 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
3221 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
3222 (TARGET_MODES_TIEABLE_P): Redefine.
3223 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
3224 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
3225 (TARGET_MODES_TIEABLE_P): Redefine.
3226 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
3227 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
3228 (arc_modes_tieable_p): New function.
3229 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
3230 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
3231 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
3232 (arm_modes_tieable_p): Make static.
3233 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
3234 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
3235 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
3236 (TARGET_MODES_TIEABLE_P): Redefine.
3237 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
3238 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
3239 (TARGET_MODES_TIEABLE_P): Redefine.
3240 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
3241 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
3242 (cr16_modes_tieable_p): New function.
3243 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
3244 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
3245 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
3246 (TRULY_NOOP_TRUNCATION): Update comment.
3247 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
3248 (TRULY_NOOP_TRUNCATION): Update comment.
3249 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
3250 (frv_modes_tieable_p): New function.
3251 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
3252 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
3253 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
3254 (TARGET_MODES_TIEABLE_P): Redefine.
3255 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
3256 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
3257 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
3258 (TARGET_MODES_TIEABLE_P): Redefine.
3259 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
3260 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
3261 (ia64_modes_tieable_p): New function.
3262 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
3263 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
3264 (iq2000_modes_tieable_p): New function.
3265 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
3266 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
3267 (lm32_modes_tieable_p): New function.
3268 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
3269 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
3270 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
3271 (TARGET_MODES_TIEABLE_P): Redefine.
3272 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
3273 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
3274 (m32r_modes_tieable_p): New function.
3275 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
3276 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
3277 (m68k_modes_tieable_p): New function.
3278 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
3279 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
3280 (mcore_modes_tieable_p): New function.
3281 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
3282 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
3284 (TARGET_MODES_TIEABLE_P): Redefine.
3285 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
3286 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
3287 * config/mips/mips.c (mips_modes_tieable_p): Make static.
3288 (TARGET_MODES_TIEABLE_P): Redefine.
3289 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
3290 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
3291 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
3292 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
3293 (mn10300_modes_tieable_p): ...this and make static.
3294 (TARGET_MODES_TIEABLE_P): Redefine.
3295 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
3296 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
3297 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
3298 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
3299 (msp430_modes_tieable_p): Make static.
3300 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
3301 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
3302 (TARGET_MODES_TIEABLE_P): Redefine.
3303 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
3304 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
3305 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
3306 (TARGET_MODES_TIEABLE_P): Redefine.
3307 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
3308 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
3309 * config/pa/pa.c (pa_modes_tieable_p): Make static.
3310 (TARGET_MODES_TIEABLE_P): Redefine.
3311 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
3312 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
3313 (pdp11_modes_tieable_p): New function.
3314 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
3315 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
3316 (rs6000_modes_tieable_p): New function.
3317 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
3318 * config/powerpcspe/powerpcspe.md: Update comment.
3319 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
3320 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
3321 (TARGET_MODES_TIEABLE_P): Redefine.
3322 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
3323 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
3324 (rl78_modes_tieable_p): New function.
3325 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
3326 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
3327 (rs6000_modes_tieable_p): New function.
3328 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
3329 * config/rs6000/rs6000.md: Update comment.
3330 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
3331 * config/rx/rx.c (rx_modes_tieable_p): New function.
3332 (TARGET_MODES_TIEABLE_P): Redefine.
3333 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
3334 * config/s390/s390.c (s390_modes_tieable_p): New function.
3335 (TARGET_MODES_TIEABLE_P): Redefine.
3336 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
3337 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
3338 (sh_modes_tieable_p): New function.
3339 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
3340 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
3341 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
3342 (sparc_modes_tieable_p): Make static.
3343 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
3344 * config/spu/spu.c (spu_modes_tieable_p): New function.
3345 (TARGET_MODES_TIEABLE_P): Redefine.
3346 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
3347 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
3348 (TARGET_MODES_TIEABLE_P): Redefine.
3349 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
3350 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
3351 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
3352 * config/v850/v850.c (v850_modes_tieable_p): New function.
3353 (TARGET_MODES_TIEABLE_P): Redefine.
3354 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
3355 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
3356 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
3357 (visium_modes_tieable_p): New function.
3358 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
3359 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
3360 (xtensa_modes_tieable_p): New function.
3361 * system.h (MODES_TIEABLE_P): Poison.
3363 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
3364 Alan Hayward <alan.hayward@arm.com>
3365 David Sherwood <david.sherwood@arm.com>
3367 * target.def (hard_regno_mode_ok): New hook.
3368 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
3369 (TARGET_HARD_REGNO_MODE_OK): ...this.
3370 * doc/tm.texi.in: Regenerate.
3371 * hooks.h (hook_bool_uint_mode_true): Declare.
3372 * hooks.c (hook_bool_uint_mode_true): New function.
3373 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
3375 * genpreds.c (write_insn_preds_c): Add an include of target.h.
3376 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
3377 instead of HARD_REGNO_MODE_OK.
3378 * caller-save.c: Include target.h.
3379 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
3381 * combine.c (can_combine_p): Likewise.
3382 (combinable_i3pat): Likewise.
3383 (can_change_dest_mode): Likewise.
3384 * expr.c (init_expr_target): Likewise.
3385 (convert_move): Likewise.
3386 (convert_modes): Likewise.
3387 * ira.c (setup_prohibited_class_mode_regs): Likewise.
3388 (setup_prohibited_mode_move_regs): Likewise.
3389 * ira.h (target_ira): Likewise.
3390 * lra-assigns.c (find_hard_regno_for_1): Likewise.
3391 * lra-constraints.c (process_alt_operands): Likewise.
3392 (split_reg): Likewise.
3393 * recog.c (peep2_find_free_register): Likewise.
3394 * ree.c (combine_reaching_defs): Likewise.
3395 * regcprop.c (maybe_mode_change): Likewise.
3396 * reginfo.c (init_reg_sets_1): Likewise.
3397 (choose_hard_reg_mode): Likewise.
3398 (simplifiable_subregs): Likewise.
3399 * regrename.c (check_new_reg_p): Likewise.
3400 * reload.c (find_valid_class): Likewise.
3401 (find_valid_class_1): Likewise.
3402 (reload_inner_reg_of_subreg): Likewise.
3403 (push_reload): Likewise.
3404 (combine_reloads): Likewise.
3405 (find_dummy_reload): Likewise.
3406 (find_reloads): Likewise.
3407 * reload1.c (find_reg): Likewise.
3408 (set_reload_reg): Likewise.
3409 (allocate_reload_reg): Likewise.
3410 (choose_reload_regs): Likewise.
3411 (reload_adjust_reg_for_temp): Likewise.
3412 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
3413 (simplify_subreg_regno): Likewise.
3414 * sel-sched.c (init_regs_for_mode): Likewise.
3415 * varasm.c (make_decl_rtl): Likewise.
3416 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
3417 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
3419 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
3420 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
3421 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3422 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
3423 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
3424 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3425 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
3426 (arc_mode_class): Delete.
3427 (HARD_REGNO_MODE_OK): Delete.
3428 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3429 (arc_hard_regno_mode_ok): Rename old array to...
3430 (arc_hard_regno_mode_ok_modes): ...this.
3431 (arc_conditional_register_usage): Update accordingly.
3432 (arc_mode_class): Make static.
3433 (arc_hard_regno_mode_ok): New function.
3434 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
3435 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
3436 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3437 (arm_hard_regno_mode_ok): Make static.
3438 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
3440 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
3441 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
3442 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
3444 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3445 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
3446 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
3447 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
3448 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
3449 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3450 * config/bfin/predicates.md (valid_reg_operand): Use
3451 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
3452 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
3453 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
3454 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3455 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
3456 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
3457 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3458 (cr16_hard_regno_mode_ok): Make static and return a bool.
3459 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
3460 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3461 (cris_hard_regno_mode_ok): New function.
3462 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
3463 (epiphany_mode_class): Delete.
3464 (HARD_REGNO_MODE_OK): Delete.
3465 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
3466 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3467 (hard_regno_mode_ok): Rename to...
3468 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
3469 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
3470 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
3472 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
3473 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
3474 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3475 (frv_hard_regno_mode_ok): Make static and return a bool.
3476 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
3478 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
3479 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
3480 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
3481 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
3483 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3484 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
3485 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
3486 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
3488 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3489 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
3490 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3491 (ia64_hard_regno_mode_ok): New function.
3492 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
3493 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3494 (iq2000_hard_regno_mode_ok): New function.
3495 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
3496 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3497 (lm32_hard_regno_mode_ok): New function.
3498 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
3499 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
3500 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
3501 instead of HARD_REGNO_MODE_OK.
3502 (m32c_hard_regno_ok): Rename to...
3503 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
3504 (m32c_cannot_change_mode_class): Update accordingly.
3505 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3506 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
3507 (m32r_mode_class): Delete.
3508 (HARD_REGNO_MODE_OK): Delete.
3509 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3510 (m32r_hard_regno_mode_ok): Rename to...
3511 (m32r_hard_regno_modes): ...this.
3512 (m32r_mode_class): Make static.
3513 (m32r_hard_regno_mode_ok): New function.
3514 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
3515 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
3516 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3517 (m68k_hard_regno_mode_ok): Make static.
3518 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
3519 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3520 (mcore_hard_regno_mode_ok): New function.
3521 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
3522 (HARD_REGNO_MODE_OK): Delete.
3523 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
3525 (microblaze_hard_regno_mode_ok_p): ...this and make static.
3526 (microblaze_hard_regno_mode_ok): New function.
3527 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3528 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
3529 (mips_hard_regno_mode_ok): Delete.
3530 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
3531 (mips_hard_regno_mode_ok_p): ...this and make static.
3532 (mips_hard_regno_mode_ok_p): Rename to...
3533 (mips_hard_regno_mode_ok_uncached): ...this.
3534 (mips_hard_regno_mode_ok): New function.
3535 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
3536 of HARD_REGNO_MODE_OK.
3537 (mips_option_override): Update after above name changes.
3538 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3539 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
3540 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
3541 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
3542 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
3543 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3544 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
3545 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
3546 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
3547 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3548 (msp430_hard_regno_mode_ok): Make static and return a bool.
3549 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
3550 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
3551 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
3553 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3554 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
3555 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
3556 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
3557 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
3558 (PA_HARD_REGNO_MODE_OK): ...this
3559 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
3560 (PA_HARD_REGNO_MODE_OK): ...this.
3561 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3562 (pa_hard_regno_mode_ok): New function.
3563 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
3564 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3565 (pdp11_hard_regno_mode_ok): New function.
3566 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
3567 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
3569 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
3571 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3572 (rs6000_hard_regno_mode_ok): Rename to...
3573 (rs6000_hard_regno_mode_ok_uncached): ...this.
3574 (rs6000_init_hard_regno_mode_ok): Update accordingly.
3575 (rs6000_hard_regno_mode_ok): New function.
3576 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
3577 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
3578 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
3579 (riscv_hard_regno_mode_ok): ...this and make static.
3580 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3581 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
3582 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
3583 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3584 (rl78_hard_regno_mode_ok): Make static and return bool.
3585 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
3586 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
3588 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
3589 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3590 (rs6000_hard_regno_mode_ok): Rename to...
3591 (rs6000_hard_regno_mode_ok_uncached): ...this.
3592 (rs6000_init_hard_regno_mode_ok): Update accordingly.
3593 (rs6000_hard_regno_mode_ok): New function.
3594 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
3595 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
3596 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3597 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
3598 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
3599 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
3600 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3601 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
3602 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
3603 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3604 (sh_hard_regno_mode_ok): Make static.
3605 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
3606 instead of HARD_REGNO_MODE_OK.
3607 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
3608 (sparc_mode_class): Delete.
3609 (HARD_REGNO_MODE_OK): Delete.
3610 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3611 (hard_regno_mode_classes): Make static.
3612 (sparc_mode_class): Likewise.
3613 (sparc_hard_regno_mode_ok): New function.
3614 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
3615 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
3616 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
3618 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3619 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
3620 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
3621 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
3622 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
3623 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3624 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
3625 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
3626 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3627 (visium_hard_regno_mode_ok): New function.
3628 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
3629 instead of HARD_REGNO_MODE_OK.
3630 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
3631 (HARD_REGNO_MODE_OK): Delete.
3632 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
3633 (xtensa_hard_regno_mode_ok_p): ...this and make static.
3634 (xtensa_option_override): Update accordingly.
3635 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3636 (xtensa_hard_regno_mode_ok): New function.
3637 * system.h (HARD_REGNO_MODE_OK): Poison.
3639 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
3640 Alan Hayward <alan.hayward@arm.com>
3641 David Sherwood <david.sherwood@arm.com>
3643 * target.def (hard_regno_call_part_clobbered): New hook.
3644 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
3645 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
3646 * doc/tm.texi: Regenerate.
3647 * hooks.h (hook_bool_uint_mode_false): Declare.
3648 * hooks.c (hook_bool_uint_mode_false): New function.
3649 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3650 * cselib.c (cselib_process_insn): Use
3651 targetm.hard_regno_call_part_clobbered instead of
3652 HARD_REGNO_CALL_PART_CLOBBERED.
3653 * ira-conflicts.c (ira_build_conflicts): Likewise.
3654 * ira-costs.c (ira_tune_allocno_costs): Likewise.
3655 * lra-constraints.c (need_for_call_save_p): Likewise.
3656 * lra-lives.c: Include target.h.
3657 (check_pseudos_live_through_calls): Use
3658 targetm.hard_regno_call_part_clobbered instead of
3659 HARD_REGNO_CALL_PART_CLOBBERED.
3660 * regcprop.c: Include target.h.
3661 (copyprop_hardreg_forward_1): Use
3662 targetm.hard_regno_call_part_clobbered instead of
3663 HARD_REGNO_CALL_PART_CLOBBERED.
3664 * reginfo.c (choose_hard_reg_mode): Likewise.
3665 * regrename.c (check_new_reg_p): Likewise.
3666 * reload.c (find_equiv_reg): Likewise.
3667 * reload1.c (emit_reload_insns): Likewise.
3668 * sched-deps.c (deps_analyze_insn): Likewise.
3669 * sel-sched.c (init_regs_for_mode): Likewise.
3670 (mark_unavailable_hard_regs): Likewise.
3671 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
3672 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3673 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
3675 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3676 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3677 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
3679 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
3681 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3682 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3683 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
3685 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3686 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3687 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
3689 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3690 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
3692 * config/powerpcspe/powerpcspe.c
3693 (rs6000_hard_regno_call_part_clobbered): New function.
3694 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3695 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3696 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
3698 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3699 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3700 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
3702 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3703 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3704 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
3706 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
3707 Alan Hayward <alan.hayward@arm.com>
3708 David Sherwood <david.sherwood@arm.com>
3710 * rtl.h (subreg_memory_offset): Declare.
3711 * emit-rtl.c (subreg_memory_offset): New function.
3712 * expmed.c (store_bit_field_1): Use it.
3713 * expr.c (undefined_operand_subword_p): Likewise.
3714 * simplify-rtx.c (simplify_subreg): Likewise.
3716 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
3718 PR rtl-optimization/57448
3721 * optabs.c (expand_atomic_load): Place compiler memory barriers if
3722 using atomic_load pattern.
3723 (expand_atomic_store): Likewise.
3725 2017-09-04 Jakub Jelinek <jakub@redhat.com>
3728 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
3729 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
3730 handling. Use replace_call_with_value with NULL instead of
3731 gsi_replace, unlink_stmt_vdef and release_defs.
3733 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
3736 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
3738 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
3741 * rtl.h (emit_library_call_value_1): Declare.
3742 (emit_library_call): Replace declaration with a series of overloads.
3743 Remove the parameter count argument.
3744 (emit_library_call_value): Likewise.
3745 * calls.c (emit_library_call_value_1): Make global. Replace varargs
3746 with an "rtx_mode_t *".
3747 (emit_library_call_value): Delete.
3748 (emit_library_call): Likewise.
3749 * asan.c (asan_emit_stack_protection): Update calls accordingly.
3750 (asan_emit_allocas_unpoison): Likewise.
3751 * builtins.c (expand_builtin_powi): Likewise.
3752 (expand_asan_emit_allocas_unpoison): Likewise.
3753 * cfgexpand.c (expand_main_function): Likewise.
3754 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
3755 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
3756 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
3757 * config/arm/arm.c (arm_trampoline_init): Likewise.
3758 (arm_call_tls_get_addr): Likewise.
3759 (arm_expand_divmod_libfunc): Likewise.
3760 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
3761 (smulsi3_highpart): Likewise.
3762 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
3763 (c6x_expand_compare): Likewise.
3764 (c6x_expand_movmem): Likewise.
3765 * config/frv/frv.c (frv_trampoline_init): Likewise.
3766 * config/i386/i386.c (ix86_trampoline_init): Likewise.
3767 (ix86_expand_divmod_libfunc): Likewise.
3768 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
3769 (ia64_expand_compare): Likewise.
3770 (ia64_profile_hook): Likewise.
3771 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
3772 (nonlocal_goto): Likewise.
3773 (restore_stack_nonlocal): Likewise.
3774 * config/m32r/m32r.c (block_move_call): Likewise.
3775 (m32r_trampoline_init): Likewise.
3776 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
3777 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
3778 (m68k_call_m68k_read_tp): Likewise.
3779 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
3780 (microblaze_expand_divide): Likewise.
3781 * config/mips/mips.h (mips_args): Likewise.
3782 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
3783 (MIPS_ICACHE_SYNC): Likewise.
3784 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
3785 (nios2_trampoline_init): Likewise.
3786 * config/pa/pa.c (hppa_tls_call): Likewise.
3787 (pa_trampoline_init): Likewise.
3788 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
3789 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
3790 (expand_strn_compare): Likewise.
3791 (rs6000_generate_compare): Likewise.
3792 (rs6000_expand_float128_convert): Likewise.
3793 (output_profile_hook): Likewise.
3794 (rs6000_trampoline_init): Likewise.
3795 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
3796 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
3797 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
3798 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
3799 (rs6000_generate_compare): Likewise.
3800 (rs6000_expand_float128_convert): Likewise.
3801 (output_profile_hook): Likewise.
3802 (rs6000_trampoline_init): Likewise.
3803 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
3804 * config/sh/sh.c (sh_trampoline_init): Likewise.
3805 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
3806 (sparc_emit_float_lib_cmp): Likewise.
3807 (sparc32_initialize_trampoline): Likewise.
3808 (sparc64_initialize_trampoline): Likewise.
3809 (sparc_profile_hook): Likewise.
3810 * config/spu/spu.c (ea_load_store): Likewise.
3811 * config/spu/spu.md (floatunssidf2): Likewise.
3812 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
3813 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
3814 * config/visium/visium.c (expand_block_move_4): Likewise.
3815 (expand_block_move_2): Likewise.
3816 (expand_block_move_1): Likewise.
3817 (expand_block_set_4): Likewise.
3818 (expand_block_set_2): Likewise.
3819 (expand_block_set_1): Likewise.
3820 (visium_trampoline_init): Likewise.
3821 (visium_profile_hook): Likewise.
3822 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
3823 (xtensa_setup_frame_addresses): Likewise.
3824 (xtensa_trampoline_init): Likewise.
3825 * except.c (sjlj_emit_function_enter): Likewise.
3826 (sjlj_emit_function_exit): Likewise.
3827 * explow.c (allocate_dynamic_stack_space): Likewise.
3828 (probe_stack_range): Likewise.
3829 * expr.c (convert_mode_scalar): Likewise.
3830 * optabs.c (expand_binop): Likewise.
3831 (expand_twoval_binop_libfunc): Likewise.
3832 (expand_unop): Likewise.
3833 (prepare_cmp_insn): Likewise.
3834 (prepare_float_lib_cmp): Likewise.
3835 (expand_float): Likewise.
3836 (expand_fix): Likewise.
3837 (expand_fixed_convert): Likewise.
3838 (maybe_emit_sync_lock_test_and_set): Likewise.
3839 (expand_atomic_compare_and_swap): Likewise.
3840 (expand_mem_thread_fence): Likewise.
3841 (expand_atomic_fetch_op): Likewise.
3843 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
3845 * doc/generic.texi (OpenACC): Adjust URL.
3846 * doc/invoke.texi (C Dialect Options): Ditto.
3848 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
3850 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
3851 predicate for operand 1. Add (m,<S>) constraint.
3852 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
3853 Prevent memory operand 1 with register operand 2.
3855 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
3857 PR rtl-optimization/82024
3858 * combine.c (try_combine): If the combination result is a PARALLEL,
3859 and we only need to retain the SET in there that would be placed
3860 at I2, check that we can place that at I3 instead, before doing so.
3862 2017-09-01 Jakub Jelinek <jakub@redhat.com>
3865 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
3867 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
3868 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
3871 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
3872 Jeff Law <law@redhat.com>
3874 * varasm.c (bss_initializer_p): Do not put constants into .bss
3875 (categorize_decl_for_section): Handle bss_initializer_p returning
3876 false when DECL_INITIAL is NULL.
3878 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3881 * config/s390/s390.c (s390_can_inline_p): New function.
3883 2017-09-01 Jeff Law <law@redhat.com>
3885 PR tree-optimization/82052
3886 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
3887 Always initialize the returned slot after a hash table miss
3888 when INSERT is true.
3890 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
3892 * config/s390/s390.md (mem_signal_fence): Remove.
3893 * doc/md.texi (mem_signal_fence): Remove.
3894 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
3896 * target-insns.def (mem_signal_fence): Remove.
3898 2017-09-01 Jakub Jelinek <jakub@redhat.com>
3901 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
3904 * asan.c (create_odr_indicator): Strip name encoding from assembler
3905 name before appending it after __odr_asan_.
3907 2017-09-01 Martin Liska <mliska@suse.cz>
3909 PR tree-optimization/82059
3910 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
3911 frequency only when an edge is redirected.
3913 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
3915 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
3916 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
3917 (arc_conditional_register_usage): Remove ARC600 lp_count
3919 (arc_file_start): Emit Tag_ARC_CPU_variation.
3920 (arc_can_use_doloop_p): New conditions to use ZOLs.
3921 (hwloop_fail): New function.
3922 (hwloop_optimize): Likewise.
3923 (hwloop_pattern_reg): Likewise.
3924 (arc_doloop_hooks): New struct, to be used with reorg_loops.
3925 (arc_reorg_loops): New function, calls reorg_loops.
3926 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
3927 (arc600_corereg_hazard): Remove ZOL checking, case handled by
3929 (arc_loop_hazard): Remove function, functionality moved into
3931 (arc_hazard): Remove arc_loop_hazard call.
3932 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
3933 into hwloop_optimize.
3934 (arc_label_align): Remove ZOL handling.
3935 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
3936 * config/arc/arc.md (doloop_begin): Remove pattern.
3937 (doloop_begin_i): Likewise.
3938 (doloop_end_i): Likewise.
3939 (doloop_fallback): Likewise.
3940 (doloop_fallback_m): Likewise.
3941 (doloop_end): Reimplement expand.
3942 (arc_lp): New pattern for LP instruction.
3943 (loop_end): New pattern.
3944 (loop_fail): Likewise.
3945 (decrement_and_branch_until_zero): Likewise.
3946 * config/arc/arc.opt (mlpc-width): New option.
3947 * doc/invoke.texi (mlpc-width): Document option.
3949 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
3951 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
3952 (arc_ccfsm_advance): Fix checking for delay slots.
3953 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
3955 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
3957 * config/arc/arc.md (movqi_insn): Add stores to save constant long
3959 (movhi_insn): Update store instruction constraint which are saving
3960 6-bit short immediates.
3961 (movsi_insn): Consider also short scaled load operations.
3962 (zero_extendhisi2_i): Use Usd constraint instead of T.
3963 (extendhisi2_i): Add q constraint.
3964 (arc_clzsi2): Add type and length attributes.
3965 (arc_ctzsi2): Likewise.
3966 * config/arc/constraints.md (Usc): Update constraint, the
3967 assembler can parse two relocations for a single instruction.
3969 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
3971 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
3972 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
3974 2017-08-31 Olivier Hainque <hainque@adacore.com>
3976 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
3977 match as powerpc-wrs-vxworks*.
3979 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
3981 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
3982 register constraint for by-element operand.
3983 (aarch64_mls_elt_merge<mode>): Likewise.
3985 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
3987 * config/arc/arc.c (arc_can_follow_jump): Check for short
3990 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
3992 * config.gcc: Use g.opt for arc.
3993 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
3994 functionality moved to ...
3995 (legitimate_scaled_address_p): New function, ...here.
3996 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
3997 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
3998 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
4000 (arc_override_options): Handle G option.
4001 (arc_output_pic_addr_const): Correct function definition.
4002 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
4003 (arc_decl_anon_ns_mem_p): Delete.
4004 (arc_in_small_data_p): Overhaul this function to take into
4005 consideration the value given via G option.
4006 (arc_rewrite_small_data_1): Renamed and corrected old
4007 arc_rewrite_small_data function.
4008 (arc_rewrite_small_data): New function.
4009 (small_data_pattern): Don't use pic_offset_table_rtx.
4010 * config/arc/arc.h (CC1_SPEC): Recognize G option.
4011 * config/arc/simdext.md (movmisalignv2hi): Use
4012 prepare_move_operands function.
4014 (movmisalign*): Likewise.
4015 * doc/invoke.texi (ARC options): Document -G option.
4017 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
4019 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
4021 * config/arc/arc.c (arc_print_operand): Output scalled address for
4022 sdata whenever is possible.
4023 (arc_in_small_data_p): Allow sdata for 64bit datum when double
4024 load/stores are available.
4025 (compact_sda_memory_operand): Check for the alignment required by
4026 code density instructions.
4027 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
4029 * config/arc/constraints.md (Usd): Update constraint.
4030 (Us0): New constraint.
4031 (Usc): Update constraint.
4033 2017-08-31 Richard Biener <rguenther@suse.de>
4036 * dwarf2out.c (dwarf2out_early_global_decl): Process each
4039 2017-08-31 Tamar Christina <tamar.christina@arm.com>
4041 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
4042 Resize type_signature.
4044 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
4045 Alan Hayward <alan.hayward@arm.com>
4046 David Sherwood <david.sherwood@arm.com>
4048 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
4049 subregs whose inner modes can be stored in GPRs.
4050 (aarch64_classify_index): Likewise.
4052 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
4053 Alan Hayward <alan.hayward@arm.com>
4054 David Sherwood <david.sherwood@arm.com>
4056 * config/aarch64/iterators.md (V_cmp_result): Rename to...
4057 (V_INT_EQUIV): ...this.
4058 (v_cmp_result): Rename to...
4059 (v_int_equiv): ...this.
4060 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
4061 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
4062 (copysign<mode>3): Likewise.
4063 (aarch64_simd_bsl<mode>_internal): Likewise.
4064 (aarch64_simd_bsl<mode>): Likewise.
4065 (vec_cmp<mode><mode>): Likewise.
4066 (vcond<mode><mode>): Likewise.
4067 (vcond<v_cmp_mixed><mode>): Likewise.
4068 (vcondu<mode><v_cmp_mixed>): Likewise.
4069 (aarch64_cm<optab><mode>): Likewise.
4070 (aarch64_cmtst<mode>): Likewise.
4071 (aarch64_fac<optab><mode>): Likewise.
4072 (vec_perm_const<mode>): Likewise.
4073 (vcond_mask_<mode><v_cmp_result>): Rename to...
4074 (vcond_mask_<mode><v_int_equiv>): ...this.
4075 (vec_cmp<mode><v_cmp_result>): Rename to...
4076 (vec_cmp<mode><v_int_equiv>): ...this.
4078 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
4079 Alan Hayward <alan.hayward@arm.com>
4080 David Sherwood <david.sherwood@arm.com>
4082 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
4084 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
4085 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
4086 (UNSPEC_LD4_DREG): New unspecs.
4087 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
4088 (aarch64_ld2<mode>_dreg_be): Replace with...
4089 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
4091 (aarch64_ld3<mode>_dreg_le)
4092 (aarch64_ld3<mode>_dreg_be): Replace with...
4093 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
4095 (aarch64_ld4<mode>_dreg_le)
4096 (aarch64_ld4<mode>_dreg_be): Replace with...
4097 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
4100 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4102 PR tree-optimization/81987
4103 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
4104 insert an initializer in a location not dominated by the stride
4107 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
4109 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
4110 on the entire header of the finally block in the fallthru case.
4112 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
4114 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
4116 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
4118 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
4119 rs6000_emit_move_from_cr and call renamed function.
4120 (rs6000_emit_prologue): Call renamed functions.
4121 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
4122 movesi_from_cr, remove volatile CRs.
4124 2017-08-30 Jon Beniston <jon@beniston.com>
4125 Richard Biener <rguenther@suse.de>
4127 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
4128 of VECTOR_MODE_P check.
4129 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
4130 element vector types.
4132 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4134 * df.h (df_read_modify_subreg_p): Remove in favor of...
4135 * rtl.h (read_modify_subreg_p): ...this new function. Take a
4136 const_rtx instead of an rtx.
4137 * cprop.c (local_cprop_find_used_regs): Update accordingly.
4138 * df-problems.c (df_word_lr_mark_ref): Likewise.
4139 * ira-lives.c (mark_pseudo_reg_live): Likewise.
4140 (mark_pseudo_reg_dead): Likewise.
4141 (mark_ref_dead): Likewise.
4142 * reginfo.c (init_subregs_of_mode): Likewise.
4143 * sched-deps.c (sched_analyze_1): Likewise.
4144 * df-scan.c (df_def_record_1): Likewise.
4145 (df_uses_record): Likewise.
4146 (df_read_modify_subreg_p): Remove in favor of...
4147 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
4148 const_rtx instead of an rtx.
4150 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4151 Alan Hayward <alan.hayward@arm.com>
4152 David Sherwood <david.sherwood@arm.com>
4154 * rtl.h (partial_subreg_p): New function.
4155 * caller-save.c (save_call_clobbered_regs): Use it.
4156 * calls.c (expand_call): Likewise.
4157 * combine.c (combinable_i3pat): Likewise.
4158 (simplify_set): Likewise.
4159 (make_extraction): Likewise.
4160 (make_compound_operation_int): Likewise.
4161 (gen_lowpart_or_truncate): Likewise.
4162 (force_to_mode): Likewise.
4163 (make_field_assignment): Likewise.
4164 (reg_truncated_to_mode): Likewise.
4165 (record_truncated_value): Likewise.
4166 (move_deaths): Likewise.
4167 * cse.c (record_jump_cond): Likewise.
4168 (cse_insn): Likewise.
4169 * cselib.c (cselib_lookup_1): Likewise.
4170 * expmed.c (extract_bit_field_using_extv): Likewise.
4171 * function.c (assign_parm_setup_reg): Likewise.
4172 * ifcvt.c (noce_convert_multiple_sets): Likewise.
4173 * ira-build.c (create_insn_allocnos): Likewise.
4174 * lra-coalesce.c (merge_pseudos): Likewise.
4175 * lra-constraints.c (match_reload): Likewise.
4176 (simplify_operand_subreg): Likewise.
4177 (curr_insn_transform): Likewise.
4178 * lra-lives.c (process_bb_lives): Likewise.
4179 * lra.c (new_insn_reg): Likewise.
4180 (lra_substitute_pseudo): Likewise.
4181 * regcprop.c (mode_change_ok): Likewise.
4182 (maybe_mode_change): Likewise.
4183 (copyprop_hardreg_forward_1): Likewise.
4184 * reload.c (push_reload): Likewise.
4185 (find_reloads): Likewise.
4186 (find_reloads_subreg_address): Likewise.
4187 * reload1.c (alter_reg): Likewise.
4188 (eliminate_regs_1): Likewise.
4189 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4191 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
4193 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
4194 back to if statements, including unpack.
4196 2017-08-30 Martin Liska <mliska@suse.cz>
4199 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
4201 (func_checker::compare_asm_inputs_outputs): ... this function.
4202 (func_checker::compare_gimple_asm): Use the function to compare
4203 also ASM constrains.
4204 * ipa-icf-gimple.h: Rename the function.
4206 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4207 Alan Hayward <alan.hayward@arm.com>
4208 David Sherwood <david.sherwood@arm.com>
4210 * coretypes.h (complex_mode): New type.
4211 * gdbhooks.py (build_pretty_printer): Handle it.
4212 * machmode.h (complex_mode): New class.
4213 (complex_mode::includes_p): New function.
4214 (is_complex_int_mode): Likewise.
4215 (is_complex_float_mode): Likewise.
4216 * genmodes.c (get_mode_class): Handle complex mode classes.
4217 * function.c (expand_function_end): Use is_complex_int_mode.
4219 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4220 Alan Hayward <alan.hayward@arm.com>
4221 David Sherwood <david.sherwood@arm.com>
4223 * coretypes.h (scalar_mode_pod): New typedef.
4224 * gdbhooks.py (build_pretty_printer): Handle it.
4225 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
4226 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
4227 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
4228 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
4231 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4232 Alan Hayward <alan.hayward@arm.com>
4233 David Sherwood <david.sherwood@arm.com>
4235 * machmode.h (mode_for_vector): Take a scalar_mode instead
4237 * stor-layout.c (mode_for_vector): Likewise.
4238 * explow.c (promote_mode): Use as_a <scalar_mode>.
4239 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
4241 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4242 Alan Hayward <alan.hayward@arm.com>
4243 David Sherwood <david.sherwood@arm.com>
4245 * target.def (preferred_simd_mode): Take a scalar_mode
4246 instead of a machine_mode.
4247 * targhooks.h (default_preferred_simd_mode): Likewise.
4248 * targhooks.c (default_preferred_simd_mode): Likewise.
4249 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
4250 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
4251 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
4252 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
4253 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
4254 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
4255 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
4256 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
4258 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
4259 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
4260 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
4261 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
4262 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
4263 * doc/tm.texi: Regenerate.
4264 * optabs-query.c (can_vec_mask_load_store_p): Return false for
4267 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4268 Alan Hayward <alan.hayward@arm.com>
4269 David Sherwood <david.sherwood@arm.com>
4271 * target.def (scalar_mode_supported_p): Take a scalar_mode
4272 instead of a machine_mode.
4273 * targhooks.h (default_scalar_mode_supported_p): Likewise.
4274 * targhooks.c (default_scalar_mode_supported_p): Likewise.
4275 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
4276 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
4277 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
4278 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
4279 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
4280 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
4281 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
4282 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
4283 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
4284 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
4285 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
4286 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
4288 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
4289 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
4290 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
4291 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
4292 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
4294 * doc/tm.texi: Regenerate.
4296 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4297 Alan Hayward <alan.hayward@arm.com>
4298 David Sherwood <david.sherwood@arm.com>
4300 * coretypes.h (opt_scalar_mode): New typedef.
4301 * gdbhooks.py (build_pretty_printers): Handle it.
4302 * machmode.h (mode_iterator::get_2xwider): Add overload for
4304 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
4306 * expr.c (convert_mode_scalar): Likewise.
4307 * omp-low.c (omp_clause_aligned_alignment): Likewise.
4308 * optabs.c (expand_float): Likewise.
4309 (expand_fix): Likewise.
4310 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
4312 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4313 Alan Hayward <alan.hayward@arm.com>
4314 David Sherwood <david.sherwood@arm.com>
4316 * optabs.c (expand_float): Explicitly check for scalars before
4317 using a branching expansion.
4318 (expand_fix): Likewise.
4320 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4321 Alan Hayward <alan.hayward@arm.com>
4322 David Sherwood <david.sherwood@arm.com>
4324 * expr.c (convert_mode): Split scalar handling out into...
4325 (convert_mode_scalar): ...this new function. Treat the modes
4328 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4329 Alan Hayward <alan.hayward@arm.com>
4330 David Sherwood <david.sherwood@arm.com>
4332 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
4334 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
4336 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4337 Alan Hayward <alan.hayward@arm.com>
4338 David Sherwood <david.sherwood@arm.com>
4340 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
4341 rather than a machine_mode.
4342 (fixed_from_string): Likewise.
4343 (fixed_convert): Likewise.
4344 (fixed_convert_from_int): Likewise.
4345 (fixed_convert_from_real): Likewise.
4346 (real_convert_from_fixed): Likewise.
4347 * fixed-value.c (fixed_from_double_int): Likewise.
4348 (fixed_from_string): Likewise.
4349 (fixed_convert): Likewise.
4350 (fixed_convert_from_int): Likewise.
4351 (fixed_convert_from_real): Likewise.
4352 (real_convert_from_fixed): Likewise.
4353 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
4355 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4356 Alan Hayward <alan.hayward@arm.com>
4357 David Sherwood <david.sherwood@arm.com>
4359 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
4360 of separate mode class checks. Do not allow vector modes here.
4361 (immed_wide_int_const): Use as_a <scalar_mode>.
4362 * explow.c (trunc_int_for_mode): Likewise.
4363 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
4364 (wi::shwi): Likewise.
4365 (wi::min_value): Likewise.
4366 (wi::max_value): Likewise.
4367 * dwarf2out.c (loc_descriptor): Likewise.
4368 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
4371 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4372 Alan Hayward <alan.hayward@arm.com>
4373 David Sherwood <david.sherwood@arm.com>
4375 * tree.h (SCALAR_TYPE_MODE): New macro.
4376 * expr.c (expand_expr_addr_expr_1): Use it.
4377 (expand_expr_real_2): Likewise.
4378 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
4379 (fold_convert_const_fixed_from_int): Likewise.
4380 (fold_convert_const_fixed_from_real): Likewise.
4381 (native_encode_fixed): Likewise
4382 (native_encode_complex): Likewise
4383 (native_encode_vector): Likewise.
4384 (native_interpret_fixed): Likewise.
4385 (native_interpret_real): Likewise.
4386 (native_interpret_complex): Likewise.
4387 (native_interpret_vector): Likewise.
4388 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
4389 (simd_clone_adjust_argument_types): Likewise.
4390 (simd_clone_init_simd_arrays): Likewise.
4391 (simd_clone_adjust): Likewise.
4392 * stor-layout.c (layout_type): Likewise.
4393 * tree.c (build_minus_one_cst): Likewise.
4394 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
4395 * tree-inline.c (estimate_move_cost): Likewise.
4396 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
4397 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
4398 (vectorizable_reduction): Likewise.
4399 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
4400 (vect_recog_mixed_size_cond_pattern): Likewise.
4401 (check_bool_pattern): Likewise.
4402 (adjust_bool_pattern): Likewise.
4403 (search_type_for_mask_1): Likewise.
4404 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
4405 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
4406 (vectorizable_load): Likewise.
4407 (vectorizable_store): Likewise.
4408 * ubsan.c (ubsan_encode_value): Likewise.
4409 * varasm.c (output_constant): Likewise.
4411 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4412 Alan Hayward <alan.hayward@arm.com>
4413 David Sherwood <david.sherwood@arm.com>
4415 * coretypes.h (scalar_mode): New class.
4416 * machmode.h (scalar_mode): Likewise.
4417 (scalar_mode::includes_p): New function.
4418 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
4419 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
4420 * genmodes.c (get_mode_class): Handle remaining scalar modes.
4421 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
4422 * expmed.c (store_bit_field_1): Likewise.
4423 (extract_bit_field_1): Likewise.
4424 * expr.c (write_complex_part): Likewise.
4425 (read_complex_part): Likewise.
4426 (emit_move_complex_push): Likewise.
4427 (expand_expr_real_2): Likewise.
4428 * function.c (assign_parm_setup_reg): Likewise.
4429 (assign_parms_unsplit_complex): Likewise.
4430 * optabs.c (expand_binop): Likewise.
4431 * rtlanal.c (subreg_get_info): Likewise.
4432 * simplify-rtx.c (simplify_immed_subreg): Likewise.
4433 * varasm.c (output_constant_pool_2): Likewise.
4435 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4436 Alan Hayward <alan.hayward@arm.com>
4437 David Sherwood <david.sherwood@arm.com>
4439 * expmed.c (extract_high_half): Use scalar_int_mode and remove
4441 (expmed_mult_highpart_optab): Likewise.
4442 (expmed_mult_highpart): Likewise.
4444 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4445 Alan Hayward <alan.hayward@arm.com>
4446 David Sherwood <david.sherwood@arm.com>
4448 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
4449 instead of a machine_mode.
4450 (builtin_memset_read_str): Likewise.
4451 * builtins.c (c_readstr): Likewise.
4452 (builtin_memcpy_read_str): Likewise.
4453 (builtin_strncpy_read_str): Likewise.
4454 (builtin_memset_read_str): Likewise.
4455 (builtin_memset_gen_str): Likewise.
4456 (expand_builtin_signbit): Use scalar_int_mode for local variables.
4457 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
4458 instead of a machine_mode.
4459 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
4461 (make_extraction): Likewise.
4462 (try_widen_shift_mode): Take and return scalar_int_modes instead
4464 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
4465 a scalar_int_mode instead of a machine_mode.
4466 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
4467 (avr_addr_space_pointer_mode): Likewise.
4468 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
4469 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
4470 (msp430_unwind_word_mode): Likewise.
4471 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
4472 (spu_addr_space_pointer_mode): Likewise.
4473 (spu_addr_space_address_mode): Likewise.
4474 (spu_libgcc_cmp_return_mode): Likewise.
4475 (spu_libgcc_shift_count_mode): Likewise.
4476 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
4477 (rl78_addr_space_pointer_mode): Likewise.
4478 (fl78_unwind_word_mode): Likewise.
4479 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
4481 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
4482 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
4483 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
4484 (mips_valid_pointer_mode): Likewise.
4485 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
4486 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
4487 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
4489 (ft32_addr_space_address_mode): Likewise.
4490 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
4491 scalar_int_mode instead of a machine_mode.
4492 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
4494 (m32c_addr_space_address_mode): Likewise.
4495 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
4496 (rs6000_eh_return_filter_mode): Likewise.
4497 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
4498 (rs6000_eh_return_filter_mode): Likewise.
4499 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
4500 (s390_libgcc_shift_count_mode): Likewise.
4501 (s390_unwind_word_mode): Likewise.
4502 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
4504 * target.def (mode_rep_extended): Likewise.
4505 (valid_pointer_mode): Likewise.
4506 (addr_space.valid_pointer_mode): Likewise.
4507 (eh_return_filter_mode): Return a scalar_int_mode rather than
4509 (libgcc_cmp_return_mode): Likewise.
4510 (libgcc_shift_count_mode): Likewise.
4511 (unwind_word_mode): Likewise.
4512 (addr_space.pointer_mode): Likewise.
4513 (addr_space.address_mode): Likewise.
4514 * doc/tm.texi: Regenerate.
4515 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
4517 (do_jump): Use scalar_int_mode for local variables.
4518 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
4519 rather than a machine_mode.
4520 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
4521 (scompare_loc_descriptor_wide): Likewise.
4522 (scompare_loc_descriptor_narrow): Likewise.
4523 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
4525 * except.c (sjlj_emit_dispatch_table): Likewise.
4526 (expand_builtin_eh_copy_values): Likewise.
4527 * explow.c (convert_memory_address_addr_space_1): Likewise.
4528 Take a scalar_int_mode rather than a machine_mode.
4529 (convert_memory_address_addr_space): Take a scalar_int_mode rather
4530 than a machine_mode.
4531 (memory_address_addr_space): Use scalar_int_mode for local variables.
4532 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
4533 rather than a machine_mode.
4534 * expmed.c (mask_rtx): Likewise.
4535 (init_expmed_one_conv): Likewise.
4536 (expand_mult_highpart_adjust): Likewise.
4537 (extract_high_half): Likewise.
4538 (expmed_mult_highpart_optab): Likewise.
4539 (expmed_mult_highpart): Likewise.
4540 (expand_smod_pow2): Likewise.
4541 (expand_sdiv_pow2): Likewise.
4542 (emit_store_flag_int): Likewise.
4543 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
4545 (extract_low_bits): Likewise.
4546 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
4548 * expr.c (pieces_addr::adjust): Likewise.
4549 (can_store_by_pieces): Likewise.
4550 (store_by_pieces): Likewise.
4551 (clear_by_pieces_1): Likewise.
4552 (expand_expr_addr_expr_1): Likewise.
4553 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
4554 (expand_expr_real_1): Likewise.
4555 (try_casesi): Likewise.
4556 * final.c (shorten_branches): Likewise.
4557 * fold-const.c (fold_convert_const_int_from_fixed): Change the
4558 type of "mode" to machine_mode.
4559 * internal-fn.c (expand_arith_overflow_result_store): Take a
4560 scalar_int_mode rather than a machine_mode.
4561 (expand_mul_overflow): Use scalar_int_mode for local variables.
4562 * loop-doloop.c (doloop_modify): Likewise.
4563 (doloop_optimize): Likewise.
4564 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
4565 than a machine_mode.
4566 (expand_doubleword_shift_condmove): Likewise.
4567 (expand_doubleword_shift): Likewise.
4568 (expand_doubleword_clz): Likewise.
4569 (expand_doubleword_popcount): Likewise.
4570 (expand_doubleword_parity): Likewise.
4571 (expand_absneg_bit): Use scalar_int_mode for local variables.
4572 (prepare_float_lib_cmp): Likewise.
4573 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
4574 rather than a machine_mode.
4575 (convert_memory_address_addr_space): Likewise.
4576 (get_mode_bounds): Likewise.
4577 (get_address_mode): Return a scalar_int_mode rather than a
4579 * rtlanal.c (get_address_mode): Likewise.
4580 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
4581 than a machine_mode.
4582 * targhooks.c (default_mode_rep_extended): Likewise.
4583 (default_valid_pointer_mode): Likewise.
4584 (default_addr_space_valid_pointer_mode): Likewise.
4585 (default_eh_return_filter_mode): Return a scalar_int_mode rather
4586 than a machine_mode.
4587 (default_libgcc_cmp_return_mode): Likewise.
4588 (default_libgcc_shift_count_mode): Likewise.
4589 (default_unwind_word_mode): Likewise.
4590 (default_addr_space_pointer_mode): Likewise.
4591 (default_addr_space_address_mode): Likewise.
4592 * targhooks.h (default_eh_return_filter_mode): Likewise.
4593 (default_libgcc_cmp_return_mode): Likewise.
4594 (default_libgcc_shift_count_mode): Likewise.
4595 (default_unwind_word_mode): Likewise.
4596 (default_addr_space_pointer_mode): Likewise.
4597 (default_addr_space_address_mode): Likewise.
4598 (default_mode_rep_extended): Take a scalar_int_mode rather than
4600 (default_valid_pointer_mode): Likewise.
4601 (default_addr_space_valid_pointer_mode): Likewise.
4602 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
4604 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
4605 rather than a machine_mode.
4606 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
4607 for local variables.
4608 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
4609 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
4610 than a machine_mode.
4612 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4613 Alan Hayward <alan.hayward@arm.com>
4614 David Sherwood <david.sherwood@arm.com>
4616 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
4617 the mode argument to scalar_int_mode.
4618 (do_jump_by_parts_zero_rtx): Likewise.
4619 (do_jump_by_parts_equality_rtx): Likewise.
4620 (do_jump_by_parts_greater): Take a mode argument.
4621 (do_jump_by_parts_equality): Likewise.
4622 (do_jump_1): Update calls accordingly.
4624 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4625 Alan Hayward <alan.hayward@arm.com>
4626 David Sherwood <david.sherwood@arm.com>
4628 * is-a.h (safe_dyn_cast): New function.
4629 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
4630 (jump_table_for_label): Likewise.
4631 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
4632 instead of an rtx_insn *.
4633 (shorten_branches): Use dyn_cast instead of LABEL_P and
4634 JUMP_TABLE_DATA_P. Use jump_table_for_label and
4635 rtx_jump_table_data::get_data_mode.
4636 (final_scan_insn): Likewise.
4638 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4639 Alan Hayward <alan.hayward@arm.com>
4640 David Sherwood <david.sherwood@arm.com>
4642 * combine.c (try_combine): Use is_a <scalar_int_mode> when
4643 trying to combine a full-register integer set with a subreg
4646 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4647 Alan Hayward <alan.hayward@arm.com>
4648 David Sherwood <david.sherwood@arm.com>
4650 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
4651 that is always either address_mode or pointer_mode.
4653 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4654 Alan Hayward <alan.hayward@arm.com>
4655 David Sherwood <david.sherwood@arm.com>
4657 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
4658 when the two are known to be equal.
4660 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4661 Alan Hayward <alan.hayward@arm.com>
4662 David Sherwood <david.sherwood@arm.com>
4664 * simplify-rtx.c (simplify_const_unary_operation): Use
4665 is_a <scalar_int_mode> instead of checking for a nonzero
4666 precision. Forcibly convert op_mode to a scalar_int_mode
4667 in that case. More clearly differentiate the operand and
4668 result modes and use the former when deciding what the value
4669 of a count-bits operation should be. Use is_int_mode instead
4670 of checking for a MODE_INT. Remove redundant check for whether
4671 this mode has a zero precision.
4673 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4674 Alan Hayward <alan.hayward@arm.com>
4675 David Sherwood <david.sherwood@arm.com>
4677 * optabs.c (widen_leading): Change the type of the mode argument
4678 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
4679 (widen_bswap): Likewise.
4680 (expand_parity): Likewise.
4681 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
4682 (expand_ffs): Likewise.
4683 (epand_unop): Check for scalar integer modes before calling the
4686 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4687 Alan Hayward <alan.hayward@arm.com>
4688 David Sherwood <david.sherwood@arm.com>
4690 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
4692 (expand_expr_real_1): Update call accordingly.
4694 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4695 Alan Hayward <alan.hayward@arm.com>
4696 David Sherwood <david.sherwood@arm.com>
4698 * expmed.c (store_bit_field_using_insv): Add op0_mode and
4699 value_mode arguments. Use scalar_int_mode internally.
4700 (store_bit_field_1): Rename the new integer mode from imode
4701 to op0_mode and use it instead of GET_MODE (op0). Update calls
4702 to store_split_bit_field, store_bit_field_using_insv and
4703 store_fixed_bit_field.
4704 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
4705 Use scalar_int_mode internally. Use a bit count rather than a mode
4706 when calculating the largest bit size for get_best_mode.
4707 Update calls to store_split_bit_field and store_fixed_bit_field_1.
4708 (store_fixed_bit_field_1): Add mode and value_mode arguments.
4709 Remove assertion that OP0 has a scalar integer mode.
4710 (store_split_bit_field): Add op0_mode and value_mode arguments.
4711 Update calls to extract_fixed_bit_field.
4712 (extract_bit_field_using_extv): Add an op0_mode argument.
4713 Use scalar_int_mode internally.
4714 (extract_bit_field_1): Rename the new integer mode from imode to
4715 op0_mode and use it instead of GET_MODE (op0). Update calls to
4716 extract_split_bit_field, extract_bit_field_using_extv and
4717 extract_fixed_bit_field.
4718 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
4719 to extract_split_bit_field and extract_fixed_bit_field_1.
4720 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
4721 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
4723 (extract_split_bit_field): Add an op0_mode argument. Update call
4724 to extract_fixed_bit_field.
4726 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4727 Alan Hayward <alan.hayward@arm.com>
4728 David Sherwood <david.sherwood@arm.com>
4730 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
4731 * explow.c (hard_function_value): Likewise.
4732 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
4733 convert_to_mode call outside the loop.
4734 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
4735 for the mode iterator. Require the mode specified by max_pieces
4737 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
4739 (copy_blkmode_to_reg): Likewise.
4740 (set_storage_via_setmem): Likewise.
4741 * optabs.c (prepare_cmp_insn): Likewise.
4742 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
4743 * stor-layout.c (finish_bitfield_representative): Likewise.
4745 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4746 Alan Hayward <alan.hayward@arm.com>
4747 David Sherwood <david.sherwood@arm.com>
4749 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
4750 * expr.c (convert_move): Use them.
4751 (convert_modes): Likewise.
4752 (store_expr_with_bounds): Likewise.
4754 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4755 Alan Hayward <alan.hayward@arm.com>
4756 David Sherwood <david.sherwood@arm.com>
4758 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
4759 parameter for the mode of "x". Remove the "known_x", "known_mode"
4760 and "known_ret" arguments. Change the type of the mode argument
4762 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
4763 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
4764 (reg_num_sign_bit_copies_for_combine): Likewise.
4765 * rtlanal.c (nonzero_bits1): Likewise.
4766 (num_sign_bit_copies1): Likewise.
4767 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
4768 (reg_num_sign_bit_copies_general): Likewise.
4769 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
4770 (reg_nonzero_bits_general): Likewise.
4772 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4773 Alan Hayward <alan.hayward@arm.com>
4774 David Sherwood <david.sherwood@arm.com>
4776 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
4777 than in subroutines. Return 1 for non-integer modes.
4778 (cached_num_sign_bit_copies): Change the type of the mode parameter
4780 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
4781 classes. Handle CONST_INT_P first and then check whether X also
4782 has a scalar integer mode. Check the same thing for inner registers
4783 of a SUBREG and for values that are being extended or truncated.
4785 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4786 Alan Hayward <alan.hayward@arm.com>
4787 David Sherwood <david.sherwood@arm.com>
4789 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
4790 in subroutines. Return the mode mask for non-integer modes.
4791 (cached_nonzero_bits): Change the type of the mode parameter
4793 (nonzero_bits1): Likewise. Remove early exit for other mode
4794 classes. Handle CONST_INT_P first and then check whether X
4795 also has a scalar integer mode.
4797 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4798 Alan Hayward <alan.hayward@arm.com>
4799 David Sherwood <david.sherwood@arm.com>
4801 * expr.c (widest_int_mode_for_size): Make the comment match the code.
4802 Return a scalar_int_mode and assert that the size is greater than
4804 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
4805 (op_by_pieces_d::op_by_pieces_d): Likewise.
4806 (op_by_pieces_d::run): Likewise.
4807 (can_store_by_pieces): Likewise.
4809 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4810 Alan Hayward <alan.hayward@arm.com>
4811 David Sherwood <david.sherwood@arm.com>
4813 * combine.c (extract_left_shift): Add a mode argument and update
4815 (make_compound_operation_int): Change the type of the mode parameter
4816 to scalar_int_mode and update the call to extract_left_shift.
4818 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4819 Alan Hayward <alan.hayward@arm.com>
4820 David Sherwood <david.sherwood@arm.com>
4822 * combine.c (simplify_and_const_int): Change the type of the mode
4823 parameter to scalar_int_mode.
4824 (simplify_and_const_int_1): Likewise. Update recursive call.
4826 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4827 Alan Hayward <alan.hayward@arm.com>
4828 David Sherwood <david.sherwood@arm.com>
4830 * combine.c (simplify_compare_const): Check that the mode is a
4831 scalar_int_mode (rather than VOIDmode) before testing its
4833 (simplify_comparison): Move COMPARISON_P handling out of the
4834 loop and restrict the latter part of the loop to scalar_int_modes.
4835 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
4836 and when considering SUBREG_REGs. Use is_int_mode instead of
4837 checking GET_MODE_CLASS against MODE_INT.
4839 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4840 Alan Hayward <alan.hayward@arm.com>
4841 David Sherwood <david.sherwood@arm.com>
4843 * combine.c (try_widen_shift_mode): Move check for equal modes to...
4844 (simplify_shift_const_1): ...here. Use scalar_int_mode for
4845 shift_unit_mode and for modes involved in scalar shifts.
4847 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4848 Alan Hayward <alan.hayward@arm.com>
4849 David Sherwood <david.sherwood@arm.com>
4851 * combine.c (force_int_to_mode): New function, split out from...
4852 (force_to_mode): ...here. Keep xmode up-to-date and use it
4853 instead of GET_MODE (x).
4855 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4856 Alan Hayward <alan.hayward@arm.com>
4857 David Sherwood <david.sherwood@arm.com>
4859 * optabs-query.h (extraction_insn::struct_mode): Change type to
4860 opt_scalar_int_mode and update comment.
4861 (extraction_insn::field_mode): Change type to scalar_int_mode.
4862 (extraction_insn::pos_mode): Likewise.
4863 * combine.c (make_extraction): Update accordingly.
4864 * optabs-query.c (get_traditional_extraction_insn): Likewise.
4865 (get_optab_extraction_insn): Likewise.
4866 * recog.c (simplify_while_replacing): Likewise.
4867 * expmed.c (narrow_bit_field_mem): Change the type of the mode
4868 parameter to opt_scalar_int_mode.
4870 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4871 Alan Hayward <alan.hayward@arm.com>
4872 David Sherwood <david.sherwood@arm.com>
4874 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
4875 to a scalar_int_mode instead of a machine_mode.
4876 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
4877 (get_best_mode): Return a boolean and use a pointer argument to store
4878 the selected mode. Replace the limit mode parameter with a bit limit.
4879 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
4880 for the values returned by bit_field_mode_iterator::next_mode.
4881 (store_bit_field): Update call to get_best_mode.
4882 (store_fixed_bit_field): Likewise.
4883 (extract_fixed_bit_field): Likewise.
4884 * expr.c (optimize_bitfield_assignment_op): Likewise.
4885 * fold-const.c (optimize_bit_field_compare): Likewise.
4886 (fold_truth_andor_1): Likewise.
4887 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
4888 Update for new type of m_mode.
4889 (get_best_mode): As above.
4891 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4892 Alan Hayward <alan.hayward@arm.com>
4893 David Sherwood <david.sherwood@arm.com>
4895 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
4896 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
4897 (store_bit_field): Check is_a <scalar_int_mode> before calling
4898 strict_volatile_bitfield_p.
4899 (extract_bit_field): Likewise.
4901 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4902 Alan Hayward <alan.hayward@arm.com>
4903 David Sherwood <david.sherwood@arm.com>
4905 * target.def (cstore_mode): Return a scalar_int_mode.
4906 * doc/tm.texi: Regenerate.
4907 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
4908 * targhooks.h (default_cstore_mode): Likewise.
4909 * targhooks.c (default_cstore_mode): Likewise, using a forced
4911 * expmed.c (emit_cstore): Expect the target of the cstore to be
4914 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4915 Alan Hayward <alan.hayward@arm.com>
4916 David Sherwood <david.sherwood@arm.com>
4918 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
4920 (niter_desc): Likewise mode.
4921 (iv_analyze): Add a mode parameter.
4923 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
4924 and change its type to scalar_int_mode.
4925 * loop-iv.c: Update commentary at head of file.
4926 (iv_constant): Pass the mode paraeter before the rtx it describes
4927 and change its type to scalar_int_mode. Remove VOIDmode handling.
4928 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
4929 (iv_extend): Likewise.
4930 (shorten_into_mode): Likewise.
4931 (iv_add): Use scalar_int_mode.
4932 (iv_mult): Likewise.
4933 (iv_shift): Likewise.
4934 (canonicalize_iv_subregs): Likewise.
4935 (get_biv_step_1): Pass the outer_mode parameter before the rtx
4936 it describes and change its mode to scalar_int_mode. Also change
4937 the type of the returned inner_mode to scalar_int_mode.
4938 (get_biv_step): Likewise, turning outer_mode from a pointer
4939 into a direct parameter. Update call to get_biv_step_1.
4940 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
4941 iv_constant and get_biv_step.
4942 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
4943 and change its type to scalar_int_mode. Don't initialise iv->mode
4944 to VOIDmode and remove later checks for its still being VOIDmode.
4945 Update calls to iv_analyze_op and iv_analyze_expr. Check
4946 is_a <scalar_int_mode> when changing the mode under consideration.
4947 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
4948 Update call to iv_analyze_expr.
4949 (iv_analyze_op): Add a mode parameter. Reject subregs whose
4950 inner register is not also a scalar_int_mode. Update call to
4952 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
4953 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
4954 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
4955 separate mode class checks. Update calls to iv_analyze. Remove
4956 fix-up of VOIDmodes after iv_analyze_biv.
4957 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
4958 don't have a scalar_int_mode. Update call to biv_p.
4960 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4961 Alan Hayward <alan.hayward@arm.com>
4962 David Sherwood <david.sherwood@arm.com>
4964 * cfgexpand.c (convert_debug_memory_address): Use
4965 as_a <scalar_int_mode>.
4966 * combine.c (expand_compound_operation): Likewise.
4967 (make_extraction): Likewise.
4968 (change_zero_ext): Likewise.
4969 (simplify_comparison): Likewise.
4970 * cse.c (cse_insn): Likewise.
4971 * dwarf2out.c (minmax_loc_descriptor): Likewise.
4972 (mem_loc_descriptor): Likewise.
4973 (loc_descriptor): Likewise.
4974 * expmed.c (init_expmed_one_mode): Likewise.
4975 (synth_mult): Likewise.
4976 (emit_store_flag_1): Likewise.
4977 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
4978 of a comparison with size.
4979 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
4980 (reduce_to_bit_field_precision): Likewise.
4981 * function.c (expand_function_end): Likewise.
4982 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
4983 * loop-doloop.c (doloop_modify): Likewise.
4984 * optabs.c (expand_binop): Likewise.
4985 (expand_unop): Likewise.
4986 (expand_copysign_absneg): Likewise.
4987 (prepare_cmp_insn): Likewise.
4988 (maybe_legitimize_operand): Likewise.
4989 * recog.c (const_scalar_int_operand): Likewise.
4990 * rtlanal.c (get_address_mode): Likewise.
4991 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4992 (simplify_cond_clz_ctz): Likewise.
4993 * tree-nested.c (get_nl_goto_field): Likewise.
4994 * tree.c (build_vector_type_for_mode): Likewise.
4995 * var-tracking.c (use_narrower_mode): Likewise.
4997 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4998 Alan Hayward <alan.hayward@arm.com>
4999 David Sherwood <david.sherwood@arm.com>
5001 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
5002 * builtins.c (expand_builtin_signbit): Use it.
5003 * cfgexpand.c (expand_debug_expr): Likewise.
5004 * dojump.c (do_jump): Likewise.
5005 (do_compare_and_jump): Likewise.
5006 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
5007 * expmed.c (make_tree): Likewise.
5008 * expr.c (expand_expr_real_2): Likewise.
5009 (expand_expr_real_1): Likewise.
5010 (try_casesi): Likewise.
5011 * fold-const-call.c (fold_const_call_ss): Likewise.
5012 * fold-const.c (unextend): Likewise.
5013 (extract_muldiv_1): Likewise.
5014 (fold_single_bit_test): Likewise.
5015 (native_encode_int): Likewise.
5016 (native_encode_string): Likewise.
5017 (native_interpret_int): Likewise.
5018 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
5019 * internal-fn.c (expand_addsub_overflow): Likewise.
5020 (expand_neg_overflow): Likewise.
5021 (expand_mul_overflow): Likewise.
5022 (expand_arith_overflow): Likewise.
5023 * match.pd: Likewise.
5024 * stor-layout.c (layout_type): Likewise.
5025 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
5026 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
5027 * tree-ssanames.c (get_range_info): Likewise.
5028 * tree-switch-conversion.c (array_value_type) Likewise.
5029 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
5030 (vect_recog_divmod_pattern): Likewise.
5031 (vect_recog_mixed_size_cond_pattern): Likewise.
5032 * tree-vrp.c (extract_range_basic): Likewise.
5033 (simplify_float_conversion_using_ranges): Likewise.
5034 * tree.c (int_fits_type_p): Likewise.
5035 * ubsan.c (instrument_bool_enum_load): Likewise.
5036 * varasm.c (mergeable_string_section): Likewise.
5037 (narrowing_initializer_constant_valid_p): Likewise.
5038 (output_constant): Likewise.
5040 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5041 Alan Hayward <alan.hayward@arm.com>
5042 David Sherwood <david.sherwood@arm.com>
5044 * machmode.h (NARROWEST_INT_MODE): New macro.
5045 * expr.c (alignment_for_piecewise_move): Use it instead of
5046 GET_CLASS_NARROWEST_MODE (MODE_INT).
5047 (push_block): Likewise.
5048 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
5050 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
5052 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5053 Alan Hayward <alan.hayward@arm.com>
5054 David Sherwood <david.sherwood@arm.com>
5056 * postreload.c (move2add_valid_value_p): Change the type of the
5057 mode parameter to scalar_int_mode.
5058 (move2add_use_add2_insn): Add a mode parameter and use it instead
5060 (move2add_use_add3_insn): Likewise.
5061 (reload_cse_move2add): Update accordingly.
5063 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5064 Alan Hayward <alan.hayward@arm.com>
5065 David Sherwood <david.sherwood@arm.com>
5067 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
5069 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
5070 * optabs.c (expand_unop): Likewise.
5072 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5073 Alan Hayward <alan.hayward@arm.com>
5074 David Sherwood <david.sherwood@arm.com>
5076 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
5077 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
5078 (popcount_loc_descriptor): Likewise.
5079 (bswap_loc_descriptor): Likewise.
5080 (rotate_loc_descriptor): Likewise.
5081 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
5082 calling the functions above.
5084 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5085 Alan Hayward <alan.hayward@arm.com>
5086 David Sherwood <david.sherwood@arm.com>
5088 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
5090 (try_combine): Likewise.
5091 (simplify_if_then_else): Likewise.
5092 * cse.c (cse_insn): Likewise.
5093 * dwarf2out.c (mem_loc_descriptor): Likewise.
5094 * emit-rtl.c (gen_lowpart_common): Likewise.
5095 * simplify-rtx.c (simplify_truncation): Likewise.
5096 (simplify_binary_operation_1): Likewise.
5097 (simplify_const_relational_operation): Likewise.
5098 (simplify_ternary_operation): Likewise.
5099 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
5101 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5102 Alan Hayward <alan.hayward@arm.com>
5103 David Sherwood <david.sherwood@arm.com>
5105 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
5106 * reload.c (push_reload): Likewise.
5107 (find_reloads): Likewise.
5109 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5110 Alan Hayward <alan.hayward@arm.com>
5111 David Sherwood <david.sherwood@arm.com>
5113 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
5114 (make_compound_operation_int): Likewise.
5115 (change_zero_ext): Likewise.
5116 * expr.c (convert_move): Likewise.
5117 (convert_modes): Likewise.
5118 * fwprop.c (forward_propagate_subreg): Likewise.
5119 * loop-iv.c (get_biv_step_1): Likewise.
5120 * optabs.c (widen_operand): Likewise.
5121 * postreload.c (move2add_valid_value_p): Likewise.
5122 * recog.c (simplify_while_replacing): Likewise.
5123 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5124 (simplify_binary_operation_1): Likewise. Remove redundant
5125 mode equality check.
5127 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5128 Alan Hayward <alan.hayward@arm.com>
5129 David Sherwood <david.sherwood@arm.com>
5131 * combine.c (combine_simplify_rtx): Add checks for
5132 is_a <scalar_int_mode>.
5133 (simplify_if_then_else): Likewise.
5134 (make_field_assignment): Likewise.
5135 (simplify_comparison): Likewise.
5136 * ifcvt.c (noce_try_bitop): Likewise.
5137 * loop-invariant.c (canonicalize_address_mult): Likewise.
5138 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5140 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5141 Alan Hayward <alan.hayward@arm.com>
5142 David Sherwood <david.sherwood@arm.com>
5144 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
5145 is_a <scalar_int_mode> instead of != BLKmode.
5147 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5148 Alan Hayward <alan.hayward@arm.com>
5149 David Sherwood <david.sherwood@arm.com>
5151 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
5152 instead of != VOIDmode.
5153 * combine.c (if_then_else_cond): Likewise.
5154 (change_zero_ext): Likewise.
5155 * dwarf2out.c (mem_loc_descriptor): Likewise.
5156 (loc_descriptor): Likewise.
5157 * rtlanal.c (canonicalize_condition): Likewise.
5158 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
5160 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5161 Alan Hayward <alan.hayward@arm.com>
5162 David Sherwood <david.sherwood@arm.com>
5164 * simplify-rtx.c (simplify_binary_operation_1): Use
5165 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
5167 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5168 Alan Hayward <alan.hayward@arm.com>
5169 David Sherwood <david.sherwood@arm.com>
5171 * wide-int.h (int_traits<unsigned char>) New class.
5172 (int_traits<unsigned short>) Likewise.
5173 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
5174 Use GET_MODE_UNIT_PRECISION and remove redundant test for
5176 * combine.c (set_nonzero_bits_and_sign_copies): Use
5177 is_a <scalar_int_mode>.
5178 (find_split_point): Likewise.
5179 (combine_simplify_rtx): Likewise.
5180 (simplify_logical): Likewise.
5181 (expand_compound_operation): Likewise.
5182 (expand_field_assignment): Likewise.
5183 (make_compound_operation): Likewise.
5184 (extended_count): Likewise.
5185 (change_zero_ext): Likewise.
5186 (simplify_comparison): Likewise.
5187 * dwarf2out.c (scompare_loc_descriptor): Likewise.
5188 (ucompare_loc_descriptor): Likewise.
5189 (minmax_loc_descriptor): Likewise.
5190 (mem_loc_descriptor): Likewise.
5191 (loc_descriptor): Likewise.
5192 * expmed.c (init_expmed_one_mode): Likewise.
5193 * lra-constraints.c (lra_constraint_offset): Likewise.
5194 * optabs.c (prepare_libcall_arg): Likewise.
5195 * postreload.c (move2add_note_store): Likewise.
5196 * reload.c (operands_match_p): Likewise.
5197 * rtl.h (load_extend_op): Likewise.
5198 * rtlhooks.c (gen_lowpart_general): Likewise.
5199 * simplify-rtx.c (simplify_truncation): Likewise.
5200 (simplify_unary_operation_1): Likewise.
5201 (simplify_binary_operation_1): Likewise.
5202 (simplify_const_binary_operation): Likewise.
5203 (simplify_const_relational_operation): Likewise.
5204 (simplify_subreg): Likewise.
5205 * stor-layout.c (bitwise_mode_for_mode): Likewise.
5206 * var-tracking.c (adjust_mems): Likewise.
5207 (prepare_call_arguments): Likewise.
5209 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5210 Alan Hayward <alan.hayward@arm.com>
5211 David Sherwood <david.sherwood@arm.com>
5213 * machmode.h (is_int_mode): New fuction.
5214 * combine.c (find_split_point): Use it.
5215 (combine_simplify_rtx): Likewise.
5216 (simplify_if_then_else): Likewise.
5217 (simplify_set): Likewise.
5218 (simplify_shift_const_1): Likewise.
5219 (simplify_comparison): Likewise.
5220 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
5221 * cse.c (notreg_cost): Likewise.
5222 (cse_insn): Likewise.
5223 * cselib.c (cselib_lookup_1): Likewise.
5224 * dojump.c (do_jump_1): Likewise.
5225 (do_compare_rtx_and_jump): Likewise.
5226 * dse.c (get_call_args): Likewise.
5227 * dwarf2out.c (rtl_for_decl_init): Likewise.
5228 (native_encode_initializer): Likewise.
5229 * expmed.c (emit_store_flag_1): Likewise.
5230 (emit_store_flag): Likewise.
5231 * expr.c (convert_modes): Likewise.
5232 (store_field): Likewise.
5233 (expand_expr_real_1): Likewise.
5234 * fold-const.c (fold_read_from_constant_string): Likewise.
5235 * gimple-ssa-sprintf.c (get_format_string): Likewise.
5236 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
5237 * optabs.c (expand_binop): Likewise.
5238 (expand_unop): Likewise.
5239 (expand_abs_nojump): Likewise.
5240 (expand_one_cmpl_abs_nojump): Likewise.
5241 * simplify-rtx.c (mode_signbit_p): Likewise.
5242 (val_signbit_p): Likewise.
5243 (val_signbit_known_set_p): Likewise.
5244 (val_signbit_known_clear_p): Likewise.
5245 (simplify_relational_operation_1): Likewise.
5246 * tree.c (vector_type_mode): Likewise.
5248 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5249 Alan Hayward <alan.hayward@arm.com>
5250 David Sherwood <david.sherwood@arm.com>
5252 * machmode.h (smallest_mode_for_size): Fix formatting.
5253 (smallest_int_mode_for_size): New function.
5254 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
5255 instead of smallest_mode_for_size.
5256 * combine.c (make_extraction): Likewise.
5257 * config/arc/arc.c (arc_expand_movmem): Likewise.
5258 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
5259 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
5260 * config/s390/s390.c (s390_expand_insv): Likewise.
5261 * config/sparc/sparc.c (assign_int_registers): Likewise.
5262 * config/spu/spu.c (spu_function_value): Likewise.
5263 (spu_function_arg): Likewise.
5264 * coverage.c (get_gcov_type): Likewise.
5265 (get_gcov_unsigned_t): Likewise.
5266 * dse.c (find_shift_sequence): Likewise.
5267 * expmed.c (store_bit_field_1): Likewise.
5268 * expr.c (convert_move): Likewise.
5269 (store_field): Likewise.
5270 * internal-fn.c (expand_arith_overflow): Likewise.
5271 * optabs-query.c (get_best_extraction_insn): Likewise.
5272 * optabs.c (expand_twoval_binop_libfunc): Likewise.
5273 * stor-layout.c (layout_type): Likewise.
5274 (initialize_sizetypes): Likewise.
5275 * targhooks.c (default_get_mask_mode): Likewise.
5276 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
5278 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5279 Alan Hayward <alan.hayward@arm.com>
5280 David Sherwood <david.sherwood@arm.com>
5282 * machmode.h (opt_mode::else_blk): New function.
5283 (int_mode_for_mode): Declare.
5284 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
5285 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
5287 * cfgexpand.c (expand_debug_expr): Likewise.
5288 * combine.c (gen_lowpart_or_truncate): Likewise.
5289 (gen_lowpart_for_combine): Likewise.
5290 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
5291 * config/avr/avr.c (avr_to_int_mode): Likewise.
5292 (avr_out_plus_1): Likewise.
5293 (avr_out_plus): Likewise.
5294 (avr_out_round): Likewise.
5295 * config/i386/i386.c (ix86_split_to_parts): Likewise.
5296 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
5297 (s390_expand_vcond): Likewise.
5298 * config/spu/spu.c (spu_split_immediate): Likewise.
5299 (spu_expand_mov): Likewise.
5300 * dse.c (get_stored_val): Likewise.
5301 * expmed.c (store_bit_field_1): Likewise.
5302 (convert_extracted_bit_field): Use int_mode_for_mode instead of
5304 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
5305 (extract_low_bits): Likewise.
5306 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
5307 handling rather than repeating the check.
5308 (emit_group_store): Likewise.
5309 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
5310 * optabs.c (expand_absneg_bit): Likewise.
5311 (expand_copysign_absneg): Likewise.
5312 (expand_copysign_bit): Likewise.
5313 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
5314 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
5315 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
5316 * var-tracking.c (prepare_call_arguments): Likewise.
5317 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
5318 int_mode_for_mode instead of mode_for_size.
5319 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
5321 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5322 Alan Hayward <alan.hayward@arm.com>
5323 David Sherwood <david.sherwood@arm.com>
5325 * machmode.h (int_mode_for_size): New function.
5326 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
5327 instead of mode_for_size.
5328 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
5330 * combine.c (expand_field_assignment): Use int_mode_for_size
5331 instead of mode_for_size.
5332 (make_extraction): Likewise.
5333 (simplify_shift_const_1): Likewise.
5334 (simplify_comparison): Likewise.
5335 * dojump.c (do_jump): Likewise.
5336 * dwarf2out.c (mem_loc_descriptor): Likewise.
5337 * emit-rtl.c (init_derived_machine_modes): Likewise.
5338 * expmed.c (flip_storage_order): Likewise.
5339 (convert_extracted_bit_field): Likewise.
5340 * expr.c (copy_blkmode_from_reg): Likewise.
5341 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
5342 * internal-fn.c (expand_mul_overflow): Likewise.
5343 * lower-subreg.c (simple_move): Likewise.
5344 * optabs-libfuncs.c (init_optabs): Likewise.
5345 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5346 * tree.c (vector_type_mode): Likewise.
5347 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
5348 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
5349 * tree-vect-generic.c (expand_vector_parallel): Likewise.
5350 * tree-vect-stmts.c (vectorizable_load): Likewise.
5351 (vectorizable_store): Likewise.
5353 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5354 Alan Hayward <alan.hayward@arm.com>
5355 David Sherwood <david.sherwood@arm.com>
5357 * coretypes.h (pod_mode): New type.
5358 (scalar_int_mode_pod): New typedef.
5359 * machmode.h (pod_mode): New class.
5360 (int_n_data_t::m): Change type to scalar_int_mode_pod.
5361 * genmodes.c (emit_mode_int_n): Update accordingly.
5362 * lower-subreg.h (target_lower_subreg): Change type to
5363 scalar_int_mode_pod.
5364 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
5365 scalar_int_mode_pod.
5367 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5368 Alan Hayward <alan.hayward@arm.com>
5369 David Sherwood <david.sherwood@arm.com>
5371 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
5372 machine_mode to scalar_int_mode.
5373 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
5374 (rs6000_option_override_internal): Remove cast to int.
5375 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
5376 machine_mode to scalar_int_mode.
5377 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
5378 (rs6000_option_override_internal): Remove cast to int.
5379 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
5380 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
5382 * config/s390/s390.c (s390_expand_builtin): Likewise.
5383 * coretypes.h (scalar_int_mode): New type.
5384 (opt_scalar_int_mode): New typedef.
5385 * machmode.h (scalar_int_mode): New class.
5386 (scalar_int_mode::includes_p): New function.
5387 (byte_mode): Change type to scalar_int_mode.
5388 (word_mode): Likewise.
5389 (ptr_mode): Likewise.
5390 * emit-rtl.c (byte_mode): Likewise.
5391 (word_mode): Likewise.
5392 (ptr_mode): Likewise.
5393 (init_derived_machine_modes): Update accordingly.
5394 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
5395 and MODE_PARTIAL_INT.
5396 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
5397 opt_scalar_int_mode.
5399 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5400 Alan Hayward <alan.hayward@arm.com>
5401 David Sherwood <david.sherwood@arm.com>
5403 * target.def (libgcc_floating_mode_supported_p): Take a
5405 * doc/tm.texi: Regenerate.
5406 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
5408 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
5409 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
5412 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5413 Alan Hayward <alan.hayward@arm.com>
5414 David Sherwood <david.sherwood@arm.com>
5416 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
5417 * doc/tm.texi: Regenerate.
5418 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
5419 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
5420 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
5421 * targhooks.h (default_floatn_mode): Likewise.
5422 * targhooks.c (default_floatn_mode): Likewise.
5423 * tree.c (build_common_tree_nodes): Update accordingly.
5425 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5426 Alan Hayward <alan.hayward@arm.com>
5427 David Sherwood <david.sherwood@arm.com>
5429 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
5430 (mode_iterator::iterate_p): Likewise.
5431 (mode_iterator::get_wider): Likewise.
5432 * expr.c (init_expr_target): Use opt_scalar_float_mode.
5434 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5435 Alan Hayward <alan.hayward@arm.com>
5436 David Sherwood <david.sherwood@arm.com>
5438 * coretypes.h (opt_scalar_float_mode): New typedef.
5439 * machmode.h (float_mode_for_size): New function.
5440 * emit-rtl.c (double_mode): Delete.
5441 (init_emit_once): Use float_mode_for_size.
5442 * stor-layout.c (layout_type): Likewise.
5443 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
5445 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5446 Alan Hayward <alan.hayward@arm.com>
5447 David Sherwood <david.sherwood@arm.com>
5449 * output.h (assemble_real): Take a scalar_float_mode.
5450 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
5451 * config/arm/arm.md (consttable_4): Likewise.
5452 (consttable_8): Likewise.
5453 (consttable_16): Likewise.
5454 * config/mips/mips.md (consttable_float): Likewise.
5455 * config/s390/s390.c (s390_output_pool_entry): Likewise.
5456 * varasm.c (assemble_real): Take a scalar_float_mode.
5457 (output_constant_pool_2): Update accordingly.
5458 (output_constant): Likewise.
5460 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5461 Alan Hayward <alan.hayward@arm.com>
5462 David Sherwood <david.sherwood@arm.com>
5464 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
5465 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
5466 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
5467 (native_encode_real): Likewise.
5468 (native_interpret_real): Likewise.
5469 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
5470 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
5472 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5473 Alan Hayward <alan.hayward@arm.com>
5474 David Sherwood <david.sherwood@arm.com>
5476 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
5477 <scalar_float_mode>. Simplify.
5478 (gen_extend_conv_libfunc): Likewise.
5480 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5481 Alan Hayward <alan.hayward@arm.com>
5482 David Sherwood <david.sherwood@arm.com>
5484 * coretypes.h (scalar_float_mode): New type.
5485 * machmode.h (mode_traits::from_int): Use machine_mode if
5486 USE_ENUM_MODES is defined.
5487 (is_a): New function.
5489 (dyn_cast): Likewise.
5490 (scalar_float_mode): New class.
5491 (scalar_float_mode::includes_p): New function.
5492 (is_float_mode): Likewise.
5493 * gdbhooks.py (MachineModePrinter): New class.
5494 (build_pretty_printer): Use it for scalar_float_mode.
5495 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
5496 (format_helper::format_helper): Turn into a template.
5497 * genmodes.c (get_mode_class): New function.
5498 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
5499 or machine_mode if none.
5500 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
5501 as_a <scalar_float_mode>.
5502 * dwarf2out.c (mem_loc_descriptor): Likewise.
5503 (insert_float): Likewise.
5504 (add_const_value_attribute): Likewise.
5505 * simplify-rtx.c (simplify_immed_subreg): Likewise.
5506 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
5507 (expand_unop): Update accordingly.
5508 (expand_abs_nojump): Likewise.
5509 (expand_copysign_absneg): Take a scalar_float_mode.
5510 (expand_copysign_bit): Likewise.
5511 (expand_copysign): Update accordingly.
5513 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5514 Alan Hayward <alan.hayward@arm.com>
5515 David Sherwood <david.sherwood@arm.com>
5517 * coretypes.h (opt_mode): New class.
5518 * machmode.h (opt_mode): Likewise.
5519 (opt_mode::else_void): New function.
5520 (opt_mode::require): Likewise.
5521 (opt_mode::exists): Likewise.
5522 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
5523 (GET_MODE_2XWIDER_MODE): Likewise.
5524 (mode_iterator::get_wider): Update accordingly.
5525 (mode_iterator::get_2xwider): Likewise.
5526 (mode_iterator::get_known_wider): Likewise, turning into a template.
5527 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
5528 forcing a wider mode to exist.
5529 * config/cr16/cr16.h (LONG_REG_P): Likewise.
5530 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
5531 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
5532 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
5533 * lower-subreg.c (init_lower_subreg): Likewise.
5534 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
5535 on the final iteration.
5536 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
5537 a wider mode exists before asking for a move pattern.
5538 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
5539 forcing a wider mode to exist.
5540 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
5541 returning false if no such mode exists.
5542 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
5543 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
5544 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
5545 Avoid checking for a MODE_INT if we already know the mode is not a
5547 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
5548 forcing a wider mode to exist.
5549 (expmed_mult_highpart_optab): Likewise.
5550 (expmed_mult_highpart): Likewise.
5551 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
5553 * lto-streamer-in.c (lto_input_mode_table): Likewise.
5554 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
5555 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
5556 * internal-fn.c (expand_mul_overflow): Update use of
5557 GET_MODE_2XWIDER_MODE.
5558 * omp-low.c (omp_clause_aligned_alignment): Likewise.
5559 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
5560 GET_MODE_WIDER_MODE.
5561 (convert_plusminus_to_widen): Likewise.
5562 * tree-switch-conversion.c (array_value_type): Likewise.
5563 * var-tracking.c (emit_note_insn_var_location): Likewise.
5564 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
5565 Return false inside rather than outside the loop if no wider mode
5567 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
5568 and GET_MODE_2XWIDER_MODE
5569 (can_compare_p): Use else_void.
5570 * gdbhooks.py (OptMachineModePrinter): New class.
5571 (build_pretty_printer): Use it for opt_mode.
5573 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5574 Alan Hayward <alan.hayward@arm.com>
5575 David Sherwood <david.sherwood@arm.com>
5577 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
5578 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
5580 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5581 Alan Hayward <alan.hayward@arm.com>
5582 David Sherwood <david.sherwood@arm.com>
5584 * machmode.h (mode_traits): New structure.
5585 (get_narrowest_mode): New function.
5586 (mode_iterator::start): Likewise.
5587 (mode_iterator::iterate_p): Likewise.
5588 (mode_iterator::get_wider): Likewise.
5589 (mode_iterator::get_known_wider): Likewise.
5590 (mode_iterator::get_2xwider): Likewise.
5591 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
5592 (FOR_EACH_MODE): Likewise.
5593 (FOR_EACH_MODE_FROM): Likewise.
5594 (FOR_EACH_MODE_UNTIL): Likewise.
5595 (FOR_EACH_WIDER_MODE): Likewise.
5596 (FOR_EACH_2XWIDER_MODE): Likewise.
5597 * builtins.c (expand_builtin_strlen): Use new mode iterators.
5598 * combine.c (simplify_comparison): Likewise
5599 * config/i386/i386.c (type_natural_mode): Likewise.
5600 * cse.c (cse_insn): Likewise.
5601 * dse.c (find_shift_sequence): Likewise.
5602 * emit-rtl.c (init_derived_machine_modes): Likewise.
5603 (init_emit_once): Likewise.
5604 * explow.c (hard_function_value): Likewise.
5605 * expmed.c (extract_fixed_bit_field_1): Likewise.
5606 (extract_bit_field_1): Likewise.
5607 (expand_divmod): Likewise.
5608 (emit_store_flag_1): Likewise.
5609 * expr.c (init_expr_target): Likewise.
5610 (convert_move): Likewise.
5611 (alignment_for_piecewise_move): Likewise.
5612 (widest_int_mode_for_size): Likewise.
5613 (emit_block_move_via_movmem): Likewise.
5614 (copy_blkmode_to_reg): Likewise.
5615 (set_storage_via_setmem): Likewise.
5616 (compress_float_constant): Likewise.
5617 * omp-low.c (omp_clause_aligned_alignment): Likewise.
5618 * optabs-query.c (get_best_extraction_insn): Likewise.
5619 * optabs.c (expand_binop): Likewise.
5620 (expand_twoval_unop): Likewise.
5621 (expand_twoval_binop): Likewise.
5622 (widen_leading): Likewise.
5623 (widen_bswap): Likewise.
5624 (expand_parity): Likewise.
5625 (expand_unop): Likewise.
5626 (prepare_cmp_insn): Likewise.
5627 (prepare_float_lib_cmp): Likewise.
5628 (expand_float): Likewise.
5629 (expand_fix): Likewise.
5630 (expand_sfix_optab): Likewise.
5631 * postreload.c (move2add_use_add2_insn): Likewise.
5632 * reg-stack.c (reg_to_stack): Likewise.
5633 * reginfo.c (choose_hard_reg_mode): Likewise.
5634 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
5635 * stor-layout.c (mode_for_size): Likewise.
5636 (smallest_mode_for_size): Likewise.
5637 (mode_for_vector): Likewise.
5638 (finish_bitfield_representative): Likewise.
5639 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
5640 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
5641 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
5642 * var-tracking.c (prepare_call_arguments): Likewise.
5644 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5645 Alan Hayward <alan.hayward@arm.com>
5646 David Sherwood <david.sherwood@arm.com>
5648 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
5649 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
5650 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
5651 * machmode.h (mode_size): Move earlier in file.
5652 (mode_precision): Likewise.
5653 (mode_inner): Likewise.
5654 (mode_nunits): Likewise.
5655 (mode_unit_size): Likewise.
5656 (unit_unit_precision): Likewise.
5657 (mode_wider): Likewise.
5658 (mode_2xwider): Likewise.
5659 (machine_mode): New class.
5660 (mode_to_bytes): New function.
5661 (mode_to_bits): Likewise.
5662 (mode_to_precision): Likewise.
5663 (mode_to_inner): Likewise.
5664 (mode_to_unit_size): Likewise.
5665 (mode_to_unit_precision): Likewise.
5666 (mode_to_nunits): Likewise.
5667 (GET_MODE_SIZE): Use mode_to_bytes.
5668 (GET_MODE_BITSIZE): Use mode_to_bits.
5669 (GET_MODE_PRECISION): Use mode_to_precision.
5670 (GET_MODE_INNER): Use mode_to_inner.
5671 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
5672 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
5673 (GET_MODE_NUNITS): Use mode_to_nunits.
5674 * system.h (ALWAYS_INLINE): New macro.
5675 * config/powerpcspe/powerpcspe-c.c
5676 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
5677 int for arg1_mode and arg2_mode.
5679 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5680 Alan Hayward <alan.hayward@arm.com>
5681 David Sherwood <david.sherwood@arm.com>
5683 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
5684 Prefix mode names with E_ in case statements.
5685 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
5686 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
5687 (aarch64_split_simd_move): Likewise.
5688 (aarch64_gen_storewb_pair): Likewise.
5689 (aarch64_gen_loadwb_pair): Likewise.
5690 (aarch64_gen_store_pair): Likewise.
5691 (aarch64_gen_load_pair): Likewise.
5692 (aarch64_get_condition_code_1): Likewise.
5693 (aarch64_constant_pool_reload_icode): Likewise.
5694 (get_rsqrte_type): Likewise.
5695 (get_rsqrts_type): Likewise.
5696 (get_recpe_type): Likewise.
5697 (get_recps_type): Likewise.
5698 (aarch64_gimplify_va_arg_expr): Likewise.
5699 (aarch64_simd_container_mode): Likewise.
5700 (aarch64_emit_load_exclusive): Likewise.
5701 (aarch64_emit_store_exclusive): Likewise.
5702 (aarch64_expand_compare_and_swap): Likewise.
5703 (aarch64_gen_atomic_cas): Likewise.
5704 (aarch64_emit_bic): Likewise.
5705 (aarch64_emit_atomic_swap): Likewise.
5706 (aarch64_emit_atomic_load_op): Likewise.
5707 (aarch64_evpc_trn): Likewise.
5708 (aarch64_evpc_uzp): Likewise.
5709 (aarch64_evpc_zip): Likewise.
5710 (aarch64_evpc_ext): Likewise.
5711 (aarch64_evpc_rev): Likewise.
5712 (aarch64_evpc_dup): Likewise.
5713 (aarch64_gen_ccmp_first): Likewise.
5714 (aarch64_gen_ccmp_next): Likewise.
5715 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
5716 (alpha_emit_xfloating_libcall): Likewise.
5717 (emit_insxl): Likewise.
5718 (alpha_arg_type): Likewise.
5719 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
5720 (arc_preferred_simd_mode): Likewise.
5721 (arc_secondary_reload): Likewise.
5722 (get_arc_condition_code): Likewise.
5723 (arc_print_operand): Likewise.
5724 (arc_legitimate_constant_p): Likewise.
5725 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
5726 * config/arc/arc.md (casesi_load): Likewise.
5727 (casesi_compact_jump): Likewise.
5728 * config/arc/predicates.md (proper_comparison_operator): Likewise.
5729 (cc_use_register): Likewise.
5730 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
5731 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
5732 (arm_init_iwmmxt_builtins): Likewise.
5733 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
5734 (neon_expand_vector_init): Likewise.
5735 (arm_attr_length_move_neon): Likewise.
5736 (maybe_get_arm_condition_code): Likewise.
5737 (arm_emit_vector_const): Likewise.
5738 (arm_preferred_simd_mode): Likewise.
5739 (arm_output_iwmmxt_tinsr): Likewise.
5740 (thumb1_output_casesi): Likewise.
5741 (thumb2_output_casesi): Likewise.
5742 (arm_emit_load_exclusive): Likewise.
5743 (arm_emit_store_exclusive): Likewise.
5744 (arm_expand_compare_and_swap): Likewise.
5745 (arm_evpc_neon_vuzp): Likewise.
5746 (arm_evpc_neon_vzip): Likewise.
5747 (arm_evpc_neon_vrev): Likewise.
5748 (arm_evpc_neon_vtrn): Likewise.
5749 (arm_evpc_neon_vext): Likewise.
5750 (arm_validize_comparison): Likewise.
5751 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
5752 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
5753 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
5754 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
5755 (c6x_preferred_simd_mode): Likewise.
5756 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
5757 (epiphany_rtx_costs): Likewise.
5758 * config/epiphany/predicates.md (proper_comparison_operator):
5760 * config/frv/frv.c (condexec_memory_operand): Likewise.
5761 (frv_emit_move): Likewise.
5762 (output_move_single): Likewise.
5763 (output_condmove_single): Likewise.
5764 (frv_hard_regno_mode_ok): Likewise.
5765 (frv_matching_accg_mode): Likewise.
5766 * config/h8300/h8300.c (split_adds_subs): Likewise.
5767 (h8300_rtx_costs): Likewise.
5768 (h8300_print_operand): Likewise.
5769 (compute_mov_length): Likewise.
5770 (output_logical_op): Likewise.
5771 (compute_logical_op_length): Likewise.
5772 (compute_logical_op_cc): Likewise.
5773 (h8300_shift_needs_scratch_p): Likewise.
5774 (output_a_shift): Likewise.
5775 (compute_a_shift_length): Likewise.
5776 (compute_a_shift_cc): Likewise.
5777 (expand_a_rotate): Likewise.
5778 (output_a_rotate): Likewise.
5779 * config/i386/i386.c (classify_argument): Likewise.
5780 (function_arg_advance_32): Likewise.
5781 (function_arg_32): Likewise.
5782 (function_arg_64): Likewise.
5783 (function_value_64): Likewise.
5784 (ix86_gimplify_va_arg): Likewise.
5785 (ix86_legitimate_constant_p): Likewise.
5786 (put_condition_code): Likewise.
5787 (split_double_mode): Likewise.
5788 (ix86_avx256_split_vector_move_misalign): Likewise.
5789 (ix86_expand_vector_logical_operator): Likewise.
5790 (ix86_split_idivmod): Likewise.
5791 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
5792 (ix86_build_const_vector): Likewise.
5793 (ix86_build_signbit_mask): Likewise.
5794 (ix86_match_ccmode): Likewise.
5795 (ix86_cc_modes_compatible): Likewise.
5796 (ix86_expand_branch): Likewise.
5797 (ix86_expand_sse_cmp): Likewise.
5798 (ix86_expand_sse_movcc): Likewise.
5799 (ix86_expand_int_sse_cmp): Likewise.
5800 (ix86_expand_vec_perm_vpermi2): Likewise.
5801 (ix86_expand_vec_perm): Likewise.
5802 (ix86_expand_sse_unpack): Likewise.
5803 (ix86_expand_int_addcc): Likewise.
5804 (ix86_split_to_parts): Likewise.
5805 (ix86_vectorize_builtin_gather): Likewise.
5806 (ix86_vectorize_builtin_scatter): Likewise.
5807 (avx_vpermilp_parallel): Likewise.
5808 (inline_memory_move_cost): Likewise.
5809 (ix86_tieable_integer_mode_p): Likewise.
5810 (x86_maybe_negate_const_int): Likewise.
5811 (ix86_expand_vector_init_duplicate): Likewise.
5812 (ix86_expand_vector_init_one_nonzero): Likewise.
5813 (ix86_expand_vector_init_one_var): Likewise.
5814 (ix86_expand_vector_init_concat): Likewise.
5815 (ix86_expand_vector_init_interleave): Likewise.
5816 (ix86_expand_vector_init_general): Likewise.
5817 (ix86_expand_vector_set): Likewise.
5818 (ix86_expand_vector_extract): Likewise.
5819 (emit_reduc_half): Likewise.
5820 (ix86_emit_i387_round): Likewise.
5821 (ix86_mangle_type): Likewise.
5822 (ix86_expand_round_sse4): Likewise.
5823 (expand_vec_perm_blend): Likewise.
5824 (canonicalize_vector_int_perm): Likewise.
5825 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
5826 (expand_vec_perm_1): Likewise.
5827 (expand_vec_perm_interleave3): Likewise.
5828 (expand_vec_perm_even_odd_pack): Likewise.
5829 (expand_vec_perm_even_odd_1): Likewise.
5830 (expand_vec_perm_broadcast_1): Likewise.
5831 (ix86_vectorize_vec_perm_const_ok): Likewise.
5832 (ix86_expand_vecop_qihi): Likewise.
5833 (ix86_expand_mul_widen_hilo): Likewise.
5834 (ix86_expand_sse2_abs): Likewise.
5835 (ix86_expand_pextr): Likewise.
5836 (ix86_expand_pinsr): Likewise.
5837 (ix86_preferred_simd_mode): Likewise.
5838 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
5839 * config/i386/sse.md (*andnot<mode>3): Likewise.
5840 (<mask_codefor><code><mode>3<mask_name>): Likewise.
5841 (*<code><mode>3): Likewise.
5842 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
5843 (ia64_expand_atomic_op): Likewise.
5844 (ia64_arg_type): Likewise.
5845 (ia64_mode_to_int): Likewise.
5846 (ia64_scalar_mode_supported_p): Likewise.
5847 (ia64_vector_mode_supported_p): Likewise.
5848 (expand_vec_perm_broadcast): Likewise.
5849 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
5850 (iq2000_function_arg_advance): Likewise.
5851 (iq2000_function_arg): Likewise.
5852 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
5853 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
5854 (m68k_libcall_value): Likewise.
5855 (m68k_function_value): Likewise.
5856 (sched_attr_op_type): Likewise.
5857 * config/mcore/mcore.c (mcore_output_move): Likewise.
5858 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
5860 (microblaze_function_arg): Likewise.
5861 * config/mips/mips.c (mips16_build_call_stub): Likewise.
5862 (mips_print_operand): Likewise.
5863 (mips_mode_ok_for_mov_fmt_p): Likewise.
5864 (mips_vector_mode_supported_p): Likewise.
5865 (mips_preferred_simd_mode): Likewise.
5866 (mips_expand_vpc_loongson_even_odd): Likewise.
5867 (mips_expand_vec_unpack): Likewise.
5868 (mips_expand_vi_broadcast): Likewise.
5869 (mips_expand_vector_init): Likewise.
5870 (mips_expand_vec_reduc): Likewise.
5871 (mips_expand_msa_cmp): Likewise.
5872 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
5873 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
5874 (cc_flags_for_mode): Likewise.
5875 * config/msp430/msp430.c (msp430_print_operand): Likewise.
5876 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
5877 (nds32_output_casesi_pc_relative): Likewise.
5878 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
5879 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
5880 (nvptx_gen_unpack): Likewise.
5881 (nvptx_gen_pack): Likewise.
5882 (nvptx_gen_shuffle): Likewise.
5883 (nvptx_gen_wcast): Likewise.
5884 (nvptx_preferred_simd_mode): Likewise.
5885 * config/pa/pa.c (pa_secondary_reload): Likewise.
5886 * config/pa/predicates.md (base14_operand): Likewise.
5887 * config/powerpcspe/powerpcspe-c.c
5888 (altivec_resolve_overloaded_builtin): Likewise.
5889 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
5891 (rs6000_preferred_simd_mode): Likewise.
5892 (output_vec_const_move): Likewise.
5893 (rs6000_expand_vector_extract): Likewise.
5894 (rs6000_split_vec_extract_var): Likewise.
5895 (reg_offset_addressing_ok_p): Likewise.
5896 (rs6000_legitimate_offset_address_p): Likewise.
5897 (rs6000_legitimize_address): Likewise.
5898 (rs6000_emit_set_const): Likewise.
5899 (rs6000_const_vec): Likewise.
5900 (rs6000_emit_move): Likewise.
5901 (spe_build_register_parallel): Likewise.
5902 (rs6000_darwin64_record_arg_recurse): Likewise.
5903 (swap_selector_for_mode): Likewise.
5904 (spe_init_builtins): Likewise.
5905 (paired_init_builtins): Likewise.
5906 (altivec_init_builtins): Likewise.
5907 (do_load_for_compare): Likewise.
5908 (rs6000_generate_compare): Likewise.
5909 (rs6000_expand_float128_convert): Likewise.
5910 (emit_load_locked): Likewise.
5911 (emit_store_conditional): Likewise.
5912 (rs6000_output_function_epilogue): Likewise.
5913 (rs6000_handle_altivec_attribute): Likewise.
5914 (rs6000_function_value): Likewise.
5915 (emit_fusion_gpr_load): Likewise.
5916 (emit_fusion_p9_load): Likewise.
5917 (emit_fusion_p9_store): Likewise.
5918 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
5919 (fusion_gpr_mem_load): Likewise.
5920 (fusion_addis_mem_combo_load): Likewise.
5921 (fusion_addis_mem_combo_store): Likewise.
5922 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
5923 (fusion_gpr_mem_load): Likewise.
5924 (fusion_addis_mem_combo_load): Likewise.
5925 (fusion_addis_mem_combo_store): Likewise.
5926 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5928 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
5929 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
5930 (rs6000_preferred_simd_mode): Likewise.
5931 (output_vec_const_move): Likewise.
5932 (rs6000_expand_vector_extract): Likewise.
5933 (rs6000_split_vec_extract_var): Likewise.
5934 (reg_offset_addressing_ok_p): Likewise.
5935 (rs6000_legitimate_offset_address_p): Likewise.
5936 (rs6000_legitimize_address): Likewise.
5937 (rs6000_emit_set_const): Likewise.
5938 (rs6000_const_vec): Likewise.
5939 (rs6000_emit_move): Likewise.
5940 (rs6000_darwin64_record_arg_recurse): Likewise.
5941 (swap_selector_for_mode): Likewise.
5942 (paired_init_builtins): Likewise.
5943 (altivec_init_builtins): Likewise.
5944 (rs6000_expand_float128_convert): Likewise.
5945 (emit_load_locked): Likewise.
5946 (emit_store_conditional): Likewise.
5947 (rs6000_output_function_epilogue): Likewise.
5948 (rs6000_handle_altivec_attribute): Likewise.
5949 (rs6000_function_value): Likewise.
5950 (emit_fusion_gpr_load): Likewise.
5951 (emit_fusion_p9_load): Likewise.
5952 (emit_fusion_p9_store): Likewise.
5953 * config/rx/rx.c (rx_gen_move_template): Likewise.
5954 (flags_from_mode): Likewise.
5955 * config/s390/predicates.md (s390_alc_comparison): Likewise.
5956 (s390_slb_comparison): Likewise.
5957 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
5958 (s390_vector_mode_supported_p): Likewise.
5959 (s390_cc_modes_compatible): Likewise.
5960 (s390_match_ccmode_set): Likewise.
5961 (s390_canonicalize_comparison): Likewise.
5962 (s390_emit_compare_and_swap): Likewise.
5963 (s390_branch_condition_mask): Likewise.
5964 (s390_rtx_costs): Likewise.
5965 (s390_secondary_reload): Likewise.
5966 (__SECONDARY_RELOAD_CASE): Likewise.
5967 (s390_expand_cs): Likewise.
5968 (s390_preferred_simd_mode): Likewise.
5969 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
5970 * config/sh/sh.c (sh_print_operand): Likewise.
5971 (dump_table): Likewise.
5972 (sh_secondary_reload): Likewise.
5973 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
5974 * config/sh/sh.md (casesi_worker_1): Likewise.
5975 (casesi_worker_2): Likewise.
5976 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
5977 (fcc_comparison_operator): Likewise.
5978 * config/sparc/sparc.c (sparc_expand_move): Likewise.
5979 (emit_soft_tfmode_cvt): Likewise.
5980 (sparc_preferred_simd_mode): Likewise.
5981 (output_cbranch): Likewise.
5982 (sparc_print_operand): Likewise.
5983 (sparc_expand_vec_perm_bmask): Likewise.
5984 (vector_init_bshuffle): Likewise.
5985 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
5986 (spu_vector_mode_supported_p): Likewise.
5987 (spu_expand_insv): Likewise.
5988 (spu_emit_branch_or_set): Likewise.
5989 (spu_handle_vector_attribute): Likewise.
5990 (spu_builtin_splats): Likewise.
5991 (spu_builtin_extract): Likewise.
5992 (spu_builtin_promote): Likewise.
5993 (spu_expand_sign_extend): Likewise.
5994 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
5995 (tilegx_simd_int): Likewise.
5996 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
5997 (tilepro_simd_int): Likewise.
5998 * config/v850/v850.c (const_double_split): Likewise.
5999 (v850_print_operand): Likewise.
6000 (ep_memory_offset): Likewise.
6001 * config/vax/vax.c (vax_rtx_costs): Likewise.
6002 (vax_output_int_move): Likewise.
6003 (vax_output_int_add): Likewise.
6004 (vax_output_int_subtract): Likewise.
6005 * config/visium/predicates.md (visium_branch_operator): Likewise.
6006 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
6007 (visium_print_operand_address): Likewise.
6008 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6009 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
6010 (xtensa_expand_conditional_branch): Likewise.
6011 (xtensa_copy_incoming_a7): Likewise.
6012 (xtensa_output_literal): Likewise.
6013 * dfp.c (decimal_real_maxval): Likewise.
6014 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
6016 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6017 Alan Hayward <alan.hayward@arm.com>
6018 David Sherwood <david.sherwood@arm.com>
6020 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
6021 (mode_nunits_inline): Likewise.
6022 (mode_inner_inline): Likewise.
6023 (mode_unit_size_inline): Likewise.
6024 (mode_unit_precision_inline): Likewise.
6025 (emit_insn_modes_h): Likewise. Also emit a #define of the
6027 (emit_mode_wider): Add an E_ prefix to mode names.
6028 (emit_mode_complex): Likewise.
6029 (emit_mode_inner): Likewise.
6030 (emit_mode_adjustments): Likewise.
6031 (emit_mode_int_n): Likewise.
6032 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
6033 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
6034 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
6035 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
6036 (CRC32_BUILTIN, ENTRY): Likewise.
6037 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
6038 (aarch64_pop_regs): Likewise.
6039 (aarch64_process_components): Likewise.
6040 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
6041 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
6042 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
6043 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
6044 * config/arm/arm.c (arm_init_libfuncs): Likewise.
6045 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
6047 * config/i386/i386-builtin.def (pcmpestr): Likewise.
6048 (pcmpistr): Likewise.
6049 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
6050 * config/mmix/mmix.c (mmix_output_condition): Likewise.
6051 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
6053 * config/rl78/rl78.c (mduc_regs): Likewise.
6054 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
6055 (htm_expand_builtin): Likewise.
6056 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
6057 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
6058 * config/xtensa/xtensa.c (print_operand): Likewise.
6059 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
6060 (NUM_MODE_VECTOR_INT): Likewise.
6061 * genoutput.c (null_operand): Likewise.
6062 (output_operand_data): Likewise.
6063 * genrecog.c (print_parameter_value): Likewise.
6064 * lra.c (debug_operand_data): Likewise.
6066 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6068 * dwarf2out.c (scompare_loc_descriptor_wide)
6069 (scompare_loc_descriptor_narrow): New functions, split out from...
6070 (scompare_loc_descriptor): ...here.
6071 * expmed.c (emit_store_flag_int): New function, split out from...
6072 (emit_store_flag): ...here.
6074 2017-08-30 Richard Biener <rguenther@suse.de>
6076 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
6077 (dwarf2out_early_finish): Move setting of AT_pubnames from
6078 early debug output to early finish.
6080 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
6082 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
6083 and -mdata-region to the assembler.
6085 2017-08-30 Richard Biener <rguenther@suse.de>
6087 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
6089 (gen_subprogram_die): Add DW_AT_object_pointer only early.
6090 (dwarf2out_early_global_decl): Only generate a DIE for the
6091 abstract origin if it doesn't already exist or is a declaration DIE.
6092 (resolve_addr): Do not add the linkage name twice when
6093 generating a stub DIE for the DW_TAG_GNU_call_site target.
6095 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6097 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6098 Use machine_mode rather than int for arg1_mode.
6100 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
6103 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
6104 that the second argument of the built-in functions to unpack
6105 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
6106 switch statement instead a lot of if statements.
6107 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
6108 Allow 64-bit values to be in Altivec registers as well as
6109 traditional floating point registers.
6110 (pack<mode>, FMOVE128_VSX iterator): Likewise.
6112 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
6114 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
6115 MAX_REGS_PER_ADDRESS == 1.
6117 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
6119 * config/i386/i386.opt (flag_fentry): Do not init to -1.
6120 * config/i386/i386.c (ix86_option_override_internal): Simplify
6121 setting of opts->x_flag_entry.
6123 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6124 Jakub Jelinek <jakub@redhat.com>
6125 Richard Biener <rguenther@suse.de>
6127 PR tree-optimization/81503
6128 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
6129 folded constant fits in the target type; reorder tests for clarity.
6131 2017-08-29 Martin Liska <mliska@suse.cz>
6133 * passes.def: Include pass_lower_switch.
6134 * stmt.c (dump_case_nodes): Remove and move to
6135 tree-switch-conversion.
6136 (case_values_threshold): Likewise.
6137 (expand_switch_as_decision_tree_p): Likewise.
6138 (emit_case_decision_tree): Likewise.
6139 (expand_case): Likewise.
6140 (balance_case_nodes): Likewise.
6141 (node_has_low_bound): Likewise.
6142 (node_has_high_bound): Likewise.
6143 (node_is_bounded): Likewise.
6144 (emit_case_nodes): Likewise.
6145 (struct simple_case_node): New struct.
6146 (add_case_node): Remove.
6147 (emit_case_dispatch_table): Use vector instead of case_list.
6148 (reset_out_edges_aux): Remove.
6149 (compute_cases_per_edge): Likewise.
6150 (expand_case): Build list of simple_case_node.
6151 (expand_sjlj_dispatch_table): Use it.
6152 * tree-switch-conversion.c (struct case_node): Moved from
6153 stmt.c and adjusted.
6154 (emit_case_nodes): Likewise.
6155 (node_has_low_bound): Likewise.
6156 (node_has_high_bound): Likewise.
6157 (node_is_bounded): Likewise.
6158 (case_values_threshold): Likewise.
6159 (reset_out_edges_aux): Likewise.
6160 (compute_cases_per_edge): Likewise.
6161 (add_case_node): Likewise.
6162 (dump_case_nodes): Likewise.
6163 (balance_case_nodes): Likewise.
6164 (expand_switch_as_decision_tree_p): Likewise.
6165 (emit_jump): Likewise.
6166 (emit_case_decision_tree): Likewise.
6167 (try_switch_expansion): Likewise.
6168 (do_jump_if_equal): Likewise.
6169 (emit_cmp_and_jump_insns): Likewise.
6170 (fix_phi_operands_for_edge): New function.
6171 (record_phi_operand_mapping): Likewise.
6172 (class pass_lower_switch): New pass.
6173 (pass_lower_switch::execute): New function.
6174 (make_pass_lower_switch): Likewise.
6175 (conditional_probability):
6176 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
6177 * tree-pass.h: Add make_pass_lower_switch.
6179 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
6182 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
6185 2017-08-29 Richard Biener <rguenther@suse.de>
6187 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
6188 we do not add a DW_AT_inline attribute twice.
6189 (gen_subprogram_die): Remove code setting DW_AT_inline on
6190 DECL_ABSTRACT_P nodes.
6192 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
6194 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
6195 calls to internal functions.
6196 (gimplify_modify_expr): Likewise.
6197 * tree-call-cdce.c (use_internal_fn): Likewise.
6198 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
6199 (convert_to_divmod): Set the nothrow flag.
6200 * tree-if-conv.c (predicate_mem_writes): Likewise.
6201 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
6202 (vectorizable_call): Likewise.
6203 (vectorizable_store): Likewise.
6204 (vectorizable_load): Likewise.
6205 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
6206 (vect_recog_mask_conversion_pattern): Likewise.
6208 2017-08-29 Martin Liska <mliska@suse.cz>
6211 * gcc.c (driver_handle_option): Add new argument.
6212 * opts-common.c (handle_option): Pass
6213 target_option_override_hook.
6214 * opts-global.c (lang_handle_option): Add new option.
6215 (set_default_handlers): Add new argument.
6216 (decode_options): Likewise.
6217 * opts.c (target_handle_option): Likewise.
6218 (common_handle_option): Call target_option_override_hook.
6219 * opts.h (struct cl_option_handler_func): Add hook for
6220 target option override.
6221 (struct cl_option_handlers): Likewise.
6222 (set_default_handlers): Add new argument.
6223 (decode_options): Likewise.
6224 (common_handle_option): Likewise.
6225 (target_handle_option): Likewise.
6226 * toplev.c (toplev::main): Pass targetm.target_option.override
6229 2017-08-29 Richard Biener <rguenther@suse.de>
6230 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
6232 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
6233 life to the active subtree.
6235 2017-08-28 Jeff Law <law@redhat.com>
6237 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
6238 derive_equivalences.
6239 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
6241 (edge_info::derive_equivalences): New private member function
6243 * tree-ssa-dom.c (class edge_info): Changed from a struct
6244 to a class. Add ctor/dtor, methods and data members.
6245 (edge_info::edge_info): Renamed from allocate_edge_info.
6246 Initialize additional members.
6247 (edge_info::~edge_info): New.
6248 (free_dom_edge_info): Delete the edge info.
6249 (record_edge_info): Use new class & associated member functions.
6250 Tighten forms for testing for edge equivalences.
6251 (record_temporary_equivalences): Iterate over the simple
6252 equivalences rather than assuming there's only one per edge.
6253 (cprop_into_successor_phis): Iterate over the simple
6254 equivalences rather than assuming there's only one per edge.
6255 (optimize_stmt): Use operand_equal_p rather than pointer
6256 equality for mini-DSE code.
6258 2017-08-28 Nathan Sidwell <nathan@acm.org>
6260 * gcc.c (execute): Fold SIGPIPE handling into switch
6261 statement. Adjust internal error message.
6263 2017-08-28 Richard Biener <rguenther@suse.de>
6266 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
6267 Do nothing for removed DIEs.
6269 2017-08-28 Richard Biener <rguenther@suse.de>
6271 PR tree-optimization/81977
6272 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
6275 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
6278 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
6279 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
6280 using targetm.gen_mem_thread_fence.
6282 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
6285 * config/i386/i386.md (*<btsc><mode>): Change operand 2
6286 predicate to register_operand. Reorder operands.
6287 (*btr<mode>): Ditto.
6288 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
6289 (*btr<mode>_mask): Ditto.
6291 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
6293 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
6294 * config/rs6000/xmmintrin.h: New file.
6295 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
6297 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6300 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
6301 parameter and_insn and return it.
6302 (recombine_lvx_pattern): Insert a copy to ensure availability of
6303 the base register of the copied masking operation at the point of
6304 the instruction replacement.
6305 (recombine_stvx_pattern): Likewise.
6307 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
6309 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
6310 undocumented switches.
6311 (-mpower9-dform-vector): Likewise.
6312 (-mpower9-dform): Likewise.
6313 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
6314 comments to delete references to -mpower9-dform* switches.
6315 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
6316 Delete reference to -mpower9-dform* switches, test for
6317 -mpower9-vector instead.
6318 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
6319 (OTHER_P9_VECTOR_MASKS): Likewise.
6320 (POWERPC_MASKS): Likewise.
6321 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
6322 tests against -mpower9-dform* to -mpower9-vector. Delete code
6323 that checked for -mpower9-dform* consistancy with other options.
6324 Add test for -mpower9-misc to enable other power9 switches.
6325 (rs6000_init_hard_regno_mode_ok): Likewise.
6326 (rs6000_option_override_internal): Likewise.
6327 (rs6000_emit_prologue): Likewise.
6328 (rs6000_emit_epilogue): Likewise.
6329 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
6330 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
6331 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
6333 (emit_fusion_p9_store): Likewise.
6334 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
6335 resetting these macros if the assembler does not support ISA 3.0
6337 (TARGET_P9_DFORM_VECTOR): Likewise.
6338 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
6339 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
6341 2017-08-25 Alan Modra <amodra@gmail.com>
6343 PR rtl-optimization/81747
6344 * cse.c (cse_extended_basic_block): Don't attempt to record
6345 equivalences for degenerate conditional branches that branch
6346 to their fall-through.
6348 2017-08-24 Martin Sebor <msebor@redhat.com>
6351 * gimple-fold.c (size_must_be_zero_p): New function.
6352 (gimple_fold_builtin_memory_op): Call it.
6354 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
6356 * config/rs6000/mm_malloc.h: New file.
6358 2017-08-24 Bin Cheng <bin.cheng@arm.com>
6360 PR tree-optimization/81913
6361 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
6362 analysis when either IVs in condition can wrap.
6364 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
6366 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
6367 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
6369 2017-08-24 Richard Biener <rguenther@suse.de>
6372 * targhooks.c (default_target_can_inline_p): Properly
6373 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
6374 is present and always compare.
6375 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
6376 infer -mfpmath=sse from TARGET_SSE_P.
6377 (ix86_can_inline_p): Properly use target_option_default_node when
6378 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
6380 2017-08-24 Richard Biener <rguenther@suse.de>
6383 * dwarf2out.c (output_die): Handle flag_generate_offload like
6385 (output_comp_unit): Likewise.
6386 (gen_array_type_die): Likewise.
6387 (dwarf2out_early_finish): Likewise.
6388 (note_variable_value_in_expr): Likewise.
6389 (dwarf2out_finish): Likewise. Adjust assert.
6390 * cgraphunit.c (symbol_table::compile): Move setting of
6391 flag_generate_offload earlier ...
6392 (symbol_table::finalize_compilation_unit): ... here, before
6393 early debug finalization.
6395 2017-08-24 Richard Biener <rguenther@suse.de>
6397 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
6398 and ipa-fnsummary.h.
6399 (ix86_can_inline_p): When ix86_fpmath flags do not match
6400 check whether the callee uses FP math at all.
6402 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
6405 * tree-ssanames.c (get_nonzero_bits): Use element_precision
6406 instead of TYPE_PRECISION.
6408 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
6409 Alan Hayward <alan.hayward@arm.com>
6410 David Sherwood <david.sherwood@arm.com>
6412 * combine.c (make_extraction): Use subreg_offset_from_lsb.
6414 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
6416 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
6418 (ix86_frame::stack_realign_allocate): New field.
6419 (struct machine_frame_state): Modify comments.
6420 (machine_frame_state::sp_realigned_fp_end): New field.
6421 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
6423 (sp_valid_at): Add assertion to assure no attempt to access invalid
6424 offset of a realigned stack.
6425 (fp_valid_at): Likewise.
6426 (choose_baseaddr): Modify comments.
6427 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
6428 ix86_expand_prologue.
6429 (ix86_expand_prologue): Modify stack realignment and allocation.
6430 (ix86_expand_epilogue): Modify comments.
6431 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
6432 avx2_runtime, avx512f, and avx512f_runtime.
6434 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
6436 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
6437 (mstackrealign): Do not init to -1.
6438 * config/i386/i386.c (ix86_option_override_internal):
6439 Check opts_set, not opts when setting default value of
6440 opts->x_ix86_force_align_arg_pointer.
6442 2017-08-23 Richard Biener <rguenther@suse.de>
6444 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
6445 lang_hooks.decl_printable_name.
6446 * print-rtl-function.c (print_rtx_function): Likewise.
6447 * tree-pretty-print.c (dump_function_header): Likewise.
6449 2017-08-23 Richard Biener <rguenther@suse.de>
6452 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
6453 -g0 at compile-time.
6455 2017-08-23 Tamar Christina <tamar.christina@arm.com>
6458 * doc/sourcebuild.texi (Other hardware attributes):
6461 2017-08-23 Tamar Christina <tamar.christina@arm.com>
6464 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
6465 Fix single-use check.
6467 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
6469 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
6471 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
6473 * doc/install.texi: Modify to add more details on running selected
6476 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
6478 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
6479 is combined with -mabi=ms.
6480 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
6483 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6485 PR tree-optimization/81488
6486 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
6487 and cached_basis fields.
6488 (MAX_SPREAD): New constant.
6489 (alloc_cand_and_find_basis): Initialize new fields.
6490 (clear_visited): New function.
6491 (create_phi_basis_1): Rename from create_phi_basis, set visited
6492 and cached_basis fields.
6493 (create_phi_basis): New wrapper function.
6494 (phi_add_costs_1): Rename from phi_add_costs, add spread
6495 parameter, set visited field, short-circuit when limits reached.
6496 (phi_add_costs): New wrapper function.
6497 (record_phi_increments_1): Rename from record_phi_increments, set
6499 (record_phi_increments): New wrapper function.
6500 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
6501 (phi_incr_cost): New wrapper function.
6502 (all_phi_incrs_profitable_1): Rename from
6503 all_phi_incrs_profitable, set visited field.
6504 (all_phi_incrs_profitable): New wrapper function.
6506 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
6507 Alan Hayward <alan.hayward@arm.com>
6508 David Sherwood <david.sherwood@arm.com>
6510 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
6511 that takes the outer and inner modes.
6512 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
6513 comparison as the canonical test for a paradoxical subreg.
6514 * combine.c (simplify_set): Use paradoxical_subreg_p.
6515 (make_extraction): Likewise.
6516 (force_to_mode): Likewise.
6517 (rtx_equal_for_field_assignment_p): Likewise.
6518 (gen_lowpart_for_combine): Likewise.
6519 (simplify_comparison): Likewise.
6520 * cse.c (equiv_constant): Likewise.
6521 * expmed.c (store_bit_field_1): Likewise.
6522 * final.c (alter_subreg): Likewise.
6523 * fwprop.c (propagate_rtx): Likewise.
6524 (forward_propagate_subreg): Likewise.
6525 * ira-conflicts.c (ira_build_conflicts): Likewise.
6526 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
6527 * lra-constraints.c (curr_insn_transform): Likewise.
6528 (split_reg): Likewise.
6529 * lra-eliminations.c (move_plus_up): Likewise.
6530 (lra_eliminate_regs_1): Likewise.
6531 * recog.c (general_operand): Likewise.
6532 * ree.c (combine_reaching_defs): Likewise.
6533 * reload.c (push_reload): Likewise.
6534 (find_reloads): Likewise.
6535 * reload1.c (elimination_effects): Likewise.
6536 (compute_reload_subreg_offset): Likewise.
6537 (choose_reload_regs): Likewise.
6538 * rtlanal.c (subreg_lsb_1): Likewise.
6539 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6540 (simplify_subreg): Likewise.
6541 * var-tracking.c (track_loc_p): Likewise.
6542 * emit-rtl.c (byte_lowpart_offset): Likewise.
6543 (paradoxical_subreg_p): Delete out-of-line definition.
6545 2017-08-22 Jeff Law <law@redhat.com>
6547 PR tree-optimization/81741
6548 PR tree-optimization/71947
6549 * tree-ssa-dom.c: Include tree-inline.h.
6550 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
6551 equivalences if one is more expensive to compute than the other.
6552 * tree-ssa-scopedtables.h (class const_or_copies): Make
6553 record_const_or_copy_raw method private.
6554 (class avail_exprs_stack): New method simplify_binary_operation.
6555 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
6556 avail_exprs_stack::simplify_binary_operation as needed.
6557 (avail_exprs_stack::simplify_binary_operation): New function.
6559 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
6561 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
6562 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
6563 (DOT_SYMBOLS): Likewise.
6564 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
6565 (RELOCATABLE_NEEDS_FIXUP): Likewise.
6566 (RS6000_ABI_NAME): Likewise.
6567 (TARGET_CMODEL): Likewise.
6568 (TOC_SECTION_ASM_OP): Likewise.
6569 (SET_CMODEL): New macro.
6570 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
6572 2017-08-22 Richard Biener <rguenther@suse.de>
6574 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
6575 to allow for free-lang-data replacements similar to verify_type_variant.
6577 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
6578 Alan Hayward <alan.hayward@arm.com>
6579 David Sherwood <david.sherwood@arm.com>
6581 * config/aarch64/aarch64.md (casesi): Use DImode rather than
6582 VOIDmode for the LABEL_REF.
6584 2017-08-22 Richard Biener <rguenther@suse.de>
6586 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
6588 2017-08-22 Richard Biener <rguenther@suse.de>
6590 * common.opt (feliminate-dwarf2-dups): Ignore.
6591 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
6592 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
6593 same_die_p_wrap, compute_section_prefix,
6594 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
6595 (comdat_symbol_id, comdat_symbol_number): Likewise.
6596 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
6598 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
6599 (output_die): Mark unreachable path unreachable.
6600 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
6601 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
6602 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
6603 (dwarf2out_early_finish): Likewise.
6605 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
6607 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
6609 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
6612 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
6613 not VAR_P. Filter attribute warnings with OPT_Wattributes.
6614 (avr_attribute_table) <io, io_low, address>: Initialize
6615 .decl_required with true.
6617 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
6619 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
6620 undocumented debugging options.
6621 (-mvsx-scalar-double): Likewise.
6622 (-mallow-df-permute): Likewise.
6623 (-mvectorize-builtins): Likewise.
6624 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
6625 (rs6000_builtin_vectorized_function): Likewise.
6626 (rs6000_builtin_md_vectorized_function): Likewise.
6627 (rs6000_opt_vars): Likewise.
6629 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
6632 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
6633 (*btrq_imm): Rename from *btrq.
6634 (*btcq_imm): Rename from *btcq.
6635 (btsc): New code attribute.
6636 (*<btsc><mode>): New insn pattern.
6637 (*btr<mode>): Ditto.
6638 (*<btsc><mode>_mask): New insn_and_split pattern.
6639 (*btr<mode>_mask): Ditto.
6641 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
6642 Alan Hayward <alan.hayward@arm.com>
6643 David Sherwood <david.sherwood@arm.com>
6645 * function.c (pad_below): Simplify padding calculation.
6647 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
6648 Alan Hayward <alan.hayward@arm.com>
6649 David Sherwood <david.sherwood@arm.com>
6651 * target.def (function_prologue): Remove frame size argument.
6652 (function_epilogue): Likewise.
6653 * doc/tm.texi: Regenerate.
6654 * final.c (final_start_function): Update call to function_prologue.
6655 (final_end_function): Update call to function_epilogue.
6656 (default_function_pro_epilogue): Remove frame size argument.
6657 * output.h (default_function_pro_epilogue): Likewise.
6658 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
6659 (arm_output_function_prologue): Likewise.
6660 * config/frv/frv.c (frv_function_prologue): Likewise.
6661 (frv_function_epilogue): Likewise.
6662 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
6663 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
6664 (ia64_output_function_epilogue): Likewise.
6665 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
6666 (m32r_output_function_epilogue): Likewise.
6667 * config/microblaze/microblaze.c (microblaze_function_prologue)
6668 (microblaze_function_epilogue): Likewise.
6669 * config/mips/mips.c (mips_output_function_prologue): Likewise.
6670 (mips_output_function_epilogue): Likewise.
6671 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
6672 (mmix_target_asm_function_epilogue): Likewise.
6673 * config/msp430/msp430.c (msp430_start_function): Likewise.
6674 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
6675 (nds32_asm_function_epilogue): Likewise.
6676 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
6677 * config/pa/pa.c (pa_output_function_prologue): Likewise.
6678 (pa_output_function_epilogue): Likewise.
6679 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
6680 (rs6000_output_function_epilogue): Likewise.
6681 * config/rl78/rl78.c (rl78_start_function): Likewise.
6682 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
6683 (rs6000_output_function_epilogue): Likewise.
6684 * config/rx/rx.c (rx_output_function_prologue): Likewise.
6685 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
6686 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
6687 (sparc_asm_function_epilogue): Likewise.
6689 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
6691 * tree.h (type_has_mode_precision_p): New function.
6692 * convert.c (convert_to_integer_1): Use it.
6693 * expr.c (expand_expr_real_2): Likewise.
6694 (expand_expr_real_1): Likewise.
6695 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
6696 * match.pd: Likewise.
6697 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
6698 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
6699 * tree-tailcall.c (process_assignment): Likewise.
6700 * tree-vect-loop.c (vectorizable_reduction): Likewise.
6701 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
6702 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
6703 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
6704 (vectorizable_assignment): Likewise.
6705 (vectorizable_shift): Likewise.
6706 (vectorizable_operation): Likewise.
6707 * tree-vrp.c (register_edge_assert_for_2): Likewise.
6709 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
6711 * match.pd: Add pow (C, x) simplification.
6713 2017-08-21 Richard Biener <rguenther@suse.de>
6715 PR tree-optimization/81900
6716 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
6717 for blocks with abnormal predecessors.
6718 (compute_antic): Do not set visited flag prematurely.
6720 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
6723 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
6725 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
6727 * stor-layout.h (vector_type_mode): Move to...
6728 * tree.h (vector_type_mode): ...here.
6729 * stor-layout.c (vector_type_mode): Move to...
6730 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
6732 2017-08-21 Richard Biener <rguenther@suse.de>
6734 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
6735 register_external_die hooks.
6736 (debug_false_tree_charstarstar_uhwistar): Declare.
6737 (debug_nothing_tree_charstar_uhwi): Likewise.
6738 * debug.c (do_nothing_debug_hooks): Adjust.
6739 (debug_false_tree_charstarstar_uhwistar): New do nothing.
6740 (debug_nothing_tree_charstar_uhwi): Likewise.
6741 * dbxout.c (dbx_debug_hooks): Adjust.
6742 (xcoff_debug_hooks): Likewise.
6743 * sdbout.c (sdb_debug_hooks): Likewise.
6744 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
6745 * dwarf2out.c (macinfo_label_base): New global.
6746 (dwarf2out_register_external_die): New function for the
6747 register_external_die hook.
6748 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
6749 (dwarf2_debug_hooks): Use them.
6750 (dwarf2_lineno_debug_hooks): Adjust.
6751 (struct die_struct): Add with_offset flag.
6752 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
6753 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
6754 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
6755 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
6756 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
6757 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
6758 defining section names for the early LTO debug variants.
6759 (reset_indirect_string): New helper.
6760 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
6761 (print_dw_val): Add support for offsetted symbol references.
6762 (get_ultimate_context): Split out from is_cxx.
6763 (is_cxx): Use get_ultimate_context.
6764 (is_fortran): Add decl overload.
6765 (compute_comp_unit_symbol): Split out worker from
6766 compute_section_prefix.
6767 (compute_section_prefix): Call compute_comp_unit_symbol and
6768 set comdat_type_p here.
6769 (output_die): Skip DIE symbol output for the LTO added one.
6770 Handle DIE symbol references with offset.
6771 (output_comp_unit): Guard section name mangling properly.
6772 For LTO debug sections emit a symbol at the section beginning
6773 which we use to refer to its DIEs.
6774 (add_abstract_origin_attribute): For DIEs registered via
6775 dwarf2out_register_external_die directly refer to the early
6776 DIE rather than indirectly through the shadow one we created.
6777 Remove obsolete call to dwarf2out_abstract_function for
6778 non-function/block origins.
6779 (gen_array_type_die): When generating early LTO debug do
6780 not emit DW_AT_string_length.
6781 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
6782 late when in LTO. As suggested place a gcc_unreachable for
6783 the DECL_ABSTRACT_P case.
6784 (gen_subprogram_die): Avoid another specification DIE
6785 for early built declarations/definitions for the late LTO case.
6786 (gen_variable_die): Add type references for late duplicated VLA dies
6788 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
6789 we have the abstract instance already.
6790 (process_scope_var): Adjust decl DIE contexts in LTO which
6791 first puts them in limbo.
6792 (gen_decl_die): Do not generate type DIEs late apart from
6793 types for VLAs or for decls we do not yet have a DIE. Do not
6794 call dwarf2out_abstract_function late.
6795 (dwarf2out_early_global_decl): Make sure to create DIEs
6796 for abstract instances of a decl first.
6797 (dwarf2out_late_global_decl): Adjust comment.
6798 (output_macinfo_op): With multiple macro sections use
6799 macinfo_label_base to distinguish labels.
6800 (output_macinfo): Likewise. Update macinfo_label_base.
6801 Pass in the line info label.
6802 (note_variable_value_in_expr): When generating LTO resolve
6803 all variable values here by generating DIEs as needed.
6804 (init_sections_and_labels): Add early LTO debug flag parameter
6805 and generate different sections and names if set. Add generation
6806 counter for the labels so we can have multiple of them.
6807 (reset_dies): Helper to allow DIEs to be output multiple times.
6808 (dwarf2out_finish): When outputting DIEs to the fat part of an
6809 LTO object first reset DIEs.
6810 (dwarf2out_early_finish): Output early DIEs when generating LTO.
6811 (modified_type_die): Check for decl_ultimate_origin being self
6813 (gen_type_die_with_usage): Likewise.
6814 (gen_typedef_die): Allow decl_ultimate_origin being self.
6815 (set_decl_abstract_flags): Remove.
6816 (set_block_abstract_flags): Likewise.
6817 (dwarf2out_abstract_function): Treat the early generated DIEs
6818 as the abstract copy and only add DW_AT_inline and
6819 DW_AT_artificial here and call set_decl_origin_self.
6820 If the DIE has an abstract origin don't do anything.
6821 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
6822 if we have none yet (Go fails to build one, PR78628).
6823 (variably_modified_type_p): Prevent endless recursion for Ada
6824 cyclic pointer types.
6825 * lto-streamer-in.c: Include debug.h.
6826 (dref_queue): New global.
6827 (lto_read_tree_1): Stream in DIE references.
6828 (lto_input_tree): Register DIE references.
6829 (input_function): Stream DECL_DEBUG_ARGS.
6830 * lto-streamer-out.c: Include debug.h.
6831 (lto_write_tree_1): Output DIE references.
6832 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
6833 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
6834 (output_function): Stream DECL_DEBUG_ARGS.
6835 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
6836 Stream DECL_ABSTRACT_ORIGIN.
6837 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
6838 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
6839 DECL_CONTEXT for file-scope decls.
6840 * lto-streamer.h (struct dref_entry): Declare.
6841 (dref_queue): Likewise.
6842 * cfgexpand.c (pass_expand::execute): Do not call the
6843 outlining_inline_function hook here.
6844 * lto-wrapper.c (debug_obj): New global.
6845 (tool_cleanup): Unlink it if required.
6846 (debug_objcopy): New function.
6847 (run_gcc): Handle early debug sections in the IL files by
6848 extracting them to separate files, partially linkin them and
6849 feeding the result back as result to the linker.
6850 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
6851 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
6852 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
6853 sections into a separate segment.
6854 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
6856 (darwin_asm_dwarf_section): Likewise.
6857 (darwin_asm_output_dwarf_offset): Likewise.
6858 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
6860 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
6861 Alan Hayward <alan.hayward@arm.com>
6862 David Sherwood <david.sherwood@arm.com>
6864 * read-md.h (md_reader::record_potential_iterator_use): Replace
6865 pointer argument with an rtx and an index.
6866 * read-rtl.c (iterator_group::apply_iterator): Likewise.
6867 (apply_mode_iterator): Likewise.
6868 (apply_code_iterator): Likewise.
6869 (apply_int_iterator): Likewise.
6870 (apply_subst_iterator): Likewise.
6871 (record_iterator_use): Likewise.
6872 (record_attribute_use): Likewise.
6873 (md_reader::record_potential_iterator_use): Likewise. Update calls
6874 to record_iterator_use and apply_iterator.
6875 (iterator_use): Replace ptr with x and index.
6876 (attribute_use): Likewise.
6877 (apply_attribute_uses): Update calls to apply_iterator.
6878 (apply_iterators): Likewise. Update initialization of iterator_use.
6879 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
6880 and record_potential_iterator_use.
6881 (rtx_reader::read_rtx_operand): Likewise.
6883 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
6884 Alan Hayward <alan.hayward@arm.com>
6885 David Sherwood <david.sherwood@arm.com>
6887 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
6890 2017-08-21 Richard Biener <rguenther@suse.de>
6893 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
6894 at struct end conservatively when comparing common bases.
6896 2017-08-21 Richard Biener <rguenther@suse.de>
6898 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
6899 (mem_ref_in_stmt): Remove.
6900 (determine_max_movement): Use ref index to get at the reference.
6901 (invariantness_dom_walker::before_dom_children): Deal with
6902 lim data already initialized.
6903 (gather_mem_refs_stmt): Initialize lim data and record ref index.
6905 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
6907 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
6908 (TARGET_ISA_ROUND): Ditto.
6909 (TARGET_ROUND): Ditto.
6910 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
6911 * config/i386/i386.md: Ditto.
6912 * config/i386/sse.md: Ditto.
6913 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
6914 with OPTION_MASK_ISA_SSE4_1.
6916 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
6919 * doc/extend.texi (x86 Built-in Functions): Correct the name of
6920 __builtin_ia32_lzcnt_u16.
6922 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
6925 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
6926 (rs6000_set_current_function): Rewrite function to use it.
6928 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
6931 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
6932 and TYPE_WARN_IF_NOT_ALIGN.
6933 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
6934 (handle_warn_if_not_align): New.
6935 (place_union_field): Call handle_warn_if_not_align.
6936 (place_field): Call handle_warn_if_not_align.
6937 Copy TYPE_WARN_IF_NOT_ALIGN.
6938 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
6939 (layout_type): Likewise.
6940 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
6942 (tree_decl_common): Add warn_if_not_align.
6943 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
6944 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
6945 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
6946 (DECL_WARN_IF_NOT_ALIGN): Likewise.
6947 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
6948 * doc/extend.texi: Document warn_if_not_aligned attribute.
6949 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
6951 2017-08-17 Martin Liska <mliska@suse.cz>
6954 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
6955 (get_data_dependence): Use it as pointer type.
6956 (distribute_loop): Likewise.
6958 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6960 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
6961 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
6962 (p8_vmrgow_<mode>_direct): New define_insn.
6963 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
6964 handle endianness for vmrgew and vmrgow permute patterns.
6966 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
6968 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
6969 * config/rs6000/rs6000-cpus.def: Remove comment.
6970 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
6971 (POWERPC_MASKS): Likewise.
6972 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
6973 use of TARGET_VSX_TIMODE.
6974 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
6975 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
6976 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
6977 (rs6000_option_override_internal): Remove dead code.
6978 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
6979 (rs6000_legitimize_reload_address): Likewise.
6980 (rs6000_legitimate_address_p): Likewise.
6981 (rs6000_opt_masks): Delete "vsx-timode".
6982 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
6983 from function comment.
6984 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
6985 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
6986 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
6988 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
6989 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
6990 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
6993 (define_peephole2): Likewise.
6995 2017-08-17 Martin Sebor <msebor@redhat.com>
6998 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
6999 past the end of an array.
7000 (test_pp_format): Add test cases.
7002 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
7004 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
7005 missing ECF_NOTHROW flags.
7007 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
7010 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
7011 operands residing in integer registers.
7012 (*vsx_le_perm_load_<mode>): Likewise.
7013 (*vsx_le_perm_store_<mode>): Likewise.
7014 (define_peephole2): Add peepholes to optimize the above.
7016 2017-08-17 Marek Polacek <polacek@redhat.com>
7019 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
7020 to mimic what shorten_compare did. Change the return type to bool.
7021 (fold_cond_expr_with_comparison): Update call to
7022 operand_equal_for_comparison_p.
7023 (fold_ternary_loc): Likewise.
7025 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
7027 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
7029 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
7031 2017-08-17 Richard Biener <rguenther@suse.de>
7033 * tree-ssa-structalias.c (solve_graph): When propagating
7034 to successors update the graphs succ edges and avoid duplicate work.
7036 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
7039 * config/i386/i386.c (ix86_option_override_internal): Save target
7040 specific options after ix86_stack_protector_guard_reg was changed.
7042 2017-08-17 Richard Biener <rguenther@suse.de>
7044 PR tree-optimization/81827
7045 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
7046 (new_var_info): Initialize it conservatively.
7047 (get_call_vi): Mark register vars.
7048 (new_scalar_tmp_constraint_exp): Likewise.
7049 (handle_rhs_call): Likewise.
7050 (handle_const_call): Likewise.
7051 (create_function_info_for): Likewise.
7052 (solve_constraints): Sort varinfos to separate register from
7053 non-register vars to pack points-to solution bitmaps during
7056 2017-08-17 Marek Polacek <polacek@redhat.com>
7058 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
7060 2017-08-17 Richard Biener <rguenther@suse.de>
7062 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
7063 to true when overflow is undefined and we saturated the result.
7065 2017-08-17 Alan Modra <amodra@gmail.com>
7068 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
7069 Don't use store multiple if only one reg needs saving.
7070 (interesting_frame_related_regno): New function.
7071 (rs6000_frame_related): Don't emit frame info for regs that
7073 (rs6000_emit_epilogue): Likewise.
7075 2017-08-16 Nathan Sidwell <nathan@acm.org>
7077 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
7078 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
7079 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
7080 * tree.c (free_lang_data_in_type): Use else-if chain. Always
7081 clear TYPE_LANG_1. Remove obsolete member-function stripping.
7082 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
7083 (verify_type): Adjust for TYPE_BINFO move.
7084 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
7085 process TYPE_BINFO directly.
7086 (hash_tree): Likewise.
7087 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
7089 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
7092 2017-08-16 David Malcolm <dmalcolm@redhat.com>
7094 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
7096 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
7099 * config/i386/i386.md (*anddi_1_btr): Change predicates of
7100 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
7101 Add ix86_binary_operator_ok to insn constraint.
7102 (*iordi_1_bts): Ditto.
7103 (*xordi_1_btc): Ditto.
7104 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
7105 Update corresponding peephole2 pattern.
7109 2017-08-16 Bin Cheng <bin.cheng@arm.com>
7111 PR tree-optimization/81832
7112 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
7113 copy loop header which has IFN_LOOP_DIST_ALIAS call.
7115 2017-08-16 Marek Polacek <polacek@redhat.com>
7118 * fold-const.c (fold_indirect_ref_1): Restore original behavior
7119 regarding size_zero_node.
7121 2017-08-16 Martin Liska <mliska@suse.cz>
7124 * config.gcc: Respect previously set extra_objs in case
7127 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
7129 PR tree-optimization/81835
7130 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
7131 the phi SSA_NAME. Check that the condition in a COND_EXPR does
7132 not depend on the phi.
7134 2017-08-16 Alan Modra <amodra@gmail.com>
7136 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
7139 2017-08-16 Alan Modra <amodra@gmail.com>
7141 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
7142 (save_reg_p): ..into this. Update all callers.
7143 (first_reg_to_save): Simplify.
7145 2017-08-16 Alan Modra <amodra@gmail.com>
7147 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
7150 2017-08-15 Joseph Myers <joseph@codesourcery.com>
7154 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
7155 constant count if that count is less than 32.
7157 2017-08-15 Nathan Sidwell <nathan@acm.org>
7159 * gcc.c (execute): Emit friendlier message if inferior is killed
7160 by an external cause.
7162 2017-08-15 Richard Biener <rguenther@suse.de>
7164 PR tree-optimization/81790
7165 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
7166 CONSTRUCTORs from simplifying and VN.
7168 2017-08-14 Martin Sebor <msebor@redhat.com>
7170 * builtin-attrs.def: Add comments.
7172 2017-08-14 Martin Sebor <msebor@redhat.com>
7175 * doc/extend.texi (attribute nonstring): Document new attribute.
7177 2017-08-14 Martin Sebor <msebor@redhat.com>
7180 * tree-diagnostic.c (default_tree_printer): Handle %G.
7181 * gimple-pretty-print.h (percent_G_format): Declare new function.
7182 * gimple-pretty-print.c (percent_G_format): Define.
7183 * tree-pretty-print.c (percent_K_format): Add argument.
7185 2017-08-14 Martin Sebor <msebor@redhat.com>
7187 PR translation/79998
7188 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
7189 Remove a stray space.
7191 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
7194 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
7195 (*iordi_1_bts): Ditto.
7196 (*xordi_1_btc): Ditto.
7198 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7201 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
7202 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7204 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
7205 quoted strings, and make more translator-friendly.
7206 (darwin_rs6000_override_options): Likewise.
7207 (rs6000_option_override_internal): Likewise.
7208 (rs6000_return_in_memory): Fix overlong line.
7209 (init_cmulative_args): Use quoted strings, and make more
7210 translator-friendly.
7211 (rs6000_pass_by_reference): Fix overlong line.
7212 (def_builtin): Use quoted strings.
7213 (altivec_expand_predicate_builtin): Use quoted strings, and make
7214 more translator-friendly.
7215 (htm_expand_builtin): Use quoted strings.
7216 (cpu_expand_builtin): Use quoted strings, and make more
7217 translator-friendly.
7218 (altivec_expand_builtin): Likewise.
7219 (paired_expand_predicate_builtin): Likewise.
7220 (rs6000_invalid_builtin): Likewise.
7221 (builtin_function_type): Use quoted strings.
7222 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
7223 more translator-friendly.
7224 (rs6000_trampoline_init): Likewise.
7225 (rs6000_handle_altivec_attribute): Likewise.
7226 (rs6000_inner_target_options): Use quoted strings.
7227 (rs6000_disable_incompatible_switches): Likewise.
7228 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
7229 strings, and make more translator-friendly.
7230 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
7232 2017-08-14 Bin Cheng <bin.cheng@arm.com>
7234 PR tree-optimization/81799
7235 * tree-loop-distribution.c (version_loop_by_alias_check): Force
7236 cond_expr to simple gimple operand.
7238 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
7241 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
7243 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
7247 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
7248 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
7249 TARGET_GASISR_PROLOGUES.
7250 * config/avr/avr.c (avr_option_override): Same.
7251 (avr_pass_pre_proep::execute): Same.
7253 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
7256 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
7257 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
7259 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
7261 * config/i386/i386.md (*load_tp_<mode>): Redefine as
7262 define_insn_and_split. Split to a memory load from 0 in
7263 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
7264 using PTR mode iterator.
7265 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
7266 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
7267 (*add_tp_<mode>): Redefine as define_insn_and_split.
7268 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
7269 address space. Merge with *add_tp_x32 using PTR mode iterator.
7270 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
7271 Split to an add with a memory load from 0 in
7272 DEFAULT_TLS_SEG_REG address space.
7274 2017-08-12 Andrew Pinski <apinski@cavium.com>
7276 * config/aarch64/aarch64-option-extensions.def (rdma):
7277 Fix feature string to what Linux prints out in /proc/cpuinfo.
7279 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
7282 * dwarf2out.c (modified_type_die): For C typedef types that have
7283 an ultimate origin, process the ultimate origin instead of the
7285 (gen_typedef_die): Assert that input DECLs have no ultimate
7287 (gen_type_die_with_usage): For typedef variants that have an
7288 ultimate origin, just call gen_decl_die on the original DECL.
7289 (process_scope_var): Avoid creating DIEs for local typedefs and
7290 concrete static variables.
7292 2017-08-12 Alan Modra <amodra@gmail.com>
7296 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
7297 match gnu-user.h startfile.
7298 (ENDFILE_LINUX_SPEC): Similarly.
7300 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
7303 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
7305 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
7307 2017-08-11 Tamar Christina <tamar.christina@arm.com>
7308 * config/aarch64/aarch64.md (mov<mode>): Change.
7309 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
7310 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
7311 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
7313 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
7315 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
7316 for storage order barriers.
7318 2017-08-11 Martin Liska <mliska@suse.cz>
7321 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
7322 variables of void type.
7324 2017-08-11 Martin Liska <mliska@suse.cz>
7326 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
7327 TARGET_SUPPORTS_ALIASES.
7328 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
7329 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
7330 (optimize_weakref): Likewise.
7331 * symtab.c (symtab_node::noninterposable_alias): Likewise.
7332 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
7333 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
7335 2017-08-11 Martin Liska <mliska@suse.cz>
7338 * config/i386/i386.c (make_resolver_func): Do complete
7339 refactoring of the function.
7341 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
7344 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
7345 * config/i386/i386.c (ix86_stack_protect_guard): Use
7346 ix86_stack_protect_guard_symbol_str to generate varible declaration.
7347 * doc/invoke.texi (x86 Options): Document
7348 -mstack-protector-guard-symbol= option.
7350 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
7352 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
7353 * config/i386/i386.c (ix86_split_stack_guard): New function.
7354 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
7355 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
7356 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
7357 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
7358 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
7359 (split_stack_space_check): Call ix86_split_stack_guard.
7361 2017-08-10 Martin Sebor <msebor@redhat.com>
7363 * print-tree.c (print_node): Print location using the established
7365 Replace spaces with colons.
7366 (debug_raw, debug): Ditto.
7368 2017-08-10 Martin Sebor <msebor@redhat.com>
7371 * pretty-print.c (pp_format): Correct the handling of %s precision.
7373 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
7376 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
7378 (ix86_finalize_stack_frame_flags): This. Also clear
7379 frame_pointer_needed if -fno-omit-frame-pointer is used without
7381 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
7382 with ix86_finalize_stack_frame_flags.
7383 (ix86_expand_epilogue): Likewise.
7384 (ix86_expand_split_stack_prologue): Likewise.
7385 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
7387 2017-08-10 Martin Liska <mliska@suse.cz>
7390 * c-attribs.c (handle_target_attribute):
7391 Report warning for an empty string argument of target attribute.
7393 2017-08-09 Jakub Jelinek <jakub@redhat.com>
7396 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
7398 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
7399 or DECL_NONLOCAL labels.
7400 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
7401 or DECL_NONLOCAL labels here.
7403 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
7405 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
7406 to indicate when early gimple folding has been disabled.
7407 (rs6000_gimple_fold_builtin): Add debug content.
7408 (rs6000_invalid_builtin): Fix whitespace.
7409 (rs6000_expand_builtin): Fix whitespace.
7410 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
7412 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
7415 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
7416 SAVE_MULTIPLE if not all the registers that saves, should be saved.
7418 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
7420 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
7421 (qdf24xx): Likewise.
7422 * config/aarch64/aarch64.md: Include falkor.md.
7423 * config/aarch64/falkor.md: New.
7425 2017-08-09 Marek Polacek <polacek@redhat.com>
7428 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
7429 * diagnostic.c (emit_diagnostic): Add a comment.
7430 (emit_diagnostic_valist): New function.
7432 2017-08-09 Marek Polacek <polacek@redhat.com>
7435 * input.c (make_location): New overload.
7436 * input.h (make_location): Declare.
7438 2017-08-08 Alan Modra <amodra@gmail.com>
7439 H.J. Lu <hongjiu.lu@intel.com>
7442 * gcc.c (NO_PIE_SPEC): Delete.
7443 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
7445 (LINK_PIE_SPEC): ..to here.
7446 (LINK_COMMAND_SPEC): Support -no-pie.
7447 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
7448 chain of crtbegin*.o selection, update for PIE_SPEC changes and
7450 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
7451 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
7452 (ENDFILE_CRTEND_SPEC): Similarly.
7454 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
7457 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
7458 (mstack-protector-guard-offset=): Ditto.
7459 * config/i386/i386.c (ix86_option_override): Handle
7460 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
7462 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
7463 ix86_stack_protect_guard_offset variables.
7464 (TARGET_STACK_PROTECT_GUARD): Always define.
7465 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
7466 and -mstack-protector-guard-offset= options.
7468 2017-08-08 Bin Cheng <bin.cheng@arm.com>
7470 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
7471 boundary case for the last candidate.
7473 2017-08-08 Bin Cheng <bin.cheng@arm.com>
7475 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
7476 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
7478 2017-08-08 Tamar Christina <tamar.christina@arm.com>
7481 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
7482 * config/aarch64/aarch64-builtins.c
7483 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
7484 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
7485 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
7487 2017-08-08 Tamar Christina <tamar.christina@arm.com>
7488 Andrew Pinski <pinskia@gmail.com>
7491 * internal-fn.def (XORSIGN): New.
7492 * optabs.def (xorsign_optab): New.
7493 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
7494 (convert_expand_mult_copysign): New.
7495 (pass_optimize_widening_mul::execute): Call
7496 convert_expand_mult_copysign.
7498 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7500 PR tree-optimization/81354
7501 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
7502 Insert on edges rather than explicitly creating landing pads.
7503 (analyze_candidates_and_replace): Commit edge inserts.
7505 2017-08-08 Richard Biener <rguenther@suse.de>
7508 * tree-ssa-loop-niter.c: Include tree-dfa.h.
7509 (expand_simple_operations): Also look through ADDR_EXPRs with
7510 MEM_REF bases treating them as POINTER_PLUS_EXPR.
7512 2017-08-08 Richard Biener <rguenther@suse.de>
7514 PR tree-optimization/81723
7515 * tree-vect-slp.c (struct bst_traits): New hash traits.
7516 (bst_fail): New global.
7517 (vect_build_slp_tree_2): New worker, split out from ...
7518 (vect_build_slp_tree): ... this now wrapping it with using
7519 bst_fail set to cache SLP tree build fails. Properly handle
7521 (vect_analyze_slp_instance): Allocate and free bst_fail.
7523 2017-08-08 Martin Liska <mliska@suse.cz>
7526 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
7527 LABEL_DECLs that can be from a different function.
7529 2017-08-08 Bin Cheng <bin.cheng@arm.com>
7531 PR tree-optimization/81744
7532 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
7533 loop's number of iterations.
7535 2017-08-08 Martin Liska <mliska@suse.cz>
7537 * asan.c: Include header files.
7538 * attribs.c (build_decl_attribute_variant): New function moved
7540 (build_type_attribute_qual_variant): Likewise.
7541 (cmp_attrib_identifiers): Likewise.
7542 (simple_cst_list_equal): Likewise.
7543 (omp_declare_simd_clauses_equal): Likewise.
7544 (attribute_value_equal): Likewise.
7545 (comp_type_attributes): Likewise.
7546 (build_type_attribute_variant): Likewise.
7547 (lookup_ident_attribute): Likewise.
7548 (remove_attribute): Likewise.
7549 (merge_attributes): Likewise.
7550 (merge_type_attributes): Likewise.
7551 (merge_decl_attributes): Likewise.
7552 (merge_dllimport_decl_attributes): Likewise.
7553 (handle_dll_attribute): Likewise.
7554 (attribute_list_equal): Likewise.
7555 (attribute_list_contained): Likewise.
7556 * attribs.h (lookup_attribute): New function moved from tree.[ch].
7557 (lookup_attribute_by_prefix): Likewise.
7558 * bb-reorder.c: Include header files.
7559 * builtins.c: Likewise.
7560 * calls.c: Likewise.
7561 * cfgexpand.c: Likewise.
7562 * cgraph.c: Likewise.
7563 * cgraphunit.c: Likewise.
7564 * convert.c: Likewise.
7565 * dwarf2out.c: Likewise.
7566 * final.c: Likewise.
7567 * fold-const.c: Likewise.
7568 * function.c: Likewise.
7569 * gimple-expr.c: Likewise.
7570 * gimple-fold.c: Likewise.
7571 * gimple-pretty-print.c: Likewise.
7572 * gimple.c: Likewise.
7573 * gimplify.c: Likewise.
7574 * hsa-common.c: Likewise.
7575 * hsa-gen.c: Likewise.
7576 * internal-fn.c: Likewise.
7577 * ipa-chkp.c: Likewise.
7578 * ipa-cp.c: Likewise.
7579 * ipa-devirt.c: Likewise.
7580 * ipa-fnsummary.c: Likewise.
7581 * ipa-inline.c: Likewise.
7582 * ipa-visibility.c: Likewise.
7584 * lto-cgraph.c: Likewise.
7585 * omp-expand.c: Likewise.
7586 * omp-general.c: Likewise.
7587 * omp-low.c: Likewise.
7588 * omp-offload.c: Likewise.
7589 * omp-simd-clone.c: Likewise.
7590 * opts-global.c: Likewise.
7591 * passes.c: Likewise.
7592 * predict.c: Likewise.
7593 * sancov.c: Likewise.
7594 * sanopt.c: Likewise.
7595 * symtab.c: Likewise.
7596 * toplev.c: Likewise.
7597 * trans-mem.c: Likewise.
7598 * tree-chkp.c: Likewise.
7599 * tree-eh.c: Likewise.
7600 * tree-into-ssa.c: Likewise.
7601 * tree-object-size.c: Likewise.
7602 * tree-parloops.c: Likewise.
7603 * tree-profile.c: Likewise.
7604 * tree-ssa-ccp.c: Likewise.
7605 * tree-ssa-live.c: Likewise.
7606 * tree-ssa-loop.c: Likewise.
7607 * tree-ssa-sccvn.c: Likewise.
7608 * tree-ssa-structalias.c: Likewise.
7609 * tree-ssa.c: Likewise.
7610 * tree-streamer-in.c: Likewise.
7611 * tree-vectorizer.c: Likewise.
7612 * tree-vrp.c: Likewise.
7614 * ubsan.c: Likewise.
7615 * varasm.c: Likewise.
7616 * varpool.c: Likewise.
7617 * tree.c: Remove functions moved to attribs.[ch].
7619 * config/aarch64/aarch64.c: Add attrs.h header file.
7620 * config/alpha/alpha.c: Likewise.
7621 * config/arc/arc.c: Likewise.
7622 * config/arm/arm.c: Likewise.
7623 * config/avr/avr.c: Likewise.
7624 * config/bfin/bfin.c: Likewise.
7625 * config/c6x/c6x.c: Likewise.
7626 * config/cr16/cr16.c: Likewise.
7627 * config/cris/cris.c: Likewise.
7628 * config/darwin.c: Likewise.
7629 * config/epiphany/epiphany.c: Likewise.
7630 * config/fr30/fr30.c: Likewise.
7631 * config/frv/frv.c: Likewise.
7632 * config/ft32/ft32.c: Likewise.
7633 * config/h8300/h8300.c: Likewise.
7634 * config/i386/winnt.c: Likewise.
7635 * config/ia64/ia64.c: Likewise.
7636 * config/iq2000/iq2000.c: Likewise.
7637 * config/lm32/lm32.c: Likewise.
7638 * config/m32c/m32c.c: Likewise.
7639 * config/m32r/m32r.c: Likewise.
7640 * config/m68k/m68k.c: Likewise.
7641 * config/mcore/mcore.c: Likewise.
7642 * config/microblaze/microblaze.c: Likewise.
7643 * config/mips/mips.c: Likewise.
7644 * config/mmix/mmix.c: Likewise.
7645 * config/mn10300/mn10300.c: Likewise.
7646 * config/moxie/moxie.c: Likewise.
7647 * config/msp430/msp430.c: Likewise.
7648 * config/nds32/nds32-isr.c: Likewise.
7649 * config/nds32/nds32.c: Likewise.
7650 * config/nios2/nios2.c: Likewise.
7651 * config/nvptx/nvptx.c: Likewise.
7652 * config/pa/pa.c: Likewise.
7653 * config/pdp11/pdp11.c: Likewise.
7654 * config/powerpcspe/powerpcspe.c: Likewise.
7655 * config/riscv/riscv.c: Likewise.
7656 * config/rl78/rl78.c: Likewise.
7657 * config/rx/rx.c: Likewise.
7658 * config/s390/s390.c: Likewise.
7659 * config/sh/sh.c: Likewise.
7660 * config/sol2.c: Likewise.
7661 * config/sparc/sparc.c: Likewise.
7662 * config/spu/spu.c: Likewise.
7663 * config/stormy16/stormy16.c: Likewise.
7664 * config/tilegx/tilegx.c: Likewise.
7665 * config/tilepro/tilepro.c: Likewise.
7666 * config/v850/v850.c: Likewise.
7667 * config/vax/vax.c: Likewise.
7668 * config/visium/visium.c: Likewise.
7669 * config/xtensa/xtensa.c: Likewise.
7671 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
7674 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
7675 constraints since the -mupper-regs-* switches have been
7677 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
7678 into a vector from a double word element that was extracted from
7679 another vector, and eliminate extra XXPERMDI instructions.
7680 (vsx_concat_<mode>_2): Likewise.
7681 (vsx_concat_<mode>_3): Likewise.
7682 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
7683 concat to allow optimizing inserts from previous extracts.
7685 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
7687 * config/i386/i386.c (ix86_stack_protect_guard): Generate
7688 memory reference to a SSP offset in TLS address space.
7689 (ix86_print_operand) <case '@'>: Remove.
7690 (ix86_print_operand_punct_valid_p): Remove '@' code.
7691 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
7693 (stack_tls_protect_set_<mode>): Remove.
7694 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
7695 (stack_tls_protect_test_<mode>): Remove.
7696 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
7698 2017-08-07 Olivier Hainque <hainque@adacore.com>
7701 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
7703 2017-08-07 Douglas Rupp <rupp@adacore.com>
7705 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
7706 variable was referenced as multidir in command.
7708 2017-08-07 Jakub Jelinek <jakub@redhat.com>
7711 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
7714 2017-08-07 Martin Liska <mliska@suse.cz>
7716 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
7717 * config/rl78/rl78.c: Add include of attribs.h.
7718 * config/sh/sh.c: Likewise.
7719 * config/v850/v850.c: Likewise.
7721 2017-08-07 Tom de Vries <tom@codesourcery.com>
7724 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
7726 2017-08-07 Martin Liska <mliska@suse.cz>
7728 * config/mips/mips.c: Include attribs.h.
7730 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
7733 * doc/invoke.texi: Document change in behvaior for -Ofast for
7736 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
7738 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
7740 (aarch64_pop_regs): Likewise.
7741 (aarch64_gen_load_pair): Likewise.
7742 (aarch64_save_callee_saves): Likewise.
7743 (aarch64_restore_callee_saves): Likewise.
7745 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
7747 * config/i386/i386.c: Revert the last change.
7749 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
7752 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
7754 (ix86_finalize_stack_frame_flags): This. Also clear
7755 frame_pointer_needed if -fno-omit-frame-pointer is used without
7757 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
7758 with ix86_finalize_stack_frame_flags.
7759 (ix86_expand_epilogue): Likewise.
7760 (ix86_expand_split_stack_prologue): Likewise.
7762 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
7765 * config/i386/i386.c (get_builtin_code_for_version): Set priority
7766 to P_AES for Westmere.
7768 2017-08-07 Jonathan Yong <10walls@gmail.com>
7770 * config/i386/mingw.opt (fset-stack-executable): Removed.
7771 * config/i386/cygming.opt (fset-stack-executable): Moved
7773 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
7775 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
7777 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
7779 2017-08-07 Marek Polacek <polacek@redhat.com>
7782 * fold-const.c (fold_indirect_ref_1): Check type_domain.
7784 2017-08-07 Martin Liska <mliska@suse.cz>
7786 * attribs.h (canonicalize_attr_name): New function.
7787 (cmp_attribs): Move from c-format.c and adjusted.
7788 (is_attribute_p): Moved from tree.h.
7789 * tree-inline.c: Add new includes.
7790 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
7791 (private_is_attribute_p): Remove.
7792 (private_lookup_attribute): Likewise.
7793 (private_lookup_attribute_by_prefix): Simplify.
7794 (remove_attribute): Use is_attribute_p.
7795 * tree.h: Remove removed declarations.
7797 2017-08-07 Jakub Jelinek <jakub@redhat.com>
7800 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
7801 instead of computing it in the function. Formatting fix.
7802 (expand_case): Don't rely on default_edge being the first edge,
7803 clear it if removing it, pass default_edge to
7804 emit_case_dispatch_table.
7805 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
7808 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
7810 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
7811 insn in the function, emit NOP after the insn.
7813 2017-08-06 Tom de Vries <tom@codesourcery.com>
7815 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
7818 2017-08-06 Tom de Vries <tom@codesourcery.com>
7820 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
7823 2017-08-04 Yury Gribov <tetra2005@gmail.com>
7825 PR tree-optimization/57371
7826 * match.pd: New pattern.
7828 2017-08-04 Marek Polacek <polacek@redhat.com>
7831 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
7832 perform the computation in offset_int.
7834 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
7836 PR tree-optimization/81136
7837 * tree-vectorizer.h: Include tree-hash-traits.h.
7838 (vec_base_alignments): New typedef.
7839 (vec_info): Add a base_alignments field.
7840 (vect_record_base_alignments): Declare.
7841 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
7843 (DR_IS_CONDITIONAL_IN_STMT): New macro.
7844 (create_data_ref): Add an is_conditional_in_stmt argument.
7845 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
7846 the is_conditional_in_stmt field.
7847 (data_ref_loc): Add an is_conditional_in_stmt field.
7848 (get_references_in_stmt): Set the is_conditional_in_stmt field.
7849 (find_data_references_in_stmt): Update call to create_data_ref.
7850 (graphite_find_data_references_in_stmt): Likewise.
7851 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
7852 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
7853 (vect_record_base_alignment): New function.
7854 (vect_record_base_alignments): Likewise.
7855 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
7856 for nested statements even if we fail to compute a misalignment.
7857 Use pooled base alignments for unconditional references.
7858 (vect_find_same_alignment_drs): Compare base addresses instead
7860 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
7861 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
7863 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
7865 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
7866 Add an explicit name for the enum. Use auto_vec for slp_instances
7868 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
7870 (_bb_vec_info): Add a constructor and destructor.
7871 (vinfo_for_stmt): Return NULL for uids of -1 as well.
7872 (destroy_loop_vec_info): Delete.
7873 (vect_destroy_datarefs): Likewise.
7874 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
7875 (vec_info::vec_info): New function.
7876 (vec_info::~vec_info): Likewise.
7877 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
7878 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
7879 destroy_loop_vec_info.
7880 * tree-vect-loop.c (new_loop_vec_info): Replace with...
7881 (_loop_vec_info::_loop_vec_info): ...this.
7882 (destroy_loop_vec_info): Replace with...
7883 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
7884 the stmt_vec_infos. Leave handling of vec_info information to its
7885 destructor. Remove explicit vector releases.
7886 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
7887 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
7888 * tree-vect-slp.c (new_bb_vec_info): Replace with...
7889 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
7890 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
7891 (destroy_bb_vec_info): Replace with...
7892 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
7893 information to its destructor.
7894 (vect_slp_analyze_bb_1): Use new and delete instead of
7895 new_bb_vec_info and destroy_bb_vec_info.
7896 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
7899 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
7901 * tree-data-ref.h (subscript): Add access_fn field.
7902 (data_dependence_relation): Add could_be_independent_p.
7903 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
7904 (same_access_functions): Move to tree-data-ref.c.
7905 * tree-data-ref.c (ref_contains_union_access_p): New function.
7906 (access_fn_component_p): Likewise.
7907 (access_fn_components_comparable_p): Likewise.
7908 (dr_analyze_indices): Add a reference to access_fn_component_p.
7909 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
7911 (constant_access_functions): Likewise.
7912 (add_other_self_distances): Likewise.
7913 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
7914 (initialize_data_dependence_relation): Use XCNEW and remove
7915 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
7916 of access functions that have the same type. Allow the
7917 subsequence to end with different bases in some circumstances.
7918 Record the chosen access functions in SUB_ACCESS_FN.
7919 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
7920 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
7921 (subscript_dependence_tester_1): Likewise dra and drb.
7922 (build_classic_dist_vector): Update calls accordingly.
7923 (subscript_dependence_tester): Likewise.
7924 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
7925 DDR_COULD_BE_INDEPENDENT_P.
7926 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
7927 comp_alias_ddrs instead of may_alias_ddrs.
7928 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
7930 (vect_analyze_data_ref_dependence): Use it if
7931 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
7932 distance vectors if that fails.
7933 (dependence_distance_ge_vf): New function.
7934 (vect_prune_runtime_alias_test_list): Use it. Don't clear
7935 LOOP_VINFO_MAY_ALIAS_DDRS.
7937 2017-08-04 Richard Biener <rguenther@suse.de>
7940 * fold-const.c (fold_binary_loc): Properly restrict
7941 minus_var0 && minus_var1 case when associating undefined overflow
7944 2017-08-04 Tom de Vries <tom@codesourcery.com>
7946 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
7948 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7950 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7951 Don't start diagnostic messages with a capital letter.
7952 * config/rs6000/rs6000.c (rs6000_option_override_internal):
7954 (rs6000_invalid_builtin): Likewise.
7955 (rs6000_trampoline_init): Likewise.
7957 2017-08-03 Jakub Jelinek <jakub@redhat.com>
7960 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
7961 after setting changeable df flags.
7963 2017-08-03 Richard Biener <rguenther@suse.de>
7965 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
7966 up if the use is in USE - X.
7968 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
7970 * toplev.c (dumpfile.h): New include.
7971 (internal_error_reentered): New static function. Use it...
7972 (internal_error_function): ...here to handle reentered internal_error.
7974 2017-08-03 Richard Biener <rguenther@suse.de>
7977 * fold-const.c (split_tree): Add minus_var and minus_con
7978 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
7979 here but always use minus_*.
7980 (associate_trees): Assert we never associate with MINUS_EXPR
7981 and NULL first operand. Do not recurse for PLUS_EXPR operands
7982 when associating as MINUS_EXPR either.
7983 (fold_binary_loc): Track minus_var and minus_con.
7985 2017-08-03 Tom de Vries <tom@codesourcery.com>
7988 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
7989 ACCEL_COMPILER, apply finish_options on
7990 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
7992 2017-08-03 Tom de Vries <tom@codesourcery.com>
7995 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
7996 function_entry_patch_area_size > 0.
7998 2017-08-03 Jakub Jelinek <jakub@redhat.com>
8001 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
8002 instead of 10??LU, perform unit multiplication in wide_int,
8003 don't change alloc_object_size_limit if the limit is larger
8006 PR tree-optimization/81655
8007 PR tree-optimization/81588
8008 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
8009 the case when ranges[i].low and high are 1 for unsigned type with
8013 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
8014 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
8016 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
8018 * tree-vrp.h: Add include guard.
8020 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
8023 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
8024 (ud2): New insn pattern.
8025 * config/i386/i386.c (ix86_expand_epilogue):
8026 For naked functions, generate ud2 instead of trap insn.
8028 2017-08-02 Marek Polacek <polacek@redhat.com>
8031 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
8033 2017-08-02 Tom de Vries <tom@codesourcery.com>
8034 Cesar Philippidis <cesar@codesourcery.com>
8036 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
8037 Add missing edge probabilities.
8039 2017-08-02 Tamar Christina <tamar.christina@arm.com>
8041 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
8044 2017-08-02 Jakub Jelinek <jakub@redhat.com>
8047 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
8048 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
8049 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
8050 prologue_seq sequences - if any.
8052 2017-08-02 Richard Biener <rguenther@suse.de>
8054 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
8055 via vectors if supported, integer extracts via punning if supported
8056 or otherwise vector extracts.
8058 2017-08-02 Richard Biener <rguenther@suse.de>
8060 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
8062 (bitmap_insert_into_set): ... this.
8064 2017-08-02 Richard Biener <rguenther@suse.de>
8066 PR tree-optimization/81633
8068 2015-08-17 Alan Hayward <alan.hayward@arm.com>
8070 PR tree-optimization/71752
8071 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
8073 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
8075 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
8076 (machine_function::call_ms2sysv_pad_out): Remove field.
8077 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
8078 (ix86_compute_frame_layout): Likewise.
8080 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
8083 * config/i386/i386.c (ix86_set_func_type): Disallow naked
8084 attribute with interrupt attribute.
8086 2017-08-01 Andrew Pinski <apinski@cavium.com>
8088 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
8089 BIT_INSERT_EXPR's operand 1
8090 to see if the types precision matches.
8092 2017-08-01 Martin Liska <mliska@suse.cz>
8095 * builtins.c (expand_builtin_memcpy_args): Remove.
8096 (expand_builtin_memcpy): Call newly added function
8097 expand_builtin_memory_copy_args.
8098 (expand_builtin_memcpy_with_bounds): Likewise.
8099 (expand_builtin_mempcpy): Remove last argument.
8100 (expand_builtin_mempcpy_with_bounds): Likewise.
8101 (expand_builtin_memory_copy_args): New function created from
8102 expand_builtin_mempcpy_args with small modifications.
8103 (expand_builtin_mempcpy_args): Remove.
8104 (expand_builtin_stpcpy): Remove unused argument.
8105 (expand_builtin): Likewise.
8106 (expand_builtin_with_bounds): Likewise.
8108 2017-08-01 Martin Liska <mliska@suse.cz>
8111 Make mempcpy more optimal (PR middle-end/70140).
8113 2017-08-01 Jakub Jelinek <jakub@redhat.com>
8116 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
8117 __builtin_vec_cmpne verify both arguments are compatible vectors
8118 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
8119 verify arg1_type is a pointer or array type. For __builtin_vec_st,
8120 move computation of aligned to after checking the argument types.
8124 * config/rs6000/vsx.md (vextract_fp_from_shorth,
8125 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
8128 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
8129 Doug Rupp <rupp@adacore.com>
8130 Olivier Hainque <hainque@adacore.com>
8132 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
8133 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
8135 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
8136 for TARGET_OS_CPP_BUILTIN.
8137 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
8138 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
8140 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
8141 passing required abi options to the assembler for EABI configurations.
8142 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
8143 of .text.hot and .text.unlikely sections for kernel modules when
8144 using ARM style exceptions.
8145 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
8146 options. Add EXTRA_CC1_SPEC.
8147 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
8149 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
8151 (ARM_TARGET2_DWARF_FORMAT): Define.
8152 * config/arm/t-vxworks: Adjust multilib control to removal of the
8153 Diab command line options.
8155 2017-08-01 Martin Liska <mliska@suse.cz>
8157 PR gcov-profile/81561
8158 * gcov.c (unblock): Make unblocking safe as we need to preserve
8159 index correspondence of blocks and block_lists.
8161 2017-08-01 Richard Biener <rguenther@suse.de>
8163 PR tree-optimization/81181
8164 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
8165 (compute_antic): ... end of iteration here.
8167 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
8169 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
8170 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
8171 (ftree-slp-vectorize): Likewise.
8172 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
8173 can no longer be set independent of flag_tree_loop_vectorize.
8174 * omp-general.c (emp_max_vf): Likewise.
8175 * opts.c (enable_fdo_optimizations): Remove references to
8176 flag_tree_vectorize, these are now implicit.
8177 (common_handle_option): Remove handling for OPT_ftree_vectorize,
8178 and leave it for the options machinery.
8180 2017-08-01 Martin Liska <mliska@suse.cz>
8183 * builtins.c (expand_builtin_memcpy_args): Remove.
8184 (expand_builtin_memcpy): Call newly added function
8185 expand_builtin_memory_copy_args.
8186 (expand_builtin_memcpy_with_bounds): Likewise.
8187 (expand_builtin_mempcpy): Remove last argument.
8188 (expand_builtin_mempcpy_with_bounds): Likewise.
8189 (expand_builtin_memory_copy_args): New function created from
8190 expand_builtin_mempcpy_args with small modifications.
8191 (expand_builtin_mempcpy_args): Remove.
8192 (expand_builtin_stpcpy): Remove unused argument.
8193 (expand_builtin): Likewise.
8194 (expand_builtin_with_bounds): Likewise.
8196 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
8199 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
8200 print "ds:" only for immediates in generic address space.
8202 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
8205 * config/i386/i386.c (ix86_funciton_naked): New prototype.
8206 (ix86_function_ok_for_sibcall): Return false for naked functions.
8208 2017-08-01 Richard Biener <rguenther@suse.de>
8210 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
8211 (compute_antic): Seed worklist with exit block predecessors.
8212 * cfganal.c (dfs_find_deadend): For a cycle return the source
8213 of the edge closing it.
8215 2017-08-01 Tamar Christina <tamar.christina@arm.com>
8217 * config/aarch64/aarch64.c
8218 (aarch64_can_const_movi_rtx_p): Move 0 check.
8220 2017-08-01 Bin Cheng <bin.cheng@arm.com>
8222 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
8223 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
8225 * match.pd: Ditto in address comparison pattern.
8227 2017-08-01 Bin Cheng <bin.cheng@arm.com>
8229 PR tree-optimization/81627
8230 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
8231 closed ssa form for store-store chain.
8233 2017-08-01 Bin Cheng <bin.cheng@arm.com>
8235 PR tree-optimization/81620
8236 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
8237 for store-store chain.
8239 2017-08-01 Jakub Jelinek <jakub@redhat.com>
8241 PR tree-optimization/81588
8242 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
8243 ranges[i].in_p, invert comparison code ccode. For >/>=,
8244 swap rhs1 and rhs2 and comparison code unconditionally,
8245 for </<= don't do that. Don't swap rhs1/rhs2 again if
8246 ranges[i].in_p, instead invert comparison code ccode if
8247 opcode or oe->rank is BIT_IOR_EXPR.
8250 * optabs.def (vec_extract_optab, vec_init_optab): Change from
8251 a direct optab to conversion optab.
8252 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
8253 with GET_MODE_INNER as last argument instead of optab_handler.
8254 * expmed.c (extract_bit_field_1): Likewise. Use vector from
8255 vector extraction if possible and optab is available.
8256 * expr.c (store_constructor): Use convert_optab_handler instead
8257 of optab_handler. Use vector initialization from smaller
8258 vectors if possible and optab is available.
8259 * tree-vect-stmts.c (vectorizable_load): Likewise.
8260 * doc/md.texi (vec_extract, vec_init): Document that the optabs
8262 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
8263 of vec_init from half-sized vectors with the same element mode.
8264 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
8265 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
8266 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
8267 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
8268 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
8269 after mode in gen_vec_extract* calls.
8270 (vec_extract<mode>): Renamed to ...
8271 (vec_extract<mode><ssescalarmodelower>): ... this.
8272 (vec_extract<mode><ssehalfvecmodelower>): New expander.
8273 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
8274 element mode after mode in gen_vec_init* calls.
8275 (VEC_INIT_HALF_MODE): New mode iterator.
8276 (vec_init<mode>): Renamed to ...
8277 (vec_init<mode><ssescalarmodelower>): ... this.
8278 (vec_init<mode><ssehalfvecmodelower>): New expander.
8279 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
8280 (vec_extractv2sfsf): ... this.
8281 (vec_initv2sf): Renamed to ...
8282 (vec_initv2sfsf): ... this.
8283 (vec_extractv2si): Renamed to ...
8284 (vec_extractv2sisi): ... this.
8285 (vec_initv2si): Renamed to ...
8286 (vec_initv2sisi): ... this.
8287 (vec_extractv4hi): Renamed to ...
8288 (vec_extractv4hihi): ... this.
8289 (vec_initv4hi): Renamed to ...
8290 (vec_initv4hihi): ... this.
8291 (vec_extractv8qi): Renamed to ...
8292 (vec_extractv8qiqi): ... this.
8293 (vec_initv8qi): Renamed to ...
8294 (vec_initv8qiqi): ... this.
8295 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
8296 (vec_init<mode>): Renamed to ...
8297 (vec_init<mode><VEC_base_l>): ... this.
8298 (vec_extract<mode>): Renamed to ...
8299 (vec_extract<mode><VEC_base_l>): ... this.
8300 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
8301 (vec_initv2sfsf): ... this.
8302 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
8303 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
8304 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
8305 element mode after mode in gen_vec_init* calls.
8306 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
8307 (vec_init<mode><Vel>): ... this.
8308 (vec_extract<mode>): Renamed to ...
8309 (vec_extract<mode><Vel>): ... this.
8310 * config/aarch64/iterators.md (Vel): New mode attribute.
8311 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
8312 Add element mode after mode in gen_vec_extract* calls.
8313 * config/s390/vector.md (non_vec_l): New mode attribute.
8314 (vec_extract<mode>): Renamed to ...
8315 (vec_extract<mode><non_vec_l>): ... this.
8316 (vec_init<mode>): Renamed to ...
8317 (vec_init<mode><non_vec_l>): ... this.
8318 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
8319 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
8321 * config/arm/iterators.md (V_elem_l): New mode attribute.
8322 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
8323 (vec_extract<mode><V_elem_l>): ... this.
8324 (vec_extractv2di): Renamed to ...
8325 (vec_extractv2didi): ... this.
8326 (vec_init<mode>): Renamed to ...
8327 (vec_init<mode><V_elem_l>): ... this.
8328 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
8329 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
8330 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
8331 Add element mode after gen_vec_extract* calls.
8332 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
8333 (vec_init<mode><unitmode>): ... this.
8334 (vec_extract<mode>): Renamed to ...
8335 (vec_extract<mode><unitmode>): ... this.
8336 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
8337 (vec_init<mode><unitmode>): ... this.
8338 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
8339 (vec_initv2sfsf): ... this.
8340 (vec_extractv2sf): Renamed to ...
8341 (vec_extractv2sfsf): ... this.
8342 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
8343 Add element mode after gen_vec_extract* calls.
8344 * config/mips/mips.md (unitmode): New mode iterator.
8345 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
8346 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
8347 * config/spu/spu.md (inner_l): New mode attribute.
8348 (vec_init<mode>): Renamed to ...
8349 (vec_init<mode><inner_l>): ... this.
8350 (vec_extract<mode>): Renamed to ...
8351 (vec_extract<mode><inner_l>): ... this.
8352 * config/sparc/sparc.md (veltmode): New mode iterator.
8353 (vec_init<VMALL:mode>): Renamed to ...
8354 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
8355 * config/ia64/vect.md (vec_initv2si): Renamed to ...
8356 (vec_initv2sisi): ... this.
8357 (vec_initv2sf): Renamed to ...
8358 (vec_initv2sfsf): ... this.
8359 (vec_extractv2sf): Renamed to ...
8360 (vec_extractv2sfsf): ... this.
8361 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
8362 (vec_init<mode>): Renamed to ...
8363 (vec_init<mode><VEC_base_l>): ... this.
8364 (vec_extract<mode>): Renamed to ...
8365 (vec_extract<mode><VEC_base_l>): ... this.
8366 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
8367 (vec_initv2sfsf): ... this.
8368 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
8369 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
8370 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
8371 gen_vec_init* calls.
8373 2017-08-01 Richard Biener <rguenther@suse.de>
8375 PR tree-optimization/81297
8376 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
8377 TREE_OVERFLOW from INTEGER_CSTs.
8379 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
8381 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
8383 2017-07-31 Carl Love <cel@us.ibm.com>
8385 * config/rs6000/rs6000-c: Add support for built-in functions
8386 vector signed char vec_xl_be (signed long long, signed char *);
8387 vector unsigned char vec_xl_be (signed long long, unsigned char *);
8388 vector signed int vec_xl_be (signed long long, signed int *);
8389 vector unsigned int vec_xl_be (signed long long, unsigned int *);
8390 vector signed long long vec_xl_be (signed long long, signed long long *);
8391 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
8392 vector signed short vec_xl_be (signed long long, signed short *);
8393 vector unsigned short vec_xl_be (signed long long, unsigned short *);
8394 vector double vec_xl_be (signed long long, double *);
8395 vector float vec_xl_be (signed long long, float *);
8396 * config/rs6000/altivec.h (vec_xl_be): Add #define.
8397 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
8398 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
8400 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
8401 (altivec_expand_builtin): Add switch statement to call
8402 altivec_expand_xl_be for each builtin.
8403 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
8404 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
8405 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
8406 __builtin_vsx_le_be_v16qi.
8407 * doc/extend.texi: Update the built-in documentation file for the
8408 new built-in functions.
8410 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
8413 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
8415 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
8417 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8419 * config.gcc: Add z14.
8420 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
8421 CPU model numbers for z13s and z14.
8422 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
8424 * config/s390/s390-opts.h (enum processor_type): Rename
8425 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
8426 * config/s390/s390.c (processor_table): Add field for CPU name to
8427 be passed to Binutils.
8428 (s390_asm_output_machine_for_arch): Use the new field in
8429 processor_table for Binutils.
8430 (s390_expand_builtin): Replace arch12 with z14.
8431 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
8432 (s390_get_sched_attrmask): Likewise.
8433 (s390_get_unit_mask): Likewise.
8434 * config/s390/s390.opt: Add z14 to processor_type enum.
8436 2017-07-31 Martin Jambor <mjambor@suse.cz>
8439 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
8440 regardless of optimization level.
8442 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
8443 Martin Liska <mliska@suse.cz>
8445 * predict.def: Remove old comment and adjust probability.
8446 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
8449 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
8452 * config/i386/i386.c (ix86_function_naked): New function.
8453 (ix86_can_use_return_insn_p): Return false for naked functions.
8454 (ix86_expand_prologue): Skip prologue for naked functions.
8455 (ix86_expand_epilogue): Skip epilogue for naked functions
8456 and emit trap instruction.
8457 (ix86_warn_func_return): New function.
8458 (ix86_attribute_table): Add "naked" attribute specification.
8459 (TARGET_WARN_FUNC_RETURN): Define.
8460 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
8462 2017-07-31 Martin Liska <mliska@suse.cz>
8464 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
8465 (dump_gimple_bb_header): Always dump BB info.
8466 (pp_cfg_jump): Do not append info about BB when dumping a jump.
8468 2017-07-31 Martin Liska <mliska@suse.cz>
8471 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
8472 also with current_function_decl non-null equality.
8474 2017-07-31 Jakub Jelinek <jakub@redhat.com>
8477 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
8478 change type to the element type, instead add eltype variable and
8479 use it where we are interested in the element type.
8481 PR tree-optimization/81603
8482 * ipa-polymorphic-call.c
8483 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
8484 offset arithmetic in offset_int, bail out if the resulting bit offset
8485 doesn't fit into shwi.
8487 2017-07-31 Martin Liska <mliska@suse.cz>
8489 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
8490 (gimplify_save_expr): Fix comment.
8492 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
8495 * config/i386/i386.c (ix86_function_arg): Update arguments for
8497 (ix86_compute_frame_layout): Set the initial stack offset to
8498 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
8499 INCOMING_FRAME_SP_OFFSET.
8500 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
8501 stack before exception handler returns.
8502 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
8503 the 'ERROR_CODE' for exception handler.
8505 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
8507 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
8508 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
8509 (ASM_OUTPUT_REG_POP): Ditto.
8510 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
8511 instead of asm_fprintf to output pure string.
8513 2017-07-29 Jakub Jelinek <jakub@redhat.com>
8515 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
8516 to imported_module_or_decl hook.
8517 (debug_nothing_tree_tree_tree_bool): Remove.
8518 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
8519 * debug.c (do_nothing_debug_hooks): Use
8520 debug_nothing_tree_tree_tree_bool_bool instead of
8521 debug_nothing_tree_tree_tree_bool.
8522 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
8523 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
8524 * sdbout.c (sdb_debug_hooks): Likewise.
8525 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
8526 (gen_namespace_die): Add DW_AT_export_symbols attribute if
8528 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
8529 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
8530 attribute, don't add anything.
8532 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8534 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
8535 (fold_build2_stat_loc): Likewise.
8536 (fold_build3_stat_loc): Likewise.
8537 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
8538 (fold_build1_loc): Remove macro.
8539 (fold_build2_loc): Likewise.
8540 (fold_build3_loc): Likewise.
8542 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8544 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
8545 (gimple_build_debug_bind_source_stat): Likewise.
8546 * gimple.h (gimple_build_debug_bind): Remove macro.
8547 (gimple_build_debug_bind_source): Likewise.
8549 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8551 * bitmap.c (bitmap_alloc): Adjust.
8552 (bitmap_gc_alloc): Likewise.
8553 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
8555 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8557 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
8558 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
8559 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
8560 (bitmap_gc_alloc_stat): Likewise.
8561 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
8563 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8565 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
8566 * rtl.h (shallow_copy_rtx): Remove macro.
8568 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8570 * emit-rtl.c (gen_raw_REG): Adjust.
8571 * gengenrtl.c (gendef): Likewise.
8572 * rtl.c (rtx_alloc_stat): Remove _stat from name.
8573 * rtl.h (rtx_alloc): Remove macro.
8575 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8577 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
8578 (build_tree_list_stat): Likewise.
8579 * tree.h (build_tree_list): Remove macro.
8580 (build_tree_list_vec): Likewise.
8582 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8584 * tree.c (make_vector_stat): Remove _stat from name.
8585 (build_vector_stat): Likewise.
8586 * tree.h (make_vector_stat): Remove macro.
8587 (build_vector_stat): Likewise.
8589 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8591 * tree.h (build_var_debug_value): Remove prototype.
8593 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8595 * tree.c (tree_cons_stat): Remove _stat from name.
8596 * tree.h (tree_cons): Remove macro.
8598 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8600 * tree.c (build_vl_exp_stat): Remove _stat from name.
8601 * tree.h (build_vl_exp): Remove macro.
8603 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8605 * tree.c (build_decl_stat): Remove _stat from name.
8606 * tree.h (build_decl): Remove macro.
8608 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8610 * gimple.c (gimple_build_with_ops_stat): Adjust.
8611 (gimple_alloc_stat): Remove _stat from name.
8612 * gimple.h (gimple_alloc): Remove macro.
8614 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8616 * tree.c (make_tree_vec_stat): Remove _stat from name.
8617 (grow_tree_vec_stat): Likewise.
8618 * tree.h (make_tree_vec_stat): Adjust prototype.
8619 (grow_tree_vec_stat): Likewise.
8620 (make_tree_vec): Remove macro.
8621 (grow_tree_vec): Likewise.
8623 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8625 * fold-const.c (fold_build1_stat_loc): Adjust.
8626 (fold_build2_stat_loc): Likewise.
8627 (fold_build3_stat_loc): Likewise.
8628 * tree.c (build0_stat): Remove _stat from name.
8629 (build1_stat): Likewise.
8630 (build2_stat): Likewise.
8631 (build3_stat): Likewise.
8632 (build4_stat): Likewise.
8633 (build5_stat): Likewise.
8634 * tree.h (build1_loc): Remove macro, and rename _stat function
8636 (build2_loc): Likewise.
8637 (build3_loc): Likewise.
8638 (build4_loc): Likewise.
8639 (build5_loc): Likewise.
8641 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8643 * tree.c (make_int_cst_stat): Remove _stat from name.
8644 * tree.h (make_int_cst_stat): Adjust prototype.
8645 (make_int_cst): Remove macro.
8647 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8649 * tree.c (make_tre_binfo_stat): Remove _stat from name.
8650 * tree.h (make_tree_binfo_stat): Adjust prototype.
8651 (make_tree_binfo): Remove.
8653 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8655 * tree.c (copy_node_stat): Rename to copy_node.
8656 (build_distinct_type_copy): Adjust.
8657 * tree.h (copy_node_stat): Adjust prototype.
8658 (copy_node): Remove macro.
8660 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8662 * tree.c (make_node_stat): rename to make_node.
8663 (build_tree_list_stat): Adjust.
8664 (build0_stat): Likewise.
8665 (build2_stat): Likewise.
8666 (build3_stat): Likewise.
8667 (build4_stat): Likewise.
8668 (build5_stat): Likewise.
8669 (build_decl_stat): Likewise.
8670 * tree.h (make_node_stat): Adjust prototype.
8671 (make_node): remove macro.
8673 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
8675 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
8676 (PPC_FEATURE2_SCV): Likewise.
8677 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
8679 2017-07-28 Tamar Christina <tamar.christina@arm.com>
8681 * config/aarch64/aarch64.c
8682 (aarch64_internal_mov_immediate): Add new special pattern.
8683 * config/aarch64/aarch64.md (*movdi_aarch64):
8684 Add reg/32bit const mov case.
8686 2017-07-28 Tamar Christina <tamar.christina@arm.com>
8687 Richard Sandiford <richard.sandiford@linaro.org>
8689 * config/aarch64/aarch64.md (mov<mode>): Generalize.
8690 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
8691 Add integer and movi cases.
8692 (movi-split-hf-df-sf split, fp16): New.
8693 (enabled): Added TARGET_FP_F16INST.
8694 * config/aarch64/iterators.md (GPF_HF): New.
8695 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
8697 2017-07-28 Tamar Christina <tamar.christina@arm.com>
8699 * config/aarch64/aarch64.c
8700 (aarch64_simd_container_mode): Add prototype.
8701 (aarch64_expand_mov_immediate): Add HI support.
8702 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
8703 (aarch64_can_const_movi_rtx_p): New.
8704 (aarch64_preferred_reload_class):
8705 Remove restrictions of using FP registers for certain SIMD operations.
8706 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
8707 (aarch64_valid_floating_const): Add integer move validation.
8708 (aarch64_simd_imm_scalar_p): Remove.
8709 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
8710 (aarch64_legitimate_constant_p): Expand list of supported cases.
8711 * config/aarch64/aarch64-protos.h
8712 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
8713 (aarch64_reinterpret_float_as_int): New.
8714 (aarch64_simd_imm_scalar_p): Remove.
8715 * config/aarch64/constraints.md (Uvi): New.
8716 (Dd): Split into Ds and new Dd.
8717 * config/aarch64/aarch64.md (*movsi_aarch64):
8719 (*movdi_aarch64): Add SIMD mov case.
8721 2017-07-28 Bin Cheng <bin.cheng@arm.com>
8723 * tree-predcom.c: (struct chain): Handle store-store chain in which
8724 stores for elimination only store loop invariant values.
8725 (execute_pred_commoning_chain): Ditto.
8726 (prepare_initializers_chain_store_elim): Ditto.
8727 (prepare_finalizers): Ditto.
8728 (is_inv_store_elimination_chain): New function.
8729 (initialize_root_vars_store_elim_1): New function.
8731 2017-07-28 Bin Cheng <bin.cheng@arm.com>
8733 * tree-predcom.c: Revise general description of the pass.
8734 (enum chain_type): New enum type for store elimination.
8735 (struct chain): New field supporting store elimination.
8736 (struct component): Ditto.
8737 (dump_chain): Dump store-stores chain.
8738 (release_chain): Release resources.
8739 (split_data_refs_to_components): Compute and create component
8740 contains only stores for elimination.
8741 (get_chain_last_ref_at): New function.
8742 (make_invariant_chain): Initialization.
8743 (make_rooted_chain): Specify chain type in parameter and record it.
8744 (add_looparound_copies): Skip for store-stores chain.
8745 (determine_roots_comp): Compute type of chain and pass it to
8747 (initialize_root_vars_store_elim_2): New function.
8748 (finalize_eliminated_stores): New function.
8749 (remove_stmt): Handle store for elimination.
8750 (execute_pred_commoning_chain): Execute predictive commoning on
8752 (determine_unroll_factor): Skip unroll for store-stores chain.
8753 (prepare_initializers_chain_store_elim): New function.
8754 (prepare_initializers_chain): Hanlde store-store chain.
8755 (prepare_finalizers_chain, prepare_finalizers): New function.
8756 (tree_predictive_commoning_loop): Return integer value indicating
8757 if loop is unrolled or lcssa form is corrupted.
8758 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
8760 2017-07-28 Bin Cheng <bin.cheng@arm.com>
8762 * tree-predcom.c (initialize_root): Delete.
8763 (execute_pred_commoning_chain): Initialize root vars and replace
8764 reference of non-combined chain directly, rather than call above
8767 2017-07-28 Bin Cheng <bin.cheng@arm.com>
8769 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
8770 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
8772 2017-07-28 Bin Cheng <bin.cheng@arm.com>
8774 * tree-predcom.c (struct chain): New field init_seq.
8775 (release_chain): Release init_seq.
8776 (prepare_initializers_chain): Record intialization stmts in above
8778 (insert_init_seqs): New function.
8779 (tree_predictive_commoning_loop): Call insert_init_seqs.
8781 2017-07-28 Bin Cheng <bin.cheng@arm.com>
8783 * tree-predcom.c (determine_roots_comp): Skip trivial components.
8785 2017-07-28 Richard Biener <rguenther@suse.de>
8787 * match.pd: Remove superfluous :c.
8788 * genmatch.c (simplify::id): Add member.
8789 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
8791 (current_id): New global.
8792 (dt_node::parent): Move from ...
8793 (dt_operand::parent): ... here. Add for_id member.
8794 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
8795 (decision_tree::find_node): Relax order requirement when
8796 merging DT_TRUE nodes to ones inbetween the current simplify
8797 and the one we try to merge with. Add diagnostic whenever
8798 we need to enforce pattern order by not merging.
8799 (decision_tree::insert): Set current_id.
8800 (decision_tree::print_node): Dump parent node and for_id.
8801 (parser::last_id): Add member.
8802 (parser::push_simplify): Assign unique id.
8803 (parser::parser): Initialize last_id.
8805 2017-07-28 Martin Liska <mliska@suse.cz>
8808 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
8809 gimple_build_debug_bind.
8811 2017-07-28 Richard Biener <rguenther@suse.de>
8813 PR tree-optimization/81502
8814 * match.pd: Add pattern combining BIT_INSERT_EXPR with
8816 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
8818 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
8819 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
8820 for BIT_FIELD_REF args.
8821 * fold-const.c (make_bit_field_ref): Likewise.
8822 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
8824 2017-07-28 Jakub Jelinek <jakub@redhat.com>
8827 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
8828 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
8829 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
8830 Or it into SANITIZER_UNDEFINED.
8831 * ubsan.c: Include gimple-fold.h and varasm.h.
8832 (ubsan_expand_ptr_ifn): New function.
8833 (instrument_pointer_overflow): New function.
8834 (maybe_instrument_pointer_overflow): New function.
8835 (instrument_object_size): Formatting fix.
8836 (pass_ubsan::execute): Call instrument_pointer_overflow
8837 and maybe_instrument_pointer_overflow.
8838 * internal-fn.c (expand_UBSAN_PTR): New function.
8839 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
8840 * sanitizer.def (__ubsan_handle_pointer_overflow,
8841 __ubsan_handle_pointer_overflow_abort): New builtins.
8842 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
8843 * internal-fn.def (UBSAN_PTR): New internal function.
8844 * opts.c (sanitizer_opts): Add pointer-overflow.
8845 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
8846 * fold-const.c (build_range_check): Compute pointer range check in
8847 integral type if pointer arithmetics would be needed. Formatting
8850 2017-07-28 Martin Liska <mliska@suse.cz>
8853 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
8854 parameters that are of a variable-length.
8856 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
8858 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
8860 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
8861 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
8862 (CRT_CALL_STATIC_FUNCTION): Likewise.
8863 (ASM_DEFAULT_SPEC): New define.
8864 (ASM_SPEC32): Likewise.
8865 (ASM_SPEC64): Likewise.
8866 (ASM_SPEC_COMMON): Likewise.
8867 (ASM_SPEC): Likewise.
8868 (INVALID_64BIT): Likewise.
8869 (LINK_OS_DEFAULT_SPEC): Likewise.
8870 (LINK_OS_SPEC32): Likewise.
8871 (LINK_OS_SPEC64): Likewise.
8872 (POWERPC_LINUX): Likewise.
8873 (PTRDIFF_TYPE): Likewise.
8874 (RESTORE_FP_PREFIX): Likewise.
8875 (RESTORE_FP_SUFFIX): Likewise.
8876 (SAVE_FP_PREFIX): Likewise.
8877 (SAVE_FP_SUFFIX): Likewise.
8878 (SIZE_TYPE): Likewise.
8879 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
8880 (TARGET_64BIT): Likewise.
8881 (TARGET_64BIT): Likewise.
8882 (TARGET_AIX): Likewise.
8883 (WCHAR_TYPE_SIZE): Likewise.
8884 (WCHAR_TYPE): Undefine.
8885 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
8886 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
8887 (CPP_OS_RTEMS_SPEC): Delete.
8888 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
8889 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
8891 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
8893 2017-07-28 Jakub Jelinek <jakub@redhat.com>
8895 PR tree-optimization/81578
8896 * tree-parloops.c (build_new_reduction): Bail out if
8897 reduction_code isn't one of the standard OpenMP reductions.
8898 Move the details printing after that decision.
8900 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
8902 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
8903 related to reload_in_progress.
8904 (splat_input_operand): Likewise.
8905 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
8907 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
8909 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
8910 (TARGET_INSTANTIATE_DECLS): Likewise.
8911 (legitimate_indexed_address_p): Delete reload_in_progress code.
8912 (rs6000_debug_legitimate_address_p): Likewise.
8913 (rs6000_eliminate_indexed_memrefs): Likewise.
8914 (rs6000_emit_le_vsx_store): Likewise.
8915 (rs6000_emit_move_si_sf_subreg): Likewise.
8916 (rs6000_emit_move): Likewise.
8917 (register_to_reg_type): Likewise.
8918 (rs6000_pre_atomic_barrier): Likewise.
8919 (rs6000_machopic_legitimize_pic_address): Likewise.
8920 (rs6000_allocate_stack_temp): Likewise.
8921 (rs6000_address_for_fpconvert): Likewise.
8922 (rs6000_address_for_altivec): Likewise.
8923 (rs6000_secondary_memory_needed_rtx): Delete function.
8924 (rs6000_check_sdmode): Likewise.
8925 (rs6000_alloc_sdmode_stack_slot): Likewise.
8926 (rs6000_instantiate_decls): Likewise.
8927 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
8928 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
8929 Delete reload_in_progress.
8930 (*vec_reload_and_plus_<mptrsize>): Likewise.
8931 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
8932 (vsx_div_v2di): Likewise.
8933 (vsx_udiv_v2di): Likewise.
8935 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
8937 * config/rs6000/rs6000.opt (mlra): Replace with stub.
8938 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
8939 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
8940 (rs6000_debug_reg_global): Delete print of LRA status.
8941 (rs6000_option_override_internal): Delete dead LRA related code.
8942 (rs6000_lra_p): Delete function.
8943 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
8945 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
8947 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
8948 * config/riscv/rtems.h: New file.
8950 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8951 Sudakshina Das <sudi.das@arm.com>
8953 * config/aarch64/aarch64.md
8954 (define_split for and<mode>3nr_compare): Move
8955 non aarch64_logical_operand to a register.
8956 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
8957 register immediate operand to a register.
8958 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
8960 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
8963 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
8965 2017-07-27 Richard Biener <rguenther@suse.de>
8967 PR tree-optimization/81573
8968 PR tree-optimization/81494
8969 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
8970 multi defuse cycle case.
8972 2017-07-27 Richard Biener <rguenther@suse.de>
8974 PR tree-optimization/81571
8975 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
8978 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
8980 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
8981 earlier and only if MASK_FPU is set. Adjust formatting.
8983 2017-07-27 Martin Liska <mliska@suse.cz>
8985 * opt-functions.awk: Add validation of value of Init.
8986 * optc-gen.awk: Pass new argument.
8988 2017-07-27 Martin Liska <mliska@suse.cz>
8990 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
8991 Fix wrong condition.
8993 2017-07-27 Martin Liska <mliska@suse.cz>
8995 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
8996 BBs and edges seen by autoFDO.
8998 2017-07-27 Richard Biener <rguenther@suse.de>
9000 PR tree-optimization/81502
9001 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
9002 with incompatible but same sized type.
9003 (execute_update_addresses_taken): Likewise.
9005 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
9007 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
9008 flag_tree_loop_vectorize rather than flag_tree_vectorize.
9010 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9013 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
9014 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
9015 Change s_operand to memory_operand.
9017 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
9019 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
9020 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
9021 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
9022 Emit instructions rather than returning an expression. Handle TFmode
9023 and KFmode by casting to TImode.
9024 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
9025 (rs6000_emit_le_vsx_store): Likewise.
9026 * config/rs6000/vsx.md (VSX_TI): New iterator.
9027 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
9028 (*vsx_le_undo_permute_<mode>): Likewise.
9029 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
9030 emit the split sequence.
9031 (*vsx_le_perm_store_<mode>): Likewise.
9033 2017-07-27 Jakub Jelinek <jakub@redhat.com>
9035 PR tree-optimization/81555
9036 PR tree-optimization/81556
9037 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
9038 if true, force CHANGED for the recursive invocation.
9039 (reassociate_bb): Remember original length of ops array, pass
9040 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
9042 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
9043 attributes for noipa attribute. For naked attribute use
9044 lookup_attribute first before lookup_attribute_spec.
9045 * final.c (rest_of_handle_final): Disable IPA RA for functions with
9047 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
9048 for functions with noipa attribute.
9049 (cgraph_externally_visible_p): Return true for functions with noipa
9051 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
9052 for functions with noipa attribute.
9053 * doc/extend.texi: Document noipa function attribute.
9054 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
9055 also for functions with noipa attribute.
9056 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
9058 2017-07-26 Andrew Pinski <apinski@cavium.com>
9060 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
9061 vec_unalign_load_cost and vec_unalign_store_cost.
9063 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
9065 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
9066 -mvsx-small-integer option.
9067 (ISA_3_0_MASKS_IEEE): Likewise.
9068 (OTHER_VSX_VECTOR_MASKS): Likewise.
9069 (POWERPC_MASKS): Likewise.
9070 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
9071 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
9072 code, only testing for DImode being allowed in non-VSX floating
9074 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
9075 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
9077 (rs6000_option_override_internal): Delete -mvsx-small-integer.
9078 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
9079 TARGET_P8_VECTOR test.
9080 (rs6000_secondary_reload_simple_move): Likewise.
9081 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
9082 since TARGET_P9_VECTOR was already tested.
9083 (rs6000_opt_masks): Remove -mvsx-small-integer.
9084 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
9085 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
9087 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
9088 test for TARGET_VEXTRACTUB was used, and that uses
9090 (p9 extract splitter): Likewise.
9091 (vsx_extract_<mode>_di_p9): Likewise.
9092 (vsx_extract_<mode>_store_p9): Likewise.
9093 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
9094 for TARGET_P9_VECTOR was used. Delete code that is now dead with
9095 the elimination of TARGET_VSX_SMALL_INTEGER.
9096 (vsx_extract_<mode>_p8): Likewise.
9097 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
9098 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
9099 (vsx_set_<mode>_p9): Likewise.
9100 (vsx_set_v4sf_p9): Likewise.
9101 (vsx_set_v4sf_p9_zero): Likewise.
9102 (vsx_insert_extract_v4sf_p9): Likewise.
9103 (vsx_insert_extract_v4sf_p9_2): Likewise.
9104 * config/rs6000/rs6000.md (sign extend splitter): Change
9105 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
9106 (floatsi<mode>2_lfiwax_mem): Likewise.
9107 (floatunssi<mode>2_lfiwzx_mem): Likewise.
9108 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
9109 since a test for TARGET_P9_VECTOR was used.
9110 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
9111 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
9112 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
9113 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
9114 TARGET_P8_VECTOR test.
9115 (fix_trunc<mode>si2_stfiwx): Likewise.
9116 (fix_trunc<mode>si2_internal): Likewise.
9117 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
9118 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
9120 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
9121 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
9122 TARGET_P8_VECTOR test.
9123 (fixuns_trunc<mode>si2_stfiwx): Likewise.
9124 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
9125 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
9127 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
9128 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
9129 since a test for TARGET_P9_VECTOR was used.
9130 (splitter for loading small constants): Likewise.
9132 2017-07-26 Andrew Pinski <apinski@cavium.com>
9134 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
9137 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
9140 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
9141 (fp_valid_at): Likewise.
9143 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
9145 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
9146 (qdf24xx_addrcost_table): Likewise.
9147 (cortexa57_tunings): Update to use generic_branch_cost.
9148 (cortexa72_tunings): Likewise.
9149 (cortexa73_tunings): Likewise.
9150 (qdf24xx_tunings): Likewise.
9152 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
9154 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
9155 (thunderx2t99_branch_cost): Likewise.
9156 (cortexa35_tunings): Update to use generic_branch_cost.
9157 (cortexa53_tunings): Likewise.
9158 (cortexa57_tunings): Likewise.
9159 (cortexa72_tunings): Likewise.
9160 (cortexa73_tunings): Likewise.
9161 (thunderx2t99_tunings): Likewise.
9163 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
9165 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
9166 (sparc_option_override): Honour MASK_FSMULD.
9167 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
9168 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
9169 * config/sparc/sparc.opt (mfsmuld): New option.
9170 * doc/invoke.texi (mfsmuld): Document option.
9172 2017-07-26 Marek Polacek <polacek@redhat.com>
9175 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
9177 2017-07-26 Richard Biener <rguenther@suse.de>
9179 * gimple-match-head.c (do_valueize): Return OP if valueize
9181 (get_def): New helper to get at the def stmt of a SSA name
9183 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
9184 do_valueize to get at the def stmt.
9185 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
9187 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
9190 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
9192 2017-07-26 Martin Liska <mliska@suse.cz>
9195 * function.c (expand_function_start): Make expansion of
9196 nonlocal_goto_save_area after parm_birth_insn.
9198 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
9200 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
9201 from all CPU target flags enable members.
9203 2017-07-26 Richard Biener <rguenther@suse.de>
9205 * genmatch.c (dt_simplify::gen): Make iterator vars const.
9206 (decision_tree::gen): Make 'type' const.
9207 (write_predicate): Likewise.
9209 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
9211 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
9212 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
9213 (rs6000_option_override_internal): Likewise.
9214 (rs6000_expand_vector_set): Likewise.
9215 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
9216 (TARGET_UPPER_REGS_SF): Likewise.
9217 (TARGET_UPPER_REGS_DI): Likewise.
9218 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
9219 (TARGET_DIRECT_MOVE_64BIT): Likewise.
9220 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
9221 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
9222 (Splitters for DI constants in Altivec registers): Likewise.
9223 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
9224 (vsx_set_v4sf_p9): Likewise.
9225 (vsx_set_v4sf_p9_zero): Likewise.
9226 (vsx_insert_extract_v4sf_p9): Likewise.
9227 (vsx_insert_extract_v4sf_p9_2): Likewise.
9229 2017-07-25 Carl Love <cel@us.ibm.com>
9231 * doc/extend.texi: Update the built-in documentation file for the
9232 existing built-in functions
9233 vector signed char vec_cnttz (vector signed char);
9234 vector unsigned char vec_cnttz (vector unsigned char);
9235 vector signed short vec_cnttz (vector signed short);
9236 vector unsigned short vec_cnttz (vector unsigned short);
9237 vector signed int vec_cnttz (vector signed int);
9238 vector unsigned int vec_cnttz (vector unsigned int);
9239 vector signed long long vec_cnttz (vector signed long long);
9240 vector unsigned long long vec_cnttz (vector unsigned long long);
9242 2017-07-25 Andrew Pinski <apinski@cavium.com>
9244 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
9245 accesses where the use is for the first operand of a BIT_INSERT.
9247 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
9250 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
9251 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
9253 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
9255 * config/i386/gstabs.h: Delete.
9256 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
9258 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
9260 * config/i386/i386.c (ix86_decompose_address): Do not check for
9261 register RTX when looking at index_reg or base_reg.
9262 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
9264 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
9266 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
9267 to update EH info here.
9269 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
9271 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
9273 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
9275 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
9277 2017-07-25 Torsten Duwe <duwe@suse.de>
9279 * common.opt: Introduce -fpatchable-function-entry
9280 command line option, and its variables function_entry_patch_area_size
9281 and function_entry_patch_area_start.
9282 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
9283 including a two-value parser.
9284 * target.def (print_patchable_function_entry): New target hook.
9285 * targhooks.h (default_print_patchable_function_entry): New function.
9286 * targhooks.c (default_print_patchable_function_entry): Likewise.
9287 * toplev.c (process_options): Switch off IPA-RA if
9288 patchable function entries are being generated.
9289 * varasm.c (assemble_start_function): Look at the
9290 patchable-function-entry command line switch and current
9291 function attributes and maybe generate NOP instructions by
9292 calling the print_patchable_function_entry hook.
9293 * doc/extend.texi: Document patchable_function_entry attribute.
9294 * doc/invoke.texi: Document -fpatchable_function_entry
9295 command line option.
9296 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
9298 * doc/tm.texi: Re-generate.
9300 2017-07-25 Jakub Jelinek <jakub@redhat.com>
9303 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
9304 TARGET_AVX512DQ rather than TARGET_AVX512BW.
9306 2017-07-25 Tamar Christina <tamar.christina@arm.com>
9308 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
9310 2017-07-25 Richard Biener <rguenther@suse.de>
9312 PR tree-optimization/81455
9313 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
9314 not walk in cycles when looking for guards.
9316 2017-07-25 Richard Biener <rguenther@suse.de>
9318 PR tree-optimization/81529
9319 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
9320 when optimizing backedge uses.
9322 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
9324 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
9326 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
9327 to dl_section_ref. On AIX, append an expression to subtract
9328 the size of the section length to dl_section_ref.
9330 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
9332 * configure.ac: If any of the config.* scripts fail, exit 1.
9333 * configure: Regenerate.
9335 2017-07-25 Richard Biener <rguenther@suse.de>
9338 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
9339 of immediate uses, be more verbose on errors.
9341 2017-07-25 Richard Biener <rguenther@suse.de>
9343 PR tree-optimization/81510
9344 * tree-vect-loop.c (vect_is_simple_reduction): When the
9345 reduction stmt is not inside the loop bail out.
9347 2017-07-25 Richard Biener <rguenther@suse.de>
9349 PR tree-optimization/81303
9350 * tree-vect-loop-manip.c (vect_loop_versioning): Build
9351 profitability check against LOOP_VINFO_NITERSM1.
9353 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
9355 * domwalk.c (cmp_bb_postorder): Simplify.
9356 (sort_bbs_postorder): New function. Use it...
9357 (dom_walker::walk): ...here to optimize common cases.
9359 2017-07-25 Martin Liska <mliska@suse.cz>
9362 * ipa-visibility.c (function_and_variable_visibility): Make the
9363 redirection just on target that supports aliasing.
9364 Fix GNU coding style.
9366 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
9369 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
9371 * config/arm/rtems.h: Likewise.
9372 * config/bfin/rtems.h: Likewise.
9373 * config/i386/rtemself.h: Likewise.
9374 * config/lm32/rtems.h: Likewise.
9375 * config/m32c/rtems.h: Likewise.
9376 * config/m68k/rtemself.h: Likewise.
9377 * config/microblaze/rtems.h: Likewise.
9378 * config/mips/rtems.h: Likewise.
9379 * config/moxie/rtems.h: Likewise.
9380 * config/nios2/rtems.h: Likewise.
9381 * config/powerpcspe/rtems.h: Likewise.
9382 * config/rs6000/rtems.h: Likewise.
9383 * config/rtems.h: Likewise.
9384 * config/sh/rtems.h: Likewise.
9385 * config/sh/rtemself.h: Likewise.
9386 * config/sparc/rtemself.h: Likewise.
9388 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
9391 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
9392 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
9393 * tree-ssa-structalias.c (alias_get_name): Same.
9395 2017-07-25 Bin Cheng <bin.cheng@arm.com>
9398 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
9399 instructions if no du chain is found.
9401 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
9403 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
9405 2017-07-25 Richard Biener <rguenther@suse.de>
9408 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
9411 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
9413 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
9415 (ISA_2_7_MASKS_SERVER): Likewise.
9416 (ISA_3_0_MASKS_IEEE): Likewise.
9417 (OTHER_P8_VECTOR_MASKS): Likewise.
9418 (OTHER_VSX_VECTOR_MASKS): Likewise.
9419 (POWERPC_MASKS): Likewise.
9420 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
9421 duplicate list of options.
9422 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
9423 explicit -mupper-regs options.
9424 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
9425 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
9426 alias for -mupper-regs-df.
9427 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
9428 (rs6000_init_hard_regno_mode_ok): Likewise.
9429 (rs6000_option_override_internal): Likewise.
9430 (rs6000_opt_masks): Likewise.
9431 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
9432 options in terms of whether -mvsx or -mpower8-vector was used.
9433 (TARGET_UPPER_REGS_DI): Likewise.
9434 (TARGET_UPPER_REGS_SF): Likewise.
9435 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
9436 -mupper-regs-* options.
9438 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
9440 * passes.c (emergency_dump_function): Print some empty lines and a
9441 header before the RTL dump.
9443 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
9445 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
9447 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
9450 * config/aarch64/aarch64.c (aarch64_classify_symbol):
9451 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
9453 2017-07-24 Carl Love <cel@us.ibm.com>
9455 * config/rs6000/rs6000-c.c: Add support for built-in functions
9456 vector float vec_extract_fp32_from_shorth (vector unsigned short);
9457 vector float vec_extract_fp32_from_shortl (vector unsigned short);
9458 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
9459 vec_extract_fp_from_shortl): Add defines for the two builtins.
9460 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
9461 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
9463 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
9464 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
9465 * doc/extend.texi: Update the built-in documentation file for the
9466 new built-in function.
9468 2017-07-24 Jakub Jelinek <jakub@redhat.com>
9471 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
9473 * doc/generic.texi: Likewise.
9474 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
9475 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
9477 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
9479 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
9480 (aarch64_mls_elt_merge<mode>): Likewise.
9482 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
9484 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
9485 having __cxa_atexit.
9487 2017-07-23 Michael Collison <michael.collison@arm.com>
9489 * config/arm/arm.c (arm_option_override): Deprecate
9490 use of -mstructure-size-boundary.
9491 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
9492 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
9494 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9497 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
9498 Reduce cost estimate for direct moves.
9500 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
9503 * config/i386/i386.c (ix86_option_override_internal): Disable
9504 BMI, BMI2 and TBM instructions for -m16.
9506 2017-07-21 Carl Love <cel@us.ibm.com>
9508 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9509 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
9510 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
9511 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
9512 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
9513 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
9514 VMULOSW): New enum "unspec" values.
9515 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
9516 altivec_vmulosw): New patterns.
9517 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
9518 VMULOSW): Add definitions.
9520 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
9522 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
9523 (qdf24xx): Likewise.
9524 * config/aarch64/aarch64-options-extensions.def (rdma); New.
9525 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
9526 (AARCH64_FL_V8_1): Renumber.
9527 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
9528 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
9529 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
9530 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
9531 rdma to feature modifiers list.
9533 2017-07-21 Yury Gribov <tetra2005@gmail.com>
9536 * ipa-visibility (function_and_variable_visibility): Convert
9537 recursive PLT call to direct call if appropriate.
9539 2017-07-21 Andrew Pinski <apinski@cavium.com>
9541 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
9542 operand 1 to see if the types precision matches.
9543 * fold-const.c (operand_equal_p): Likewise.
9545 2017-07-21 Richard Biener <rguenther@suse.de>
9547 PR tree-optimization/81303
9548 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
9549 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
9550 (vect_peeling_hash_get_lowest_cost): Adjust.
9551 (vect_enhance_data_refs_alignment): Likewise. Use
9552 vect_get_peeling_costs_all_drs to compute the penalty for no
9553 peeling to match up costs.
9555 2017-07-21 Richard Biener <rguenther@suse.de>
9557 PR tree-optimization/81500
9558 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
9559 we didn't identify a reduction path.
9561 2017-07-21 Tom de Vries <tom@codesourcery.com>
9562 Cesar Philippidis <cesar@codesourcery.com>
9564 PR gcov-profile/81442
9565 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
9568 2017-07-21 Tom de Vries <tom@codesourcery.com>
9571 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
9573 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
9574 nvptx_override_options_after_change.
9576 2017-07-21 Ulrich Drepper <drepper@redhat.com>
9578 * dwarf2out.c (output_file_names): Avoid double testing for
9581 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
9583 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
9585 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
9587 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
9589 (try_crossjump_to_edge): Do not punt on partitioned functions.
9591 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
9593 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
9594 Put all BBs reachable only via paths crossing cold region to cold
9596 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
9598 2016-07-21 Richard Biener <rguenther@suse.de>
9600 PR tree-optimization/81303
9601 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
9602 into account prologue and epilogue iterations when raising
9603 min_profitable_iters to sth at least covering one vector iteration.
9605 2017-07-21 Tamar Christina <tamar.christina@arm.com>
9607 * config/arm/arm.c (arm_test_cpu_arch_dat):
9610 2017-07-20 Nathan Sidwell <nathan@acm.org>
9612 Remove TYPE_METHODS.
9613 * tree.h (TYPE_METHODS): Delete.
9614 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
9615 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
9616 (dbxout_type_methods): Scan TYPE_FIELDS.
9617 (dbxout_type): Don't check TYPE_METHODS here.
9618 * function.c (use_register_for_decl): Always ignore register for
9619 class types when not optimizing.
9620 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
9621 * tree.c (free_lang_data_in_type): Stitch out member functions and
9622 templates from TYPE_FIELDS.
9623 (build_distinct_type_copy, verify_type_variant,
9624 verify_type): Member fns are on TYPE_FIELDS.
9625 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
9626 * tree-pretty-print.c (dump_generic_node): Likewise.
9628 2017-07-20 Jakub Jelinek <jakub@redhat.com>
9631 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
9632 V2TImode and V4TImode.
9633 (ix86_expand_vector_extract): Likewise.
9634 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
9635 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
9636 (ssescalarmode): Handle V4TImode and V2TImode.
9637 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
9638 (*vec_extractv2ti, *vec_extractv4ti): New insns.
9639 (VEXTRACTI128_MODE): New mode iterator.
9640 (splitter for *vec_extractv?ti first element): New.
9641 (VEC_INIT_MODE): New mode iterator.
9642 (vec_init<mode>): Consolidate 3 expanders into one using
9643 VEC_INIT_MODE mode iterator.
9645 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
9647 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
9648 non_spilled_static_chain_regno_p.
9650 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
9652 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
9654 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
9656 * bb-reorder.c (connect_traces): Allow copying of blocks within
9659 2017-07-20 Richard Biener <rguenther@suse.de>
9661 * gimple.h (gimple_phi_result): Add gphi * overload.
9662 (gimple_phi_result_ptr): Likewise.
9663 (gimple_phi_arg): Likewise. Adjust index assert to only
9664 allow actual argument accesses rather than all slots available
9666 (gimple_phi_arg_def): Add gphi * overload.
9667 * tree-phinodes.c (make_phi_node): Initialize only actual
9669 (resize_phi_node): Clear memory not covered by old node,
9670 do not initialize excess argument slots.
9671 (reserve_phi_args_for_new_edge): Initialize new argument slot
9674 2017-07-20 Bin Cheng <bin.cheng@arm.com>
9676 PR tree-optimization/81388
9678 2016-07-21 Bin Cheng <bin.cheng@arm.com>
9680 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
9681 by removing computation of may_be_zero.
9683 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
9684 Tom de Vries <tom@codesourcery.com>
9687 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
9688 when gimple level profile disagrees with what RTL expander did.
9690 2017-07-20 Richard Biener <rguenther@suse.de>
9692 PR tree-optimization/61171
9693 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
9694 (vect_analyze_stmt): Add slp instance parameter.
9695 (vectorizable_reduction): Likewise.
9696 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
9697 (vect_is_simple_reduction): Deal with chains not detected
9698 as SLP reduction chain, specifically not properly associated
9699 chains containing a mix of plus/minus.
9700 (get_reduction_op): Remove.
9701 (get_initial_defs_for_reduction): Simplify, pass in whether
9702 this is a reduction chain, pass in the SLP node for the PHIs.
9703 (vect_create_epilog_for_reduction): Get the SLP instance as
9705 (vectorizable_reduction): Get the SLP instance as arg.
9706 During analysis remember the SLP node with the PHIs in the
9707 instance. Simplify getting at the vectorized reduction PHIs.
9708 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
9709 through SLP instance.
9710 (vect_slp_analyze_operations): Likewise.
9711 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
9712 (vect_transform_stmt): Likewise.
9714 2017-07-20 Tom de Vries <tom@codesourcery.com>
9716 PR tree-optimization/81489
9717 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
9718 read of phi arg location to before loop that modifies phi.
9720 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
9722 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
9725 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
9728 * except.c (execute): Fix ordering issue.
9730 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
9732 PR rtl-optimization/81423
9733 * combine.c (make_compound_operation_int): Don't try to optimize
9734 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
9736 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
9738 PR rtl-optimization/81423
9739 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
9740 with a constant that is -1 in the truncated to mode.
9742 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
9744 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
9745 (determine_unlikely_bbs): ... here.
9746 * predict.h (propagate_unlikely_bbs_forward): Declare.
9747 * cfgexpand.c (pass_expand::execute): Use it.
9748 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
9750 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
9751 propagate_unlikely_bbs_forward.
9753 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
9756 * except.c (maybe_add_nop_after_section_switch): New function.
9759 2017-07-19 Tom de Vries <tom@codesourcery.com>
9761 * gimple.h (gimple_phi_set_arg): Make assert more strict.
9763 2017-07-19 Tom de Vries <tom@codesourcery.com>
9765 * gimple.h (gimple_phi_arg): Make assert more strict.
9767 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
9769 * config.gcc (powerpc*-*-*): Add mmintrin.h.
9770 * config/rs6000/mmintrin.h: New file.
9771 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
9773 2017-07-19 Jakub Jelinek <jakub@redhat.com>
9775 PR tree-optimization/81346
9776 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
9778 2017-07-19 Tom de Vries <tom@codesourcery.com>
9780 * config/nvptx/nvptx.md (VECIM): Add V2DI.
9782 2017-07-19 Tom de Vries <tom@codesourcery.com>
9784 * config/nvptx/nvptx-modes.def: Add V2DImode.
9785 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
9786 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
9787 (nvptx_output_mov_insn): Handle lack of mov.b128.
9788 (nvptx_print_operand): Handle 'H' and 'L' codes.
9789 (nvptx_vector_mode_supported): Allow V2DImode.
9790 (nvptx_preferred_simd_mode): New function.
9791 (nvptx_data_alignment): New function.
9792 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
9793 nvptx_preferred_simd_mode.
9794 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
9796 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
9798 2017-07-19 Tom de Vries <tom@codesourcery.com>
9800 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
9801 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
9802 (nvptx_vector_mode_supported): New function. Allow V2SImode.
9803 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
9804 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
9805 (mov<VECIM>_insn): New define_insn.
9806 (define_expand "mov<VECIM>): New define_expand.
9808 2017-07-19 Tom de Vries <tom@codesourcery.com>
9810 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
9812 2017-07-19 Jakub Jelinek <jakub@redhat.com>
9814 PR tree-optimization/81346
9815 * fold-const.h (fold_div_compare, range_check_type): Declare.
9816 * fold-const.c (range_check_type): New function.
9817 (build_range_check): Use range_check_type.
9818 (fold_div_compare): No longer static, rewritten into
9819 a match.pd helper function.
9820 (fold_comparison): Don't call fold_div_compare here.
9821 * match.pd (X / C1 op C2): New optimization using fold_div_compare
9824 2017-07-19 Nathan Sidwell <nathan@acm.org>
9826 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
9827 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
9828 * tree.c (find_decls_types_r, verify_type): Use
9829 TYPE_{MIN,MAX}_VALUE_RAW.
9830 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
9831 (hash_tree): Likewise.
9832 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
9834 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
9837 2017-07-18 Tom de Vries <tom@codesourcery.com>
9840 * omp-expand.c (expand_omp_for_static_chunk): Handle
9841 equal-argument loop exit phi.
9843 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
9846 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
9847 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
9848 operand 2 predicate.
9849 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
9850 operand 2 predicate.
9851 (ror,rol -> rorx splitters): Use const_int_operand as
9852 operand 2 predicate.
9854 2017-06-18 Richard Biener <rguenther@suse.de>
9856 PR tree-optimization/81410
9857 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
9858 the gap in the ! slp_perm SLP case after each group.
9860 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
9863 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
9866 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
9869 * predict.c (set_even_probabilities): Cleanup; do not affect
9870 probabilities that are already known.
9871 (combine_predictions_for_bb): Call even when count is set.
9873 2017-07-18 Nathan Sidwell <nathan@acm.org>
9875 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
9878 2017-07-18 Bin Cheng <bin.cheng@arm.com>
9881 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
9882 optimization for loop niter analysis.
9884 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
9887 * config/avr/avr.c (avr_optimize_casesi): Don't use
9888 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
9890 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
9892 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
9893 body_cost_vec from _vect_peel_extended_info.
9894 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
9895 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
9898 2017-07-18 Bin Cheng <bin.cheng@arm.com>
9900 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
9902 2017-07-18 Richard Biener <rguenther@suse.de>
9904 PR tree-optimization/80620
9905 PR tree-optimization/81403
9906 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
9907 info when re-using a VN table entry.
9909 2017-07-18 Richard Biener <rguenther@suse.de>
9911 PR tree-optimization/81418
9912 * tree-vect-loop.c (vectorizable_reduction): Properly compute
9913 vectype_in. Verify that with lane-reducing reduction operations
9914 we have a single def-use cycle.
9916 2017-07-17 Carl Love <cel@us.ibm.com>
9918 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
9920 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9921 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
9922 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
9923 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
9924 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
9925 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
9926 VMULOSW): New enum "unspec" values.
9927 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
9928 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
9929 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
9930 altivec_vmulosw): New patterns.
9931 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
9932 VMULOSW): Add definitions.
9933 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
9935 * config/alpha/alpha.c: Include predict.h.
9937 2017-07-17 Yury Gribov <tetra2005@gmail.com>
9939 * tree-vrp.c (compare_assert_loc): Fix comparison function
9940 to return predictable results.
9942 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
9944 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
9947 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
9948 * doc/invoke.texi (mexpand-adddi): Update text.
9950 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
9952 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
9953 that also clobbers the CC register. The old expand code is moved
9955 (*arc_clzsi2): ... here.
9956 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
9957 the CC register. The old expand code is moved to ...
9958 (arc_ctzsi2): ... here.
9960 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
9962 * config/arc/arc.opt (mindexed-loads): Use initial value
9963 TARGET_INDEXED_LOADS_DEFAULT.
9964 (mauto-modify-reg): Use initial value
9965 TARGET_AUTO_MODIFY_REG_DEFAULT.
9966 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
9967 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
9968 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
9969 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
9971 2017-07-17 Martin Liska <mliska@suse.cz>
9974 * opts.c (finish_options): Do not allow -fgnu-tm
9975 w/ -fsanitize={kernel-,}address. Say sorry.
9977 2017-07-17 Bin Cheng <bin.cheng@arm.com>
9980 * tree-loop-distribution.c (classify_partition): Only assert on
9981 numer of iterations.
9982 (merge_dep_scc_partitions): Delete prameter. Update function call.
9983 (distribute_loop): Remove code handling loop with unknown niters.
9984 (pass_loop_distribution::execute): Skip loop with unknown niters.
9986 2017-07-17 Bin Cheng <bin.cheng@arm.com>
9989 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
9990 function sort_partitions_by_post_order.
9992 2017-07-17 Bin Cheng <bin.cheng@arm.com>
9994 PR tree-optimization/81374
9995 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
9996 the max index of basic blocks, rather than number of basic blocks.
9998 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
10000 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
10002 (arc_legitimate_pic_operand_p): Likewise.
10003 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
10005 (arc_needs_pcl_p): Likewise.
10006 (arc_legitimate_pc_offset_p): Likewise.
10007 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
10008 function is also used in constrains.md.
10009 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
10010 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
10011 PLUS. Only return true/false in known cases, otherwise assert.
10012 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
10013 is already called in arc_legitimate_constant_p.
10014 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
10016 (LEGITIMATE_PIC_OPERAND_P): Use
10017 arc_raw_symbolic_reference_mentioned_p function.
10018 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
10023 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
10024 Andrew Burgess <andrew.burgess@embecosm.com>
10026 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
10027 (arc_return_address_register): New function.
10028 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
10029 (arc_handle_fndecl_attribute): Add naked attribute.
10030 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
10031 (TARGET_WARN_FUNC_RETURN): Likewise.
10032 (arc_allocate_stack_slots_for_args): New function.
10033 (arc_warn_func_return): Likewise.
10034 (machine_function): Change type fn_type.
10035 (arc_compute_function_type): Consider new naked function type,
10036 change function return type.
10037 (arc_must_save_register): Adapt to handle new
10038 arc_compute_function_type's return type.
10039 (arc_expand_prologue): Likewise.
10040 (arc_expand_epilogue): Likewise.
10041 (arc_return_address_regs): Delete.
10042 (arc_return_address_register): New function.
10043 (arc_epilogue_uses): Use above function.
10044 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
10045 (arc_function_type): Change encoding, add naked type.
10046 (ARC_INTERRUPT_P): Change to handle the new encoding.
10047 (ARC_FAST_INTERRUPT_P): Likewise.
10048 (ARC_NORMAL_P): Define.
10049 (ARC_NAKED_P): Likewise.
10050 (arc_compute_function_type): Delete prototype.
10051 * config/arc/arc.md (in_ret_delay_slot): Use
10052 arc_return_address_register function.
10053 (simple_return): Likewise.
10054 (p_return_i): Likewise.
10056 2017-07-17 Jakub Jelinek <jakub@redhat.com>
10058 PR tree-optimization/81428
10059 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
10060 can't be built for those types.
10062 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
10064 Remove stuff dead since r239246.
10066 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
10067 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
10068 (avr_inform_devices): Remove dead stuff.
10070 2017-07-17 Tamar Christina <tamar.christina@arm.com>
10072 * config/arm/arm_neon.h: Fix softp typo.
10074 2017-07-17 Jakub Jelinek <jakub@redhat.com>
10076 PR tree-optimization/81365
10077 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
10078 aggregate moves onto bb predecessor edges, make sure there are no
10079 loads that could alias the lhs in between the start of bb and the
10082 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
10085 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
10086 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
10087 [LSHIFTRT, outer_code = TRUNCATE]: Same.
10089 2017-07-17 Jakub Jelinek <jakub@redhat.com>
10091 PR tree-optimization/81396
10092 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
10093 (init_symbolic_number): Initialize it to 1.
10094 (perform_symbolic_merge): Add n_ops from both operands into the new
10096 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
10097 without base_addr as useless if they need more than one operation.
10098 (bswap_replace): Handle !bswap case for NULL base_addr.
10100 2017-07-17 Tom de Vries <tom@codesourcery.com>
10103 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
10106 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
10108 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
10109 conditional builtin define __FIX_LEON3FT_B2BST.
10111 2017-07-17 Daniel Cederman <cederman@gaisler.com>
10113 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
10114 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
10117 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
10119 PR rtl-optimization/81424
10120 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
10121 to remove potential trapping from operands if -fnon-call-exceptions.
10123 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10125 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
10126 profile_proability for scalling.
10127 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
10129 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10131 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
10133 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10135 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
10136 fixpoint arithmetics.
10138 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10140 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
10141 fixpoint arithmetics.
10143 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10145 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
10146 fixpoint arithmetics.
10148 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10150 * profile-count.h (profile_probability::from_reg_br_prob_note,
10151 profile_probability::to_reg_br_prob_note): New functions.
10152 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
10153 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
10154 * predict.c (probability_reliable_p): Update.
10155 (edge_probability_reliable_p): Update.
10156 (br_prob_note_reliable_p): Update.
10157 (invert_br_probabilities): Update.
10158 (add_reg_br_prob_note): New function.
10159 (combine_predictions_for_insn): Update.
10160 * asan.c (asan_clear_shadow): Update.
10161 * cfgbuild.c (compute_outgoing_frequencies): Update.
10162 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
10163 (update_br_prob_note): Update.
10164 (rtl_verify_edges): Update.
10165 (purge_dead_edges): Update.
10166 (fixup_reorder_chain): Update.
10167 * emit-rtl.c (try_split): Update.
10168 * ifcvt.c (cond_exec_process_insns): Update.
10169 (cond_exec_process_if_block): Update.
10170 (dead_or_predicable): Update.
10171 * internal-fn.c (expand_addsub_overflow): Update.
10172 (expand_neg_overflow): Update.
10173 (expand_mul_overflow): Update.
10174 * loop-doloop.c (doloop_modify): Update.
10175 * loop-unroll.c (compare_and_jump_seq): Update.
10176 * optabs.c (emit_cmp_and_jump_insn_1): Update.
10177 * predict.h: Update.
10178 * reorg.c (mostly_true_jump): Update.
10180 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
10181 * config/alpha/alpha.c (emit_unlikely_jump): Update.
10182 * config/arc/arc.c: (emit_unlikely_jump): Update.
10183 * config/arm/arm.c: (emit_unlikely_jump): Update.
10184 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
10185 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
10186 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
10187 (ix86_print_operand): Update.
10188 (ix86_split_fp_branch): Update.
10189 (predict_jump): Update.
10190 * config/ia64/ia64.c (ia64_print_operand): Update.
10191 * config/mmix/mmix.c (mmix_print_operand): Update.
10192 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
10193 (rs6000_expand_split_stack_prologue): Update.
10194 * config/rs6000/rs6000.c: Update.
10195 * config/s390/s390.c (s390_expand_vec_strlen): Update.
10196 (s390_expand_vec_movstr): Update.
10197 (s390_expand_cs_tdsi): Update.
10198 (s390_expand_split_stack_prologue): Update.
10199 * config/sh/sh.c (sh_print_operand): Update.
10200 (expand_cbranchsi4): Update.
10201 (expand_cbranchdi4): Update.
10202 * config/sparc/sparc.c (output_v9branch): Update.
10203 * config/spu/spu.c (get_branch_target): Update.
10204 (ea_load_store_inline): Update.
10205 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
10206 * config/tilepro/tilepro.c: Update.
10208 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
10210 * gimplify.c (mostly_copy_tree_r): Revert latest change.
10211 (gimplify_save_expr): Likewise.
10213 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
10215 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
10217 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
10219 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
10221 * timevar.def (TV_IPA_FNSUMMARY): Define.
10223 2017-07-16 Daniel Cederman <cederman@gaisler.com>
10225 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
10226 to back store errata sensitive sequence from being generated.
10227 (sqrtdf2_fix): Likewise.
10229 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
10231 * tree-ssa-threadupdate.c (compute_path_counts,
10232 update_joiner_offpath_counts): Use profile_probability.
10234 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
10237 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
10239 * config/arm/arm-c.c (arm_cpu_builtins): Define
10240 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
10242 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
10244 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10245 array entries to represent __ieee128 versions of the
10246 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
10247 scalar_extract_sig, and scalar_insert_exp built-in functions.
10248 (altivec_resolve_overloaded_builtin): Add special case handling
10249 for the __builtin_scalar_insert_exp function, as represented by
10250 the P9V_BUILTIN_VEC_VSIEDP constant.
10251 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
10252 exponent support for __ieee128 argument.
10253 (VSESQP): Add scalar extract signature support for __ieee128
10255 (VSTDCNQP): Add scalar test negative support for __ieee128
10257 (VSIEQP): Add scalar insert exponent support for __int128 argument
10258 with __ieee128 result.
10259 (VSIEQPF): Add scalar insert exponent support for __ieee128
10260 argument with __ieee128 result.
10261 (VSTDCQP): Add scalar test data class support for __ieee128
10263 (VSTDCNQP): Add overload support for scalar test negative with
10264 __ieee128 argument.
10265 (VSTDCQP): Add overload support for scalar test data class
10266 __ieee128 argument.
10267 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
10268 UNSPEC_VSX_SXSIGDP.
10269 (UNSPEC_VSX_SIEXPQP): New constant.
10270 (xsxexpqp): New insn for VSX scalar extract exponent quad
10272 (xsxsigqp): New insn for VSX scalar extract significand quad
10274 (xsiexpqpf): New insn for VSX scalar insert exponent quad
10275 precision with floating point argument.
10276 (xststdcqp): New expand for VSX scalar test data class quad
10278 (xststdcnegqp): New expand for VSX scalar test negative quad
10280 (xststdcqp): New insn to match expansions for VSX scalar test data
10281 class quad precision and VSX scalar test negative quad precision.
10282 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
10283 special case operand checking to enforce that second operand of
10284 VSX scalar test data class with quad precision argument is a 7-bit
10286 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
10287 prototypes and descriptions of __ieee128 versions of
10288 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
10289 scalar_test_data_class, and scalar_test_neg built-in functions.
10291 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10293 PR tree-optimization/81162
10294 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
10295 replace a negate with an add.
10297 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
10299 * doc/invoke.texi (arm/-mcpu): Document +crypto.
10301 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
10303 * config/arm/arm-c.c (arm_cpu_builtins): Define
10304 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
10306 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
10308 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
10309 (armv8-r): Set ARM Cortex-R52 as default CPU.
10310 * config/arm/arm-tables.opt: Regenerate.
10311 * config/arm/arm-tune.md: Regenerate.
10312 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
10314 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
10315 extension for -mcpu=cortex-r52.
10317 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
10319 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
10320 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
10321 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
10322 (fp-armv8): Define it as FP_ARMv8 only.
10323 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
10324 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
10326 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
10327 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
10328 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
10329 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
10330 than TARGET_FPU_ARMV8.
10331 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
10332 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
10333 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
10335 * config/arm/neon.md (neon_vrint): Likewise.
10336 (neon_vcvt): Likewise.
10337 (neon_<fmaxmin_op><mode>): Likewise.
10338 (<fmaxmin><mode>3): Likewise.
10339 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
10340 * config/arm/predicates.md (arm_cond_move_operator): Check against
10341 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
10343 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
10345 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
10346 to top of function.
10348 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10350 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
10351 loop in comment with memset.
10353 2017-07-14 Martin Liska <mliska@suse.cz>
10355 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
10356 * dwarf2out.c (is_java): Remove the function.
10357 (output_pubname): Remove usage of the function.
10358 (lower_bound_default): Remove usage of DW_LANG_Java.
10359 (gen_compile_unit_die): Likewise.
10360 * gcc.c: Remove compiler defaults for .java and .zip files.
10361 * gimple-expr.c (remove_suffix): Change as there's no longer
10362 extension than 4-letter one.
10363 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
10364 (gimplify_save_expr): Likewise.
10365 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
10366 as it's possible even for other languages than Java.
10367 * langhooks.h (struct lang_hooks): Remove Java from a comment.
10368 * lto-opts.c (lto_write_options): Remove reference to Java.
10369 * opts.c (strip_off_ending): Update file extension handling.
10370 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
10371 * tree-eh.c (lower_resx): Likewise.
10372 * tree.c (free_lang_data_in_type): Remove dead code.
10373 (find_decls_types_r): Likewise.
10374 (build_common_builtin_nodes): Remove Java from a comment.
10375 (verify_type): Remove dead code.
10376 * varasm.c (assemble_external): Remove Java from a comment.
10378 2017-07-14 Martin Liska <mliska@suse.cz>
10380 * opts.c (finish_options): Add quotes.
10381 (common_handle_option): Likewise.
10383 2017-07-14 Martin Liska <mliska@suse.cz>
10385 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
10386 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
10387 Remove N_SO_PASCAL.
10388 * dwarf2out.c (lower_bound_default): Do not handle
10390 (gen_compile_unit_die): Likewise.
10391 * gcc.c: Remove default extension binding for GNU Pascal.
10392 * stmt.c: Remove Pascal language from a comment.
10393 * xcoffout.c: Likewise.
10395 2017-07-13 David Malcolm <dmalcolm@redhat.com>
10398 * diagnostic-show-locus.c (fixit_cmp): New function.
10399 (layout::layout): Sort m_fixit_hints.
10400 (column_range::column_range): Assert that the values are valid.
10401 (struct char_span): New struct.
10402 (correction::overwrite): New method.
10403 (struct source_line): New struct.
10404 (line_corrections::add_hint): Add assertions. Reimplement memcpy
10405 calls in terms of classes source_line and char_span, and
10406 correction::overwrite.
10407 (selftest::test_overlapped_fixit_printing_2): New function.
10408 (selftest::diagnostic_show_locus_c_tests): Call it.
10410 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
10412 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
10413 early if there is no lhs.
10415 2017-07-13 Martin Liska <mliska@suse.cz>
10417 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
10418 (gen_reference_type_die): Likewise.
10419 * stor-layout.c: Remove Pascal-related comment.
10421 2017-07-13 Martin Liska <mliska@suse.cz>
10423 * opts.c (finish_options): Add quotes to error messages.
10424 (parse_sanitizer_options): Likewise.
10426 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
10428 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
10430 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
10432 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
10434 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
10436 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
10438 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
10440 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
10443 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
10444 provides the hardware capability bits, define the macro
10445 __BUILTIN_CPU_SUPPORTS__.
10446 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
10447 if GLIBC does not provide the hardware capability bits. Add a
10448 gcc_unreachable call if the built-in cpu function is neither
10449 __builtin_cpu_is nor __builtin_cpu_supports.
10450 (rs6000_get_function_versions_dispatcher): Change the warning
10451 that an old GLIBC is used which does not export the capability
10452 bits to be an error.
10453 * doc/extend.texi (target_clones attribute): Document the
10454 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
10455 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
10456 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
10457 the macros defined by GCC if the newer GLIBC is available.
10459 2017-07-12 Jeff Law <law@redhat.com>
10461 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
10462 remaining includes slightly.
10463 * config/riscv/riscv-builtins.c: Include profile-count.h.
10465 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
10468 * config/avr/avr.c (avr_set_current_function): In diagnostic
10469 messages: Quote keywords and (parts of) identifiers.
10470 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
10473 2017-07-12 Carl Love <cel@us.ibm.com>
10475 * config/rs6000/rs6000-c.c: Add support for built-in functions
10476 vector bool char vec_revb (vector bool char);
10477 vector bool short vec_revb (vector short char);
10478 vector bool int vec_revb (vector bool int);
10479 vector bool long long vec_revb (vector bool long long);
10480 * doc/extend.texi: Update the built-in documentation file for the
10481 new built-in functions.
10483 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10485 * config/s390/s390.md: Remove movcc splitter.
10487 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10489 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
10490 load/store on condition.
10492 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
10495 * config/avr/avr.c (avr_encode_section_info)
10496 [progmem && !TREE_READONLY]: Error if progmem object needs
10499 2017-07-11 Michael Collison <michael.collison@arm.com>
10501 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
10504 2017-07-11 Carl Love <cel@us.ibm.com>
10506 * config/rs6000/rs6000-c.c: Add support for builtins
10507 vector unsigned int vec_parity_lsbb (vector signed int);
10508 vector unsigned int vec_parity_lsbb (vector unsigned int);
10509 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
10510 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
10511 vector unsigned long long vec_parity_lsbb (vector signed long long);
10512 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
10513 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
10514 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
10515 * doc/extend.texi: Update the built-in documentation file for the
10516 new built-in functions.
10518 2017-07-11 David Malcolm <dmalcolm@redhat.com>
10520 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
10521 (layout::m_primary_loc): New field.
10522 (layout::layout): Initialize new field. Move location filtering
10523 logic from here to...
10524 (layout::maybe_add_location_range): ...this new method. Add
10525 support for filtering to just the lines already specified by other
10527 (layout::will_show_line_p): New method.
10528 (gcc_rich_location::add_location_if_nearby): New method.
10529 (selftest::test_add_location_if_nearby): New test function.
10530 (selftest::diagnostic_show_locus_c_tests): Call it.
10531 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
10534 2017-07-11 Tom de Vries <tom@codesourcery.com>
10536 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
10537 (bb_first_real_insn): New function.
10538 (nvptx_single): Add extra initialization of broadcasted condition
10541 2017-07-11 Nathan Sidwell <nathan@acm.org>
10543 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
10545 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
10547 * doc/extend.texi (AVR Function Attributes): Remove weblink to
10548 Binutils doc as TEXI will mess them up.
10549 * doc/invoke.texi (AVR Options): Same here.
10551 2017-07-11 Daniel Cederman <cederman@gaisler.com>
10553 * config/sparc/sparc.opt (mfix-ut700): New option.
10554 (mfix-gr712rc): Likewise.
10555 (sparc_fix_b2bst): New variable.
10556 * doc/invoke.texi (SPARC options): Document them.
10557 (ARM options): Fix warnings.
10558 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
10559 instructions to prevent sequences that can trigger the store-store
10560 errata for certain LEON3FT processors.
10561 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
10562 (sparc_option_override): Set sparc_fix_b2bst appropriately.
10563 * config/sparc/sparc.md (fix_b2bst): New attribute.
10564 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
10566 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
10569 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
10571 (*rsqrtsf2_sse): Ditto.
10573 (div<mode>3): Macroize insn from divdf3 and divsf3
10574 using MODEF mode iterator.
10576 2017-07-10 Martin Sebor <msebor@redhat.com>
10578 PR tree-optimization/80397
10579 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
10580 instead of testing for equality to INTEGER_TYPE.
10582 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
10584 * config.gcc: Remove uclibc from arc target spec.
10586 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
10588 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
10590 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
10593 * lto-wrapper.c (remove_option): New function.
10594 (merge_and_complain): Merge PIC/PIE options more realistically.
10596 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
10598 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
10602 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
10603 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
10604 * config.in: Regenerate.
10605 * configure: Regenerate.
10606 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
10607 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
10608 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
10609 (TARGET_GASISR_PROLOGUES): ...target mask.
10610 * common/config/avr/avr-common.c
10611 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
10612 Set -mgas-isr-prologues.
10613 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
10614 INSERT_PASS_BEFORE for it.
10615 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
10616 * config/avr/avr.c (avr_option_override)
10617 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
10618 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
10619 (avr_attribute_table) <no_gccisr>: Add new function attribute.
10620 (avr_set_current_function) <is_no_gccisr>: Init machine field.
10621 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
10623 (make_avr_pass_pre_proep): New function.
10624 (emit_push_sfr) <treg>: Add argument to function and use it
10625 instead of TMP_REG.
10626 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
10627 and set machine->gasisr.yes.
10628 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
10629 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
10630 __gcc_isr.n_pushed to .L__stack_usage.
10631 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
10632 (avr_asm_final_postscan_insn): ...this new static function.
10633 * config/avr/avr.h (machine_function)
10634 <is_no_gccisr, use_L__stack_usage>: New fields.
10635 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
10636 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
10637 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
10638 (gasisr, *gasisr): New expander and insn.
10639 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
10640 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
10641 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
10643 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
10645 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
10648 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
10650 Move jump-tables out of .text again.
10653 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
10654 (ASM_OUTPUT_ADDR_VEC): New function.
10655 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
10656 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
10657 INSN_ADDRESSes as asm comment.
10658 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
10659 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
10660 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
10661 * config/avr/avr.md (*tablejump): Adjust comment.
10662 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
10663 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
10665 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
10666 (avr_output_addr_vec): New proto.
10667 (avr_log_t) <insn_addresses>: New field.
10669 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
10672 * config/i386/i386.c (ix86_function_arg_advance): Set
10673 outgoing_args_on_stack to true if there are outgoing arguments
10675 (ix86_function_arg): Likewise.
10676 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
10677 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
10678 * config/i386/i386.h (machine_function): Add
10679 outgoing_args_on_stack.
10681 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
10683 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
10684 supporting pthreds.
10685 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
10687 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
10689 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
10690 (REAL_H): Remove $(MACHMODE_H).
10691 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
10693 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
10694 $(MACHMODE_H) and double-int.h.
10695 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
10697 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
10700 2017-07-07 Andrew Pinski <apinski@cavium.com>
10702 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
10703 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
10705 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
10707 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
10708 Add warning if GCC was not configured to link against a GLIBC that
10709 exports the hardware capability bits.
10710 (make_resolver_func): Make resolver function private and not a
10711 COMDAT function. Create the name with clone_function_name instead
10712 of make_unique_name.
10715 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
10716 correct operand in doing the split.
10718 2017-07-07 Carl Love <cel@us.ibm.com>
10720 * config/rs6000/rs6000-c: Add support for built-in function
10721 vector unsigned short vec_pack_to_short_fp32 (vector float,
10723 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
10724 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
10725 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
10726 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
10727 (convert_4f32_8i16): Add define_expand.
10728 * doc/extend.texi: Update the built-in documentation file for the
10729 new built-in function.
10731 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10733 * config/sparc/m8.md: New file.
10734 * config/sparc/sparc.md: Include m8.md.
10736 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10738 * config/sparc/sparc.opt: New option -mvis4b.
10739 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
10740 (sparc_option_override): Handle VIS4B.
10741 (enum sparc_builtins): Define
10742 SPARC_BUILTIN_DICTUNPACK{8,16,32},
10743 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
10744 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
10745 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
10746 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
10747 (check_constant_argument): New function.
10748 (sparc_vis_init_builtins): Define builtins
10749 __builtin_vis_dictunpack{8,16,32},
10750 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
10751 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
10752 __builtin_vis_fpcmpde{8,16,32}shl and
10753 __builtin_vis_fpcmpur{8,16,32}shl.
10754 (sparc_expand_builtin): Check that the constant operands to
10755 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
10756 constant and in range.
10757 * config/sparc/sparc-c.c (sparc_target_macros): Handle
10759 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
10760 (SPARC_IMM5_P): Likewise.
10761 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
10762 (enabled): Handle vis4b.
10763 (UNSPEC_DICTUNPACK): New unspec.
10764 (UNSPEC_FPCMPSHL): Likewise.
10765 (UNSPEC_FPUCMPSHL): Likewise.
10766 (UNSPEC_FPCMPDESHL): Likewise.
10767 (UNSPEC_FPCMPURSHL): Likewise.
10768 (cpu_feature): New CPU feature `vis4b'.
10769 (dictunpack{8,16,32}): New insns.
10770 (FPCSMODE): New mode iterator.
10771 (fpcscond): New code iterator.
10772 (fpcsucond): Likewise.
10773 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
10774 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
10775 (fpcmpde{8,16,32}{si,di}shl): Likewise.
10776 (fpcmpur{8,16,32}{si,di}shl): Likewise.
10777 * config/sparc/constraints.md: Define constraints `q' for unsigned
10778 2-bit integer constants and `t' for unsigned 5-bit integer
10780 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
10782 (imm5_operand_dictunpack16): Likewise.
10783 (imm5_operand_dictunpack32): Likewise.
10784 (imm2_operand): Likewise.
10785 * doc/invoke.texi (SPARC Options): Document -mvis4b.
10786 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
10787 ditunpack* and fpcmp*shl builtins.
10789 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10791 * config.gcc: Handle m8 in --with-{cpu,tune} options.
10792 * config.in: Add HAVE_AS_SPARC6 define.
10793 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
10795 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
10797 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
10798 (CPP_CPU_SPEC): Handle m8.
10799 (ASM_CPU_SPEC): Likewise.
10800 * config/sparc/sparc-opts.h (enum processor_type): Add
10802 * config/sparc/sparc.c (m8_costs): New struct.
10803 (sparc_option_override): Handle TARGET_CPU_m8.
10804 (sparc32_initialize_trampoline): Likewise.
10805 (sparc64_initialize_trampoline): Likewise.
10806 (sparc_issue_rate): Likewise.
10807 (sparc_register_move_cost): Likewise.
10808 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
10809 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
10810 (ASM_CPU64_DEFAULT_SPEC): Likewise.
10811 (CPP_CPU_SPEC): Handle M8.
10812 (ASM_CPU_SPEC): Likewise.
10813 (AS_M8_FLAG): Define.
10814 * config/sparc/sparc.md: Add m8 to the cpu attribute.
10815 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
10816 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
10818 * configure: Regenerate.
10819 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
10822 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10824 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
10826 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
10827 ("*movdi_insn_sp32"): Do not set v3pipe.
10828 ("*movsi_insn"): Likewise.
10829 ("*movdi_insn_sp64"): Likewise.
10830 ("*movsf_insn"): Likewise.
10831 ("*movdf_insn_sp32"): Likewise.
10832 ("*movdf_insn_sp64"): Likewise.
10833 ("*zero_extendsidi2_insn_sp64"): Likewise.
10834 ("*sign_extendsidi2_insn"): Likewise.
10835 ("*mov<VM32:mode>_insn"): Likewise.
10836 ("*mov<VM64:mode>_insn_sp64"): Likewise.
10837 ("*mov<VM64:mode>_insn_sp32"): Likewise.
10838 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
10839 ("<vlop:code><VL:mode>3"): Likewise.
10840 ("*not_<vlop:code><VL:mode>3"): Likewise.
10841 ("*nand<VL:mode>_vis"): Likewise.
10842 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
10843 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
10844 ("one_cmpl<VL:mode>2"): Likewise.
10845 ("faligndata<VM64:mode>_vis"): Likewise.
10846 ("alignaddrsi_vis"): Likewise.
10847 ("alignaddrdi_vis"): Likweise.
10848 ("alignaddrlsi_vis"): Likewise.
10849 ("alignaddrldi_vis"): Likewise.
10850 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
10851 ("bmaskdi_vis"): Likewise.
10852 ("bmasksi_vis"): Likewise.
10853 ("bshuffle<VM64:mode>_vis"): Likewise.
10854 ("cmask8<P:mode>_vis"): Likewise.
10855 ("cmask16<P:mode>_vis"): Likewise.
10856 ("cmask32<P:mode>_vis"): Likewise.
10857 ("pdistn<P:mode>_vis"): Likewise.
10858 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
10860 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10862 * config/sparc/sparc.md ("subtype"): New insn attribute.
10863 ("*wrgsr_sp64"): Set insn subtype.
10864 ("*rdgsr_sp64"): Likewise.
10865 ("alignaddrsi_vis"): Likewise.
10866 ("alignaddrdi_vis"): Likewise.
10867 ("alignaddrlsi_vis"): Likewise.
10868 ("alignaddrldi_vis"): Likewise.
10869 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
10870 ("fexpand_vis"): Likewise.
10871 ("fpmerge_vis"): Likewise.
10872 ("faligndata<VM64:mode>_vis"): Likewise.
10873 ("bshuffle<VM64:mode>_vis"): Likewise.
10874 ("cmask8<P:mode>_vis"): Likewise.
10875 ("cmask16<P:mode>_vis"): Likewise.
10876 ("cmask32<P:mode>_vis"): Likewise.
10877 ("fchksm16_vis"): Likewise.
10878 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
10879 ("fmean16_vis"): Likewise.
10880 ("fp<plusminus_insn>64_vis"): Likewise.
10881 ("<plusminus_insn>v8qi3"): Likewise.
10882 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
10883 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
10884 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
10885 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
10886 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
10887 ("*movqi_insn"): Likewise.
10888 ("*movhi_insn"): Likewise.
10889 ("*movsi_insn"): Likewise.
10890 ("movsi_pic_gotdata_op"): Likewise.
10891 ("*movdi_insn_sp32"): Likewise.
10892 ("*movdi_insn_sp64"): Likewise.
10893 ("movdi_pic_gotdata_op"): Likewise.
10894 ("*movsf_insn"): Likewise.
10895 ("*movdf_insn_sp32"): Likewise.
10896 ("*movdf_insn_sp64"): Likewise.
10897 ("*zero_extendhisi2_insn"): Likewise.
10898 ("*zero_extendqihi2_insn"): Likewise.
10899 ("*zero_extendqisi2_insn"): Likewise.
10900 ("*zero_extendqidi2_insn"): Likewise.
10901 ("*zero_extendhidi2_insn"): Likewise.
10902 ("*zero_extendsidi2_insn_sp64"): Likewise.
10903 ("ldfsr"): Likewise.
10904 ("prefetch_64"): Likewise.
10905 ("prefetch_32"): Likewise.
10906 ("tie_ld32"): Likewise.
10907 ("tie_ld64"): Likewise.
10908 ("*tldo_ldub_sp32"): Likewise.
10909 ("*tldo_ldub1_sp32"): Likewise.
10910 ("*tldo_ldub2_sp32"): Likewise.
10911 ("*tldo_ldub_sp64"): Likewise.
10912 ("*tldo_ldub1_sp64"): Likewise.
10913 ("*tldo_ldub2_sp64"): Likewise.
10914 ("*tldo_ldub3_sp64"): Likewise.
10915 ("*tldo_lduh_sp32"): Likewise.
10916 ("*tldo_lduh1_sp32"): Likewise.
10917 ("*tldo_lduh_sp64"): Likewise.
10918 ("*tldo_lduh1_sp64"): Likewise.
10919 ("*tldo_lduh2_sp64"): Likewise.
10920 ("*tldo_lduw_sp32"): Likewise.
10921 ("*tldo_lduw_sp64"): Likewise.
10922 ("*tldo_lduw1_sp64"): Likewise.
10923 ("*tldo_ldx_sp64"): Likewise.
10924 ("*mov<VM32:mode>_insn"): Likewise.
10925 ("*mov<VM64:mode>_insn_sp64"): Likewise.
10926 ("*mov<VM64:mode>_insn_sp32"): Likewise.
10928 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10930 * config/sparc/sparc.md ("type"): New insn type viscmp.
10931 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
10933 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
10934 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
10935 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
10936 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
10938 ("n7_vis_logical_11cycle"): Likewise.
10939 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
10940 * config/sparc/niagara2.md ("niag3_vis": Likewise.
10941 * config/sparc/niagara.md ("niag_vis"): Likewise.
10942 * config/sparc/ultra3.md ("us3_fga"): Likewise.
10943 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
10945 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10947 * config/sparc/sparc.md: New instruction type `bmask'.
10948 (bmaskdi_vis): Use the `bmask' type.
10949 (bmasksi_vis): Likewise.
10950 * config/sparc/ultra3.md (us3_array): Likewise.
10951 * config/sparc/niagara7.md (n7_array): Likewise.
10952 * config/sparc/niagara4.md (n4_array): Likewise.
10953 * config/sparc/niagara2.md (niag2_vis): Likewise.
10954 (niag3_vis): Likewise.
10955 * config/sparc/niagara.md (niag_vis): Likewise.
10957 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
10959 * ipa-comdats.c: Remove optimize check from gate.
10960 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
10961 for functions not optimized.
10962 (ipa_fn_summary_read): Skip optimize check.
10963 (ipa_fn_summary_write): Likewise.
10964 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
10966 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
10968 (can_inline_edge_p): Check flag_pcc_struct_return for match.
10969 (check_callers): Give up on caller which is not optimized.
10970 (inline_small_functions): Likewise.
10971 (ipa_inline): Do not give up when not optimizing.
10972 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
10973 away unoptimizes cdtors.
10974 (whole_program_function_and_variable_visibility): Do
10975 ipa_discover_readonly_nonaddressable_vars in LTO mode.
10976 * ipa.c (process_references): Do not check optimize.
10977 (symbol_table::remove_unreachable_nodes): Update optimize check.
10978 (set_writeonly_bit): Update optimize check.
10979 (pass_ipa_cdtor_merge::gate): Do not check optimize.
10980 (pass_ipa_single_use::gate): Remove.
10982 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
10984 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
10985 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
10986 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
10987 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
10988 permute_load, permute_store, adjust_extract, adjust_splat,
10989 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
10990 replace_swap_with_copy, dump_swap_insn_table,
10991 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
10992 recombine_lvx_pattern, recombine_stvx_pattern,
10993 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
10994 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
10995 to file rs6000-p8swap.c.
10996 * config/rs6000/rs6000-p8swap.c: New file.
10997 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
10998 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
10999 and rs6000*-*-* targets.
11001 2017-07-06 David Malcolm <dmalcolm@redhat.com>
11003 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
11005 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
11007 * lto-wrapper.c (merge_and_complain): Do not merge
11008 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
11009 fsigned_zeros, ftrapping_math, fwrapv.
11010 (append_compiler_options): Do not track these options.
11011 (append_linker_options): Likewie
11013 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
11015 * cgraphunit.c (cgraph_node::finalize_function): When
11016 !flag_toplevel_reorde set no_reorder flag.
11017 (varpool_node::finalize_decl): Likewise.
11018 (symbol_table::compile): Drop no toplevel reorder path.
11020 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
11022 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
11023 edges; zero probability is not better than uninitialized.
11025 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
11027 * asan.h (asan_sanitize_allocas_p): Declare.
11028 * asan.c (asan_sanitize_allocas_p): New function.
11029 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
11030 (handle_builtin_alloca): Likewise.
11031 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
11032 if !asan_sanitize_allocas_p.
11033 * params.def (asan-instrument-allocas): Add new option.
11034 * params.h (ASAN_PROTECT_ALLOCAS): Define.
11035 * opts.c (common_handle_option): Disable allocas sanitization for
11038 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
11040 * asan.c: Include gimple-fold.h.
11041 (get_last_alloca_addr): New function.
11042 (handle_builtin_stackrestore): Likewise.
11043 (handle_builtin_alloca): Likewise.
11044 (asan_emit_allocas_unpoison): Likewise.
11045 (get_mem_refs_of_builtin_call): Add new parameter, remove const
11046 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
11047 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
11048 (instrument_builtin_call): Pass gimple iterator to
11049 get_mem_refs_of_builtin_call.
11050 (last_alloca_addr): New global.
11051 * asan.h (asan_emit_allocas_unpoison): Declare.
11052 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
11053 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
11054 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
11055 if function calls alloca.
11056 * gimple-fold.c (replace_call_with_value): Remove static keyword.
11057 * gimple-fold.h (replace_call_with_value): Declare.
11058 * internal-fn.c: Include asan.h.
11059 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
11060 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
11062 2017-07-06 David Malcolm <dmalcolm@redhat.com>
11064 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
11065 (C_SELFTEST_FLAGS): New.
11066 (CPP_SELFTEST_FLAGS): New.
11067 (SELFTEST_DEPS): New, from deps of s-selftest.
11068 (C_SELFTEST_DEPS): New, from deps of s-selftest.
11069 (CPP_SELFTEST_DEPS): New.
11070 (selftest): Add dependency on s-selftest-c++.
11071 (s-selftest): Rename to...
11072 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
11073 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
11074 than SELFTEST_FLAGS.
11075 (selftest-gdb): Rename to...
11076 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
11078 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
11079 (selftest-valgrind): Rename to...
11080 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
11082 (selftest-valgrind): Reintroduce as an alias for
11083 selftest-c-valgrind.
11084 (s-selftest-c++): New.
11085 (selftest-c++-gdb): New.
11086 (selftest-c++-valgrind): New.
11088 2017-07-06 Olivier Hainque <hainque@adacore.com>
11090 * gcc.c (process_command): When deciding if undefined variables
11091 should be ignored when processing specs, accept "gcc -v" as well.
11093 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
11095 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
11096 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
11098 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
11100 * config/arm/arm-cpus.in (armv8-r): Add new entry.
11101 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
11102 * config/arm/arm-tables.opt: Regenerate.
11103 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
11105 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
11107 2017-07-06 Carl Love <cel@us.ibm.com>
11109 * ChangeLog: Clean up from mid air collision
11111 2017-07-06 Carl Love <cel@us.ibm.com>
11113 * config/rs6000/rs6000-c.c: Add support for built-in functions
11114 vector signed int vec_subc (vector signed int, vector signed int);
11115 vector signed __int128 vec_subc (vector signed __int128,
11116 vector signed __int128);
11117 vector unsigned __int128 vec_subc (vector unsigned __int128,
11118 vector unsigned __int128);
11119 vector signed int vec_sube (vector signed int, vector signed int,
11120 vector signed int);
11121 vector unsigned int vec_sube (vector unsigned int,
11122 vector unsigned int,
11123 vector unsigned int);
11124 vector signed __int128 vec_sube (vector signed __int128,
11125 vector signed __int128,
11126 vector signed__int128);
11127 vector unsigned __int128 vec_sube (vector unsigned __int128,
11128 vector unsigned __int128,
11129 vector unsigned __int128);
11130 vector signed int vec_subec (vector signed int, vector signed int,
11131 vector signed int);
11132 vector unsigned int vec_subec (vector unsigned int,
11133 vector unsigned int,
11134 vector unsigned int);
11135 vector signed __int128 vec_subec (vector signed __int128,
11136 vector signed __int128,
11137 vector signed__int128);
11138 vector unsigned __int128 vec_subec (vector unsigned __int128,
11139 vector unsigned __int128,
11140 vector unsigned __int128);
11141 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
11142 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
11143 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
11144 BU_ALTIVEC_OVERLOAD_X definitions.
11145 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
11146 * doc/extend.texi: Update the built-in documentation file for the new
11147 built-in functions.
11149 2017-07-06 David Malcolm <dmalcolm@redhat.com>
11152 * diagnostic-show-locus.c (layout::layout): Use start and finish
11153 spelling location for the start and finish of each range.
11154 * genmatch.c (linemap_client_expand_location_to_spelling_point):
11155 Add unused aspect param.
11156 * input.c (expand_location_1): Add "aspect" param, and use it
11157 to access the correct part of the location.
11158 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
11160 (expand_location_to_spelling_point): Likewise.
11161 (linemap_client_expand_location_to_spelling_point): Add "aspect"
11162 param, and pass it to expand_location_1.
11164 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
11166 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
11167 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
11168 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
11169 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
11170 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
11171 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
11172 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
11173 _mm_maskz_getmant_ss): New intrinsics.
11174 (__builtin_ia32_getexpss128_mask): Changed to ...
11175 __builtin_ia32_getexpss128_round ... this.
11176 (__builtin_ia32_getexpsd128_mask): Changed to ...
11177 __builtin_ia32_getexpsd128_round ... this.
11178 * config/i386/i386-builtin-types.def
11179 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
11180 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
11181 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
11182 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
11183 __builtin_ia32_getmantss_mask_round): New builtins.
11184 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
11185 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
11186 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
11187 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
11188 * config/i386/sse.md
11189 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
11190 avx512f_sgetexp<mode><mask_scalar_name>
11191 <round_saeonly_scalar_name> ... this.
11192 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
11193 %0, %1, %2<round_saeonly_op3>}): Changed to ...
11194 vgetexp<ssescalarmodesuffix>
11195 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
11196 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
11197 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
11198 avx512f_vgetmant<mode><mask_scalar_name>
11199 <round_saeonly_scalar_name> ... this.
11200 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
11201 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
11202 vgetmant<ssescalarmodesuffix>
11203 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
11204 %0<mask_scalar_operand4>, %1, %2
11205 <round_saeonly_scalar_mask_op4>, %3} ... this.
11206 * config/i386/subst.md (mask_scalar_operand4,
11207 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
11208 round_saeonly_scalar_nimm_predicate): New subst attributes.
11210 2017-07-06 Julia Koval <julia.koval@intel.com>
11212 * config/i386/i386.c (ix86_erase_embedded_rounding):
11213 Remove code for old rounding pattern.
11215 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
11217 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
11219 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
11221 * doc/sourcebuild.texi (Test Directives, Variants of
11222 dg-require-support): Add documentation for dg-require-stack-check.
11224 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
11226 * config/i386/subst.md (mask_scalar, round_scalar,
11227 round_saeonly_scalar): New meta-templates.
11228 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
11229 round_scalar_mask_operand3, round_scalar_mask_op3,
11230 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
11231 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
11232 round_saeonly_scalar_constraint,
11233 round_saeonly_scalar_prefix): New subst attribute.
11234 * config/i386/sse.md
11235 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
11236 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
11237 <round_scalar_name> ... this.
11238 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
11239 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
11240 <round_scalar_name> ... this.
11241 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
11242 <sse>_vm<code><mode>3<mask_scalar_name>
11243 <round_saeonly_scalar_name> ... this.
11244 (v<plusminus_mnemonic><ssescalarmodesuffix>
11245 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
11246 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
11247 v<plusminus_mnemonic><ssescalarmodesuffix>
11248 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
11249 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
11250 (v<multdiv_mnemonic><ssescalarmodesuffix>
11251 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
11252 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
11253 v<multdiv_mnemonic><ssescalarmodesuffix>
11254 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
11255 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
11256 (v<maxmin_float><ssescalarmodesuffix>
11257 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
11258 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
11259 v<maxmin_float><ssescalarmodesuffix>
11260 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
11261 %0<mask_scalar_operand3>, %1, %<iptr>2
11262 <round_saeonly_scalar_mask_op3>} ... this.
11264 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
11266 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
11267 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
11269 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
11270 Alan Hayward <alan.hayward@arm.com>
11271 David Sherwood <david.sherwood@arm.com>
11273 * combine.c (simplify_if_then_else): Remove "enum" before
11275 * compare-elim.c (can_eliminate_compare): Likewise.
11276 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
11278 (aarch64_lookup_simd_builtin_type): Likewise.
11279 (aarch64_simd_builtin_type): Likewise.
11280 (aarch64_init_simd_builtin_types): Likewise.
11281 (aarch64_simd_expand_args): Likewise.
11282 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
11284 (aarch64_reverse_mask): Likewise.
11285 (aarch64_simd_emit_reg_reg_move): Likewise.
11286 (aarch64_gen_adjusted_ldpstp): Likewise.
11287 (aarch64_ccmp_mode_to_code): Likewise.
11288 (aarch64_operands_ok_for_ldpstp): Likewise.
11289 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
11290 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
11292 (aarch64_min_divisions_for_recip_mul): Likewise.
11293 (aarch64_reassociation_width): Likewise.
11294 (aarch64_get_condition_code_1): Likewise.
11295 (aarch64_simd_emit_reg_reg_move): Likewise.
11296 (aarch64_simd_attr_length_rglist): Likewise.
11297 (aarch64_reverse_mask): Likewise.
11298 (aarch64_operands_ok_for_ldpstp): Likewise.
11299 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
11300 (aarch64_gen_adjusted_ldpstp): Likewise.
11301 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
11303 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
11304 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
11305 (arm_lookup_simd_builtin_type): Likewise.
11306 (arm_simd_builtin_type): Likewise.
11307 (arm_init_simd_builtin_types): Likewise.
11308 (arm_expand_builtin_args): Likewise.
11309 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
11310 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
11311 (ft32_setup_incoming_varargs): Likewise.
11312 (ft32_function_arg): Likewise.
11313 (ft32_function_arg_advance): Likewise.
11314 (ft32_pass_by_reference): Likewise.
11315 (ft32_arg_partial_bytes): Likewise.
11316 (ft32_valid_pointer_mode): Likewise.
11317 (ft32_addr_space_pointer_mode): Likewise.
11318 (ft32_addr_space_legitimate_address_p): Likewise.
11319 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
11321 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
11322 (ix86_emit_outlined_ms2sysv_restore): Likewise.
11323 (iamcu_alignment): Likewise.
11324 (canonicalize_vector_int_perm): Likewise.
11325 (ix86_noce_conversion_profitable_p): Likewise.
11326 (ix86_mpx_bound_mode): Likewise.
11327 (ix86_operands_ok_for_move_multiple): Likewise.
11328 * config/microblaze/microblaze-protos.h
11329 (microblaze_expand_conditional_branch_reg): Likewise.
11330 * config/microblaze/microblaze.c
11331 (microblaze_expand_conditional_branch_reg): Likewise.
11332 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
11334 (rs6000_reassociation_width): Likewise.
11335 (rs6000_invalid_binary_op): Likewise.
11336 (fusion_p9_p): Likewise.
11337 (emit_fusion_p9_load): Likewise.
11338 (emit_fusion_p9_store): Likewise.
11339 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
11341 (riscv_hard_regno_mode_ok_p): Likewise.
11342 (riscv_address_insns): Likewise.
11343 (riscv_split_symbol): Likewise.
11344 (riscv_legitimize_move): Likewise.
11345 (riscv_function_value): Likewise.
11346 (riscv_hard_regno_nregs): Likewise.
11347 (riscv_expand_builtin): Likewise.
11348 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
11349 (riscv_build_integer): Likewise.
11350 (riscv_split_integer): Likewise.
11351 (riscv_legitimate_constant_p): Likewise.
11352 (riscv_cannot_force_const_mem): Likewise.
11353 (riscv_regno_mode_ok_for_base_p): Likewise.
11354 (riscv_valid_base_register_p): Likewise.
11355 (riscv_valid_offset_p): Likewise.
11356 (riscv_valid_lo_sum_p): Likewise.
11357 (riscv_classify_address): Likewise.
11358 (riscv_legitimate_address_p): Likewise.
11359 (riscv_address_insns): Likewise.
11360 (riscv_load_store_insns): Likewise.
11361 (riscv_force_binary): Likewise.
11362 (riscv_split_symbol): Likewise.
11363 (riscv_force_address): Likewise.
11364 (riscv_legitimize_address): Likewise.
11365 (riscv_move_integer): Likewise.
11366 (riscv_legitimize_const_move): Likewise.
11367 (riscv_legitimize_move): Likewise.
11368 (riscv_address_cost): Likewise.
11369 (riscv_subword): Likewise.
11370 (riscv_output_move): Likewise.
11371 (riscv_canonicalize_int_order_test): Likewise.
11372 (riscv_emit_int_order_test): Likewise.
11373 (riscv_function_arg_boundary): Likewise.
11374 (riscv_pass_mode_in_fpr_p): Likewise.
11375 (riscv_pass_fpr_single): Likewise.
11376 (riscv_pass_fpr_pair): Likewise.
11377 (riscv_get_arg_info): Likewise.
11378 (riscv_function_arg): Likewise.
11379 (riscv_function_arg_advance): Likewise.
11380 (riscv_arg_partial_bytes): Likewise.
11381 (riscv_function_value): Likewise.
11382 (riscv_pass_by_reference): Likewise.
11383 (riscv_setup_incoming_varargs): Likewise.
11384 (riscv_print_operand): Likewise.
11385 (riscv_elf_select_rtx_section): Likewise.
11386 (riscv_save_restore_reg): Likewise.
11387 (riscv_for_each_saved_reg): Likewise.
11388 (riscv_register_move_cost): Likewise.
11389 (riscv_hard_regno_mode_ok_p): Likewise.
11390 (riscv_hard_regno_nregs): Likewise.
11391 (riscv_class_max_nregs): Likewise.
11392 (riscv_memory_move_cost): Likewise.
11393 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
11394 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
11395 (rl78_addr_space_address_mode): Likewise.
11396 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11398 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
11399 (rs6000_reassociation_width): Likewise.
11400 (rs6000_invalid_binary_op): Likewise.
11401 (fusion_p9_p): Likewise.
11402 (emit_fusion_p9_load): Likewise.
11403 (emit_fusion_p9_store): Likewise.
11404 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
11405 (ok_for_simple_move_operands): Likewise.
11406 (ok_for_simple_move_strict_operands): Likewise.
11407 (ok_for_simple_arith_logic_operands): Likewise.
11408 (visium_legitimize_reload_address): Likewise.
11409 (visium_select_cc_mode): Likewise.
11410 (output_cbranch): Likewise.
11411 (visium_split_double_move): Likewise.
11412 (visium_expand_copysign): Likewise.
11413 (visium_expand_int_cstore): Likewise.
11414 (visium_expand_fp_cstore): Likewise.
11415 * config/visium/visium.c (visium_pass_by_reference): Likewise.
11416 (visium_function_arg): Likewise.
11417 (visium_function_arg_advance): Likewise.
11418 (visium_libcall_value): Likewise.
11419 (visium_setup_incoming_varargs): Likewise.
11420 (visium_legitimate_constant_p): Likewise.
11421 (visium_legitimate_address_p): Likewise.
11422 (visium_legitimize_address): Likewise.
11423 (visium_secondary_reload): Likewise.
11424 (visium_register_move_cost): Likewise.
11425 (visium_memory_move_cost): Likewise.
11426 (prepare_move_operands): Likewise.
11427 (ok_for_simple_move_operands): Likewise.
11428 (ok_for_simple_move_strict_operands): Likewise.
11429 (ok_for_simple_arith_logic_operands): Likewise.
11430 (visium_function_value_1): Likewise.
11431 (rtx_ok_for_offset_p): Likewise.
11432 (visium_legitimize_reload_address): Likewise.
11433 (visium_split_double_move): Likewise.
11434 (visium_expand_copysign): Likewise.
11435 (visium_expand_int_cstore): Likewise.
11436 (visium_expand_fp_cstore): Likewise.
11437 (visium_split_cstore): Likewise.
11438 (visium_select_cc_mode): Likewise.
11439 (visium_split_cbranch): Likewise.
11440 (output_cbranch): Likewise.
11441 (visium_print_operand_address): Likewise.
11442 * expmed.c (flip_storage_order): Likewise.
11443 * expmed.h (emit_cstore): Likewise.
11444 (flip_storage_order): Likewise.
11445 * genrecog.c (validate_pattern): Likewise.
11446 * hsa-gen.c (gen_hsa_addr): Likewise.
11447 * internal-fn.c (expand_arith_overflow): Likewise.
11448 * ira-color.c (allocno_copy_cost_saving): Likewise.
11449 * lra-assigns.c (find_hard_regno_for_1): Likewise.
11450 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
11451 (process_invariant_for_inheritance): Likewise.
11452 * lra-eliminations.c (move_plus_up): Likewise.
11453 * omp-low.c (lower_oacc_reductions): Likewise.
11454 * simplify-rtx.c (simplify_subreg): Likewise.
11455 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
11456 (TARGET_CHKP_BOUND_MODE): Likewise..
11457 * targhooks.c (default_chkp_bound_mode): Likewise.
11458 (default_setup_incoming_vararg_bounds): Likewise.
11459 * targhooks.h (default_chkp_bound_mode): Likewise.
11460 (default_setup_incoming_vararg_bounds): Likewise.
11461 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
11462 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
11463 (have_whole_vector_shift): Likewise.
11464 * tree-vect-stmts.c (vectorizable_load): Likewise.
11465 * doc/tm.texi: Regenerate.
11467 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
11469 Graceful degrade if Binutils PR21472 is not available.
11472 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
11473 .rodata in flash test fails.
11474 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
11475 * confgure: Regenerate.
11476 * config.in: Regenerate.
11477 * config/avr/avr.c (avr_asm_named_section)
11478 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
11479 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
11480 (avr_asm_init_sections): Same.
11482 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
11484 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
11485 (fma<VH:mode>4_intrinsic): Likewise.
11486 (*fmsub<VCVTF:mode>4): Likewise.
11487 (*fmsub<VH:mode>4_intrinsic): Likewise.
11489 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
11492 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
11493 Don't depend on "optimize > 0".
11494 (out_movhi_r_mr, out_movqi_mr_r): Same.
11495 (out_movhi_mr_r, out_movqi_r_mr): Same.
11496 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
11497 io_address_operand on "optimize > 0".
11499 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11501 * tree-loop-distribution.c: Add general explanantion on the pass.
11502 (generate_loops_for_partition): Mark distributed loop.
11503 (pg_add_dependence_edges): New parameter. Handle alias data
11504 dependence specially and record it in the parameter if asked.
11505 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
11506 (init_partition_graph_vertices, add_partition_graph_edge): New.
11507 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
11508 (free_partition_graph_vdata, build_partition_graph): New.
11509 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
11510 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
11511 (data_ref_segment_size, latch_dominated_by_data_ref): New.
11512 (compute_alias_check_pairs, version_loop_by_alias_check): New.
11513 (version_for_distribution_p, finalize_partitions): New.
11514 (distribute_loop): Handle alias data dependence specially. Factor
11515 out loop fusion code as functions and call these functions.
11517 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11519 * tree-loop-distribution.c (classify_partition): New parameter and
11520 better handle reduction statement.
11521 (rdg_build_partitions): Revise comment.
11522 (distribute_loop): Compute statements in all partitions and pass it
11523 to classify_partition.
11525 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11527 * tree-loop-distribution.c (enum partition_type): New.
11528 (struct partition): New field type.
11529 (partition_merge_into): Add parameter. Update partition type.
11530 (data_dep_in_cycle_p, update_type_for_merge): New functions.
11531 (build_rdg_partition_for_vertex): Compute partition type.
11532 (rdg_build_partitions): Dump partition type.
11533 (distribute_loop): Update calls to partition_merge_into.
11535 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11537 * tree-loop-distribution.c (struct ddr_hasher): New.
11538 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
11540 (classify_partition): Call get_data_dependence.
11541 (pg_add_dependence_edges): Ditto.
11542 (distribute_loop): Release data dependence hash table.
11544 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11546 * tree-loop-distribution.c (ref_base_address): Delete.
11547 (similar_memory_accesses): Rename ...
11548 (share_memory_accesses): ... to this. Check if partitions access
11549 the same memory reference.
11550 (distribute_loop): Call share_memory_accesses.
11552 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11554 * tree-loop-distribution.c (struct partition): New field recording
11555 its data reference.
11556 (partition_alloc, partition_free): Init and release data refs.
11557 (partition_merge_into): Merge data refs.
11558 (build_rdg_partition_for_vertex): Collect data refs for partition.
11559 (pg_add_dependence_edges): Change parameters from vector to bitmap.
11561 (distribute_loop): Remve data refs from vertice data of partition
11564 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11566 * tree-loop-distribution.c (params.h): Include header file.
11567 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
11568 (datarefs_vec): New global var.
11569 (create_rdg_vertices): Use datarefs_vec directly.
11570 (free_rdg): Don't free data references.
11571 (build_rdg): Update use. Don't free data references.
11572 (distribute_loop): Compute global variable for data references.
11573 Bail out if there are too many data references.
11575 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11577 * tree-loop-distribution.c (loop_nest): New global var.
11578 (build_rdg): Use loop directly, rather than loop nest.
11579 (pg_add_dependence_edges): Remove loop nest parameter. Use global
11581 (distribute_loop): Compute global variable loop nest. Update use.
11583 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11585 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
11586 (partition_merge_into): New parameter. Dump reason for fusion.
11587 (distribute_loop): Update use of partition_merge_into.
11589 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11591 * tree-loop-distribution.c (bb_top_order_index): New.
11592 (bb_top_order_index_size, bb_top_order_cmp): New.
11593 (stmts_from_loop): Use topological order.
11594 (pass_loop_distribution::execute): Compute and release topological
11595 order for basic blocks.
11597 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11599 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
11602 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11604 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
11605 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
11606 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
11607 * internal-fn.def (LOOP_DIST_ALIAS): New.
11608 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
11609 (fold_loop_internal_call): ... this.
11610 (vect_loop_dist_alias_call): New function.
11611 (set_uid_loop_bbs): Call fold_loop_internal_call.
11612 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
11615 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
11618 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
11619 Require dead FLAGS_REG at the beginning of a peephole.
11621 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
11624 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
11625 arguments in the call to __builtin_ia32_sbb_u32.
11626 (_subborrow_u64): Swap _X and _Y arguments in the call to
11627 __builtin_ia32_sbb_u64.
11629 2017-07-04 Jakub Jelinek <jakub@redhat.com>
11632 * tree-vrp.c (compare_assert_loc): Turn into a function template
11633 with stable template parameter. Only test if a->e is NULL,
11634 !a->e == !b->e has been verified already. Use e == NULL or
11635 e != NULL instead of e or ! e tests. If stable is true, don't use
11636 iterative_hash_expr, on the other side allow a or b or both NULL
11637 and sort the NULLs last.
11638 (process_assert_insertions): Sort using compare_assert_loc<false>
11639 instead of compare_assert_loc, later sort using
11640 compare_assert_loc<true> before calling process_assert_insertions_for
11641 in a loop. Use break instead of continue once seen NULL pointer.
11643 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
11645 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
11646 Cortex-R7 and Cortex-R8 processors.
11648 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
11650 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
11651 uninitialized while src is not.
11653 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
11655 * common/config/arm/arm-common.c: Adjust include path for
11657 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
11658 (arm-cpu.h): Create in build directory. Adjust dependency rules.
11659 (arm-cpu-data.h): Likewise.
11660 (arm-cpu-cdata.h): Likewise.
11661 * config/arm/arm-cpu.h: Delete.
11662 * config/arm/arm-cpu-cdata.h: Delete.
11663 * config/arm/arm-cpu-data.h: Delete.
11665 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
11667 * config/arm/arm-cpus.in (cortex-a55): New.
11668 (cortex-a75): Likewise.
11669 (cortex-a75.cortex-a55): Likewise.
11670 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
11672 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
11673 * config/arm/arm-cpu-cdata.h: Regenerate.
11674 * config/arm/arm-cpu-data.h: Regenerate.
11675 * config/arm/arm-cpu.h: Regenerate.
11676 * config/arm/arm-tables.opt: Regenerate.
11677 * config/arm/arm-tune.md: Regenerate.
11679 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
11681 * haifa-sched.c (sched_create_recovery_edges): Update profile.
11683 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
11685 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
11688 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
11690 PR tree-optimization/81292
11691 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
11692 full_string_p, also call adjust_related_strinfos if the adjustment
11693 is simple, otherwise invalidate related strinfos.
11695 2017-07-04 Martin Liska <mliska@suse.cz>
11698 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
11699 newly created variable as DECL_IGNORED_P.
11701 2017-07-04 Martin Liska <mliska@suse.cz>
11704 * ipa-inline.c (inline_small_functions):
11705 Use xstrdup_for_dump.
11707 2017-07-04 Tom de Vries <tom@codesourcery.com>
11709 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
11711 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
11714 * config/darwin.c (darwin_function_switched_text_sections):
11717 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
11719 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
11721 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
11723 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
11725 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11727 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
11728 min_profitable_iters, and th as inclusive lower bounds.
11729 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
11730 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
11731 for min_profitable_iters and min_profitable_estimate.
11732 (vect_transform_loop): Treat th as an inclusive lower bound.
11733 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
11735 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
11738 * config/darwin.c (darwin_function_switched_text_sections):
11739 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
11740 in two pieces, and suppress the use of buf.
11742 2017-07-03 Nathan Sidwell <nathan@acm.org>
11744 * hash-table.h (hash_table_mod1): Fix indentation.
11746 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11748 PR middle-end/81290
11749 * predict.c (force_edge_cold): Be more careful about propagation
11751 * profile-count.h (profile_probability::guessed,
11752 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
11754 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
11756 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
11758 * doc/invoke.texi (rcpc architecture extension): Document it.
11760 2017-07-03 Richard Biener <rguenther@suse.de>
11762 PR tree-optimization/60510
11763 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
11764 the scalar reduction PHI and use it.
11765 (vectorizable_reduction): Properly guard the single_defuse_cycle
11766 path for non-SLP reduction chains where we cannot use it.
11767 Rework reduc_def/index and vector type deduction. Rework
11768 vector operand gathering during reduction op code-gen.
11769 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
11770 chains dissolve the chain and leave it to non-SLP reduction
11773 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11775 * tree-data-ref.h (dr_alignment): Declare.
11776 * tree-data-ref.c (dr_alignment): New function.
11777 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
11778 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
11780 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
11782 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11784 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
11785 and base_misalignment fields.
11786 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
11787 * tree-data-ref.c: Include builtins.h.
11788 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
11789 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
11790 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
11791 * tree-vect-data-refs.c: Include tree-cfg.h.
11792 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
11793 fields instead of calculating an alignment here.
11794 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
11795 innermost_loop_behavior fields.
11797 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11799 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
11801 (DR_STEP_ALIGNMENT): New macro.
11802 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
11803 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
11804 (create_data_ref): Print it.
11805 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
11806 to tell whether the step preserves vector (mis)alignment.
11807 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
11808 Move the check for an integer step and generalise to all INTEGER_CST.
11809 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
11810 Print the outer step alignment.
11812 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11814 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
11815 with offset_alignment.
11816 (DR_ALIGNED_TO): Delete.
11817 (DR_OFFSET_ALIGNMENT): New macro.
11818 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
11819 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
11820 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
11821 (create_data_ref): Likewise.
11822 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
11823 (vect_analyze_data_refs): Likewise.
11824 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
11825 creating dummy innermost behavior.
11827 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11829 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
11830 with a "innermost_loop_behavior *" and refeence tree.
11831 * tree-data-ref.c (dr_analyze_innermost): Likewise.
11832 (create_data_ref): Update call accordingly.
11833 * tree-predcom.c (find_looparound_phi): Likewise.
11835 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11837 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
11838 fields with dr_wrt_vec_loop.
11839 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
11840 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
11841 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
11842 (vect_dr_behavior): New function.
11843 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
11844 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
11845 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
11846 track whether the step preserves the misalignment.
11847 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
11848 Use vect_dr_behavior.
11849 (vect_setup_realignment): Update call accordingly.
11850 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
11851 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
11852 call to vect_create_addr_base_for_vector_ref.
11853 (vect_create_cond_for_align_checks): Likewise.
11854 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
11855 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
11856 (vect_recog_mask_conversion_pattern): Likewise.
11857 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
11858 (new_stmt_vec_info): Remove redundant zeroing.
11860 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
11862 * common/config/arm/arm-common.c (arm_be8_option): New function.
11863 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
11864 (ISA_ARMv6): Add isa_bit_be8.
11865 * config/arm/arm.h (arm_be8_option): Add prototype.
11866 (BE8_SPEC_FUNCTION): New define.
11867 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
11868 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
11869 (mlittle-endian): Similarly.
11870 (mbe8, mbe32): New options.
11871 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
11872 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
11874 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11876 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
11878 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11880 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
11881 (cleanup_tree_cfg_bb): Use it.
11882 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
11884 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
11886 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11889 * loop-doloop.c (add_test): Update profile.
11891 2017-07-03 Martin Liska <mliska@suse.cz>
11894 * sanopt.c (rewrite_usage_of_param): New function.
11895 (sanitize_rewrite_addressable_params): Likewise.
11896 (pass_sanopt::execute): Call rewrite_usage_of_param.
11898 2017-07-03 Richard Biener <rguenther@suse.de>
11900 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
11901 back to using VIEW_CONVERT_EXPR.
11903 2017-07-03 Martin Liska <mliska@suse.cz>
11906 * doc/extend.texi: Document when a resolver function is
11907 generated for target_clones.
11909 2017-07-03 Martin Liska <mliska@suse.cz>
11911 * asan.c (asan_emit_stack_protection): Unpoison just red zones
11912 and shadow memory of auto variables which are subject of
11913 use-after-scope sanitization.
11914 (asan_expand_mark_ifn): Add do set only when is_poison.
11916 2016-07-03 Richard Biener <rguenther@suse.de>
11918 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
11920 (vect_force_simple_reduction): Record reduction def -> phi mapping.
11921 (vectorizable_reduction): Perform reduction PHI creation when
11922 visiting a reduction PHI and adjust and simplify code generation
11923 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
11924 (vect_transform_loop): Visit reduction PHIs.
11925 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
11926 defs into the SLP tree.
11927 (vect_build_slp_tree): Reduction defs terminate the recursion.
11928 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
11930 (vect_get_vec_defs_for_stmt_copy): Export.
11931 (vect_get_vec_defs): Likewise.
11932 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
11934 (vect_get_vec_defs_for_stmt_copy): Declare.
11935 (vect_get_vec_defs): Likewise.
11937 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11939 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
11940 parameter with a "loop" parameter and use it instead of the
11941 loop containing DR_STMT. Don't check simple_iv when doing
11942 BB analysis. Describe the two analysis modes in the comment.
11944 2017-07-03 Tom de Vries <tom@codesourcery.com>
11946 PR tree-optimization/69468
11947 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
11948 (find_same_succ_bb): Handle ignore_edge_flags.
11950 2017-07-03 Tom de Vries <tom@codesourcery.com>
11952 PR tree-optimization/81192
11953 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
11955 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
11957 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
11959 2017-07-03 Tom de Vries <tom@codesourcery.com>
11961 PR tree-optimization/81192
11962 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
11963 BB_SAME_SUCC (bb) == NULL.
11965 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11967 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
11970 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11972 * dumpfile.c: Include profile-count.h
11973 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
11975 (insert_cond_bb): Update profile.
11976 * tree-cfg.h (insert_cond_bb): Update prototype.
11977 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
11978 * tree-dump.c: Do not include tree-cfg.
11980 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11982 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
11984 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11986 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
11989 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11991 * tree-complex.c (expand_complex_div_wide): update profile.
11993 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
11994 Alan Hayward <alan.hayward@arm.com>
11995 David Sherwood <david.sherwood@arm.com>
11997 * Makefile.in (MACHMODE_H): Remove insn-modes.h
11998 (CORETYPES_H): New define.
11999 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
12000 (insn-modes-inline.h, s-modes-inline-h): New rules.
12001 (generated_files): Add insn-modes-inline.h.
12002 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
12003 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
12004 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
12005 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
12006 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
12007 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
12008 (build/gencodes.o, build/genconditions.o): Likewise.
12009 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
12010 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
12011 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
12012 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
12013 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
12014 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
12015 * coretypes.h: Include everything up to real.h for generators.
12016 Include insn-modes.h first. Include wide-int-print.h after
12017 wide-int.h. Include insn-modes-inline.h and then machmode.h.
12018 * machmode.h: Don't include insn-modes.h here.
12019 * function-tests.c: Remove includes of signop.h, machmode.h,
12020 double-int.h and wide-int.h.
12022 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
12024 * optc-save-gen.awk: Likewise.
12025 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
12026 * godump.c: Remove include of wide-int-print.h.
12027 * pretty-print.h: Likewise.
12028 * wide-int-print.cc: Likewise.
12029 * wide-int.cc: Likewise.
12030 * hash-map-tests.c: Remove include of signop.h.
12031 * hash-set-tests.c: Likewise.
12032 * rtl-tests.c: Likewise.
12033 * mkconfig.sh: Remove include of machmode.h.
12034 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
12036 (emit_insn_modes_inline_h): ...this new function. Emit the code
12037 into an insn-modes-inline.h header file, adding appropriate
12038 include guards and end comments.
12039 (emit_insn_modes_c_header): Remove include of machmode.h.
12040 (emit_min_insn_modes_c_header): Include coretypes.h rather than
12042 (main): Handle -i flag and call emit_insn_modes_inline_h when
12045 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
12047 * tree-ssa-strlen.c (strinfo): Rename the length field to
12048 nonzero_chars. Add a full_string_p field.
12049 (compare_nonzero_chars, zero_length_string_p): New functions.
12050 (get_addr_stridx): Add an offset_out parameter.
12051 Use compare_nonzero_chars.
12052 (get_stridx): Update accordingly. Use compare_nonzero_chars.
12053 (new_strinfo): Update after above changes to strinfo.
12054 (set_endptr_and_length): Set full_string_p.
12055 (get_string_length): Update after above changes to strinfo.
12056 (unshare_strinfo): Update call to new_strinfo.
12057 (maybe_invalidate): Likewise.
12058 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
12059 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
12060 as a uhwi instead of an shwi. Update after above changes to
12061 strinfo and new_strinfo.
12062 (zero_length_string): Assert that chainsi contains full strings.
12063 Use zero_length_string_p. Update call to new_strinfo.
12064 (adjust_related_strinfos): Update after above changes to strinfo.
12065 Copy full_string_p from origsi.
12066 (adjust_last_stmt): Use zero_length_string_p.
12067 (handle_builtin_strlen): Update after above changes to strinfo and
12068 new_strinfo. Install the lhs as the string length if the previous
12069 entry didn't describe a full string.
12070 (handle_builtin_strchr): Update after above changes to strinfo
12072 (handle_builtin_strcpy): Likewise.
12073 (handle_builtin_strcat): Likewise.
12074 (handle_builtin_malloc): Likewise.
12075 (handle_pointer_plus): Likewise.
12076 (handle_builtin_memcpy): Likewise. Track nonzero characters
12077 that aren't necessarily followed by a nul terminator.
12078 (handle_char_store): Likewise.
12080 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
12082 PR tree-optimization/80769
12083 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
12084 for malloc and calloc. Document the new invariant that all related
12085 strinfos have delayed lengths or none do.
12086 (verify_related_strinfos): Move earlier in file.
12087 (set_endptr_and_length): New function, split out from...
12088 (get_string_length): ...here. Also set the lengths of related
12090 (zero_length_string): Assert that chainsi has known (rather than
12092 (adjust_related_strinfos): Likewise.
12094 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
12096 PR tree-optimization/81136
12097 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
12098 assert that two references with the same misalignment have the same
12099 compile-time misalignment if those compile-time misalignments
12102 2017-07-01 Andi Kleen <ak@linux.intel.com>
12104 * print-tree.c (print_node): Print all attributes.
12106 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12108 * cfg.c (scale_bbs_frequencies): New function.
12109 * cfg.h (scale_bbs_frequencies): Declare it.
12110 * cfgloopanal.c (single_likely_exit): Cleanup.
12111 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
12113 (scale_loop_profile): Likewise.
12114 (loop_version): Likewise.
12115 (create_empty_loop_on_edge): Update.
12116 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
12117 scale_loop_frequencies, scale_loop_profile, loopify,
12118 loop_version): Update prototypes.
12119 * modulo-sched.c (sms_schedule): Update.
12120 * predict.c (unlikely_executed_edge_p): Also check probability.
12121 (probably_never_executed_edge_p): Fix typo.
12122 * tree-if-conv.c (version_loop_for_if_conversion): Update.
12123 * tree-parloops.c (gen_parallel_loop): Update.
12124 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
12125 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
12126 * tree-ssa-loop-split.c (split_loop): Update.
12127 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
12128 * tree-vect-loop-manip.c (vect_do_peeling): Update.
12129 (vect_loop_versioning): Update.
12130 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
12132 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12134 * trans-mem.c (split_bb_make_tm_edge): Update profile.
12136 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12138 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
12139 to keep profile consistent.
12141 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12143 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
12144 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
12145 * profile-count.h (max_safe_multiplier): Make unsigned.
12146 (profile_count::guessed_zero): New.
12148 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12150 * bb-reorder.c (fix_up_crossing_landing_pad,
12151 fix_crossing_conditional_branches): Use make_single_succ_edge
12152 to keep profile consistent.
12154 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12156 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
12159 2017-07-01 Jakub Jelinek <jakub@redhat.com>
12162 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
12163 the right scopes, make sure cond_jump isn't preserved between multiple
12164 iterations. Search for fallthru edge whenever there are 3+ edges and
12165 use find_fallthru_edge for it.
12167 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
12169 Patch by Alexander Monakov <amonakov@ispras.ru>
12170 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
12171 probabilities consistently.
12173 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
12175 * pa.c (pa_expand_compare_and_swap_loop): Update call of
12176 emit_cmp_and_jump_insns.
12178 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
12181 * tree-inline.c (expand_call_inline): Combine profile statuses.
12183 2017-06-30 Andrew Pinski <apinski@cavium.com>
12185 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
12186 fold_stmt returned true.
12188 2017-06-30 Nathan Sidwell <nathan@acm.org>
12190 * ggc.h (empty_string): Delete.
12191 * cfgexpand.c (expand_asm_stmt): Use plain "".
12192 * optabs.c (expand_asm_memory_barrier): Likewise.
12193 * stringpool.c (empty_string): Delete.
12194 (digit_vector, digit_string): Delete.
12195 (ggc_alloc_string): Use plain "", don't optimize single digit
12196 strings. Use ggc_alloc_atomic.
12198 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
12200 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
12201 comparison set and one other set, use the cost of the non-comparison
12204 2017-06-30 Nathan Sidwell <nathan@acm.org>
12206 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
12209 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
12211 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
12212 loops. Remove now unneeded calls to gimple_switch_set_label() that
12213 just set removed labels to NULL_TREE.
12215 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
12217 * tree-ssanames.c (set_range_info_raw): Abstract from ...
12218 (set_range_info): ...here. Only call set_range_info_raw if domain
12220 (set_nonzero_bits): Call set_range_info_raw.
12221 * tree-ssanames.h (set_range_info_raw): New.
12223 2017-06-30 Jakub Jelinek <jakub@redhat.com>
12226 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
12227 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
12228 of nonimmediate_operand and <store_mask_constraint> instead of m
12229 for the input operand. For V8FI iterator, always split if input
12230 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
12231 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
12232 <store_mask_predicate> instead of register_operand and
12233 <store_mask_constraint> instead of v for the input operand. Make
12234 sure both operands aren't MEMs for if not <mask_applied>.
12236 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
12238 * lto-wrapper.c (copy_file) Close both file descriptors before
12241 2017-06-30 Martin Liska <mliska@suse.cz>
12244 * multiple_target.c (create_dispatcher_calls): Make ifunc
12245 also for function that don't have calls or are not referenced.
12247 2017-06-30 Richard Biener <rguenther@suse.de>
12249 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
12250 analyze the first scalar stmt. Move vector type computation
12251 for the BB case here from ...
12252 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
12253 live operation processing in the SLP case properly.
12255 2017-06-30 Richard Biener <rguenther@suse.de>
12257 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
12259 2017-06-30 Martin Liska <mliska@suse.cz>
12262 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
12263 before BUILT_IN_UNWIND_RESUME when ASAN is used.
12265 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
12267 * doc/invoke.texi (AArch64): Add missing options and remove redundant
12270 2017-06-30 Richard Biener <rguenther@suse.de>
12272 PR tree-optimization/81249
12273 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
12274 condition reduction result to original scalar type.
12276 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12278 * profile-count.h (enum profile_quality): Fix typos and whitespace
12281 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12283 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
12284 type for branch probabilities.
12286 2017-06-29 Julian Brown <julian@codesourcery.com>
12287 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
12289 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
12290 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
12291 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
12292 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
12294 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
12296 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
12297 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
12298 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
12299 CC usage from generic code to here.
12300 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
12301 CC usage into the target macros.
12303 2017-06-29 Maya Rashish <coypu@sdf.org>
12305 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
12308 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
12310 * arm/arm-builtins.c: Include profile-count.h
12311 * except.c (sjlj_emit_function_enter): Use
12312 profile_probability::unlikely.
12314 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
12316 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
12317 and tocrel_offset be pointer args rather than implicitly using
12319 (legitimate_constant_pool_address_p, rs6000_emit_move,
12320 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
12321 tocrel_offset and use in toc_relative_expr_p call.
12322 (print_operand, print_operand_address): Use static tocrel_base_oac
12323 and tocrel_offset_oac.
12324 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
12327 2017-06-29 Maya Rashish <coypu@sdf.org>
12329 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
12331 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
12333 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
12334 objects, take into account only the alignment of 'op0' and 'mode1' if
12337 2017-06-29 Steve Ellcey <sellcey@cavium.com>
12339 * ccmp.c (ccmp_tree_comparison_p): New function.
12340 (ccmp_candidate_p): Update to use above function.
12341 (get_compare_parts): New function.
12342 (expand_ccmp_next): Update to use new functions.
12343 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
12345 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
12346 take mode as argument.
12347 * ccmp.h (expand_ccmp_expr): Add mode as argument.
12348 * expr.c (expand_expr_real_1): Pass mode as argument.
12350 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
12352 * combine.c (combine_instructions): Print insns to dump_file, together
12355 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
12357 * asan.c (asan_emit_stack_protection): Update.
12358 (create_cond_insert_point): Update.
12359 * auto-profile.c (afdo_propagate_circuit): Update.
12360 * basic-block.h (struct edge_def): Turn probability to
12361 profile_probability.
12362 (EDGE_FREQUENCY): Update.
12363 * bb-reorder.c (find_traces_1_round): Update.
12364 (better_edge_p): Update.
12365 (sanitize_hot_paths): Update.
12366 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
12367 (make_single_succ_edge): Update.
12368 (check_bb_profile): Update.
12369 (dump_edge_info): Update.
12370 (update_bb_profile_for_threading): Update.
12371 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
12372 probabilitycount to 0.
12373 * cfgbuild.c (compute_outgoing_frequencies): Update.
12374 * cfgcleanup.c (try_forward_edges): Update.
12375 (outgoing_edges_match): Update.
12376 (try_crossjump_to_edge): Update.
12377 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
12378 (expand_gimple_tailcall): Update.
12379 (construct_init_block): Use make_single_succ_edge.
12380 (construct_exit_block): Use make_single_succ_edge.
12381 * cfghooks.c (verify_flow_info): Update.
12382 (redirect_edge_succ_nodup): Update.
12383 (split_edge): Update.
12384 (account_profile_record): Update.
12385 * cfgloopanal.c (single_likely_exit): Update.
12386 * cfgloopmanip.c (scale_loop_profile): Update.
12387 (set_zero_probability): Remove.
12388 (duplicate_loop_to_header_edge): Update.
12389 * cfgloopmanip.h (loop_version): Update prototype.
12390 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
12391 (force_nonfallthru_and_redirect): Update.
12392 (update_br_prob_note): Update.
12393 (rtl_verify_edges): Update.
12394 (purge_dead_edges): Update.
12395 (rtl_lv_add_condition_to_bb): Update.
12396 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
12397 * cgraphunit.c (init_lowered_empty_function): Update.
12398 (cgraph_node::expand_thunk): Update.
12399 * cilk-common.c: Include profile-count.h
12400 * dojump.c (inv): Remove.
12401 (jumpifnot): Update.
12402 (jumpifnot_1): Update.
12403 (do_jump_1): Update.
12405 (do_jump_by_parts_greater_rtx): Update.
12406 (do_compare_rtx_and_jump): Update.
12407 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
12408 do_jump_1. do_compare_rtx_and_jump): Update prototype.
12409 * dwarf2cfi.c: Include profile-count.h
12410 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
12411 (sjlj_emit_dispatch_table): Likewise.
12412 * explow.c: Include profile-count.h
12413 * expmed.c (emit_store_flag_force): Update.
12414 (do_cmp_and_jump): Update.
12415 * expr.c (compare_by_pieces_d::generate): Update.
12416 (compare_by_pieces_d::finish_mode): Update.
12417 (emit_block_move_via_loop): Update.
12418 (store_expr_with_bounds): Update.
12419 (store_constructor): Update.
12420 (expand_expr_real_2): Update.
12421 (expand_expr_real_1): Update.
12422 * expr.h (try_casesi, try_tablejump): Update prototypes.
12423 * gimple-pretty-print.c (dump_probability): Update.
12424 (dump_profile): New.
12425 (dump_gimple_label): Update.
12426 (dump_gimple_bb_header): Update.
12427 * graph.c (draw_cfg_node_succ_edges): Update.
12428 * hsa-gen.c (convert_switch_statements): Update.
12429 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
12430 (find_if_case_1): Update.
12431 (find_if_case_2): Update.
12432 * internal-fn.c (expand_arith_overflow_result_store): Update.
12433 (expand_addsub_overflow): Update.
12434 (expand_neg_overflow): Update.
12435 (expand_mul_overflow): Update.
12436 (expand_vector_ubsan_overflow): Update.
12437 * ipa-cp.c (good_cloning_opportunity_p): Update.
12438 * ipa-split.c (split_function): Use make_single_succ_edge.
12439 * ipa-utils.c (ipa_merge_profiles): Update.
12440 * loop-doloop.c (add_test): Update.
12441 (doloop_modify): Update.
12442 * loop-unroll.c (compare_and_jump_seq): Update.
12443 (unroll_loop_runtime_iterations): Update.
12444 * lra-constraints.c (lra_inheritance): Update.
12445 * lto-streamer-in.c (input_cfg): Update.
12446 * lto-streamer-out.c (output_cfg): Update.
12447 * mcf.c (adjust_cfg_counts): Update.
12448 * modulo-sched.c (sms_schedule): Update.
12449 * omp-expand.c (expand_omp_for_init_counts): Update.
12450 (extract_omp_for_update_vars): Update.
12451 (expand_omp_ordered_sink): Update.
12452 (expand_omp_for_ordered_loops): Update.
12453 (expand_omp_for_generic): Update.
12454 (expand_omp_for_static_nochunk): Update.
12455 (expand_omp_for_static_chunk): Update.
12456 (expand_cilk_for): Update.
12457 (expand_omp_simd): Update.
12458 (expand_omp_taskloop_for_outer): Update.
12459 (expand_omp_taskloop_for_inner): Update.
12460 * omp-simd-clone.c (simd_clone_adjust): Update.
12461 * optabs.c (expand_doubleword_shift): Update.
12462 (expand_abs): Update.
12463 (emit_cmp_and_jump_insn_1): Update.
12464 (expand_compare_and_swap_loop): Update.
12465 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
12466 * predict.c (predictable_edge_p): Update.
12467 (edge_probability_reliable_p): Update.
12468 (set_even_probabilities): Update.
12469 (combine_predictions_for_insn): Update.
12470 (combine_predictions_for_bb): Update.
12471 (propagate_freq): Update.
12472 (estimate_bb_frequencies): Update.
12473 (force_edge_cold): Update.
12474 * profile-count.c (profile_count::dump): Add missing space into dump.
12475 (profile_count::debug): Add newline.
12476 (profile_count::differs_from_p): Explicitly convert to unsigned.
12477 (profile_count::stream_in): Update.
12478 (profile_probability::dump): New member function.
12479 (profile_probability::debug): New member function.
12480 (profile_probability::differs_from_p): New member function.
12481 (profile_probability::differs_lot_from_p): New member function.
12482 (profile_probability::stream_in): New member function.
12483 (profile_probability::stream_out): New member function.
12484 * profile-count.h (profile_count_quality): Rename to ...
12485 (profile_quality): ... this one.
12486 (profile_probability): New.
12487 (profile_count): Update.
12488 * profile.c (compute_branch_probabilities): Update.
12489 * recog.c (peep2_attempt): Update.
12490 * sched-ebb.c (schedule_ebbs): Update.
12491 * sched-rgn.c (find_single_block_region): Update.
12492 (compute_dom_prob_ps): Update.
12493 (schedule_region): Update.
12494 * sel-sched-ir.c (compute_succs_info): Update.
12495 * stmt.c (struct case_node): Update.
12496 (do_jump_if_equal): Update.
12497 (get_outgoing_edge_probs): Update.
12498 (conditional_probability): Update.
12499 (emit_case_dispatch_table): Update.
12500 (expand_case): Update.
12501 (expand_sjlj_dispatch_table): Update.
12502 (emit_case_nodes): Update.
12503 * targhooks.c: Update.
12504 * tracer.c (better_p): Update.
12505 (find_best_successor): Update.
12506 * trans-mem.c (expand_transaction): Update.
12507 * tree-call-cdce.c: Update.
12508 * tree-cfg.c (gimple_split_edge): Upate.
12509 (move_sese_region_to_fn): Upate.
12510 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
12511 * tree-eh.c (lower_resx): Upate.
12512 (cleanup_empty_eh_move_lp): Upate.
12513 * tree-if-conv.c (version_loop_for_if_conversion): Update.
12514 * tree-inline.c (copy_edges_for_bb): Update.
12515 (copy_cfg_body): Update.
12516 * tree-parloops.c (gen_parallel_loop): Update.
12517 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
12518 (gimple_gen_time_profiler): Update.
12519 * tree-ssa-dce.c (remove_dead_stmt): Update.
12520 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
12521 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
12522 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
12523 (unloop_loops): Update.
12524 (try_peel_loop): Update.
12525 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
12526 * tree-ssa-loop-split.c (connect_loops): Update.
12527 (split_loop): Update.
12528 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
12529 (hoist_guard): Update.
12530 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
12531 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
12532 (value_replacement): Update.
12533 * tree-ssa-reassoc.c (branch_fixup): Update.
12534 * tree-ssa-tail-merge.c (replace_block_by): Update.
12535 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
12536 (create_edge_and_update_destination_phis): Update.
12537 (compute_path_counts): Update.
12538 (recompute_probabilities): Update.
12539 (update_joiner_offpath_counts): Update.
12540 (freqs_to_counts_path): Update.
12541 (duplicate_thread_path): Update.
12542 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
12543 (struct switch_conv_info): Update.
12544 (gen_inbound_check): Update.
12545 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
12546 (vect_do_peeling): Update.
12547 (vect_loop_versioning): Update.
12548 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
12549 (optimize_mask_stores): Update.
12550 * ubsan.c (ubsan_expand_null_ifn): Update.
12551 * value-prof.c (gimple_divmod_fixed_value): Update.
12552 (gimple_divmod_fixed_value_transform): Update.
12553 (gimple_mod_pow2): Update.
12554 (gimple_mod_pow2_value_transform): Update.
12555 (gimple_mod_subtract): Update.
12556 (gimple_mod_subtract_transform): Update.
12557 (gimple_ic): Update.
12558 (gimple_stringop_fixed_value): Update.
12559 (gimple_stringops_transform): Update.
12560 * value-prof.h: Update.
12562 2017-06-29 Carl Love <cel@us.ibm.com>
12564 * config/rs6000/rs6000-c.c: Add support for built-in functions
12565 vector signed int vec_signed (vector float);
12566 vector signed long long vec_signed (vector double);
12567 vector signed int vec_signed2 (vector double, vector double);
12568 vector signed int vec_signede (vector double);
12569 vector signed int vec_signedo (vector double);
12570 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
12571 instruction generator.
12572 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
12573 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
12574 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
12576 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
12577 vunsignede_v2df): Add define_expands.
12578 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
12579 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
12580 VEC_UNSIGNEDO): Add definitions.
12581 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
12582 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
12583 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
12584 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
12585 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
12586 * config/rs6000/altivec.h (vec_signed, vec_signed2,
12587 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
12588 vec_unsignede, vec_unsignedo): Add builtin defines.
12589 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
12591 * doc/extend.texi: Update the built-in documentation file for the
12592 new built-in functions.
12594 2017-06-29 Richard Biener <rguenther@suse.de>
12596 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
12597 reduction chains to LOOP_VINFO_REDUCTIONS.
12598 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
12599 SLP reductions after processing reduction chains.
12601 2017-06-29 Nathan Sidwell <nathan@acm.org>
12603 * builtins.c (fold_builtin_FUNCTION): Use
12604 lang_hooks.decl_printable_name.
12606 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
12608 PR middle-end/81194
12609 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
12610 with only one label.
12611 * stmt.c (expand_case): Assert NCASES is greater than one.
12613 2017-06-29 Richard Biener <rguenther@suse.de>
12615 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
12617 (group_case_labels): Likewise.
12618 (find_taken_edge): Push sanity checking on val to workers...
12619 (find_taken_edge_cond_expr): ... here
12620 (find_taken_edge_switch_expr): ... and here, handle cases
12621 with just a default label.
12622 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
12623 (group_case_labels): Likewise.
12624 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
12625 group_case_labels does anything cleanup the CFG again.
12627 2017-06-29 Bin Cheng <bin.cheng@arm.com>
12629 PR tree-optimization/81196
12630 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
12631 exit condition comparing two IVs.
12633 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
12635 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
12636 profile to the dummy entry at the end of the list of architectures.
12637 * config/arm/arm-cpu-cdata.h: Regenerated.
12639 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12640 Michael Collison <michael.collison@arm.com>
12643 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
12645 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
12646 (*aarch64_reg_<mode>3_minus_mask): New pattern.
12647 (*aarch64_<optab>_reg_di3_mask2): New pattern.
12648 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
12649 of shift when the shift amount is masked with constant equal to
12650 the size of the mode.
12651 * config/aarch64/predicates.md (subreg_lowpart_operator): New
12654 2017-06-29 Martin Liska <mliska@suse.cz>
12656 * config/i386/i386.opt: Change range from [1,5] to [0,5].
12658 2017-06-29 Yury Gribov <tetra2005@gmail.com>
12661 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
12664 (edge_growth_cache_entry::edge_growth_cache_entry): New
12666 (reset_edge_growth_cache): Update to use constructor.
12668 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
12670 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
12671 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
12672 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
12674 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
12676 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
12677 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
12679 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
12681 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
12682 (*-linux-uclibc*): Add t-uclibc tmake_file.
12683 * config/t-musl: New.
12684 * config/t-uclibc: New.
12686 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
12688 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
12690 (gen_comm_data): Emit architectural setting of arch_prof.
12691 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
12693 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
12694 (armv8-m.base, armv8-m.main): Likewise.
12695 * arm-protos.h (arm_build_target): Add profile field.
12696 (arch_option): Likewise.
12697 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
12699 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
12700 arm_active_target.profile.
12702 2017-06-28 Richard Biener <rguenther@suse.de>
12704 PR middle-end/81227
12705 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
12706 TYPE_OVERFLOW_WRAPS.
12707 * match.pd (negate_expr_p): Likewise.
12708 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
12709 fold_build2, not fold_binary.
12711 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
12713 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
12714 Convert memory address to Pmode.
12715 (aarch64_print_operand): Assert MEM operands are always Pmode.
12717 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
12720 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
12721 Remove redundant if.
12722 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
12723 * config/arm/aarch-common-protos.h
12724 (aarch_forward_to_shift_is_not_shifted_re): Remove.
12725 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
12727 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
12730 * multiple_target.c (create_dispatcher_calls): Set the default
12731 clone to be static, not public.
12733 2017-06-28 Richard Biener <rguenther@suse.de>
12735 * tree-vect-loop.c (vectorizable_reduction): Move special
12736 cond reduction IV var creation ...
12737 (vect_create_epilog_for_reduction): ... here. Remove induction_index
12738 parameter. Use STMT_VINFO_VECTYPE.
12739 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
12742 2017-06-28 Martin Liska <mliska@suse.cz>
12745 * ipa-visibility.c (non_local_p): Handle visibility.
12747 2017-06-28 Martin Liska <mliska@suse.cz>
12750 * common.opt: Add IntegerRange to various options.
12751 * opt-functions.awk (integer_range_info): New function.
12752 * optc-gen.awk: Add integer_range_info to cl_options struct.
12753 * opts-common.c (decode_cmdline_option): Handle
12754 CL_ERR_INT_RANGE_ARG.
12755 (cmdline_handle_error): Likewise.
12756 * opts.c (print_filtered_help): Show valid interval in
12757 when --help is provided.
12758 * opts.h (struct cl_option): Add range_min and range_max fields.
12759 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
12761 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
12763 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
12764 (x * C EQ/NE y * C): New transformation.
12766 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
12768 * genmultilib (combination_space): Accept '+' in option names.
12770 2017-06-28 Martin Liska <mliska@suse.cz>
12773 * asan.c (instrument_derefs): Bail out inner references
12774 that are hard register variables.
12776 2017-06-28 Jakub Jelinek <jakub@redhat.com>
12779 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
12780 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
12782 2017-06-28 Richard Biener <rguenther@suse.de>
12784 * tree-vectorizer.h (vect_get_vec_defs): Remove.
12785 (vect_get_slp_defs): Adjust.
12786 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
12788 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
12790 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
12791 get_initial_defs_for_reduction instead of vect_get_vec_defs.
12792 (vectorizable_reduction): Adjust.
12793 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
12795 (vect_get_slp_defs): Likewise.
12796 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
12797 (vectorizable_bswap): Adjust.
12798 (vectorizable_call): Likewise.
12799 (vectorizable_conversion): Likewise.
12800 (vectorizable_assignment): Likewise.
12801 (vectorizable_shift): Likewise.
12802 (vectorizable_operation): Likewise.
12803 (vectorizable_store): Likewise.
12804 (vectorizable_condition): Likewise.
12805 (vectorizable_comparison): Likewise.
12807 2017-06-28 Michael Collison <michael.collison@arm.com>
12810 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
12812 (arm_gen_movmemqi): Removed unused variable 'i'.
12813 Convert 'for' loop into 'while' loop.
12814 (arm_expand_prologue): Remove last unnecessary set of insn.
12815 (thumb_pop): Remove unused variable 'pushed_words'.
12816 (thumb_exit): Remove last unnecessary set of regs_to_pop.
12818 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12820 * config/s390/predicates.md: Use s390_rel_address_ok_p.
12821 * config/s390/s390-protos.h: Add prototype of
12822 s390_rel_address_ok_p.
12823 * config/s390/s390.c (s390_got_symbol): New function.
12824 (s390_rel_address_ok_p): New function.
12825 (legitimize_pic_address): Use s390_rel_address_ok_p.
12826 (s390_load_got): Use s390_got_symbol.
12827 (s390_option_override): Issue error if
12828 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
12829 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
12831 * config/s390/s390.opt: New option mpic-data-is-text-relative.
12833 2017-06-27 Andrew Pinski <apinski@cavium.com>
12835 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
12836 (X * copysign (1.0, X)): New pattern.
12837 (X * copysign (1.0, -X)): New pattern.
12838 (copysign (-1.0, CST)): New pattern.
12840 2017-06-27 Joseph Myers <joseph@codesourcery.com>
12842 * genmultilib (combination_space): Remove variable.
12843 Validate reuse rules against regular expression for any sequence
12844 of multilib options in any order.
12846 2017-06-27 Michael Collison <michael.collison@arm.com>
12848 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
12849 call aarch64_split_simd_combine.
12850 * (aarch64_combine_internal<mode>): Delete pattern.
12851 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
12852 Allow register and subreg operands.
12854 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
12856 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
12857 specific need, just fallback on defaults.
12858 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
12860 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
12861 Olivier Hainque <hainque@adacore.com>
12863 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
12865 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
12866 targets. Pick a default if no particular attempt applied.
12867 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
12870 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
12872 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
12873 (x86_64-wrs-vxworks7): Likewise.
12875 2017-06-27 Marek Polacek <polacek@redhat.com>
12878 * ubsan.c (instrument_null): Check get_base_address's result for null.
12880 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
12882 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
12884 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
12886 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
12887 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
12888 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
12889 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
12890 New function types.
12891 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
12892 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
12893 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
12894 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
12895 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
12896 BUILT_IN_FEUPDATEENV): New builtins.
12897 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
12898 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
12899 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
12900 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
12902 (builtin_structptr_types): Adjust size.
12903 * tree.c (builtin_structptr_types): Add four entries.
12905 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
12906 Olivier Hainque <hainque@adacore.com>
12908 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
12909 (TLS_SYM): New local macro, forcing reference to __tls__ on
12910 link command lines for VxWorks 7 RTPs, triggering initialization
12912 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
12913 OS features TLS support, true for RTPs on VxWorks 7.
12914 * config/vxworks.c (vxworks_override_options): Setup emutls
12917 2017-06-27 Jakub Jelinek <jakub@redhat.com>
12919 * predict.c (test_prediction_value_range): Use -1U instead of -1
12920 to avoid narrowing conversion warning.
12921 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
12922 to avoid narrowing conversion warning.
12923 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
12925 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
12927 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
12929 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
12930 64bit configurations.
12931 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
12932 (SIZE_TYPE): Likewise.
12933 * config/vxworks.c (vxworks_emutls_var_fields): Use
12934 long_unsigned_type_node instead of unsigned_type_node as the offset
12935 field type, which is "pointer" mode in emutls.c.
12937 2017-06-27 Jakub Jelinek <jakub@redhat.com>
12940 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
12942 PR middle-end/81207
12943 * gimple-fold.c (replace_call_with_call_and_fold): Handle
12944 gimple_vuse copying separately from gimple_vdef copying.
12946 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
12948 * value-prof.c (free_hist): Remove call to memset and the enclosing if
12951 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
12952 Olivier Hainque <hainque@adacore.com>
12954 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
12955 for all vxworks7 targets.
12956 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
12957 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
12958 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
12959 variations for VX6/VX7 and 32/64bits later on in ...
12960 (VXWORKS_LIB_SPEC): Leverage new macros.
12961 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
12962 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
12964 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
12966 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
12967 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
12969 2017-06-26 Carl Love <cel@us.ibm.com>
12971 * config/rs6000/rs6000-c.c: Add support for built-in functions
12972 vector bool char vec_reve (vector bool char);
12973 vector signed char vec_reve (vector signed char);
12974 vector unsigned char vec_reve (vector unsigned char);
12975 vector bool int vec_reve (vector bool int);
12976 vector signed int vec_reve (vector signed int);
12977 vector unsigned int vec_reve (vector unsigned int);
12978 vector bool long long vec_reve (vector bool long long);
12979 vector signed long long vec_reve (vector signed long long);
12980 vector unsigned long long vec_reve (vector unsigned long long);
12981 vector bool short vec_reve (vector bool short);
12982 vector signed short vec_reve (vector signed short);
12983 vector double vec_reve (vector double);
12984 vector float vec_reve (vector float);
12985 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
12986 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
12987 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
12988 (altivec_vreve): New pattern.
12989 * config/rs6000/altivec.h (vec_reve): New define.
12990 * doc/extend.texi (vec_rev): Update the built-in documentation file
12991 for the new built-in functions.
12993 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12995 PR tree-optimization/71815
12996 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
12998 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
13000 (slsr_process_phi): Likewise.
13001 (replace_uncond_cands_and_profitable_phis): Don't replace a
13002 multiply candidate with a stride of 1 (copy or cast).
13003 (phi_incr_cost): Call uses_consumed_by_stmt rather than
13005 (lowest_cost_path): Likewise.
13006 (total_savings): Likewise.
13008 2017-06-26 Richard Biener <rguenther@suse.de>
13011 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
13012 Use def_builtin_pure for all gather builtins.
13014 2017-06-26 Richard Biener <rguenther@suse.de>
13016 PR tree-optimization/81203
13017 * tree-tailcall.c (find_tail_calls): Do not move stmts into
13018 non-dominating BBs.
13020 2017-06-26 Marek Polacek <polacek@redhat.com>
13023 * doc/invoke.texi: Document -Wmultistatement-macros.
13025 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
13027 * doc/sourcebuild.texi (ARM-specific attributes): Document new
13028 arm_neon_ok_no_float_abi effective target.
13030 2017-06-26 Richard Biener <rguenther@suse.de>
13032 PR tree-optimization/80928
13033 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
13034 (copy_bbs): Set BB_DUPLICATED flag early.
13035 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
13037 (execute_on_shrinking_pred): Likewise.
13038 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
13039 BB_DUPLICATED blocks.
13040 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
13041 iterate over all PHIs considering removal of *gsi.
13043 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
13045 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
13048 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
13050 * config/rs6000/rs6000-string.c: (expand_block_clear,
13051 do_load_for_compare, select_block_compare_mode,
13052 compute_current_alignment, expand_block_compare,
13053 expand_strncmp_align_check, expand_strn_compare,
13054 expand_block_move, rs6000_output_load_multiple)
13055 Move functions related to string/block move/compare
13056 to a separate file.
13057 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
13058 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
13059 for this function which is now used in two files.
13060 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
13061 * config.gcc: Add rs6000-string.o to extra_objs for
13062 targets powerpc*-*-* and rs6000*-*-*.
13064 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
13067 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
13068 32-bit, since indexed is not valid for DImode.
13069 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
13070 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
13071 (define_peephole2 for Altivec d-form load): Add 32-bit support.
13072 (define_peephole2 for Altivec d-form store): Likewise.
13075 * multiple_target.c (create_dispatcher_calls): Only create the
13076 dispatcher call if the function is the default clone of a
13077 versioned function.
13079 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
13081 PR middle-end/80902
13082 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
13083 a call, force the call to not be a tail call.
13085 2017-06-23 Jeff Law <law@redhat.com>
13087 * doc/contrib.texi: Add entry for Steven Pemberton's work on
13090 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
13092 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
13093 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
13094 handling for early expansion of vector shifts (sl,sr,sra,rl).
13095 (builtin_function_type): Add vector shift right instructions
13096 to the unsigned argument list.
13098 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
13100 rtl-optimizatoin/79286
13101 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
13102 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
13103 trap. PIC register plus a const unspec without offset can never trap.
13105 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
13107 * tree.h (builtin_structptr_type): New type.
13108 (builtin_structptr_types): Declare new array.
13109 * tree.c (builtin_structptr_types): New array.
13110 (free_lang_data, build_common_tree_nodes): Use it.
13112 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
13115 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
13118 2017-06-22 Matt Turner <mattst88@gmail.com>
13120 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
13121 Lake models to skylake case. Assume skylake for unknown
13122 models with clflushopt.
13124 2017-06-22 Jeff Law <law@redhat.com>
13126 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
13127 frame sizes that do not satisfy aarch64_uimm12_shift.
13129 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
13131 * profile-count.h (apply_probability,
13132 apply_scale, probability_in): Fix checks for zero.
13134 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
13136 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
13137 * doc/cppdiropts.texi (-I @var{dir}): Document it.
13139 2016-06-22 Richard Biener <rguenther@suse.de>
13141 * tree-vect-loop.c (vect_model_reduction_cost): Handle
13142 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
13143 REDUC_MAX_EXPR support.
13144 (vectorizable_reduction): Likewise.
13145 (vect_create_epilog_for_reduction): Likewise.
13147 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
13149 * match.pd (A / (1 << B) -> A >> B): New.
13150 * generic-match-head.c: Include optabs-tree.h.
13151 * gimple-match-head.c: Likewise.
13152 * optabs-tree.h (target_supports_op_p): New.
13153 * optabs-tree.c (target_supports_op_p): New.
13155 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
13157 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
13158 $gcc_cv_ld --help output.
13159 (gcc_cv_ld_demangle): Likewise.
13160 (gcc_cv_ld_eh_frame_hdr): Likewise.
13161 (gcc_cv_ld_pie): Likewise.
13162 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
13163 (gcc_cv_ld_buildid): Likewise.
13164 (gcc_cv_ld_sysroot): Likewise.
13165 (ld_bndplt_support): Likewise.
13166 (ld_pushpopstate_support): Likewise.
13167 * configure: Regenerate.
13168 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
13170 2017-06-21 Jakub Jelinek <jakub@redhat.com>
13173 * config/i386/sse.md (round<mode>2): Renumber match_dup and
13174 operands indexes to avoid gap between operands and match_dups.
13176 2017-06-21 Andrew Pinski <apinski@cavium.com>
13178 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
13179 Increment Arith_shift and Arith_shift_reg by 1.
13180 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
13182 * config/aarch64/aarch64.c (thunderx_tunings): Enable
13183 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
13184 (aarch64_strip_extend): Add new argument and test for it.
13185 (aarch64_cheap_mult_shift_p): New function.
13186 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
13187 add a cost if it is true.
13188 Update calls to aarch64_strip_extend.
13189 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
13191 2017-06-21 Andrew Pinski <apinski@cavium.com>
13193 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
13195 (thunderxt88): Likewise.
13196 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
13197 (thunderx_prefetch_tune): New variable.
13198 (thunderx2t99_prefetch_tune): Update for the correct values.
13199 (thunderxt88_tunings): New variable.
13200 (thunderx_tunings): Use thunderx_prefetch_tune instead of
13201 generic_prefetch_tune.
13202 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
13204 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13206 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
13207 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
13208 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
13209 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
13210 (aarch64_atomic_cas<mode>, GPI): Likewise.
13212 2017-06-21 Martin Liska <mliska@suse.cz>
13214 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
13215 statements on cold and hot labels.
13216 * predict.c (tree_estimate_probability_bb): Remove the
13217 prediction from this place.
13219 2017-06-21 Martin Liska <mliska@suse.cz>
13221 PR tree-optimization/79489
13222 * gimplify.c (maybe_add_early_return_predict_stmt): New
13224 (gimplify_return_expr): Call the function.
13225 * predict.c (tree_estimate_probability_bb): Remove handling
13227 * predict.def: Update comment about early return predictor.
13228 * gimple-predict.h (is_gimple_predict): New function.
13229 * predict.def: Change default value of early return to 66.
13230 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
13232 * passes.def: Put pass_strip_predict_hints to the beginning of
13235 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
13237 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
13238 FUNCTION_DECL declarations.
13239 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
13241 (dwaf2out_decl): Likewise.
13242 * godump.c (go_early_global_decl): Skip call to the real debug hook
13243 for FUNCTION_DECL declarations.
13244 * passes.c (rest_of_decl_compilation): Skip call to the
13245 early_global_decl debug hook for FUNCTION_DECL declarations, unless
13246 -fdump-go-spec is passed.
13248 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
13250 * config/i386/i386.c (struct builtin_isa): New field pure_p.
13251 Reorder for compactness.
13252 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
13253 (def_builtin_pure, def_builtin_pure2): New functions.
13254 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
13256 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
13258 * match.pd (nop_convert): New predicate.
13259 ((A +- CST1) +- CST2): Allow some NOP conversions.
13261 2017-06-21 Jakub Jelinek <jakub@redhat.com>
13264 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
13265 with ctors/dtors if GOVD_SHARED is set.
13267 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
13269 * config/aarch64/aarch64.md (movti_aarch64):
13270 Emit mov rather than orr.
13271 (movtf_aarch64): Likewise.
13272 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
13273 Emit mov rather than orr.
13275 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
13277 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
13278 Swap alternatives, make integer dup more expensive.
13280 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
13282 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
13283 Return true for non-tls symbols.
13285 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
13287 * config/aarch64/aarch64-cores.def (cortex-a55): New.
13288 (cortex-a75): Likewise.
13289 (cortex-a75.cortex-a55): Likewise.
13290 * config/aarch64/aarch64-tune.md: Regenerate.
13291 * doc/invoke.texi (-mtune): Document new values for -mtune.
13293 2017-06-21 Tom de Vries <tom@codesourcery.com>
13295 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
13296 stack_size feature.
13297 (Effective-Target Keywords, Other attributes): Suggest using
13298 dg-add-options stack_size feature to get stack limit in stack_size
13299 effective target documentation.
13301 2017-06-21 Julian Brown <julian@codesourcery.com>
13302 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
13304 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
13305 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
13306 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
13308 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
13309 attribute type list for neon_multiply.
13310 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
13311 attribute type list for neon_multiply.
13312 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
13313 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
13314 attribute type list for neon_multiply.
13315 * config/arm/types.md (crypto_pmull): Add.
13316 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
13317 attribute type list.
13319 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
13321 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
13324 2017-06-20 Jakub Jelinek <jakub@redhat.com>
13326 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
13327 to make sure not to dereference a NULL cost_classes_ptr pointer.
13329 2017-06-20 Carl Love <cel@us.ibm.com>
13331 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13332 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
13333 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
13334 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
13335 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
13336 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
13337 VMULOSW): New enum "unspec" values.
13338 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
13339 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
13340 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
13341 altivec_vmulosw): New patterns.
13342 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
13343 VMULOSW): Add definitions.
13345 2017-06-20 Julia Koval <julia.koval@intel.com>
13347 * config/i386/i386.c: Fix rounding expand for new pattern.
13348 * config/i386/subst.md: Fix pattern (parallel -> unspec).
13350 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
13352 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
13353 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
13355 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
13357 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
13360 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
13362 * config/aarch64/aarch64-cores.def: Rearrange to sort by
13363 architecture, then by implementer ID.
13364 * config/aarch64/aarch64-tune.md: Regenerate.
13366 2017-06-20 Richard Biener <rguenther@suse.de>
13368 PR middle-end/81097
13369 * fold-const.c (split_tree): Fold to type before negating.
13371 2017-06-20 David Malcolm <dmalcolm@redhat.com>
13373 * diagnostic-show-locus.c
13374 (selftest::test_fixit_deletion_affecting_newline): New function.
13375 (selftest::diagnostic_show_locus_c_tests): Call it.
13377 2017-06-20 Andreas Schwab <schwab@suse.de>
13380 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
13383 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
13385 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
13386 __ARM_FEATURE_COPROC according to support.
13388 2017-06-20 Jakub Jelinek <jakub@redhat.com>
13390 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
13391 Rewritten to avoid overflow for > 32-bit pointers.
13394 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
13395 by removing enum keyword.
13396 (ubsan_type_descriptor): Likewise. Formatting fix.
13399 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
13400 splitter): Require TARGET_SSE2 in the condition.
13402 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
13405 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
13406 for doing vector set of SFmode on ISA 3.0.
13407 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
13408 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
13410 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
13411 SFmode value into a V4SF variable that was extracted from another
13412 V4SF variable without converting the element to double precision
13413 and back to single precision vector format.
13414 (vsx_insert_extract_v4sf_p9_2): Likewise.
13416 2017-06-19 Jakub Jelinek <jakub@redhat.com>
13418 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
13419 in UWHI to avoid undefined overflow.
13422 * ubsan.h (enum ubsan_encode_value_phase): New.
13423 (ubsan_encode_value): Change second argument to
13424 enum ubsan_encode_value_phase with default value of
13425 UBSAN_ENCODE_VALUE_GENERIC.
13426 * ubsan.c (ubsan_encode_value): Change second argument to
13427 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
13428 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
13429 create_tmp_var_raw instead of create_tmp_var and use a
13431 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
13432 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
13433 ubsan_encode_value callers.
13436 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
13437 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
13438 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
13440 2017-06-19 Richard Biener <rguenther@suse.de>
13442 PR middle-end/81118
13443 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
13444 estimates if we changed anything.
13446 2017-06-19 Richard Biener <rguenther@suse.de>
13448 PR tree-optimization/80887
13449 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
13450 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
13451 simplified lookups, then reset mprts_hook.
13452 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
13454 (try_to_simplify): Likewise.
13456 2017-06-19 Martin Liska <mliska@suse.cz>
13459 * gimplify.c (gimplify_switch_expr):
13460 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
13462 2017-06-19 Martin Liska <mliska@suse.cz>
13464 * doc/install.texi: Document that PGO runs in 4 stages.
13466 2017-06-19 Martin Liska <mliska@suse.cz>
13469 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
13470 to dispatcher function name.
13471 * multiple_target.c (replace_function_decl): New function.
13472 (create_dispatcher_calls): Redirect both edges and references.
13474 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
13476 * profile-count.c (profile_count::dump): Dump quality.
13477 (profile_count::differs_from_p): Update for unsigned val.
13478 * profile-count.h (profile_count_quality): New enum.
13479 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
13481 2017-06-19 Richard Biener <rguenther@suse.de>
13483 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
13484 struct function as arg.
13485 (estimate_numbers_of_iterations): Export overload with loop arg.
13486 (free_numbers_of_iterations_estimates_loop): Use an overload of
13487 free_numbers_of_iterations_estimates instead.
13488 * tree-cfg.c (remove_bb): Adjust.
13489 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
13490 * tree-parloops.c (gen_parallel_loop): Likewise.
13491 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
13493 (tree_unroll_loops_completely): Likewise.
13494 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
13495 Use an overload instead and export.
13496 (estimated_loop_iterations): Adjust.
13497 (max_loop_iterations): Likewise.
13498 (likely_max_loop_iterations): Likewise.
13499 (estimate_numbers_of_iterations): Take struct function as arg
13501 (loop_exits_before_overflow): Adjust.
13502 (free_numbers_of_iterations_estimates_loop): Use an overload.
13503 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
13504 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
13506 2017-06-19 Richard Biener <rguenther@suse.de>
13509 * ipa-prop.c (find_constructor_constant_at_offset): Handle
13510 RANGE_EXPR conservatively.
13512 2017-06-16 Carl Love <cel@us.ibm.com>
13514 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13515 definitions for vec_float, vec_float2, vec_floato,
13516 vec_floate built-ins.
13517 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
13518 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
13520 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
13521 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
13522 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
13523 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
13524 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
13525 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
13526 vec_floato): Add builtin defines.
13527 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
13528 Update the built-in documentation file for the new built-in
13531 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13533 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
13534 (mthumb): Mark as the negative of -marm.
13536 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13538 * doc/invoke.texi (ARM Options, -mcpu): Document supported
13540 (ARM Options, -mtune): Document that this accepts the same
13541 extension options as -mcpu.
13542 (ARM Options, -mfpu): Document addition of -mfpu=auto.
13544 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13546 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
13547 permitted extensions.
13549 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13551 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
13552 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
13553 (armv8-m.main): Add option +nodsp.
13554 * config/arm/arm-cpu-cdata.h: Regenerated.
13556 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13558 * config/arm/t-fuchsia: New file.
13559 * config.gcc (arm*-*-fuchsia*): Use it.
13561 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13563 * config/arm/t-symbian: Rewrite for new option infrastructure.
13565 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13567 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
13568 (MULTILIB_REQUIRED): Likewise.
13570 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13572 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
13573 (MULTILIB_RESUE): Likewise.
13574 (MULTILIB_MATCHES): Likewise.
13575 (MULTLIB_REQUIRED): Likewise.
13577 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13579 * config/arm/t-rtems: Rewrite for new option framework.
13581 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13583 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
13584 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
13585 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
13586 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
13587 * config/arm/t-multilib: ... here.
13588 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
13589 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
13590 armv7-a and armv8*-a when A-profile libraries have not been built.
13591 * config/arm/t-rmprofile: Rewrite.
13593 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13595 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
13596 with a backslash. Remove the backslash after substituting unescaped
13598 * doc/fragments.texi (MULTILIB_REUSE): Document it.
13600 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13602 * config.gcc: (arm*-*-*): When building a-profile libraries, force
13603 the driver to pass through the default setting of -mfloat-abi.
13604 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
13606 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
13607 (all_feat_combs): New rule.
13608 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
13610 * config/arm/t-aprofile: Rewrite.
13612 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13614 * config/arm/arm.h (FPUTYPE_AUTO): Define.
13615 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
13616 fpu is not specified by the user/command-line.
13617 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
13618 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
13619 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
13620 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
13621 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
13622 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
13624 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13626 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
13627 * config/arm/t-arm-elf: Rewritten.
13629 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13631 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
13632 have some floating-point instructions.
13633 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
13634 (TARGET_MAYBE_HARD_FLOAT): New macro.
13635 * config/arm/arm-builtins.c (arm_init_builtins): Use
13636 TARGET_MAYBE_HARD_FLOAT.
13637 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
13639 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13641 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
13642 (configargs.h): Include it.
13643 (arm_print_hint_for_fpu_option): New function.
13644 (arm_parse_fpu_option): New function.
13645 (candidate_extension): New class.
13646 (arm_canon_for_multilib): New function.
13647 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
13648 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
13649 (ARCH_CANONICAL_SPECS): New macro.
13650 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
13652 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13654 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
13655 are set after handling multilib fragments. Set target_cpu_default2
13658 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13660 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
13662 (arm*-*-*): Set target_cpu_default2 to a quoted string.
13663 * config/arm/parsecpu.awk (check_cpu): Validate any extension
13665 (check_arch): Likewise.
13666 * config/arm/arm.c (arm_configure_build_target): Handle
13667 TARGET_CPU_DEFAULT being a string constant. Scan any feature
13668 options in the default.
13670 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13672 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
13673 when an option is an alias of another.
13674 * config/arm/parsecpu.awk (optalias): New parser token.
13675 (gen_comm_data): Mark non-alias options as such. Emit entries
13676 for extension aliases.
13677 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
13678 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
13679 (armv6kz, armv6zk, armv6t2): Likewise.
13680 (armv7): Make vfpv3-d16 an alias.
13681 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
13683 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
13684 Sort in canonical order.
13685 (armv8-a): Sort in canonical order.
13686 (armv8.1-a, armv8.2-a): Likewise.
13687 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
13689 (cortex-a9): Sort in canonical order.
13690 * config/arm/arm.c (selftests.h): Include it.
13691 (arm_test_cpu_arch_data): New function.
13692 (arm_run_self_tests): New function.
13693 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
13694 (targetm): Move declaration to the end of the file.
13695 * arm-cpu-cdata.h: Regenerated.
13697 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13699 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
13700 call to target_mode_check describing the type of option passed.
13701 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
13702 (arm_target_thumb_only): Use arm_parse_arch_option_name or
13703 arm_parse_cpu_option_name to match parameters against list of
13705 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
13706 arm_arch_core_flags data structure.
13707 * config/arm/arm-cpu_cdata.h: Regenerated.
13709 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13711 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
13713 (arm_print_hint_for_cpu_option): Likewise.
13714 (arm_print_hint_for_arch_option): Likewise.
13715 (arm_parse_cpu_option_name): Likewise.
13716 (arm_parse_arch_option_name): Likewise.
13717 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
13718 of entries in the all_fpus list.
13719 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
13720 (arm_parse_cpu_option_name): Declare.
13721 (arm_parse_arch_option_name): Declare.
13722 (arm_parse_option_features): Declare.
13723 (arm_intialize_isa): Declare.
13724 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
13726 (gen_comm_data): ... here. Make definitions non-static.
13727 * config/arm/arm-cpu-data.h: Regenerated.
13728 * config/arm/arm-cpu-cdata.h: Regenerated.
13730 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13732 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
13733 (cpu_arch_extension): New structure.
13734 (cpu_arch_option, arch_option, cpu_option): New structures.
13735 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
13736 architecture types.
13737 (gen_data): Generate new format data tables.
13738 * config/arm/arm.c (cpu_tune): New structure.
13739 (cpu_option, processors): Delete.
13740 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
13741 (arm_print_hint_for_cpu_option): ... this and ...
13742 (arm_print_hint_for_arch_option): ... this.
13743 (arm_parse_arch_cpu_name): Delete. Replace with ...
13744 (arm_parse_cpu_option_name): ... this and ...
13745 (arm_parse_arch_option_name): ... this.
13746 (arm_unrecognized_feature): Change type of target parameter to
13748 (arm_parse_arch_cpu_features): Delete. Replace with ...
13749 (arm_parse_option_features): ... this.
13750 (arm_configure_build_target): Rework to use new configuration data
13752 (arm_print_tune_info): Rework for new configuration data tables.
13753 * config/arm/arm-cpu-data.h: Regenerated.
13754 * config/arm/arm-cpu.h: Regenerated.
13756 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13758 * Makefile.in (OBJS): Move sbitmap.o from here ...
13759 (OBJS-libcommon): ... to here.
13761 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13763 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
13764 (ISA_ALL_CRYPTO): New macro.
13765 (ISA_ALL_SIMD): New macro
13766 (ISA_ALL_FP): New macro.
13767 * config/arm/arm.c (fpu_bitlist): Update initializer.
13768 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
13770 (arm9e): Add fpu. Add option for nofp
13771 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
13772 (arm926ej-s, arm1026ej-s): Likewise.
13773 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
13774 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
13775 neon-fp16, neon-vfpv4, nofp and nosimd.
13776 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
13777 (cortex-a8): Add fpu. Add option for nofp.
13778 (cortex-a9): Add fpu. Add options for nosimd and nofp.
13779 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
13780 (cortex-r4f): Add fpu.
13781 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
13782 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
13784 (cortex-r8): Likewise.
13785 (cortex-m4): Add fpu. Add option for nofp.
13786 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
13787 (cortex-a17.cortex-a7): Likewise.
13788 (cortex-a32): Add fpu. Add options for crypto and nofp.
13789 (cortex-a35, cortex-a53): Likewise.
13790 (cortex-a57): Add fpu. Add option for crypto.
13791 (cortex-a72, cortex-a73): Likewise.
13792 (exynos-m1): Likewise.
13793 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
13794 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
13795 (cortex-m33): Add fpu. Add option for nofp.
13796 * config/arm/arm-cpu-cdata.h: Regenerated
13797 * config/arm/arm-cpu-data.h: Regenerated.
13799 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13801 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
13802 (armv5te, armv5tej): Likewise.
13803 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
13804 (armv7): Add options fp and vfpv3-d16.
13805 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
13806 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
13808 (armv7ve): Likewise.
13809 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
13810 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
13811 (armv8-a): Add nocrypto option.
13812 (armv8.1-a, armv8.2-a): Likewise.
13813 (armv8-m.main): add options fp, fp.dp and nofp.
13815 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13817 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
13819 (armv8-a+crc): Delete.
13820 (armv8.1-a): Add options simd, crypto and nofp.
13821 (armv8.2-a): Add options fp16, simd, crypto and nofp.
13822 (armv8.2-a+fp16): Delete.
13823 (armv8-m.main): Add option dsp.
13824 (armv8-m.main+dsp): Delete.
13825 (cortex-a8): Add fpu. Add nofp option.
13826 (cortex-a9): Add fpu. Add nofp and nosimd options.
13827 * config/arm/parsecpu.awk (gen_data): Generate option tables and
13828 link to main cpu and architecture data structures.
13829 (gen_comm_data): Only put isa attributes from the main architecture
13831 (option): New statement for architecture and CPU entries.
13832 * arm.c (struct cpu_option): New structure.
13833 (struct processors): Add entry for options.
13834 (arm_unrecognized_feature): New function.
13835 (arm_parse_arch_cpu_name): Ignore any characters after the first
13837 (arm_parse_arch_cpu_feature): New function.
13838 (arm_configure_build_target): Separate out any CPU and architecture
13839 features and parse separately. Don't error out if -mfpu=auto is
13840 used with only an architecture string.
13841 (arm_print_asm_arch_directives): New function.
13842 (arm_file_start): Call it.
13843 * config/arm/arm-cpu-cdata.h: Regenerated.
13844 * config/arm/arm-cpu-data.h: Likewise.
13845 * config/arm/arm-tables.opt: Likewise.
13847 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13849 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
13850 assembler when it is not -mfpu=auto.
13852 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13854 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
13855 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
13856 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
13857 (ASM_CPU_SPEC): Rewrite.
13858 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
13859 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
13860 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
13861 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
13862 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
13863 copied string is NUL-terminated. Also strip any characters prefixed
13865 (arm_rewrite_selected_arch): New function.
13866 (arm_rewrite_march): New function.
13868 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13870 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
13871 (x_arm_cpu_string, x_arm_tune_string): Likewise.
13872 (march, mcpu, mtune): Convert to string-based options.
13873 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
13874 (arm_parse_arch_cpu_name): New function.
13875 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
13876 identify selected architecture or CPU.
13877 (arm_option_save): New function.
13878 (TARGET_OPTION_SAVE): Redefine.
13879 (arm_option_restore): Restore string options.
13880 (arm_option_print): Print string options.
13882 2017-06-16 Martin Sebor <msebor@redhat.com>
13884 PR tree-optimization/80933
13885 PR tree-optimization/80934
13886 * builtins.c (fold_builtin_3): Do not handle bcmp here.
13887 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
13888 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
13889 (gimple_fold_builtin): Call them.
13891 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13893 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
13894 as unlikely; update profile.
13896 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13898 * predict.c (force_edge_cold): Handle declaring edges impossible
13901 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13903 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
13905 (try_unroll_loop_completely): Fix reporting.
13907 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13909 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
13911 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
13914 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
13915 if given a non-constant argument for an intrinsic which requires a
13918 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13920 * profile.c (compare_freqs): New function.
13921 (branch_prob): Sort edge list.
13922 (find_spanning_tree): Assume that the list is priority sorted.
13924 2017-06-16 Richard Biener <rguenther@suse.de>
13926 PR tree-optimization/81090
13927 * passes.def (pass_record_bounds): Remove.
13928 * tree-pass.h (make_pass_record_bounds): Likewise.
13929 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
13930 make_pass_record_bounds): Likewise.
13931 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
13932 not free niter estimates at the beginning but at the end.
13933 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
13935 2017-06-16 Richard Biener <rguenther@suse.de>
13937 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
13938 initializer to workaround ICE in host GCC 4.8.
13940 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13942 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
13944 (clone_inlined_nodes): Update.
13946 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
13948 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
13949 prefetch settings, and enable prefetching by default at -O3.
13951 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
13953 * config/aarch64/aarch64.c (aarch64_override_options_internal):
13954 Set flag_prefetch_loop_arrays according to tuning data.
13956 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
13958 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
13959 New tune structure.
13960 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
13961 [Unrelated to main purpose of the patch] Place the pointer field last
13962 to enable type checking errors when tune structure are wrongly merged.
13963 * config/aarch64/aarch64.c (generic_prefetch_tune,)
13964 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
13965 (thunderx2t99_prefetch_tune): New tune constants.
13966 (tune_params *_tunings): Update all tunings (no functional change).
13967 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
13968 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
13969 from tunings structures.
13971 2017-06-16 Jakub Jelinek <jakub@redhat.com>
13974 * ubsan.c (instrument_null): Add T argument, use it instead
13975 of computing it based on IS_LHS.
13976 (instrument_object_size): Likewise.
13977 (pass_ubsan::execute): Adjust instrument_null and
13978 instrument_object_size callers to pass gimple_get_lhs or
13979 gimple_assign_rhs1 result to it. Use instrument_null instead of
13980 calling get_base_address and instrument_mem_ref. Handle
13981 aggregate call arguments for object-size sanitization.
13983 2017-06-16 Yury Gribov <tetra2005@gmail.com>
13985 PR tree-optimization/81089
13986 * tree-vrp.c (is_masked_range_test): Validate operands of
13989 2017-06-15 Martin Sebor <msebor@redhat.com>
13992 * dumpfile.c (dump_register): Avoid calling memset to initialize
13993 a class with a default ctor.
13994 * gcc.c (struct compiler): Remove const qualification.
13995 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
13996 * hash-table.h: Ditto.
13997 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
13999 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
14000 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
14002 * params.h (struct param_info): Make struct members non-const.
14003 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
14004 with default initialization.
14005 * vec.h (vec_copy_construct, vec_default_construct): New helper
14007 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
14008 with vec_copy_construct.
14009 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
14010 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
14011 * doc/invoke.texi (-Wclass-memaccess): Document.
14013 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
14015 * emit-rtl.h (is_leaf): Update comment about local
14016 register allocator.
14018 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
14021 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
14022 for a variable to have a section before checking if the section has a
14024 Set section to.persistent if persistent attribute is set.
14025 Warn if .persistent attribute is used on an automatic variable.
14027 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
14029 PR rtl-optimization/80474
14030 * reorg.c (update_block): Do not ignore instructions in a delay slot.
14032 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
14034 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
14037 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
14039 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
14040 (casesi): Emit bounds checking as RTL.
14041 (casesi_internal_mips16_<mode>): Remove bounds checking.
14043 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
14045 * config/xtensa/xtensa.c (xtensa_option_override): Append
14046 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
14047 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
14048 xtensa_doloop_hooks): Define unconditionally.
14049 (xtensa_reorg_loops): Only call reorg_loops in the presence of
14051 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
14052 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
14053 for it in xtensa_option_override.
14054 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
14055 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
14057 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
14059 * doc/cppopts.texi: Document '-' special value to -MF.
14061 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
14063 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
14064 (cortex_a53_fconst): Likewise.
14065 (cortex_a53_fpmul): Likewise.
14066 (cortex_a53_f_load_64): Likewise.
14067 (cortex_a53_f_load_many): Likewise.
14068 (cortex_a53_advsimd_alu): Likewise.
14069 (cortex_a53_advsimd_alu_q): Likewise.
14070 (cortex_a53_advsimd_mul): Likewise.
14071 (cortex_a53_advsimd_mul_q): Likewise.
14072 (fpmac bypass): Add new bypass for fpmac-fpmac case.
14073 Add missing fmul, r2f_cvt and fconst cases.
14075 2017-06-14 Richard Biener <rguenther@suse.de>
14077 PR middle-end/81088
14078 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
14080 (fold_binary_loc): When associating do not treat pre-existing
14081 TREE_OVERFLOW on literal constants as a reason to allow
14082 TREE_OVERFLOW on associated literal constants.
14084 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
14086 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
14087 (MASK_FEATURES): New macro.
14088 * config/sparc/sparc.c (sparc_option_override): Remove the special
14089 handling of -mfpu and generalize it to all MASK_FEATURES switches.
14091 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
14093 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
14094 a division of 0 if non-call exceptions are enabled.
14096 2017-06-14 Andrew Pinski <apinski@cavium.com>
14097 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
14100 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
14101 Improve vector initialization code gen for only variable case.
14103 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
14105 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
14107 2017-06-14 Richard Biener <rguenther@suse.de>
14109 PR tree-optimization/81083
14110 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
14113 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14115 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
14116 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
14117 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
14118 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
14119 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
14120 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
14122 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14124 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
14125 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
14127 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14129 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
14131 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14133 * config/rs6000/t-rtems: Don't handle SPE.
14135 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14137 * config/rs6000/t-linux: Don't handle SPE.
14139 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14141 * config/rs6000/eabispe.h: Delete file.
14143 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14145 * config/rs6000/t-spe: Delete file.
14147 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14149 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
14150 (rs6000_legitimate_offset_address_p): Return false for anything in
14151 V2SImode or V2SFmode.
14153 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14155 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
14156 except V2SF and V2SI. Rearrange the vector modes, and add comments.
14157 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
14159 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
14160 (rs6000_legitimate_offset_address_p): Ditto.
14161 (rs6000_emit_move): Ditto.
14162 (rs6000_init_builtins): Remove V4HI_type_node.
14164 2017-06-13 Martin Liska <mliska@suse.cz>
14167 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
14168 (gate_asan): Likewise.
14169 * asan.h (asan_no_sanitize_address_p): Remove the function.
14170 (sanitize_flags_p): New function.
14171 * builtins.def: Fix coding style.
14172 * common.opt: Use renamed enum value.
14173 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
14174 * doc/extend.texi: Document no_sanitize attribute.
14175 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
14176 to SANITIZE_UNDEFINED_NONDEFAULT.
14177 * gcc.c (sanitize_spec_function): Use the renamed enum value.
14178 * gimple-fold.c (optimize_atomic_compare_exchange_p):
14179 Use sanitize_flags_p.
14180 * gimplify.c (gimplify_function_tree): Likewise.
14181 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
14182 * opts.c (parse_no_sanitize_attribute): New function.
14183 (common_handle_option): Use renamed enum value.
14184 * opts.h (parse_no_sanitize_attribute): Declare.
14185 * tree.c (sanitize_flags_p): New function.
14186 * tree.h: Declared here.
14187 * tsan.c: Use sanitize_flags_p.
14188 * ubsan.c (ubsan_expand_null_ifn): Likewise.
14189 (instrument_mem_ref): Likewise.
14190 (instrument_bool_enum_load): Likewise.
14191 (do_ubsan_in_current_function): Remove the function.
14192 (pass_ubsan::execute): Use sanitize_flags_p.
14193 * ubsan.h: Remove do_ubsan_in_current_function
14194 * tree-cfg.c (print_no_sanitize_attr_value): New function.
14195 (dump_function_to_file): Use it here.
14197 2017-06-13 Martin Jambor <mjambor@suse.cz>
14199 PR tree-optimization/80803
14200 PR tree-optimization/81063
14201 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
14202 (propagate_subaccesses_across_link): Enqueue subtree whenever
14203 necessary instead of relying on the caller.
14205 2017-06-13 Martin Jambor <mjambor@suse.cz>
14207 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
14208 that have a first_link.
14209 (sort_and_splice_var_accesses): Do not check first_link before
14211 (subtree_mark_written_and_enqueue): Likewise.
14212 (propagate_all_subaccesses): Likewise and do not stop at first
14213 parent with a first_link.
14215 2017-06-13 Martin Jambor <mjambor@suse.cz>
14217 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
14220 2017-06-13 Yury Gribov <tetra2005@gmail.com>
14222 * match.pd: New pattern.
14224 2017-06-13 Yury Gribov <tetra2005@gmail.com>
14226 * tree-vrp.c (is_masked_range_test): New function.
14227 (register_edge_assert_for): Determine ranges for
14230 2017-06-13 Yury Gribov <tetra2005@gmail.com>
14232 PR tree-optimization/67328
14233 * fold-const.c (maskable_range_p): New function.
14234 (build_range_check): Generate bittests if possible.
14236 2017-06-13 Martin Liska <mliska@suse.cz>
14238 * gimple-pretty-print.c (dump_probability): Add new argument.
14239 (dump_edge_probability): Dump both probability and count.
14240 (dump_gimple_label): Likewise.
14241 (dump_gimple_bb_header): Likewise.
14243 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
14246 * config/avr/avr-devices.c: Fix indentation.
14247 * config/avr/gen-avr-mmcu-specs.c: Dito.
14249 2017-06-13 Richard Biener <rguenther@suse.de>
14251 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
14252 instead get vector type from stmt_info.
14253 (vectorizable_reduction): Adjust. Remove dead code.
14255 2017-06-13 Richard Biener <rguenther@suse.de>
14257 PR middle-end/81065
14258 * fold-const.c (extract_muldiv_1): Remove bogus distribution
14259 case of C * (x * C2 + C3).
14260 (fold_addr_of_array_ref_difference): Properly fold index difference.
14262 2017-06-12 David S. Miller <davem@davemloft.net>
14265 * config/sparc/sparc.md (return expander): Emit frame blockage if
14266 function uses alloca.
14268 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
14270 * combine.c (make_field_assignment): Check len rather than the mode
14271 precision when calling force_to_mode.
14273 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
14275 Support multilibs and devices that see flash in RAM address range.
14278 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
14279 (avr_mcu_t) <flash_pm_offset>: New field.
14280 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
14281 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
14282 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
14283 (AVR_TINY_PM_OFFSET): Remove macro.
14284 * config/avr/avr.opt (-mshort-calls): New option.
14285 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
14286 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
14287 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
14288 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
14289 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
14290 instead of avr_arch->have_jmp_call.
14291 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
14292 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
14293 avr_arch->flash_pm_offset to define.
14294 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
14295 new field flash_pm_offset. Add entry for avrxmega3.
14296 (avr_texinfo): Add entry for avrxmega3.
14297 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
14298 attiny212, attiny214,
14299 attiny412, attiny414, attiny416, attiny417,
14300 attiny814, attiny816, attiny817,
14301 attiny1614, attiny1616, attiny1617,
14302 attiny3214, attiny3216, attiny3217.
14303 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
14304 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
14305 (avr_print_operand_address) [AVR_TINY]: Same.
14306 (avr_asm_init_sections) <readonly_data_section>: Only patch
14307 callback if avr_arch->flash_pm_offset = 0.
14308 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
14309 for rodata if avr_arch->flash_pm_offset != 0.
14310 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
14311 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
14312 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
14313 (m_options): Append opt_rcall.
14314 (m_dirnames): Append dir_rcall.
14315 * config/avr/t-multilib: Regenerate.
14317 * configure.ac [target=avr]: Check whether avrxmega3 default
14318 linker description file works as needed.
14319 * configure: Regenerate.
14320 * doc/avr-mmcu.texi: Regenerate.
14321 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
14322 <__AVR_ARCH__>: Document avrxmega3 and 103.
14323 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
14324 <__AVR_SHORT_CALLS__>: Document it.
14325 <__AVR_PM_BASE_ADDRESS__>: Document it.
14326 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
14327 (AVR Variable Attributes) <progmem>: Document this is
14328 not needed for avrxmega3.
14329 (AVR Named Address Spaces) <__flash>: Dito.
14331 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
14333 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
14335 2017-06-12 Doug Rupp <rupp@adacore.com>
14337 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
14338 Append vxworks-stdint.h to the tm_file list.
14339 * config/vxworks-stdint.h: New file.
14341 2017-06-12 Martin Liska <mliska@suse.cz>
14343 PR tree-optimization/81041
14344 * tree-profile.c (gimple_gen_ic_func_profiler):
14345 Create an extra BB in profile-generate
14346 (gimple_gen_time_profiler): Likewise.
14348 2017-06-12 Jakub Jelinek <jakub@redhat.com>
14350 PR tree-optimization/81003
14351 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
14352 (update_range_test): Use it instead of force_gimple_operand_gsi.
14354 2017-06-12 Richard Biener <rguenther@suse.de>
14356 PR tree-optimization/81053
14357 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
14358 with backedge value not defined in loop. Simplify def stmt
14361 2017-06-11 Tom de Vries <tom@codesourcery.com>
14364 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
14366 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
14367 nvptx_cannot_force_const_mem.
14369 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
14371 * opts.c (finish_options): Move test for flag_split_stack after
14372 it has been initialized.
14374 2017-06-11 Jason Merrill <jason@redhat.com>
14376 * tree.h (id_equal): New.
14377 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
14378 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
14379 instead of strcmp of IDENTIFIER_POINTER.
14381 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
14383 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
14384 (mark_all_inlined_calls_cdtor): Fix formating.
14385 (inline_transform): Rescale profile before inlining.
14387 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
14389 * cgraph.h (cgraph_edge::clone): Update prototype.
14390 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
14391 (cgraph_node::create_clone): Update.
14392 (cgraph_node::create_version_clone): Update.
14393 * tree-inline.c (copy_bb): Update.
14394 (expand_call_inline): Update.
14396 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
14398 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
14399 factored out from ...
14400 (rs6000_emit_prologue): ... here.
14402 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
14404 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
14405 factored out from ...
14406 (rs6000_emit_prologue): ... here.
14408 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
14410 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
14412 (handle_missing_profiles): Fix computation of tp_first_run.
14413 (counts_to_freqs): Do not touch freqs when count is 0.
14415 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
14417 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
14420 2017-06-10 Tom de Vries <tom@codesourcery.com>
14422 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
14423 attributes): Document signal effective target.
14425 2017-06-10 Tom de Vries <tom@codesourcery.com>
14427 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
14428 Document effective target stack_size.
14430 2017-06-09 David Malcolm <dmalcolm@redhat.com>
14432 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
14433 to the edit_context if they can be auto-applied.
14435 2017-06-9 Ian Lance Taylor <iant@golang.org>
14437 * opts.c (finish_options): If -fsplit-stack, disable implicit
14438 -forder-blocks-and-partition.
14439 * doc/invoke.texi (Optimize Options): Document that when using
14440 -fsplit-stack -forder-blocks-and-partition is not implicitly
14443 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
14445 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
14446 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
14447 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
14448 * builtins.def (abort, trap, unreachable): Declare cold.
14449 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
14450 * tree-core.h (ECF_COLD): New.
14451 * tree.c (set_call_expr_flags): Handle ECF_COLD.
14452 (build_common_builtin_nodes): Mark unreachable and abort as cold.
14454 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
14456 * predict.c (unlikely_executed_stmt_p): Cleanup.
14458 2017-06-09 Richard Biener <rguenther@suse.de>
14460 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
14461 model if the ref is always written to.
14463 2017-06-09 Tamar Christina <tamar.christina@arm.com>
14465 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
14467 2017-06-09 Tamar Christina <tamar.christina@arm.com>
14469 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
14472 2017-06-09 Tom de Vries <tom@codesourcery.com>
14475 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
14476 "target cannot support label values" when encountering LABEL_REF.
14478 2017-06-09 Martin Liska <mliska@suse.cz>
14480 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
14481 (gimple_gen_ic_func_profiler): Emit direct comparison
14482 of __gcov_indirect_call_callee with NULL.
14483 (gimple_gen_time_profiler): Change probability from
14484 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
14486 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
14488 * profile.c (edge_gcov_counts): Turn to pointer.
14489 (compute_branch_probabilities, compute_branch_probabilities): Update.
14490 (branch_prob): Do not clear edge_gcov_count.
14491 * profile.h (edge_gcov_counts): Turn to pointer.
14492 (edge_gcov_count): Update.
14494 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
14496 * gimple.h (gimple_check_failed): Mark cold.
14498 2017-06-09 Richard Biener <rguenther@suse.de>
14500 PR tree-optimization/66623
14501 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
14502 refactor check_reduction into two parts, properly computing
14503 whether we have to check reduction validity for outer loop
14506 2017-06-09 Richard Biener <rguenther@suse.de>
14508 PR tree-optimization/79483
14509 * graphite-scop-detection.c (order): New global.
14510 (get_order): Compute bb to order mapping that satisfies code
14511 generation constraints.
14512 (cmp_pbbs): New helper.
14513 (build_scops): Start domwalk at entry block, sort generated
14516 2017-06-09 Richard Biener <rguenther@suse.de>
14518 PR middle-end/81007
14519 * ipa-polymorphic-call.c
14520 (ipa_polymorphic_call_context::restrict_to_inner_class):
14521 Skip FIELD_DECLs with error_mark_node type.
14522 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
14525 2017-06-09 Martin Liska <mliska@suse.cz>
14527 * predict.c (struct branch_predictor): New struct.
14528 (test_prediction_value_range): New test.
14529 (predict_c_tests): New function.
14530 * selftest-run-tests.c (selftest::run_tests): Run the function.
14531 * selftest.h: Declare new tests.
14533 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
14536 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
14537 gen_add3_insn did not fail.
14538 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
14539 r0, construct that number in a temporary reg and add that reg to r0.
14540 If asked to put the result in r0 as well, fail.
14542 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
14544 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14545 for early expansion of vec_eqv.
14547 2017-06-08 Jakub Jelinek <jakub@redhat.com>
14549 PR middle-end/81005
14550 * ubsan.c (instrument_null): Avoid pointless code temporary.
14551 (pass_ubsan::execute): Instrument aggregate arguments of calls.
14553 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
14557 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
14560 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
14563 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
14565 * predict.c (unlikely_executed_edge_p): Move ahead.
14566 (probably_never_executed_edge_p): Use it.
14568 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
14570 PR middle-end/79988
14571 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
14572 gimple_call_builtin_p call.
14574 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
14576 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
14577 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
14578 rtl_check_failed_type2, rtl_check_failed_code1,
14579 rtl_check_failed_code2, rtl_check_failed_code_mode,
14580 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
14581 rtvec_check_failed_bounds, rtl_check_failed_flag,
14582 _fatal_insn_not_found, _fatal_insn): Likewise.
14583 * tree.h (tree_contains_struct_check_failed,
14584 tree_check_failed, tree_not_check_failed,
14585 tree_class_check_failed, tree_range_check_failed,
14586 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
14587 tree_vec_elt_check_failed, phi_node_elt_check_failed,
14588 tree_operand_check_failed, omp_clause_check_failed,
14589 omp_clause_operand_check_failed, omp_clause_range_check_failed):
14592 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
14594 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
14595 flag_branch_probabilities.
14596 * ipa-inline.c (edge_badness): Likewise.
14597 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
14598 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
14599 * predict.c (maybe_hot_frequency_p): Likewise.
14600 (probably_never_executed): Likewise.
14601 * sched-ebb.c (schedule_ebbs): Likewise.
14602 * sched-rgn.c (find_single_block_region): Likewise.
14603 * tracer.c (tail_duplicate): Likewise.
14605 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
14607 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
14608 longer requires x_flag_profile_use.
14610 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
14612 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
14613 instead of flag_reorder_blocks_and_partition.
14614 * dbxout.c (dbxout_function_end): Likewise.
14615 * dwarf2out.c (gen_subprogram_die): Likewise.
14616 * haifa-sched.c (sched_create_recovery_edges): Likewise.
14617 * hw-doloop.c (reorg_loops): Likewise.
14618 * varasm.c (assemble_start_function,
14619 assemble_end_function): Likewise.
14620 (decide_function_section): Do not check for
14621 flag_reorder_blocks_and_partition.
14623 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
14625 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
14627 (chkp_get_hard_register_fake_addr_expr): Ditto.
14628 (chkp_build_addr_expr): Add check for hard reg case.
14629 (chkp_parse_array_and_component_ref): Ditto.
14630 (chkp_find_bounds_1): Ditto.
14631 (chkp_process_stmt): Don't generate bounds store for
14634 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
14636 * predict.c (maybe_hot_bb_p): Do not check profile status.
14637 (maybe_hot_edge_p): Likewise.
14638 (probably_never_executed): Check for zero counts even if profile
14640 (unlikely_executed_edge_p): New function.
14641 (unlikely_executed_stmt_p): New function.
14642 (unlikely_executed_bb_p): New function.
14643 (set_even_probabilities): Use unlikely predicates.
14644 (combine_predictions_for_bb): Likewise.
14645 (predict_paths_for_bb): Likewise.
14646 (predict_paths_leading_to_edge): Likewise.
14647 (determine_unlikely_bbs): New function.
14648 (estimate_bb_frequencies): Use it.
14649 (compute_function_frequency): Use zero counts even if profile is
14651 * profile-count.h: Fix typo.
14653 2017-08-08 Julia Koval <julia.koval@intel.com>
14655 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
14656 _mm512_mask_cvtsepi16_storeu_epi8,
14657 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
14658 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
14659 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
14660 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
14661 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
14662 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
14663 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
14664 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
14665 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
14666 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
14667 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
14668 __builtin_ia32_pmovuswb256mem_mask,
14669 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
14670 __builtin_ia32_pmovwb512mem_mask): New builtins.
14672 2017-08-08 Julia Koval <julia.koval@intel.com>
14674 PR target/73350,80862
14675 * config/i386/subst.md (round): Fix round pattern.
14676 * config/i386/i386.c (ix86_erase_embedded_rounding):
14677 Fix erasing rounding for the fixed pattern.
14679 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
14681 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
14683 2017-06-08 Martin Liska <mliska@suse.cz>
14685 PR gcov-profile/80911
14686 * gcov.c (block_info::block_info): New constructor.
14688 2017-06-07 Carl Love <cel@us.ibm.com>
14690 * config/rs6000/rs6000-c: The return type of the following
14691 built-in functions was implemented as int not long long. Fix sign
14692 of return value for the unsigned version of vec_mulo and vec_mule.
14693 vector unsigned long long vec_bperm (vector unsigned long long,
14694 vector unsigned char)
14695 vector signed long long vec_mule (vector signed int,
14697 vector unsigned long long vec_mule (vector unsigned int,
14698 vector unsigned int)
14699 vector signed long long vec_mulo (vector signed int,
14701 vector unsigned long long vec_mulo (vector unsigned int,
14702 vector unsigned int)
14703 * doc/extend.texi: Fix the documentation for the built-in
14706 2017-06-07 Carl Love <cel@us.ibm.com>
14709 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
14712 2017-06-07 Carl Love <cel@us.ibm.com>
14714 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
14715 support, Generate doublehv for signed int/float for BE case only.
14717 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
14719 * doc/invoke.texi (mcx16): Rewrite.
14721 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14723 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
14724 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
14725 *mov<mode>_softfloat, and an anonymous splitter): Use
14726 nonimmediate_operand instead of rs6000_nonimmediate_operand.
14728 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14730 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
14732 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
14733 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
14734 (rs6000_debug_reg_global): Adjust.
14735 (rs6000_init_hard_regno_mode_ok): Adjust.
14736 (rs6000_dbx_register_number): Adjust.
14737 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
14738 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
14739 Remove SPE_ACC and SPEFSCR.
14740 (REG_ALLOC_ORDER): Ditto.
14741 (FRAME_POINTER_REGNUM): Change to 111.
14742 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
14743 (REG_CLASS_NAMES): Ditto.
14744 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
14745 (REGISTER_NAMES): Ditto.
14746 (ADDITIONAL_REG_NAMES): Ditto.
14747 (rs6000_reg_names): Ditto.
14748 * config/rs6000/rs6000.md: Renumber some register number
14751 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14753 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
14755 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
14756 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
14758 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
14759 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
14760 Delete the SPE high registers.
14761 (REG_ALLOC_ORDER): Ditto.
14762 (enum reg_class): Remove SPE_HIGH_REGS.
14763 (REG_CLASS_NAMES): Ditto.
14764 (REG_CLASS_CONTENTS): Delete the SPE high registers.
14765 (REGISTER_NAMES): Ditto.
14766 (rs6000_reg_names): Ditto.
14767 * doc/tm.texi.in: Remove SPE as example.
14768 * doc/tm.texi: Regenerate.
14770 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14772 * config/rs6000/8540.md (ppc8540_brinc): Delete.
14773 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
14774 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
14775 * config/rs6000/rs6000.md (type): Remove "brinc".
14777 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14779 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
14780 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
14781 * config/rs6000/linuxspe.h: Delete file.
14782 * config/rs6000/rs6000.md: Don't include spe.md.
14783 * config/rs6000/spe.h: Delete file.
14784 * config/rs6000/spe.md: Delete file.
14785 * config/rs6000/t-rs6000: Remove spe.md.
14787 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14789 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
14790 (reg_or_none500mem_operand): Delete.
14791 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
14792 instead of reg_or_none500mem_operand.
14794 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14796 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
14797 handling of SPE flags.
14798 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
14800 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14802 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
14804 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
14805 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
14806 paired_divv2sf3): Similar.
14807 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
14808 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
14809 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
14811 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
14812 Rename the paired_* instruction patterns.
14813 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
14815 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
14816 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
14817 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
14818 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
14819 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
14820 PAIRED_VECTOR_MODE.
14821 (struct machine_function): Delete field spe_insn_chain_scanned_p.
14822 (spe_func_has_64bit_regs_p): Delete.
14823 (spe_expand_predicate_builtin): Delete.
14824 (spe_expand_evsel_builtin): Delete.
14825 (TARGET_DWARF_REGISTER_SPAN): Do not define.
14826 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
14827 (invalid_e500_subreg): Delete.
14828 (rs6000_legitimize_address): Always force_reg op2 as well, for
14829 paired single memory accesses.
14830 (rs6000_member_type_forces_blk): Delete.
14831 (rs6000_spe_function_arg): Delete.
14832 (rs6000_expand_unop_builtin): Delete SPE handling.
14833 (rs6000_expand_binop_builtin): Ditto.
14834 (spe_expand_stv_builtin): Delete.
14835 (bdesc_2arg_spe): Delete.
14836 (spe_expand_builtin): Delete.
14837 (spe_expand_predicate_builtin): Delete.
14838 (spe_expand_evsel_builtin): Delete.
14839 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
14840 (spe_init_builtins): Delete.
14841 (spe_func_has_64bit_regs_p): Delete.
14842 (savres_routine_name): Delete "info" parameter. Adjust callers.
14843 (rs6000_emit_stack_reset): Ditto.
14844 (rs6000_dwarf_register_span): Delete.
14845 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
14846 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
14847 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
14849 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
14851 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
14852 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
14853 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
14854 mulv2sf3, divv2sf3): Delete expanders.
14856 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14858 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
14860 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14862 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
14863 * config/rs6000/rs6000.c: Ditto.
14865 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14867 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
14868 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
14869 comparison_operator.
14871 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14873 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
14874 * config/rs6000/rs6000.opt: Ditto.
14875 * config/rs6000/t-rtems: Ditto.
14877 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14879 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
14880 TARGET_E500_SINGLE by 0, simplify.
14881 * config/rs6000/rs6000.c: Ditto.
14882 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
14883 (spe_build_register_parallel): Delete.
14884 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
14885 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
14886 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
14887 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
14888 (E500_CONVERT): Delete.
14889 * config/rs6000/spe.md: Remove many patterns and all define_constants.
14891 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14893 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
14894 * config/rs6000/dfp.md: Ditto.
14895 (negdd2, *negdd2_fpr): Merge.
14896 (absdd2, *absdd2_fpr): Merge.
14897 (negtd2, *negtd2_fpr): Merge.
14898 (abstd2, *abstd2_fpr): Merge.
14899 * config/rs6000/e500.h: Delete file.
14900 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
14901 TARGET_FPRS by 1 and simplify.
14902 * config/rs6000/rs6000-c.c: Ditto.
14903 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
14904 TARGET_DF_SPE by 0.
14905 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
14907 * config/rs6000/rs6000.md: Ditto.
14908 (floatdidf2, *floatdidf2_fpr): Merge.
14909 (move_from_CR_gt_bit): Delete.
14910 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
14911 (E500_CR_IOR_COMPARE): Delete.
14912 (All patterns that require !TARGET_FPRS): Delete.
14913 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
14915 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14917 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
14919 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14921 * graphds.c (add_edge): Intitialize edge's attached data.
14922 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
14923 pointer parameter. Call pointed function on each edge during
14924 graph traversing. Skip traversing the edge when the function
14926 (graphds_dfs, graphds_scc): Ditto.
14927 (for_each_edge): New parameter. Pass the new parameter to callback
14929 * graphds.h (skip_edge_callback): New function pointer type.
14930 (graphds_dfs, graphds_scc): New function pointer parameter.
14931 (graphds_edge_callback, for_each_edge): New parameter.
14933 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14935 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
14936 out code checking if runtime alias check is possible to below ...
14937 Call the new function.
14938 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
14939 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
14941 2017-06-07 Marek Polacek <polacek@redhat.com>
14944 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
14945 TYPE_OVERFLOW_WRAPS check.
14947 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14949 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
14950 if versioning is required.
14951 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
14952 peeling with the check for versioning.
14954 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14956 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
14957 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
14958 Set true to new parameter if new ssa variable is defined.
14959 (vect_gen_vector_loop_niters): Refactor. Set range information
14960 for the new vector loop bound variable.
14961 (vect_do_peeling): Ditto.
14963 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14965 * tree-affine.c (ssa.h): Include header file.
14966 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
14967 has wrapping overflow behavior.
14969 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14971 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
14973 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14975 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
14976 (tree_to_aff_combination): ... here.
14978 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14980 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
14981 reg_pressure model function.
14982 (ivopts_global_cost_for_size): Delete.
14983 (determine_set_costs, iv_ca_recount_cost): Call new model function
14984 ivopts_estimate_reg_pressure.
14986 2017-06-07 Tamar Christina <tamar.christina@arm.com>
14988 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
14989 expensive than udiv. Remove floating point cases from mod.
14991 2017-06-07 Tamar Christina <tamar.christina@arm.com>
14993 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
14994 Increase idiv cost.
14996 2017-06-07 Tamar Christina <tamar.christina@arm.com>
14998 * config/aarch64/aarch64.md
14999 (copysignsf3): Fix mask generation.
15001 2017-06-07 Jakub Jelinek <jakub@redhat.com>
15003 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
15005 (class dump_manager): Add register_dumps method.
15006 * dumpfile.c: Include langhooks.h.
15007 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
15008 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
15009 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
15010 (dump_manager::dump_register): Start with 512 entries instead of 32.
15011 (dump_manager::register_dumps): New method.
15012 * toplev.c (general_init): Instead of invoking register_dumps
15013 langhook, invoke register_dumps method on the dump manager.
15014 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
15017 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
15019 * doc/md.texi: Clarify the restrictions on a define_insn condition.
15020 Say that # requires an associated define_split to exist, and that
15021 the define_split must be suitable for use after register allocation.
15023 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15025 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
15026 (compute_outgoing_frequencies): Also initialize zero counts.
15027 (find_many_sub_basic_blocks): Do not produce uninitialized profile
15028 around loops; preserve more of profile when nothing changes.
15030 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
15032 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
15034 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
15035 * config/arm/arm-cpu-cdata.h: Regenerate.
15036 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
15037 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
15038 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
15039 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
15040 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
15042 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
15044 * config/arm/t-rmprofile: Likewise.
15045 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
15047 2017-06-06 David S. Miller <davem@davemloft.net>
15050 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
15051 blockage if function uses alloca.
15053 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
15055 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
15056 New "uid" fields to hold pretty-print IDs of group and ref.
15057 Memory references are now identified as <group_id>:<ref_id>
15058 instead of using [random] addresses.
15059 (dump_mem_details): Simplify, no functional change.
15060 (dump_mem_ref): Simplify and make output more concise.
15061 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
15062 (find_or_create_group): Initialize group uid.
15063 (record_ref): Initialize ref uid. Improve debug output.
15064 (prune_group_by_reuse, should_issue_prefetch_p,)
15065 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
15066 (mark_nontemporal_store, determine_loop_nest_reuse):
15067 Improve debug output.
15069 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
15071 * dbgcnt.def (prefetch): New debug counter.
15072 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
15073 (schedule_prefetches): Stop issueing prefetches if debug counter
15076 2017-06-06 Tom de Vries <tom@codesourcery.com>
15078 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
15079 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
15081 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15083 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
15084 Use aarch64_reg_or_zero predicate for operand 4.
15085 (aarch64_compare_and_swap<mode> define_insn_and_split):
15086 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
15087 (aarch64_store_exclusive<mode>): Likewise for operand 2.
15089 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
15091 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
15092 (arm_compute_save_core_reg_mask): This.
15093 (thumb1_compute_save_reg_mask): Rename into ...
15094 (thumb1_compute_save_core_reg_mask): This.
15095 (arm_compute_save_reg0_reg12_mask): Adapt comment.
15096 (arm_compute_frame_layout): Likewise.
15098 2017-06-06 Richard Biener <rguenther@suse.de>
15100 PR tree-optimization/80974
15101 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
15102 keep or clear leaders SSA info.
15104 2017-06-06 Tom de Vries <tom@codesourcery.com>
15106 * config/nvptx/nvptx.c (split_mode_p): New function.
15107 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
15109 2017-06-06 Tom de Vries <tom@codesourcery.com>
15111 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
15113 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15116 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
15119 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15121 * shrink-wrap.c (handle_simple_exit): Update profile.
15122 (try_shrink_wrapping): Upate profile.
15124 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15126 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
15127 (tree_guess_outgoing_edge_probabilities): New.
15128 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
15129 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
15131 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15133 * ipa-split.c (split_function): Initialize return bb profile.
15135 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15137 * profile.c (compute_branch_probabilities): Also initialize
15138 EXIT_BLOCK profile.
15140 2017-06-06 Richard Biener <rguenther@suse.de>
15142 PR tree-optimization/80928
15143 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
15144 (vect_analyze_loop_operations): Properly guard analysis for
15146 (vect_transform_loop): Likewise.
15147 (vect_analyze_loop_2): Also reset SLP type on PHIs.
15148 (vect_model_induction_cost): Do not cost for pure SLP.
15149 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
15150 of induction in inner loop vectorization.
15151 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
15152 (vect_get_and_check_slp_defs): Handle vect_induction_def.
15153 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
15155 (vect_analyze_slp_cost_1): Cost induction.
15156 (vect_detect_hybrid_slp_stmts): Handle PHIs.
15157 (vect_get_slp_vect_defs): Likewise.
15158 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
15159 (vect_transform_stmt): Handle SLP reductions.
15160 * tree-vectorizer.h (vectorizable_induction): Adjust.
15162 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
15164 * config/rs6000/rs6000.c (make_resolver_func): Update
15165 init_lowered_empty_function call.
15167 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
15169 * doc/invoke.texi: Document the -fprofile-abs-path option.
15170 * common.opt (fprofile-abs-path): New option.
15171 * gcov-io.h (gcov_write_filename): Declare.
15172 * gcov-io.c (gcov_write_filename): New function.
15173 * coverage.c (coverage_begin_function): Use gcov_write_filename.
15174 * profile.c (output_location): Likewise.
15176 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
15178 * shring-wrap.c: Revert accidental commit.
15180 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
15182 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
15184 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
15186 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
15188 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
15189 profile in callgraph edge.
15190 * profile-count.h (apply_probability): If THIS is 0, then result is 0
15191 (apply_scale): Likewise.
15192 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
15193 Also scale profile when inlining function with zero profile.
15194 (initialize_cfun): Update exit block profile even when it is zero.
15195 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
15196 when profile is read.
15198 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
15200 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
15201 (CLONE_*): New constants to define the processors we can generate
15202 code for with the target_clone attribute.
15203 (rs6000_clone_map): New array to identify which clone processors
15204 the current program is running on.
15205 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
15206 target_clone attribute.
15207 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
15208 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
15209 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
15210 (cpu_expand_builtin): Add support for target_clone attribute.
15211 (rs6000_valid_attribute_p): Allow "default" attribute.
15212 (get_decl_name): New debug function to simplify printing the
15213 current function name in debugging statements.
15214 (rs6000_clone_priority): New functions to support the target_clone
15215 attribute, and be able to generate code to switch between ISA 2.05
15216 through ISA 3.0 (power6 through power9).
15217 (rs6000_compare_version_priority): Likewise.
15218 (rs6000_get_function_versions_dispatcher): Likewise.
15219 (make_resolver_func): Likewise.
15220 (add_condition_to_bb): Likewise.
15221 (dispatch_function_versions): Likewise.
15222 (rs6000_generate_version_dispatcher_body): Likewise.
15223 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
15224 (fusion_gpr_load_p): Fix a spacing issue.
15225 * doc/extend.texi (Common Function Attributes): Document that the
15226 PowerPC supports the target_clone attribute.
15228 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
15230 * config/arm/arm.h: explain F symbol found in description of ARM
15231 register allocation in its legend.
15233 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
15235 * config/mips/frame-header-opt.c: Include profile-count.h.
15236 * config/riscv/riscv.c: Include profile-count.h
15238 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
15240 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
15242 (sm_set_flag_if_changed): Add bbs field.
15243 (execute_sm_if_changed_flag_set): Pass BBS.
15244 (execute_sm): Update.
15246 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15248 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
15251 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15253 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
15254 (peephole2): New peephole2 to emit the above.
15255 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
15257 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15259 * config/aarch64/aarch64.c (define_peephole2 above
15260 *sub_<shift>_<mode>): New peephole.
15262 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
15264 * config/i386/i386.c (make_resolver_func): Update.
15265 * Makefile.in: Add profile-count.h and profile-count.o
15266 * auto-profile.c (afdo_indirect_call): Update to new API.
15267 (afdo_set_bb_count): Update.
15268 (afdo_propagate_edge): Update.
15269 (afdo_propagate_circuit): Update.
15270 (afdo_calculate_branch_prob): Update.
15271 (afdo_annotate_cfg): Update.
15272 * basic-block.h: Include profile-count.h
15273 (struct edge_def): Turn count to profile_count.
15274 (struct basic_block_def): Likewie.
15275 (REG_BR_PROB_BASE): Move to profile-count.h
15276 (RDIV): Move to profile-count.h
15277 * bb-reorder.c (max_entry_count): Turn to profile_count.
15278 (find_traces): Update.
15279 (rotate_loop):Update.
15280 (connect_traces):Update.
15281 (sanitize_hot_paths):Update.
15282 * bt-load.c (migrate_btr_defs): Update.
15283 * cfg.c (RDIV): Remove.
15284 (init_flow): Use alloc_block.
15285 (alloc_block): Uninitialize count.
15286 (unchecked_make_edge): Uninitialize count.
15287 (check_bb_profile): Update.
15288 (dump_edge_info): Update.
15289 (dump_bb_info): Update.
15290 (update_bb_profile_for_threading): Update.
15291 (scale_bbs_frequencies_int): Update.
15292 (scale_bbs_frequencies_gcov_type): Update.
15293 (scale_bbs_frequencies_profile_count): New.
15294 * cfg.h (update_bb_profile_for_threading): Update.
15295 (scale_bbs_frequencies_profile_count): Declare.
15296 * cfgbuild.c (compute_outgoing_frequencies): Update.
15297 (find_many_sub_basic_blocks): Update.
15298 * cfgcleanup.c (try_forward_edges): Update.
15299 (try_crossjump_to_edge): Update.
15300 * cfgexpand.c (expand_gimple_tailcall): Update.
15301 (construct_exit_block): Update.
15302 * cfghooks.c (verify_flow_info): Update.
15303 (dump_bb_for_graph): Update.
15304 (split_edge): Update.
15305 (make_forwarder_block): Update.
15306 (duplicate_block): Update.
15307 (account_profile_record): Update.
15308 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
15309 (get_estimated_loop_iterations): Update.
15310 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
15311 (single_likely_exit): Update.
15312 * cfgloopmanip.c (scale_loop_profile): Update.
15314 (set_zero_probability): Update.
15315 (lv_adjust_loop_entry_edge): Update.
15316 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
15317 (purge_dead_edges): Update.
15318 (rtl_account_profile_record): Update.
15319 * cgraph.c (cgraph_node::create): Uninitialize count.
15320 (symbol_table::create_edge): Uninitialize count.
15321 (cgraph_update_edges_for_call_stmt_node): Update.
15322 (cgraph_edge::dump_edge_flags): Update.
15323 (cgraph_node::dump): Update.
15324 (cgraph_edge::maybe_hot_p): Update.
15325 * cgraph.h: Include profile-count.h
15326 (create_clone), create_edge, create_indirect_edge): Update.
15327 (cgraph_node): Turn count to profile_count.
15328 (cgraph_edge0: Likewise.
15329 (make_speculative, clone): Update.
15330 (create_edge): Update.
15331 (init_lowered_empty_function): Update.
15332 * cgraphclones.c (cgraph_edge::clone): Update.
15333 (duplicate_thunk_for_node): Update.
15334 (cgraph_node::create_clone): Update.
15335 * cgraphunit.c (cgraph_node::analyze): Update.
15336 (cgraph_node::expand_thunk): Update.
15337 * final.c (dump_basic_block_info): Update.
15338 * gimple-streamer-in.c (input_bb): Update.
15339 * gimple-streamer-out.c (output_bb): Update.
15340 * graphite.c (print_global_statistics): Update.
15341 (print_graphite_scop_statistics): Update.
15342 * hsa-brig.c: Include basic-block.h.
15343 * hsa-dump.c: Include basic-block.h.
15344 * hsa-gen.c (T sum_slice): Update.
15345 (convert_switch_statements):Update.
15346 * hsa-regalloc.c: Include basic-block.h.
15347 * ipa-chkp.c (chkp_produce_thunks): Update.
15348 * ipa-cp.c (struct caller_statistics): Update.
15349 (init_caller_stats): Update.
15350 (gather_caller_stats): Update.
15351 (ipcp_cloning_candidate_p): Update.
15352 (good_cloning_opportunity_p): Update.
15353 (get_info_about_necessary_edges): Update.
15354 (dump_profile_updates): Update.
15355 (update_profiling_info): Update.
15356 (update_specialized_profile): Update.
15357 (perhaps_add_new_callers): Update.
15358 (decide_about_value): Update.
15359 (ipa_cp_c_finalize): Update.
15360 * ipa-devirt.c (struct odr_type_warn_count): Update.
15361 (struct decl_warn_count): Update.
15362 (struct final_warning_record): Update.
15363 (possible_polymorphic_call_targets): Update.
15364 (ipa_devirt): Update.
15365 * ipa-fnsummary.c (redirect_to_unreachable): Update.
15366 * ipa-icf.c (sem_function::merge): Update.
15367 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
15368 * ipa-inline.c (compute_uninlined_call_time): Update.
15369 (compute_inlined_call_time): Update.
15370 (want_inline_small_function_p): Update.
15371 (want_inline_self_recursive_call_p): Update.
15372 (edge_badness): Update.
15373 (lookup_recursive_calls): Update.
15374 (recursive_inlining): Update.
15375 (inline_small_functions): Update.
15376 (dump_overall_stats): Update.
15377 (dump_inline_stats): Update.
15378 * ipa-profile.c (ipa_profile_generate_summary): Update.
15379 (ipa_propagate_frequency): Update.
15380 (ipa_profile): Update.
15381 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
15382 * ipa-utils.c (ipa_merge_profiles): Update.
15383 * loop-doloop.c (doloop_modify): Update.
15384 * loop-unroll.c (report_unroll): Update.
15385 (unroll_loop_runtime_iterations): Update.
15386 * lto-cgraph.c (lto_output_edge): Update.
15387 (lto_output_node): Update.
15388 (input_node): Update.
15389 (input_edge): Update.
15390 (merge_profile_summaries): Update.
15391 * lto-streamer-in.c (input_cfg): Update.
15392 * lto-streamer-out.c (output_cfg): Update.
15393 * mcf.c (create_fixup_graph): Update.
15394 (adjust_cfg_counts): Update.
15395 (sum_edge_counts): Update.
15396 * modulo-sched.c (sms_schedule): Update.
15397 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
15398 * predict.c (maybe_hot_count_p): Update.
15399 (probably_never_executed): Update.
15400 (dump_prediction): Update.
15401 (combine_predictions_for_bb): Update.
15402 (propagate_freq): Update.
15403 (handle_missing_profiles): Update.
15404 (counts_to_freqs): Update.
15405 (rebuild_frequencies): Update.
15406 (force_edge_cold): Update.
15407 * predict.h: Include profile-count.h
15408 (maybe_hot_count_p, counts_to_freqs): UPdate.
15409 * print-rtl-function.c: Do not include cfg.h
15410 * print-rtl.c: Include basic-block.h
15411 * profile-count.c: New file.
15412 * profile-count.h: New file.
15413 * profile.c (is_edge_inconsistent): Update.
15414 (correct_negative_edge_counts): Update.
15415 (is_inconsistent): Update.
15416 (set_bb_counts): Update.
15417 (read_profile_edge_counts): Update.
15418 (compute_frequency_overlap): Update.
15419 (compute_branch_probabilities): Update; Initialize and deinitialize
15421 (branch_prob): Update.
15422 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
15423 (edge_gcov_count): New.
15424 (bb_gcov_count): New.
15425 * shrink-wrap.c (try_shrink_wrapping): Update.
15426 * tracer.c (better_p): Update.
15427 * trans-mem.c (expand_transaction): Update.
15428 (ipa_tm_insert_irr_call): Update.
15429 (ipa_tm_insert_gettmclone_call): Update.
15430 * tree-call-cdce.c: Update.
15431 * tree-cfg.c (gimple_duplicate_sese_region): Update.
15432 (gimple_duplicate_sese_tail): Update.
15433 (gimple_account_profile_record): Update.
15434 (execute_fixup_cfg): Update.
15435 * tree-inline.c (copy_bb): Update.
15436 (copy_edges_for_bb): Update.
15437 (initialize_cfun): Update.
15438 (freqs_to_counts): Update.
15439 (copy_cfg_body): Update.
15440 (expand_call_inline): Update.
15441 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
15442 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
15443 (try_unroll_loop_completely): Update.
15444 (try_peel_loop): Update.
15445 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
15446 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
15447 * tree-ssa-loop-split.c (connect_loops): Update.
15448 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
15449 * tree-ssa-reassoc.c (branch_fixup): Update.
15450 * tree-ssa-tail-merge.c (replace_block_by): Update.
15451 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
15452 (compute_path_counts): Update.
15453 (update_profile): Update.
15454 (recompute_probabilities): Update.
15455 (update_joiner_offpath_counts): Update.
15456 (estimated_freqs_path): Update.
15457 (freqs_to_counts_path): Update.
15458 (clear_counts_path): Update.
15459 (ssa_fix_duplicate_block_edges): Update.
15460 (duplicate_thread_path): Update.
15461 * tree-switch-conversion.c (case_bit_test_cmp): Update.
15462 (struct switch_conv_info): Update.
15463 * tree-tailcall.c (decrease_profile): Update.
15464 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
15465 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
15466 * value-prof.c (check_counter): Update.
15467 (gimple_divmod_fixed_value): Update.
15468 (gimple_mod_pow2): Update.
15469 (gimple_mod_subtract): Update.
15470 (gimple_ic_transform): Update.
15471 (gimple_stringop_fixed_value): Update.
15472 * value-prof.h (gimple_ic): Update.
15474 2017-06-02 Carl Love <cel@us.ibm.com>
15476 * config/rs6000/rs6000-c: Add support for built-in functions
15477 vector double vec_doublee (vector signed int);
15478 vector double vec_doublee (vector unsigned int);
15479 vector double vec_doublee (vector float);
15480 vector double vec_doubleh (vector signed int);
15481 vector double vec_doubleh (vector unsigned int);
15482 vector double vec_doubleh (vector float);
15483 vector double vec_doublel (vector signed int);
15484 vector double vec_doublel (vector unsigned int);
15485 vector double vec_doublel (vector float);
15486 vector double vec_doubleo (vector signed int);
15487 vector double vec_doubleo (vector unsigned int);
15488 vector double vec_doubleo (vector float);.
15489 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
15490 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
15492 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
15493 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
15494 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
15496 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
15497 vec_doublel, vec_doubleh.
15498 * doc/extend.texi: Update the built-in documentation file for the
15499 new built-in functions.
15501 2017-06-02 David Malcolm <dmalcolm@redhat.com>
15504 * ipa-inline-analysis.c (free_growth_caches): Set
15505 edge_removal_hook_holder to NULL after removing it.
15507 2017-06-02 Sudakshina Das <sudi.das@arm.com>
15509 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
15510 comparision with zero.
15512 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
15513 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
15514 for early expansion of vec_min and vec_max builtins.
15515 (builtin_function_type): Add min/max unsigned variants to those
15516 identified as having unsigned arguments.
15518 2017-06-02 Olivier Hainque <hainque@adacore.com>
15520 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
15522 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15524 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
15525 Use VALL_F16 iterator rather than VALL.
15527 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15529 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
15530 Emit CBNZ inside loop when doing a strong exchange and comparing
15531 against zero. Generate the CC flags after the loop.
15533 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
15535 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
15536 (dl_section_ref): New.
15537 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
15538 On AIX, append an expression to subtract the size of the
15539 section length to dl_section_ref.
15541 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
15543 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
15544 for early expansion of vector absolute builtins.
15546 2017-06-02 Richard Biener <rguenther@suse.de>
15548 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
15549 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
15551 2017-06-02 Richard Biener <rguenther@suse.de>
15553 PR tree-optimization/80948
15554 * tree-tailcall.c (find_tail_calls): Track stmts to move in
15555 stmt order as well.
15557 2017-06-02 Richard Biener <rguenther@suse.de>
15559 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
15561 * tree-vect-stmts.c (process_use): Do not mark backedge defs
15562 for inductions as relevant.
15564 2017-06-02 Richard Biener <rguenther@suse.de>
15566 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
15567 (vectorizable_induction): ... this. Remove dead code.
15569 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
15571 * builtins. (expand_builtin_alloca): Remove second parameter and
15572 infer its value from the first parameter instead.
15573 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
15575 2017-06-02 Jakub Jelinek <jakub@redhat.com>
15577 PR rtl-optimization/80903
15578 * loop-doloop.c (add_test): Unshare sequence.
15580 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
15582 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
15584 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
15586 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
15588 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
15589 xlogue_layout::get_instance, logue_layout::xlogue_layout,
15590 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
15591 (xlogue_layout::get_stub_rtx): Make static.
15592 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
15593 (xlogue_layout::compute_stub_managed_regs): Rename to...
15594 (xlogue_layout::count_stub_managed_regs): ...this.
15595 (xlogue_layout::is_stub_managed_reg): New function.
15596 (xlogue_layout::m_stub_names): Rename to...
15597 (xlogue_layout::s_stub_names): ...this, make static.
15598 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
15599 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
15600 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
15601 xlogue_layout::s_stub_names): Instantiate statics.
15602 (stub_managed_regs): Remove.
15603 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
15604 (disable_call_ms2sysv_xlogues): Rename to...
15605 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
15606 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
15608 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
15609 change after reload_completed.
15610 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
15612 (ix86_expand_prologue): Likewise.
15613 (ix86_expand_epilogue): Likewise.
15614 (ix86_expand_split_stack_prologue): Likewise.
15615 (ix86_compute_frame_layout): Remove frame parameter ...
15616 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
15617 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
15619 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
15620 (ix86_frame): Move from here ...
15621 * config/i386/i386.h (ix86_frame): ... to here.
15622 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
15623 complete ix86_frame data structure instead. Remove some_ld_name.
15625 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
15627 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
15628 symbols that hold a DECL_VALUE_EXPR.
15630 2017-06-01 Martin Jambor <mjambor@suse.cz>
15632 PR tree-optimization/80898
15633 * tree-sra.c (process_subtree_disqualification): Removed.
15634 (disqualify_candidate): Do not acll
15635 process_subtree_disqualification.
15636 (subtree_mark_written_and_enqueue): New function.
15637 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
15638 RHS has been disqualified and re-queue LHS if necessary. Apart
15639 from that, ignore disqualified RHS.
15641 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15643 * config/s390/s390.c (s390_emit_epilogue): Disable early return
15644 address fetch for z10 or later.
15646 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15648 * config/arc/arc.md (tst_movb): Add guard when splitting.
15650 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15652 * config/arc/arc.c (arc_can_eliminate): Test against
15653 arc_frame_pointer_needed.
15655 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15657 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
15658 to prevent store reordering.
15659 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
15660 (type): Add block type.
15661 (stack_tie): Define special instruction to be used in
15664 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15666 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
15667 constraint. It is not valid for the pattern.
15668 (noncommutative_binary_comparison): Likewise.
15670 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15672 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
15675 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15677 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
15678 be used by the reg-alloc.
15680 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15682 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
15683 reg-alloc when having mul64 or mul32x16 instructions.
15684 (mulsidi3): Likewise.
15685 (umulsidi3): Likewise.
15686 (mulsi32x16): New pattern.
15687 (mulsi64): Likewise.
15688 (mulsidi64): Likewise.
15689 (umulsidi64): Likewise.
15690 (MUL32x16_REG): Define.
15691 (mul64_600): Use MUL32x16_REG.
15692 (mac64_600): Likewise.
15693 (umul64_600): Likewise.
15694 (umac64_600): Likewise.
15696 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15698 * config/arc/arc.md (mulsi3_700): Make it commutative.
15700 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
15702 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
15704 (*sign_extendsidi2_insn): Likewise for movstosw.
15706 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
15708 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
15709 the type of the input discriminant value. Convert the
15710 discriminant value of signedness vary.
15712 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
15714 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
15715 Add to -Wall section.
15717 2017-06-01 Richard Biener <rguenther@suse.de>
15719 PR middle-end/66313
15720 * fold-const.c (fold_plusminus_mult_expr): If the factored
15721 factor may be zero use a wrapping type for the inner operation.
15722 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
15723 and handle moved defs.
15724 (process_assignment): Properly guard the unary op case. Return a
15725 tri-state indicating that moving the stmt before the call may allow
15726 to continue. Pass through to_move.
15727 (find_tail_calls): Handle moving unrelated defs before
15730 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
15733 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
15734 splitter result in the canonical way.
15736 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
15738 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
15739 also for 32bit target. Update insn attributes.
15740 (zero-extendsidi2 splitter): Allow all registers for operand 1.
15742 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
15744 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
15745 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
15746 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
15747 (_mm_maskz_min_ss): New intrinsics.
15749 2017-05-31 Martin Liska <mliska@suse.cz>
15751 * tree-vect-loop.c (vect_create_epilog_for_reduction):
15752 Change comment style to one we normally use.
15753 (vectorizable_reduction): Likewise.
15754 (vectorizable_induction): Likewise.
15755 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
15756 (vectorizable_call): Likewise.
15757 (vectorizable_simd_clone_call): Likewise.
15758 (vectorizable_conversion): Likewise.
15759 (vectorizable_assignment): Likewise.
15760 (vectorizable_shift): Likewise.
15761 (vectorizable_operation): Likewise.
15762 (vectorizable_store): Likewise.
15763 (vectorizable_load): Likewise.
15764 * tree-vectorizer.h: Likewise.
15766 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
15768 * passes.c (emergency_dump_function): New.
15769 * tree-pass.h (emergency_dump_function): Declare.
15770 * plugin.c (plugins_internal_error_function): Remove.
15771 * plugin.h (plugins_internal_error_function): Remove declaration.
15772 * toplev.c (internal_error_function): New static function. Use it...
15773 (general_init): ...here.
15775 2017-05-31 Graham Markall <graham.markall@embecosm.com>
15777 * config/arc/arc.c (arc_print_operand): Handle constant operands.
15778 (arc_rtx_costs): Add costs for new patterns.
15779 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
15780 * config/arc/predicates.md: Add _1_2_3_operand predicate.
15782 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
15784 * tree-ssa-strlen.c (get_next_strinfo): New function.
15785 (get_stridx_plus_constant): Use it.
15786 (zero_length_string): Likewise.
15787 (adjust_related_strinfos): Likewise.
15788 (adjust_last_stmt): Likewise.
15790 2017-05-31 Richard Biener <rguenther@suse.de>
15793 * config/i386/i386.c (ix86_expand_builtin): Remove assert
15794 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
15796 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
15798 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
15799 loop_vinfo argument and use of dependence distance vectors.
15800 Check instead whether the two references differ only in their
15801 initial value and assume that they have the same alignment if the
15802 difference is a multiple of the vector alignment.
15803 (vect_analyze_data_refs_alignment): Update call accordingly.
15805 2017-05-31 Martin Liska <mliska@suse.cz>
15808 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
15810 2017-05-31 Bin Cheng <bin.cheng@arm.com>
15812 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
15813 (create_intersect_range_checks): Move from ...
15814 * tree-data-ref.c (create_intersect_range_checks_index)
15815 (create_intersect_range_checks): ... to here.
15816 (create_runtime_alias_checks): New function factored from ...
15817 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
15818 here. Call above function.
15819 * tree-data-ref.h (create_runtime_alias_checks): New function.
15821 2017-05-31 Bin Cheng <bin.cheng@arm.com>
15823 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
15824 segment length for dr_b and compute it in wide_int.
15826 2017-05-31 Richard Biener <rguenther@suse.de>
15828 PR tree-optimization/80906
15829 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
15830 and pass through iv_map.
15831 (copy_bb_and_scalar_dependences): Adjust.
15832 (translate_pending_phi_nodes): Likewise.
15833 (copy_loop_close_phi_args): Handle code-generating IVs instead
15836 2017-05-30 David Malcolm <dmalcolm@redhat.com>
15838 * diagnostic-color.c (color_dict): Add "type-diff".
15839 (parse_gcc_colors): Update comment.
15840 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
15841 -fdiagnostics-show-template-tree and -fno-elide-type.
15842 (GCC_COLORS): Add type-diff to example.
15844 (-fdiagnostics-show-template-tree): New.
15845 (-fno-elide-type): New.
15846 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
15847 the pp_format_decoder callback. Call any m_format_postprocessor's
15849 (pretty_printer::pretty_printer): Initialize
15850 m_format_postprocessor.
15851 (pretty_printer::~pretty_printer): Delete any
15852 m_format_postprocessor.
15853 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
15854 (class format_postprocessor): New class.
15855 (struct pretty_printer::format_decoder): Document the new parameters.
15856 (struct pretty_printer::m_format_postprocessor): New field.
15857 * tree-diagnostic.c (default_tree_printer): Update for new
15858 bool and const char ** params.
15859 * tree-diagnostic.h (default_tree_printer): Likewise.
15861 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
15863 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
15864 (lwa_operand): Delete rs6000_gen_cell_microcode test.
15865 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
15866 rs6000_gen_cell_microcode code.
15867 (rs6000_final_prescan_insn): Delete.
15868 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
15869 "warn-cell-microcode" entries.
15870 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
15871 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
15872 throughout. Change cc_reg_not_micro_cr0_operand to
15873 cc_reg_not_cr0_operand throughout.
15874 (*extendhi<mode>2_noload): Delete.
15875 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
15876 (mwarn-cell-microcode): Delete.
15877 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
15878 -mgen-cell-microcode and -mwarn-cell-microcode.
15880 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
15883 * config/i386/constraints.md (Yd): New constraint.
15885 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
15886 and (?Yd, r) alternatives. Update insn attributes.
15887 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
15888 and (?*Yd, r) alternatives. Update insn attributes.
15889 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
15891 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
15893 * gimplify.c (gimplify_modify_expr): Don't create a
15894 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
15897 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
15899 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
15901 2017-05-30 Richard Biener <rguenther@suse.de>
15903 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
15904 and reduc_def fields.
15905 (STMT_VINFO_REDUC_TYPE): New define.
15906 (STMT_VINFO_REDUC_DEF): Likewise.
15907 (vect_force_simple_reduction): Adjust prototype.
15908 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
15909 (vect_is_simple_reduction): Remove check_reduction argument.
15910 (vect_force_simple_reduction): Adjust and set
15911 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
15912 (vectorizable_reduction): Do not re-do reduction analysis
15913 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
15914 * tree-parloops.c (gather_scalar_reductions): Adjust.
15916 2017-05-30 Richard Biener <rguenther@suse.de>
15918 PR middle-end/80901
15919 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
15922 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
15924 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
15925 Introduce unknown_misalignment parameter and remove vf.
15926 (vect_peeling_hash_get_lowest_cost):
15927 Pass unknown_misalignment parameter.
15928 (vect_enhance_data_refs_alignment):
15929 Fix unsupportable data ref treatment.
15931 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
15933 * tree-vect-data-refs.c (vect_get_data_access_cost):
15934 Workaround for SLP handling.
15935 (vect_enhance_data_refs_alignment):
15936 Compute costs for doing no peeling at all, compare to the best
15937 peeling costs so far and avoid peeling if cheaper.
15939 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
15941 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
15942 Return peeling info and set costs to zero for unlimited cost
15944 (vect_enhance_data_refs_alignment): Also inspect all datarefs
15945 with unknown misalignment. Compute and costs for unknown
15946 misalignment, compare them to the costs for known misalignment
15947 and choose the cheapest for peeling.
15949 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
15951 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
15952 (vect_get_peeling_costs_all_drs): Create function.
15953 (vect_peeling_hash_get_lowest_cost):
15954 Use vect_get_peeling_costs_all_drs.
15955 (vect_peeling_supportable): Create function.
15956 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
15958 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
15960 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
15961 DR_HAS_NEGATIVE_STEP.
15962 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
15963 (vect_enhance_data_refs_alignment): Use.
15964 (vect_duplicate_ssa_name_ptr_info): Use.
15965 * tree-vectorizer.h (dr_misalignment): Use.
15966 (known_alignment_for_access_p): Use.
15968 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
15971 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
15973 (has_section_name): New function.
15975 2017-05-30 Martin Liska <mliska@suse.cz>
15978 * auto-profile.c (get_function_decl_from_block): Fix
15981 2017-05-30 Richard Biener <rguenther@suse.de>
15983 PR middle-end/80876
15984 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
15986 2017-05-30 Martin Liska <mliska@suse.cz>
15988 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
15989 * dumpfile.h (struct dump_file_info): Remove ctors.
15991 2017-05-30 Martin Liska <mliska@suse.cz>
15993 * predict.def: Fix GNU coding style.
15995 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
15997 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
15998 Mark 'to' argument with ATTRIBUTE_UNUSED.
16000 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
16002 * config/xtensa/xtensa.c (xtensa_emit_call): Use
16003 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
16004 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
16007 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
16009 * doc/install.texi (Options specification): Restore entry of
16010 --enable-sjlj-exceptions.
16012 2017-05-27 Michael Eager <eager@eagercon.com>
16015 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
16017 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
16019 * config/microblaze/microblaze.h
16020 (FIXED_REGISTERS): Update in macro.
16021 (CALL_USED_REGISTERS): Update in macro.
16023 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
16025 * doc/install.texi: Add links to macOS binary distributions.
16027 2017-05-27 Jakub Jelinek <jakub@redhat.com>
16031 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
16033 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
16035 2017-05-26 Martin Liska <mliska@suse.cz>
16037 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
16039 2017-05-26 Martin Liska <mliska@suse.cz>
16041 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
16042 always leading ';; '.
16043 (dump_bb_info): Likewise.
16044 (brief_dump_cfg): Likewise.
16045 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
16046 * dumpfile.c: Remove usage of TDF_VERBOSE.
16047 * dumpfile.h (enum dump_kind): Likewise.
16048 (dump_gimple_bb_header): Do not use TDF_COMMENT.
16049 * print-tree.c (debug_verbose): Remove.
16050 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
16051 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
16052 * tree-diagnostic.c (default_tree_printer): Replace
16053 TDF_DIAGNOSTIC with TDF_SLIM.
16055 2017-05-26 Bin Cheng <bin.cheng@arm.com>
16057 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
16058 in parameter loop, rather than loop_vinfo.
16059 (create_intersect_range_checks): Ditto.
16060 (vect_create_cond_for_alias_checks): Update call to above functions.
16062 2017-05-26 Bin Cheng <bin.cheng@arm.com>
16064 PR tree-optimization/80815
16065 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
16066 for merging runtime alias checks. Handle negative DR_STEPs.
16068 2017-05-26 Bin Cheng <bin.cheng@arm.com>
16070 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
16072 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
16073 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
16074 out code pruning runtime alias checks.
16075 * tree-data-ref.c (prune_runtime_alias_test_list): New function
16076 factored out from above.
16077 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
16079 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
16081 (prune_runtime_alias_test_list): New decalaration.
16083 2017-05-26 Bin Cheng <bin.cheng@arm.com>
16085 * tree-vect-data-refs.c (compare_tree): Rename and move ...
16086 * tree-data-ref.c (data_ref_compare_tree): ... to here.
16087 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
16088 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
16089 (operator==, comp_dr_with_seg_len_pair): Ditto.
16090 (vect_prune_runtime_alias_test_list): Ditto.
16092 2017-05-26 Martin Liska <mliska@suse.cz>
16095 * params.def: Bound partial-inlining-entry-probability param.
16097 2017-05-26 Marek Polacek <polacek@redhat.com>
16100 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
16103 2017-05-26 Richard Biener <rguenther@suse.de>
16105 PR tree-optimization/80842
16106 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
16109 2017-05-26 Richard Biener <rguenther@suse.de>
16111 PR tree-optimization/80844
16112 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
16114 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
16116 * doc/md.texi (Machine Constraints): Update x86 family
16117 machine constraints section to match 'config/i386/constraints.md'.
16119 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
16121 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
16123 2017-05-25 Nathan Sidwell <nathan@acm.org>
16125 * doc/invoke.texi (--enable-languages): Update documentation.
16127 2017-05-25 Martin Liska <mliska@suse.cz>
16129 * dumpfile.c: Add TDF_FOLDING.
16130 * dumpfile.h (enum dump_kind): Likewise.
16131 * genmatch.c (dt_simplify::gen_1): Use it.
16133 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
16135 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
16137 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
16139 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
16140 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
16142 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
16144 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
16145 * match.pd (X == C): Rewrite it here.
16146 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
16147 with_certain_nonzero_bits2): New predicates.
16148 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
16150 2017-05-24 Nathan Sidwell <nathan@acm.org>
16152 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
16155 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
16158 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
16160 * config/powerpcspe: New port. Files are copied from the rs6000
16161 port, with "rs6000" in filenames replaced by "powerpcspe".
16163 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
16165 PR rtl-optimization/80754
16166 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
16168 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
16170 * config/sparc/sparc.md (length): Return the correct value for -mflat
16171 sibcalls to match output_sibcall.
16173 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
16177 * config/rs6000/rs6000.c (struct machine_function): Add new field
16179 (rs6000_get_separate_components): Init that field, use it.
16180 (rs6000_components_for_bb): Use the field.
16182 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
16184 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
16186 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
16188 PR middle-end/80823
16189 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
16191 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16194 * config/s390/s390.c (s390_check_qrst_address): Check incoming
16195 address against address_operand predicate.
16196 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
16198 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
16200 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
16201 parameters passed indirectly.
16203 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
16205 * config/i386/i386.md (*movdi_internal): Remove SSE4
16206 alternative 18 (?r, *v). Update insn attributes.
16207 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
16208 Update insn attributes.
16209 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
16210 Update insn attributes.
16211 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
16212 alternative 1 (r, v). Remove isa attribute.
16213 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
16214 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
16215 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
16217 2017-05-23 Tom de Vries <tom@codesourcery.com>
16219 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
16222 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
16224 * cgraphunit.c (symbol_table::process_new_functions): Update.
16225 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
16226 (inline_generate_summary): Rename to ...
16227 (ipa_fn_summary_generate): ... this one.
16228 (inline_read_summary): Rename to ...
16229 (ipa_fn_summary_read): ... this one.
16230 (inline_write_summary): Rename to ...
16231 (ipa_fn_summary_write): ... this one.
16232 (inline_free_summary): Rename to ...
16233 (ipa_free_fn_summary): ... this one.
16234 (pass_data_local_fn_summary, pass_local_fn_summary,
16235 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
16236 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
16237 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
16238 make_pass_ipa_fn_summary): New.
16239 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
16240 inline_write_summary, inline_free_summary): Remove.
16241 (ipa_free_fn_summary) : New.
16242 * ipa-inline.c (ipa_inline): Update.
16243 (pass_ipa_inline): Do not generate summaries.
16244 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
16246 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
16247 and add pass_ipa_fn_summary.
16248 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
16250 (make_pass_inline_parameters): Remove.
16252 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
16254 * omp-low.c (struct omp_context): Remove "default_kind" member.
16257 * omp-offload.c (execute_oacc_device_lower): Remove the
16258 parallelism dimensions function attributes for unparallelized
16259 OpenACC kernels constructs.
16261 2017-05-23 Martin Liska <mliska@suse.cz>
16263 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
16265 (cgraph_edge::make_speculative): Likewise.
16266 (cgraph_edge::resolve_speculation): Likewise.
16267 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
16268 (cgraph_node::dump): Likewise.
16269 * cgraph.h: Likewise.
16270 * cgraphunit.c (analyze_functions): Likewise.
16271 (symbol_table::compile): Likewise.
16272 * ipa-cp.c (print_all_lattices): Likewise.
16273 (determine_versionability): Likewise.
16274 (initialize_node_lattices): Likewise.
16275 (ipcp_verify_propagated_values): Likewise.
16276 (estimate_local_effects): Likewise.
16277 (update_profiling_info): Likewise.
16278 (create_specialized_node): Likewise.
16279 (perhaps_add_new_callers): Likewise.
16280 (decide_about_value): Likewise.
16281 (decide_whether_version_node): Likewise.
16282 (identify_dead_nodes): Likewise.
16283 (ipcp_store_bits_results): Likewise.
16284 * ipa-devirt.c (dump_targets): Likewise.
16285 (ipa_devirt): Likewise.
16286 * ipa-icf.c (sem_item::dump): Likewise.
16287 (sem_function::equals): Likewise.
16288 (sem_variable::equals): Likewise.
16289 (sem_item_optimizer::read_section): Likewise.
16290 (sem_item_optimizer::execute): Likewise.
16291 (congruence_class::dump): Likewise.
16292 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
16293 (dump_inline_summary): Likewise.
16294 (estimate_node_size_and_time): Likewise.
16295 (inline_analyze_function): Likewise.
16296 * ipa-inline-transform.c (inline_call): Likewise.
16297 * ipa-inline.c (report_inline_failed_reason): Likewise.
16298 (want_early_inline_function_p): Likewise.
16299 (edge_badness): Likewise.
16300 (update_edge_key): Likewise.
16301 (inline_small_functions): Likewise.
16302 * ipa-profile.c (ipa_profile): Likewise.
16303 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
16304 (ipa_make_edge_direct_to_target): Likewise.
16305 (remove_described_reference): Likewise.
16306 (ipa_impossible_devirt_target): Likewise.
16307 (propagate_controlled_uses): Likewise.
16308 (ipa_print_node_params): Likewise.
16309 (ipcp_transform_function): Likewise.
16310 * ipa-pure-const.c (pure_const_read_summary): Likewise.
16311 (propagate_pure_const): Likewise.
16312 * ipa-reference.c (generate_summary): Likewise.
16313 (read_write_all_from_decl): Likewise.
16314 (propagate): Likewise.
16315 (ipa_reference_read_optimization_summary): Likewise.
16316 * ipa-utils.c (ipa_merge_profiles): Likewise.
16317 * ipa.c (walk_polymorphic_call_targets): Likewise.
16318 (symbol_table::remove_unreachable_nodes): Likewise.
16319 (ipa_single_use): Likewise.
16320 * passes.c (execute_todo): Likewise.
16321 * predict.c (drop_profile): Likewise.
16322 * symtab.c (symtab_node::get_dump_name): New function.
16323 (symtab_node::dump_name): Likewise.
16324 (symtab_node::dump_asm_name): Likewise.
16325 (symtab_node::dump_references): Likewise.
16326 (symtab_node::dump_referring): Likewise.
16327 (symtab_node::dump_base): Likewise.
16328 (symtab_node::debug_symtab): Likewise.
16329 * tree-sra.c (convert_callers_for_node): Likewise.
16330 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
16331 * value-prof.c (init_node_map): Likewise.
16333 2017-05-23 Martin Liska <mliska@suse.cz>
16335 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
16336 and symtab_node::debug_symtab to symbol_table::debug.
16337 * cgraphunit.c (analyze_functions): Use the renamed function.
16338 (symbol_table::compile): Likewise.
16339 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
16340 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
16341 * passes.c (execute_todo): Likewise.
16342 * symtab.c (symbol_table::dump): New function.
16343 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
16345 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
16347 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
16348 that nonconst implies exec.
16350 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
16352 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
16353 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
16354 (inline_edge_summary_vec): Turn into ...
16355 (ipa_call_summaries): ... this one.
16356 (redirect_to_unreachable, edge_set_predicate,
16357 evaluate_properties_for_edge, inline_summary_alloc,
16358 reset_ipa_call_summary, reset_inline_summary,
16359 inline_summary_t::duplicate): Update.
16360 (inline_edge_duplication_hook): Turn to ...
16361 (ipa_call_summary_t::duplicate): ... this one.
16362 (inline_edge_removal_hook): Turn to ...
16363 (ipa_call_summary_t::remove): ... this one.
16364 (dump_inline_edge_summary): Turn to ...
16365 (dump_ipa_call_summary): ... this one.
16366 (estimate_function_body_sizes): Update.
16367 (inline_update_callee_summaries): Update.
16368 (remap_edge_change_prob): Update.
16369 (remap_edge_summaries): Update.
16370 (inline_merge_summary): Update.
16371 (do_estimate_edge_time): Update.
16372 (inline_generate_summary): Update.
16373 (inline_read_section): Update.
16374 (inline_read_summary): Update.
16375 (inline_free_summary): Update.
16376 * ipa-inline.c (can_inline_edge_p): Update.
16377 (compute_inlined_call_time): Update.
16378 (want_inline_small_function_p): Update.
16379 (edge_badness): Update.
16380 (early_inliner): Update.
16381 * ipa-inline.h (inline_edge_summary): Turn to ...
16382 (ipa_call_summary): ... this one.
16383 (ipa_call_summary_t): New class.
16384 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
16385 (ipa_call_summaries): New.
16386 (inline_edge_summary): Remove.
16387 (estimate_edge_growth): Update.
16388 * ipa-profile.c (ipa_propagate_frequency_1): Update.
16389 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
16390 * ipa-split.c (execute_split_functions): Update.
16391 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
16393 2017-05-23 Tom de Vries <tom@codesourcery.com>
16395 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
16396 attributes): Document rdrand effective target.
16398 2017-05-23 Tom de Vries <tom@codesourcery.com>
16400 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
16401 attributes): Sort alphabetically.
16403 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
16405 * config/avr/genmultilib.awk: Use gsub instead of gensub.
16407 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
16410 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
16411 V2DF/V2DI splat into two separate patterns, one that handles
16412 registers, and the other that only handles memory. Drop support
16413 for splatting from a GPR on ISA 2.07 and then splitting the
16414 splat into direct move and splat.
16415 (vsx_splat_<mode>_reg): Likewise.
16416 (vsx_splat_<mode>_mem): Likewise.
16418 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
16420 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
16422 2017-05-22 Jakub Jelinek <jakub@redhat.com>
16424 PR middle-end/80809
16425 * omp-low.c (finish_taskreg_remap): New function.
16426 (finish_taskreg_scan): If unit size of ctx->record_type
16427 is non-constant, unshare the size expression and replace
16428 decls in it with possible outer var refs.
16430 PR middle-end/80809
16431 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
16432 GOVD_SHARED rather than GOVD_PRIVATE with it.
16433 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
16434 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
16436 PR middle-end/80853
16437 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
16438 as last argument to build_outer_var_ref for pointer bases of array
16439 section reductions.
16441 2017-05-19 Martin Sebor <msebor@redhat.com>
16443 * print-tree.c (print_node): Print DECL_READ_P flag.
16445 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
16447 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
16448 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
16449 * cgraph.c: Likewise.
16450 * cgraphunit.c: Likewise.
16451 * gengtype.c: Likewise.
16452 * ipa-cp.c: Likewise.
16453 * ipa-devirt.c: Likewise.
16454 * ipa-icf.c: Likewise.
16455 * ipa-predicate.c: Likewise.
16456 * ipa-profile.c: Likewise.
16457 * ipa-prop.c: Likewise.
16458 * ipa-split.c: Likewise.
16460 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
16461 edge_predicate_pool, dump_inline_hints,
16462 inline_summary::account_size_time, redirect_to_unreachable,
16463 edge_set_predicate, set_hint_predicate,
16464 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
16465 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
16466 inline_summary_t::remove, remap_hint_predicate_after_duplication,
16467 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
16468 ipa_call_summary_t::remove, initialize_growth_caches,
16469 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
16470 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
16471 mark_modified, unmodified_parm_1, unmodified_parm,
16472 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
16473 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
16474 compute_bb_predicates, will_be_nonconstant_expr_predicate,
16475 will_be_nonconstant_predicate, record_modified_bb_info,
16476 get_minimal_bb, record_modified, param_change_prob,
16477 phi_result_unknown_predicate, predicate_for_phi_result,
16478 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
16479 estimate_function_body_sizes, compute_inline_parameters,
16480 compute_inline_parameters_for_curren, pass_data_inline_parameters,
16481 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
16482 inline_update_callee_summaries, remap_edge_change_prob,
16483 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
16484 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
16485 inline_analyze_function, inline_summary_t::insert,
16486 inline_generate_summary, read_ipa_call_summary, inline_read_section,
16487 inline_read_summary, write_ipa_call_summary, inline_write_summary,
16488 inline_free_summary): Move to ipa-fnsummary.h
16489 (predicate_t): Remove.
16490 * ipa-fnsummary.c: New file.
16491 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
16492 (enum inline_hints_vals, inline_hints, agg_position_info,
16493 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
16494 inline_summaries, ipa_call_summary, ipa_call_summary_t,
16495 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
16496 dump_inline_summary, dump_inline_hints, inline_generate_summary,
16497 inline_read_summary, inline_write_summary, inline_free_summary,
16498 inline_analyze_function, initialize_inline_failed,
16499 inline_merge_summary, inline_update_overall_summary,
16500 compute_inline_parameters): Move to ipa-fnsummary.h
16501 * ipa-fnsummary.h: New file.
16502 * ipa-inline-transform.h: Include ipa-inline.h.
16503 * ipa-inline.c: LIkewise.
16505 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
16507 * ipa-inline.c (edge_badness): Use inlined_time instead of
16508 inline_summaries->get.
16510 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
16512 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
16514 2017-05-22 Nathan Sidwell <nathan@acm.org>
16516 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
16517 (fdump-lang): Document 'raw' option.
16518 * dumpfile.h (TDI_tu): Delete.
16519 * dumpfile.c (dump_files): Remove translation-unit.
16520 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
16522 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
16524 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
16525 command option from $(AWK) call.
16526 * config/avr/genmultilib.awk: Simplify and rewrite so that it
16527 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
16528 [FORMAT]: Remove handling of variable.
16529 * config/avr/t-multilib: Regenerate.
16531 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
16533 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
16535 (dump_inline_summary): Do not print self_time.
16536 (estimate_function_body_sizes): Do not set self_time.
16537 (compute_inline_parameters): Likewise.
16538 (inline_read_section, inline_write_summary): Do not stream self_time.
16539 * ipa-inline.h (inline_summary): Drop self_time.
16541 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
16543 * ipa-inline-analysis.c (account_size_time): Rename to ...
16544 (inline_summary::account_size_time): ... this one.
16545 (reset_ipa_call_summary): Turn to ...
16546 (ipa_call_summary::reset): ... this one.
16547 (reset_inline_summary): Turn to ...
16548 (inline_summary::reset): ... this one.
16549 (inline_summary_t::remove): Update.
16550 (inline_summary_t::duplicate): Update.
16551 (ipa_call_summary_t::remove): Update.
16552 (dump_inline_summary): Update.
16553 (estimate_function_body_sizes): Update.
16554 (compute_inline_parameters): Update.
16555 (estimate_node_size_and_time): Update.
16556 (inline_merge_summary): Update.
16557 (inline_update_overall_summary): Update.
16558 (inline_read_section): Update.
16559 (inline_write_summary): Update.
16560 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
16561 add account_size_time and reset member functions.
16562 (ipa_call_summary): Add reset function.
16563 * ipa-predicate.h (predicate::operator &): Constify.
16565 2017-05-22 Richard Biener <rguenther@suse.de>
16567 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
16569 2017-05-19 Jason Merrill <jason@redhat.com>
16571 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
16573 2017-05-19 Marek Polacek <polacek@redhat.com>
16576 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
16577 TYPE_OVERFLOW_WRAPS checks.
16579 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
16581 * tree-core.h (enum omp_clause_default_kind): Add
16582 "OMP_CLAUSE_DEFAULT_PRESENT".
16583 * tree-pretty-print.c (dump_omp_clause): Handle it.
16584 * gimplify.c (enum gimplify_omp_var_data): Add
16585 "GOVD_MAP_FORCE_PRESENT".
16586 (gimplify_adjust_omp_clauses_1): Map it to
16587 "GOMP_MAP_FORCE_PRESENT".
16588 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
16590 * gimplify.c (oacc_default_clause): Clarify.
16592 2017-05-19 Nathan Sidwell <nathan@acm.org>
16594 LANG_HOOK_REGISTER_DUMPS
16595 * toplev.c (general_init): Call register dump lang hook.
16596 * doc/invoke.texi: Document -fdump-lang option family.
16597 * dumpfile.c (dump_files): Remove class dump here.
16598 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
16599 * dumpfile.h (tree_dump_index): Remove TDI_class.
16600 * langhooks-def.h (lhd_register_dumps): Declare.
16601 (LANG_HOOKS_REGISTER_DUMPS): Define.
16602 (LANG_HOOKS_INITIALIZER): Add it.
16603 * langhooks.c (lhd_register_dumps): Define.
16604 * langhooks.h (struct lang_hooks): Add register_dumps.
16606 2017-05-19 Nathan Sidwell <nathan@acm.org>
16608 * context.h (context::set_passes): New.
16609 * context.c (context::context): Do not create pass manager.
16610 * toplev.c (general_init): Create pass manager here.
16612 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
16614 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
16615 use this splitter if two add or or instructions would also work for
16616 the constant we want to generate.
16618 2017-05-19 Richard Biener <rguenther@suse.de>
16621 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
16622 predicate evaluation.
16624 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
16626 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
16628 * ipa-inline.c (want_inline_small_function_p): Do not cast to
16631 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
16633 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
16634 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
16635 (inline_edge_summary_vec): Turn into ...
16636 (ipa_call_summaries): ... this one.
16637 (redirect_to_unreachable, edge_set_predicate,
16638 evaluate_properties_for_edge, inline_summary_alloc,
16639 reset_ipa_call_summary, reset_inline_summary,
16640 inline_summary_t::duplicate): Update.
16641 (inline_edge_duplication_hook): Turn to ...
16642 (ipa_call_summary_t::duplicate): ... this one.
16643 (inline_edge_removal_hook): Turn to ...
16644 (ipa_call_summary_t::remove): ... this one.
16645 (dump_inline_edge_summary): Turn to ...
16646 (dump_ipa_call_summary): ... this one.
16647 (estimate_function_body_sizes): Update.
16648 (inline_update_callee_summaries): Update.
16649 (remap_edge_change_prob): Update.
16650 (remap_edge_summaries): Update.
16651 (inline_merge_summary): Update.
16652 (do_estimate_edge_time): Update.
16653 (inline_generate_summary): Update.
16654 (inline_read_section): Update.
16655 (inline_read_summary): Update.
16656 (inline_free_summary): Update.
16657 * ipa-inline.c (can_inline_edge_p): Update.
16658 (compute_inlined_call_time): Update.
16659 (want_inline_small_function_p): Update.
16660 (edge_badness): Update.
16661 (early_inliner): Update.
16662 * ipa-inline.h (inline_edge_summary): Turn to ...
16663 (ipa_call_summary): ... this one.
16664 (ipa_call_summary_t): New class.
16665 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
16666 (ipa_call_summaries): New.
16667 (inline_edge_summary): Remove.
16668 (estimate_edge_growth): Update.
16669 * ipa-profile.c (ipa_propagate_frequency_1): Update.
16670 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
16671 * ipa-split.c (execute_split_functions): Update.
16672 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
16674 2017-05-19 Richard Biener <rguenther@suse.de>
16676 PR middle-end/80764
16677 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
16679 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
16681 * config/rs6000/rs6000.c (struct machine_function): Add field
16682 fpr_is_wrapped_separately.
16683 (rs6000_get_separate_components): Use 64 components. Handle the
16684 new FPR components.
16685 (rs6000_components_for_bb): Handle the FPR components.
16686 (rs6000_emit_prologue_components): Handle the FPR components.
16687 (rs6000_emit_epilogue_components): Handle the FPR components.
16688 (rs6000_set_handled_components): Handle the FPR components.
16689 (rs6000_emit_prologue): Don't output prologue code for those FPRs
16690 that are already separately shrink-wrapped.
16691 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
16692 that are already separately shrink-wrapped.
16694 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
16697 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
16700 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
16701 (define_peephole2 for Altivec d-form load): Add peepholes to catch
16702 cases where the register allocator uses a move and an offsettable
16703 memory operation to/from a FPR register on ISA 2.06/2.07.
16704 (define_peephole2 for Altivec d-form store): Likewise.
16706 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
16709 * config/i386/mmx.md (*mov<mode>_internal): Enable
16710 alternatives 11, 12, 13 and 14 also for 32bit targets.
16711 Remove alternatives 15, 16, 17 and 18.
16712 * config/i386/sse.md (vec_concatv2di): Change
16713 alternative (!x, *y) to (x, ?!*Yn).
16715 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
16717 * dumpfile.h (enum dump_kind): Remove stray comma.
16719 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
16721 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
16722 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
16723 predicate::num_conditions
16724 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
16725 (CHANGED): turn into predicate::changed.
16726 (agg_position_info): Move to ipa-predicate.h
16727 (add_condition, predicate::add_clause, predicate::operator &=,
16728 predicate::or_with, predicate::evaluate, predicate::probability,
16729 dump_condition, dump_clause, predicate::dump,
16730 predicate::remap_after_duplication, predicate::remap_after_inlining,
16731 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
16732 (evaluate_conditions_for_known_args): Update.
16733 (set_cond_stmt_execution_predicate): Update.
16734 * ipa-inline.h: Include ipa-predicate.h
16735 (condition, inline_param_summary, conditions, agg_position_info,
16736 predicate): Move to ipa-predicate.h
16737 * ipa-predicate.c: New file.
16738 * ipa-predicate.h: New file.
16740 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
16742 * final.c (leaf_function_p): Check we are not in a sequence.
16744 2017-05-18 Martin Liska <mliska@suse.cz>
16746 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
16747 * dumpfile.c (dump_register): Use new enum dump_kind.
16748 (get_dump_file_name): Likewise.
16749 (dump_enable_all): Likewise.
16750 (dump_switch_p_1): Likewise.
16751 (enable_rtl_dump_file): Remove usage of TDF_RTL.
16752 * dumpfile.h (enum dump_kind): New enum type.
16753 (struct dump_file_info): Create constructor and
16754 format fields and comments.
16755 * passes.c (pass_manager::register_one_dump_file):
16757 * statistics.c (statistics_early_init): Likewise.
16758 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
16759 TDF_TREE with TDF_SLIM.
16760 (gather_memory_references_ref): Likewise.
16762 2017-05-18 Martin Liska <mliska@suse.cz>
16764 * vec.h (struct vnull): Use it.
16766 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
16768 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
16769 (true_predicate, false_predicate, true_predicate_p,
16770 false_predicate_p): Remove.
16771 (single_cond_predicate, not_inlined_predicate): Turn to member function
16773 (add_condition): Update.
16774 (add_clause): Turn to...
16775 (predicate::add_clause): ... this one; update; allow passing NULL
16777 (and_predicates): Turn to ...
16778 (predicate::operator &=): ... this one.
16779 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
16780 (or_predicates): Turn to ...
16781 (predicate::or_with): ... this one.
16782 (evaluate_predicate): Turn to ...
16783 (predicate::evaluate): ... this one.
16784 (predicate_probability): Turn to ...
16785 (predicate::probability): ... this one.
16786 (dump_condition): Update.
16787 (dump_predicate): Turn to ...
16788 (predicate::dump): ... this one.
16789 (account_size_time): Update.
16790 (edge_set_predicate): Update.
16791 (set_hint_predicate): UPdate.
16792 (evaluate_conditions_for_known_args): Update.
16793 (evaluate_properties_for_edge): Update.
16794 (remap_predicate_after_duplication): Turn to...
16795 (predicate::remap_after_duplication): ... this one.
16796 (remap_hint_predicate_after_duplication): Update.
16797 (inline_summary_t::duplicate): UPdate.
16798 (dump_inline_edge_summary): Update.
16799 (dump_inline_summary): Update.
16800 (set_cond_stmt_execution_predicate): Update.
16801 (set_switch_stmt_execution_predicate): Update.
16802 (compute_bb_predicates): Update.
16803 (will_be_nonconstant_expr_predicate): Update.
16804 (will_be_nonconstant_predicate): Update.
16805 (phi_result_unknown_predicate): Update.
16806 (predicate_for_phi_result): Update.
16807 (array_index_predicate): Update.
16808 (estimate_function_body_sizes): Update.
16809 (estimate_node_size_and_time): Update.
16810 (estimate_ipcp_clone_size_and_time): Update.
16811 (remap_predicate): Rename to ...
16812 (predicate::remap_after_inlining): ... this one.
16813 (remap_hint_predicate): Update.
16814 (inline_merge_summary): Update.
16815 (inline_update_overall_summary): Update.
16816 (estimate_size_after_inlining): Update.
16817 (read_predicate): Rename to ...
16818 (predicate::stream_in): ... this one.
16819 (read_inline_edge_summary): Update.
16820 (write_predicate): Rename to ...
16821 (predicate::stream_out): ... this one.
16822 (write_inline_edge_summary): Update.
16823 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
16824 (clause_t): Turn to uint32_t
16825 (predicate): Turn to class; implement constructor and operators
16827 (size_time_entry): Update.
16828 (inline_summary): Update.
16829 (inline_edge_summary): Update.
16831 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
16833 * fold-const.c (fold_binary_loc): Move transformation...
16834 * match.pd (C - X CMP X): ... here.
16836 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
16838 * config/sparc/sparc.c (sparc_option_override): Set function
16839 alignment for -mcpu=niagara7 to 64 to match the I$ line.
16840 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
16842 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
16844 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
16846 2017-05-18 Marek Polacek <polacek@redhat.com>
16849 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
16850 (pass_ubsan::execute): Call gimple_assign_single_p instead of
16851 gimple_assign_load_p.
16853 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
16855 PR middle-end/80692
16856 * real.c (do_compare): Give decimal_do_compare preference over
16857 comparing just the signs.
16859 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
16861 * doc/md.texi (Canonicalization of Instructions): Describe the
16862 canonical form of instructions that inherently set a condition
16865 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
16867 PR middle-end/80775
16868 * tree-cfg.c: Move deletion of unreachable case statements to after
16869 the merging of consecutive case labels.
16871 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
16873 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
16874 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
16875 restoring of callee-saved registers.
16877 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
16879 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
16880 * config/visium/visium.c (single_set_and_flags): Likewise.
16881 * config/visium/visium.md (Substitutions): Likewise.
16883 2017-05-17 Martin Liska <mliska@suse.cz>
16885 * cfg.c: Introduce dump_flags_t type and
16886 use it instead of int type.
16888 * cfghooks.c: Likewise.
16889 * cfghooks.h (struct cfg_hooks): Likewise.
16890 * cfgrtl.c: Likewise.
16891 * cfgrtl.h: Likewise.
16892 * cgraph.c (cgraph_node::get_body): Likewise.
16893 * coretypes.h: Likewise.
16894 * domwalk.c: Likewise.
16895 * domwalk.h: Likewise.
16896 * dumpfile.c (struct dump_option_value_info): Likewise.
16897 (dump_enable_all): Likewise.
16898 (dump_switch_p_1): Likewise.
16899 (opt_info_switch_p): Likewise.
16900 * dumpfile.h (enum tree_dump_index): Likewise.
16901 (struct dump_file_info): Likewise.
16902 * genemit.c: Likewise.
16903 * generic-match-head.c: Likewise.
16904 * gengtype.c (open_base_files): Likewise.
16905 * gimple-pretty-print.c: Likewise.
16906 * gimple-pretty-print.h: Likewise.
16907 * graph.c (print_graph_cfg): Likewise.
16908 * graphite-scop-detection.c (dot_all_sese): Likewise.
16909 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
16910 * loop-unroll.c (report_unroll): Likewise.
16911 * passes.c (pass_manager::register_one_dump_file): Likewise.
16912 * print-tree.c: Likewise.
16913 * statistics.c: Likewise.
16914 * tree-cfg.c: Likewise.
16915 * tree-cfg.h: Likewise.
16916 * tree-dfa.c: Likewise.
16917 * tree-dfa.h: Likewise.
16918 * tree-dump.c (dump_function): Likewise.
16919 * tree-dump.h (struct dump_info): Likewise.
16920 * tree-pretty-print.c: Likewise.
16921 * tree-pretty-print.h: Likewise.
16922 * tree-ssa-live.c: Likewise.
16923 * tree-ssa-live.h: Likewise.
16924 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
16925 * tree-vect-loop.c: Likewise.
16926 * tree-vect-slp.c: Likewise.
16928 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
16929 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16931 PR tree-optimization/80457
16932 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
16933 of all arguments to a statement as scalar_to_vec operations.
16934 (vectorizable_call): Adjust call to vect_model_simple_cost for
16936 (vectorizable_conversion): Likewise.
16937 (vectorizable_assignment): Likewise.
16938 (vectorizable_shift): Likewise.
16939 (vectorizable_operation): Likewise.
16940 (vectorizable_comparison): Likewise.
16941 (vect_is_simple_cond): Record the def types for operands.
16942 (vectorizable_condition): Likewise, call vect_model_simple_cost.
16943 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
16944 for statement argument count.
16946 2017-05-16 Carl Love <cel@us.ibm.com>
16948 * config/rs6000/rs6000-c: Add support for built-in functions
16949 vector unsigned long long vec_bperm (vector unsigned long long,
16950 vector unsigned char)
16951 vector signed long long vec_mule (vector signed int,
16953 vector unsigned long long vec_mule (vector unsigned int,
16954 vector unsigned int)
16955 vector signed long long vec_mulo (vector signed int,
16957 vector unsigned long long vec_mulo (vector unsigned int,
16958 vector unsigned int)
16959 vector signed char vec_sldw (vector signed char,
16960 vector signed char,
16962 vector unsigned char vec_sldw (vector unsigned char,
16963 vector unsigned char,
16965 vector signed short vec_sldw (vector signed short,
16966 vector signed short,
16968 vector unsigned short vec_sldw (vector unsigned short,
16969 vector unsigned short,
16971 vector signed int vec_sldw (vector signed int,
16974 vector unsigned int vec_sldw (vector unsigned int,
16975 vector unsigned int,
16977 vector signed long long vec_sldw (vector signed long long,
16978 vector signed long long,
16980 vector unsigned long long vec_sldw (vector unsigned long long,
16981 vector unsigned long long,
16983 * config/rs6000/rs6000-c: Add support for built-in functions
16984 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
16985 * config/rs6000/altivec.h: Add defintion for vec_sldw.
16986 * doc/extend.texi: Update the built-in documentation for the
16987 new built-in functions.
16989 2017-05-16 Marek Polacek <polacek@redhat.com>
16993 * tree.c (save_expr): Don't fold the expression.
16995 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
16997 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
16998 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
16999 and (?*y,m). Update insn attributes.
17001 2017-05-16 Martin Liska <mliska@suse.cz>
17003 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
17004 flags argument of print_gimple_stmt, print_gimple_expr,
17005 print_generic_stmt and print_generic_expr.
17006 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
17007 * coretypes.h: Likewise.
17008 * except.c (dump_eh_tree): Likewise.
17009 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
17010 * gimple-pretty-print.h: Likewise.
17011 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
17012 (backprop::push_to_worklist): Likewise.
17013 (backprop::pop_from_worklist): Likewise.
17014 (backprop::process_use): Likewise.
17015 (backprop::intersect_uses): Likewise.
17016 (note_replacement): Likewise.
17017 * gimple-ssa-store-merging.c
17018 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
17019 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
17020 (pass_store_merging::execute): Likewise.
17021 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
17022 (ssa_base_cand_dump_callback): Likewise.
17023 (dump_incr_vec): Likewise.
17024 (replace_refs): Likewise.
17025 (replace_mult_candidate): Likewise.
17026 (create_add_on_incoming_edge): Likewise.
17027 (create_phi_basis): Likewise.
17028 (insert_initializers): Likewise.
17029 (all_phi_incrs_profitable): Likewise.
17030 (introduce_cast_before_cand): Likewise.
17031 (replace_one_candidate): Likewise.
17032 * gimplify.c (gimplify_expr): Likewise.
17033 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
17034 (set_rename): Likewise.
17035 (rename_uses): Likewise.
17036 (copy_loop_phi_nodes): Likewise.
17037 (add_close_phis_to_merge_points): Likewise.
17038 (copy_loop_close_phi_args): Likewise.
17039 (copy_cond_phi_args): Likewise.
17040 (graphite_copy_stmts_from_block): Likewise.
17041 (translate_pending_phi_nodes): Likewise.
17042 * graphite-poly.c (print_pdr): Likewise.
17043 (dump_gbb_cases): Likewise.
17044 (dump_gbb_conditions): Likewise.
17045 (print_scop_params): Likewise.
17046 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
17047 (build_cross_bb_scalars_use): Likewise.
17048 (gather_bbs::before_dom_children): Likewise.
17049 * hsa-dump.c (dump_hsa_immed): Likewise.
17050 * ipa-cp.c (print_ipcp_constant_value): Likewise.
17051 (get_replacement_map): Likewise.
17052 * ipa-inline-analysis.c (dump_condition): Likewise.
17053 (estimate_function_body_sizes): Likewise.
17054 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
17055 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
17056 * ipa-prop.c (ipa_dump_param): Likewise.
17057 (ipa_print_node_jump_functions_for_edge): Likewise.
17058 (ipa_modify_call_arguments): Likewise.
17059 (ipa_modify_expr): Likewise.
17060 (ipa_dump_param_adjustments): Likewise.
17061 (ipa_dump_agg_replacement_values): Likewise.
17062 (ipcp_modif_dom_walker::before_dom_children): Likewise.
17063 * ipa-pure-const.c (check_stmt): Likewise.
17064 (pass_nothrow::execute): Likewise.
17065 * ipa-split.c (execute_split_functions): Likewise.
17066 * omp-offload.c (dump_oacc_loop_part): Likewise.
17067 (dump_oacc_loop): Likewise.
17068 * trans-mem.c (tm_log_emit): Likewise.
17069 (tm_memopt_accumulate_memops): Likewise.
17070 (dump_tm_memopt_set): Likewise.
17071 (dump_tm_memopt_transform): Likewise.
17072 * tree-cfg.c (gimple_verify_flow_info): Likewise.
17073 (print_loop): Likewise.
17074 * tree-chkp-opt.c (chkp_print_addr): Likewise.
17075 (chkp_gather_checks_info): Likewise.
17076 (chkp_get_check_result): Likewise.
17077 (chkp_remove_check_if_pass): Likewise.
17078 (chkp_use_outer_bounds_if_possible): Likewise.
17079 (chkp_reduce_bounds_lifetime): Likewise.
17080 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
17081 (chkp_mark_completed_bounds): Likewise.
17082 (chkp_register_incomplete_bounds): Likewise.
17083 (chkp_mark_invalid_bounds): Likewise.
17084 (chkp_maybe_copy_and_register_bounds): Likewise.
17085 (chkp_build_returned_bound): Likewise.
17086 (chkp_get_bound_for_parm): Likewise.
17087 (chkp_build_bndldx): Likewise.
17088 (chkp_get_bounds_by_definition): Likewise.
17089 (chkp_generate_extern_var_bounds): Likewise.
17090 (chkp_get_bounds_for_decl_addr): Likewise.
17091 * tree-chrec.c (chrec_apply): Likewise.
17092 * tree-data-ref.c (dump_data_reference): Likewise.
17093 (dump_subscript): Likewise.
17094 (dump_data_dependence_relation): Likewise.
17095 (analyze_overlapping_iterations): Likewise.
17096 * tree-inline.c (expand_call_inline): Likewise.
17097 (tree_function_versioning): Likewise.
17098 * tree-into-ssa.c (dump_defs_stack): Likewise.
17099 (dump_currdefs): Likewise.
17100 (dump_names_replaced_by): Likewise.
17101 (dump_update_ssa): Likewise.
17102 (update_ssa): Likewise.
17103 * tree-object-size.c (pass_object_sizes::execute): Likewise.
17104 * tree-parloops.c (build_new_reduction): Likewise.
17105 (try_create_reduction_list): Likewise.
17106 (ref_conflicts_with_region): Likewise.
17107 (oacc_entry_exit_ok_1): Likewise.
17108 (oacc_entry_exit_single_gang): Likewise.
17109 * tree-pretty-print.h: Likewise.
17110 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
17111 (get_scalar_evolution): Likewise.
17112 (add_to_evolution): Likewise.
17113 (get_loop_exit_condition): Likewise.
17114 (analyze_evolution_in_loop): Likewise.
17115 (analyze_initial_condition): Likewise.
17116 (analyze_scalar_evolution): Likewise.
17117 (instantiate_scev): Likewise.
17118 (number_of_latch_executions): Likewise.
17119 (gather_chrec_stats): Likewise.
17120 (final_value_replacement_loop): Likewise.
17121 (scev_const_prop): Likewise.
17122 * tree-sra.c (dump_access): Likewise.
17123 (disqualify_candidate): Likewise.
17124 (create_access): Likewise.
17125 (reject): Likewise.
17126 (maybe_add_sra_candidate): Likewise.
17127 (create_access_replacement): Likewise.
17128 (analyze_access_subtree): Likewise.
17129 (analyze_all_variable_accesses): Likewise.
17130 (sra_modify_assign): Likewise.
17131 (initialize_constant_pool_replacements): Likewise.
17132 (find_param_candidates): Likewise.
17133 (decide_one_param_reduction): Likewise.
17134 (replace_removed_params_ssa_names): Likewise.
17135 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
17136 * tree-ssa-copy.c (dump_copy_of): Likewise.
17137 (copy_prop_visit_cond_stmt): Likewise.
17138 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
17139 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
17140 (record_equivalences_from_stmt): Likewise.
17141 * tree-ssa-dse.c (compute_trims): Likewise.
17142 (delete_dead_call): Likewise.
17143 (delete_dead_assignment): Likewise.
17144 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
17145 (forward_propagate_into_cond): Likewise.
17146 (pass_forwprop::execute): Likewise.
17147 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
17148 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
17150 (move_computations_worker): Likewise.
17151 (execute_sm): Likewise.
17152 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
17153 (remove_exits_and_undefined_stmts): Likewise.
17154 (remove_redundant_iv_tests): Likewise.
17155 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
17156 (adjust_iv_update_pos): Likewise.
17157 * tree-ssa-math-opts.c (bswap_replace): Likewise.
17158 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
17159 (value_replacement): Likewise.
17160 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
17161 * tree-ssa-pre.c (print_pre_expr): Likewise.
17162 (get_representative_for): Likewise.
17163 (create_expression_by_pieces): Likewise.
17164 (insert_into_preds_of_block): Likewise.
17165 (eliminate_insert): Likewise.
17166 (eliminate_dom_walker::before_dom_children): Likewise.
17167 (eliminate): Likewise.
17168 (remove_dead_inserted_code): Likewise.
17169 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
17170 * tree-ssa-reassoc.c (get_rank): Likewise.
17171 (eliminate_duplicate_pair): Likewise.
17172 (eliminate_plus_minus_pair): Likewise.
17173 (eliminate_not_pairs): Likewise.
17174 (undistribute_ops_list): Likewise.
17175 (eliminate_redundant_comparison): Likewise.
17176 (update_range_test): Likewise.
17177 (optimize_range_tests_var_bound): Likewise.
17178 (optimize_vec_cond_expr): Likewise.
17179 (rewrite_expr_tree): Likewise.
17180 (rewrite_expr_tree_parallel): Likewise.
17181 (linearize_expr): Likewise.
17182 (break_up_subtract): Likewise.
17183 (linearize_expr_tree): Likewise.
17184 (attempt_builtin_powi): Likewise.
17185 (attempt_builtin_copysign): Likewise.
17186 (transform_stmt_to_copy): Likewise.
17187 (transform_stmt_to_multiply): Likewise.
17188 (dump_ops_vector): Likewise.
17189 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
17190 (print_scc): Likewise.
17191 (set_ssa_val_to): Likewise.
17192 (visit_reference_op_store): Likewise.
17193 (visit_use): Likewise.
17194 (sccvn_dom_walker::before_dom_children): Likewise.
17195 (run_scc_vn): Likewise.
17196 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
17198 (expr_hash_elt::print): Likewise.
17199 (const_and_copies::pop_to_marker): Likewise.
17200 (const_and_copies::record_const_or_copy_raw): Likewise.
17201 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
17202 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
17203 (dump_predicates): Likewise.
17204 (find_uninit_use): Likewise.
17205 (warn_uninitialized_phi): Likewise.
17206 (pass_late_warn_uninitialized::execute): Likewise.
17207 * tree-ssa.c (verify_vssa): Likewise.
17208 (verify_ssa): Likewise.
17209 (maybe_optimize_var): Likewise.
17210 * tree-vrp.c (dump_value_range): Likewise.
17211 (dump_all_value_ranges): Likewise.
17212 (dump_asserts_for): Likewise.
17213 (register_edge_assert_for_2): Likewise.
17214 (vrp_visit_cond_stmt): Likewise.
17215 (vrp_visit_switch_stmt): Likewise.
17216 (vrp_visit_stmt): Likewise.
17217 (vrp_visit_phi_node): Likewise.
17218 (simplify_cond_using_ranges_1): Likewise.
17219 (fold_predicate_in): Likewise.
17220 (evrp_dom_walker::before_dom_children): Likewise.
17221 (evrp_dom_walker::push_value_range): Likewise.
17222 (evrp_dom_walker::pop_value_range): Likewise.
17223 (execute_early_vrp): Likewise.
17225 2017-05-16 Richard Biener <rguenther@suse.de>
17227 * dwarf2out.c (loc_list_from_tree_1): Do not create
17228 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
17230 2017-05-16 Richard Biener <rguenther@suse.de>
17232 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
17234 (note_variable_value_in_expr): If we resolved the decl ref
17235 do not push to the stack.
17237 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
17239 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
17240 operations in fast-math mode.
17241 (vaddq_f16): Likewise.
17242 (vmul_f16): Likewise.
17243 (vmulq_f16): Likewise.
17244 (vsub_f16): Likewise.
17245 (vsubq_f16): Likewise.
17246 * config/arm/neon.md (add<mode>3): New.
17248 (fma:<VH:mode>3): New. Also remove outdated comment.
17251 2017-05-16 Martin Liska <mliska@suse.cz>
17255 * ipa-devirt.c (warn_types_mismatch): Fix typo.
17256 (odr_types_equivalent_p): Likewise.
17258 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
17260 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
17262 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
17265 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
17266 non-interunit SSE move alternatives with '?'.
17267 (zero-extendsidi peephole2): New peephole to skip intermediate
17268 general register in SSE zero-extend sequence.
17270 2017-05-15 Jeff Law <law@redhat.com>
17272 * reorg.c (relax_delay_slots): Create a new variable to hold
17273 the temporary target rather than clobbering TARGET_LABEL.
17275 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
17276 missing argument to extract_bit_field call.
17277 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
17279 2017-05-15 Martin Liska <mliska@suse.cz>
17282 * gcc.c (process_command): Do not allow empty argument of -o option.
17284 2017-05-15 Renlin Li <renlin.li@arm.com>
17286 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
17287 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
17288 * config/aarch64/constraints.md (Usf): Add long call check.
17289 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
17290 (call_value): Likewise.
17291 (sibcall): Likewise.
17292 (sibcall_value): Likewise.
17294 (call_value_insn): New.
17295 (sibcall_insn): Update rtx pattern.
17296 (sibcall_value_insn): Likewise.
17297 (call_internal): Remove.
17298 (call_value_internal): Likewise.
17299 (sibcall_internal): Likewise.
17300 (sibcall_value_internal): Likewise.
17301 (call_reg): Likewise.
17302 (call_symbol): Likewise.
17303 (call_value_reg): Likewise.
17304 (call_value_symbol): Likewise.
17306 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
17309 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
17311 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
17313 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
17314 compatible with CCGOCmode and with CCZmode.
17316 2017-05-14 Martin Sebor <msebor@redhat.com>
17318 PR middle-end/77671
17319 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
17320 (gimple_fold_builtin_snprintf): Same.
17321 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
17322 (gimple_fold_builtin_snprintf): Same.
17323 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
17324 of character types.
17325 (is_call_safe): New function.
17326 (try_substitute_return_value): Call it.
17327 (try_simplify_call): New function.
17328 (pass_sprintf_length::handle_gimple_call): Call it.
17330 2017-05-14 Martin Sebor <msebor@redhat.com>
17332 PR middle-end/80669
17333 * builtins.c (expand_builtin_stpncpy): Simplify.
17335 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
17337 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
17338 * config/i386/i386.h
17339 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
17340 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
17341 (struct machine_function): Add new members call_ms2sysv,
17342 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
17343 (struct machine_frame_state): New fields sp_realigned and
17344 sp_realigned_offset.
17345 * config/i386/i386.c
17346 (enum xlogue_stub): New enum.
17347 (enum xlogue_stub_sets): New enum.
17348 (class xlogue_layout): New class.
17349 (struct ix86_frame): New fields stack_realign_allocate_offset,
17350 stack_realign_offset and outlined_save_offset. Modify comments to
17351 detail stack layout when using out-of-line stubs.
17352 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
17353 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
17354 -mcall-ms2sysv-xlogues.
17355 (stub_managed_regs): New static variable.
17356 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
17357 registers managed by out-of-line stub.
17358 (disable_call_ms2sysv_xlogues): New function.
17359 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
17360 m->call_ms2sysv when appropriate and compute frame layout for
17362 (sp_valid_at, fp_valid_at): New inline functions.
17363 (choose_basereg): New function.
17364 (choose_baseaddr): Add align parameter, use choose_basereg and modify
17366 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
17367 Use align parameter of choose_baseaddr to generated aligned SSE movs
17369 (pro_epilogue_adjust_stack): Modify to track
17370 machine_frame_state::sp_realigned.
17371 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
17372 (ix86_nsaved_sseregs): Likewise.
17373 (ix86_emit_save_regs): Likewise.
17374 (ix86_emit_save_regs_using_mov): Likewise.
17375 (ix86_emit_save_sse_regs_using_mov): Likewise.
17376 (get_scratch_register_on_entry): Likewise.
17377 (gen_frame_set): New function.
17378 (gen_frame_load): Likewise.
17379 (gen_frame_store): Likewise.
17380 (emit_outlined_ms2sysv_save): Likewise.
17381 (emit_outlined_ms2sysv_restore): Likewise.
17382 (ix86_expand_prologue): Modify stack re-alignment code and call
17383 emit_outlined_ms2sysv_save when appropriate.
17384 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
17385 parameter rtx_insn *insn, which allows the function to be used to only
17386 generate the notes.
17387 (ix86_expand_epilogue): Modify validity checks of frame and stack
17388 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
17389 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
17390 * config/i386/predicates.md
17391 (save_multiple): New predicate.
17392 (restore_multiple): Likewise.
17393 * config/i386/sse.md
17394 (save_multiple<mode>): New pattern.
17395 (save_multiple_realign<mode>): Likewise.
17396 (restore_multiple<mode>): Likewise.
17397 (restore_multiple_and_return<mode>): Likewise.
17398 (restore_multiple_leave_return<mode>): Likewise.
17399 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
17401 2017-05-14 Julia Koval <julia.koval@intel.com>
17403 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
17404 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
17405 (__builtin_ia32_xsetbv): New builtins.
17406 * config/i386/i386.c (ix86_expand_special_args_builtin):
17408 (ix86_expand_builtin): Special expand for new intrinsics.
17409 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
17410 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
17411 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
17413 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17415 * cfganal.c (inverted_post_order_compute): Change argument type
17417 * cfganal.h (inverted_post_order_compute): Adjust prototype.
17418 * df-core.c (rest_of_handle_df_initialize): Adjust.
17419 (rest_of_handle_df_finish): Likewise.
17420 (df_analyze_1): Likewise.
17421 (df_analyze): Likewise.
17422 (loop_inverted_post_order_compute): Change argument to be a vec *.
17423 (df_analyze_loop): Adjust.
17424 (df_get_n_blocks): Likewise.
17425 (df_get_postorder): Likewise.
17426 * df.h (struct df_d): Change field to be a vec.
17427 * lcm.c (compute_laterin): Adjust.
17428 (compute_available): Likewise.
17429 * lra-lives.c (lra_create_live_ranges_1): Likewise.
17430 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
17431 * tree-ssa-pre.c (compute_antic): Likewise.
17433 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17435 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
17436 (depth_first_search::depth_first_search): Change structure init
17437 function to this constructor.
17438 (depth_first_search::add_bb): Rename function to this member.
17439 (depth_first_search::execute): Likewise.
17440 (flow_dfs_compute_reverse_finish): Adjust.
17442 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17444 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
17445 (longest_simple_path): Likewise.
17446 * shrink-wrap.c (spread_components): Likewise.
17447 (disqualify_problematic_components): Likewise.
17448 (emit_common_heads_for_components): Likewise.
17449 (emit_common_tails_for_components): Likewise.
17450 (insert_prologue_epilogue_for_components): Likewise.
17452 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17454 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
17457 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17459 * df-core.c (df_set_blocks): Start using auto_bitmap.
17460 (df_compact_blocks): Likewise.
17461 * df-problems.c (df_rd_confluence_n): Likewise.
17462 * df-scan.c (df_insn_rescan_all): Likewise.
17463 (df_process_deferred_rescans): Likewise.
17464 (df_update_entry_block_defs): Likewise.
17465 (df_update_exit_block_uses): Likewise.
17466 (df_entry_block_bitmap_verify): Likewise.
17467 (df_exit_block_bitmap_verify): Likewise.
17468 (df_scan_verify): Likewise.
17469 * lra-constraints.c (lra_constraints): Likewise.
17470 (undo_optional_reloads): Likewise.
17471 (lra_undo_inheritance): Likewise.
17472 * lra-remat.c (calculate_gen_cands): Likewise.
17473 (do_remat): Likewise.
17474 * lra-spills.c (assign_spill_hard_regs): Likewise.
17475 (spill_pseudos): Likewise.
17476 * tree-ssa-pre.c (bitmap_set_and): Likewise.
17477 (bitmap_set_subtract_values): Likewise.
17479 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17481 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
17482 management with auto_bitmap.
17483 (fix_inter_tick): Likewise.
17484 (fix_recovery_deps): Likewise.
17485 * ira.c (add_store_equivs): Likewise.
17486 (find_moveable_pseudos): Likewise.
17487 (split_live_ranges_for_shrink_wrap): Likewise.
17488 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
17489 (rtx_reuse_manager::seen_def_p): Likewise.
17490 (rtx_reuse_manager::set_seen_def): Likewise.
17491 * print-rtl.h (class rtx_reuse_manager): Likewise.
17493 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17495 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
17497 (migrate_btr_def): Likewise.
17498 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
17499 * df-core.c (loop_post_order_compute): Likewise.
17500 (loop_inverted_post_order_compute): Likewise.
17501 * hsa-common.h: Likewise.
17502 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
17503 * init-regs.c (initialize_uninitialized_regs): Likewise.
17504 * ipa-inline.c (resolve_noninline_speculation): Likewise.
17505 (inline_small_functions): Likewise.
17506 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
17507 * ira.c (combine_and_move_insns): Likewise.
17508 (build_insn_chain): Likewise.
17509 * loop-invariant.c (find_invariants): Likewise.
17510 * lower-subreg.c (propagate_pseudo_copies): Likewise.
17511 * predict.c (tree_predict_by_opcode): Likewise.
17512 (predict_paths_leading_to): Likewise.
17513 (predict_paths_leading_to_edge): Likewise.
17514 (estimate_loops_at_level): Likewise.
17515 (estimate_loops): Likewise.
17516 * shrink-wrap.c (try_shrink_wrapping): Likewise.
17517 (spread_components): Likewise.
17518 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
17519 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
17520 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
17521 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
17522 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
17523 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
17524 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
17525 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
17526 (mark_threaded_blocks): Likewise.
17527 (thread_through_all_blocks): Likewise.
17528 * tree-ssa.c (verify_ssa): Likewise.
17529 (execute_update_addresses_taken): Likewise.
17530 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
17532 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17534 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
17536 (post_order_compute): Likewise.
17537 (inverted_post_order_compute): Likewise.
17538 (pre_and_rev_post_order_compute_fn): Likewise.
17540 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17542 * genrecog.c (int_set::int_set): Explicitly construct our
17543 auto_vec base class.
17544 * vec.h (auto_vec::auto_vec): New constructor.
17546 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17548 * bitmap.h (class auto_bitmap): New constructor taking
17549 bitmap_obstack * argument.
17551 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17553 * bitmap.h (class auto_bitmap): Change type of m_bits to
17554 bitmap_head, and adjust ctor / dtor and member operators.
17556 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
17558 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
17559 when returned register mode doesn't match original mode.
17561 2017-05-12 Jeff Law <law@redhat.com>
17562 Jakub Jelinek <jakub@redhat.com>
17564 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
17565 we look for cc setter after the compare-elim changes.
17566 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
17567 within the vector to match what compare-elim now expects.
17568 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
17569 (xorsi3_flags, one_cmplsi2_flags): Likewise.
17571 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
17572 after the compare-elim changes.
17573 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
17574 the vector to match what compare-elim now expects.
17575 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
17576 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
17577 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
17578 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
17579 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
17581 * config/visium/visium.c (single_set_and_flags): Fix where
17582 we look for cc setter after the compare-elim changes.
17583 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
17584 with the vector to match what compare-elim now expects.
17585 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
17586 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
17587 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
17588 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
17589 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
17590 (neg<mode>2_insn_set_overflow): Likewise.
17592 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
17594 PR middle-end/79794
17595 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
17596 maybe_expand_insn call, set ops[0].target. If still set after call,
17597 set alt_rtl. Add extra arg to recursive calls.
17598 (extract_bit_field): Add alt_rtl argument. Pass to
17600 * expmed.h (extract_bit_field): Fix prototype.
17601 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
17602 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
17603 to extract_bit_field_calls.
17604 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
17605 Pass alt_rtl to extract_bit_field calls.
17606 * calls.c (store_unaligned_arguments_into_psuedos)
17607 load_register_parameters): Pass extra NULL to extract_bit_field calls.
17608 * optabs.c (maybe_legitimize_operand): Clear op->target when call
17610 * optabs.h (struct expand_operand): Add target bitfield.
17612 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
17614 * compare-elim.c (try_eliminate_compare): Canonicalize
17615 operation with embedded compare to
17616 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
17617 (set (reg) (operation)].
17619 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
17621 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
17624 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
17625 cost of adding a carry flag for ADC instruction.
17626 [case MINUS]: Ignore the cost of subtracting a carry flag
17627 for SBB instruction.
17629 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
17631 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
17633 * config/rs6000/bmiintrin.h: New file.
17634 * config/rs6000/bmi2intrin.h: New file.
17635 * config/rs6000/x86intrin.h: New file.
17637 2017-05-12 Jeff Law <law@redhat.com>
17639 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
17642 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
17644 PR middle-end/80707
17645 * tree-cfg.c: Remove cfg edges of unreachable case statements.
17647 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
17649 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
17650 early expansion of vector divide builtins.
17651 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
17652 builtins identified as having unsigned arguments.
17654 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
17656 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
17657 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
17658 expansion of vector logical operations (and, andc, or, xor,
17661 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
17663 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
17664 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
17666 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
17668 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
17669 early GIMPLE expansion of vector multiplies.
17671 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
17673 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
17674 TARGET_HAVE_MOVT conditional.
17675 (movt splitter): Likewise.
17677 2017-05-12 Richard Biener <rguenther@suse.de>
17679 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
17680 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
17681 Fold all stmts not inplace.
17683 2017-05-12 Richard Biener <rguenther@suse.de>
17685 PR tree-optimization/80713
17686 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
17687 inserted_exprs bit for not removed stmts.
17689 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
17691 PR middle-end/69921
17692 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
17693 parallelized" attribute for parallelized OpenACC kernels.
17694 * omp-offload.c (execute_oacc_device_lower): Use it.
17696 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
17697 Set "oacc kernels" attribute.
17698 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
17699 parameter. Adjust all users.
17700 (oacc_fn_attrib_kernels_p): Remove function.
17701 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
17702 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
17703 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
17704 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
17705 assert "oacc kernels" attribute is set.
17707 2017-05-11 Carl Love <cel@us.ibm.com>
17709 * config/rs6000/rs6000-c: Add support for built-in functions
17710 vector unsigned char vec_popcnt (vector signed char)
17711 vector unsigned char vec_popcnt (vector unsigned char)
17712 vector unsigned short vec_popcnt (vector signed short)
17713 vector unsigned short vec_popcnt (vector unsigned short)
17714 vector unsigned int vec_popcnt (vector signed int)
17715 vector unsigned int vec_popcnt (vector unsigned int)
17716 vector unsigned long long vec_popcnt (vector signed long long)
17717 vector unsigned long long vec_popcnt (vector unsigned long long)
17718 vector signed long long vec_slo (vector signed long long,
17719 vector signed char)
17720 vector signed long long vec_slo (vector signed long long,
17721 vector unsigned char)
17722 vector unsigned long long vec_slo (vector unsigned long long,
17723 vector signed char)
17724 vector unsigned long long vec_slo (vector unsigned long long,
17725 vector unsigned char)
17726 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
17727 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
17728 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
17729 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
17730 * doc/extend.texi: Update the built-in documentation file for the
17731 new built-in functions.
17733 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
17735 * attribs.h (sorted_attr_string): Move machine independent
17736 functions for target clone support from the i386 port to common
17737 code. Rename ix86_function_versions to common_function_versions.
17738 Rename make_name to make_unique_name.
17739 (common_function_versions): Likewise.
17740 (make_unique_name): Likewise.
17741 (make_dispatcher_decl): Likewise.
17742 (is_function_default_version): Likewise.
17743 * attribs.c (attr_strcmp): Likewise.
17744 (sorted_attr_string): Likewise.
17745 (common_function_versions): Likewise.
17746 (make_unique_name): Likewise.
17747 (make_dispatcher_decl): Likewise.
17748 (is_function_default_version): Likewise.
17749 * config/i386/i386.c (attr_strcmp): Likewise.
17750 (sorted_attr_string): Likewise.
17751 (ix86_function_versions): Likewise.
17752 (make_name): Likewise.
17753 (make_dispatcher_decl): Likewise.
17754 (is_function_default_version): Likewise.
17755 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
17757 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17760 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
17761 Account for direct move costs for vec_construct of integer
17764 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
17767 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
17768 (UNSPEC_STX_ATOMIC): Ditto.
17769 (loaddi_via_sse): New insn.
17770 (storedi_via_sse): Ditto.
17771 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
17772 Update corresponding peephole2 patterns.
17773 (atomic_storedi_fpu): Ditto.
17775 2017-05-11 Julia Koval <julia.koval@intel.com>
17777 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
17778 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
17780 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
17781 (__builtin_ia32_rsqrt14ss_mask): New builtins.
17782 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
17784 2017-05-11 Nathan Sidwell <nathan@acm.org>
17786 * graphite-poly.c: Include dumpfile.h.
17788 * dumpfle.h (dump_function): Declare here ...
17789 * tree-dump.h (dump_function): ... not here.
17790 * dumpfile.c: #include tree-cfg.h.
17791 (dump_function): Move here from ...
17792 * tree-dump.c (dump_function): ... here.
17793 * gimplify.c: #include splay-tree.h, not tree-dump.h.
17794 * graphite-poly.c: Don't include tree-dump.h.
17795 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
17796 * print-tree.c: Likewise.
17797 * stor-layout.c: Likewise.
17798 * tree-nested.c: Likewise.
17800 * dumpfile.c (dump_start): Use TDF_FLAGS.
17801 (dump_enable_all): Fix TDF_KIND check thinko.
17803 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
17805 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17806 array entries to represent two legal parameterizations of the
17807 overloaded __builtin_cmpb function, as represented by the
17808 P6_OV_BUILTIN_CMPB constant.
17809 (altivec_resolve_overloaded_builtin): Add special case handling
17810 for the __builtin_cmpb function, as represented by the
17811 P6_OV_BUILTIN_CMPB constant.
17812 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
17813 (BU_P6_64BIT_2): New macro.
17814 (BU_P6_OVERLOAD_2): New macro
17815 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
17816 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
17817 (CMPB): Add overload support to represent both 32-bit and 64-bit
17818 compare-bytes function.
17819 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
17820 support for TARGET_CMPB.
17821 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
17822 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
17823 documentation of the __builtin_cmpb overloaded built-in function.
17825 2017-05-11 Richard Biener <rguenther@suse.de>
17827 PR tree-optimization/80705
17828 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
17829 bases are not vectorizable.
17831 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17833 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
17834 when counting register pressure.
17836 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17838 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
17839 (iv_ca_more_deps): Renamed to ...
17840 (iv_ca_compare_deps): ... this.
17841 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
17843 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17845 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
17847 (determine_group_iv_costs): ... here.
17848 (find_inv_vars_cb): Record inv var if it's not recorded before.
17850 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17852 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
17853 (get_shiftadd_cost): Ditto.
17855 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17857 * tree-ssa-address.c: Include header file.
17858 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
17860 (add_to_parts): Refactor.
17861 (addr_to_parts): New parameter. Update use of move_hint_to_base.
17862 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
17865 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17867 PR tree-optimization/53090
17868 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
17870 (extract_cond_operands): Detect condition with IV on both sides
17871 and return COMP_IV_EXPR_2.
17872 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
17873 (rewrite_use_compare): Simplify by removing call to function
17874 extract_cond_operands.
17876 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17878 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
17879 (extract_cond_operands): Detect condition comparing against non-
17880 invariant bound and return appropriate enum value.
17881 (find_interesting_uses_cond): Update use of extract_cond_operands.
17882 Handle its return value accordingly.
17883 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
17885 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17887 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
17888 nonlinear iv_use computation in loop invariant sensitive way.
17890 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17892 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
17893 (find_iv_candidates): Call relate_compare_use_with_all_cands.
17895 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17897 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
17898 (dump_cand): Support iv_cand.inv_exprs.
17899 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
17901 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
17904 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17906 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
17908 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
17909 as local function. Include necessary header files.
17910 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
17912 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17914 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
17916 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17918 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
17919 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
17920 RSHIFT_EXPR and BIT_NOT_EXPR.
17922 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17924 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
17925 (adjust_setup_cost): New parameter supporting round up adjustment.
17926 (struct address_cost_data): Delete.
17927 (force_expr_to_var_cost): Don't bound cost with spill_cost.
17928 (split_address_cost, ptr_difference_cost): Delete.
17929 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
17930 (struct ainc_cost_data): New struct.
17931 (get_address_cost_ainc): New function.
17932 (get_address_cost, get_computation_cost): Reimplement.
17933 (determine_group_iv_cost_address): Record inv_expr for all uses of
17935 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
17936 (iv_ca_has_deps): Reimplemented to ...
17937 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
17939 (iv_ca_extend): Call iv_ca_more_deps.
17941 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17943 * tree-ssa-address.c (struct mem_address): Move to header file.
17944 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
17945 * tree-ssa-address.h (struct mem_address): Move from C file.
17946 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
17948 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17950 * tree-affine.h (aff_combination_type): New interface.
17951 (aff_combination_zero_p): Remove static.
17952 (aff_combination_const_p): New interface.
17953 (aff_combination_singleton_var_p): New interfaces.
17955 2017-05-11 Richard Biener <rguenther@suse.de>
17957 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
17958 Skip unreachable blocks and destinations.
17959 (eliminate): Move stmt removal and fixup ...
17960 (fini_eliminate): ... here. Skip inserted exprs.
17961 (pass_pre::execute): Move fini_pre after fini_eliminate.
17962 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
17963 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
17964 PRE to get rid of dead code that has invalid SSA form and
17965 split critical edges again.
17967 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17969 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
17971 2017-05-11 Richard Biener <rguenther@suse.de>
17973 * passes.c (execute_function_todo): Verify loops if they are
17974 said to be up-to-date.
17975 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
17976 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
17978 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
17981 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
17982 handle calling assemble_external ourself.
17985 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
17986 modes with zero size. Enhance comment.
17988 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17990 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
17991 built-ins for vec_xl and vec_xst with short and char pointer
17994 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
17996 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
17997 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
17998 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
17999 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
18000 (_mm_maskz_min_round_ss): New intrinsics.
18001 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
18002 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
18003 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
18004 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
18005 (__builtin_ia32_minss_mask_round): New builtins.
18006 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
18007 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
18008 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
18010 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
18011 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
18013 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
18016 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
18018 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
18019 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
18020 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
18021 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
18022 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
18023 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
18024 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
18025 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
18026 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
18027 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
18028 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
18029 (__builtin_ia32_mulss_mask_round): New builtins.
18030 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
18031 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
18032 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
18034 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
18035 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
18037 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
18040 2017-05-10 Julia Koval <julia.koval@intel.com>
18042 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
18043 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
18044 (_mm256_setr_m128i): New intrinsics.
18046 2017-05-10 Julia Koval <julia.koval@intel.com>
18048 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
18049 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
18050 (_mm_maskz_rcp14_ss): New intrinsics.
18051 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
18052 (__builtin_ia32_rcp14ss_mask): New builtins.
18053 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
18055 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
18057 PR tree-optimization/51513
18058 * tree-cfg.c (gimple_seq_unreachable_p): New function.
18059 (assert_unreachable_fallthru_edge_p): Use it.
18060 (group_case_labels_stmt): Likewise.
18061 * tree-cfg.h: Prototype it.
18062 * stmt.c: Include cfghooks.h and tree-cfg.h.
18063 (emit_case_dispatch_table) <gap_label>: New local variable.
18064 Use it to fill dispatch table gaps.
18065 Test for default_label before updating probabilities.
18066 (expand_case) <default_label>: Remove unneeded initialization.
18067 Test for unreachable default case statement and remove its edge.
18068 Set default_label accordingly.
18069 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
18071 2017-05-10 Carl Love <cel@us.ibm.com>
18073 * config/rs6000/rs6000-c: Add support for built-in functions
18074 vector signed char vec_neg (vector signed char)
18075 vector signed short int vec_neg (vector short int)
18076 vector signed int vec_neg (vector signed int)
18077 vector signed long long vec_neg (vector signed long long)
18078 vector float vec_neg (vector float)
18079 vector double vec_neg (vector double)
18080 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
18082 * config/rs6000/altivec.h: Add define for vec_neg
18083 * doc/extend.texi: Update the built-in documentation for the
18084 new built-in functions.
18086 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18088 PR tree-optimization/77644
18089 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
18091 2017-05-10 Nathan Sidwell <nathan@acm.org>
18093 * dumpfile.h (TDI_lang_all): New.
18094 (TDF_KIND): New. Renumber others
18095 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
18097 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
18099 (get_dump_file_name): Adjust suffix generation.
18100 (dump_enable_all): Use TDF_KIND.
18101 * doc/invoke.texi (-fdump-lang-all): Document.
18103 * dumpfile.h: Tabify.
18105 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
18108 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
18109 Move member access before delete.
18111 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
18113 * tree-inline.c (expand_call_inline): Split block at stmt
18116 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
18119 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
18120 are now unused after splitting mov{sf,sd}_hardfloat.
18121 (f32_lr2): Likewise.
18122 (f32_lm): Likewise.
18123 (f32_lm2): Likewise.
18124 (f32_li): Likewise.
18125 (f32_li2): Likewise.
18126 (f32_lv): Likewise.
18127 (f32_sr): Likewise.
18128 (f32_sr2): Likewise.
18129 (f32_sm): Likewise.
18130 (f32_sm2): Likewise.
18131 (f32_si): Likewise.
18132 (f32_si2): Likewise.
18133 (f32_sv): Likewise.
18134 (f32_dm): Likewise.
18135 (f32_vsx): Likewise.
18136 (f32_av): Likewise.
18137 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
18138 For movsf, order stores so the VSX stores occur before the GPR
18139 store which encourages the register allocator to use a traditional
18140 FPR instead of a GPR. For movsd, order the stores so that the GPR
18141 store comes before the VSX stores to allow the power6 to work.
18142 This is due to the power6 not having a 32-bit integer store
18143 instruction from a FPR.
18144 (movsf_hardfloat): Likewise.
18145 (movsd_hardfloat): Likewise.
18147 2017-05-09 Martin Sebor <msebor@redhat.com>
18149 PR translation/80280
18150 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
18153 PR translation/80280
18154 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
18155 data member added in r247778.
18156 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
18158 2017-05-09 Nathan Sidwell <nathan@acm.org>
18160 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
18162 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
18165 2017-05-09 Marek Polacek <polacek@redhat.com>
18167 * doc/invoke.texi: Fix typo.
18169 2017-05-09 Richard Biener <rguenther@suse.de>
18171 * tree-vrp.c (vrp_val_is_max): Adjust comment.
18172 (vrp_val_is_min): Likewise.
18173 (set_value_range_to_value): Likewise.
18174 (set_value_range_to_nonnegative): Likewise.
18175 (gimple_assign_nonzero_p): Likewise.
18176 (gimple_stmt_nonzero_p): Likewise.
18177 (vrp_int_const_binop): Likewise. Remove unreachable case.
18178 (adjust_range_with_scev): Adjust comments.
18179 (compare_range_with_value): Likewise.
18180 (extract_range_from_phi_node): Likewise.
18181 (test_for_singularity): Likewise.
18183 2017-05-09 Richard Biener <rguenther@suse.de>
18185 * tree-vrp.c (get_single_symbol): Add assert that we don't
18186 get overflowed constants as invariant part.
18187 (compare_values_warnv): Add comment before the TREE_NO_WARNING
18188 checks. Use wi::cmp instead of recursing for integer constants.
18189 (compare_values): Just ignore whether we assumed undefined
18190 overflow instead of failing the compare.
18191 (extract_range_for_var_from_comparison_expr): Add comment before the
18192 TREE_NO_WARNING sets.
18193 (test_for_singularity): Likewise.
18194 (extract_range_from_comparison): Do not disable optimization
18195 when we assumed undefined overflow.
18196 (extract_range_basic): Remove init of unused var.
18198 2017-05-09 Richard Biener <rguenther@suse.de>
18200 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
18201 (extract_range_from_multiplicative_op_1): Adjust.
18202 (extract_range_from_binary_expr_1): Use int_const_binop.
18204 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
18207 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
18208 rs6000_store_data_bypass_p in seven define_bypass directives and
18209 in several comments.
18210 * config/rs6000/rs6000-protos.h: Add prototype for
18211 rs6000_store_data_bypass_p function.
18212 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
18213 function implements slightly different (rs6000-specific) semantics
18214 than store_data_bypass_p, returning false rather than aborting
18215 with assertion error when arguments do not satisfy the
18216 requirements of store data bypass.
18217 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
18218 rs6000_store_data_bypass_p.
18220 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
18222 * config/xtensa/xtensa-protos.h
18223 (xtensa_initial_elimination_offset): New declaration.
18224 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
18225 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
18226 macro definition, add case for FRAME_POINTER_REGNUM when
18227 FRAME_GROWS_DOWNWARD.
18228 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
18229 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
18230 xtensa_initial_elimination_offset.
18232 2017-05-08 Nathan Sidwell <nathan@acm.org>
18234 * doc/invoke.texi: Alphabetize -fdump options.
18236 2017-05-08 Martin Sebor <msebor@redhat.com>
18238 PR translation/80280
18239 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
18241 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
18243 * target.def (compute_frame_layout): New optional target hook.
18244 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
18245 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
18246 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
18248 * reload1.c (verify_initial_elim_offsets): Likewise.
18249 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
18250 (use_simple_return_p): Call arm_compute_frame_layout if needed.
18251 (arm_get_frame_offsets): Split up into this ...
18252 (arm_compute_frame_layout): ... and this function.
18254 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
18256 * config/aarch64/constraints.md (Usa): New constraint.
18257 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
18259 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
18261 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
18262 with_multilib_list after it has been checked.
18264 2017-05-08 Richard Biener <rguenther@suse.de>
18266 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
18267 (bitmap_set_subtract_values): Likewise.
18269 2017-05-08 Richard Biener <rguenther@suse.de>
18271 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
18272 (gimple_assign_nonzero): ... this and remove strict_overflow_p
18274 (gimple_stmt_nonzero_warnv_p): Rename to ...
18275 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
18277 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
18278 (extract_range_basic): Adjust, do not disable propagation on
18279 strict overflow sensitive simplification.
18280 (vrp_visit_cond_stmt): Likewise.
18282 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
18284 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
18285 body size unconditionally.
18287 2017-05-07 Jeff Law <law@redhat.com>
18290 2017-05-06 Jeff Law <law@redhat.com>
18291 PR tree-optimization/78496
18292 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
18295 PR tree-optimization/78496
18296 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
18297 (simplify_stmt_using_ranges): Call it.
18298 (vrp_dom_walker::before_dom_children): Extract equivalences
18299 from an ASSERT_EXPR with an equality comparison against a
18302 2017-05-06 Jeff Law <law@redhat.com>
18304 PR tree-optimization/78496
18305 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
18308 PR tree-optimization/78496
18309 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
18310 (simplify_stmt_using_ranges): Call it.
18311 (vrp_dom_walker::before_dom_children): Extract equivalences
18312 from an ASSERT_EXPR with an equality comparison against a
18315 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
18317 * lra-constraints.c (lra_copy_reg_equiv): New function.
18318 (split_reg): Use it to copy equivalence information from the
18319 original register to the spill register.
18321 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
18323 PR rtl-optimization/75964
18324 * simplify-rtx.c (simplify_const_relational_operation): Remove
18325 invalid handling of comparisons of integer ABS.
18327 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
18329 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
18330 initialize to zero.
18331 (init_regs): Remove declaration.
18332 (function_arg_advance_32): Initialize error_p as boolean variable.
18334 2017-05-05 Nathan Sidwell <nathan@acm.org>
18336 * store-motion.c (remove_reachable_equiv_notes): Reformat long
18337 lines. Use for (;;).
18339 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18341 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
18342 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
18343 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
18344 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
18345 VF=2 that require versioning.
18347 2017-05-05 David Malcolm <dmalcolm@redhat.com>
18349 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
18352 2017-05-05 David Malcolm <dmalcolm@redhat.com>
18354 * diagnostic.h (diagnostic_override_option_index): Convert from
18355 macro to inline function.
18357 2017-05-05 David Malcolm <dmalcolm@redhat.com>
18359 * diagnostic.c (last_module_changed_p): New function.
18360 (set_last_module): New function.
18361 (diagnostic_report_current_module): Convert macro usage to
18362 the above functions.
18363 * diagnostic.h (diagnostic_context::last_module): Strengthen
18364 from const line_map * to const line_map_ordinary *.
18365 (diagnostic_last_module_changed): Delete macro.
18366 (diagnostic_set_last_module): Delete macro.
18368 2017-05-05 David Malcolm <dmalcolm@redhat.com>
18370 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
18371 with diagnostic_report_diagnostic.
18372 (diagnostic_n_impl_richloc): Likewise.
18373 * diagnostic.h (report_diagnostic): Delete macro.
18374 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
18375 with diagnostic_report_diagnostic.
18376 * substring-locations.c (format_warning_va): Likewise.
18378 2017-05-05 David Malcolm <dmalcolm@redhat.com>
18380 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
18381 save/restor of format_spec. Move option-printing code to...
18382 (print_option_information): ...this new function, and
18383 reimplement by simply printing to the pretty_printer,
18384 rather than appending to the format string.
18386 2017-05-05 David Malcolm <dmalcolm@redhat.com>
18388 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
18389 handling logic into...
18390 (update_effective_level_from_pragmas): ...this new function.
18392 2017-05-04 Andrew Waterman <andrew@sifive.com>
18394 * config/riscv/riscv.opt (mstrict-align): New option.
18395 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
18396 (SLOW_UNALIGNED_ACCESS): Define.
18397 (riscv_slow_unaligned_access): Declare.
18398 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
18400 (riscv_slow_unaligned_access): New variable.
18401 (rocket_tune_info): Set slow_unaligned_access to true.
18402 (optimize_size_tune_info): Set slow_unaligned_access to false.
18403 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
18404 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
18405 (riscv_option_override): Set riscv_slow_unaligned_access.
18406 * doc/invoke.texi: Add -mstrict-align to RISC-V.
18408 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
18410 * config/riscv/riscv.md: Unify indentation.
18412 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
18417 * config/rs6000/rs6000.md (u code attribute): Add FIX and
18419 (extendsi<mode>2): Add support for doing sign extension via
18420 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
18421 don't have ISA 3.0 instructions.
18422 (extendsi<mode>2 splitter): Likewise.
18423 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
18424 generate the normal insns since SImode can now go in vector
18425 registers. Disallow the special UNSPECs needed for previous
18426 machines to hide SImode being used. Add new insns
18427 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
18428 (fix_trunc<mode>si2_stfiwx): Likewise.
18429 (fix_trunc<mode>si2_internal): Likewise.
18430 (fixuns_trunc<mode>si2): Likewise.
18431 (fixuns_trunc<mode>si2_stfiwx): Likewise.
18432 (fctiw<u>z_<mode>_smallint): Likewise.
18433 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
18434 of floating point to 32-bit integer from doing a direct move to
18435 the GPR registers to do a store.
18436 (fctiwz_<mode>): Break long line.
18438 2017-05-05 Bin Cheng <bin.cheng@arm.com>
18440 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
18441 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
18442 (addr_list, addr_offset_valid_p): New.
18443 (split_address_groups): Check offset validity with above function.
18444 (gt-tree-ssa-loop-ivopts.h): Include header file.
18446 2017-05-05 Nathan Sidwell <nathan@acm.org>
18448 * config.gcc (arm*-*-*): Add missing 'fi'.
18450 2017-05-05 Steve Ellcey <sellcey@cavium.com>
18452 * doc/invoke.texi (-fopt-info): Explicitly say order of options
18453 included in -fopt-info does not matter.
18454 * doc/optinfo.texi (-fopt-info): Fix description of default
18455 behavour. Explicitly say order of options included in -fopt-info
18458 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
18460 * config.gcc: Allow combinations of aprofile and rmprofile values for
18461 --with-multilib-list.
18462 * config/arm/t-multilib: New file.
18463 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
18464 variables. Remove setting of ISA and floating-point ABI in
18465 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
18466 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
18467 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
18468 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
18470 * config/arm/t-rmprofile: Likewise except for the matches changes.
18471 * doc/install.texi (--with-multilib-list): Document the combination of
18472 aprofile and rmprofile values and warn about pitfalls in doing that.
18474 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
18476 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
18477 (movdi_aarch64): Likewise.
18479 2017-05-05 Jakub Jelinek <jakub@redhat.com>
18481 PR tree-optimization/80632
18482 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
18484 (build_arrays): Initialize it for virtual phis.
18485 (fix_phi_nodes): Use it for virtual phis.
18487 PR tree-optimization/80558
18488 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
18489 [x, y] op z into [x op, y op z] for op & or | if conditions
18492 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
18493 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
18496 * config/arm/arm.md (use_literal_pool): Remove.
18497 (64-bit immediate split): No longer takes cost into consideration
18498 if arm_disable_literal_pool is enabled.
18499 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
18500 used when arm_disable_literal_pool is enabled.
18501 (arm_max_const_double_inline_cost): Remove use of
18502 arm_disable_literal_pool.
18503 (push_minipool_fix): Add assert.
18504 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
18505 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
18506 (no_literal_pool_sf_immediate): New.
18508 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18510 PR tree-optimization/80613
18511 * tree-ssa-dce.c (propagate_necessity): Remove cases for
18512 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
18514 2017-05-05 Richard Biener <rguenther@suse.de>
18516 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
18518 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
18520 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
18521 of this flag from insn conditions due to removal from r247495.
18523 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
18525 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
18527 (arm_early_store_addr_dep_ptr): Likewise.
18528 * config/arm/aarch-common-protos.h
18529 (arm_early_load_addr_dep_ptr): Add prototype.
18530 (arm_early_store_addr_dep_ptr): Likewise.
18531 * config/arm/cortex-a53.md: Add new bypasses.
18533 2017-05-05 Jakub Jelinek <jakub@redhat.com>
18535 * tree.c (next_type_uid): Change type to unsigned.
18536 (type_hash_canon): Decrement back next_type_uid if
18537 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
18538 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
18541 2017-05-04 Martin Sebor <msebor@redhat.com>
18543 * builtins.c: Fix a trivial typo in a comment.
18545 PR middle-end/79234
18546 * builtins.c (check_sizes): Adjust to handle reading past the end.
18547 Avoid printing excessive upper bound of ranges. Use %E to print
18548 tree nodes instead of converting them to %wu.
18549 (expand_builtin_memchr): New function.
18550 (compute_dest_size): Rename...
18551 (compute_objsize): ...to this.
18552 (expand_builtin_memcpy): Adjust.
18553 (expand_builtin_mempcpy): Adjust.
18554 (expand_builtin_strcat): Adjust.
18555 (expand_builtin_strcpy): Adjust.
18556 (check_strncat_sizes): Adjust.
18557 (expand_builtin_strncat): Adjust.
18558 (expand_builtin_strncpy): Adjust and simplify.
18559 (expand_builtin_memset): Adjust.
18560 (expand_builtin_bzero): Adjust.
18561 (expand_builtin_memcmp): Adjust.
18562 (expand_builtin): Handle memcmp.
18563 (maybe_emit_chk_warning): Check strncat just once.
18565 2017-05-04 Martin Sebor <msebor@redhat.com>
18567 PR preprocessor/79214
18568 PR middle-end/79222
18569 PR middle-end/79223
18570 * builtins.c (check_sizes): Add inlining context and issue
18571 warnings even when -Wno-system-headers is set.
18572 (check_strncat_sizes): Same.
18573 (expand_builtin_strncat): Same.
18574 (expand_builtin_memmove): New function.
18575 (expand_builtin_stpncpy): Same.
18576 (expand_builtin): Handle memmove and stpncpy.
18578 2017-05-04 Bin Cheng <bin.cheng@arm.com>
18580 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
18581 which is not used any more.
18583 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
18585 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
18587 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
18589 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
18590 (cortexa53_tunings): Likewise.
18591 (cortexa57_tunings): Likewise.
18592 (cortexa72_tunings): Likewise.
18593 (cortexa73_tunings): Likewise.
18595 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
18597 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
18598 Set loop alignment to 8.
18600 2017-05-04 Martin Sebor <msebor@redhat.com>
18602 PR translation/80280
18603 * builtins.c (expand_builtin_object_size): Add missing quoting to
18604 %D and like directives.
18605 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
18606 (hsa_type_for_tree_type): Same.
18607 (verify_function_arguments): Same.
18608 * symtab.c (symbol_table::change_decl_assembler_name): Same.
18609 * varasm.c (get_section): Same.
18612 2017-05-04 Martin Sebor <msebor@redhat.com>
18614 PR translation/80280
18615 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
18617 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
18619 * config/aarch64/aarch64.c (generic_addrcost_table):
18620 Change HI/TI mode setting.
18622 2017-05-04 Martin Jambor <mjambor@suse.cz>
18624 PR tree-optimization/80622
18625 * tree-sra.c (comes_initialized_p): New function.
18626 (build_accesses_from_assign): Only set write lazily when
18627 comes_initialized_p is false.
18628 (analyze_access_subtree): Use comes_initialized_p.
18629 (propagate_subaccesses_across_link): Assert !comes_initialized_p
18630 instead of testing for PARM_DECL.
18632 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18634 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
18635 constraint on operand 0 to allow more general addressing modes.
18636 Adjust output template.
18637 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
18639 * config/aarch64/aarch64-protos.h
18640 (aarch64_address_valid_for_prefetch_p): Declare prototype.
18641 * config/aarch64/constraints.md (Dp): New address constraint.
18642 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
18645 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
18647 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
18648 update use of estimate_ipcp_clone_size_and_time.
18649 (estimate_local_effects): Update use of
18650 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
18651 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
18652 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
18653 Return nonspecialized time.
18655 2017-05-04 Richard Biener <rguenther@suse.de>
18657 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
18658 for the last VUSE which def dominates the PHI. Directly call
18660 (get_continuation_for_phi_1): Remove.
18662 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
18664 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
18665 to explain the use of truncating division. Cap the number of
18666 iterations to the maximum given by nb_iterations_upper_bound,
18669 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
18671 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
18672 * configure: Regenerate.
18673 * config.in: Regenerate.
18674 * config/i386/driver-mingw32.c: new file.
18675 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
18676 * config.host: Link driver-mingw32.o on MinGW host.
18677 * doc/install.texi: Document new --enable-mingw-wildcard configure
18680 2017-05-04 Marek Polacek <polacek@redhat.com>
18682 PR tree-optimization/80612
18683 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
18685 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
18686 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
18688 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
18689 (movt splitter): Likewise.
18690 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
18691 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
18692 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
18693 block for Thumb-1 with MOVT.
18694 (thumb2_legitimate_address_p): Move code block ...
18695 (can_avoid_literal_pool_for_label_p): ... into this new function.
18696 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
18698 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
18699 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
18700 "M-profile targets with the MOVT instruction".
18702 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
18704 * config/arm/arm-builtins.c (arm_init_builtins): Rename
18705 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
18706 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
18708 2017-05-04 Martin Liska <mliska@suse.cz>
18710 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
18711 variable cond_code.
18713 2017-05-04 Richard Biener <rguenther@suse.de>
18715 * tree.c (array_at_struct_end_p): Handle arrays at struct
18716 end with flexarrays more conservatively. Refactor and treat
18717 arrays of arrays or aggregates more strict. Fix
18718 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
18719 * tree.c (array_at_struct_end_p): Adjust prototype.
18720 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
18721 * gimple-fold.c (get_range_strlen): Likewise.
18722 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
18724 2017-05-04 Richard Biener <rguenther@suse.de>
18726 PR tree-optimization/31130
18727 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
18729 (supports_overflow_infinity): Likewise.
18730 (is_negative_overflow_infinity): Likewise.
18731 (is_positive_overflow_infinity): Likewise.
18732 (is_overflow_infinity): Likewise.
18733 (stmt_overflow_infinity): Likewise.
18734 (overflow_infinity_range_p): Likewise.
18735 (usable_range_p): Remove as always returning true.
18736 (make_overflow_infinity): Remove.
18737 (negative_overflow_infinity): Likewise.
18738 (positive_overflow_infinity): Likewise.
18739 (avoid_overflow_infinity): Likewise.
18740 (set_value_range): Adjust accordingly.
18741 (set_value_range_to_nonnegative): Likewise, remove now unused
18742 overflow_infinity arg.
18743 (vrp_operand_equal_p): Adjust.
18744 (update_value_range): Likewise.
18745 (range_int_cst_singleton_p): Likewise.
18746 (operand_less_p): Likewise.
18747 (compare_values_warnv): Likewise.
18748 (extract_range_for_var_from_comparison_expr): Likewise.
18749 (vrp_int_const_binop): Likewise.
18750 (zero_nonzero_bits_from_vr): Likewise.
18751 (extract_range_from_multiplicative_op_1): Likewise.
18752 (extract_range_from_binary_expr_1): Likewise.
18753 (extract_range_from_unary_expr): Likewise.
18754 (extract_range_from_comparison): Likewise.
18755 (extract_range_basic): Likewise.
18756 (adjust_range_with_scev): Likewise.
18757 (compare_ranges): Likewise.
18758 (compare_range_with_value): Likewise.
18759 (dump_value_range): Likewise.
18760 (test_for_singularity): Likewise, remove strict_overflow_p parameter
18762 (simplify_cond_using_ranges): Adjust.
18764 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
18766 * brig-builtins.def: Added a builtin for class_f64.
18767 * builtin-types.def: Added a builtin type needed by class_f64.
18769 2017-05-03 Jason Merrill <jason@redhat.com>
18771 * timevar.def: Add TV_CONSTEXPR.
18773 2017-05-03 David Malcolm <dmalcolm@redhat.com>
18775 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
18777 2017-05-03 Martin Jambor <mjambor@suse.cz>
18779 * ipa-prop.c (ipa_update_after_lto_read): Removed.
18780 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
18781 * ipa-cp.c (ipcp_propagate_stage): Do not call
18782 ipa_update_after_lto_read.
18783 * ipa-inline.c (ipa_inline): Likewise.
18785 2017-05-03 Martin Jambor <mjambor@suse.cz>
18787 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
18788 tag. Added a default constructor and a destructor.
18789 (ipa_edge_args_sum_t): New class;
18790 (ipa_edge_args_sum): Declare.
18791 (ipa_edge_args_vector): Remove declaration.
18792 (IPA_EDGE_REF): Use ipa_edge_args_sum.
18793 (ipa_free_edge_args_substructures): Remove declaration.
18794 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
18795 (ipa_edge_args_info_available_for_edge_p): Likewise.
18796 * ipa-prop.c (ipa_edge_args_vector): Removed.
18797 (edge_removal_hook_holder): Likewise.
18798 (edge_duplication_hook_holder): Likewise.
18799 (ipa_edge_args_sum): New variable.
18800 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
18801 ipa_edge_args_vector.
18802 (ipa_free_edge_args_substructures): Likewise.
18803 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
18804 ipa_edge_args_vector.
18805 (ipa_edge_removal_hook): Turned into method
18806 ipa_edge_args_sum_t::remove.
18807 (ipa_edge_duplication_hook): Turned into method
18808 ipa_edge_args_sum_t::duplicate.
18809 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
18810 registering edge hooks.
18811 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
18812 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
18813 ipa_edge_args_sum instead of ipa_edge_args_vector.
18814 * ipa-profile.c (ipa_profile): Likewise.
18816 2017-05-03 Martin Jambor <mjambor@suse.cz>
18818 * symbol-summary.h (function_summary): New method exists.
18819 (function_summary::symtab_removal): Deallocate through release.
18820 (call_summary): New class.
18821 (gt_ggc_mx): New overload.
18822 (gt_pch_nx): Likewise.
18823 (gt_pch_nx): Likewise.
18825 2017-05-03 Jeff Law <law@redhat.com>
18827 PR tree-optimization/78496
18828 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
18829 from simplify_cond_using_ranges. Split off code to walk
18830 backwards through casts into ...
18831 (simplify_cond_using_ranges_2): New function.
18832 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
18833 (execute_vrp): After identifying jump threads, call
18834 simplify_cond_using_ranges_2.
18836 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
18839 * ipa-inline.h (inline_summary): Add ctor.
18840 (create_ggc): Do not use ggc_cleared_alloc.
18842 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
18843 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18845 * gcc.c (handle_braces): Support escaping in switch matching
18847 * doc/invoke.texi (Spec Files): Document it.
18848 Remove superfluous @code markup in items.
18850 2017-05-03 David Malcolm <dmalcolm@redhat.com>
18852 * diagnostic-show-locus.c (struct column_range): New struct.
18853 (get_affected_columns): New function.
18854 (get_printed_columns): New function.
18855 (struct correction): New struct.
18856 (correction::ensure_capacity): New function.
18857 (correction::ensure_terminated): New function.
18858 (struct line_corrections): New struct.
18859 (line_corrections::~line_corrections): New dtor.
18860 (line_corrections::add_hint): New function.
18861 (layout::print_trailing_fixits): Reimplement in terms of the new
18863 (selftest::test_overlapped_fixit_printing): New function.
18864 (selftest::diagnostic_show_locus_c_tests): Call it.
18866 2017-05-03 Nathan Sidwell <nathan@acm.org>
18868 Canonicalize canonical type hashing
18869 * tree.h (type_hash_canon_hash): Declare.
18870 * tree.c (type_hash_list, attribute_hash_list): Move into
18871 type_hash_canon_hash.
18872 (build_type_attribute_qual_variant): Break out hash code calc into
18873 type_hash_canon_hash.
18874 (type_hash_canon_hash): New. Generic type hash computation.
18875 (build_range_type_1, build_array_type_1, build_function_type,
18876 build_method_type_directly, build_offset_type, build_complex_type,
18877 make_vector_type): Call it.
18879 2017-05-03 Richard Biener <rguenther@suse.de>
18881 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
18882 When all DRs have unknown misaligned do not always peel
18883 when there is a store but apply the same costing model as if
18884 there were only loads.
18886 2017-05-03 Richard Biener <rguenther@suse.de>
18889 PR tree-optimization/80492
18890 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
18891 compare_base_decls returning dont-know properly.
18893 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
18895 * config/arm/iterators.md (CCSI): New mode iterator.
18896 (arch): New mode attribute.
18897 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
18898 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
18899 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
18900 code iterator for success result mode.
18901 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
18902 the corresponding new insn generators.
18904 2017-05-03 Bin Cheng <bin.cheng@arm.com>
18907 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18908 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
18910 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
18912 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
18913 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
18914 (DDR_A): Wrap DDR argument in brackets.
18915 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
18916 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
18917 (DDR_REVERSED_P): Likewise.
18919 2017-05-03 Jakub Jelinek <jakub@redhat.com>
18921 PR tree-optimization/79472
18922 * tree-switch-conversion.c (struct switch_conv_info): Add
18923 contiguous_range and default_case_nonstandard fields.
18924 (collect_switch_conv_info): Compute contiguous_range and
18925 default_case_nonstandard fields, don't clear final_bb if
18926 contiguous_range and only the default case doesn't have the required
18928 (check_all_empty_except_final): Set default_case_nonstandard instead
18929 of failing if contiguous_range and the default case doesn't have empty
18931 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
18932 and only the default case doesn't have the required constants. Skip
18934 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
18935 if default_case_nonstandard.
18936 (build_constructors): Build constant 1 just once. Assert that default
18937 values aren't inserted in between cases if contiguous_range. Skip
18939 (build_arrays): Skip virtual phis.
18940 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
18941 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
18942 Handle virtual phis.
18943 (gen_inbound_check): Handle default_case_nonstandard case.
18944 (process_switch): Adjust check_final_bb caller. Call
18945 gather_default_values with the first non-default case instead of
18946 default case if default_case_nonstandard.
18948 2017-05-02 Nathan Sidwell <nathan@acm.org>
18950 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
18951 check. Fix formatting.
18953 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
18955 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
18956 errors when comparing specialized and unspecialized times.
18958 2017-05-02 David Malcolm <dmalcolm@redhat.com>
18960 * diagnostic-show-locus.c
18961 (layout::should_print_annotation_line_p): Make private.
18962 (layout::print_annotation_line): Make private.
18963 (layout::annotation_line_showed_range_p): Make private.
18964 (layout::show_ruler): Make private.
18965 (layout::print_source_line): Make private. Pass in line and
18966 line_width, rather than calling location_get_source_line. Drop
18968 (layout::print_leading_fixits): New method.
18969 (layout::print_any_fixits): Rename to...
18970 (layout::print_trailing_fixits): ...this, and make private.
18971 Don't print newline fixits.
18972 (diagnostic_show_locus): Move logic for printing one row into...
18973 (layout::print_line): ...this new function. Move the
18974 location_get_source_line call and error-handling from
18975 print_source_line to here. Call print_leading_fixits, and rename
18976 print_any_fixits to print_trailing_fixits.
18977 (selftest::test_fixit_insert_containing_newline): Update now that
18978 newlines are partially supported.
18979 (selftest::test_fixit_insert_containing_newline_2): New test.
18980 (selftest::test_fixit_replace_containing_newline): Update comments.
18981 (selftest::diagnostic_show_locus_c_tests): Call the new test.
18982 * edit-context.c (class added_line): New class.
18983 (class edited_line): Describe newline handling in comment.
18984 (edited_line::actually_edited_p): New method.
18985 (edited_line::print_content): Delete redundant decl.
18986 (edited_line::m_predecessors): New field.
18987 (edited_file::print_content): Call edited_line::print_content.
18988 (edited_file::print_diff): Update to support newlines.
18989 (edited_file::print_diff_hunk): Likewise.
18990 (edited_file::print_run_of_changed_lines): New function.
18991 (edited_file::print_diff_line): Convert to...
18992 (print_diff_line): ...this.
18993 (edited_file::get_effective_line_count): New function.
18994 (edited_line::edited_line): Initialize new field m_predecessors.
18995 (edited_line::~edited_line): Clean up m_predecessors.
18996 (edited_line::apply_fixit): Handle newlines.
18997 (edited_line::get_effective_line_count): New function.
18998 (edited_line::print_content): New function.
18999 (edited_line::print_diff_lines): New function.
19000 (selftest::test_applying_fixits_insert_containing_newline): New
19002 (selftest::test_applying_fixits_replace_containing_newline): New
19004 (selftest::insert_line): New function.
19005 (selftest::test_applying_fixits_multiple_lines): Add example of
19007 (selftest::edit_context_c_tests): Call the new tests.
19009 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19011 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
19012 parameter cand. Update dump information.
19013 (get_computation_cost): Update uses.
19015 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19017 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
19018 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
19019 (get_computation_at, rewrite_use_address): Update use of
19020 get_computation_aff.
19022 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19024 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
19025 (get_computation): Delete.
19026 (get_computation_cost): Implement like get_computation_cost_at.
19027 Use get_computation_at.
19028 (get_computation_cost_at): Delete.
19029 (rewrite_use_nonlinear_expr): Use get_computation_at.
19030 (rewrite_use_compare, remove_unused_ivs): Ditto.
19032 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19034 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
19036 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19038 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
19039 (ivopts_global_cost_for_size): Rename parameter and update uses.
19040 (iv_ca_recount_cost): Update uses.
19041 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
19042 candidates seperately in n_invs and n_cands.
19043 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
19045 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19047 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
19048 (find_inv_vars_cb): New.
19049 (find_depends): Renamed to ...
19050 (find_inv_vars): ... this.
19051 (add_candidate_1, force_var_cost): Call find_inv_vars.
19052 (split_address_cost, determine_group_iv_cost_cond): Ditto.
19054 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19056 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
19057 inv_vars. Add inv_exprs.
19058 (struct iv_cand): Rename depends_on to inv_vars.
19059 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
19060 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
19061 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
19062 (dump_cand): Dump inv_vars.
19063 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
19064 (record_invariant, find_depends, add_candidate_1): Ditto.
19065 (set_group_iv_cost, force_var_cost): Ditto.
19066 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
19067 (get_computation_cost_at, get_computation_cost): Ditto.
19068 (determine_group_iv_cost_generic): Ditto.
19069 (determine_group_iv_cost_address): Ditto.
19070 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
19071 (determine_group_iv_costs): Ditto.
19072 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
19073 (iv_ca_set_remove_invariants): Renamed to ...
19074 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
19075 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
19076 (iv_ca_set_add_invariants): Renamed to ...
19077 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
19078 (iv_ca_set_cp): Use iv_ca_set_add_invs.
19079 (iv_ca_has_deps): Support inv_vars and inv_exprs.
19080 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
19081 (create_new_ivs): Remove useless dump.
19083 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19085 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
19087 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
19088 (iv_ca_set_no_cp, create_new_iv): Ditto.
19090 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19092 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
19094 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
19096 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
19097 function tree_check2.
19099 2017-05-02 Martin Liska <mliska@suse.cz>
19101 * doc/gcov.texi: Add missing preposition.
19102 * gcov.c (function_info::function_info): Properly fill up
19103 all member variables.
19105 2017-05-02 Tamar Christina <tamar.christina@arm.com>
19107 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
19109 2017-05-02 Tamar Christina <tamar.christina@arm.com>
19111 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
19113 2017-05-02 Martin Liska <mliska@suse.cz>
19116 * lto-streamer-in.c (lto_read_tree_1): Remove
19117 LTO_STREAMER_DEBUG.
19118 * lto-streamer.c (struct tree_hash_entry): Likewise.
19119 (struct tree_entry_hasher): Likewise.
19120 (tree_entry_hasher::hash): Likewise.
19121 (tree_entry_hasher::equal): Likewise.
19122 (lto_streamer_init): Likewise.
19123 (lto_orig_address_map): Likewise.
19124 (lto_orig_address_get): Likewise.
19125 (lto_orig_address_remove): Likewise.
19126 * lto-streamer.h: Likewise.
19127 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
19128 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
19130 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
19132 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
19133 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
19134 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
19135 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
19136 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
19137 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
19138 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
19139 (mm_maskz_sub_ss): New intrinsics.
19140 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
19141 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
19142 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
19143 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
19144 (__builtin_ia32_subss_mask_round): New builtins.
19145 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
19146 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
19147 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
19149 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
19150 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
19152 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
19155 2017-05-02 Martin Jambor <mjambor@suse.cz>
19157 PR tree-optimization/78687
19158 * tree-sra.c (access): New field parent.
19159 (process_subtree_disqualification): New function.
19160 (disqualify_candidate): Call it.
19161 (build_accesses_from_assign): Reset write flag if creating an
19163 (build_access_subtree): Fill in parent field and also prpagate
19164 down grp_write flag.
19165 (create_artificial_child_access): New parameter set_grp_write, set
19166 grp_write to its value.
19167 (propagate_subaccesses_across_link): Also propagate grp_write flag
19169 (propagate_all_subaccesses): Push the closest parent back to work
19170 queue if add_access_to_work_queue returned true.
19172 2017-05-02 Richard Biener <rguenther@suse.de>
19174 * common.opt (fstrict-overflow): Alias negative to fwrapv.
19175 * doc/invoke.texi (fstrict-overflow): Remove all traces of
19176 -fstrict-overflow documentation.
19177 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
19178 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
19179 flag_strict_overflow.
19180 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
19181 * lto-opts.c (lto_write_options): Do not stream it.
19182 * lto-wrapper.c (merge_and_complain): Do not handle it.
19183 * opts.c (default_options_table): Do not set -fstrict-overflow.
19184 (finish_options): Likewise do not clear it when sanitizing.
19185 * simplify-rtx.c (simplify_const_relational_operation): Do not
19186 test flag_strict_overflow.
19188 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
19190 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
19191 using enabled attribute.
19192 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
19193 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
19194 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
19195 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
19196 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
19197 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
19198 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
19199 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
19200 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
19201 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
19203 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
19205 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
19207 2017-05-02 Richard Biener <rguenther@suse.de>
19209 PR tree-optimization/80591
19211 2017-04-10 Richard Biener <rguenther@suse.de>
19213 * tree-ssa-structalias.c (find_func_aliases): Properly handle
19216 2017-05-02 Richard Biener <rguenther@suse.de>
19218 PR tree-optimization/80549
19219 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
19220 (cleanup_tree_cfg_noloop): Create forwarders to known loop
19221 headers if they do not have a preheader.
19223 2017-05-02 Martin Liska <mliska@suse.cz>
19226 * common.opt: Fix typo.
19227 * doc/invoke.texi: Likewise.
19229 2017-05-01 Jan Beulich <jbeulich@suse.com>
19231 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
19232 swapping, add (x,x,m,x,n) alternative.
19234 2017-05-01 Nathan Sidwell <nathan@acm.org>
19236 * calls.c (combine_pending_stack_adjustment_and_call): Remove
19237 unnecessary unadjusted_alignment check.
19239 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
19242 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
19244 * gimplify.c (gimplify_cilk_detach): New function.
19245 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
19246 * tree-core.h: Document EXPR_CILK_SPAWN.
19247 * tree.h (EXPR_CILK_SPAWN): Define.
19249 2017-05-01 David Malcolm <dmalcolm@redhat.com>
19251 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
19252 to use new fixit_hint representation, using the "replace" logic.
19253 (get_line_span_for_fixit_hint): Likewise.
19254 (layout::print_any_fixits): Likewise.
19255 (selftest::test_one_liner_many_fixits): Rename to...
19256 (selftest::test_one_liner_many_fixits_1): ...this, and update
19257 comment and expected output to reflect that the multiple fix-it
19258 hints are now consolidated into one insertion.
19259 (selftest::test_one_liner_many_fixits_2): New test.
19260 (selftest::test_diagnostic_show_locus_one_liner): Update for
19262 (selftest::test_fixit_consolidation): Update for fix-it API
19264 * diagnostic.c (print_parseable_fixits): Likewise.
19265 * edit-context.c (edited_line::m_line_events): Convert from
19266 auto_vec <line_event *> to auto_vec <line_event>.
19267 (class line_event): Convert from abstract base class to a concrete
19268 class, taking over the role of replace_event.
19269 (class insert_event): Delete.
19270 (class replace_event): Rename to class line_event. Convert to
19272 (edit_context::add_fixits): Reimplement.
19273 (edit_context::apply_insert): Delete.
19274 (edit_context::apply_replace): Rename to...
19275 (edit_context::apply_fixit): ...this. Convert to half-open range.
19276 (edited_file::apply_insert): Delete.
19277 (edited_file::apply_replace): Rename to...
19278 (edited_file::apply_fixit): ...this.
19279 (edited_line::~edited_line): Drop deletion of events.
19280 (edited_line::apply_insert): Delete.
19281 (edited_line::apply_replace): Rename to...
19282 (edited_line::apply_fixit): ...this. Convert to half-open range.
19283 Update for change to type of m_line_events.
19284 * edit-context.h (edit_context::apply_insert): Delete.
19285 (edit_context::apply_replace): Rename to...
19286 (edit_context::apply_fixit): ...this.
19288 2017-05-01 Martin Sebor <msebor@redhat.com>
19290 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
19293 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
19296 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
19297 __get_cpuid_max returns 0.
19298 (__get_cpuid_count): Ditto.
19300 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
19302 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
19303 replacement expression is another instance of one of its arguments.
19305 2017-05-01 Jakub Jelinek <jakub@redhat.com>
19308 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
19309 check for stack push/pop autoinc.
19310 * config/i386/i386.c (ix86_agi_dependent): Return false
19311 if the only reason why modified_in_p returned true is that
19312 addr is SP based and set_insn is a push or pop.
19314 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
19316 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
19319 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
19322 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
19323 (account_size_time): Use two predicates - exec_pred and
19325 (evaluate_conditions_for_known_args): Compute both clause and
19327 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
19328 (inline_summary_t::duplicate): Update.
19329 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
19331 (compute_inline_parameters): Likewise.
19332 (estimate_edge_size_and_time): Update caluclation of time.
19333 (estimate_node_size_and_time): Compute both time and nonspecialized
19335 (estimate_ipcp_clone_size_and_time): Update.
19336 (inline_merge_summary): Update.
19337 (do_estimate_edge_time): Update.
19338 (do_estimate_edge_size): Update.
19339 (do_estimate_edge_hints): Update.
19340 (inline_read_section, inline_write_summary): Stream both new predicates.
19341 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
19343 (compute_inlined_call_time): Cleanup.
19344 (big_speedup_p): Update.
19345 (edge_badness): Update.
19346 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
19347 (size_time_entry): Replace predicate by exec_predicate and
19348 nonconst_predicate.
19349 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
19350 (estimate_edge_time): Return also nonspec_time.
19351 (reset_edge_growth_cache): Update.
19353 2017-04-29 Jakub Jelinek <jakub@redhat.com>
19355 PR rtl-optimization/80491
19356 * ifcvt.c (noce_process_if_block): When looking for x setter
19357 with missing else_bb, don't check only the insn right before
19358 cond_earliest, but look for the last insn that x is modified in
19359 within the same bb.
19361 PR rtl-optimization/80491
19362 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
19364 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
19366 PR tree-optimization/80487
19367 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
19369 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19371 PR tree-optimization/79697
19372 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
19373 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
19374 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
19376 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
19377 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
19379 2017-04-28 Martin Sebor <msebor@redhat.com>
19381 PR tree-optimization/80523
19382 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
19383 (init_target_to_host_charmap, target_to_host, target_strtol10): New
19385 (maybe_warn, format_directive, parse_directive): Use new functions.
19386 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
19388 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
19390 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
19392 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
19394 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
19395 target_header_dir): Set correctly.
19396 * configure: Regenerated.
19397 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
19398 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
19399 instead of SYSTEM_HEADER_DIR.
19401 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
19403 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
19404 (estimate_local_effects): Likewise.
19405 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
19406 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
19407 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
19408 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
19409 do_estimate_edge_time, estimate_edge_time): Likewise.
19410 * ipa-inline-analysis.c (estimate_node_size_and_time,
19411 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
19412 (estimate_time_after_inlining): Remove.
19414 2017-04-28 Martin Liska <mliska@suse.cz>
19416 * doc/gcov.texi: Enhance documentation of gcov.
19418 2017-04-28 Martin Liska <mliska@suse.cz>
19420 * doc/gcov.texi: Sort options in alphabetic order.
19421 * doc/gcov-dump.texi: Likewise.
19422 * doc/gcov-tool.texi: Likewise.
19423 * gcov.c (print_usage): Likewise.
19424 * gcov-dump.c (print_usage): Likewise.
19425 * gcov-tool.c (print_merge_usage_message): Likewise.
19426 (print_rewrite_usage_message): Likewise.
19427 (print_overlap_usage_message): Likewise.
19429 2017-04-28 Martin Liska <mliska@suse.cz>
19431 PR gcov-profile/53915
19432 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
19434 2017-04-28 Martin Liska <mliska@suse.cz>
19436 PR gcov-profile/79891
19437 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
19438 is marked by compiler as living on a line.
19439 (get_cycles_count): Remove usage of the union.
19440 (output_intermediate_file): Likewise.
19441 (find_source): Fix GNU coding style.
19442 (accumulate_line_counts): Remove old non-all block mode.
19443 (output_lines): Remove usage of the union.
19444 * profile.c (output_location): Include all BBs, even if
19445 belonging to a same line (and file) as a previous BB.
19447 2017-04-28 Martin Liska <mliska@suse.cz>
19449 * gcov.c (process_args): Handle new argument 'w'.
19450 (read_graph_file): Assign ID to BBs.
19451 (output_branch_count): Display BB # if verbose flag is set.
19452 (output_lines): Likewise for arcs.
19453 (print_usage): Add '--verbose' option help.
19454 * doc/gcov.texi: Document --verbose (-w) option.
19456 2017-04-28 Martin Liska <mliska@suse.cz>
19458 * gcov.c (struct block_location_info): New struct.
19459 (process_file): Fill up the new structure.
19460 (read_graph_file): Replace usage of encoding by the newly added
19462 (add_line_counts): Likewise.
19463 (accumulate_line_counts): Remove usage of the union.
19464 (function_info::function_info): New function.
19465 (function_info::~function_info): Likewise.
19466 (process_file): Call delete instead of release_function.
19467 (release_function): Release the function.
19468 (release_structures): Call delete instead of release_function.
19469 (solve_flow_graph): Replace usage of num_blocks.
19470 (find_exception_blocks): Likewise.
19471 (output_lines): Fix GNU coding style.
19473 2017-04-28 Martin Liska <mliska@suse.cz>
19476 * coverage.c (coverage_remove_note_file): New function.
19477 * coverage.h: Declare the function.
19478 * toplev.c (finalize): Clean if an error has been seen.
19480 2017-04-28 Martin Liska <mliska@suse.cz>
19482 PR gcov-profile/80031
19483 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
19484 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
19485 * gcov.c (read_graph_file): Read just number of blocks.
19486 * profile.c (branch_prob): Do not stream 0 flags per a basic
19489 2017-04-28 Martin Liska <mliska@suse.cz>
19491 * gcov-dump.c (tag_*): Add new argument to declarations.
19492 (dump_gcov_file): Likewise.
19493 (tag_blocks): Add and use new argument depth.
19494 (tag_arcs): Likewise.
19495 (tag_lines): Likewise.
19496 (tag_counters): Likewise.
19497 (tag_summary): Likewise.
19498 (dump_working_sets): Use depth to do a proper indentation.
19500 2017-04-28 Jakub Jelinek <jakub@redhat.com>
19503 * cgraph.h (symtab_node::debug_symtab): No longer inline.
19504 * symtab.c (symtab_node::debug_symtab): Move definition here.
19506 2017-04-28 Richard Biener <rguenther@suse.de>
19508 * lto-streamer.h (LTO_major_version): Bump to 7.
19510 2017-04-28 Richard Biener <rguenther@suse.de>
19512 * tree-vrp.c (assert_info): New struct.
19513 (add_assert_info): New helper.
19514 (register_edge_assert_for_2): Refactor to add asserts to a vector
19516 (register_edge_assert_for_1): Likewise.
19517 (register_edge_assert_for): Likewise.
19518 (finish_register_edge_assert_for): New helper actually registering
19519 asserts where live on edge.
19520 (find_conditional_asserts): Adjust.
19521 (find_switch_asserts): Likewise.
19522 (evrp_dom_walker::try_find_new_range): Generalize.
19523 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
19525 2017-04-27 Marek Polacek <polacek@redhat.com>
19528 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
19529 arg10 and arg11 to itype.
19531 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
19533 * doc/extend.texi (Object Size Checking): Improve grammar.
19535 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
19538 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
19539 that the logic for permitting reciprocal estimates matches that
19542 2017-04-27 Jakub Jelinek <jakub@redhat.com>
19545 * tree.c (type_cache_hasher::equal): Only compare
19546 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
19547 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
19548 non-aggregate element types.
19549 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
19550 about the flag on ARRAY_TYPEs in the comment, formatting fix.
19552 2017-04-27 Richard Biener <rguenther@suse.de>
19554 PR middle-end/80533
19555 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
19556 stripping ARRAY_REFs from MEM_EXPR make sure we're not
19557 keeping a reference to a trailing array.
19559 2017-04-27 Richard Biener <rguenther@suse.de>
19561 PR middle-end/80539
19562 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
19563 being in loop-closed SSA form conservatively.
19564 (chrec_fold_multiply_poly_poly): Likewise.
19566 2017-04-27 Tamar Christina <tamar.christina@arm.com>
19568 PR middle-end/79665
19569 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
19570 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
19572 2017-04-27 Jakub Jelinek <jakub@redhat.com>
19575 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
19576 (aarch64_function_arg_alignment): Return unsigned int again, but still
19577 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
19578 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
19579 Don't emit -Wpsabi note.
19580 (aarch64_function_arg_boundary): Likewise.
19581 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
19584 2017-04-26 Nathan Sidwell <nathan@acm.org>
19586 * tree.h (crc32_unsigned_n): Declare.
19587 (crc32_unsigned, crc32_unsigned): Make inline.
19588 * tree.c (crc32_unsigned_bits): Replace with ...
19589 (crc32_unsigned_n): ... this.
19590 (crc32_unsigned, crc32_byte): Remove.
19591 (crc32_string): Remove unnecessary braces.
19593 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
19595 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
19596 * ipa-inline-analysis.c (MAX_TIME): Remove.
19597 (account_size_time): Use sreal for time.
19598 (dump_inline_summary): Update.
19599 (estimate_function_body_sizes): Update.
19600 (estimate_edge_size_and_time): Update.
19601 (estimate_calls_size_and_time): Update.
19602 (estimate_node_size_and_time): Update.
19603 (inline_merge_summary): Update.
19604 (inline_update_overall_summary): Update.
19605 (estimate_time_after_inlining): Update.
19606 (inline_read_section): Update.
19607 (inline_write_summary): Update.
19608 * ipa-inline.c (compute_uninlined_call_time): Update.
19609 (compute_inlined_call_time): Update.
19610 (recursive_inlining): Update.
19611 (inline_small_functions): Update.
19612 (dump_overall_stats): Update.
19613 * ipa-inline.h: Include sreal.h.
19614 (size_time_entry): Turn time to sreal.
19615 (inline_summary): Turn self_time nad time to sreal.
19617 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
19619 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
19621 (sreal::stream_out, sreal::stream_in): New.
19622 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
19624 2017-04-25 Jakub Jelinek <jakub@redhat.com>
19626 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
19629 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
19632 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
19634 (dimode_scalar_chain::compute_convert_gain): Ditto.
19635 (dimode_scalar_chain::make_vector_copies): Ditto.
19636 (dimode_scalar_chain::convert_reg): Ditto.
19637 (dimode_scalar_chain::convert_insn): Ditto.
19638 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
19639 (VI248_AVX512BW_1): New mode iterator.
19640 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
19641 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
19644 2017-04-25 Martin Sebor <msebor@redhat.com>
19646 PR tree-optimization/80497
19647 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
19648 constants are representable in HOST_WIDE_INT.
19649 (parse_directive): Ditto.
19651 2017-04-25 Martin Sebor <msebor@redhat.com>
19654 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
19655 (new_zero_array): Adjust signature.
19656 (dom_info::dom_init): Used unsigned rather that size_t.
19657 (dom_info::dom_info): Same.
19659 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19660 Jakub Jelinek <jakub@redhat.com>
19663 * config/arm/arm.c: Include gimple.h.
19664 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
19665 returns negative, increment ncrn only if it returned positive.
19666 (arm_needs_doubleword_align): Return int instead of bool,
19667 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
19668 members, but if there is any such non-FIELD_DECL
19669 > PARM_BOUNDARY aligned decl, return -1 instead of false.
19670 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
19671 returns negative, increment nregs only if it returned positive.
19672 (arm_setup_incoming_varargs): Likewise.
19673 (arm_function_arg_boundary): Emit -Wpsabi note if
19674 arm_needs_doubleword_align returns negative, return
19675 DOUBLEWORD_ALIGNMENT only if it returned positive.
19677 2017-04-25 Marek Polacek <polacek@redhat.com>
19680 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
19681 first argument to type.
19683 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
19686 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
19687 type checks to test for compatibility instead of equality.
19689 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19690 Jakub Jelinek <jakub@redhat.com>
19693 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
19695 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
19696 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
19697 the alignment computation, but return their maximum in warn_alignment.
19698 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
19699 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
19701 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
19702 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
19705 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19707 * config/arc/simdext.md (dmpyh): Fix typo.
19709 2017-04-25 Richard Biener <rguenther@suse.de>
19711 PR tree-optimization/80492
19712 * alias.c (compare_base_decls): Handle registers with asm
19713 specification conservatively.
19714 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
19715 compare_base_decls returning dont-know properly.
19717 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19719 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
19720 (legitimate_offset_address_p): New function.
19721 (arc_legitimate_address_p): Use above function.
19723 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19725 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
19727 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19729 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
19730 ACCH registers whenever they are available.
19732 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19734 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
19735 double regs fix when not used.
19737 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19739 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
19741 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
19742 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
19744 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19746 * config/arc/arc.c (arc_output_addsi): Check for h-register class
19747 when emitting short ADD instructions.
19749 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19751 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
19753 (cmpsi_cc_c_insn): Likewise.
19754 (cbranchsi4_scratch): Compute proper instruction length using
19755 compact_hreg_operand.
19756 * config/arc/predicates.md (compact_hreg_operand): New predicate.
19758 2017-04-25 Richard Biener <rguenther@suse.de>
19760 PR middle-end/80509
19761 * passes.c (pass_manager::pass_manager): Initialize
19762 m_name_to_pass_map.
19764 2017-04-25 Richard Biener <rguenther@suse.de>
19766 PR tree-optimization/79201
19767 * tree-ssa-sink.c (statement_sink_location): Handle calls.
19769 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19772 * config/s390/vector.md: Split MEM->GPR vector moves for
19773 non-s_operand addresses.
19775 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19778 * config/s390/predicates.md (reload_const_wide_int_operand): New
19780 * config/s390/s390.md ("movti"): Remove d/P alternative.
19781 ("movti_bigconst"): New pattern definition.
19783 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
19786 * s390-protos.h (s390_expand_cs_hqi): Removed.
19787 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
19788 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
19789 modes as well as CCZ1mode and CCZmode.
19790 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
19791 signature of s390_emit_compare_and_swap.
19792 (s390_expand_cs_hqi): Likewise, make static.
19793 (s390_expand_cs_tdsi): Generate an explicit compare before trying
19794 compare-and-swap, in some cases.
19795 (s390_expand_cs): Wrapper function.
19796 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
19798 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
19799 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
19800 patterns for small and large integers. Forbid symref memory operands.
19801 Move expander to s390.c. Require cc register.
19802 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
19803 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
19804 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
19805 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
19806 symref memory operands. Remove CC mode and call s390_match_ccmode
19808 ("atomic_exchange<mode>"): Allow and implement all integer modes.
19810 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
19812 * config/s390/s390.md (define_peephole2): New peephole to help
19813 combining the load-and-test pattern with volatile memory.
19815 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
19817 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
19818 with CCZmode for TARGET_Z196.
19820 2017-04-25 Jakub Jelinek <jakub@redhat.com>
19822 PR rtl-optimization/80501
19823 * combine.c (make_compound_operation_int): Set subreg_code to SET
19824 even for AND with mask of the sign bit of mode.
19826 PR rtl-optimization/80500
19827 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
19828 sum's initial value.
19830 2017-04-25 Julian Brown <julian@codesourcery.com>
19831 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19833 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
19835 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
19837 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
19839 2017-04-25 Julian Brown <julian@codesourcery.com>
19840 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19842 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
19843 (thunderx2t99_sha): New Reservation.
19845 2017-04-25 Julian Brown <julian@codesourcery.com>
19846 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19848 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
19849 type for 1-element load.
19851 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
19853 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
19855 2017-04-24 Martin Jambor <mjambor@suse.cz>
19857 PR tree-optimization/80293
19858 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
19859 char arrays not totally scalarizable if it is false.
19860 (analyze_all_variable_accesses): Pass correct value in the new
19861 parameter. Add a statistics counter.
19863 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
19865 PR middle-end/79931
19866 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
19868 2017-04-24 Richard Biener <rguenther@suse.de>
19870 PR tree-optimization/80494
19871 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
19872 out for complex types.
19874 2017-04-24 Richard Biener <rguenther@suse.de>
19876 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
19877 * tree-ssa-sccvn.c (print_scc): Print SCC size.
19878 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
19879 (DFS): Adjust and never fail.
19880 (sccvn_dom_walker::fail): Remove.
19881 (sccvn_dom_walker::before_dom_children): Adjust.
19882 (run_scc_vn): Likewise and never fail.
19883 * tree-ssa-pre.c (pass_pre::execute): Adjust.
19884 (pass_fre::execute): Likewise.
19886 2017-04-24 Richard Biener <rguenther@suse.de>
19888 PR tree-optimization/79725
19889 * tree-ssa-sink.c (statement_sink_location): Return whether
19890 failure reason was zero uses. Move that check later.
19891 (sink_code_in_bb): Deal with zero uses by removing the stmt
19894 2017-04-24 Richard Biener <rguenther@suse.de>
19897 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
19898 pointer-based references.
19900 2017-04-24 Richard Biener <rguenther@suse.de>
19903 * pass_manager.h (pass_manager::operator new): Remove.
19904 (pass_manager::operator delete): Likewise.
19905 * passes.c (pass_manager::operator new): Remove.
19906 (pass_manager::operator delete): Likewise.
19907 (pass_manager::pass_manager): Zero individual pass members.
19909 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
19912 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
19913 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
19914 Check "XEXP (src, 1)" operand here.
19915 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
19916 Check "XEXP (src, 1)" operand here.
19917 (dimode_scalar_chain::make_vector_copies): Detect count register
19918 of a shift instruction. Zero extend count register from QImode
19919 to DImode to satisfy vector shift pattern count operand predicate.
19920 Substitute vector shift count operand with a DImode copy.
19921 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
19924 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
19926 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
19927 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
19928 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
19929 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
19930 (UNSPEC_NOREX_MEM): Remove definition.
19932 2017-04-21 Richard Biener <rguenther@suse.de>
19934 PR tree-optimization/79547
19935 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
19936 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
19937 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
19938 without any constraints.
19940 2017-04-21 Richard Biener <rguenther@suse.de>
19942 PR tree-optimization/78847
19943 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
19945 2017-04-21 Richard Biener <rguenther@suse.de>
19947 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
19948 (build_distinct_type_copy): Likewise.
19949 (build_variant_type_copy): Likewise.
19950 * tree.c (build_qualified_type): Pass down mem-stat info.
19951 (build_distinct_type_copy): Likewise.
19952 (build_variant_type_copy): Likewise.
19954 2017-04-21 Richard Biener <rguenther@suse.de>
19956 PR tree-optimization/80237
19957 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
19959 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
19960 for a simplified result.
19962 2016-04-21 Richard Biener <rguenther@suse.de>
19964 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
19965 sth as strict as a simple_iv but a chrec without symbols and an
19966 operand defined in the loop we are peeling (and not some subloop).
19967 (propagate_constants_for_unrolling): Propagate all constants.
19969 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
19972 * config/i386/i386.c (print_reg): Remove assert for disalowed
19973 regno values, call output_operand_lossage instead.
19975 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
19978 * config/i386/constraints.md (Yc): New register constraint.
19979 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
19980 Use Yc constraint for alternative 2 of operand 0. Remove
19981 preferred_for_speed attribute.
19983 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
19985 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
19986 lastprivate clauses in SIMT case.
19988 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
19990 * doc/invoke.texi (-Wextra-semi): Document new warning option.
19992 2017-04-20 Richard Biener <rguenther@suse.de>
19994 PR tree-optimization/57796
19995 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
19996 as N scalar stores.
19997 (vect_model_load_cost): Cost gathers as N scalar loads.
19999 2017-04-20 Richard Biener <rguenther@suse.de>
20001 * ggc-page.c (ggc_allocated_p): Rename to ...
20002 (safe_lookup_page_table_entry): ... this and return the lookup
20004 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
20006 2017-04-20 Richard Biener <rguenther@suse.de>
20008 PR tree-optimization/80453
20009 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
20010 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
20011 from the conditions.
20012 (vn_phi_eq): Pass them down.
20013 (vn_phi_lookup): Record them.
20014 (vn_phi_insert): Likewise.
20016 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
20018 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
20019 uninitialized variable warning to avoid buffer overrun.
20021 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
20024 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
20025 is suppressed for '{ 0 }' in C.
20027 2017-04-20 Jakub Jelinek <jakub@redhat.com>
20029 * BASE-VER: Set to 8.0.0.
20031 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
20033 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
20034 priority .init_array and .fini_array section with SECTION_NOTYPE
20037 2017-04-20 Jakub Jelinek <jakub@redhat.com>
20039 PR middle-end/80423
20040 * tree.h (build_array_type): Add typeless_storage default argument.
20041 * tree.c (type_cache_hasher::equal): Also compare
20042 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
20043 (build_array_type): Add typeless_storage argument, set
20044 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
20046 (build_nonshared_array_type): Adjust build_array_type_1 caller.
20047 (build_array_type): Likewise. Add typeless_storage argument.
20049 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
20050 Jakub Jelinek <jakub@redhat.com>
20052 PR tree-optimization/80426
20053 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
20054 operation on symbolic operands, also compute the overflow for the
20055 invariant part when the operation degenerates into a negation.
20057 2017-04-19 Jakub Jelinek <jakub@redhat.com>
20060 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
20061 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
20064 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
20066 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
20069 * config/avr/avr.c (tree.h): Include it.
20070 (cgraph.h): Include it.
20071 (avr_encode_section_info): Don't warn for uninitialized progmem
20072 variable if it's just an alias.
20074 2017-04-19 Richard Biener <rguenther@suse.de>
20077 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
20078 when needed by AutoPGO.
20080 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
20083 * doc/lto.texi: Remove an extra 'that'.
20085 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
20087 PR rtl-optimization/80429
20088 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
20089 are only used in debug insns.
20091 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
20092 Vladimir Makarov <vmakarov@redhat.com>
20094 * config/sparc/predicates.md (input_operand): Add comment. Return
20095 true for any memory operand when LRA is in progress.
20096 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
20098 2017-04-18 Jeff Law <law@redhat.com>
20101 * mips.md ({return,simple_return}_internal): Do not overwrite
20104 2017-04-18 Jakub Jelinek <jakub@redhat.com>
20106 PR tree-optimization/80443
20107 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
20108 instead of adding 1, subtract -1 and similarly instead of subtracting
20111 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
20113 PR rtl-optimization/80357
20114 * haifa-sched.c (tmp_bitmap): New variable.
20115 (model_recompute): Handle duplicate use records.
20116 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
20117 (free_global_sched_pressure_data): Free it.
20119 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
20122 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
20123 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
20124 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
20125 instead of SYSTEM_HEADER_DIR.
20127 2017-04-18 Jeff Law <law@redhat.com>
20129 PR middle-end/80422
20130 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
20131 predecessors after walking up the insn chain.
20133 2017-04-18 Jakub Jelinek <jakub@redhat.com>
20136 * dwarf2out.c (modified_type_die): Try harder not to emit internal
20137 sizetype type into debug info.
20139 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
20142 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
20143 unneeded test for TARGET_UPPER_REGS_SF.
20144 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
20146 2017-04-18 Jakub Jelinek <jakub@redhat.com>
20149 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
20150 instead of gsi_after_labels.
20152 2017-04-18 Jeff Law <law@redhat.com>
20154 * regcprop.c (maybe_mode_change): Avoid creating copies of the
20158 2017-04-13 Jeff Law <law@redhat.com>
20159 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
20160 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
20162 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
20165 * config/avr/avr.c (intl.h): Include it.
20166 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
20168 2017-04-18 Martin Liska <mliska@suse.cz>
20170 PR gcov-profile/78783
20171 * gcov-tool.c (gcov_output_files): Validate that destination
20172 file is either removed by the tool or by a user.
20174 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
20175 Guy Benyei <guybe@mellanox.com>
20177 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
20178 block, and do not negate it, the stored id is already negative.
20180 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
20182 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
20184 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
20187 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
20188 masks of options that should be turned off if the VSX vector
20189 options are turned off.
20190 (OTHER_P8_VECTOR_MASKS): Likewise.
20191 (OTHER_VSX_VECTOR_MASKS): Likewise.
20192 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
20193 rs6000_disable_incompatible_switches to validate no type switches
20195 (rs6000_incompatible_switch): New function to disallow turning on
20196 other vector options if -mno-vsx, -mno-power8-vector, or
20197 -mno-power9-vector are specified.
20199 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
20201 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
20203 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
20205 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
20206 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
20207 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
20208 (ARG_POINTER_CFA_OFFSET): Likewise.
20210 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
20212 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
20213 conditions to take advantage of various optimizations.
20215 2017-04-13 Jeff Law <law@redhat.com>
20217 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
20218 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
20219 (zero_extendsidi2_dext): Likewise.
20221 2017-04-13 Jakub Jelinek <jakub@redhat.com>
20224 * fold-const.c (fold_ternary_loc): Revert
20225 use op0 instead of fold_convert_loc (loc, type, arg0) part of
20228 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
20230 PR rtl-optimization/80343
20231 * lra-remat.c (update_scratch_ops): Assign original hard reg to
20232 new scratch pseudo.
20234 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
20237 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
20238 to ubsan_encode_value.
20240 2017-04-13 Jeff Law <law@redhat.com>
20242 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
20243 appearing in DEBUG_INSNs.
20245 2017-04-13 Martin Liska <mliska@suse.cz>
20247 PR gcov-profile/80413
20248 * gcov-io.c (gcov_write_string): Copy to buffer just when
20249 allocated size is greater than zero.
20251 2017-04-13 Jakub Jelinek <jakub@redhat.com>
20254 * dwarf2out.c (decls_for_scope): Ignore declarations of
20255 current_function_decl in BLOCK_NONLOCALIZED_VARS.
20257 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
20260 * ipa-visibility.c (non_local_p): Fix typos.
20261 (localize_node): When localizing symbol in same comdat group,
20262 dissolve the group only when we know external symbols are going
20264 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
20266 2017-04-12 Jakub Jelinek <jakub@redhat.com>
20268 PR tree-optimization/79390
20269 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
20270 order does not result in usable sequence, retry with reversed operand
20276 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
20277 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
20278 op0 instead of fold_convert_loc (loc, type, arg0).
20280 2017-04-12 Jeff Law <law@redhat.com>
20282 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
20283 has a delay slot in the generated code.
20285 * config/cris/cris.md (cris_preferred_reload_class): Return
20286 GENNONACR_REGS rather than GENERAL_REGS.
20288 2017-04-12 Jakub Jelinek <jakub@redhat.com>
20291 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
20292 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
20293 signedness of the result type.
20295 2017-04-12 Richard Biener <rguenther@suse.de>
20296 Jeff Law <law@redhat.com>
20298 PR tree-optimization/80359
20299 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
20300 trim stores to TARGET_MEM_REFs.
20302 2017-04-12 Richard Biener <rguenther@suse.de>
20304 PR tree-optimization/79390
20305 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
20306 threading case even more.
20308 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
20311 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
20312 for quad_address_p for TImode, instead of just not indexed_address.
20314 2017-04-12 Richard Biener <rguenther@suse.de>
20315 Bernd Edlinger <bernd.edlinger@hotmail.de>
20317 PR middle-end/79671
20318 * alias.c (component_uses_parent_alias_set_from): Handle
20319 TYPE_TYPELESS_STORAGE.
20320 (get_alias_set): Likewise.
20321 * tree-core.h (tree_type_common): Add typeless_storage flag.
20322 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
20323 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
20324 for types containing members with TYPE_TYPELESS_STORAGE.
20325 (place_field): Likewise.
20326 (layout_type): Likewise for ARRAY_TYPE.
20327 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
20328 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
20329 TYPE_TYPELESS_STORAGE.
20330 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
20332 2017-04-12 Jakub Jelinek <jakub@redhat.com>
20335 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
20336 first argument to type.
20338 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20342 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
20343 CONST0_RTX (mode) rather than const0_rtx where appropriate.
20344 (rs6000_expand_binop_builtin): Likewise.
20345 (rs6000_expand_ternop_builtin): Likewise; also add missing
20346 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
20347 vshasigma built-ins.
20348 * doc/extend.texi: Document that vec_xxpermdi's third argument
20349 must be a constant.
20351 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
20353 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
20354 Use shift_const cost parameter when calculating gain of STV shifts.
20356 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
20358 PR rtl-optimization/70478
20359 * lra-constraints.c (process_alt_operands): Check memory for
20360 disfavoring memory insn operand.
20362 2017-04-11 Jakub Jelinek <jakub@redhat.com>
20364 PR middle-end/80100
20365 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
20366 left shift in unsigned HOST_WIDE_INT type.
20368 PR rtl-optimization/80385
20369 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
20370 (not (neg X)) into (plus X -1) for complex or non-integral modes.
20373 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
20374 if they have any depend clauses.
20376 2017-04-11 Martin Liska <mliska@suse.cz>
20379 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
20380 * ipa-split.c (split_function): Create a local comdat symbol
20381 if caller is in a comdat group.
20383 2017-04-11 Martin Liska <mliska@suse.cz>
20386 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
20389 2017-04-11 Martin Sebor <msebor@redhat.com>
20391 PR middle-end/80364
20392 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
20393 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
20395 (directive::set_width, directive::set_precision, format_character):
20397 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
20400 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
20403 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
20404 conflict, set target->arch_name instead of target->cpu_name.
20406 2017-04-11 Richard Biener <rguenther@suse.de>
20408 PR tree-optimization/80374
20409 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
20410 build_zero_cst, remove fold_convertible_p check again.
20412 2017-04-11 Martin Liska <mliska@suse.cz>
20415 * ubsan.c (instrument_object_size): Do not instrument register
20418 2017-04-11 Jakub Jelinek <jakub@redhat.com>
20421 * config/i386/i386-builtin-types.def
20422 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
20423 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
20424 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
20425 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
20426 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
20427 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
20428 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
20429 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
20430 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
20431 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
20432 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
20433 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
20434 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
20435 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
20436 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
20437 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
20438 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
20439 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
20440 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
20441 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
20442 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
20443 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
20444 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
20445 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
20446 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
20447 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
20448 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
20449 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
20450 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
20451 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
20452 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
20453 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
20454 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
20455 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
20456 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
20457 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
20458 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
20459 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
20460 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
20461 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
20462 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
20463 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
20464 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
20465 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
20466 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
20467 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
20469 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
20470 flag to second_arg_count, handle 4 argument function type _COUNT
20471 aliases, handle second_arg_count on second argument rather than last.
20473 2017-04-10 Jeff Law <law@redhat.com>
20475 PR tree-optimization/80374
20476 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
20477 record anything if we can not convert integer_zero_node to the
20480 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
20483 * config/rs6000/rs6000.c (rs6000_option_override_internal):
20484 Enhance special handling given to the TARGET_P9_MINMAX option in
20485 relation to certain other options.
20487 2017-04-10 Bin Cheng <bin.cheng@arm.com>
20489 PR tree-optimization/80153
20490 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
20491 remove POINTER_PLUS_EXPR's base part directly, rather than through
20494 2017-04-10 Richard Biener <rguenther@suse.de>
20495 Bin Cheng <bin.cheng@arm.com>
20497 PR tree-optimization/80153
20498 * tree-affine.c (aff_combination_to_tree): Get base pointer from
20499 the first element of pointer type aff_tree. Build result expr in
20501 (add_elt_to_tree): Convert to type unconditionally. Remove other
20502 fold_convert calls.
20503 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
20504 (rewrite_use_nonlinear_expr): Check invariant using iv information.
20506 2017-04-10 Richard Biener <rguenther@suse.de>
20508 * tree-ssa-structalias.c (find_func_aliases): Properly handle
20511 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
20513 PR rtl-optimization/70478
20514 * lra-constraints.c (curr_small_class_check): New.
20515 (update_and_check_small_class_inputs): New.
20516 (process_alt_operands): Update curr_small_class_check. Disfavor
20517 alternative insn memory operands. Check available regs for small
20520 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
20523 * config/mips/mips.opt (-mvirt): Update description.
20524 * doc/invoke.texi (-mvirt): Likewise.
20526 2017-04-10 Richard Biener <rguenther@suse.de>
20528 PR middle-end/80362
20529 * fold-const.c (fold_binary_loc): Look at unstripped ops when
20530 looking for NEGATE_EXPR in -A / -B to A / B folding.
20532 2017-04-10 Martin Liska <mliska@suse.cz>
20534 PR gcov-profile/80224
20535 * gcov.c (print_usage): Fix usage string.
20536 (get_gcov_intermediate_filename): Remove.
20537 (output_gcov_file): Use both for normal and intermediate format.
20538 (generate_results): Do not initialize special file for
20539 intermediate format.
20541 2017-04-10 Richard Biener <rguenther@suse.de>
20543 PR tree-optimization/80304
20544 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
20547 2017-04-10 Nathan Sidwell <nathan@acm.org>
20550 * config/rs6000/rs6000.c (rs6000_vector_type): New.
20551 (rs6000_init_builtins): Use it.
20553 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20555 * config/arm/arm.md (<mrc>): Add mode to SET source.
20556 (<mrrc>): Likewise.
20558 2017-04-10 Richard Biener <rguenther@suse.de>
20560 PR middle-end/80344
20561 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
20563 2017-04-10 Jakub Jelinek <jakub@redhat.com>
20566 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
20567 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
20568 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
20569 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
20570 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
20571 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
20572 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
20573 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
20574 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
20575 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
20576 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
20577 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
20578 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
20579 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
20580 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
20581 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
20582 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
20583 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
20584 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
20585 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
20586 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
20587 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
20588 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
20590 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
20592 PR rtl-optimization/70478
20593 * lra-constraints.c: Reverse the last patch.
20595 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
20597 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
20598 Add comment for WCHAR_T.
20600 2017-04-08 Martin Liska <mliska@suse.cz>
20603 2017-04-07 Martin Liska <mliska@suse.cz>
20606 * ipa-split.c (split_function): Add function part to a same comdat
20609 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20612 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
20614 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
20616 * rs6000/rs6000.c (vec_load_pendulum): Rename...
20617 (vec_pairing): ...to this.
20618 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
20619 (rs6000_sched_init): Adjust for name change.
20620 (struct rs6000_sched_context): Likewise.
20621 (rs6000_init_sched_context): Likewise.
20622 (rs6000_set_sched_context): Likewise.
20624 2017-04-07 Jakub Jelinek <jakub@redhat.com>
20630 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
20632 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
20633 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
20634 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
20636 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
20638 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
20640 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
20642 PR rtl-optimization/70703
20643 * ira-color.c (update_conflict_hard_regno_costs): Use
20644 int64_t instead of HOST_WIDE_INT.
20646 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
20648 PR rtl-optimization/70478
20649 * lra-constraints.c (process_alt_operands): Disfavor alternative
20650 insn memory operands.
20652 2017-04-07 Jeff Law <law@redhat.com>
20654 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
20655 CALL and NOTE_INSN_CALL_ARG_LOCATION.
20657 2017-04-07 Martin Liska <mliska@suse.cz>
20660 * config/aarch64/aarch64.c (aarch64_process_target_attr):
20661 Show error message instead of an ICE.
20663 2017-04-07 Martin Liska <mliska@suse.cz>
20666 * ipa-split.c (split_function): Add function part to a same comdat
20669 2017-04-07 Richard Biener <rguenther@suse.de>
20671 PR middle-end/80341
20672 * tree.c (get_unwidened): Also handle ! for_type case for
20674 * convert.c (do_narrow): Split out from ...
20675 (convert_to_integer_1): ... here. Do not pass final truncation
20676 type to get_unwidened for TRUNC_DIV_EXPR.
20678 2017-04-07 Richard Biener <rguenther@suse.de>
20680 * tree-affine.c (wide_int_ext_for_comb): Take type rather
20682 (aff_combination_const): Adjust.
20683 (aff_combination_scale): Likewise.
20684 (aff_combination_add_elt): Likewise.
20685 (aff_combination_add_cst): Likewise.
20686 (aff_combination_convert): Likewise.
20687 (add_elt_to_tree): Likewise. Remove unused argument.
20688 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
20690 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
20692 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
20694 * config/arm/arm.c (arm_default_short_enums): Use
20695 ARM_DEFAULT_SHORT_ENUMS.
20696 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
20698 2017-04-06 Jakub Jelinek <jakub@redhat.com>
20701 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
20702 members with redundant out-of-class redeclaration.
20704 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
20707 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
20708 * config/i386/i386.md (*zero_extendsidi2):
20709 Add (?*x,*x) and (?*v,*v) alternatives.
20711 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
20714 * config/i386/i386.c (ix86_expand_builtin)
20715 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
20716 mode from insn data. Convert operands to insn operand mode.
20717 Copy operands that don't satisfy insn predicate to a register.
20719 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
20721 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
20724 2017-04-06 Richard Biener <rguenther@suse.de>
20726 PR tree-optimization/80334
20727 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
20728 preserve alignment of accesses.
20730 2017-04-06 Richard Biener <rguenther@suse.de>
20732 PR tree-optimization/80262
20733 * tree-sra.c (build_ref_for_offset): Preserve address-space
20735 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
20736 Drop useless address-space information on MEM_REF offsets.
20738 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
20740 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
20742 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
20744 PR rtl-optimization/70703
20745 * ira-color.c (update_conflict_hard_regno_costs): Use
20746 HOST_WIDE_INT instead of long.
20748 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
20751 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
20752 not defined for x86_64 target. Add -mmmx target option when __SSE2__
20754 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
20755 for x86_64 target. Handle -m3dnowa option.
20757 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
20759 PR rtl-optimization/70703
20760 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
20761 (update_conflict_hard_regno_costs): Use long instead of unsigned
20762 arithmetic for cost calculation.
20764 2017-04-05 Jakub Jelinek <jakub@redhat.com>
20765 Bernd Edlinger <bernd.edlinger@hotmail.de>
20768 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
20771 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
20774 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
20775 ptr_mode with Pmode throughout.
20776 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
20777 into probe_stack_range and use DImode.
20779 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
20782 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
20783 call_eh_return is true.
20785 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20787 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
20788 Initialize last_match_fntype_index.
20790 2017-04-05 Jakub Jelinek <jakub@redhat.com>
20793 * tree-nvr.c: Include internal-fn.h.
20794 (pass_return_slot::execute): Ignore internal calls without
20797 2017-04-04 Jakub Jelinek <jakub@redhat.com>
20798 Richard Biener <rguenther@suse.de>
20801 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
20802 captures used multiple times, except for the last use.
20803 * generic-match-head.c: Include gimplify.h.
20805 2017-04-04 Jakub Jelinek <jakub@redhat.com>
20807 PR tree-optimization/79390
20808 * target.h (struct noce_if_info): Declare.
20809 * targhooks.h (default_noce_conversion_profitable_p): Declare.
20810 * target.def (noce_conversion_profitable_p): New target hook.
20811 * ifcvt.h (struct noce_if_info): New type, moved from ...
20812 * ifcvt.c (struct noce_if_info): ... here.
20813 (noce_conversion_profitable_p): Renamed to ...
20814 (default_noce_conversion_profitable_p): ... this. No longer
20816 (noce_try_store_flag_constants, noce_try_addcc,
20817 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
20818 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
20819 instead of noce_conversion_profitable_p.
20820 * config/i386/i386.c: Include ifcvt.h.
20821 (ix86_option_override_internal): Don't override
20822 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
20823 (ix86_noce_conversion_profitable_p): New function.
20824 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
20825 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
20826 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
20827 * doc/tm.texi: Regenerated.
20829 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20831 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
20834 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
20837 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
20838 instructions for small multiply cores.
20840 2017-04-04 Jeff Law <law@redhat.com>
20842 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
20844 (mips_expand_vec_perm_const): Initialize elements in orig_perm
20845 that are not set by the loop over the elements.
20847 2017-04-04 Jakub Jelinek <jakub@redhat.com>
20850 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
20851 int mode, convert_modes it to mode as unsigned, otherwise use
20852 lowpart_subreg to mode rather than SImode.
20853 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
20854 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
20855 Use DImode instead of SImode for the shift count operand.
20856 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
20859 2017-04-04 Richard Biener <rguenther@suse.de>
20861 PR middle-end/80281
20862 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
20863 arithmetic done for the negate or the plus. Simplify.
20864 (A - (-B) -> A + B): Likewise.
20865 * fold-const.c (split_tree): Make sure to not negate pointers.
20867 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
20869 PR rtl-optimization/60818
20870 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
20871 a compare of comparisons with the thing compared if this results
20872 in a different machine mode.
20874 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
20876 * alias.c (base_alias_check): Fix typo in comment.
20877 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
20878 * cgraphunit.c (symbol_table::compile): Likewise.
20879 * collect2.c (maybe_run_lto_and_relink): Likewise.
20880 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
20881 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
20882 * config/avr/avr.c (avr_map_op_t): Likewise.
20883 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
20884 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
20885 * config/epiphany/epiphany.md (movcc): Likewise.
20886 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
20887 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
20889 * config/mips/mips.c (mips_save_restore_reg): Likewise.
20890 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
20891 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
20892 * config/sh/sh.c (sh_rtx_costs): Likewise.
20893 * fold-const.c (fold_truth_andor): Likewise.
20894 * genautomata.c (collapse_flag): Likewise.
20895 * gengtype.h (struct type::u::s): Likewise.
20896 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
20897 * input.c (FORMAT_AMOUNT): Likewise.
20898 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
20899 (known_aggs_to_agg_replacement_list): Likewise.
20900 * ipa-inline-analysis.c: Likewise.
20901 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
20902 * ipa-polymorphic-call.c
20903 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
20904 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
20905 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
20907 * modulo-sched.c (apply_reg_moves): Likewise.
20908 * omp-expand.c (build_omp_regions_1): Likewise.
20909 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
20910 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
20911 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
20912 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
20913 * value-prof.c: Likewise.
20914 * var-tracking.c (val_reset): Likewise.
20916 2017-04-03 Richard Biener <rguenther@suse.de>
20918 PR tree-optimization/80275
20919 * fold-const.c (split_address_to_core_and_offset): Handle
20922 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
20924 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
20925 descriptors is at least equal to that of functions.
20927 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
20929 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
20931 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
20934 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
20935 (mov<IMOD4:mode>): New expander.
20936 (*mov<IMOD4:mode>_internal): New insn and split pattern.
20938 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
20940 PR rtl-optimization/79405
20941 * fwprop.c (propagations_left): New variable.
20942 (forward_propagate_into): Decrement it.
20943 (fwprop_init): Initialize it.
20944 (fw_prop): If the variable has reached zero, stop propagating.
20945 (fwprop_addr): Ditto.
20947 2017-03-31 Jakub Jelinek <jakub@redhat.com>
20950 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
20951 a FUNCTION_DECL, pass it as decl instead of origin to
20954 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
20956 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
20959 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
20962 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
20963 TARGET_VSX_SMALL_INTEGER.
20965 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20967 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
20968 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
20970 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
20972 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
20973 extraction from odd-numbered MSA register.
20975 2017-03-31 Jakub Jelinek <jakub@redhat.com>
20977 PR middle-end/80173
20978 * expmed.c (store_bit_field_1): Don't attempt to create
20979 a word subreg out of hard registers wider than word if they
20980 have HARD_REGNO_NREGS of 1 for their mode.
20982 PR middle-end/80163
20983 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
20984 conversions to integer types wider than word and pointer.
20987 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
20988 (rtx_equal_for_cselib_p): Pass 0 to it.
20989 * cselib.c (cselib_hasher::equal): Likewise.
20990 (rtx_equal_for_cselib_1): Add depth argument. If depth
20991 is 128, don't look up VALUE locs and punt. Increment
20992 depth in recursive calls when walking VALUE locs.
20994 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
20996 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
20997 (make_gcov_file_name): Use the canonical path name for generating
20999 (read_line): Fix handling of files with ascii null bytes.
21001 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
21003 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
21004 to initialise a vector register instead
21005 of using a const_int.
21007 2017-03-30 Jakub Jelinek <jakub@redhat.com>
21009 PR translation/80189
21010 * gimplify.c (omp_default_clause): Use %qs instead of %s in
21011 diagnostic messages.
21013 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
21016 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
21017 (dfp_diex_<mode>): Update mode of operand 1.
21018 * doc/extend.texi (dxex, dxexq): Document change to return type.
21019 (diex, diexq): Document change to argument type.
21021 2017-03-30 Martin Jambor <mjambor@suse.cz>
21024 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
21025 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
21026 it reflects the signature changes performed at the callee side.
21027 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
21028 to cgraph_build_function_type_skip_args.
21029 (build_function_decl_skip_args): Adjust call to the above function.
21031 2017-03-30 Jakub Jelinek <jakub@redhat.com>
21034 * config/i386/sse.md
21035 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
21036 register as dest whenever it is a MEM not rtx_equal_p to the
21037 corresponding dup operand, and when forcing into reg move the
21038 reg into the memory afterwards.
21039 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
21040 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
21041 for the force_reg mode.
21042 (avx512vl_vextractf128<mode>): Use register as dest either
21043 always when a MEM, or when it is a MEM not rtx_equal_p to the
21044 corresponding dup operand, or even not when it is a CONST_VECTOR
21045 depending on the mode and lo vs. hi.
21046 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
21048 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
21049 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
21050 Likewise. Require that operands[2] is even.
21051 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
21052 Remove extraneous parens. Require that operands[2] is a multiple
21054 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
21055 operands[0] is a MEM if <mask_applied>, the predicates/constraints
21056 disallow memory then.
21058 2017-03-30 Richard Biener <rguenther@suse.de>
21060 PR tree-optimization/77498
21061 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
21062 to non-constants over backedges.
21064 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
21066 PR rtl-optimization/80233
21067 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
21068 as last_combined_insn. Do not test for BARRIER_P separately.
21070 2017-03-29 Andreas Schwab <schwab@suse.de>
21073 * calls.c (prepare_call_address): Convert funexp to Pmode before
21074 copying to temp reg.
21076 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21078 PR tree-optimization/80158
21079 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
21080 Handle possible future case of more than one alternate
21082 (replace_rhs_if_not_dup): Likewise.
21083 (replace_one_candidate): Likewise.
21085 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
21087 PR rtl-optimization/80193
21088 * ira.c (ira): Do not check allocation for LRA.
21090 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
21092 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
21093 (nvptx_output_simt_exit): Declare.
21094 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
21095 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
21096 (init_softstack_frame): Move initialization of crtl->is_leaf to...
21097 (nvptx_declare_function_name): ...here. Emit declaration of local
21098 memory space buffer for omp_simt_enter insn.
21099 (nvptx_output_unisimt_switch): New.
21100 (nvptx_output_softstack_switch): New.
21101 (nvptx_output_simt_enter): New.
21102 (nvptx_output_simt_exit): New.
21103 * config/nvptx/nvptx.h (struct machine_function): New fields
21104 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
21105 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
21106 (UNSPECV_SIMT_EXIT): Ditto.
21107 (omp_simt_enter_insn): New insn.
21108 (omp_simt_enter): New expansion.
21109 (omp_simt_exit): New insn.
21110 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
21112 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
21113 (expand_GOMP_SIMT_ENTER_ALLOC): New.
21114 (expand_GOMP_SIMT_EXIT): New.
21115 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
21116 (GOMP_SIMT_ENTER_ALLOC): Ditto.
21117 (GOMP_SIMT_EXIT): Ditto.
21118 * target-insns.def (omp_simt_enter): New insn.
21119 (omp_simt_exit): Ditto.
21120 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
21122 (lower_rec_simd_input_clauses): Implement SIMT privatization.
21123 (lower_rec_input_clauses): Likewise.
21124 (lower_lastprivate_clauses): Handle SIMT privatization.
21126 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
21127 (ompdevlow_adjust_simt_enter): New.
21128 (find_simtpriv_var_op): New.
21129 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
21130 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
21132 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
21133 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
21134 (copy_decl_for_dup_finish): Ditto.
21136 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
21138 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
21141 * config/i386/i386.c (ix86_option_override_internal): Always
21142 allow -mpreferred-stack-boundary=3 for 64-bit targets.
21144 2017-03-28 Bin Cheng <bin.cheng@arm.com>
21146 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
21148 2017-03-28 Bin Cheng <bin.cheng@arm.com>
21150 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
21151 mark new edge's irreducible flag accordign to it.
21152 (vect_do_peeling): Check loop preheader edge's irreducible flag
21153 and pass it to function slpeel_add_loop_guard.
21155 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
21157 PR tree-optimization/80218
21158 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
21159 Update block frequencies and counts.
21161 2017-03-28 Richard Biener <rguenther@suse.de>
21163 PR tree-optimization/78644
21164 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
21165 of a simplification result we may not use it at all.
21167 2017-03-28 Richard Biener <rguenther@suse.de>
21170 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
21171 without arguments, generate default definition of a SSA name.
21173 2017-03-28 Richard Biener <rguenther@suse.de>
21175 PR middle-end/80222
21176 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
21177 TYPE_REF_CAN_ALIAS_ALL references.
21178 * fold-const.c (fold_indirect_ref_1): Likewise.
21180 2017-03-28 Martin Liska <mliska@suse.cz>
21183 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
21184 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
21186 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
21187 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
21189 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
21190 (EXTRA_SPECS): Define.
21191 (SUBTARGET_EXTRA_SPECS): Likewise.
21192 (SUBTARGET_CPP_SPEC): Likewise.
21193 * config/arc/elf.h (EXTRA_SPECS): Renamed to
21194 SUBTARGET_EXTRA_SPECS.
21195 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
21197 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
21199 * config/arc/simdext.md (vst64_insn): Update pattern.
21200 (vld32wh_insn): Likewise.
21201 (vld32wl_insn): Likewise.
21202 (vld64_insn): Likewise.
21203 (vld32_insn): Likewise.
21205 2017-03-28 Marek Polacek <polacek@redhat.com>
21208 * fold-const.c (fold_comparison): Use protected_set_expr_location
21209 instead of SET_EXPR_LOCATION.
21211 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
21213 * tree.c (add_expr): Avoid name lookup warning.
21215 2017-03-27 Jeff Law <law@redhat.com>
21217 PR tree-optimization/80216
21218 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
21219 function name. Limit recursion depth.
21220 (record_temporary_equivalences): Corresponding changes.
21222 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
21224 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
21227 2017-03-27 Jakub Jelinek <jakub@redhat.com>
21230 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
21232 * cfgcleanup.c (reg_note_cfa_p): New array.
21233 (insns_have_identical_cfa_notes): New function.
21234 (old_insns_match_p): Don't cross-jump in between /f
21235 and non-/f instructions. If both i1 and i2 are frame related,
21236 verify all CFA notes, their order and content.
21238 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
21241 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
21242 HImode and SImode with zero extend to DImode to one insn.
21243 (bswap<mode>2_extenddi): Likewise.
21244 (bswapsi2_extenddi): Likewise.
21245 (bswaphi2_extendsi): Likewise.
21246 (bswaphi2): Combine bswap HImode and SImode into one insn.
21247 Separate memory insns from swapping register.
21248 (bswapsi2): Likewise.
21249 (bswap<mode>2): Likewise.
21250 (bswaphi2_internal): Delete, no longer used.
21251 (bswapsi2_internal): Likewise.
21252 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
21253 store, and gpr<-gpr swap insns.
21254 (bswap<mode>2_store): Likewise.
21255 (bswaphi2_reg): Register only splitter, combine with the splitter.
21256 (bswaphi2 splitter): Likewise.
21257 (bswapsi2_reg): Likewise.
21258 (bswapsi2 splitter): Likewise.
21259 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
21260 the insns into load, store, and register/register insns.
21261 (bswapdi2_ldbrx): Likewise.
21262 (bswapdi2_load): Likewise.
21263 (bswapdi2_store): Likewise.
21264 (bswapdi2_reg): Likewise.
21266 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
21268 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
21269 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
21271 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
21274 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
21276 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
21277 special handling for target option conflicts between dform
21278 options (-mpower9-dform, -mpower9-dform-vector,
21279 -mpower9-dform-scalar) and -mno-direct-move.
21281 2017-03-27 Richard Biener <rguenther@suse.de>
21283 PR tree-optimization/80181
21284 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
21286 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
21288 * config/arc/predicates.md (move_double_src_operand): Replace the
21289 call to move_double_src_operand with a call to address_operand.
21291 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
21293 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
21294 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
21295 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
21297 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
21299 * config/arc/predicates.md (long_immediate_loadstore_operand):
21300 Consider scaled addresses cases.
21302 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
21304 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
21305 restored when in interrupt.
21306 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
21307 doesn't have delay slot.
21309 2017-03-27 Richard Biener <rguenther@suse.de>
21312 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
21313 inlined thunk clones.
21315 2017-03-27 Jakub Jelinek <jakub@redhat.com>
21318 * asan.c (instrument_derefs): Copy over last operand from
21319 original COMPONENT_REF to the new COMPONENT_REF with
21320 DECL_BIT_FIELD_REPRESENTATIVE.
21321 * ubsan.c (instrument_object_size): Likewise.
21323 2017-03-27 Richard Biener <rguenther@suse.de>
21325 PR tree-optimization/80170
21326 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
21327 sure DR/SCEV didnt fold in constants we do not see when looking
21328 at the reference base alignment.
21330 2017-03-27 Richard Biener <rguenther@suse.de>
21332 PR middle-end/80171
21333 * gimple-fold.c (fold_ctor_reference): Properly guard against
21334 NULL return value from canonicalize_constructor_val.
21336 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
21339 * config/i386/i386.c (ix86_expand_builtin)
21340 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
21341 flags reg setting and flags reg using instructions.
21342 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
21343 clobbering instructions to zero extend op2.
21345 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
21347 * doc/install.texi (Configuration) <--with-aix-soname>:
21348 Update link to AIX ld.
21350 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
21352 PR rtl-optimization/80160
21353 PR rtl-optimization/80159
21354 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
21355 reg_alternate_class into account.
21357 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
21360 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
21361 to consider in curr_insn_transform.
21363 2017-03-24 Jakub Jelinek <jakub@redhat.com>
21365 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
21366 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
21367 and emit_mode_inner.
21369 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21371 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
21372 argument to the overloaded builtin variants. Use the new flag to
21373 deprecate certain builtin variants.
21374 * config/s390/s390-builtin-types.def: Add new builtin types.
21375 * config/s390/s390-builtins.h: Support new flags field for
21376 overloaded builtins.
21377 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
21378 (s390_macro_to_expand): Enable vector float data type.
21379 (s390_cpu_cpp_builtins_internal): Indicate support of the new
21380 builtins by incrementing the __VEC__ version number.
21381 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
21383 (s390_resolve_overloaded_builtin): Emit error messages depending
21384 on the builtin flags.
21385 * config/s390/s390.c (s390_expand_builtin): Support additional
21386 flags argument. Change error message to match the messages
21387 emitted in s390-c.c.
21388 * config/s390/s390.md: New UNSPEC_* constants.
21389 (op_type): Add new instruction types.
21390 * config/s390/vecintrin.h: Add new builtins and test data class
21392 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
21393 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
21394 (VEC_INEXACT, VEC_NOINEXACT): New constants.
21395 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
21396 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
21397 ("vec_mergel<mode>"): V_HW -> VEC_HW.
21399 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
21400 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
21401 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
21402 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
21404 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
21405 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
21406 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
21407 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
21409 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
21410 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
21411 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
21412 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
21413 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
21414 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
21415 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
21417 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
21418 ("vec_scatter_element<V_HW_4:mode>_DI")
21419 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
21420 ("vec_fpint<mode>", "vflls")
21421 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
21422 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
21423 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
21424 ("*vec_cmphe<mode>_cc"): ... these.
21426 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
21427 mode constant instead of magic value.
21429 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21431 * config/s390/s390.c (s390_expand_vec_compare): Support other
21432 vector floating point modes than just V2DF.
21433 (s390_expand_vcond): Likewise.
21434 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
21435 (s390_cannot_change_mode_class): Prevent mode changes between TF
21436 and V1TF in vector registers.
21437 * config/s390/s390.md (DF, SF): New mode attributes.
21438 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
21439 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
21440 SFmode support for VRs.
21441 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
21443 (VFT, VF_HW): New mode iterators.
21444 (vw, sdx): New mode attributes.
21445 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
21446 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
21447 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
21448 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
21449 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
21450 also the new vector floating point modes. Renaming to ...
21452 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
21453 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
21454 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
21455 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
21456 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
21457 ("vec_unordered<mode>"): ... these.
21459 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
21460 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
21461 ("*vec_extendv2df"): New insn definitions.
21463 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21465 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
21466 ("mulditi3_2", "*muldi3_sign"): New patterns.
21467 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
21468 rename the pattern definition.
21470 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21472 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
21474 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
21476 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21478 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
21479 instruction if possible.
21480 * config/s390/vector.md (vec_halfnumelts): New mode
21482 ("*vec_vllezlf<mode>"): New pattern.
21484 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21486 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
21487 ("popcountv4si2", "popcountv2di2"): Rename to ...
21488 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
21489 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
21491 ("popcount<mode>2_vxe"): New pattern.
21493 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21495 * common/config/s390/s390-common.c (processor_flags_table): Add
21497 * config.gcc: Add arch12.
21498 * config/s390/driver-native.c (s390_host_detect_local_cpu):
21499 Default to arch12 for unknown CPU model numbers.
21500 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
21501 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
21502 PROCESSOR_max sanity check.
21503 * config/s390/s390-opts.h (enum processor_type): Add
21505 * config/s390/s390.c (processor_table): Add arch12.
21506 (s390_expand_builtin): Add check for B_VXE flag.
21507 (s390_issue_rate): Add PROCESSOR_ARCH12.
21508 (s390_get_sched_attrmask): Likewise.
21509 (s390_get_unit_mask): Likewise.
21510 (s390_sched_score): Enable z13 scheduling for arch12.
21511 (s390_sched_reorder): Likewise.
21512 (s390_sched_variable_issue): Likewise.
21513 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
21515 (s390_tune_attr): Use z13 scheduling also for arch12.
21516 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
21517 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
21518 (TARGET_VXE_P): New macros.
21519 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
21520 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
21521 * config/s390/s390.opt: Add arch12 as processor_type.
21523 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21525 * config/s390/s390.md
21526 ("fixuns_truncdddi2", "fixuns_trunctddi2")
21527 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
21528 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
21530 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
21531 Rename expanders to ...
21533 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
21534 ("fixuns_truncdddi2_emu"): ... these.
21536 ("fixuns_trunc<mode>si2_emu"): New expander.
21538 ("*fixuns_truncdfdi2_z13"): Rename to ...
21539 ("*fixuns_truncdfdi2_vx"): ... this.
21541 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21543 * config/s390/2964.md: Remove the single element vector compare
21544 instructions which are no longer used.
21545 * config/s390/s390.c (s390_select_ccmode): Remove handling of
21547 (s390_canonicalize_comparison): Remove handling of DFmode
21549 (s390_expand_vec_compare_scalar): Remove function.
21550 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
21551 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
21553 ("*cmp<mode>_ccs"): Add wfcdb instruction.
21555 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21557 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
21559 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
21560 will anyway by matched by mov<mode>_64dfp.
21562 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21564 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
21565 vlef/vstef. Add missing operand to vleif.
21567 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21569 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
21570 pair for all vector types with 64 bit elements.
21571 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
21572 * config/s390/vector.md (V_HW_64): ... here.
21573 (V_128_NOSINGLE): New mode iterator.
21574 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
21575 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
21576 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
21577 ("*vec_load_pairv2di"): Change to ...
21578 ("*vec_load_pair<mode>"): ... this one.
21580 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21582 * config/s390/constraints.md: Add comments.
21583 (jKK): Reject element sizes > 8 bytes.
21584 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
21586 * config/s390/s390.md: Add the s_operand checks formerly in
21587 s390_split_ok_p to various splitters where they are still
21589 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
21590 for 128 bit vectors. Plus two splitters.
21592 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21594 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
21597 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21600 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
21601 error if the boundary argument is not constant.
21603 2017-03-24 Jakub Jelinek <jakub@redhat.com>
21605 PR rtl-optimization/80112
21606 * loop-doloop.c (doloop_condition_get): Don't check condition
21607 if cmp isn't SET with IF_THEN_ELSE src.
21609 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21611 PR tree-optimization/80158
21612 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
21613 replacing a candidate statement, also replace it for the
21614 candidate's alternate interpretation.
21615 (replace_rhs_if_not_dup): Likewise.
21616 (replace_one_candidate): Likewise.
21618 2017-03-24 Richard Biener <rguenther@suse.de>
21620 PR tree-optimization/80167
21621 * graphite-isl-ast-to-gimple.c
21622 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
21624 (translate_isl_ast_to_gimple::get_rename): Likewise.
21626 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
21628 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
21629 handling of certain combinations of target options, including the
21630 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
21631 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
21633 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21636 * config/arm/arm.md (*load_multiple): Add reload_completed to
21637 matching condition.
21639 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21640 Richard Biener <rguenth@suse.de>
21642 PR tree-optimization/79908
21643 PR tree-optimization/80136
21644 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
21645 been cast away, gimplify_and_add suffices.
21647 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
21649 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
21651 2017-03-23 Richard Biener <rguenther@suse.de>
21653 PR tree-optimization/80032
21654 * gimplify.c (gimple_push_cleanup): Forced unconditional
21655 cleanups still have to go to the conditional_cleanups
21658 2017-03-22 Jakub Jelinek <jakub@redhat.com>
21660 PR tree-optimization/80072
21661 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
21663 (next_operand_entry_id): Change type to unsigned int.
21664 (sort_by_operand_rank): Make sure to return the right return value
21665 even if unsigned fields are bigger than INT_MAX.
21666 (struct oecount): Change cnt and id type to unsigned int.
21667 (oecount_hasher::equal): Formatting fix.
21668 (oecount_cmp): Make sure to return the right return value
21669 even if unsigned fields are bigger than INT_MAX.
21670 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
21673 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
21674 TREE_READONLY on result if writing it more than once.
21677 * doc/invoke.texi (-fsanitize=thread): Document that with
21678 -fnon-call-exceptions atomics are not able to throw
21682 * tsan.c: Include tree-eh.h.
21683 (instrument_builtin_call): Call maybe_clean_eh_stmt or
21684 maybe_clean_or_replace_eh_stmt where needed.
21685 (instrument_memory_accesses): Add cfg_changed argument.
21686 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
21687 if it returned true.
21688 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
21690 PR rtl-optimization/63191
21691 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
21692 wrapper function, moved the whole old content into ...
21693 (ix86_delegitimize_address_1): ... this. New inline function.
21694 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
21695 true as last argument instead of ix86_delegitimize_address.
21697 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
21699 * config/aarch64/aarch64.c (generic_branch_cost): Copy
21700 cortexa57_branch_cost.
21702 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
21704 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
21706 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21709 * doc/md.texi (Constraints): Document wA constraint.
21710 * config/rs6000/constraints.md (wA): New.
21711 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
21712 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
21713 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
21714 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
21716 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
21719 * gimplify.c (is_oacc_declared): New function.
21720 (oacc_default_clause): Use it to set default flags for acc declared
21721 variables inside parallel regions.
21722 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
21723 declared variables.
21724 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
21725 declare attribute to any decl as necessary.
21727 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
21730 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
21731 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
21732 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
21733 (arm_arch_lpae): This.
21734 * config/arm/arm.c (arm_arch7ve): Rename into ...
21735 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
21736 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
21739 2017-03-22 Martin Liska <mliska@suse.cz>
21742 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
21743 error message instead of an ICE.
21745 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21747 * doc/extend.texi (6.11 Additional Floating Types): Revise.
21749 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
21751 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
21753 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
21756 2017-03-21 Martin Sebor <msebor@redhat.com>
21758 * doc/extend.texi: Use "cannot" instead of "can't."
21759 * doc/hostconfig.texi: Same.
21760 * doc/install.texi: Same.
21761 * doc/invoke.texi: Same.
21762 * doc/loop.texi: Same.
21763 * doc/md.texi: Same.
21764 * doc/objc.texi: Same.
21765 * doc/rtl.texi: Same.
21766 * doc/tm.texi: Same.
21767 * doc/tm.texi.in: Same.
21768 * doc/trouble.texi: Same.
21770 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
21773 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
21774 (collect_checksum_attributes): Set it.
21775 (die_checksum_ordered): Use it.
21777 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21779 PR tree-optimization/79908
21780 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
21781 change: For a VA_ARG whose LHS has been cast away, use
21782 force_gimple_operand to construct the side effects.
21784 2017-03-21 David Malcolm <dmalcolm@redhat.com>
21786 PR translation/80001
21787 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
21788 more amenable to translation.
21789 (oacc_loop_auto_partitions): Likewise.
21791 2017-03-21 Marek Polacek <polacek@redhat.com>
21792 Martin Sebor <msebor@redhat.com>
21794 PR tree-optimization/80109
21795 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
21796 on INTEGRAL_TYPE_P.
21798 2017-03-21 Jakub Jelinek <jakub@redhat.com>
21799 Segher Boessenkool <segher@kernel.crashing.org>
21802 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
21803 check reg_used_between_p between insn and one of succ or succ2
21804 depending on if succ is artificial insn not inserted into insn
21807 2017-03-21 Martin Liska <mliska@suse.cz>
21809 PR gcov-profile/80081
21810 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
21811 * doc/gcc.texi: Include gcov-dump stuff.
21812 * doc/gcov-dump.texi: New file.
21814 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
21816 PR rtl-optimization/79150
21817 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
21818 conditional jump, if the jump is the last insn of the loop.
21820 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21821 Richard Biener <rguenth@suse.de>
21823 PR tree-optimization/79908
21824 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
21825 been cast away, use force_gimple_operand to construct the side
21828 2017-03-21 Martin Liska <mliska@suse.cz>
21830 PR libfortran/79956
21831 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
21834 2017-03-21 Brad Spengler <spender@grsecurity.net>
21837 * plugin.c (htab_hash_plugin): New function.
21838 (add_new_plugin): Use it and adjust.
21839 (parse_plugin_arg_opt): Adjust.
21840 (init_one_plugin): Likewise.
21842 2017-03-21 Richard Biener <rguenther@suse.de>
21844 PR tree-optimization/80032
21845 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
21846 if set force the cleanup to happen unconditionally.
21847 (gimplify_target_expr): Push inserted clobbers with force_uncond
21848 to avoid them being removed by control-dependent DCE.
21850 2017-03-21 Richard Biener <rguenther@suse.de>
21852 PR tree-optimization/80122
21853 * tree-inline.c (copy_bb): Do not expans va-arg packs or
21854 va_arg_pack_len when the inlined call stmt requires pack
21856 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
21858 2017-03-21 Jakub Jelinek <jakub@redhat.com>
21861 * tsan.c (instrument_builtin_call): If the memory model argument
21862 is not a constant, assume it is valid.
21865 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
21868 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
21870 PR rtl-optimization/79910
21871 * combine.c (can_combine_p): Do not allow combining an I0 or I1
21872 if its dest is used by an insn before I2 (other than the combined
21873 insns themselves, which are properly handled already).
21875 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
21878 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
21880 * combine.c (record_used_regs): New static function.
21881 (try_combine): Handle situations where there is an additional
21882 instruction between I2 and I3 which needs to have a LOG_LINK
21886 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
21888 * combine.c (try_combine): Delete redundant i1 test. Call
21889 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
21891 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21894 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
21895 alternatives 13/14.
21897 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21899 PR tree-optimization/80054
21900 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
21901 the optimization if a PHI or any of its arguments is not dominated
21902 by the candidate's basis. Use gphi* rather than gimple* as
21904 (replace_profitable_candidates): Clean up a gimple* variable that
21905 should be a gphi* variable.
21907 2017-03-20 Martin Sebor <msebor@redhat.com>
21910 * doc/extend.texi (attribute constructor): Document present limitation.
21912 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
21915 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
21916 __POWER9_VECTOR__ #ifdef control, change template definition to
21917 use Power9-specific built-in function.
21918 (vec_any_eq): Likewise.
21919 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
21920 to control outcomes from this test.
21921 (vector_ae_<mode>p): For VEC_F modes, likewise.
21923 2017-03-20 Ian Lance Taylor <iant@google.com>
21925 * config/i386/i386.c (ix86_function_regparm): Save an extra
21926 register for -fsplit-stack with DECL_STATIC_CHAIN.
21928 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
21931 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
21932 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
21934 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
21936 * config/riscv/riscv.c (riscv_print_operand): Use "fence
21938 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
21941 2017-03-20 Marek Polacek <polacek@redhat.com>
21944 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
21946 2017-03-20 Richard Biener <rguenther@suse.de>
21948 PR tree-optimization/80113
21949 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
21950 allocate extra SSA name for PHI def.
21951 (add_close_phis_to_outer_loops): Likewise.
21952 (add_close_phis_to_merge_points): Likewise.
21953 (copy_loop_close_phi_args): Likewise.
21954 (copy_cond_phi_nodes): Likewise.
21956 2017-03-20 Martin Liska <mliska@suse.cz>
21958 PR middle-end/79753
21959 * tree-chkp.c (chkp_build_returned_bound): Do not build
21960 returned bounds for a LHS that's not a BOUNDED_P type.
21962 2017-03-20 Martin Liska <mliska@suse.cz>
21966 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
21967 COMPLEX_CST and VECTOR_CST.
21969 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21972 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
21973 target operand. A new splitter adds the clobber statement in case
21974 the target operand is dead anyway.
21976 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
21978 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
21979 to age-old versions of binutils and glibc.
21981 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
21983 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
21985 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
21987 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
21989 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
21991 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
21992 requirement for binutils 2.13.
21994 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
21996 * combine.c (try_combine): Delete redundant i1 test. Call
21997 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
21999 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
22001 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
22002 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
22004 <riscv64-*-elf>: Re-arrange section
22005 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
22006 <riscv32-*-linux>: Likewise.
22007 <riscv64-*-elf>: Likewise
22008 <riscv64-*-linux>: Likewise.
22010 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
22013 * aarch64.opt(verbose-cost-dump): Fix typo.
22015 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
22018 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
22019 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
22021 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
22023 * reload.c (find_reloads): When reloading a nonoffsettable address,
22024 use RELOAD_OTHER for it and its address reloads.
22026 PR rtl-optimization/79910
22027 * combine.c (record_used_regs): New static function.
22028 (try_combine): Handle situations where there is an additional
22029 instruction between I2 and I3 which needs to have a LOG_LINK
22032 2017-03-17 Jeff Law <law@redhat.com>
22034 PR tree-optimization/71437
22035 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
22036 conditional in the hash table first.
22037 (vrp_dom_walker::before_dom_children): Extract condition from
22038 ASSERT_EXPR. Record condition, its inverion and any implied
22039 conditions as well.
22041 2017-03-17 Marek Polacek <polacek@redhat.com>
22042 Markus Trippelsdorf <markus@trippelsdorf.de>
22044 PR tree-optimization/80079
22045 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
22048 2017-03-17 Richard Biener <rguenther@suse.de>
22050 PR middle-end/80075
22051 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
22052 Properly verify the LHS before the RHS possibly claims to be
22054 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
22057 2017-03-17 Martin Jambor <mjambor@suse.cz>
22059 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
22060 (List of -O2 options): Likewise.
22061 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
22064 2017-03-17 Tom de Vries <tom@codesourcery.com>
22066 * gcov-dump.c (print_usage): Print bug_report_url.
22068 2017-03-17 Richard Biener <rguenther@suse.de>
22070 PR middle-end/80050
22071 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
22072 (parser::peek): Likewise.
22074 2017-03-17 Richard Biener <rguenther@suse.de>
22076 PR tree-optimization/80048
22077 * sese.c (free_sese_info): Properly release rename_map and
22078 copied_bb_map elements.
22080 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
22082 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
22083 Add linked-list forward and backlinks. Insert on
22084 construction, remove on destruction.
22085 (class pass_store_merging): Add m_stores_head field.
22086 (pass_store_merging::terminate_and_process_all_chains):
22087 Iterate over m_stores_head list.
22088 (pass_store_merging::terminate_all_aliasing_chains):
22090 (pass_store_merging::execute): Check for debug stmts first.
22091 Push new chains onto the m_stores_head stack.
22093 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
22096 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
22097 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
22098 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
22100 2017-03-16 Jeff Law <law@redhat.com>
22102 PR tree-optimization/71437
22103 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
22104 member function. Implementation moved into after_dom_children
22105 member function and into the threader's thread_outgoing_edges
22107 (dom_opt_dom_walker::after_dom_children): Simplify by moving
22108 some code into new thread_outgoing_edges.
22109 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
22110 definition. Simplify marker handling (do it here). Assume we always
22111 have the available expression and the const/copies tables.
22112 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
22114 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
22115 * tree-vrp.c (equiv_stack): No longer file scoped.
22116 (vrp_dom_walker): New class.
22117 (vrp_dom_walker::before_dom_children): New member function.
22118 (vrp_dom_walker::after_dom_children): Likewise.
22119 (identify_jump_threads): Setup domwalker. Use it rather than
22120 walking edges in a random order by hand. Simplify setup/finalization.
22121 (finalize_jump_threads): Remove.
22122 (vrp_finalize): Do not call identify_jump_threads here.
22123 (execute_vrp): Do it here instead and call thread_through_all_blocks
22126 PR tree-optimization/71437
22127 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
22129 (simplify_stmt_for_jump_threading): Add basic_block argument. All
22131 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
22132 (dom_opt_dom_walker::thread_across_edge): Remove
22133 handle_dominating_asserts argument. All callers changed.
22134 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
22135 changes. Remove calls to lhs_of_dominating_assert. Other
22136 uses of handle_dominating_asserts turn into unconditional code
22137 (simplify_control_stmt_condition_1): Likewise.
22138 (simplify_control_stmt_condition): Likewise.
22139 (thread_through_normal_block, thread_across_edge): Likewise.
22140 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
22141 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
22142 object if it is not an SSA_NAME.
22143 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
22144 before calling into the VRP specific simplifiers.
22145 (identify_jump_threads): Remove handle_dominating_asserts
22148 2017-03-16 Jakub Jelinek <jakub@redhat.com>
22151 * tree-diagnostic.c (default_tree_printer): No longer static.
22152 * tree-diagnostic.h (default_tree_printer): New prototype.
22154 2017-03-16 Tamar Christina <tamar.christina@arm.com>
22156 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
22157 Change ins into fmov.
22159 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22161 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
22162 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
22163 Use h_con constraint for operand 1.
22164 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
22165 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
22167 2017-03-15 Jeff Law <law@redhat.com>
22169 PR tree-optimization/71437
22170 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
22171 (record_temporary_equivalences): Use it.
22173 PR tree-optimization/71437
22174 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
22175 tree-ssa-scopedtables.
22176 (lookup_avail_expr, build_and_record_new_cond): Likewise.
22177 (record_conditions, record_cond, vuse_eq): Likewise.
22178 (record_edge_info): Adjust to API tweak of record_conditions.
22179 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
22180 (record_temporary_equivalences, optimize_stmt): Likewise.
22181 (eliminate_redundant_computations): Likewise.
22182 (record_equivalences_from_stmt): Likewise.
22183 * tree-ssa-scopedtables.c: Include options.h and params.h.
22184 (vuse_eq): New function, moved from tree-ssa-dom.c
22185 (build_and_record_new_cond): Likewise.
22186 (record_conditions): Likewise. Accept vector of conditions rather
22187 than edge_equivalence structure for first argument.
22188 for the first argument.
22189 (avail_exprs_stack::lookup_avail_expr): New member function, moved
22190 from tree-ssa-dom.c.
22191 (avail_exprs_stack::record_cond): Likewise.
22192 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
22193 from tree-ssa-dom.c.
22194 (avail_exprs_stack): Add new member functions lookup_avail_expr
22196 (record_conditions): Declare.
22198 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
22201 * lra-constraints.c (process_alt_operands): Increase reject for
22202 reloading an input/output operand.
22204 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
22207 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
22208 insns to convert from signed/unsigned char/short to IEEE 128-bit
22210 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
22212 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
22215 * config/i386/i386.c (ix86_vector_duplicate_value): Create
22216 subreg of inner mode for values already in registers.
22218 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
22220 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
22221 iteration reg is used after the loop.
22223 2017-03-14 Martin Sebor <msebor@redhat.com>
22225 PR tree-optimization/79800
22226 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
22227 precision in negative-positive range.
22228 (format_floating): Call non-const overload with adjusted precision.
22230 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
22233 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
22236 2017-03-14 Martin Sebor <msebor@redhat.com>
22238 PR middle-end/80020
22239 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
22240 * builtins.def (aligned_alloc): Use it.
22243 * Makefile.in (GTFILES): Add calls.c.
22244 * calls.c: Include "gt-calls.h".
22246 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
22248 PR rtl-optimization/79728
22249 * regs.h (struct target_regs): New field
22250 x_contains_allocatable_regs_of_mode.
22251 (contains_allocatable_regs_of_mode): New macro.
22252 * reginfo.c (init_reg_sets_1): Initialize it, and change
22253 contains_reg_of_mode so it includes global regs as well.
22254 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
22255 rather than contains_regs_of_mode.
22257 2017-03-14 Martin Liska <mliska@suse.cz>
22259 * doc/invoke.texi: Document options that can't be combined with
22260 -fcheck-pointer-bounds.
22262 2017-03-14 Martin Liska <mliska@suse.cz>
22264 PR middle-end/79831
22265 * doc/invoke.texi (-Wchkp): Document the option.
22267 2017-03-14 Martin Liska <mliska@suse.cz>
22269 * Makefile.in: Install gcov-dump.
22271 2017-03-14 Martin Liska <mliska@suse.cz>
22273 * multiple_target.c (expand_target_clones): Bail out for
22274 an invalid attribute.
22276 2017-03-14 Richard Biener <rguenther@suse.de>
22278 * alias.c (struct alias_set_entry): Pack properly.
22279 * cfgloop.h (struct loop): Likewise.
22280 * cse.c (struct set): Likewise.
22281 * ipa-utils.c (struct searchc_env): Likewise.
22282 * loop-invariant.c (struct invariant): Likewise.
22283 * lra-remat.c (struct cand): Likewise.
22284 * recog.c (struct change_t): Likewise.
22285 * rtl.h (struct address_info): Likewise.
22286 * symbol-summary.h (function_summary): Likewise.
22287 * tree-loop-distribution.c (struct partition): Likewise.
22288 * tree-object-size.c (struct object_size_info): Likewise.
22289 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
22290 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
22291 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
22292 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
22293 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
22294 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
22295 (struct _stmt_vec_info): Likewise.
22297 2017-03-14 Martin Liska <mliska@suse.cz>
22300 * multiple_target.c (create_dispatcher_calls): Check that
22301 a target can create a function dispatcher.
22303 2017-03-14 Martin Liska <mliska@suse.cz>
22306 * multiple_target.c (expand_target_clones): Drop local.local
22307 flag for default implementation.
22309 2017-03-14 Richard Biener <rguenther@suse.de>
22311 PR tree-optimization/80030
22312 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
22314 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
22316 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
22317 gcc_fallthrough() instead of __attribute__((fallthrough));
22319 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
22321 * doc/gcc.texi: Remove "up" link to (DIR).
22322 * doc/gccint.texi: Ditto.
22324 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
22326 * doc/install.texi (Specific) <avr>: Remove reference to
22329 2017-03-13 Jeff Law <law@redhat.com>
22331 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
22332 attribute rather than comments.
22334 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
22335 match_scratch operand is highest.
22337 2017-03-13 Martin Liska <mliska@suse.cz>
22339 PR middle-end/78339
22340 * ipa-pure-const.c (warn_function_noreturn): If the declarations
22341 is a CHKP clone, use original declaration.
22343 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
22345 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
22346 (arc_conditional_register_usage): Use a different allocation order
22347 when optimizing for size.
22348 * common/config/arc/arc-common.c (arc_option_optimization_table):
22349 Section anchors default on when optimizing for size.
22351 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
22353 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
22355 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
22357 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
22358 * config/arc/arc.md (cpu_facility): Add cd variant.
22359 (*movqi_insn): Add code density variant.
22360 (*movhi_insn): Likewise.
22361 (*movqi_insn): Likewise.
22362 (*addsi3_mixed): Likewise.
22363 (subsi3_insn): Likewise.
22365 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
22367 * config/arc/arc.md (movsi_cond_exec): Update constraint.
22369 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
22371 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
22372 expressions with MINUS and UNARY ops.
22374 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22377 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
22379 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
22380 between vec_select and vector argument.
22381 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
22382 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
22383 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
22384 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
22385 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
22386 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
22388 2017-03-13 Richard Biener <rguenther@suse.de>
22391 * params.def (vect-max-peeling-for-alignment): Fix typo.
22393 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
22395 * doc/install.texi (Specific) <mips-*-*>: Remove description of
22396 issue that only occurred with binutils below 2.18.
22398 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
22400 * doc/install.texi (Specific) <cris-axis-elf>: No longer
22401 refer to binutils 2.11/2.12 minimum.
22403 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
22405 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
22406 ftp.kernel.org and simplify binutils requirement.
22408 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
22410 * doc/invoke.texi (Warning Options): Fix spelling of link-time
22412 (Optimize Options): Ditto. Also remove redundancy.
22414 2017-03-10 David Malcolm <dmalcolm@redhat.com>
22416 PR translation/79848
22417 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
22419 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
22420 to G_ to avoid double translation.
22422 2017-03-10 David Malcolm <dmalcolm@redhat.com>
22424 PR translation/79923
22425 * auto-profile.c (get_combined_location): Convert leading
22426 character of diagnostics to lower case and remove trailing period.
22427 (read_profile): Likewise for various diagnostics.
22428 * config/arm/arm.c (arm_option_override): Remove trailing period
22429 from various diagnostics.
22430 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
22431 (msp430_expand_delay_cycles): Likewise.
22433 2017-03-10 David Malcolm <dmalcolm@redhat.com>
22436 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
22437 full command-line argument, rather than just "str".
22438 (aarch64_validate_march): Likewise.
22439 (aarch64_validate_mtune): Likewise.
22441 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
22443 PR rtl-optimization/78911
22444 * lra-assigns.c (must_not_spill_p): New function.
22445 (spill_for): Use it.
22447 2017-03-10 Jakub Jelinek <jakub@redhat.com>
22449 PR tree-optimization/79981
22450 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
22451 ATOMIC_COMPARE_EXCHANGE ifn result.
22452 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
22453 IFN_ATOMIC_COMPARE_EXCHANGE.
22455 2017-03-10 David Malcolm <dmalcolm@redhat.com>
22458 * opts.c (parse_sanitizer_options): Add missing question mark to
22459 "did you mean" message.
22461 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22463 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
22465 (VMULEUH_UNS): Likewise.
22466 (VMULOUB_UNS): Likewise.
22467 (VMULOUH_UNS): Likewise.
22468 * config/rs6000/rs6000.c (builtin_function_type): Remove
22469 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
22471 2017-03-10 David Malcolm <dmalcolm@redhat.com>
22474 * read-rtl-function.c (function_reader::read_rtx_operand): Update
22475 x with result of extra_parsing_for_operand_code_0.
22476 (function_reader::extra_parsing_for_operand_code_0): Convert
22477 return type from void to rtx, returning x. When reading
22478 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
22479 larger size containing struct block_symbol.
22481 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
22483 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
22484 -mfloat128-hardware without -m64.
22486 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
22489 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
22490 entries to the case statement that marks unsigned arguments to
22491 overloaded functions.
22493 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
22495 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
22496 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
22498 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
22501 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
22502 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
22504 2017-03-10 Martin Liska <mliska@suse.cz>
22508 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
22509 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
22512 2017-03-10 Olivier Hainque <hainque@adacore.com>
22514 * tree-switch-conversion (array_value_type): Start by resetting
22515 candidate type to it's main variant.
22517 2017-03-10 Jakub Jelinek <jakub@redhat.com>
22519 PR rtl-optimization/79909
22520 * combine.c (try_combine): Use simplify_replace_rtx on individual
22521 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
22522 of the whole CALL_INSN_FUNCTION_USAGE.
22524 PR tree-optimization/79972
22525 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
22526 get_range_info on SSA_NAMEs. Formatting fixes.
22528 2017-03-10 Richard Biener <rguenther@suse.de>
22529 Jakub Jelinek <jakub@redhat.com>
22531 PR tree-optimization/77975
22532 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
22533 edge to be constant.
22534 (get_val_for): For constant x return it. Formatting fix.
22535 (loop_niter_by_eval): Avoid pointless looping if the next iteration
22536 would use the same bases as the current one.
22538 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22540 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
22541 instead of vec_select for V1TImode.
22542 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
22544 (VSX_LE_128): Add V1TI to this mode iterator.
22545 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
22546 (*vsx_le_perm_store_<mode>): Likewise.
22547 (pre-reload splitter for VSX stores): Likewise.
22548 (post-reload splitter for VSX stores): Likewise.
22549 (*vsx_xxpermdi2_le_<mode>): Likewise.
22550 (*vsx_lxvd2x2_le_<mode>): Likewise.
22551 (*vsx_stxvd2x2_le_<mode>): Likewise.
22553 2017-03-09 Michael Eager <eager@eagercon.com>
22555 Correct failures with --enable-checking=yes,rtl.
22557 * config/microblaze/microblaze.c (microblaze_expand_shift):
22558 Replace GET_CODE test with CONST_INT_P and INTVAL test with
22559 test for const0_rtx.
22560 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
22561 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
22563 2017-03-09 Richard Biener <rguenther@suse.de>
22565 PR tree-optimization/79977
22566 * graphite-scop-detection.c (scop_detection::merge_sese):
22567 Handle the case of extra exits to blocks dominating the entry.
22569 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
22571 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
22574 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
22576 PR rtl-optimization/79949
22577 * lra-constraints.c (process_alt_operands): Check memory when
22578 trying to predict a cycle. Print about the overall increase.
22580 2017-03-09 Richard Biener <rguenther@suse.de>
22582 PR middle-end/79971
22583 * gimple-expr.c (useless_type_conversion_p): Preserve
22584 TYPE_SATURATING for fixed-point types.
22586 2017-03-09 Richard Biener <rguenther@suse.de>
22589 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
22590 alignment of BLKmode params.
22592 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22595 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
22596 (VALL_NO_V2Q): Likewise.
22598 * config/aarch64/aarch64-simd.md
22599 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
22601 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
22602 VALL_NO_V2Q mode iterator.
22603 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
22605 2017-03-09 Martin Liska <mliska@suse.cz>
22607 PR tree-optimization/79631
22608 * tree-chkp-opt.c (chkp_is_constant_addr): Call
22609 tree_int_cst_sign_bit just for INTEGER constants.
22611 2017-03-09 Martin Liska <mliska@suse.cz>
22615 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
22618 2017-03-09 Marek Polacek <polacek@redhat.com>
22621 * tree.c (inchash::add_expr): Handle TREE_VEC.
22623 2017-03-09 Martin Liska <mliska@suse.cz>
22626 (chkp_narrow_size_and_offset): New function.
22627 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
22628 (void chkp_parse_bit_field_ref): New function.
22629 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
22630 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
22632 2017-03-09 Martin Liska <mliska@suse.cz>
22635 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
22636 (chkp_find_bounds_1): Remove gcc_unreachable.
22638 2017-03-09 Jakub Jelinek <jakub@redhat.com>
22641 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
22642 BUILT_IN_SYNC*, determine the access type from the size suffix and
22643 always build a MEM_REF with that type. Handle forgotten
22644 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
22647 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
22648 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
22649 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
22650 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
22651 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
22652 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
22653 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
22654 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
22655 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
22656 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
22657 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
22658 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
22659 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
22660 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
22661 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
22662 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
22663 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
22664 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
22665 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
22666 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
22667 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
22668 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
22669 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
22670 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
22671 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
22672 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
22673 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
22674 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
22675 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
22676 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
22677 definitions outside of __OPTIMIZE__ guarded section.
22680 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
22681 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
22682 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
22683 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
22686 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22688 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
22689 ("vfenez<mode>"): Add missing constraints.
22691 2017-03-08 Martin Sebor <msebor@redhat.com>
22694 * config/nds32/nds32.c (nds32_option_override):
22695 Fix misspelled diagnostic.
22697 2017-03-08 Jakub Jelinek <jakub@redhat.com>
22700 * gimplify.c (gimplify_omp_for): Replace index var in outer
22701 taskloop statement with an artificial variable and add
22702 OMP_CLAUSE_PRIVATE clause for it.
22704 2017-03-08 Richard Biener <rguenther@suse.de>
22706 PR tree-optimization/79955
22707 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
22708 for accesses that are completely outside of the variable.
22710 2017-03-08 Andrew Haley <aph@redhat.com>
22712 PR tree-optimization/79943
22713 * tree-ssa-loop-split.c (compute_new_first_bound): When
22714 calculating the new upper bound, (END-BEG) should be added, not
22717 2017-03-08 Jakub Jelinek <jakub@redhat.com>
22719 * config/avr/avr.md (setmemhi): Make sure match_dup
22720 operand number comes before match_scratch.
22722 2017-03-08 Richard Biener <rguenther@suse.de>
22724 PR tree-optimization/79920
22725 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
22726 with ncopies == 1 to ...
22727 (vect_transform_slp_perm_load): ... here. Properly compute
22728 all element loads by iterating VF times over the group. Do
22729 not handle ncopies (computed in a broken way) in
22730 vect_create_mask_and_perm.
22732 2017-03-08 Jakub Jelinek <jakub@redhat.com>
22735 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
22736 is a uniform vector, use uniform_vector_p return value instead of
22737 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
22739 2017-03-07 Marek Polacek <polacek@redhat.com>
22741 PR middle-end/79809
22742 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
22743 (alloca_call_type): Likewise.
22745 2017-03-07 Martin Liska <mliska@suse.cz>
22747 * gcov.c (process_args): Put comment to correct location.
22749 2017-03-07 Martin Liska <mliska@suse.cz>
22751 PR middle-end/68270
22752 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
22753 Use array_at_struct_end_p instead of DECL_CHAIN (field).
22754 (chkp_narrow_bounds_for_field): Likewise.
22755 (chkp_parse_array_and_component_ref): Pass one more argument to
22758 2017-03-07 Richard Biener <rguenther@suse.de>
22760 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
22763 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
22765 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
22766 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
22768 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22771 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
22772 to end of description.
22773 (PARAM_MAX_STORES_TO_MERGE): Likewise.
22775 2017-03-07 Jakub Jelinek <jakub@redhat.com>
22777 PR rtl-optimization/79901
22778 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
22780 (*avx512f_<code><mode>3<mask_name>): ... this.
22781 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
22782 iterator instead of VI8_AVX2_AVX512BW.
22784 PR rtl-optimization/79901
22785 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
22786 min/max expander, expand it using expand_vec_cond_expr.
22789 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
22792 2017-03-06 Jakub Jelinek <jakub@redhat.com>
22795 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
22796 to void * for PCH reasons.
22797 * dwarf2out.c (output_loc_operands, output_die): Cast
22798 v.val_vec.array to unsigned char *.
22800 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
22803 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
22806 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
22808 PR rtl-optimization/79571
22809 * lra-constraints.c (process_alt_operands): Calculate static
22810 reject and subtract it from overall when only addresses will be
22813 2017-03-06 Julia Koval <julia.koval@intel.com>
22816 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
22817 incoming stack boundary to 128 for 64-bit targets.
22819 2017-03-06 Richard Biener <rguenther@suse.de>
22821 PR tree-optimization/79894
22822 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
22823 to NULL after folding it.
22825 2017-03-06 Richard Biener <rguenther@suse.de>
22827 PR tree-optimization/79824
22828 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
22829 check disabling peeling for gaps.
22831 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
22833 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
22834 attributes): Document gettimeofday.
22836 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
22838 * config/s390/s390.c (s390_option_override_internal): Set
22839 PARAM_MIN_VECT_LOOP_BOUND
22841 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
22843 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
22844 * config/s390/s390.md: Likewise.
22846 2017-03-06 Jakub Jelinek <jakub@redhat.com>
22849 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
22850 (<avx2_avx512>_perm<mode>): Rename to ...
22851 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
22853 (<avx512>_perm<mode>_mask): Rename to ...
22854 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
22856 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
22857 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
22858 instead of VI8F_256_512.
22859 (avx512f_perm<mode>): New define_expand.
22860 (avx512f_perm<mode>_mask): Likewise.
22861 (avx512f_perm<mode>_1<mask_name>): New define_insn.
22862 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
22864 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
22866 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
22867 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
22869 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
22871 2017-03-06 Martin Liska <mliska@suse.cz>
22874 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
22875 when having a SSA NAME w/o VAR_DECL assigned to it.
22877 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
22879 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
22880 msa_dpsub_<su>_d): Fix MODE for vec_select.
22882 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
22884 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
22886 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
22888 2017-03-06 Richard Biener <rguenther@suse.de>
22890 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
22891 * plugin.c (register_plugin_info): Likewise.
22892 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
22894 2017-03-05 Jakub Jelinek <jakub@redhat.com>
22896 * config/i386/sse.md (sse_storehps, sse_storelps,
22897 avx_<castmode><avxsizesuffix>_<castmode>,
22898 avx512f_<castmode><avxsizesuffix>_<castmode>,
22899 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
22900 in condition that at least one operand is not a MEM.
22902 2017-03-03 Jakub Jelinek <jakub@redhat.com>
22904 PR middle-end/79805
22905 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
22906 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
22908 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
22909 gimple_call_nothrow_p flag based on whether original builtin can throw.
22910 If it can, emit following stmts on the fallthrough edge.
22911 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
22912 don't create new bb if inserting just debug stmts on the edge, try to
22913 insert them on the fallthru bb or just reset debug stmts.
22915 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
22918 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
22919 restore recog_data (including the operand rtxes inside it) around
22920 the call to get_insn_template.
22922 2017-03-03 Martin Sebor <msebor@redhat.com>
22924 PR tree-optimization/79699
22925 * context.c (context::~context): Free MPFR caches to avoid
22926 a memory leak on program exit.
22928 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22930 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
22931 Use wide_int::ulow () instead of .elt (0).
22933 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
22935 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
22936 (*pushxf): Limit oF constraint to 32bit targets and add oC
22937 constraint for 64bit targets.
22938 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
22939 (*pushdf): Change rmF constraint to rmC.
22941 2017-03-03 Martin Liska <mliska@suse.cz>
22943 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
22944 Remove unused variable.
22946 2017-03-03 Jakub Jelinek <jakub@redhat.com>
22949 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
22950 is a memory operand, increase num_memory.
22951 (ix86_expand_args_builtin): Likewise.
22953 2017-03-03 Jan Hubicka <jh@suse.cz>
22956 * ipa-devirt.c (maybe_record_node): Properly handle
22957 __cxa_pure_virtual visibility.
22959 2017-03-03 Martin Liska <mliska@suse.cz>
22961 PR tree-optimization/79803
22962 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
22964 (pass_loop_prefetch::execute): Disabled optimization if an
22965 assumption about L1 cache size is not met.
22967 2017-03-03 Martin Liska <mliska@suse.cz>
22969 PR rtl-optimization/79574
22970 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
22971 (hash_scan_set): Likewise.
22972 (dump_hash_table): Likewise.
22973 (hoist_code): Likewise.
22975 2017-03-03 Richard Biener <rguenther@suse.de>
22977 * fixed-value.c (fixed_from_string): Restore use of elt (1)
22978 in place of uhigh ().
22979 (fixed_convert_from_real): Likewise.
22981 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
22984 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
22986 2017-03-03 Richard Biener <rguenther@suse.de>
22988 PR middle-end/79818
22989 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
22990 TYPE_OVERFLOW_UNDEFINED check.
22992 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22994 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
22996 (vector_ae_<mode>_p): Likewise.
22997 (vector_nez_<mode>_p): Likewise.
22998 (vector_ne_v2di_p): Likewise.
22999 (vector_ae_v2di_p): Likewise.
23000 (vector_ne_<mode>_p): Likewise.
23001 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
23003 (vsx_tsqrt<mode>2_fe): Likewise.
23005 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
23008 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
23010 2017-03-02 Jakub Jelinek <jakub@redhat.com>
23012 PR rtl-optimization/79780
23013 * cprop.c (one_cprop_pass): When second and further conditional trap
23014 in a single basic block is turned into an unconditional trap, turn it
23015 into a deleted note to avoid RTL verification failures.
23017 2017-03-02 Richard Biener <rguenther@suse.de>
23019 * fold-const.c (const_binop): Use ulow () instead of elt (0).
23021 2017-03-02 Richard Biener <rguenther@suse.de>
23023 PR tree-optimization/79345
23025 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
23026 param and abort the walk, returning -1 if it is hit.
23027 (walk_aliased_vdefs): Take a limit param and pass it on.
23028 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
23029 defaulting to 0 and return a signed int.
23030 * tree-ssa-uninit.c (struct check_defs_data): New struct.
23031 (check_defs): New helper.
23032 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
23033 about uninitialized memory.
23034 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
23035 bogus uninitialized warning.
23036 (fixed_convert_from_real): Likewise.
23038 2017-03-02 Bin Cheng <bin.cheng@arm.com>
23040 PR tree-optimization/66768
23041 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
23042 iv_use if base object can't be determined.
23044 2017-03-02 Jakub Jelinek <jakub@redhat.com>
23046 PR tree-optimization/79345
23047 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
23048 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
23049 (get_pattern_stats): Initialize it.
23050 * genemit.c (gen_expand): Verify match_scratch numbers come after
23051 match_operand/match_dup numbers.
23052 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
23053 match_scratch numbers.
23054 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
23056 * config/s390/s390.md (trunctdsd2): Likewise.
23058 2017-03-02 Richard Biener <rguenther@suse.de>
23060 * wide-int.h (wide_int_storage::operator=): Implement in terms
23063 2017-03-02 Richard Biener <rguenther@suse.de>
23065 PR tree-optimization/79777
23066 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
23067 the to insert expression to sth existing.
23069 2017-03-01 Martin Sebor <msebor@redhat.com>
23071 PR middle-end/79692
23072 * gimple-ssa-sprintf.c
23073 (directive::known_width_and_precision): New function.
23074 (format_integer): Use it.
23075 (get_mpfr_format_length): Consider the full range of precision
23076 when computing %g output with the # flag. Set the likely byte
23077 count to 3 rather than 1 when precision is indeterminate.
23078 (format_floating): Correct the lower bound of precision.
23080 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23082 * doc/invoke.texi: Document default code model for 64-bit Linux.
23084 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23087 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
23088 udiv rather than div since input pattern is unsigned.
23090 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
23092 * config/i386/i386.c (print_reg): Warn for values of
23093 unsupported size in integer register.
23095 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
23098 * config/rs6000/predicates.md (current_file_function_operand): Do
23099 not allow self calls to be local if the function is replaceable.
23101 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
23104 * config/rs6000/altivec.h (vec_ctz and others): Change the
23105 preprocessor macro that controls conditional compilation from
23106 _ARCH_PWR9 to __POWER9_VECTOR__.
23107 (vec_all_ne): Change parameterization of __altivec_scalar_pred
23108 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
23109 control (instead of _ARCH_PWR9 control) so that template
23110 definition uses power9-specific function.
23111 (vec_any_eq): Likewise.
23112 (vec_all_ne): Change macro definition to use a power9-specific
23113 expansion under #ifdef __POWER9_VECTOR__ control (instead of
23114 _ARCH_PWR9 control).
23115 (vec_any_eq) Likewise.
23116 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
23117 expansion for CMPNEF to remove support for xvcmpnesp instruction.
23118 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
23119 support for xvcmpnedp instruction.
23120 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
23121 macro expansion so that Power9 implementation of vec_all_ne does
23122 not use the AltiVec predicate framework.
23123 (VCMPNEH_P): Likewise.
23124 (VCMPNEW_P): Likewise.
23125 (VCMPNED_P): Likewise.
23126 (VCMPNEFP_P): Likewise.
23127 (VCMPNEDP_P): Likewise.
23128 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
23129 implementation of vec_any_eq to not use AltiVec predicate
23131 (VCMPAEH_P): Likewise.
23132 (VCMPAEW_P): Likewise.
23133 (VCMPAED_P): Likewise.
23134 (VCMPAEFP_P): Likewise.
23135 (VCMPAEDP_P): Likewise.
23136 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
23137 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
23138 not use the AltiVec predicate framework.
23139 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
23140 of vec_any_eq to not use AltiVec predicate framework.
23141 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
23142 support for predefined __POWER9_VECTOR__ macro to indicate that
23143 Power9 instruction selection is enabled.
23144 (altivec_overloaded_builtins): Remove extraneous
23145 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
23146 function argument types RS6000_BTI_bool_V16QI and
23147 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
23148 entry for overloaded function argument types RS6000_BTI_bool_V4SI
23149 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
23150 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
23151 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
23152 Power9 for implementations of vec_cmpne. Change the signature for
23153 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
23154 (representing vec_all_ne) to remove the previously described first
23155 argument of type RS6000_BTI_INTSI, as this was an artifact of
23156 reliance on the AltiVec predicate framework, which is no longer
23157 used in the implementation of these functions. Add
23158 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
23159 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
23160 since, unlike the AltiVec predicate framework implementation, we
23161 do not share function descriptors between vec_alle and vec_anyeq.
23162 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
23163 set of modes that receive special treatment even when
23164 TARGET_P9_VECTOR is true. The special treatment emits code that
23165 does not depend on Power9 instructions.
23166 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
23167 define_expand to not rely on AltiVec predicate framework.
23168 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
23170 (vector_ne_v2di_p): Change this define_expand to not rely on
23171 AltiVec predicate framework.
23172 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
23174 (vector_ne_<mode>_p): Change this define_expand to not rely on
23175 AltiVec predicate framework.
23176 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
23178 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
23179 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
23180 define_insn pattern.
23181 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
23182 define_insn pattern because the xvcmpne<VSs>. instruction is not
23184 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
23185 instruction is not supported.
23187 2017-03-01 Jakub Jelinek <jakub@redhat.com>
23189 * config/nvptx/nvptx.c: Include intl.h.
23191 2017-03-01 Martin Jambor <mjambor@suse.cz>
23194 * ipa-prop.h (ipa_bits): Removed field known.
23195 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
23196 to pointers. Adjusted their comments to warn about their sharing.
23197 (ipcp_transformation_summary): Change bits to a vector of pointers.
23198 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
23199 (ipa_get_ipa_bits_for_value): Declare.
23200 * tree-vrp.h (value_range): Mark as GTY((for_user)).
23201 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
23202 (ipa_bits_hash_table): Likewise.
23203 (ipa_vr_ggc_hash_traits): Likewise.
23204 (ipa_vr_hash_table): Likewise.
23205 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
23206 being pointers and vr_known being removed.
23207 (ipa_set_jf_unknown): Likewise.
23208 (ipa_get_ipa_bits_for_value): New function.
23209 (ipa_set_jfunc_bits): Likewise.
23210 (ipa_get_value_range): New overloaded functions.
23211 (ipa_set_jfunc_vr): Likewise.
23212 (ipa_compute_jump_functions_for_edge): Use the above functions to
23213 construct bits and vr parts of jump functions.
23214 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
23215 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
23217 (ipcp_grow_transformations_if_necessary): Also allocate
23218 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
23220 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
23221 them. Fix too long lines.
23222 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
23223 vr_known being removed.
23224 (ipa_read_jump_function): Use new setter functions to construct bits
23225 and vr parts of jump functions or set them to NULL.
23226 (write_ipcp_transformation_info): Adjust for bits being pointers.
23227 (read_ipcp_transformation_info): Likewise.
23228 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
23230 Include gt-ipa-prop.h.
23231 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
23233 (ipcp_store_bits_results): Likewise.
23234 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
23235 Do not write to existing jump functions but use a temporary instead.
23237 2017-03-01 Jakub Jelinek <jakub@redhat.com>
23240 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
23241 attempt to use its first operand as BIT_FIELD_REF base.
23243 2017-03-01 Richard Biener <rguenther@suse.de>
23245 PR middle-end/79721
23246 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
23247 interpolating formula in wrapping arithmetic.
23248 (chrec_apply): Convert chrec_evaluate return value to wanted type.
23250 2017-03-01 Jakub Jelinek <jakub@redhat.com>
23252 PR tree-optimization/79734
23253 * tree-vect-generic.c (expand_vector_condition): Optimize
23254 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
23255 Handle VEC_COND_EXPR where comparison has different inner width from
23256 type's inner width.
23258 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
23260 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
23261 markup, and similar issues. Remove @opindex entries for things
23262 that aren't options. Add missing -mmpy-option entries.
23264 2017-02-28 Jakub Jelinek <jakub@redhat.com>
23266 PR tree-optimization/79737
23267 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
23268 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
23269 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
23270 instead of byte_size. Formatting fix.
23271 (shift_bytes_in_array_right): Formatting fix.
23273 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
23276 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
23277 condition on optimize for the leaf function test.
23279 2017-02-28 Martin Liska <mliska@suse.cz>
23282 * read-rtl-function.c (function_reader::handle_unknown_directive):
23283 Bail out when one uses -flto.
23285 2017-02-28 Martin Liska <mliska@suse.cz>
23287 * common.opt: Replace space with tabular for options of <number>
23289 * config/i386/i386.opt: Show <number> value for
23290 -mlarge-data-threshold.
23291 * opts.c (print_filtered_help): Do not display number in hexadecimal
23294 2017-02-28 Martin Liska <mliska@suse.cz>
23296 * common.opt: Fix --help=option -Q for options which are of
23299 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
23301 * config/i386/i386.c (print_reg): Error out for values
23302 of 8-bit size in invalid integer register.
23304 2017-02-28 Martin Sebor <msebor@redhat.com>
23306 PR tree-optimization/79691
23307 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
23309 2017-02-28 Jakub Jelinek <jakub@redhat.com>
23312 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
23313 gcc_unreachable with output_operand_lossage.
23315 2017-02-28 Richard Biener <rguenther@suse.de>
23317 PR tree-optimization/79740
23318 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
23320 (visit_nary_op): Insert the nary into the hashtable if we
23321 pattern-matched sth.
23322 * tree-ssa-pre.c (eliminate_insert): Robustify.
23324 2017-02-28 Richard Biener <rguenther@suse.de>
23326 PR middle-end/79731
23327 * fold-const.c (decode_field_reference): Reject out-of-bound
23330 2017-02-28 Jakub Jelinek <jakub@redhat.com>
23332 * config/i386/i386.c: Include intl.h.
23333 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
23334 instead of just cond ? "..." : "...".
23335 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
23336 * coverage.c (read_counts_file): Likewise.
23337 * omp-offload.c: Include intl.h.
23338 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
23339 of just cond ? "..." : "...".
23340 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
23341 of just cond ? "..." : "...".
23343 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
23346 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
23348 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
23349 'tune for' CPU name.
23350 * config/arm/arm-cpu-data.h: Regenerated.
23352 2017-02-28 Richard Biener <rguenther@suse.de>
23354 PR tree-optimization/79732
23355 * tree-inline.c (expand_call_inline): Do not shadow var.
23357 2017-02-28 Richard Biener <rguenther@suse.de>
23359 PR tree-optimization/79723
23360 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
23361 address-space properly.
23363 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
23365 * doc/optinfo.texi (Optimization groups): Fix option used for
23367 * doc/invoke.texi (-fopt-info): Document "omp".
23368 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
23369 (OPTGROUP_ALL): Add OPTGROUP_OMP.
23370 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
23371 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
23372 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
23374 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
23376 * dumpfile.c (optgroup_options): Instead of "openmp", associate
23377 OPTGROUP_OMP with "omp".
23379 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
23382 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
23383 for arithmetic shift of unsigned V2DI.
23385 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
23387 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
23388 arc/linux.h headers.
23389 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
23390 (LINK_SPEC): Likewise.
23391 (ARC_TLS_EXTRA_START_SPEC): Likewise.
23392 (EXTRA_SPECS): Likewise.
23393 (STARTFILE_SPEC): Likewise.
23394 (ENDFILE_SPEC): Likewise.
23395 (LIB_SPEC): Likewise.
23396 (TARGET_SDATA_DEFAULT): Likewise.
23397 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
23398 (MULTILIB_DEFAULTS): Likewise.
23399 (DWARF2_UNWIND_INFO): Likewise.
23400 * config/arc/big.h: New file.
23401 * config/arc/elf.h: Likewise.
23402 * config/arc/linux.h: Likewise.
23403 * config/arc/t-uClibc: Remove.
23405 2017-02-27 Bin Cheng <bin.cheng@arm.com>
23407 PR tree-optimization/77536
23408 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
23409 (tree_transform_and_unroll_loop): Use above function to compute the
23410 estimated niter of unrolled loop and use it when scaling profile.
23411 Also use count info rather than frequency if it's non-zero.
23412 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
23413 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
23414 (vect_transform_loop): Call above function.
23416 2017-02-27 Richard Biener <rguenther@suse.de>
23418 PR tree-optimization/45397
23419 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
23420 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
23421 (visit_nary_op): Add pattern matching for CSEing sign-changed
23422 or truncated operations with wider ones.
23424 2017-02-27 Richard Biener <rguenther@suse.de>
23426 PR tree-optimization/79690
23427 * tree-vect-stmts.c (vectorizable_store): Use vector type
23428 built from the DR with address-space.
23430 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
23432 * doc/invoke.texi (Optimize Options): Refine the description
23433 of asan-use-after-return.
23435 2017-02-25 Alan Modra <amodra@gmail.com>
23437 PR rtl-optimization/79584
23438 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
23439 base, not ad->base_term, the reg within base. Remove assertion
23440 that ad->base == ad->base_term. Replace gen_int_mode using
23441 bogus mode with const0_rtx.
23443 2017-02-25 Jakub Jelinek <jakub@redhat.com>
23445 PR middle-end/79396
23446 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
23447 FMA_EXPR like tcc_binary or tcc_unary.
23449 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
23452 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
23454 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
23455 (output_loc_operands): Handle DW_OP_call_ref and
23456 DW_OP_GNU_variable_value.
23457 (struct variable_value_struct): New type.
23458 (struct variable_value_hasher): Likewise.
23459 (variable_value_hash): New variable.
23460 (string_types): Remove.
23461 (copy_loc_descr): New function.
23462 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
23463 (prepend_loc_descr_to_each): New function.
23464 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
23465 instead of add_loc_descr_to_each if the first argument is single
23466 location list and the second has multiple.
23467 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
23468 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
23469 when looking for variable value which doesn't have other location info.
23470 (loc_list_from_tree): Formatting fix.
23471 (gen_array_type_die): Simplify DW_AT_string_length handling.
23472 (adjust_string_types): Remove.
23473 (gen_subprogram_die): Don't call adjust_string_types nor test/set
23474 string_types. Call resolve_variable_values.
23475 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
23476 (resolve_addr_in_expr): Likewise. Add A argument.
23477 (copy_deref_exprloc): Remove deref argument. Adjust for the
23478 original expression being DW_OP_GNU_variable_value with optionally
23479 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
23480 optionally after it.
23481 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
23482 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
23483 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
23484 (variable_value_hasher::hash, variable_value_hasher::equal): New
23486 (resolve_variable_value_in_expr, resolve_variable_value,
23487 resolve_variable_values, note_variable_value_in_expr,
23488 note_variable_value): New functions.
23489 (dwarf2out_early_finish): Call note_variable_value on all toplevel
23492 2017-02-24 Jakub Jelinek <jakub@redhat.com>
23495 * opts.h (handle_generated_option): Add GENERATED_P argument.
23496 * opts-common.c (handle_option): Adjust function comment.
23497 (handle_generated_option): Add GENERATED_P argument, pass it to
23499 (control_warning_option): Pass false to handle_generated_option
23501 * opts.c (maybe_default_option): Pass true to handle_generated_option
23503 * optc-gen.awk: Likewise.
23505 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
23507 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
23508 a REG, look at the REG it is a SUBREG of.
23509 (splitter for cmpeqsi_t): Ditto.
23511 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
23513 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
23514 the special USEs with the pattern of the insn, not the insn itself.
23516 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
23519 * doc/invoke.texi: Document -mload-store-pairs.
23521 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
23522 Sandra Loosemore <sandra@codesourcery.com>
23524 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
23525 argument isn't a CONST_INT.
23526 (nios2_alternate_compare_const): Assert op is a CONST_INT.
23527 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
23528 (nios2_validate_compare): Bypass alternate compare logic if *op2
23529 is not a CONST_INT.
23530 (ldstwm_operation_p): Return false if first_base is not a REG or
23531 if first_offset is not a CONST_INT.
23533 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
23535 * config/cris/cris.md: Use correct operand in a define_peephole2.
23537 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
23539 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
23541 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
23543 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
23544 this_insn if it is an INSN or JUMP_INSN.
23545 (force_offsettable): Look at base, not at addr.
23546 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
23547 on things that aren't necessarily CONST_INTs.
23549 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
23551 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
23552 -mfpmath=sse is the default also for x86-32 targets with SSE2
23553 instruction set when @option{-ffast-math} is enabled
23555 2017-02-24 Jeff Law <law@redhat.com>
23557 PR rtl-optimizatoin/79286
23558 * ira.c (update_equiv_regs): Drop may_trap_p exception to
23561 2017-02-24 Richard Biener <rguenther@suse.de>
23563 PR tree-optimization/79389
23564 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
23567 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
23569 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
23570 function comment to reflect reality.
23571 (loop_exits_before_overflow): Fix typo in function description.
23573 2017-02-24 Richard Biener <rguenther@suse.de>
23575 PR tree-optimization/79389
23576 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
23577 properly that a threading opportunity exists. Detect conditional
23578 copy/constant propagation opportunities.
23580 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
23582 * config/visium/visium.md (type): Add trap.
23583 (b): New mode attribute.
23584 (*btst): Rename into...
23585 (*btst<mode>): ...this and adjust.
23586 (*cbranchsi4_btst_insn): Rename into...
23587 (*cbranch<mode>4_btst_insn): ...this and adjust.
23588 (trap): New define_insn.
23590 2017-02-23 Jakub Jelinek <jakub@redhat.com>
23592 PR tree-optimization/79389
23593 * ifcvt.c (struct noce_if_info): Add rev_cond field.
23594 (noce_reversed_cond_code): New function.
23595 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
23596 reversed_comparison_code. Formatting fix.
23597 (noce_try_store_flag): Test rev_cond != NULL in addition to
23598 reversed_comparison_code.
23599 (noce_try_store_flag_constants): Likewise.
23600 (noce_try_store_flag_mask): Likewise.
23601 (noce_try_addcc): Use rev_cond if non-NULL instead of
23602 reversed_comparison_code.
23603 (noce_try_cmove_arith): Likewise. Formatting fixes.
23604 (noce_try_minmax, noce_try_abs): Clear rev_cond.
23605 (noce_find_if_block): Initialize rev_cond.
23606 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
23607 instead of false as last argument never attempt to reverse it
23610 2017-02-23 Bin Cheng <bin.cheng@arm.com>
23612 PR tree-optimization/79663
23613 * tree-predcom.c (combine_chains): Process refs in reverse order
23614 only for ZERO length chains, and add explaining comment.
23616 2017-02-23 Jeff Law <law@redhat.com>
23618 PR tree-optimization/79578
23619 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
23620 in call to operand_equal_p.
23622 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
23625 * config/i386/cpuid.h: Fix another undefined behavior.
23627 2017-02-23 Richard Biener <rguenther@suse.de>
23629 PR tree-optimization/79683
23630 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
23631 vector types for data-refs.
23633 2017-02-23 Martin Liska <mliska@suse.cz>
23635 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
23637 2017-02-23 Jakub Jelinek <jakub@redhat.com>
23639 PR middle-end/79665
23640 * internal-fn.c (get_range_pos_neg): Moved to ...
23641 * tree.c (get_range_pos_neg): ... here. No longer static.
23642 * tree.h (get_range_pos_neg): New prototype.
23643 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
23644 are known to be in between 0 and signed maximum inclusive, try to
23645 expand both unsigned and signed divmod and use the cheaper one from
23648 2017-02-22 Jeff Law <law@redhat.com>
23650 PR tree-optimization/79578
23651 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
23652 to compare base operands.
23654 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
23657 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
23658 gpc_reg_operand instead of fpr_reg_operand.
23660 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
23662 * config/mips/mips.c (mips_return_in_memory): Force FP
23663 vector types to be returned in memory for o32 ABI.
23665 2017-02-22 Jakub Jelinek <jakub@redhat.com>
23667 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
23668 instead of DW_TAG_member for static data member declarations and don't
23669 set no_linkage_name for static inline data members.
23670 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
23673 2017-02-22 Martin Liska <mliska@suse.cz>
23675 * doc/invoke.texi: Replace inequality signs with square brackets
23678 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23680 PR tree-optimization/68644
23681 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
23683 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
23686 * lra-constraints.c (simplify_operand_subreg): Handle
23687 WORD_REGISTER_OPERATIONS targets.
23689 2017-02-22 Jakub Jelinek <jakub@redhat.com>
23692 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
23693 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
23694 elimination by swapping fld*.
23696 2017-02-22 Richard Biener <rguenther@suse.de>
23698 PR tree-optimization/79673
23699 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
23700 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
23701 irrelevant address-space qualifiers and avoiding a
23702 ADDR_SPACE_CONVERT_EXPR from fold_convert.
23704 2017-02-22 Richard Biener <rguenther@suse.de>
23706 PR tree-optimization/79666
23707 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
23708 to not symbolically negate if that may introduce undefined
23711 2017-02-22 Martin Liska <mliska@suse.cz>
23714 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
23715 * data-streamer-out.c (streamer_write_gcov_count_stream):
23717 * value-prof.c (stream_out_histogram_value): Make assert more
23718 precise based on type of counter.
23720 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
23723 * config/i386/i386.md (standard_x87sse_constant_load splitter):
23724 Use nonimmediate_operand instead of memory_operand for operand 1.
23725 (float-extend standard_x87sse_constant_load splitter): Ditto.
23727 2017-02-21 Jeff Law <law@redhat.com>
23729 PR tree-optimization/79621
23730 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
23731 blocks with edges to themselves.
23733 2017-02-21 Jakub Jelinek <jakub@redhat.com>
23736 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
23737 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
23738 Use gimple_call_builtin_p.
23741 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
23742 on temporarily removed DEBUG_INSNs.
23744 PR tree-optimization/79649
23745 * tree-loop-distribution.c (classify_partition): Give up on
23746 non-generic address space loads/stores.
23748 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
23750 * doc/loop.texi (Loop manipulation): Remove nonexistent
23751 tree_ssa_loop_version from the documentation.
23752 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
23754 2017-02-21 Jakub Jelinek <jakub@redhat.com>
23757 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
23758 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
23759 * config/rs6000/rs6000.c: Include except.h.
23760 (rs6000_expand_split_stack_prologue): Call
23761 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
23763 2017-02-21 Martin Jambor <mjambor@suse.cz>
23766 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
23767 have been analyzed.
23769 2017-02-21 Martin Jambor <mjambor@suse.cz>
23771 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
23772 for backward compatibility only.
23773 * doc/invoke.texi (Option Summary): Remove all references to
23774 -fipa-cp-alignment.
23776 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
23780 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
23782 * lra-constraints.c (curr_insn_transform): Handle
23783 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
23785 2017-02-21 Martin Liska <mliska@suse.cz>
23787 * config/i386/i386.opt: Replace -masm-dialect with -masm.
23789 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
23791 PR translation/79638
23792 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
23794 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
23797 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
23798 (arm_function_ok_for_sibcall): Return false for an indirect call by
23799 descriptor if all the argument registers are used.
23800 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
23801 alignment of the function.
23803 2017-02-21 Jakub Jelinek <jakub@redhat.com>
23805 PR tree-optimization/61441
23806 * simplify-rtx.c (simplify_const_unary_operation): For
23807 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
23808 the sNaN unmodified.
23810 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
23812 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
23813 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
23814 instead of SYSTEM_HEADER_DIR.
23816 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
23817 Martin Liška <mliska@suse.cz>
23819 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
23820 Fix typos and grammar, use active voice, and clarify.
23822 2017-02-20 Marek Polacek <polacek@redhat.com>
23824 PR middle-end/79537
23825 * gimplify.c (gimplify_expr): Handle unused *&&L;.
23828 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
23830 2017-02-20 Jakub Jelinek <jakub@redhat.com>
23833 * config/i386/i386.c (ix86_expand_builtin): Handle
23834 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
23835 ix86_builtins_isa[fcode].isa as a requirement of those
23836 flags and any other flag in the bitmask.
23837 (ix86_init_mmx_sse_builtins): Use 0 instead of
23838 ~OPTION_MASK_ISA_64BIT as mask.
23839 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
23840 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
23841 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
23842 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
23844 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
23847 * lra-constraints.c (split_reg): Check requested split mode
23848 is supported by the register.
23850 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
23852 * lra-constraints.c (simplify_operand_subreg): Remove early
23855 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
23858 * lra-constraints.c (curr_insn_transform): Tighten condition
23859 for converting SUBREG reloads from OP_OUT to OP_INOUT.
23861 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
23864 * lra-constraints.c (curr_insn_transform): Handle
23865 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
23867 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
23870 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
23872 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
23874 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
23877 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
23880 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23882 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
23885 2017-02-18 Jakub Jelinek <jakub@redhat.com>
23887 * final.c (last_columnnum, override_columnnum): New variables.
23888 (final_start_function): Set last_columnnum, pass it to begin_prologue
23889 hook and pass 0 to dwarf2out_begin_prologue.
23890 (final_scan_insn): Update override_columnnum. Pass last_columnnum
23891 to source_line debug hook.
23892 (notice_source_line): Compute last_columnnum and for debug_column_info
23893 return true on column changes.
23894 * debug.h (struct gcc_debug_hooks): Add column argument to
23895 source_line and begin_prologue hooks.
23896 (debug_nothing_int_charstar_int_bool): Remove prototype.
23897 (debug_nothing_int_int_charstar,
23898 debug_nothing_int_int_charstar_int_bool): New prototypes.
23899 (dwarf2out_begin_prologue): Add column argument.
23900 * debug.c (do_nothing_debug_hooks): Adjust source_line and
23901 begin_prologue hooks.
23902 (debug_nothing_int_charstar_int_bool): Remove.
23903 (debug_nothing_int_int_charstar,
23904 debug_nothing_int_int_charstar_int_bool): New functions.
23905 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
23906 through to dwarf2out_source_line.
23907 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
23908 (dwarf2out_source_line): Add column argument, emit it if requested.
23909 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
23911 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
23912 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
23913 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
23914 through to dwarf2out_begin_prologue.
23915 (vmsdbgout_source_line): Add column argument, pass it through to
23916 dwarf2out_source_line.
23917 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
23918 dbxout_source_line caller.
23919 (dbxout_source_line): Add column argument.
23921 * common.opt (gno-column-info, gcolumn-info): New options.
23922 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
23923 (check_die): Also test for multiple DW_AT_decl_column attributes.
23924 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
23925 DW_AT_decl_column if requested.
23926 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
23928 (gen_variable_die): Likewise.
23929 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
23930 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
23933 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
23934 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
23935 (ix86_handle_option): Handle OPT_m3dnowa.
23936 * doc/invoke.texi (-m3dnowa): Document.
23937 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
23938 -m3dnowa instead of -m3dnow -march=athlon.
23941 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
23942 instead of gcc_assert for K, r and R code checks. Formatting fixes.
23944 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23947 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
23948 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
23949 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
23950 generator for vsx_xxpermdi_<mode>_be.
23951 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
23952 force big-endian semantics.
23953 (vsx_xxpermdi_<mode>_be): New define_expand with same
23954 implementation as previous version of vsx_xxpermdi_<mode>.
23956 2017-02-17 Jakub Jelinek <jakub@redhat.com>
23958 PR tree-optimization/79327
23959 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
23960 variable, its initialization and use.
23962 2017-02-17 Julia Koval <julia.koval@intel.com>
23964 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
23965 (OPTION_MASK_ISA_PKU_UNSET): New.
23966 (ix86_handle_option): Handle -mrdpid.
23967 * config/i386/cpuid.h (bit_RDPID): New.
23968 * config/i386/driver-i386.c (host_detect_local_cpu):
23969 Detect RDPID feature.
23970 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
23971 * config/i386/i386-c.c (ix86_target_macros_internal):
23973 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
23974 (ix86_valid_target_attribute_inner_p): Add "rdpid".
23975 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
23976 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
23977 * config/i386/i386.md (define_insn "rdpid"): New.
23978 * config/i386/i386.opt Add -mrdpid.
23979 * config/i386/immintrin.h (_rdpid_u32): New.
23981 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
23983 PR rtl-optimization/79541
23984 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
23985 instead of transforming it into USE.
23987 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
23989 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
23990 If HONOR_SNANS (SFmode) force the input to a register.
23991 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
23992 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
23993 an frsp or similar insn.
23995 2017-02-17 Martin Liska <mliska@suse.cz>
23997 PR rtl-optimization/79577
23998 * params.def (selsched-max-sched-times): Increase minimum to 1.
24000 2017-02-17 Martin Liska <mliska@suse.cz>
24002 PR rtl-optimization/79574
24003 * gcse.c (want_to_gcse_p): Prevent integer overflow.
24005 2017-02-17 Martin Liska <mliska@suse.cz>
24007 PR tree-optimization/79529
24008 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
24009 ssa_defined_default_def_p to handle cases which are implicitly
24011 * tree-ssa.c (ssa_defined_default_def_p): New function.
24012 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
24013 which are implicitly defined.
24014 * tree-ssa.h (ssa_defined_default_def_p): Declare.
24016 2017-02-17 Richard Biener <rguenther@suse.de>
24018 PR middle-end/79576
24019 * params.def (max-ssa-name-query-depth): Limit to 10.
24021 2017-02-17 Richard Biener <rguenther@suse.de>
24023 PR tree-optimization/79552
24024 * tree-ssa-structalias.c (visit_loadstore): Properly verify
24027 2017-02-17 Richard Biener <rguenther@suse.de>
24030 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
24032 2017-02-17 Marek Polacek <polacek@redhat.com>
24034 PR middle-end/79536
24035 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
24036 (fold_negate_expr): New wrapper.
24038 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
24040 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
24041 Correct terminology and de-emphasize pre-standard behavior.
24043 2017-02-16 Alan Modra <amodra@gmail.com>
24045 PR rtl-optimization/79286
24046 * ira.c (def_dominates_uses): New function.
24047 (update_equiv_regs): Don't create an equivalence for insns that
24048 may trap where the register def does not dominate the use.
24050 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
24052 PR rtl-optimization/78127
24053 * lra.c (lra): Call lra_eliminate before finish the loop after
24056 2017-02-16 Richard Biener <rguenther@suse.de>
24058 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
24060 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
24061 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
24062 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
24063 (isl_val_int_from_wi): New function.
24064 (extract_affine_gmp): Rename to ...
24065 (extract_affine_wi): ... this, take a widest_int.
24066 (extract_affine_int): Just wrap extract_affine_wi.
24067 (add_param_constraints): Use isl_val_int_from_wi.
24068 (add_loop_constraints): Likewise, and extract_affine_wi.
24070 2017-02-15 Jeff Law <law@redhat.com>
24072 PR middle-end/79521
24073 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
24074 ira_init_register_move_cost_if_necessary.
24076 2017-02-15 Martin Sebor <msebor@redhat.com>
24078 PR middle-end/32003
24079 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
24080 removed in a prior commit.
24082 2017-02-15 Bin Cheng <bin.cheng@arm.com>
24084 PR tree-optimization/79347
24085 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
24086 counters during peeling.
24088 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
24090 * Makefile.in (site.exp): Remove "set ISLVER".
24092 2017-02-15 Jakub Jelinek <jakub@redhat.com>
24095 * real.c (real_from_integer): Call real_convert even for decimal.
24097 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
24100 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
24102 2017-02-14 Andrew Pinski <apinski@cavium.com>
24104 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
24105 cores and change the partno/implementer to be correct.
24106 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
24107 the 'B" as the implementer.
24108 * config/aarch64/aarch64-tune.md: Regenerate.
24110 2017-02-14 Carl Love <cel@us.ibm.com>
24112 * config/rs6000/rs6000.c: Add case statement entry to make the
24113 xvcvuxdsp built-in argument unsigned.
24114 * config/rs6000/vsx.md: Fix the source and return operand types so they
24115 match the instruction definitions from the ISA document. Fix typo
24116 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
24119 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
24122 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
24123 member early_clobber_alts.
24124 * lra-lives.c (reg_early_clobber_p): New.
24125 (process_bb_lives): Use it.
24126 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
24127 (debug_operand_data): Initialize early_clobber_alts.
24128 (setup_operand_alternative): Set up early_clobber_alts.
24129 (collect_non_operand_hard_regs): Ditto. Pass early clobber
24130 alternatives to new_insn_reg.
24131 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
24133 (lra_update_insn_regno_info): Pass the new arg.
24135 2017-02-14 Jakub Jelinek <jakub@redhat.com>
24137 PR middle-end/79505
24138 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
24139 (new_oacc_loop_raw): Don't clear already cleared fields.
24142 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
24143 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
24144 _mm512_prefetch_i64gather_ps): New inline functions and macros.
24146 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
24149 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
24151 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
24154 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
24155 the extra instruction to the right place to store 128-bit constant
24158 2017-02-14 Martin Sebor <msebor@redhat.com>
24160 PR middle-end/79448
24161 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
24162 warning for strings of unknown length.
24164 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
24166 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
24168 2017-02-14 Jeff Law <law@redhat.com>
24171 * ira-costs.c (scan_one_insn): Initialize register move costs
24172 for pseudos seen in USE/CLOBBER insns.
24174 PR tree-optimization/79095
24175 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
24176 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
24177 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
24178 if the operands are known to be not equal, then the resulting range
24180 (intersect_ranges): If the new range is ~[0,0] and the old range is
24181 wide, then prefer ~[0,0].
24182 * tree-vrp.c (overflow_comparison_p_1): New function.
24183 (overflow_comparison_p): New function.
24184 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
24185 if NAME is used in an overflow test.
24186 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
24187 overflow check that can be expressed as an equality test, then adjust
24188 ops to be that equality test.
24190 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24192 * config/s390/s390-builtin-types.def: Remove flags argument.
24193 * config/s390/s390.c (s390_init_builtins): Likewise.
24195 2017-02-14 Martin Liska <mliska@suse.cz>
24197 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
24198 vector. Fix trailing white spaces.
24200 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
24202 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
24205 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24207 PR rtl-optimization/68664
24208 * config/arm/arm.c (arm_sched_can_speculate_insn):
24209 New function. Declare prototype.
24210 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
24212 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24214 PR rtl-optimization/68664
24215 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
24217 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
24219 2017-02-14 Amit Pawar <amit.pawar@amd.com>
24221 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
24222 max skip bytes for function, loop and jump.
24224 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24226 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
24227 ABS_EXPR for gimple dump.
24229 2017-02-14 Jakub Jelinek <jakub@redhat.com>
24232 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
24234 PR tree-optimization/79408
24235 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
24236 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
24237 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
24238 also if rhs1 is INTEGER_CST.
24240 2017-02-14 Richard Biener <rguenther@suse.de>
24242 PR middle-end/79432
24243 * tree-into-ssa.c (insert_phi_nodes): When the function can
24244 have abnormal edges rewrite SSA names with broken use-def
24245 dominance out of SSA and register them for PHI insertion.
24247 2017-02-13 Martin Sebor <msebor@redhat.com>
24249 PR middle-end/79496
24250 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
24251 clearing info.nowrite flag when snprintf size argument is a range.
24253 2017-02-13 Jakub Jelinek <jakub@redhat.com>
24255 * cprop.c (cprop_jump): Add missing space in string literal.
24256 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
24257 (get_constraint_for_component_ref): Likewise.
24258 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
24259 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
24260 * lra-constraints.c (process_alt_operands): Likewise.
24261 * ipa-inline.c (inline_small_functions): Likewise.
24262 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
24263 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
24264 * trans-mem.c (diagnose_tm_1_op): Likewise.
24265 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
24266 (grid_parallel_clauses_gridifiable): Likewise.
24268 * config/nvptx/mkoffload.c (process): Add space in between
24271 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
24272 "MOD4_SSE_REGS" and "ALL_REGS".
24274 * spellcheck.c (test_data): Add , in between "foo" and "food".
24276 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24279 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
24280 boundary crossing check and subsequent code generation agree.
24282 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24284 * config/aarch64/aarch64.c (has_memory_op): Delete.
24285 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
24288 2017-02-13 Jakub Jelinek <jakub@redhat.com>
24290 PR rtl-optimization/79388
24291 PR rtl-optimization/79450
24292 * combine.c (distribute_notes): When removing TEM_INSN for which
24293 corresponding dest has last value recorded, invalidate that last
24296 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24298 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
24299 of explicit '@'. Add missing assembly comment marker on branch costs
24302 2017-02-13 Nathan Sidwell <nathan@acm.org>
24304 * gengtype-lex.l (<in_struct>): Add '/'.
24306 2017-02-13 Martin Liska <mliska@suse.cz>
24309 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
24311 2017-02-13 Richard Biener <rguenther@suse.de>
24313 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
24315 * configure: Re-generate.
24316 * config.in: Likewise.
24317 * graphite-dependences.c: Simplify as if
24318 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
24319 * graphite-isl-ast-to-gimple.c: Likewise.
24320 * graphite-optimize-isl.c: Likewise.
24321 * graphite-poly.c: Likewise.
24322 * graphite-sese-to-poly.c: Likewise.
24323 * graphite.h: Likewise.
24324 * toplev.c: Include isl/version.h and use isl_version () for
24325 printing the ISL version.
24326 * doc/install.texi: Update ISL requirement.
24328 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
24330 * doc/standards.texi (Standards): Update reference to
24333 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
24335 * doc/extend.texi (Named Address Spaces): sourceware.org now
24337 * doc/install.texi (Binaries): Ditto.
24340 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
24342 * doc/cpp.texi: Replace "stringify"/"stringification" with C
24343 standard terminology "stringize"/"stringizing" throughout.
24344 * doc/cppinternals.texi: Likewise.
24346 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
24348 * doc/extend.texi: Fix some spelling mistakes and typos.
24349 * doc/invoke.texi: Likewise.
24351 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
24354 * params.def (inline-min-speedup) Change from 10 to 8.
24356 2017-02-11 Jakub Jelinek <jakub@redhat.com>
24358 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
24361 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
24364 * ipa-inline-analysis.c (get_minimal_bb): New function.
24365 (record_modified): Use it.
24366 (remap_edge_change_prob): Handle also ancestor functions.
24368 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
24370 * doc/contrib.texi (Contributors): Remove broken link into
24371 the Mauve CVS repository.
24373 2017-02-11 Jakub Jelinek <jakub@redhat.com>
24375 PR middle-end/79454
24376 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
24377 result computation whenever lhs doesn't have vector mode, not
24378 just when it has BLKmode.
24380 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
24382 * doc/makefile.texi (profiledbootstrap): Refer to the
24383 installation instructions only in textual form.
24385 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24388 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
24390 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
24392 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
24393 (Specific): Update mingw-w64 reference.
24395 (Specific): Remove broken link to Renesas RX processor.
24397 2017-02-10 Richard Biener <rguenther@suse.de>
24399 * toplev.c (process_options): Do not mention obsolete graphite
24400 options when printing sorry message about missing graphite support.
24401 Mention -floop-nest-optimize.
24403 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
24405 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
24406 (vtst_p16): Likewise.
24407 (vtstq_p8): Likewise.
24408 (vtstq_p16): Likewise.
24410 (vtstq_p64): Likewise.
24411 * config/arm/arm_neon.h (vgetq_lane_p64): New.
24412 (vset_lane_p64): New.
24413 (vsetq_lane_p64): New.
24415 2017-02-10 Jakub Jelinek <jakub@redhat.com>
24417 PR tree-optimization/79411
24418 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
24419 stmt operands are SSA_NAMEs used in abnormal phis.
24420 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
24423 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
24426 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
24429 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
24431 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
24433 2017-02-09 Jakub Jelinek <jakub@redhat.com>
24435 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
24439 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
24440 not arbitrary TREE_CONSTANT.
24443 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
24444 "omp declare target link" attribute unless is_global_var.
24445 * omp-offload.c (find_link_var_op): Likewise.
24447 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
24448 Chung-Lin Tang <cltang@codesourcery.com>
24450 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
24452 (gimplify_adjust_omp_clauses): Don't delete TILE.
24453 (gimplify_omp_for): Deal with TILE.
24454 * internal-fn.c (expand_GOACC_TILE): New function.
24455 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
24457 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
24458 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
24460 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
24461 avoid DIV for outermost collapse var.
24462 (expand_oacc_for): Insert tile element loop as needed. Adjust.
24463 Remove out of date comments, fix whitespace.
24464 * omp-general.c (omp_extract_for_data): Deal with tiling.
24465 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
24466 adjust OLF_DIM_BASE value.
24467 (struct omp_for_data): Add tiling field.
24468 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
24469 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
24470 for auto loops. Remove default auto determining, moved to
24471 oacc_loop_fixed_partitions.
24472 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
24473 stmts, add e_mask field.
24474 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
24475 (oacc_thread_numbers): Use oacc_dim_call.
24476 (oacc_xform_tile): New.
24477 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
24478 (finish_oacc_loop): Adjust for ifns vector.
24479 (oacc_loop_discover_walk): Append loop abstraction sites to list,
24480 add case for GOACC_TILE fns.
24481 (oacc_loop_xform_loop): Delete.
24482 (oacc_loop_process): Iterate over call list directly, and add
24483 handling for GOACC_TILE fns.
24484 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
24486 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
24487 vector partitioning to outer loops. Assign 2 partitions to loops
24488 when available. Add TILE handling.
24489 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
24490 (execite_oacc_device_lower): Process GOACC_TILE fns,
24491 ignore unknown specs.
24492 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
24493 * tree.c (omp_clause_num_ops): Adjust TILE ops.
24494 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
24496 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
24498 * configure.ac (ACX_BUGURL): Update.
24499 * configure: Regenerate.
24501 2017-02-09 Richard Biener <rguenther@suse.de>
24503 PR tree-optimization/69823
24504 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
24505 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
24507 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
24509 * config/arc/arc-c.def: Add __NPS400__ definition.
24510 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
24511 (TARGET_NPS400): Define.
24513 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
24515 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
24517 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
24518 pointer, arch_info.
24519 (arc_cpu_types): Fill the arch_info field with a pointer into the
24520 arc_arch_types table.
24521 (arc_selected_cpu): Declare.
24522 * config/arc/arc.c (arc_selected_cpu): Make global.
24523 (arc_selected_arch): Delete.
24524 (arc_base_cpu): Delete.
24525 (arc_override_options): Remove references to deleted variables,
24526 update access to arch information.
24527 (ARC_OPT): Update access to arch information.
24528 (ARC_OPTX): Likewise.
24529 * config/arc/arc.h (arc_base_cpu): Remove declaration.
24530 (TARGET_ARC600): Update access to arch information.
24531 (TARGET_ARC601): Likewise.
24532 (TARGET_ARC700): Likewise.
24533 (TARGET_EM): Likewise.
24534 (TARGET_HS): Likewise.
24535 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
24538 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
24541 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
24542 condition/operands for integer GE/LE/GEU/LEU operations.
24544 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
24546 PR translation/79397
24547 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
24550 2017-02-08 Martin Jambor <mjambor@suse.cz>
24553 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
24554 whether allocation happened.
24555 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
24556 nothing was allocated.
24558 2017-02-08 Jakub Jelinek <jakub@redhat.com>
24560 PR tree-optimization/79408
24561 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
24562 constant, but SSA_NAME with a known integer range, use the minimum
24563 of that range instead of op1 to determine if modulo can be replaced
24564 with its first operand.
24566 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24568 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
24570 2017-02-08 Richard Biener <rguenther@suse.de>
24572 PR tree-optimization/71824
24573 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
24574 Check all loops contained in the merged region.
24576 2017-02-07 Andrew Pinski <apinski@cavium.com>
24578 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
24580 2017-02-07 Andrew Pinski <apinski@cavium.com>
24582 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
24583 (thunderxt88): Likewise.
24584 (thunderxt81): Disable LSE and change v8.1 to v8.
24585 (thunderxt83): Likewise.
24587 2017-02-07 Jakub Jelinek <jakub@redhat.com>
24588 Richard Biener <rguenther@suse.de>
24590 PR middle-end/79399
24591 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
24592 type from int to size_t.
24593 * ira-costs.c (struct_costs_size): Change type from int to size_t.
24595 2017-02-07 Jakub Jelinek <jakub@redhat.com>
24597 PR rtl-optimization/79386
24598 * cprop.c (bypass_conditional_jumps): Initialize
24599 bypass_last_basic_block already before splitting bbs after
24600 unconditional traps...
24601 (bypass_conditional_jumps): ... rather than here.
24604 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
24605 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
24606 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
24607 fix -masm=intel patterns.
24609 2017-02-07 Richard Biener <rguenther@suse.de>
24611 PR tree-optimization/79256
24612 PR middle-end/79278
24613 * builtins.c (get_object_alignment_2): Use min_align_of_type
24614 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
24615 and ADJUST_FIELD_ALIGN.
24617 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
24619 * doc/tm.texi: Regenerate.
24620 * stor-layout.c (layout_decl): Adjust.
24621 (update_alignment_for_field): Likewise.
24622 (place_field): Likewise.
24623 (min_align_of_type): Likewise.
24624 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
24625 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
24626 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
24627 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
24628 * config/frv/frv.c (frv_adjust_field_align): Likewise.
24629 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
24630 * config/i386/i386.c (x86_field_alignment): Likewise.
24631 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
24632 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
24633 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
24634 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
24635 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
24636 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
24640 2017-01-30 Richard Biener <rguenther@suse.de>
24642 PR tree-optimization/79256
24643 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
24644 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
24647 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
24649 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
24650 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
24651 builtins to SImode and emit a zero-extend, if necessary.
24653 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
24655 * docs/invoke.texi (RISC-V Options): Alphabetize.
24657 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
24659 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
24662 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
24664 * config/riscv/riscv.c: New file.
24665 * common/config/riscv/riscv-common.c: Likewise.
24666 * config.gcc: Likewise.
24667 * config/riscv/constraints.md: Likewise.
24668 * config/riscv/elf.h: Likewise.
24669 * config/riscv/generic.md: Likewise.
24670 * config/riscv/linux.h: Likewise.
24671 * config/riscv/multilib-generator: Likewise.
24672 * config/riscv/peephole.md: Likewise.
24673 * config/riscv/pic.md: Likewise.
24674 * config/riscv/predicates.md: Likewise.
24675 * config/riscv/riscv-builtins.c: Likewise.
24676 * config/riscv/riscv-c.c: Likewise.
24677 * config/riscv/riscv-ftypes.def: Likewise.
24678 * config/riscv/riscv-modes.def: Likewise.
24679 * config/riscv/riscv-opts.h: Likewise.
24680 * config/riscv/riscv-protos.h: Likewise.
24681 * config/riscv/riscv.h: Likewise.
24682 * config/riscv/riscv.md: Likewise.
24683 * config/riscv/riscv.opt: Likewise.
24684 * config/riscv/sync.md: Likewise.
24685 * config/riscv/t-elf-multilib: Likewise.
24686 * config/riscv/t-linux: Likewise.
24687 * config/riscv/t-linux-multilib: Likewise.
24688 * config/riscv/t-riscv: Likewise.
24689 * configure.ac: Likewise.
24690 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
24691 Waterman as RISC-V maintainers.
24692 * doc/install.texi: Add RISC-V entries.
24693 * doc/invoke.texi: Add RISC-V options section.
24694 * doc/md.texi: Add RISC-V constraints section.
24695 * configure: Regenerated.
24697 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
24700 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
24701 false values to be constant vectors with all 0 or all 1 bits set.
24702 (vcondu<mode><mode>): Likewise.
24703 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
24705 (fpmask_comparison_operator): Update comment.
24706 (vecint_comparison_operator): New predicate.
24707 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
24708 vector conditionals when the true and false values are constant
24709 vectors with all 0 bits or all 1 bits set.
24711 2017-02-06 Martin Sebor <msebor@redhat.com>
24713 PR tree-optimization/79376
24714 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
24716 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
24718 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
24719 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
24720 to simplify split condition.
24722 2017-02-06 Jakub Jelinek <jakub@redhat.com>
24724 * omp-expand.c (oxpand_omp_atomic_fetch_op,
24725 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
24728 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
24730 PR rtl-optimization/68664
24731 * target.def (can_speculate_insn): New hook.
24732 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
24733 * doc/tm.texi: Regenerate.
24734 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
24735 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
24736 (rs6000_sched_can_speculate_insn): New function.
24738 2017-02-06 Jakub Jelinek <jakub@redhat.com>
24740 PR tree-optimization/79284
24741 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
24742 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
24743 vectorizable_mask_load_store, vectorizable_operation,
24744 vect_is_simple_cond, get_same_sized_vectype): Use it instead
24745 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
24746 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
24747 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
24748 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
24749 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
24750 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
24751 is_gimple_assign (stmt). Replace another such test with
24752 is_gimple_assign (stmt).
24754 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
24757 * config/avr/avr.c (rtl-iter.h): Include it.
24758 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
24759 (avr_legitimate_combined_insn): ...and implementation.
24761 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
24763 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
24764 * config/s390/s390.c (s390_const_operand_ok)
24765 (s390_canonicalize_comparison, s390_extract_part)
24766 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
24767 (s390_contiguous_bitmask_p, s390_rtx_costs)
24768 (legitimize_pic_address): Likewise.
24769 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
24770 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
24771 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
24772 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
24773 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
24775 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
24777 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
24778 REGNO($0) == REGNO($1).
24780 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24782 * config/s390/linux.h(SIZE_TYPE): Add comment.
24784 2017-02-06 Julian Brown <julian@codesourcery.com>
24785 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24786 Virendra Pathak <virendra.pathak@broadcom.com>
24788 * config/aarch64/aarch64-cores.def: Change the scheduler
24790 * config/aarch64/aarch64.md: Include thunderx2t99.md.
24791 * config/aarch64/thunderx2t99.md: New file.
24793 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
24795 * doc/standards.texi (Go Language): Update link to language
24798 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
24800 * tree-eh.c (lower_resx): Sanitize profile.
24801 (cleanup_empty_eh_move_lp): Likewise.
24803 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
24806 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
24808 * cfgloopmanip.h (loop_version): Update prototype.
24809 * modulo-sched.c (sms_schedule): Update call of loop_version.
24810 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
24811 * tree-parloops.c (gen_parallel_loop): Likewise.
24812 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
24813 * tree-ssa-loop-split.c (split_loop): Likewise.
24814 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
24815 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
24817 2017-02-05 Martin Liska <mliska@suse.cz>
24820 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
24822 (print_operand_address): Initialize a struct to zero.
24824 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
24826 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
24827 garbage collector only in textual form.
24829 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
24831 * doc/extend.texi (x86 specific memory model extensions for
24832 transactional memory): Simplify a phrase.
24834 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
24837 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
24838 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
24839 (atomic_storedi_1): Likewise.
24841 2017-02-04 Jakub Jelinek <jakub@redhat.com>
24843 PR tree-optimization/79338
24844 * tree-parloops.c (gather_scalar_reductions): Don't call
24845 vect_analyze_loop_form for loop->inner before destroying loop's
24848 2017-02-03 Martin Sebor <msebor@redhat.com>
24850 PR tree-optimization/79327
24851 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
24852 when precision has resulted in leading zeros.
24853 (format_integer): Adjust the likely counter to assume an unknown
24854 argument that may be zero is non-zero.
24856 2017-02-03 Jason Merrill <jason@redhat.com>
24859 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
24860 avoid copying non-taken branch.
24862 2017-02-03 Jakub Jelinek <jakub@redhat.com>
24864 PR tree-optimization/79340
24865 * tree-vect-loop.c (vectorizable_reduction): Release
24866 vec_defs elements after safe_splicing them into other vectors.
24869 PR tree-optimization/79327
24870 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
24871 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
24873 (format_integer): Use wide_int_to_tree instead of build_int_cst
24874 + to_?hwi. If argmin is NULL, just set argmin and argmax to
24875 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
24876 of shortest and longest sequence.
24878 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
24880 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
24881 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
24883 2017-02-03 Walter Lee <walt@tilera.com>
24886 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
24887 after initial stackframe link reg save.
24888 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
24890 2017-02-03 Jakub Jelinek <jakub@redhat.com>
24893 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
24894 wu for stxssp alternative.
24896 2017-02-03 Martin Sebor <msebor@redhat.com>
24898 PR tree-optimization/79352
24899 * gimple-fold.c (get_range_strlen): Add argument.
24900 (get_range_strlen): Change return type to bool.
24901 (get_maxval_strlen): Pass in a dummy argument.
24902 * gimple-fold.h (get_range_strlen): Change return type to bool.
24903 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
24904 * tree.h (array_at_struct_end_p): Add argument.
24905 * tree.c (array_at_struct_end_p): Handle it.
24907 2017-02-03 Martin Liska <mliska@suse.cz>
24910 * multiple_target.c (create_dispatcher_calls): Redirect edge
24911 from a caller of a dispatcher.
24912 (expand_target_clones): Make the clones local.
24913 (ipa_target_clone): Do both target clones and resolvers.
24914 (ipa_dispatcher_calls): Remove the pass.
24915 (pass_dispatcher_calls::gate): Likewise.
24916 (make_pass_dispatcher_calls): Likewise.
24917 * passes.def (pass_target_clone): Put as very first IPA early
24920 2017-02-03 Martin Liska <mliska@suse.cz>
24922 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
24923 in case of a function with ifunc attribute.
24925 2017-02-03 Martin Liska <mliska@suse.cz>
24927 * cgraph.c (cgraph_node::dump): Dump function version info.
24928 * symtab.c (symtab_node::dump_base): Add missing new line.
24930 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
24932 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
24933 (ifcombine_ifandif): Use it.
24935 2017-02-03 Martin Liska <mliska@suse.cz>
24937 * doc/invoke.texi: Document default value for
24938 use-after-scope-direct-emission-threshold.
24940 2017-02-03 Martin Liska <mliska@suse.cz>
24942 PR tree-optimization/79339
24943 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
24944 (format_floating): Likewise.
24946 2017-02-03 Martin Liska <mliska@suse.cz>
24949 * ipa-prop.c (ipa_node_params_t::insert): Remove current
24951 (ipa_node_params_t::remove): Likewise.
24952 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
24953 initialization from removed ipa_node_params_t::insert.
24954 (ipa_node_params::~ipa_node_params): Move from removed
24955 ipa_node_params_t::release.
24956 * symbol-summary.h (symbol_summary::m_released): New member.
24957 Do not release a summary twice. Do not allow to call finalizer
24958 for types of a summary that live in GGC memory.
24960 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24962 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
24965 2017-02-02 Martin Sebor <msebor@redhat.com>
24967 PR middle-end/79275
24968 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
24969 (format_string): Tighten up the range of output for non-constant
24970 strings and correct the expected range for wide non-constant strings.
24972 2017-02-02 Martin Sebor <msebor@redhat.com>
24974 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
24976 PR middle-end/32003
24977 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
24979 (-fdump-tree-@var): Add to index and document how to come up
24980 with pass-specific option and dump file names.
24981 (-fdump-passes): Clarify where to look for output.
24983 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
24985 PR middle-end/77445
24986 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
24987 statistics of the analyzed path; allow threading for speed when
24988 any of BBs along the path are optimized for speed.
24990 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
24992 PR middle-end/78468
24993 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
24994 settings of the virtual registers.
24997 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
24999 * explow.c (get_dynamic_stack_size): Take known alignment of stack
25000 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
25003 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25005 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
25006 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
25008 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25010 * config/s390/s390.md: Add missing comments with the expanded
25012 * config/s390/vector.md: Likewise.
25013 * config/s390/vx-builtins.md: Likewise.
25015 2017-02-02 Jakub Jelinek <jakub@redhat.com>
25018 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
25019 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
25020 conditions on a single line.
25022 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25024 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
25025 __S390_VX__ to __VX__.
25027 2017-02-01 Andrew Pinski <apinski@cavium.com>
25029 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
25030 stmt_info to record_stmt_cost.
25031 (vect_get_known_peeling_cost): Pass stmt_info if known to
25033 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
25034 cpu_vector_cost field into
25035 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
25036 field into vec_int_stmt_cost and vec_fp_stmt_cost.
25037 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
25038 splitting of scalar_stmt_cost and vec_stmt_cost.
25039 (thunderx_vector_cost): Likewise.
25040 (cortexa57_vector_cost): LIkewise.
25041 (exynosm1_vector_cost): Likewise.
25042 (xgene1_vector_cost): Likewise.
25043 (thunderx2t99_vector_cost): Improve after the splitting of the two
25045 (aarch64_builtin_vectorization_cost): Update for the splitting of
25046 scalar_stmt_cost and vec_stmt_cost.
25048 2017-02-01 Torvald Riegel <triegel@redhat.com>
25049 Richard Henderson <rth@redhat.com>
25051 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
25052 conditional on existance of a fast atomic load.
25053 * optabs-query.c (can_atomic_load_p): New function.
25054 * optabs-query.h (can_atomic_load_p): Declare it.
25055 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
25056 no fast atomic load is available for the particular size of access.
25057 (expand_atomic_compare_and_swap): Likewise.
25058 (expand_atomic_load): Likewise.
25059 (expand_atomic_store): Likewise.
25060 (expand_atomic_fetch_op): Likewise.
25061 * testsuite/lib/target-supports.exp
25062 (check_effective_target_sync_int_128): Remove x86 because it provides
25063 no fast atomic load.
25064 (check_effective_target_sync_int_128_runtime): Likewise.
25066 2017-02-01 Richard Biener <rguenther@suse.de>
25068 * graphite.c: Include tree-vectorizer.h for find_loop_location.
25069 (graphite_transform_loops): Provide opt-info for optimized nests.
25070 * tree-parloop.c (parallelize_loops): Provide opt-info for
25071 parallelized loops.
25073 2017-02-01 Richard Biener <rguenther@suse.de>
25075 PR middle-end/79315
25076 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
25077 was not set before.
25079 2017-02-01 Richard Biener <rguenther@suse.de>
25081 PR tree-optimization/71824
25082 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
25083 Verify the loops are valid in the merged SESE region.
25084 (scop_detection::can_represent_loop_1): Check analyzing the
25085 evolution of the number of iterations in the region succeeds.
25087 2017-01-31 Ian Lance Taylor <iant@golang.org>
25089 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
25090 REG_ARGS_SIZE note to 32-bit push insns and call insn.
25092 2017-01-31 David Malcolm <dmalcolm@redhat.com>
25094 PR preprocessor/79210
25095 * input.c (get_substring_ranges_for_loc): Replace line_width
25096 assertion with error-handling.
25098 2017-01-31 Richard Biener <rguenther@suse.de>
25100 PR tree-optimization/77318
25101 * graphite-sese-to-poly.c (extract_affine): Fix assert.
25102 (create_pw_aff_from_tree): Take loop parameter.
25103 (add_condition_to_pbb): Pass loop of the condition to
25104 create_pw_aff_from_tree.
25106 2017-01-31 Jakub Jelinek <jakub@redhat.com>
25108 * config/s390/s390.c (s390_asan_shadow_offset): New function.
25109 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
25111 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
25115 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
25117 (convert_int_to_float128): Likewise.
25118 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
25119 (convert_int_to_float128): Likewise.
25120 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
25121 (UNSPEC_IEEE128_CONVERT): Likewise.
25122 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
25123 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
25124 Use local variables for IBM extended format.
25125 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
25126 (fix_trunc<mode>si2_fprs): Likewise.
25127 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
25128 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
25129 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
25130 to know that we can now have integers of all sizes in vector
25132 (fix<uns>_<mode>di2_hw): Likewise.
25133 (float<uns>_<mode>si2_hw): Likewise.
25134 (fix_<mode>si2_hw): Likewise.
25135 (fixuns_<mode>si2_hw): Likewise.
25136 (float<uns>_<mode>di2_hw): Likewise.
25137 (float_<mode>di2_hw): Likewise.
25138 (float_<mode>si2_hw): Likewise.
25139 (floatuns_<mode>di2_hw): Likewise.
25140 (floatuns_<mode>si2_hw): Likewise.
25141 (xscvqp<su>wz_<mode>): Delete, no longer used.
25142 (xscvqp<su>dz_<mode>): Likewise.
25143 (xscv<su>dqp_<mode>): Likewise.
25144 (ieee128_mfvsrd_64bit): Likewise.
25145 (ieee128_mfvsrd_32bit): Likewise.
25146 (ieee128_mfvsrwz): Likewise.
25147 (ieee128_mtvsrw): Likewise.
25148 (ieee128_mtvsrd_64bit): Likewise.
25149 (ieee128_mtvsrd_32bit): Likewise.
25151 2017-01-31 Martin Liska <mliska@suse.cz>
25154 * ipa-prop.c (ipa_free_all_node_params): Call release method
25155 instead of ~sumbol_summary to not to trigger double times
25158 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
25160 PR tree-optimization/71691
25161 * bitmap.h (class auto_bitmap): New.
25162 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
25163 is_maybe_undefined instead of ssa_undefined_value_p.
25165 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25167 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
25168 __S390_ARCH_LEVEL__ to __ARCH__.
25170 2017-01-31 Jakub Jelinek <jakub@redhat.com>
25172 PR tree-optimization/79267
25173 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
25174 if should_remove_lhs_p is true.
25176 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
25179 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
25180 (add_alignment_attribute): New.
25181 (base_type_die): Add alignment attribute.
25182 (subrange_type_die): Likewise.
25183 (modified_type_die): Likewise.
25184 (gen_array_type_die): Likewise.
25185 (gen_descr_array_type_die: Likewise.
25186 (gen_enumeration_type_die): Likewise.
25187 (gen_subprogram_die): Likewise.
25188 (gen_variable_die): Likewise.
25189 (gen_field_die): Likewise.
25190 (gen_ptr_to_mbr_type_die): Likewise.
25191 (gen_struct_or_union_type_die): Likewise.
25192 (gen_subroutine_type_die): Likewise.
25193 (gen_typedef_die): Likewise.
25194 (base_type_cmp): Compare alignment attribute.
25196 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25199 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
25200 (setb_unsigned) New pattern for setb with CCUNS.
25201 * config/rs6000/rs6000.c (expand_block_compare): Use a different
25202 subfc./subfe sequence to avoid overflow problems. Generate a
25203 shorter sequence with cmpld/setb for power9.
25204 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
25205 for generating subfc. instruction.
25206 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
25207 now uses this instruction.
25209 2017-01-30 Ian Lance Taylor <iant@google.com>
25212 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
25213 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
25215 2017-01-30 Martin Sebor <msebor@redhat.com>
25217 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
25218 Move constant to the right of a relational operator.
25219 (get_mpfr_format_length, format_character, format_string): Ditto.
25220 (should_warn_p, maybe_warn): Same.
25222 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
25224 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
25227 * asan.c (get_translation_unit_decl): Remove function.
25228 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
25230 2017-01-30 Martin Liska <mliska@suse.cz>
25232 PR gcov-profile/79259
25233 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
25234 -fprofile-generate.
25236 2017-01-30 Martin Liska <mliska@suse.cz>
25239 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
25240 Initialize variables with NULL value.
25242 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
25245 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
25247 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
25249 2017-01-30 Richard Biener <rguenther@suse.de>
25251 PR tree-optimization/79276
25252 * tree-vrp.c (process_assert_insertions): Properly adjust common
25253 when removing a duplicate.
25255 * gcc.dg/torture/pr79276.c: New testcase.
25257 2017-01-30 Richard Biener <rguenther@suse.de>
25259 PR tree-optimization/79256
25260 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
25261 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
25263 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
25265 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
25268 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
25269 ("*r<noxa>sbg_<mode>_sll_bitmask")
25270 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
25271 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
25272 Use contiguous_bitmask_nowrap_operand.
25274 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25277 * config/rs6000/altivec.h (vec_xl): Revise #define.
25278 (vec_xst): Likewise.
25280 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
25282 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
25284 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
25286 PR rtl-optimization/79194
25287 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
25288 traps before call to bypass_conditional_jumps.
25290 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
25292 PR tree-optimization/71374
25293 * lra-constraints.c (check_conflict_input_operands): New.
25294 (match_reload): Use it.
25296 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
25299 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
25300 account to calculate conflict_set.
25302 2017-01-27 Bin Cheng <bin.cheng@arm.com>
25304 PR rtl-optimization/78559
25305 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
25306 other_insn in combine.
25308 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
25310 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
25311 uint16_type_node for BT_UINT16.
25313 2017-01-27 David Malcolm <dmalcolm@redhat.com>
25315 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
25316 "RTL Tests" to menu.
25317 (GIMPLE Tests): New node.
25318 (RTL Tests): New node.
25320 2017-01-27 Richard Biener <rguenther@suse.de>
25322 PR tree-optimization/79245
25323 * tree-loop-distribution.c (distribute_loop): Apply cost
25324 modeling also to detected patterns.
25326 2017-01-27 Richard Biener <rguenther@suse.de>
25328 PR tree-optimization/71433
25329 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
25330 (compare_assert_loc): New function.
25331 (process_assert_insertions): Sort and optimize assert locations
25332 to remove duplicates and push down identical assertions on
25333 edges to their destination block.
25335 2017-01-27 Richard Biener <rguenther@suse.de>
25337 PR tree-optimization/79244
25338 * tree-vrp.c (remove_range_assertions): Forcefully propagate
25339 out SSA names even if abnormal.
25341 2017-01-27 Jakub Jelinek <jakub@redhat.com>
25343 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
25344 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
25345 instead of MPFR_RNDN.
25347 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
25350 * arm.c (arm_option_override): Don't call build_target_option_node
25351 until after doing all option overrides.
25352 (arm_valid_target_attribute_tree): Likewise.
25354 2017-01-27 Martin Liska <mliska@suse.cz>
25356 * doc/invoke.texi (-fprofile-arcs): Document profiling support
25357 for {cd}tors and C++ {cd}tors.
25359 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
25361 * config/s390/s390.md ("*setmem_long_and")
25362 ("*setmem_long_and_31z"): Use zero_extend instead of and.
25364 2017-01-26 Martin Sebor <msebor@redhat.com>
25366 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
25369 2017-01-26 Martin Sebor <msebor@redhat.com>
25371 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
25372 HAVE_DFmode before using XFmode or DFmode.
25373 (parse_directive): Avoid using the z length modifier to avoid
25374 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
25376 PR middle-end/78703
25377 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
25378 to accept adjustment as an array.
25379 (get_int_range): New function.
25380 (struct directive): Make width and prec arrays.
25381 (directive::set_width, directive::set_precision): Call get_int_range.
25382 (format_integer, format_floating): Handle width and precision ranges.
25383 (format_string, parse_directive): Same.
25385 2017-01-26 Jakub Jelinek <jakub@redhat.com>
25388 * dwarf2out.c (generate_skeleton_bottom_up): For children with
25389 comdat_type_p set, just clone them, but keep the children in the
25393 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
25394 which have direct callers with -fvar-tracking-assignments enabled
25396 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
25397 inside of type units.
25399 2017-01-26 Martin Sebor <msebor@redhat.com>
25401 PR middle-end/78703
25402 * gimple-ssa-sprintf.c (struct result_range): Add likely and
25404 (struct format_result): Replace number_chars, number_chars_min,
25405 and number_chars_max with a single member of struct result_range.
25407 (format_result::operator+=): Adjust.
25408 (struct fmtresult): Remove bounded. Handle likely and unlikely
25410 (fmtresult::adjust_for_width_or_precision): New function.
25411 (fmtresult:type_max_digits): New function.
25412 (bytes_remaining): Handle likely and unlikely counters.
25413 (min_bytes_remaining): Remove.
25414 (format_percent): Simplify.
25415 (format_integer, format_floating): Set likely and unlikely counters.
25416 (get_string_length, format_character, format_string): Same.
25417 (format_plain, should_warn_p): New function.
25418 (maybe_warn): Call should_warn_p. Update diagnostic messages
25419 and handle those for all directives, including plain strings.
25420 (format_directive): Handle likely and unlikely counters.
25421 Remove unnecessary quoting from diagnostics. Add an informational
25423 (add_bytes): Remove.
25424 (pass_sprintf_length::compute_format_length): Simplify.
25425 (try_substitute_return_value): Handle likely and unlikely counters.
25427 2017-01-26 Carl Love <cel@us.ibm.com>
25429 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
25430 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
25432 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
25435 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
25436 endianess for subregs into account.
25437 * lra-constraints.c (lra_constraints): Do risky transformations
25438 always on the first iteration.
25439 * lra-lives.c (check_pseudos_live_through_calls): Add arg
25440 last_call_used_reg_set.
25441 (process_bb_lives): Define and use last_call_used_reg_set.
25442 * lra.c (lra): Always continue after lra_constraints on the first
25445 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
25447 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
25449 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
25451 2017-01-26 Jakub Jelinek <jakub@redhat.com>
25453 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
25454 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
25455 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
25456 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
25457 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
25458 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
25459 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
25460 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
25461 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
25463 2017-01-26 Marek Polacek <polacek@redhat.com>
25466 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
25467 for the third operand.
25469 2017-01-26 Jakub Jelinek <jakub@redhat.com>
25471 PR middle-end/79236
25472 * omp-low.c (struct omp_context): Add simt_stmt field.
25473 (scan_omp_for): Return omp_context *.
25474 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
25475 context to the _simt_ SIMD stmt.
25476 (lower_omp_for): For combined SIMD with sibling _simt_
25477 SIMD, make sure to use the same decls in _looptemp_
25478 clauses as in the sibling.
25480 2017-01-26 David Sherwood <david.sherwood@arm.com>
25482 PR middle-end/79212
25483 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
25486 2017-01-26 Jakub Jelinek <jakub@redhat.com>
25489 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
25490 emit fld b; fld a; if possible.
25492 * brig-builtins.def: Update copyright years.
25493 * config/arm/arm_acle_builtins.def: Update copyright years.
25495 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
25498 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
25499 constraint instead of o for the stxsd instruction.
25501 2017-01-25 Carl Love <cel@us.ibm.com>
25503 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
25504 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
25506 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
25508 * doc/invoke.texi (C++ Dialect Options): Fix typo.
25510 2017-01-25 Richard Biener <rguenther@suse.de>
25512 PR tree-optimization/69264
25513 * target.def (vector_alignment_reachable): Improve documentation.
25514 * doc/tm.texi: Regenerate.
25515 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
25517 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
25518 earlier changes with respect to TYPE_USER_ALIGN.
25519 (vector_alignment_reachable_p): Likewise. Improve dumping.
25521 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25524 * config/arm/arm.md (xordi3): Force constant operand into a register
25527 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25529 * doc/invoke.texi (-fstore-merging): Correct default optimization
25530 levels at which it is enabled.
25531 (-O): Move -fstore-merging from list to...
25534 2017-01-25 Richard Biener <rguenther@suse.de>
25537 * omp-expand.c: Include debug.h.
25538 (expand_omp_taskreg): Make sure to generate early debug before
25539 outlining anything from a function.
25540 (expand_omp_target): Likewise.
25541 (grid_expand_target_grid_body): Likewise.
25543 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
25546 * asan.c (get_translation_unit_decl): New function.
25547 (asan_add_global): Extract modules file name from globals
25548 TRANSLATION_UNIT_DECL name.
25550 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
25553 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
25554 for long calls with APCS frame and VFP.
25556 2017-01-24 David Malcolm <dmalcolm@redhat.com>
25558 * cfg.c (original_copy_tables_initialized_p): New function.
25559 * cfg.h (original_copy_tables_initialized_p): New decl.
25560 * cfgrtl.c (relink_block_chain): Guard the call to
25561 free_original_copy_tables with a call to
25562 original_copy_tables_initialized_p.
25563 * cgraph.h (symtab_node::native_rtl_p): New decl.
25564 * cgraphunit.c (symtab_node::native_rtl_p): New function.
25565 (symtab_node::needed_p): Don't assert for early assembly output
25566 for __RTL functions.
25567 (cgraph_node::finalize_function): Set "force_output" for __RTL
25569 (cgraph_node::analyze): Bail out early for __RTL functions.
25570 (analyze_functions): Update assertion to support __RTL functions.
25571 (cgraph_node::expand): Bail out early for __RTL functions.
25572 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
25574 * function.h (struct function): Update comment for field
25576 * gimple-expr.c: Include "tree-pass.h".
25577 (gimple_has_body_p): Return false for __RTL functions.
25578 * Makefile.in (OBJS): Add run-rtl-passes.o.
25579 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
25581 (gcc::pass_manager::get_clean_slate): New accessor.
25582 * passes.c: Include "insn-addr.h".
25583 (should_skip_pass_p): Add logging. Update logic for running
25584 "expand" to be compatible with both __GIMPLE and __RTL. Guard
25585 property-provider override so it is only done for gimple passes.
25587 (skip_pass): New function.
25588 (execute_one_pass): Call skip_pass when skipping passes.
25589 * read-md.c (md_reader::read_char): Support filtering
25590 the input to a subset of line numbers.
25591 (md_reader::md_reader): Initialize fields
25592 m_first_line and m_last_line.
25593 (md_reader::read_file_fragment): New function.
25594 * read-md.h (md_reader::read_file_fragment): New decl.
25595 (md_reader::m_first_line): New field.
25596 (md_reader::m_last_line): New field.
25597 * read-rtl-function.c (function_reader::create_function): Only
25598 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
25599 curr_properties. Set DECL_INITIAL to a dummy block.
25600 (read_rtl_function_body_from_file_range): New function.
25601 * read-rtl-function.h (read_rtl_function_body_from_file_range):
25603 * run-rtl-passes.c: New file.
25604 * run-rtl-passes.h: New file.
25606 2017-01-24 Jeff Law <law@redhat.com>
25608 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
25611 2017-01-24 Bin Cheng <bin.cheng@arm.com>
25613 PR tree-optimization/79159
25614 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
25615 (record_nonwrapping_iv): Improve boundary using above function if no
25616 value range information.
25618 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
25619 Martin Jambor <mjambor@suse.cz>
25621 * brig-builtins.def: New file.
25622 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
25623 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
25624 (DEF_HSAIL_SAT_BUILTIN): Likewise.
25625 (DEF_HSAIL_INTR_BUILTIN): Likewise.
25626 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
25627 * builtin-types.def (BT_INT8): New.
25628 (BT_INT16): Likewise.
25629 (BT_UINT8): Likewise.
25630 (BT_UINT16): Likewise.
25631 (BT_FN_ULONG): Likewise.
25632 (BT_FN_UINT_INT): Likewise.
25633 (BT_FN_UINT_ULONG): Likewise.
25634 (BT_FN_UINT_LONG): Likewise.
25635 (BT_FN_UINT_PTR): Likewise.
25636 (BT_FN_ULONG_PTR): Likewise.
25637 (BT_FN_INT8_FLOAT): Likewise.
25638 (BT_FN_INT16_FLOAT): Likewise.
25639 (BT_FN_UINT32_FLOAT): Likewise.
25640 (BT_FN_UINT16_FLOAT): Likewise.
25641 (BT_FN_UINT8_FLOAT): Likewise.
25642 (BT_FN_UINT64_FLOAT): Likewise.
25643 (BT_FN_UINT16_UINT32): Likewise.
25644 (BT_FN_UINT32_UINT16): Likewise.
25645 (BT_FN_UINT16_UINT16_UINT16): Likewise.
25646 (BT_FN_INT_PTR_INT): Likewise.
25647 (BT_FN_UINT_PTR_UINT): Likewise.
25648 (BT_FN_LONG_PTR_LONG): Likewise.
25649 (BT_FN_ULONG_PTR_ULONG): Likewise.
25650 (BT_FN_VOID_UINT64_UINT64): Likewise.
25651 (BT_FN_UINT8_UINT8_UINT8): Likewise.
25652 (BT_FN_INT8_INT8_INT8): Likewise.
25653 (BT_FN_INT16_INT16_INT16): Likewise.
25654 (BT_FN_INT_INT_INT): Likewise.
25655 (BT_FN_UINT_FLOAT_UINT): Likewise.
25656 (BT_FN_FLOAT_UINT_UINT): Likewise.
25657 (BT_FN_ULONG_UINT_UINT): Likewise.
25658 (BT_FN_ULONG_UINT_PTR): Likewise.
25659 (BT_FN_ULONG_ULONG_ULONG): Likewise.
25660 (BT_FN_UINT_UINT_UINT): Likewise.
25661 (BT_FN_VOID_UINT_PTR): Likewise.
25662 (BT_FN_UINT_UINT_PTR: Likewise.
25663 (BT_FN_UINT32_UINT64_PTR): Likewise.
25664 (BT_FN_INT_INT_UINT_UINT): Likewise.
25665 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
25666 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
25667 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
25668 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
25669 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
25670 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
25671 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
25672 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
25673 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
25674 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
25675 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
25676 * doc/frontends.texi: List BRIG FE.
25677 * doc/install.texi (Testing): Add BRIG tesring requirements.
25678 * doc/invoke.texi (Overall Options): Mention BRIG.
25679 * doc/standards.texi (Standards): Doucment BRIG HSA version.
25681 2017-01-24 Richard Biener <rguenther@suse.de>
25683 PR translation/79208
25684 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
25686 2017-01-24 Martin Jambor <mjambor@suse.cz>
25689 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
25690 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
25691 and known_contexts.
25693 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
25695 PR middle-end/79123
25696 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
25697 casts from signed to unsigned really don't have a range.
25699 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
25701 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
25702 GMP_RNDx for compatiblity.
25704 2017-01-24 Martin Liska <mliska@suse.cz>
25707 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
25708 that would prevent us to call alloca with -1 as argument.
25710 2017-01-24 Jakub Jelinek <jakub@redhat.com>
25712 * dwarf2out.c (output_compilation_unit_header, output_file_names):
25713 Avoid -Wformat-security warning.
25715 2017-01-23 Andrew Pinski <apinski@cavium.com>
25717 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
25720 2017-01-23 Martin Sebor <msebor@redhat.com>
25722 PR middle-end/78703
25723 * gimple-ssa-sprintf.c (warn_level): New global.
25724 (format_integer): Use it here and throughout the rest of the file.
25725 Use the same switch to compute sign as base.
25726 (maybe_warn): New function.
25727 (format_directive): Factor out warnings into maybe_warn.
25728 Add debugging output. Use warn_level.
25729 (add_bytes): Use warn_level.
25730 (pass_sprintf_length::compute_format_length): Add debugging output.
25731 (try_substitute_return_value): Same.
25732 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
25734 PR middle-end/78703
25735 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
25736 (struct fmtresult, format_integer, format_floating): Adjust.
25737 (fmtresult::fmtresult): Set max correctly in two argument ctor.
25738 (get_string_length, format_string,format_directive): Same.
25739 (pass_sprintf_length::compute_format_length): Same.
25740 (try_substitute_return_value): Simplify slightly.
25742 PR middle-end/78703
25743 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
25744 (fmtresult::operator+=): Outlined.
25745 (struct fmtresult): Add ctors.
25746 (struct conversion_spec): Rename...
25747 (struct directive): ...to this. Add and remove data members.
25748 (directive::set_width, directive::set_precision): New functions.
25749 (format_percent): Use fmtresult ctor.
25750 (get_width_and_precision): Remove.
25751 (format_integer): Make naming changes. Avoid computing width and
25753 (format_floating): Same. Adjust indentation.
25754 (format_character, format_none): New functions.
25755 (format_string): Moved character handling to format_character.
25756 (format_directive): Remove arguments, change return type.
25757 (parse_directive): New function.
25758 (pass_sprintf_length::compute_format_length): Move directive
25759 parsing to parse_directive.
25761 2017-01-23 Jakub Jelinek <jakub@redhat.com>
25763 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
25764 (assign_assembler_name_if_needed): ... this.
25765 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
25766 (assign_assembler_name_if_needed): ... this.
25767 (free_lang_data_in_cgraph): Adjust callers.
25768 * cgraphunit.c (cgraph_node::analyze): Likewise.
25769 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
25772 2017-01-23 Richard Biener <rguenther@suse.de>
25774 PR tree-optimization/79088
25775 PR tree-optimization/79188
25776 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
25777 resetting loop bounds after last path deletion. Reset loop
25778 bounds of the target loop, make code match the comments.
25779 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
25780 Make sure loops need no fixups.
25782 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
25784 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
25785 exponent support with double type for first argument.
25786 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
25787 type returned by __builtin_vec_extract_sig,
25788 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
25789 functions from "vector int" to "vector unsigned int" or from
25790 "vector long long int" to "vector unsigned long long int".
25791 Changed type returned by __builtin_vec_extract_exp,
25792 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
25793 functions from "vector int" to "vector unsigned int" or from
25794 "vector long long int" to "vector unsigned long long int".
25795 Changed return type of __builtin_vec_test_data_class,
25796 __builtin_vec_test_data_class_sp, and
25797 __builtin_vec_test_data_class_dp from "vector int" to
25798 "vector bool int" or from "vector long long int" to "vector bool
25799 long long int" and changed second argument type from "unsigned
25800 int" to "int". Added new overloaded function forms "vector float
25801 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
25802 "vector float __builtin_vec_insert_exp_sp (vector float, vector
25803 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
25804 double, vector unsigned long long int)" and "vector double
25805 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
25806 long int)". Changed return type of
25807 __builtin_scalar_test_data_class and
25808 __builtin_scalar_test_data_class_sp and
25809 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
25810 int" and changed second argument from "unsigned int" to "int".
25811 Changed type returned by __builtin_scalar_test_neg,
25812 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
25813 from "int" to "bool int". Added new overloaded function form
25814 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
25815 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
25816 exponent double-precision with floating point first argument.
25817 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
25818 documentation of scalar_test_data_class, scalar_test_neg,
25819 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
25820 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
25821 vec_test_data_class built-in functions to reflect refinements in
25822 their type signatures.
25824 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
25826 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
25828 (aarch64_elf_asm_destructor): Likewise.
25830 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
25832 PR rtl-optimization/78634
25833 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
25834 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
25835 * ifcvt.c (noce_try_cmove): Add missing cost check.
25837 PR rtl-optimization/71724
25838 * combine.c (if_then_else_cond): Look for situations where it is
25839 beneficial to undo the work of one of the recursive calls.
25841 2017-01-23 Bin Cheng <bin.cheng@arm.com>
25843 PR tree-optimization/70754
25844 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
25845 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
25846 combined stmt before it if not NULL.
25847 (combine_chains): Process refs reversely and compute dominance point
25850 2017-01-23 Martin Liska <mliska@suse.cz>
25852 PR tree-optimization/79196
25853 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
25854 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
25856 (strlen_optimize_stmt): Call the renamed function.
25858 2017-01-23 Michael Matz <matz@suse.de>
25860 PR tree-optimization/78384
25861 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
25863 2017-01-23 Richard Biener <rguenther@suse.de>
25865 PR tree-optimization/79186
25866 * tree-vrp.c (register_new_assert_for): Make sure we've seen
25867 both incoming edges before moving an assert.
25869 2017-01-23 Martin Jambor <mjambor@suse.cz>
25871 * ipa-prop.c (load_from_param_1): Removed.
25872 (load_from_unmodified_param): Bits from load_from_param_1 put back
25874 (load_from_param): Removed.
25875 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
25876 with stmt. Reverted back to use of load_from_unmodified_param.
25878 2017-01-23 Martin Jambor <mjambor@suse.cz>
25881 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
25882 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
25883 field a pointer to garbage collected vector, mark lattices and
25884 ipcp_orig_node with GTY((skip)).
25885 (ipa_get_param_count): Adjust to descriptors being a pointer.
25886 (ipa_get_param): Likewise.
25887 (ipa_get_type): Likewise.
25888 (ipa_get_param_move_cost): Likewise.
25889 (ipa_set_param_used): Likewise.
25890 (ipa_get_controlled_uses): Likewise.
25891 (ipa_set_controlled_uses): Likewise.
25892 (ipa_is_param_used): Likewise.
25893 (ipa_node_params_t): Move into garbage collector. New methods insert
25895 (ipa_node_params_sum): Annotate wth GTY(()).
25896 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
25898 (ipa_load_from_parm_agg): Adjust declaration.
25899 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
25900 * ipa-profile.c (ipa_profile): Likewise.
25901 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
25902 (ipa_populate_param_decls): Make descriptors parameter garbage
25904 (ipa_dump_param): Adjust to descriptors being a pointer.
25905 (ipa_alloc_node_params): Likewise.
25906 (ipa_initialize_node_params): Likewise.
25907 (load_from_param_1): Make descriptors parameter garbage collected.
25908 (load_from_unmodified_param): Likewise.
25909 (load_from_param): Likewise.
25910 (ipa_load_from_parm_agg): Likewise.
25911 (ipa_node_params::~ipa_node_params): Removed.
25912 (ipa_free_all_node_params): Remove call to delete operator.
25913 (ipa_node_params_t::insert): New.
25914 (ipa_node_params_t::remove): Likewise.
25915 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
25916 copy known_csts and known_contexts vectors.
25917 (ipa_read_node_info): Adjust to descriptors being a pointer.
25918 (ipcp_modif_dom_walker): Make m_descriptors field garbage
25920 (ipcp_transform_function): Make descriptors variable garbage
25923 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
25925 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
25926 * config/i386/avx512dqintrin.h: Ditto.
25927 * config/i386/avx512fintrin.h: Ditto.
25928 * config/i386/i386.c: Handle new builtins.
25929 * config/i386/i386-builtin.def: Add new builtins.
25930 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
25931 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
25933 2017-01-23 Jakub Jelinek <jakub@redhat.com>
25934 Martin Liska <mliska@suse.cz>
25936 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
25937 * asan.c (asan_expand_poison_ifn): Support stores and use
25938 appropriate ASAN report function.
25939 * internal-fn.c (expand_ASAN_POISON_USE): New function.
25940 * internal-fn.def (ASAN_POISON_USE): Declare.
25941 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
25942 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
25943 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
25944 ASAN_POISON calls w/o LHS.
25945 * tree-ssa.c (execute_update_addresses_taken): Create clobber
25946 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
25947 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
25948 * gimplify.c (asan_poison_variables): Add attribute
25949 use_after_scope_memory to variables that really needs to live
25951 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
25952 having the attribute.
25954 2017-01-23 Martin Liska <mliska@suse.cz>
25956 * asan.c (create_asan_shadow_var): New function.
25957 (asan_expand_poison_ifn): Likewise.
25958 * asan.h (asan_expand_poison_ifn): New declaration.
25959 * internal-fn.c (expand_ASAN_POISON): Likewise.
25960 * internal-fn.def (ASAN_POISON): New builtin.
25961 * sanopt.c (pass_sanopt::execute): Expand
25962 asan_expand_poison_ifn.
25963 * tree-inline.c (copy_decl_for_dup_finish): Make function
25965 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
25966 * tree-ssa.c (is_asan_mark_p): New function.
25967 (execute_update_addresses_taken): Rewrite local variables
25968 (identified just by use-after-scope as addressable) into SSA.
25970 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
25972 * doc/install.texi (Specific): opensource.apple.com uses https
25973 now. Remove trailing slash.
25975 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
25977 * README.Portability: Remove note on an Irix compatibility issue.
25979 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
25981 * gcov.c (INCLUDE_ALGORITHM): Define.
25982 (INCLUDE_VECTOR): Define.
25983 No longer include <vector> and <algorithm> directly.
25985 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
25987 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
25989 * doc/invoke.texi (Code Gen Options): Ditto.
25991 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
25994 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
25996 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
25998 rtl-optimization/79125
25999 * cprop.c (local_cprop_pass): Handle cases where we make an
26000 unconditional trap.
26002 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
26006 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
26007 read from, for big endian.
26009 2017-01-20 Jiong Wang <jiong.wang@arm.com>
26011 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
26012 register pauth builtins for LP64 only.
26014 2017-01-20 Marek Polacek <polacek@redhat.com>
26017 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
26020 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
26022 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
26024 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
26025 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
26026 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
26028 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26031 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
26032 in big-endian mode when they are not a single duplicated value.
26034 2017-01-20 Richard Biener <rguenther@suse.de>
26036 * BASE-VER: Bump to 7.0.1.
26038 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
26040 * omp-low.c (omplow_simd_context): New struct. Use it...
26041 (lower_rec_simd_input_clauses): ...here and...
26042 (lower_rec_input_clauses): ...here to hold common data. Adjust all
26043 references to idx, lane, max_vf, is_simt.
26045 2017-01-20 Graham Markall <graham.markall@embecosm.com>
26047 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
26050 2017-01-20 Martin Jambor <mjambor@suse.cz>
26052 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
26053 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
26055 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
26056 (GTFILES): Rename hsa.c to hsa-common.c.
26057 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
26058 * hsa-dump.c: Likewise.
26059 * hsa-gen.c: Likewise.
26060 * hsa-regalloc.c: Likewise.
26061 * ipa-hsa.c: Likewise.
26062 * omp-expand.c: Likewise.
26063 * omp-low.c: Likewise.
26064 * toplev.c: Likewise.
26066 2017-01-20 Marek Polacek <polacek@redhat.com>
26069 * doc/invoke.texi: Document -Wduplicated-branches.
26070 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
26071 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
26072 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
26073 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
26074 return 0 only when not OEP_LEXICOGRAPHIC.
26075 (fold_build_cleanup_point_expr): Use the expression
26076 location when building CLEANUP_POINT_EXPR.
26077 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
26078 * tree.c (add_expr): Handle error_mark_node.
26080 2017-01-20 Martin Liska <mliska@suse.cz>
26083 * tree-profile.c (init_ic_make_global_vars): Do not call
26085 (gimple_init_gcov_profiler): Likewise.
26087 2017-01-20 Martin Liska <mliska@suse.cz>
26090 * cgraph.h (maybe_create_reference): Remove argument and
26092 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
26094 * ipa-cp.c (create_specialized_node): Likewise.
26095 * symtab.c (symtab_node::maybe_create_reference): Handle
26096 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
26098 2017-01-20 Martin Liska <mliska@suse.cz>
26100 * read-rtl-function.c (function_reader::create_function): Use
26101 build_decl instread of build_decl_stat.
26103 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
26105 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
26106 * config/i386/avx512dqintrin.h: Ditto.
26107 * config/i386/avx512fintrin.h: Ditto.
26108 * config/i386/i386-builtin-types.def: Add new types.
26109 * config/i386/i386.c: Handle new types.
26110 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
26111 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
26112 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
26113 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
26114 (__builtin_ia32_kshiftridi): New.
26115 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
26117 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
26121 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
26122 define to rs6000_init_stack_protect_guard.
26123 (rs6000_init_stack_protect_guard): New function.
26125 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
26126 Yunqiang Su <yunqiang.su@imgtec.com>
26128 * config.gcc (supported_defaults): Add madd4.
26129 (with_madd4): Add validation.
26130 (all_defaults): Add madd4.
26131 * config/mips/mips.opt (mmadd4): New option.
26132 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
26134 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
26136 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
26137 (ISA_HAS_FUSED_MADD4): Likewise.
26138 * doc/invoke.texi (-mmadd4): Document the new option.
26139 * doc/install.texi (--with-madd4): Document the new option.
26141 2017-01-19 Jiong Wang <jiong.wang@arm.com>
26143 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
26144 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
26145 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
26146 (aarch64_init_pauth_hint_builtins): New.
26147 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
26148 (aarch64_expand_builtin): Expand new builtins.
26150 2017-01-19 Jiong Wang <jiong.wang@arm.com>
26152 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
26153 * combine-stack-adj.c (no_unhandled_cfa): Handle
26154 REG_CFA_TOGGLE_RA_MANGLE.
26155 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
26156 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
26157 info for return address signing.
26158 (aarch64_expand_epilogue): Likewise.
26160 2017-01-19 Jiong Wang <jiong.wang@arm.com>
26162 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
26163 * config/aarch64/aarch64-protos.h
26164 (aarch64_return_address_signing_enabled): New declaration.
26165 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
26167 (aarch64_expand_prologue): Sign return address before it's pushed onto
26169 (aarch64_expand_epilogue): Authenticate return address fetched from
26171 (aarch64_override_options): Sanity check for ILP32 and ISA level.
26172 (aarch64_attributes): New function attributes for "sign-return-address".
26173 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
26174 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
26175 ("*do_return"): Generate combined instructions according to key index.
26176 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
26177 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
26179 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
26180 * config/aarch64/aarch64.opt (msign-return-address=): New.
26181 * doc/extend.texi (AArch64 Function Attributes): Documents
26182 "sign-return-address=".
26183 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
26185 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
26187 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
26188 overall option summary.
26190 2017-01-19 Jiong Wang <jiong.wang@arm.com>
26192 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
26193 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
26194 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
26195 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
26197 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
26199 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
26200 -mpower9-minmax by default for -mcpu=power9.
26201 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
26202 128-bit floating point.
26204 2017-01-20 Alan Modra <amodra@gmail.com>
26206 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
26207 optimizing for size.
26209 2017-01-20 Alan Modra <amodra@gmail.com>
26212 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
26213 for strcmp and strncmp from corresponding builtin decl.
26215 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
26217 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
26218 instead of i386/rtems-64.h.
26219 * config/i386/rtems-64.h: Remove.
26221 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
26225 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
26227 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
26229 2017-01-19 Tamar Christina <tamar.christina@arm.com>
26231 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
26232 Change int to HOST_WIDE_INT.
26233 * config/aarch64/aarch64-protos.h
26234 (aarch64_simd_gen_const_vector_dup): Likewise.
26235 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
26237 2017-01-19 David Malcolm <dmalcolm@redhat.com>
26239 * langhooks-def.h (lhd_type_for_size): New decl.
26240 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
26241 * langhooks.c (lhd_type_for_size): New function, taken from
26244 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
26246 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
26247 define_bypass for CR latency.
26248 (power9-cracked-alu): Update bypass latency and remove power9-branch.
26249 (power9-alu2): Add define_bypass for CR latency.
26251 (power9-mul): Update insn latency.
26252 (power9-mul-compare): Update insn latency, bypass latency and remove
26255 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26257 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
26259 * config/aarch64/aarch64.md
26260 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
26261 aarch64_nopcrelative_literal_loads.
26262 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
26264 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
26266 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
26267 TARGET_LOONGSON_3A.
26268 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
26270 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
26273 * config.gcc (supported_defaults): Add lxc1-sxc1.
26274 (with_lxc1_sxc1): Add validation.
26275 (all_defaults): Add lxc1-sxc1.
26276 * config/mips/mips.opt (mlxc1-sxc1): New option.
26277 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
26279 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
26280 __mips_no_lxc1_sxc1.
26281 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
26282 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
26283 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
26285 2017-01-19 Richard Biener <rguenther@suse.de>
26287 PR tree-optimization/72488
26288 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
26289 sure to restore SSA info.
26290 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
26292 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
26294 PR rtl-optimization/79121
26295 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
26296 of the inner type when shifting an extended value.
26298 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
26301 * symtab.c (symtab_node::equal_address_to): Fix comparing of
26302 interposable aliases.
26304 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
26307 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
26308 Use the evmergelohi instruction.
26309 (mov_si<mode>_e500_subreg4_2_le): Likewise.
26310 (mov_sitf_e500_subreg8_2_be): Likewise.
26311 (mov_sitf_e500_subreg12_2_le): Likewise.
26312 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
26313 (mov_si<mode>_e500_subreg4_2_be): Likewise.
26314 (mov_sitf_e500_subreg8_2_le): Likewise.
26315 (mov_sitf_e500_subreg12_2_be): Likewise.
26317 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26319 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
26320 attribute from vecsimple to vecperm.
26321 (altivec_vbpermq2): Likewise.
26323 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26326 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
26328 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26329 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
26330 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
26331 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
26332 case where N arg is SIZE_MAX.
26333 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
26334 (cmpstrsi): Add pattern.
26336 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
26338 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26339 __builtin_vec_revb builtins.
26340 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
26341 built-in functions to support generation of the ISA 3.0 XXBR<x>
26342 vector byte reverse instructions.
26343 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
26344 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
26345 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
26346 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
26347 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
26348 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
26349 (P9V_BUILTIN_VEC_REVB): Likewise.
26350 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
26351 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
26352 (p9_xxbrq_v16qi): Likewise.
26353 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
26354 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
26355 (p9_xxbrh_v8hi): Likewise.
26356 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
26357 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
26358 vec_revb built-in functions.
26360 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
26362 PR rtl-optimization/78952
26363 * config/i386/i386.md (any_extract): New code iterator.
26364 (*insvqi_2): Use any_extract for source operand.
26365 (*insvqi_3): Use any_shiftrt for source operand.
26367 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
26369 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
26371 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
26373 2017-01-18 Matthias Klose <doko@ubuntu.com>
26375 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
26377 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26379 * config/rs6000/altivec.h (vec_bperm): Change #define.
26380 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
26381 (altivec_vbpermq2): New define_insn.
26382 (altivec_vbpermd): Likewise.
26383 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
26384 function interface.
26385 (VBPERMD): Likewise.
26386 (VBPERM): New polymorphic function interface.
26387 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
26388 Add entries for P9V_BUILTIN_VEC_VBPERM.
26389 * doc/extend.texi: Add interfaces for vec_bperm.
26391 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26393 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
26394 first letter of error messages.
26395 (s390_resolve_overloaded_builtin): Likewise.
26396 * config/s390/s390.c (s390_expand_builtin): Likewise.
26397 (s390_invalid_arg_for_unprototyped_fn): Likewise.
26398 (s390_valid_target_attribute_inner_p): Likewise.
26399 * config/s390/s390.md ("tabort"): Likewise.
26401 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
26403 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
26404 (ISA_AVOID_DIV_HILO): New macro.
26405 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
26406 (ISA_HAS_DDIV): Likewise.
26408 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
26410 * doc/invoke.texi (fabi-version): Correct number of occurrences.
26412 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
26414 * doc/invoke.texi (fabi-version): Spelling fix.
26416 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
26419 * doc/invoke.texi (fabi-version): Mention mangling fix for
26422 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
26425 * doc/invoke.texi (fabi-version): Document discriminator mangling.
26427 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
26430 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
26431 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
26433 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
26435 (stack_protect_test): Ditto.
26436 * config/rs6000/rs6000.opt (mstack-protector-guard=,
26437 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
26439 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
26440 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
26441 -mstack-protector-guard-offset=.
26442 (RS/6000 and PowerPC Options): Ditto.
26444 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
26446 * config/i386/i386.h (MASK_CLASS_P): New define.
26447 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
26448 there are no registers from different register sets also when
26449 mask registers are used. Update function comment.
26450 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
26451 to (*k/*r) and (*k/*km) alternatives.
26453 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
26455 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
26456 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
26457 (EH_RETURN_HANDLER_RTX): New define.
26458 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
26459 Force frame pointer in EH return functions.
26460 (aarch64_expand_epilogue): Add barrier for eh_return.
26461 (aarch64_final_eh_return_addr): Remove.
26462 (aarch64_eh_return_handler_rtx): New function.
26463 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
26465 (aarch64_eh_return_handler_rtx): New prototype.
26467 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26469 * config/rs6000/altivec.h (vec_rlmi): New #define.
26470 (vec_vrlnm): Likewise.
26471 (vec_rlnm): Likewise.
26472 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
26473 (UNSPEC_VRLNM): Likewise.
26474 (VIlong): New mode iterator.
26475 (altivec_vrl<VI_char>mi): New define_insn.
26476 (altivec_vrl<VI_char>nm): Likewise.
26477 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
26479 (VRLDNM): Likewise.
26480 (RLNM): New polymorphic function entry.
26481 (VRLWMI): New monomorphic function entry.
26482 (VRLDMI): Likewise.
26483 (RLMI): New polymorphic function entry.
26484 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
26485 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
26486 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
26489 2017-01-17 Jakub Jelinek <jakub@redhat.com>
26492 * dwarf2out.c (field_byte_offset): Restore the
26493 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
26494 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
26495 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
26498 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
26501 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
26503 2017-01-17 Jakub Jelinek <jakub@redhat.com>
26506 * dwarf2out.c (add_data_member_location_attribute): For constant
26507 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
26508 instead of DW_AT_data_member_location, DW_AT_bit_offset and
26509 DW_AT_byte_size attributes.
26511 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
26513 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
26514 after forcing to constant memory when the code model is medium.
26516 2017-01-17 Julia Koval <julia.koval@intel.com>
26519 * config/i386/avx512fintrin.h
26520 (_mm512_i32gather_ps): Change __addr type to void const*.
26521 (_mm512_mask_i32gather_ps): Ditto.
26522 (_mm512_i32gather_pd): Ditto.
26523 (_mm512_mask_i32gather_pd): Ditto.
26524 (_mm512_i64gather_ps): Ditto.
26525 (_mm512_mask_i64gather_ps): Ditto.
26526 (_mm512_i64gather_pd): Ditto.
26527 (_mm512_mask_i64gather_pd): Ditto.
26528 (_mm512_i32gather_epi32): Ditto.
26529 (_mm512_mask_i32gather_epi32): Ditto.
26530 (_mm512_i32gather_epi64): Ditto.
26531 (_mm512_mask_i32gather_epi64): Ditto.
26532 (_mm512_i64gather_epi32): Ditto.
26533 (_mm512_mask_i64gather_epi32): Ditto.
26534 (_mm512_i64gather_epi64): Ditto.
26535 (_mm512_mask_i64gather_epi64): Ditto.
26536 (_mm512_i32scatter_ps): Change __addr type to void*.
26537 (_mm512_mask_i32scatter_ps): Ditto.
26538 (_mm512_i32scatter_pd): Ditto.
26539 (_mm512_mask_i32scatter_pd): Ditto.
26540 (_mm512_i64scatter_ps): Ditto.
26541 (_mm512_mask_i64scatter_ps): Ditto.
26542 (_mm512_i64scatter_pd): Ditto.
26543 (_mm512_mask_i64scatter_pd): Ditto.
26544 (_mm512_i32scatter_epi32): Ditto.
26545 (_mm512_mask_i32scatter_epi32): Ditto.
26546 (_mm512_i32scatter_epi64): Ditto.
26547 (_mm512_mask_i32scatter_epi64): Ditto.
26548 (_mm512_i64scatter_epi32): Ditto.
26549 (_mm512_mask_i64scatter_epi32): Ditto.
26550 (_mm512_i64scatter_epi64): Ditto.
26551 (_mm512_mask_i64scatter_epi64): Ditto.
26552 * config/i386/avx512pfintrin.h
26553 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
26554 (_mm512_mask_prefetch_i32gather_ps): Ditto.
26555 (_mm512_mask_prefetch_i64gather_pd): Ditto.
26556 (_mm512_mask_prefetch_i64gather_ps): Ditto.
26557 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
26558 (_mm512_prefetch_i32scatter_ps): Ditto.
26559 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
26560 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
26561 (_mm512_prefetch_i64scatter_pd): Ditto.
26562 (_mm512_prefetch_i64scatter_ps): Ditto.
26563 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
26564 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
26565 * config/i386/avx512vlintrin.h
26566 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
26567 (_mm_mmask_i32gather_ps): Ditto.
26568 (_mm256_mmask_i32gather_pd): Ditto.
26569 (_mm_mmask_i32gather_pd): Ditto.
26570 (_mm256_mmask_i64gather_ps): Ditto.
26571 (_mm_mmask_i64gather_ps): Ditto.
26572 (_mm256_mmask_i64gather_pd): Ditto.
26573 (_mm_mmask_i64gather_pd): Ditto.
26574 (_mm256_mmask_i32gather_epi32): Ditto.
26575 (_mm_mmask_i32gather_epi32): Ditto.
26576 (_mm256_mmask_i32gather_epi64): Ditto.
26577 (_mm_mmask_i32gather_epi64): Ditto.
26578 (_mm256_mmask_i64gather_epi32): Ditto.
26579 (_mm_mmask_i64gather_epi32): Ditto.
26580 (_mm256_mmask_i64gather_epi64): Ditto.
26581 (_mm_mmask_i64gather_epi64): Ditto.
26582 (_mm256_i32scatter_ps): Change __addr type to void*.
26583 (_mm256_mask_i32scatter_ps): Ditto.
26584 (_mm_i32scatter_ps): Ditto.
26585 (_mm_mask_i32scatter_ps): Ditto.
26586 (_mm256_i32scatter_pd): Ditto.
26587 (_mm256_mask_i32scatter_pd): Ditto.
26588 (_mm_i32scatter_pd): Ditto.
26589 (_mm_mask_i32scatter_pd): Ditto.
26590 (_mm256_i64scatter_ps): Ditto.
26591 (_mm256_mask_i64scatter_ps): Ditto.
26592 (_mm_i64scatter_ps): Ditto.
26593 (_mm_mask_i64scatter_ps): Ditto.
26594 (_mm256_i64scatter_pd): Ditto.
26595 (_mm256_mask_i64scatter_pd): Ditto.
26596 (_mm_i64scatter_pd): Ditto.
26597 (_mm_mask_i64scatter_pd): Ditto.
26598 (_mm256_i32scatter_epi32): Ditto.
26599 (_mm256_mask_i32scatter_epi32): Ditto.
26600 (_mm_i32scatter_epi32): Ditto.
26601 (_mm_mask_i32scatter_epi32): Ditto.
26602 (_mm256_i32scatter_epi64): Ditto.
26603 (_mm256_mask_i32scatter_epi64): Ditto.
26604 (_mm_i32scatter_epi64): Ditto.
26605 (_mm_mask_i32scatter_epi64): Ditto.
26606 (_mm256_i64scatter_epi32): Ditto.
26607 (_mm256_mask_i64scatter_epi32): Ditto.
26608 (_mm_i64scatter_epi32): Ditto.
26609 (_mm_mask_i64scatter_epi32): Ditto.
26610 (_mm256_i64scatter_epi64): Ditto.
26611 (_mm256_mask_i64scatter_epi64): Ditto.
26612 (_mm_i64scatter_epi64): Ditto.
26613 (_mm_mask_i64scatter_epi64): Ditto.
26614 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
26615 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
26616 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
26617 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
26618 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
26619 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
26620 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
26621 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
26622 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
26623 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
26624 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
26625 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
26626 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
26627 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
26628 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
26629 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
26630 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
26631 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
26632 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
26633 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
26634 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
26635 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
26636 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
26637 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
26638 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
26639 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
26640 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
26641 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
26642 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
26643 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
26644 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
26645 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
26646 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
26647 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
26648 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
26649 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
26650 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
26651 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
26652 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
26653 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
26654 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
26655 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
26656 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
26657 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
26658 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
26659 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
26660 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
26661 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
26662 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
26663 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
26664 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
26665 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
26666 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
26667 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
26668 definitions accordingly.
26670 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
26671 Kuan-Lin Chen <kuanlinchentw@gmail.com>
26674 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
26677 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
26680 * ira-conflicts.c (ira_build_conflicts): Update total conflict
26681 hard regs for inner regno.
26683 2017-01-17 Martin Liska <mliska@suse.cz>
26686 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
26687 assumption and add comment.
26689 2017-01-17 Nathan Sidwell <nathan@acm.org>
26691 * ipa-visibility.c (localize_node): New function, broken out of ...
26692 (function_and_variable_visibility): ... here. Call it.
26694 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
26696 PR middle-end/77445
26697 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
26698 correctly set frequency of oudgoing edge.
26699 (duplicate_thread_path): Fix profile updating.
26701 2017-01-17 Jakub Jelinek <jakub@redhat.com>
26704 * configure.ac: Add GCC_BASE_VER.
26705 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
26706 version from BASE-VER file.
26707 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
26708 (gcc.o): Depend on $(BASEVER).
26709 * common.opt (dumpfullversion): New option.
26710 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
26711 * doc/invoke.texi: Document -dumpfullversion.
26712 * doc/install.texi: Document --with-gcc-major-version-only.
26713 * configure: Regenerated.
26715 2017-01-17 Richard Biener <rguenther@suse.de>
26717 PR tree-optimization/71433
26718 * tree-vrp.c (register_new_assert_for): Merge same asserts
26719 on all incoming edges.
26720 (process_assert_insertions_for): Handle insertions at the
26723 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
26725 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
26726 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
26728 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
26731 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
26734 2017-01-17 Alan Modra <amodra@gmail.com>
26737 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
26738 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
26739 symbolic stack limit when pic.
26741 2017-01-16 Martin Sebor <msebor@redhat.com>
26743 PR tree-optimization/78608
26744 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
26746 2017-01-16 Jeff Law <law@redhat.com>
26749 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
26750 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
26751 for several include directories that may be relative to sysroot.
26752 * config/i386/x-mingw32 (gplus_includedir): Define.
26753 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
26754 (native_system_includedir): Likewise.
26755 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
26756 override if TARGET_SYSTEM_ROOT is defined.
26757 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
26759 PR tree-optimization/79090
26760 PR tree-optimization/33562
26761 PR tree-optimization/61912
26762 PR tree-optimization/77485
26763 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
26764 and computed trims into the dump file.
26766 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
26768 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
26770 2017-01-16 Jakub Jelinek <jakub@redhat.com>
26773 * gimplify.c (gimplify_init_constructor): If want_value and
26774 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
26778 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
26779 sequence. Formatting fixes.
26780 (doloop_optimize): Formatting fixes.
26783 * gcc.c (debug_level_greater_than_spec_func): New function.
26784 (static_spec_functions): Add debug-level-gt spec function.
26785 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
26787 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
26788 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
26789 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
26790 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
26791 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
26792 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
26794 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
26796 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
26797 QImode fixups to general and mask registers only.
26799 2017-01-16 Carl Love <cel@us.ibm.com>
26801 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
26802 for built-in functions
26803 vector signed char vec_nabs (vector signed char)
26804 vector signed short vec_nabs (vector signed short)
26805 vector signed int vec_nabs (vector signed int)
26806 vector signed long long vec_nabs (vector signed long long)
26807 vector float vec_nabs (vector float)
26808 vector double vec_nabs (vector double)
26809 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
26811 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
26812 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
26813 * doc/extend.texi: Update the documentation file for the new built-in
26816 2017-01-16 Martin Sebor <msebor@redhat.com>
26818 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
26821 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26823 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
26824 UNSPEC_VSX__XXSPLTD to require special splat handling.
26826 2017-01-16 David Malcolm <dmalcolm@redhat.com>
26829 * system.h: Poison strndup.
26831 2017-01-16 Alan Modra <amodra@gmail.com>
26834 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
26837 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
26839 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
26841 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
26843 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
26844 call recog here. Assert that INSN_CODE (insn) is non-negative.
26846 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
26849 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
26851 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
26852 in the currently scheduled RTL fragment.
26854 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
26856 PR rtl-optimization/78751
26857 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
26860 2017-01-14 Jeff Law <law@redhat.com>
26862 PR tree-optimization/79090
26863 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
26864 variable length stores.
26865 (compute_trims): Delete dead assignment to *trim_tail.
26866 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
26869 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
26871 PR rtl-optimization/78626
26872 PR rtl-optimization/78727
26873 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
26874 of a block, and split such blocks after everything else is finished.
26876 2017-01-14 Alan Modra <amodra@gmail.com>
26879 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
26880 target legitimate_combined_insn.
26881 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
26882 (rs6000_legitimate_combined_insn): New function.
26883 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
26885 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
26886 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
26887 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
26889 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
26891 * doc/frontends.texi (G++ and GCC): Remove references to Java.
26893 2017-01-13 Jeff Law <law@redhat.com>
26895 PR tree-optimization/33562
26896 PR tree-optimization/61912
26897 PR tree-optimization/77485
26898 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
26900 (delete_dead_assignment): Likewise.
26901 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
26902 statement to delete_dead_call and delete_dead_assignment.
26904 2017-01-13 David Malcolm <dmalcolm@redhat.com>
26907 * substring-locations.c (format_warning_va): Strengthen case 1 so
26908 that both endpoints of the substring must be within the format
26909 range for just the substring to be printed.
26911 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
26913 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
26914 * config/i386/i386.c (ix86_target_string): Add missing options
26915 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
26916 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
26917 flags_other and ix86_target_other to flags2_other. Display unknown
26919 (ix86_valid_target_attribute_inner_p): Add missing options and
26920 reorder options by implied ISAs, as in ix86_target_string.
26922 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
26924 * hash-table.h (hash_table::too_empty_p): New function.
26925 (hash_table::expand): Use it.
26926 (hash_table::traverse): Likewise.
26927 (hash_table::empty_slot): Use sizeof (value_type) instead of
26928 sizeof (PTR) to convert bytes to elements. Shrink the table
26929 if the current size is excessive for the current number of
26932 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
26934 * ira-costs.c (record_reg_classes): Break from the inner loop
26935 early once alt_fail is known to be true. Update outer loop
26936 handling accordingly.
26938 2017-01-13 Jeff Law <law@redhat.com>
26940 * tree-ssa-dse.c (decrement_count): New function.
26941 (increment_start_addr, maybe_trim_memstar_call): Likewise.
26942 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
26943 when we know the partially dead statement is a mem* function.
26945 PR tree-optimization/61912
26946 PR tree-optimization/77485
26947 * tree-ssa-dse.c: Include expr.h.
26948 (maybe_trim_constructor_store): New function.
26949 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
26951 PR tree-optimization/33562
26952 PR tree-optimization/61912
26953 PR tree-optimization/77485
26954 * doc/invoke.texi: Document new dse-max-object-size param.
26955 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
26956 * tree-ssa-dse.c: Include params.h.
26957 (dse_store_status): New enum.
26958 (initialize_ao_ref_for_dse): New, partially extracted from
26960 (valid_ao_ref_for_dse, normalize_ref): New.
26961 (setup_live_bytes_from_ref, compute_trims): Likewise.
26962 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
26963 (maybe_trim_partially_dead_store): Likewise.
26964 (maybe_trim_complex_store): Likewise.
26965 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
26966 Track what bytes live from the original store. Return tri-state
26967 for dead, partially dead or live.
26968 (dse_dom_walker): Add constructor, destructor and new private members.
26969 (delete_dead_call, delete_dead_assignment): New extracted from
26971 (dse_optimize_stmt): Make a member of dse_dom_walker.
26972 Use initialize_ao_ref_for_dse.
26974 PR tree-optimization/33562
26975 PR tree-optimization/61912
26976 PR tree-optimization/77485
26977 * sbitmap.h (bitmap_count_bits): Prototype.
26978 (bitmap_clear_range, bitmap_set_range): Likewise.
26979 * sbitmap.c (bitmap_clear_range): New function.
26980 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
26982 2017-01-13 Martin Liska <mliska@suse.cz>
26985 * function.c (set_cfun): Add new argument force.
26986 * function.h (set_cfun): Likewise.
26987 * ipa-inline-transform.c (inline_call): Use the function when
26988 strict alising from is dropped for function we inline to.
26990 2017-01-13 Richard Biener <rguenther@suse.de>
26992 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
26993 for dumping GIMPLE INTEGER_CSTs.
26995 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26997 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
26998 to 201112L since C++17.
27000 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
27003 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
27004 if -fsanitize=kernel-address is present.
27006 2017-01-13 Richard Biener <rguenther@suse.de>
27008 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
27009 as _Literal ( type ) number in case usual suffixes do not
27010 preserve all information.
27012 2017-01-13 Richard Biener <rguenther@suse.de>
27014 PR tree-optimization/77283
27015 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
27016 and ssa-iterators.h.
27017 (is_feasible_trace): Implement a cost model based on joiner
27020 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
27023 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
27024 char or short to __float128/_Float128 directly.
27026 2017-01-12 Martin Sebor <msebor@redhat.com>
27028 to -Wformat-overflow.
27029 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
27030 (min_bytes_remaining): Same.
27031 (get_string_length): Same.
27032 (format_string): Same.
27033 (format_directive): Same.
27035 (pass_sprintf_length::handle_gimple_call): Same.
27037 2017-01-12 Jakub Jelinek <jakub@redhat.com>
27039 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
27040 info.nowrite calls with no lhs that can't throw. Return bool
27041 whether gsi_remove has been called or not.
27042 (pass_sprintf_length::handle_gimple_call): Return bool whether
27043 try_substitute_return_value called gsi_remove. Formatting fix.
27044 (pass_sprintf_length::execute): Don't use gsi_remove if
27045 handle_gimple_call returned true.
27048 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
27049 be removed due to side-effects, don't remove following barrier nor
27050 turn the successor edge into fallthru edge.
27052 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27055 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
27056 element-reversing loads and stores as not swappable.
27058 2017-01-12 Nathan Sidwell <nathan@acm.org>
27059 Nicolai Stange <nicstange@gmail.com>
27061 * combine.c (try_combine): Don't ignore result of overlap checking
27062 loop. Combine overlap & asm check into single loop.
27064 2017-01-12 Richard Biener <rguenther@suse.de>
27066 * tree-pretty-print.c (dump_generic_node): Provide -gimple
27067 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
27069 2017-01-12 Richard Biener <rguenther@suse.de>
27071 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
27072 and TS_TARGET_OPTION directly derive from TS_BASE.
27073 * tree-core.h (tree_optimization_option): Derive from tree_base.
27074 (tree_target_option): Likewise.
27076 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
27078 * config/i386/i386.c (memory_address_length): Increase len
27079 only when rip_relative_addr_p returns false.
27081 2017-01-11 Julia Koval <julia.koval@intel.com>
27083 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
27084 (OPTION_MASK_ISA_SGX_SET): New.
27085 (ix86_handle_option): Handle OPT_msgx.
27086 * config.gcc: Added sgxintrin.h.
27087 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
27088 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
27089 * config/i386/i386.c (ix86_target_string): Add -msgx.
27091 (ix86_option_override_internal): Handle new options.
27092 (ix86_valid_target_attribute_inner_p): Add sgx.
27093 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
27094 * config/i386/i386.opt: Add msgx.
27095 * config/i386/sgxintrin.h: New file.
27096 * config/i386/x86intrin.h: Add sgxintrin.h.
27098 2017-01-11 Jakub Jelinek <jakub@redhat.com>
27101 * fold-const.c (maybe_nonzero_address): Return 1 for function
27103 (tree_single_nonzero_warnv_p): Don't handle function local objects
27107 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
27110 2017-01-11 David Malcolm <dmalcolm@redhat.com>
27113 * opts.c: Include "spellcheck.h"
27114 (struct string_fragment): New struct.
27115 (struct edit_distance_traits<const string_fragment &>): New
27117 (get_closest_sanitizer_option): New function.
27118 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
27120 2017-01-11 Jakub Jelinek <jakub@redhat.com>
27122 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
27124 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
27125 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
27126 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
27127 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
27128 for initial die_offset if dwarf_split_debug_info.
27129 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
27130 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
27132 (output_skeleton_debug_sections): Formatting fix. Use
27133 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
27134 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
27136 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
27138 * config/arm/cortex-a53.md: Add bypasses for
27139 cortex_a53_r2f_cvt.
27140 (cortex_a53_r2f): Only use for transfers.
27141 (cortex_a53_f2r): Likewise.
27142 (cortex_a53_r2f_cvt): Add reservation for conversions.
27143 (cortex_a53_f2r_cvt): Likewise.
27145 2017-01-11 Tamar Christina <tamar.christina@arm.com>
27147 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
27148 to all inlined functions, change static to extern.
27150 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
27153 * config/arm/arm.c (legitimize_pic_address): Handle reference to
27155 (arm_assemble_integer): Likewise.
27157 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
27159 * config.gcc: Use new awk script to check CPU, FPU and architecture
27160 parameters for --with-... options.
27161 * config/arm/parsecpu.awk: New file
27162 * config/arm/arm-cpus.in: New file.
27163 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
27165 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
27167 * config/arm/t-arm: Update dependency rules.
27168 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
27169 of processing .def files.
27170 * config/arm/genopt.sh: Deleted.
27171 * config/arm/gentune.sh: Deleted.
27172 * config/arm/arm-cores.def: Deleted.
27173 * config/arm/arm-arches.def: Deleted.
27174 * config/arm/arm-fpus.def: Deleted.
27175 * config/arm/arm-tune.md: Regenerated.
27176 * config/arm/arm-tables.opt: Regenerated.
27177 * config/arm/arm-cpu.h: New generated file.
27178 * config/arm/arm-cpu-data.h: New generated file.
27179 * config/arm/arm-cpu-cdata.h: New generated file.
27181 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
27184 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
27186 (input_varpool_node): Unpack dynamically_initialized bit.
27188 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
27190 PR rtl-optimization/79032
27191 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
27192 the alignment of the adjusted memory reference against that of MODE,
27193 instead of the alignment of the original memory reference.
27195 2017-01-11 Martin Jambor <mjambor@suse.cz>
27197 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
27199 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
27200 decorated functions.
27202 2017-01-11 Richard Biener <rguenther@suse.de>
27204 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
27205 set range/nonnull info for PHI results. Do not set it on
27206 stmts marked for removal.
27208 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
27210 * expr.c (store_field): In the bitfield case, fetch the return value
27211 from the registers before applying a single big-endian adjustment.
27212 Always do a final load for a BLKmode value not larger than a word.
27214 2017-01-10 David Malcolm <dmalcolm@redhat.com>
27217 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
27218 that we correctly handle column numbers greater than
27219 LINE_MAP_MAX_COLUMN_NUMBER.
27221 2017-01-10 Martin Sebor <msebor@redhat.com>
27223 PR middle-end/78245
27224 * gimple-ssa-sprintf.c (get_destination_size): Call
27225 {init,fini}object_sizes.
27226 * tree-object-size.c (addr_object_size): Adjust.
27227 (pass_through_call): Adjust.
27228 (pass_object_sizes::execute): Adjust.
27229 * tree-object-size.h (fini_object_sizes): Declare.
27231 2017-01-10 Martin Sebor <msebor@redhat.com>
27233 PR tree-optimization/78775
27234 * builtins.c (get_size_range): Move...
27235 * calls.c: ...to here.
27236 (alloc_max_size): Accept zero argument.
27237 (operand_signed_p): Remove.
27238 (maybe_warn_alloc_args_overflow): Call get_size_range.
27239 * calls.h (get_size_range): Declare.
27241 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
27243 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
27244 from TI's devices.csv file as of September 2016.
27245 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
27247 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
27249 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
27250 * doc/invoke.texi: Likewise.
27251 * doc/md.texi: Likewise.
27252 * doc/objc.texi: Likewise.
27254 2017-01-10 Joshua Conner <joshconner@google.com>
27256 * config/arm/fuchsia-elf.h: New file.
27257 * config/fuchsia.h: New file.
27258 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
27259 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
27261 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
27263 2016-01-10 Richard Biener <rguenther@suse.de>
27265 PR tree-optimization/79034
27266 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
27267 Propagate out degenerate PHIs in the joiner.
27269 2017-01-10 Martin Liska <mliska@suse.cz>
27271 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
27272 (sort_congruence_classes_by_decl_uid): Likewise.
27273 (sort_congruence_class_groups_by_decl_uid): Likewise.
27274 (sem_item_optimizer::merge_classes): Sort class, groups in these
27275 classes and members in the groups by DECL_UID of declarations.
27276 This would make merge operations stable.
27278 2017-01-10 Martin Liska <mliska@suse.cz>
27280 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
27281 usage of m_classes_vec.
27282 (sem_item_optimizer::~sem_item_optimizer): Likewise.
27283 (sem_item_optimizer::get_group_by_hash): Likewise.
27284 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
27285 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
27286 (sem_item_optimizer::verify_classes): Likewise.
27287 (sem_item_optimizer::process_cong_reduction): Likewise.
27288 (sem_item_optimizer::dump_cong_classes): Likewise.
27289 (sem_item_optimizer::merge_classes): Likewise.
27290 * ipa-icf.h (congruence_class_hash): Rename from
27291 congruence_class_group_hash. Remove declaration of m_classes_vec.
27293 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
27295 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
27296 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
27297 * config.gcc: Add avx512vpopcntdqintrin.h.
27298 * config/i386/avx512vpopcntdqintrin.h: New.
27299 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
27300 * config/i386/i386-builtin-types.def: Add new types.
27301 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
27302 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
27303 __builtin_ia32_vpopcountq_v8di_mask): New.
27304 * config/i386/i386-c.c (ix86_target_macros_internal): Define
27305 __AVX512VPOPCNTDQ__.
27306 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
27307 (PTA_AVX512VPOPCNTDQ): Define.
27308 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
27309 TARGET_AVX512VPOPCNTDQ_P): Define.
27310 * config/i386/i386.opt: Add mavx512vpopcntdq.
27311 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
27312 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
27314 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
27316 PR middle-end/77484
27317 * predict.def (PRED_CALL): Set to 67.
27319 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
27321 * expr.c (store_field): In the bitfield case, if the value comes from
27322 a function call and is of an aggregate type returned in registers, do
27323 not modify the field mode; extract the value in all cases if the mode
27324 is BLKmode and the size is not larger than a word.
27326 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
27329 * config/i386/cpuid.h: Fix undefined behavior.
27331 2017-01-04 Jeff Law <law@redhat.com>
27333 PR tree-optimization/79007
27334 PR tree-optimization/67955
27335 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
27336 conservative for pt.null when flag_non_call_exceptions is on.
27338 2017-01-09 Jakub Jelinek <jakub@redhat.com>
27340 PR translation/79019
27341 PR translation/79020
27342 * params.def (PARAM_INLINE_MIN_SPEEDUP,
27343 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
27344 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
27346 * config/avr/avr.opt (maccumulate-args): Likewise.
27347 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
27348 * common.opt (freport-bug): Likewise.
27349 * cif-code.def (CIF_FINAL_ERROR): Likewise.
27350 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
27351 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
27352 translatable string.
27353 * config/i386/i386.c (function_value_32): Likewise.
27354 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
27355 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
27357 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
27358 * common/config/msp430/msp430-common.c (msp430_handle_option):
27360 * symtab.c (symtab_node::verify_base): Likewise.
27361 * opts.c (set_debug_level): Likewise.
27362 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
27363 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
27364 missing whitespace to translatable strings.
27365 * config/avr/avr.md (bswapsi2): Fix typo in comment.
27366 * config/sh/superh.h: Likewise.
27367 * config/i386/xopintrin.h: Likewise.
27368 * config/i386/znver1.md: Likewise.
27369 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
27370 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
27371 * double-int.h (struct double_int): Likewise.
27372 * double-int.c (div_and_round_double): Likewise.
27373 * wide-int.cc: Likewise.
27374 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
27375 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
27376 * cfgcleanup.c (crossjumps_occured): Renamed to ...
27377 (crossjumps_occurred): ... this.
27378 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
27381 PR tree-optimization/78899
27382 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
27383 returning bool return struct loop *, NULL for failure and the new
27385 (versionable_outer_loop_p): Don't version outer loop if it has
27386 dont_vectorized bit set.
27387 (tree_if_conversion): When versioning outer loop, ensure
27388 tree_if_conversion is performed also on the inner loop of the
27389 non-vectorizable outer loop copy.
27390 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
27391 LOOP_VECTORIZED in inner loop of the scalar outer loop and
27392 prevent vectorization of it.
27393 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
27394 the outer loop vectorization of the non-scalar version is attempted
27395 before vectorization of the inner loop in scalar version. If
27396 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
27397 vectorization of its inner loop.
27398 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
27399 has 2 inner loops, rename also on edges from bb whose single pred
27400 is outer_loop->header. Fix typo in function comment.
27402 2017-01-09 Martin Sebor <msebor@redhat.com>
27405 * asan.c (asan_emit_stack_protection): Increase local buffer size
27406 to avoid snprintf truncation warning.
27408 2017-01-09 Andrew Pinski <apinski@cavium.com>
27410 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
27411 to reference thunderx2t99 for the tuning structure
27412 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
27414 (thunderx2t99_extra_costs): This.
27415 * config/aarch64/aarch64-tune.md: Regenerate.
27416 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
27417 (vulcan_addrcost_table): This.
27418 (vulcan_regmove_cost): Rename to ...
27419 (thunderx2t99_regmove_cost): This.
27420 (vulcan_vector_cost): Rename to ...
27421 (thunderx2t99_vector_cost): this.
27422 (vulcan_branch_cost): Rename to ...
27423 (thunderx2t99_branch_cost): This.
27424 (vulcan_tunings): Rename to ...
27425 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
27426 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
27428 2017-01-09 Martin Jambor <mjambor@suse.cz>
27432 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
27433 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
27434 (propagate_vr_accross_jump_function): Use the above function for all
27435 value range computations for pass-through jump functions and type
27436 converasion from explicit value range values.
27437 (ipcp_propagate_stage): Do not attempt to deduce types of formal
27438 parameters from TYPE_ARG_TYPES.
27439 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
27440 (ipa_write_node_info): Stream type of the actual argument.
27441 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
27443 2017-01-09 Martin Liska <mliska@suse.cz>
27446 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
27447 (lookup_compiler): Do not show error message with have_E.
27449 2017-01-09 Jakub Jelinek <jakub@redhat.com>
27451 PR tree-optimization/78938
27452 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
27453 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
27454 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
27455 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
27458 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27460 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
27463 2017-01-09 Richard Biener <rguenther@suse.de>
27465 PR tree-optimization/78997
27466 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
27467 name condition properly.
27469 2017-01-09 Richard Biener <rguenther@suse.de>
27472 * dwarf2out.c (is_cxx): New overload with context.
27473 (is_naming_typedef_decl): Use it.
27475 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
27477 * invoke.texi (Option Summary): Correct spacing in option lists
27478 and add line breaks to fix over-long lines.
27480 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
27482 PR middle-end/17660
27484 * extend.texi (Common Variable Attributes): Add xref to GCC
27485 Internals manual to explain mode attribute keywords.
27487 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
27490 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
27491 and Preprocessor Options.
27492 (Options for Linking): Document -pthread here....
27493 (RS/6000 and PowerPC Options): ...not here.
27494 (Solaris 2 Options): ...or here.
27495 * doc/cppopts.texi: Document -pthread.
27497 2017-01-08 Martin Sebor <msebor@redhat.com>
27499 PR middle-end/77708
27500 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
27501 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
27502 New member functions.
27503 (format_directive): Used them.
27505 (pass_sprintf_length::handle_gimple_call): Same.
27506 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
27507 to avoid truncation for any argument.
27508 (extract_affine_mul): Same.
27509 * tree.c (get_file_function_name): Same.
27511 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
27513 PR middle-end/77484
27514 * predict.def (PRED_INDIR_CALL): Set to 86.
27516 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
27518 PR preprocessor/54124
27519 * doc/cppopts.texi: Reformat -d subtable to list the full name
27520 of the options. Add cross-reference to the docs for the general
27521 compiler -d options.
27522 * doc/invoke.texi (Developer Options): Add cross-reference to the
27523 preprocessor-specific -d option documentation.
27525 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
27527 PR preprocessor/13498
27528 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
27529 redudant material, and reflect new command-line options.
27530 (System Headers): Likewise.
27532 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
27534 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
27535 -isystem, and -idirafter. Copy-edit.
27536 * doc/cppopts.texi: Copy-edit. Remove contradiction about
27537 default for -ftrack-macro-expansion. Delete obsolete and
27538 badly-formatted implementation details about -fdebug-cpp output.
27539 * doc/cppwarnopts.texi: Copy-edit.
27541 2017-01-07 David Malcolm <dmalcolm@redhat.com>
27544 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
27545 that the transition from a max line width >= 1<<10 to narrower
27546 lines works correctly.
27548 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
27550 * doc/options.texi (PerFunction): New.
27551 * opt-functions.awk (switch_flags): Map both Optimization and
27552 PerFunction to CL_OPTIMIZATION.
27553 * opth-gen.awk: Test for PerFunction flag along with
27555 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
27556 it only when the latter is present. Skip those that don't in
27557 the hash function generator.
27558 * common.opt (fvar-tracking): Mark as PerFunction instead of
27560 (fvar-tracking-assignments): Likewise.
27561 (fvar-tracking-assignments-toggle): Likewise.
27562 (fvar-tracking-uninit): Likewise.
27564 2017-01-07 Jakub Jelinek <jakub@redhat.com>
27566 PR translation/79018
27567 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
27570 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
27573 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
27574 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
27575 TARGET_LONG_JUMP_TABLE_OFFSETS.
27576 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
27577 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
27578 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
27579 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
27580 * config/m68k/m68k.md (tablejump expander): Likewise.
27581 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
27582 TARGET_LONG_JUMP_TABLE_OFFSETS.
27583 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
27584 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
27586 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
27587 David Holsgrove <david.holsgrove@xilinx.com>
27589 * common/config/microblaze/microblaze-common.c
27590 (TARGET_EXCEPT_UNWIND_INFO): Remove.
27591 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
27593 * config/microblaze/microblaze.c (microblaze_must_save_register)
27594 (microblaze_expand_epilogue, microblaze_return_addr): Handle
27596 (microblaze_eh_return): New function.
27597 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
27598 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
27599 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
27600 * config/microblaze/microblaze.md (eh_return): New pattern.
27602 2017-01-06 Jakub Jelinek <jakub@redhat.com>
27604 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
27605 GCC_DIAGNOSTIC_STRINGIFY): Define.
27607 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
27609 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
27611 * config/arm/arm.md (<mcrr>): New.
27613 * config/arm/arm.c (arm_arch5te): New.
27614 (arm_option_override): Set arm_arch5te.
27615 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
27617 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
27618 (arm_mcrr_qualifiers): ... this. New.
27619 (MRRC_QUALIFIERS): Define to...
27620 (arm_mrrc_qualifiers): ... this. New.
27621 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
27623 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
27624 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
27625 (MRRCI, mrrc, MRRC): New.
27626 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
27627 VUNSPEC_MRRC2): New.
27629 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
27631 * config/arm/arm.md (<mcr>): New.
27633 * config/arm/arm.c (arm_coproc_builtin_available): Add
27634 support for mcr, mrc, mcr2 and mrc2.
27635 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
27636 (arm_mcr_qualifiers): ... this. New.
27637 (MRC_QUALIFIERS): Define to ...
27638 (arm_mrc_qualifiers): ... this. New.
27639 (MCR_QUALIFIERS): Define to ...
27640 (arm_mcr_qualifiers): ... this. New.
27641 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
27643 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
27644 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
27645 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
27646 VUNSPEC_MRC2): New.
27648 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
27650 * config/arm/arm.md (*ldc): New.
27654 * config/arm/arm.c (arm_coproc_builtin_available): Add
27655 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
27656 (arm_coproc_ldc_stc_legitimate_address): New.
27657 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
27658 'qualifier_const_pointer'.
27659 (LDC_QUALIFIERS): Define to...
27660 (arm_ldc_qualifiers): ... this. New.
27661 (STC_QUALIFIERS): Define to...
27662 (arm_stc_qualifiers): ... this. New.
27663 * config/arm/arm-protos.h
27664 (arm_coproc_ldc_stc_legitimate_address): New.
27665 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
27666 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
27667 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
27668 stc2, stcl, stc2l): New.
27669 * config/arm/constraints.md (Uz): New.
27670 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
27671 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
27672 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
27673 VUNSPEC_STC2L): New.
27675 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
27677 * config/arm/arm.md (<cdp>): New.
27678 * config/arm/arm.c (neon_const_bounds): Rename this ...
27679 (arm_const_bounds): ... this.
27680 (arm_coproc_builtin_available): New.
27681 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
27682 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
27683 (CDP_QUALIFIERS): Define to...
27684 (arm_cdp_qualifiers): ... this. New.
27686 (arm_expand_builtin_args): Add case for 6 arguments.
27687 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
27688 (arm_const_bounds): ... this.
27689 (arm_coproc_builtin_available): New.
27690 * config/arm/arm_acle.h (__arm_cdp): New.
27692 * config/arm/arm_acle_builtins.def (cdp): New.
27694 * config/arm/iterators.md (CDPI,CDP,cdp): New.
27695 * config/arm/neon.md: Rename all 'neon_const_bounds' to
27696 'arm_const_bounds'.
27697 * config/arm/types.md (coproc): New.
27698 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
27699 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
27700 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
27701 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
27703 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
27705 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
27706 (UBINOP_QUALIFIERS): New.
27708 (acle_builtin_data): New. Change comment.
27709 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
27710 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
27711 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
27712 arm_acle_builtins.def.
27713 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
27714 (arm_init_acle_builtins): New.
27715 (CRC32_BUILTIN): Remove.
27716 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
27717 crc32cb, crc32ch and crc32cw.
27718 (arm_init_crc32_builtins): Remove.
27719 (arm_init_builtins): Use arm_init_acle_builtins rather
27720 than arm_init_crc32_builtins.
27721 (arm_expand_acle_builtin): New.
27722 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
27723 * config/arm/arm_acle_builtins.def: New.
27725 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
27727 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
27728 (arm_builtin_datum): ... this.
27729 (arm_init_neon_builtin): Rename to ...
27730 (arm_init_builtin): ... this. Add a new parameters PREFIX
27731 and USE_SIG_IN_NAME.
27732 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
27733 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
27734 'arm_builtin_datum'.
27735 (arm_init_vfp_builtins): Likewise.
27736 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
27737 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
27738 (arm_expand_neon_args): Rename to ...
27739 (arm_expand_builtin_args): ... this. Rename builtin_arg
27740 enum values and differentiate between ARG_BUILTIN_MEMORY
27741 and ARG_BUILTIN_NEON_MEMORY.
27742 (arm_expand_neon_builtin_1): Rename to ...
27743 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
27744 values, arm_expand_builtin_args and add bool parameter NEON.
27745 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
27746 (arm_expand_vfp_builtin): Likewise.
27747 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
27749 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
27751 PR middle-end/77484
27752 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
27753 * predict.c (tree_estimate_probability_bb): Reverse direction of
27754 polymorphic call predictor.
27756 2017-01-06 David Malcolm <dmalcolm@redhat.com>
27758 * passes.c (execute_one_pass): Split out pass-skipping logic into...
27759 (determine_pass_name_match): ...this new function and...
27760 (should_skip_pass_p): ...this new function.
27762 2017-01-06 Nathan Sidwell <nathan@acm.org>
27764 * ipa-visibility.c (function_and_variable_visibility): Reformat
27765 comments and long lines. Remove extrneous if.
27766 * symtab.c (symtab_node::make_decl_local): Fix code format.
27767 (symtab_node::set_section_for_node): Fix comment typo.
27769 2017-01-06 Martin Liska <mliska@suse.cz>
27772 * lra-constraints.c: Rename invariant to lra_invariant.
27773 * predict.c (set_even_probabilities): Initialize e to NULL.
27775 2017-01-05 Martin Sebor <msebor@redhat.com>
27777 PR tree-optimization/78910
27778 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
27779 (format_integer): Correct off-by-one error in the handling
27780 of precision with negative numbers in signed conversions..
27782 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
27784 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
27786 2017-01-05 Jakub Jelinek <jakub@redhat.com>
27788 PR tree-optimization/71016
27789 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
27790 factor_out_conditional_conversion. Formatting fix.
27791 (factor_out_conditional_conversion): Add cond_stmt argument.
27792 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
27793 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
27796 2017-01-05 David Malcolm <dmalcolm@redhat.com>
27798 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
27799 read-rtl-function.o, and selftest-rtl.o.
27800 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
27801 (selftest::aarch64_test_loading_full_dump): New function.
27802 (selftest::aarch64_run_selftests): New function.
27803 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
27804 selftest::aarch64_run_selftests.
27805 * config/i386/i386.c
27806 (selftest::ix86_test_loading_dump_fragment_1): New function.
27807 (selftest::ix86_test_loading_call_insn): New function.
27808 (selftest::ix86_test_loading_full_dump): New function.
27809 (selftest::ix86_test_loading_unspec): New function.
27810 (selftest::ix86_run_selftests): Call the new functions.
27811 * emit-rtl.c (maybe_set_max_label_num): New function.
27812 * emit-rtl.h (maybe_set_max_label_num): New decl.
27813 * function.c (instantiate_decls): Guard call to
27814 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
27815 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
27817 * gensupport.c (gen_reader::gen_reader): Pass "false"
27818 for new "compact" param of rtx_reader.
27819 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
27820 rather than an empty string for NULL strings.
27821 * read-md.c: Potentially include config.h rather than bconfig.h.
27822 Wrap include of errors.h with #ifdef GENERATOR_FILE.
27823 (have_error): New global, copied from errors.c.
27824 (md_reader::read_name): Rename to...
27825 (md_reader::read_name_1): ...this, adding "out_loc" param,
27826 and converting "missing name or number" to returning false, rather
27828 (md_reader::read_name): Reimplement in terms of read_name_1.
27829 (md_reader::read_name_or_nil): New function.
27830 (md_reader::read_string): Handle "(nil)" by returning NULL.
27831 (md_reader::md_reader): Add new param "compact".
27832 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
27833 (md_reader::read_file): New method.
27834 * read-md.h (md_reader::md_reader): Add new param "compact".
27835 (md_reader::read_file): New method.
27836 (md_reader::is_compact): New accessor.
27837 (md_reader::read_name): Convert return type from void to file_location.
27838 (md_reader::read_name_or_nil): New decl.
27839 (md_reader::read_name_1): New decl.
27840 (md_reader::m_compact): New field.
27841 (noop_reader::noop_reader): Pass "false" for new "compact" param
27843 (rtx_reader::rtx_reader): Add new "compact" param.
27844 (rtx_reader::read_rtx_operand): Make virtual and convert return
27845 type from void to rtx.
27846 (rtx_reader::read_until): New decl.
27847 (rtx_reader::handle_any_trailing_information): New virtual function.
27848 (rtx_reader::postprocess): New virtual function.
27849 (rtx_reader::finalize_string): New virtual function.
27850 (rtx_reader::m_in_call_function_usage): New field.
27851 (rtx_reader::m_reuse_rtx_by_id): New field.
27852 * read-rtl-function.c: New file.
27853 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
27854 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
27855 (selftest::verify_three_block_rtl_cfg): New decl.
27856 * read-rtl-function.h: New file.
27857 * read-rtl.c: Potentially include config.h rather than bconfig.h.
27858 For host, include function.h, memmodel.h, and emit-rtl.h.
27859 (one_time_initialization): New function.
27860 (struct compact_insn_name): New struct.
27861 (compact_insn_names): New array.
27862 (find_code): Handle insn codes in compact dumps.
27863 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
27864 (bind_subst_iter_and_attr): Likewise.
27865 (add_condition_to_string): Likewise.
27866 (add_condition_to_rtx): Likewise.
27867 (apply_attribute_uses): Likewise.
27868 (add_current_iterators): Likewise.
27869 (apply_iterators): Likewise.
27870 (initialize_iterators): Guard usage of apply_subst_iterator with
27871 #ifdef GENERATOR_FILE.
27872 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
27873 (md_reader::read_mapping): Likewise.
27874 (add_define_attr_for_define_subst): Likewise.
27875 (add_define_subst_attr): Likewise.
27876 (read_subst_mapping): Likewise.
27877 (check_code_iterator): Likewise.
27878 (rtx_reader::read_rtx): Likewise. Move one-time initialization
27880 (one_time_initialization): New function.
27881 (rtx_reader::read_until): New method.
27882 (read_flags): New function.
27883 (parse_reg_note_name): New function.
27884 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
27885 Handle reuse_rtx ids.
27886 Wrap iterator lookup within #ifdef GENERATOR_FILE.
27887 Add parsing support for RTL dumps, mirroring the special-cases in
27888 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
27889 values, and calling handle_any_trailing_information.
27890 (rtx_reader::read_rtx_operand): Convert return type from void
27891 to rtx, returning return_rtx. Handle case 'e'. Call
27892 finalize_string on XSTR and XTMPL fields.
27893 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
27894 "(nil)" values were omitted. Call the postprocess vfunc on the
27896 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
27897 class ctor. Initialize m_in_call_function_usage. Call
27898 one_time_initialization.
27899 * rtl-tests.c (selftest::test_uncond_jump): Call
27900 set_new_first_and_last_insn.
27901 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
27902 * selftest-rtl.c: New file.
27903 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
27904 (selftest::get_insn_by_uid): New decl.
27905 * selftest-run-tests.c (selftest::run_tests): Call
27906 read_rtl_function_c_tests.
27907 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
27908 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
27911 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
27913 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
27914 operands in a special way. Assert that pos+len <= mode precision.
27916 2017-01-05 Jakub Jelinek <jakub@redhat.com>
27918 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
27919 3 argument Alias with unlimited for the negative form.
27920 (fno-vect-cost-model): Removed.
27922 2017-01-05 Martin Liska <mliska@suse.cz>
27924 * hsa-gen.c (gen_hsa_divmod): New function.
27925 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
27927 2017-01-05 Martin Liska <mliska@suse.cz>
27930 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
27933 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27935 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
27936 small constant length operands.
27938 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27940 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
27941 between loop iterations.
27943 2017-01-05 Martin Liska <mliska@suse.cz>
27946 * gimplify.c (gimplify_decl_expr): Compare to
27947 asan_poisoned_variables instread of checking flags.
27948 (gimplify_target_expr): Likewise.
27949 (gimplify_expr): Likewise.
27950 (gimplify_function_tree): Conditionally initialize
27951 asan_poisoned_variables.
27953 2017-01-04 Jeff Law <law@redhat.com>
27955 PR tree-optimizatin/78812
27956 * rtl.h (contains_mem_rtx_p): Prototype.
27957 * ifcvt.c (containts_mem_rtx_p): Move from here to...
27958 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
27959 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
27960 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
27961 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
27963 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
27965 * input.c (assert_char_at_range): Default-initialize actual_range.
27967 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
27969 * df-scan.c (df_ref_create_structure): Make regno unsigned,
27970 to match the caller.
27972 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
27974 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
27975 insns after final jump in test to emit dummy move.
27977 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
27979 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
27980 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
27982 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
27984 * multiple_target.c (create_dispatcher_calls): Init e_next.
27985 * tree-ssa-loop-split.c (split_loop): Init border.
27986 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
27989 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
27994 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
27995 (altivec_register_operand): Do not return true if the operand
27996 contains a SUBREG mixing SImode and SFmode.
27997 (vsx_register_operand): Likewise.
27998 (vsx_reg_sfsubreg_ok): New predicate.
27999 (vfloat_operand): Do not return true if the operand contains a
28000 SUBREG mixing SImode and SFmode.
28001 (vint_operand): Likewise.
28002 (vlogical_operand): Likewise.
28003 (gpc_reg_operand): Likewise.
28004 (int_reg_operand): Likewise.
28005 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
28006 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
28007 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
28009 (rs6000_emit_move_si_sf_subreg): New helper function.
28010 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
28011 fixup SUBREGs involving SImode and SFmode.
28012 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
28013 numbers for the new peephole2 optimization.
28014 (peephole2 for SFmode unions): New peephole2 to optimize cases in
28015 the GLIBC math library that do AND/IOR/XOR operations on single
28016 precision floating point.
28017 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
28018 target macros to say whether we need to avoid SUBREGs mixing
28020 (TARGET_ALLOW_SF_SUBREG): Likewise.
28021 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
28022 (UNSPEC_SI_FROM_SF): Likewise.
28023 (iorxor): Change spacing.
28024 (and_ior_xor): New iterator for AND, IOR, and XOR.
28025 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
28026 (movdi_from_sf_zero_ext): Likewise.
28027 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
28028 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
28029 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
28030 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
28031 (fms<mode>4): Likewise.
28032 (fnma<mode>4): Likewise.
28033 (fnms<mode>4): Likewise.
28034 (nfma<mode>4): Likewise.
28035 (nfms<mode>4): Likewise.
28037 2017-01-04 Marek Polacek <polacek@redhat.com>
28040 * doc/invoke.texi: Document -Wpointer-compare.
28042 2017-01-04 Jakub Jelinek <jakub@redhat.com>
28044 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
28047 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
28048 descriptions for -gdwarf-5 and emit them as uleb128 instead of
28051 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
28054 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
28055 documentation of the powerpc_popcntb_ok attribute.
28056 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
28057 code to issue warning messages if a requested CPU configuration is
28058 not supported by the binary (assembler and loader) toolchain.
28059 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
28060 made to define a built-in function that has been disabled.
28061 (paired_init_builtins): Add assertion to prevent ICE if attempt is
28062 made to define a built-in function that has been disabled.
28063 (altivec_init_builtins): Add comment explaining why definition
28064 of the DST built-in functions is not preceded by an assertion
28065 check. Add assertions to prevent ICE if attempts are made to
28066 define an altivec predicate or an abs* built-in function that has
28068 (htm_init_builtins): Add comment explaining why definition of the
28069 htm built-in functions is not preceded by an assertion check.
28071 2017-01-04 Jeff Law <law@redhat.com>
28073 PR tree-optimizatin/67955
28074 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
28075 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
28076 the points-to solution does not include pt_null. Use DECL_PT_UID
28079 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
28081 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
28082 Use gen_int_mode instead of gen_lopwart for const_int operands.
28084 2017-01-04 Jakub Jelinek <jakub@redhat.com>
28086 PR tree-optimization/71563
28087 * match.pd: Simplify X << Y into X if Y is known to be 0 or
28088 out of range value - has low bits known to be zero.
28090 2017-01-04 Alan Modra <amodra@gmail.com>
28092 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
28093 * configure: Regenerate.
28094 * config.in: Regenerate.
28096 2017-01-04 Jakub Jelinek <jakub@redhat.com>
28099 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
28100 a substring of the message, but strcmp with the whole message. Ifdef
28101 ENABLE_NLS, translate the message first using dgettext.
28103 2017-01-03 Jeff Law <law@redhat.com>
28105 PR tree-optimizatin/78856
28106 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
28107 (mark_threaded_blocks): Remove code to truncate thread paths that
28108 cross multiple loop headers. Instead invalidate the cached loop
28109 iteration information and handle case of a thread path walking
28110 into an irreducible region.
28112 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
28115 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
28116 assertions. Add support for doing the signbit if the IEEE 128-bit
28117 floating point value is in a GPR.
28118 * config/rs6000/rs6000.md (Fsignbit): Delete.
28119 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
28120 Update the length attribute if the value is in a GPR.
28121 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
28122 the sign or zero extension instruction, since the value is always 0/1.
28123 (signbit<mode>2_dm2): Delete using <Fsignbit>.
28126 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
28127 extracting SImode to a GPR register so that we can generate a
28128 store, limit the vector to be in a traditional Altivec register
28129 for the vextuwrx instruction.
28131 2017-01-03 Ian Lance Taylor <iant@google.com>
28133 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
28135 2017-01-03 Martin Sebor <msebor@redhat.com>
28137 PR tree-optimization/78696
28138 * gimple-ssa-sprintf.c (format_floating): Correct handling of
28139 precision. Use MPFR for %f for greater fidelity. Correct handling
28141 (pass_sprintf_length::compute_format_length): Set width and precision
28142 specified by asrerisk to void_node for vararg functions.
28143 (try_substitute_return_value): Adjust dump output.
28145 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
28147 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
28149 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
28151 * doc/invoke.texi (SPARC options): Document -mlra as the default.
28152 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
28153 -mlra/-mno-lra was passed to the compiler.
28155 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
28157 PR rtl-optimization/65618
28158 * emit-rtl.c (try_split): Move initialization of "before" and
28159 "after" to just before the call to emit_insn_after_setloc.
28161 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
28163 * doc/md.texi (Standard Names): Remove reference to Java frontend.
28165 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
28167 * dwarf2out.c (gen_enumeration_type_die): When
28168 -gno-strict-dwarf, add a DW_AT_encoding attribute.
28170 2017-01-03 Jakub Jelinek <jakub@redhat.com>
28172 PR tree-optimization/78965
28173 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
28174 Change first argument from const call_info & to call_info &. For %n
28175 set info.nowrite to false.
28177 PR middle-end/78901
28178 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
28179 possibly throwing calls.
28181 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
28182 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
28183 and fns handling, rather than in a separate case SSA_NAME.
28185 2017-01-02 Jeff Law <law@redhat.com>
28187 * config/darwin-driver.c (darwin_driver_init): Const-correctness
28188 fixes for first_period and second_period variables.
28190 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
28193 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
28194 (*insvqi_1): New insn pattern.
28195 (*insvqi_1_mem_rex64): Ditto.
28196 (*insvqi_2): Ditto.
28197 (*insvqi_3): Rename from *insvqi.
28199 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
28201 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
28203 * doc/cfg.texi (Edges): Remove reference to Java.
28204 (Maintaining the CFG): Ditto.
28206 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
28208 PR middle-end/77674
28209 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
28210 transparent aliases.
28212 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
28214 PR middle-end/77484
28215 * predict.def (PRED_CALL): Update hitrate.
28216 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
28217 * predict.c (tree_estimate_probability_bb): Split CALL predictor
28218 into direct/indirect/polymorphic variants.
28220 2017-01-01 Jakub Jelinek <jakub@redhat.com>
28222 Update copyright years.
28224 * gcc.c (process_command): Update copyright notice dates.
28225 * gcov-dump.c (print_version): Ditto.
28226 * gcov.c (print_version): Ditto.
28227 * gcov-tool.c (print_version): Ditto.
28228 * gengtype.c (create_file): Ditto.
28229 * doc/cpp.texi: Bump @copying's copyright year.
28230 * doc/cppinternals.texi: Ditto.
28231 * doc/gcc.texi: Ditto.
28232 * doc/gccint.texi: Ditto.
28233 * doc/gcov.texi: Ditto.
28234 * doc/install.texi: Ditto.
28235 * doc/invoke.texi: Ditto.
28237 Copyright (C) 2017 Free Software Foundation, Inc.
28239 Copying and distribution of this file, with or without modification,
28240 are permitted in any medium without royalty provided the copyright
28241 notice and this notice are preserved.