1 2017-09-26 Jakub Jelinek <jakub@redhat.com>
4 * config/i386/i386.c (ix86_print_operand_address_as): Only test
5 REGNO (base) == SP_REG if base is a REG.
8 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
9 if it is different SSA_NAME.
10 (optimize_range_tests_cmp_bitwise): New function.
11 (optimize_range_tests): Call it.
13 2017-09-26 Richard Biener <rguenther@suse.de>
15 PR tree-optimization/82321
16 * graphite.c (canonicalize_loop_closed_ssa): Properly check
17 for the def being inside the loop.
19 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
23 * config/s390/s390-builtins.def: Fix constraint on op4.
25 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
28 independent expanders.
29 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
30 ("vec_ordered", "vec_unordered"): New expanders.
32 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
37 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
41 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
43 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
45 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
46 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
47 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
49 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
51 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
53 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
55 ("*vec_slb<mode>"): New insn pattern.
56 ("vec_shr_<mode>"): New expander.
57 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
58 and force the shift count operand to V16QImode.
59 ("vec_srb<mode>"): Set shift count mode to V16QI.
61 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
63 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
64 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
65 ("vec_widen_smult_hi_<mode>"): New expander definitions.
67 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
70 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
72 2017-09-26 Richard Biener <rguenther@suse.de>
74 PR tree-optimization/82320
75 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
78 2017-09-25 Jeff Law <law@redhat.com>
80 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
81 prototype for new argument.
82 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
83 mostly extracted from rs6000_emit_allocate_stack.
84 (rs6000_emit_probe_stack_range_stack_clash): New function.
85 (rs6000_emit_allocate_stack): Call
86 rs6000_emit_probe_stack_range_stack_clash as needed.
87 (rs6000_emit_probe_stack_range): Add additional argument
88 to call to gen_probe_stack_range{si,di}.
89 (output_probe_stack_range): New.
90 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
91 (output_probe_stack_range_stack_clash): New.
92 (rs6000_emit_prologue): Emit notes into dump file as requested.
93 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
94 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
95 Add additional operand and pass it to output_probe_stack_range.
97 2017-09-25 Bin Cheng <bin.cheng@arm.com>
99 PR tree-optimization/82163
100 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
101 (checking_verify_loop_closed_ssa): New parameter.
102 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
103 (check_loop_closed_ssa_stmt): Delete.
104 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
105 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
106 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
109 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
111 * brig-builtins.def: Treat HSAIL barrier builtins as
112 setjmp/longjump style functions.
114 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
116 * target.def (constant_alignment): New hook.
117 * defaults.h (CONSTANT_ALIGNMENT): Delete.
118 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
119 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
120 * doc/tm.texi: Regenerate.
121 * targhooks.h (default_constant_alignment): Declare.
122 (constant_alignment_word_strings): Likewise.
123 * targhooks.c (default_constant_alignment): New function.
124 (constant_alignment_word_strings): Likewise.
125 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
126 instead of CONSTANT_ALIGNMENT.
127 * varasm.c (align_variable, get_variable_align, build_constant_desc)
128 (force_const_mem): Likewise.
129 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
130 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
131 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
132 (TARGET_CONSTANT_ALIGNMENT): Redefine.
133 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
135 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
136 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
137 constant_alignment_word_strings.
138 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
139 (CONSTANT_ALIGNMENT): Likewise.
140 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
141 (arm_constant_alignment): New function.
142 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
143 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
144 constant_alignment_word_strings.
145 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
146 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
147 constant_alignment_word_strings.
148 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
149 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
150 (cris_constant_alignment): New function.
151 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
152 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
153 (epiphany_constant_alignment): New function.
154 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
155 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
156 constant_alignment_word_strings.
157 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
158 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
159 constant_alignment_word_strings.
160 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
161 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
162 constant_alignment_word_strings.
163 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
164 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
165 * config/i386/i386.c (ix86_constant_alignment): Make static.
166 Use the same interface as the target hook.
167 (TARGET_CONSTANT_ALIGNMENT): Redefine.
168 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
169 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
170 constant_alignment_word_strings.
171 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
172 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
173 (TARGET_CONSTANT_ALIGNMENT): Redefine.
174 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
175 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
176 constant_alignment_word_strings.
177 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
178 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
179 constant_alignment_word_strings.
180 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
181 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
182 constant_alignment_word_strings.
183 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
184 * config/microblaze/microblaze.c (microblaze_constant_alignment):
186 (TARGET_CONSTANT_ALIGNMENT): Redefine.
187 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
188 * config/mips/mips.c (mips_constant_alignment): New function.
189 (TARGET_CONSTANT_ALIGNMENT): Redefine.
190 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
191 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
192 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
193 (mmix_constant_alignment): Make static. Use the same interface
195 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
196 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
197 constant_alignment_word_strings.
198 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
199 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
200 constant_alignment_word_strings.
201 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
202 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
203 constant_alignment_word_strings.
204 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
205 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
206 (rs6000_constant_alignment): New function.
207 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
208 * config/riscv/riscv.c (riscv_constant_alignment): New function.
209 (TARGET_CONSTANT_ALIGNMENT): Redefine.
210 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
211 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
212 (rs6000_constant_alignment): New function.
213 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
214 * config/s390/s390.c (s390_constant_alignment): New function.
215 (TARGET_CONSTANT_ALIGNMENT): Redefine.
216 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
217 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
218 constant_alignment_word_strings.
219 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
220 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
221 (sparc_constant_alignment): New function.
222 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
223 * config/spu/spu.c (spu_constant_alignment): New function.
224 (TARGET_CONSTANT_ALIGNMENT): Redefine.
225 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
226 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
227 constant_alignment_word_strings.
228 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
229 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
230 constant_alignment_word_strings.
231 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
232 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
233 constant_alignment_word_strings.
234 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
235 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
236 (visium_constant_alignment): New function.
237 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
238 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
239 (xtensa_constant_alignment): New function.
240 * system.h (CONSTANT_ALIGNMENT): Poison.
242 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
244 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
245 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
246 (rs6000_builtin_valid_without_lhs): New helper function.
247 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
248 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
250 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
252 * target.h (vec_perm_indices): Use unsigned short rather than
254 (auto_vec_perm_indices): Likewise.
255 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
256 Use unsigned int rather than unsigned char.
257 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
259 2017-09-25 Richard Biener <rguenther@suse.de>
261 * cfgloop.h (sort_sibling_loops): Declare.
262 * cfgloop.c (sort_sibling_loops_cmp): New helper.
263 (sort_sibling_loops): New function sorting the sibling loop list
265 * graphite.c (graphite_transform_loops): Sort sibling loops.
267 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
269 * target.def (vec_perm_const_ok): Change sel parameter to
271 * optabs-query.c (can_vec_perm_p): Update accordingly.
272 * doc/tm.texi: Regenerate.
273 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
274 auto_vec_perm_indices and remove separate nelt field.
275 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
276 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
277 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
278 (aarch64_expand_vec_perm_const): Update accordingly.
279 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
281 * config/arm/arm.c (expand_vec_perm_d): Change perm to
282 auto_vec_perm_indices and remove separate nelt field.
283 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
284 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
285 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
287 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
289 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
290 sel to vec_perm_indices.
291 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
292 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
293 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
295 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
298 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
301 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
302 on the FUNCTION_DECL function context if it has a DIE that is a
305 2017-09-25 Richard Biener <rguenther@suse.de>
307 PR tree-optimization/82285
308 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
311 2017-09-25 Tom de Vries <tom@codesourcery.com>
315 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
318 2017-09-25 Richard Biener <rguenther@suse.de>
320 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
321 ISL errors other than isl_error_quota happen. Dump if the
322 schedule is the same.
323 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
324 errors instead of aborting inside ISL.
326 2017-09-25 Iain Sandoe <iain@codesourcery.com>
329 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
330 of libgcc_eh for m64.
331 * config/i386/darwin64.h: Likewise.
333 2017-09-25 Richard Biener <rguenther@suse.de>
336 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
337 attribute for incomplete types nor twice for complete ones.
339 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
342 * config/i386/i386.c (ix86_print_operand_address_as): Encode
343 %esp as %rsp to avoid 0x67 prefix if there is no index or base
346 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
349 * config/i386/i386.opt (mprefer-avx256): Use
350 ix86_target_flags variable.
351 * config/i386/i386.c (ix86_target_string): Move
352 -mprefer-avx256 to flag2_opts.
354 2017-09-22 Jakub Jelinek <jakub@redhat.com>
357 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
358 and x != -1 | y != -1 into (x & y) != -1.
360 2017-09-22 Steve Ellcey <sellcey@cavium.com>
362 * config.gcc: Add new case statement to set
363 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
364 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
365 s390x-*-linux* case statements. Added aarch64 to the list of
366 supported architectures.
368 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
370 PR tree-optimization/82289
371 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
372 STMT_VINFO_RELEVANT_P.
374 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
375 Alan Hayward <alan.hayward@arm.com>
376 David Sherwood <david.sherwood@arm.com>
378 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
379 for VR_RANGE only; don't allow VR_ANTI_RANGE.
380 (extract_range_from_binary_expr_1): Don't call
381 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
383 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
384 Alan Hayward <alan.hayward@arm.com>
385 David Sherwood <david.sherwood@arm.com>
387 * target.def (preferred_vector_alignment): New hook.
388 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
390 * doc/tm.texi: Regenerate.
391 * targhooks.h (default_preferred_vector_alignment): Declare.
392 * targhooks.c (default_preferred_vector_alignment): New function.
393 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
395 (DR_TARGET_ALIGNMENT): New macro.
396 (aligned_access_p): Update commentary.
397 (vect_known_alignment_in_bytes): New function.
398 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
400 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
401 Calculate the misalignment based on the target alignment rather than
403 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
404 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
405 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
406 the target alignment, rather than masking the element misalignment
407 with the number of elements in a vector. Also use the target
408 alignment when calculating the maximum number of peels.
409 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
410 instead of TYPE_ALIGN_UNIT.
411 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
412 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
413 (vect_create_addr_base_for_vector_ref): Update call accordingly.
414 (vect_create_data_ref_ptr): Likewise.
415 (vect_setup_realignment): Realign by ANDing with
416 -DR_TARGET_MISALIGNMENT.
417 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
418 the number of peels based on DR_TARGET_ALIGNMENT.
419 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
420 with the guaranteed alignment boundary when deciding whether
422 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
423 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
424 (ensure_base_align): Remove stmt_info parameter. Get the
425 target base alignment from DR_TARGET_ALIGNMENT.
426 (vectorizable_store): Update call accordingly. Interpret
427 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
429 (vectorizable_load): Likewise.
431 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
432 Alan Hayward <alan.hayward@arm.com>
433 David Sherwood <david.sherwood@arm.com>
435 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
436 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
437 (vect_enhance_data_refs_alignment): Likewise.
439 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
441 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
442 error. Only quit immediately if parsing is complete.
443 (BEGIN): Initialize fatal_err and parse_done.
444 (begin fpu, end fpu): Check number of arguments.
445 (begin arch, end arch): Likewise.
446 (begin cpu, end cpu): Likewise.
447 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
448 (optalias): Likewise.
450 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
452 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
453 * config/arm/arm-isa.h: Delete. Move definitions to ...
454 * arm-cpus.in: ... here. Use new feature and fgroup values.
455 * config/arm/arm.c (arm_option_override): Use lower case for feature
457 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
458 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
459 * config/arm/parsecpu.awk (END): Add new command 'isa'.
461 (print_isa_bits_for): New function.
462 (gen_isa): New function.
463 (gen_comm_data): Use print_isa_bits_for.
464 (define feature): New keyword.
465 (define fgroup): New keyword.
466 * config/arm/t-arm (TM_H): Remove.
467 (GTM_H): Add arm-isa.h.
468 (arm-isa.h): Add rule to generate file.
469 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
470 case for feature bit names.
472 2017-09-22 Richard Biener <rguenther@suse.de>
474 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
476 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
477 print of no dependency loops ...
478 * graphite.c (graphite_transform_loops): ... here.
479 (canonicalize_loop_closed_ssa_form): Work from inner to outer
481 (same_close_phi_node, remove_duplicate_close_phi,
482 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
483 (canonicalize_loop_closed_ssa): ... here and simplify.
484 * graphite-optimize-isl.c: Include tree-vectorizer.h.
485 (optimize_isl): Use dump_printf_loc to tell when we stopped
486 optimizing because of an ISL timeout.
488 2017-09-22 Richard Biener <rguenther@suse.de>
490 PR tree-optimization/82291
491 * tree-if-conv.c (predicate_mem_writes): Make sure to
492 remove writes in blocks predicated with false.
494 2017-09-22 Richard Biener <rguenther@suse.de>
496 * sese.c: Include cfganal.h.
497 (if_region_set_false_region): Remove.
498 (create_if_region_on_edge): Likewise.
499 (move_sese_in_condition): Re-implement without destroying
502 2017-09-22 Richard Biener <rguenther@suse.de>
504 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
505 Verify both BBs contain loop PHI nodes before dispatching to
507 (graphite_regenerate_ast_isl): Do not recompute dominators,
508 do not verify three times. Restructure for clarity.
509 * graphite-scop-detection.c (same_close_phi_node,
510 remove_duplicate_close_phi, make_close_phi_nodes_unique,
511 defined_in_loop_p, canonicalize_loop_closed_ssa,
512 canonicalize_loop_closed_ssa_form): Simplify, remove excess
513 checking and SSA rewrite, move to ...
514 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
515 (graphite_initialize): Do not pass in ctx, do not reset the
516 SCEV cache, compute only dominators.
517 (graphite_transform_loops): Allocate ISL ctx after
518 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
519 Maintain post-dominators only around build_scops.
520 * sese.c (if_region_set_false_region): Make static. Free
521 and recompute dominators.
522 (move_sese_in_condition): Assert we don't get called with
523 post-dominators computed.
524 * sese.h (if_region_set_false_region): Remove.
526 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
528 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
529 mode attribute for TARGET_AVX512VL.
531 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
533 * config/i386/i386.opt (mprefer-avx256): New option.
534 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
536 (ix86_preferred_simd_mode): Return 256-bit AVX modes
537 for TARGET_PREFER_AVX256.
538 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
540 2017-09-21 Jeff Law <law@redhat.com>
542 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
543 Fix dump output if the only stack space is for pushed registers.
545 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
547 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
550 2017-09-21 Martin Sebor <msebor@redhat.com>
553 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
554 code (in C++) or code that triggers warnings.
556 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
558 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
560 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
562 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
563 * sched-rgn.c: Ditto.
564 * sel-sched-ir.c: Ditto.
566 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
568 * toplev.h (set_random_seed): Adjust return type.
569 * toplev.c (init_local_tick): Move eager initialization of random_seed
570 to get_random_seed. Adjust comment.
571 (init_random_seed): Inline to get_random_seed, delete.
572 (get_random_seed): Initialize random_seed lazily.
573 (set_random_seed): Do not return previous value.
574 (print_switch_value): Do not call get_random_seed.
576 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
578 * cgraph.c (delete_function_version): New, broken out from...
579 (cgraph_node::delete_function_version): ...here. Rename to
580 cgraph_node::delete_function_version_by_decl. Update all uses.
581 (cgraph_node::remove): Call delete_function_version.
583 2017-09-21 Jakub Jelinek <jakub@redhat.com>
586 * tree-inline.c (expand_call_inline): Emit clobber stmts for
587 VAR_DECLs to which addressable non-volatile parameters are mapped
588 and for id->retvar after the return value assignment. Clear
589 id->retval and id->retbnd after inlining.
591 2017-09-21 Richard Biener <rguenther@suse.de>
593 PR tree-optimization/82276
594 PR tree-optimization/82244
595 * tree-vrp.c (build_assert_expr_for): Set
596 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
598 (remove_range_assertions): Revert earlier change.
600 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
603 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
605 2017-09-21 Richard Biener <rguenther@suse.de>
607 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
608 Restore valid IL after code generation errors.
609 * graphite.c (graphite_transform_loops): Diagnose code
610 generation issues as MSG_MISSED_OPTIMIZATION and continue
611 with processing SCOPs.
613 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
614 Alan Hayward <alan.hayward@arm.com>
615 David Sherwood <david.sherwood@arm.com>
617 * calls.c (compute_argument_addresses): Use simplify_gen_binary
618 rather than choosing between plus_constant and gen_rtx_<CODE>.
619 * expr.c (emit_push_insn): Likewise.
620 (expand_expr_real_2): Likewise.
622 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
623 Alan Hayward <alan.hayward@arm.com>
624 David Sherwood <david.sherwood@arm.com>
626 * loop-unroll.c (split_iv): Call copy_rtx on the step.
628 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
629 Alan Hayward <alan.hayward@arm.com>
630 David Sherwood <david.sherwood@arm.com>
632 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
633 calling tree_to_uhwi.
635 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
636 Alan Hayward <alan.hayward@arm.com>
637 David Sherwood <david.sherwood@arm.com>
639 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
640 INTEGER_CST rather than a negative test for ADDR_EXPR.
642 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
643 Alan Hayward <alan.hayward@arm.com>
644 David Sherwood <david.sherwood@arm.com>
646 * tree-vrp.c (extract_range_from_binary_expr_1): Check
647 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
649 2017-09-21 Richard Biener <rguenther@suse.de>
651 PR tree-optimization/71351
652 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
653 graphite_create_new_loop_guard): Remove, fold remaining parts
655 (translate_isl_ast_node_for): ... here and simplify.
657 2017-09-21 Jakub Jelinek <jakub@redhat.com>
660 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
661 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
662 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
663 alternative always use QI mode, for -Os imov (=R,R) alternative
664 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
667 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
668 Jeff Law <law@redhat.com>
670 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
671 (allocate_stack_space): New function, partially extracted from
673 (s390_emit_prologue): Track offset to most recent stack probe.
674 Code to allocate space moved into allocate_stack_space.
675 Dump actions when no stack is allocated.
676 (s390_prologue_plus_offset): New function.
677 (s390_emit_stack_probe): Likewise.
679 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
681 * common.opt (Wa, Wl, Wp, g, gz=): Add
683 (gno-column-info): Remove.
684 (gcolumn-info): Drop RejectNegative.
686 (gno-record-gcc-switches): Remove.
687 (grecord-gcc-switches): Drop RejectNegative.
688 (gno-split-dwarf): Remove.
689 (gsplit-dwarf): Drop RejectNegative.
690 (gno-strict-dwarf): Remove.
691 (gstrict-dwarf): Drop RejectNegative.
692 * config/darwin.opt (gfull, gused): Add RejectNegative.
693 * dwarf2out.c (gen_producer_string): Drop
694 gno-record-gcc-switches handler.
695 * optc-gen.awk: Add g to prefixes with negative forms.
696 * opts-common.c (remapping_prefix_p): New.
697 (find_opt): Check it.
698 (generate_canonical_option): Test g prefix.
699 (option_map): Add -gno- mapping.
700 (add_misspelling_candidates): Check remapping_prefix_p.
702 2017-09-20 Jeff Law <law@redhat.com>
704 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
705 thinko in stack clash protection support.
707 * explow.c (compute_stack_clash_protection_loop_data): Use
708 CONST_INT_P instead of explicit test. Verify object is a
709 CONST_INT_P before looking at INTVAL.
710 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
711 instead of explicit test.
713 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
716 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
717 address instead of to r1 and r11.
719 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
721 * config.gcc: Support "knm".
722 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
723 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
725 * config/i386/i386.c (m_KNM): Define.
726 (processor_target_table): Add "knm".
728 (ix86_option_override_internal): Add "knm".
729 (ix86_issue_rate): Add PROCESSOR_KNM.
730 (ix86_adjust_cost): Ditto.
731 (ia32_multipass_dfa_lookahead): Ditto.
732 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
733 (fold_builtin_cpu): Add M_INTEL_KNM.
734 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
735 (processor_type): Add PROCESSOR_KNM.
736 * config/i386/x86-tune.def: Add m_KNM.
737 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
739 2017-09-20 Richard Biener <rguenther@suse.de>
741 PR tree-optimization/80213
742 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
743 are allowed in empty BBs as well.
744 (canonicalize_loop_closed_ssa): Also look for other complex
746 (scop_detection::get_sese): Include the loop-closed PHI block
748 (scop_detection::merge_sese): Remove code adding extra blocks.
749 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
750 (build_scops): Assert the final returned scop is invalid.
752 2017-09-20 Richard Biener <rguenther@suse.de>
754 PR tree-optimization/82264
755 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
757 (vn_phi_lookup): Likewise.
758 (vn_phi_insert): Likewise.
760 2017-09-20 Jakub Jelinek <jakub@redhat.com>
762 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
763 that fits into uhwi or shwi, add DW_AT_const_value regardless
764 of early_dwarf without going through RTL, using add_AT_unsigned
767 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
768 (DEBUG_LTO_ABBREV_SECTION): Likewise.
769 (DEBUG_LTO_MACINFO_SECTION): Likewise.
770 (DEBUG_MACRO_SECTION): Likewise.
771 (DEBUG_LTO_MACRO_SECTION): Likewise.
772 (DEBUG_STR_DWO_SECTION): Likewise.
773 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
774 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
775 (DEBUG_LTO_DWO_LINE_SECTION): Define.
776 (DEBUG_LTO_LINE_STR_SECTION): Define.
777 (init_sections_and_labels): Initialize debug_line_str_section
778 variable. Initialize debug_loc_section for -gdwarf-5 to
779 DEBUG_LOCLISTS_SECTION. Formatting fixes.
781 2017-09-20 Richard Biener <rguenther@suse.de>
783 * graphite-sese-to-poly.c (extract_affine): Properly handle
784 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
786 2017-09-20 Richard Biener <rguenther@suse.de>
788 PR tree-optimization/81373
789 * graphite-scop-detection.c (build_cross_bb_scalars_def):
790 Force SESE live-out defs to be handled even if they are
793 2017-09-19 Jeff Law <law@redhat.com>
795 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
796 nothing for stack adjustments with REG_STACK_CHECK.
797 * sched-deps.c (parse_add_or_inc): Reject insns with
798 REG_STACK_CHECK from dependency breaking.
799 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
800 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
801 * reg-notes.def (STACK_CHECK): New note.
803 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
804 (ix86_expand_prologue): Dump stack clash info as needed.
805 Call ix86_adjust_stack_and_probe_stack_clash as needed.
807 * function.c (dump_stack_clash_frame_info): New function.
808 * function.h (dump_stack_clash_frame_info): Prototype.
809 (enum stack_clash_probes): New enum.
811 * config/alpha/alpha.c (alpha_expand_prologue): Also check
812 flag_stack_clash_protection.
813 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
814 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
815 (arm_frame_pointer_required): Likewise.
816 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
817 (ia64_expand_prologue): Likewise.
818 * config/mips/mips.c (mips_expand_prologue): Likewise.
819 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
820 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
821 (sparc_flat_expand_prologue): Likewise.
822 * config/spu/spu.c (spu_expand_prologue): Likewise.
824 * explow.c: Include "params.h".
825 (anti_adjust_stack_and_probe_stack_clash): New function.
826 (get_stack_check_protect): Likewise.
827 (compute_stack_clash_protection_loop_data): Likewise.
828 (emit_stack_clash_protection_loop_start): Likewise.
829 (emit_stack_clash_protection_loop_end): Likewise.
830 (allocate_dynamic_stack_space): Use get_stack_check_protect.
831 Use anti_adjust_stack_and_probe_stack_clash.
832 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
833 (emit_stack_clash_protection_loop_start): Likewise.
834 (emit_stack_clash_protection_loop_end): Likewise.
835 * rtl.h (get_stack_check_protect): Prototype.
836 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
837 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
838 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
840 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
842 * doc/tm.texi: Rebuilt.
843 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
844 get_stack_check_protect.
845 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
846 * config/arm/arm.c (arm_expand_prologue): Likewise.
847 (arm_frame_pointer_required): Likewise.
848 * config/i386/i386.c (ix86_expand_prologue): Likewise.
849 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
850 * config/mips/mips.c (mips_expand_prologue): Likewise.
851 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
852 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
853 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
854 (sparc_flat_expand_prologue): Likewise.
856 * common.opt (-fstack-clash-protection): New option.
857 * flag-types.h (enum stack_check_type): Note difference between
858 -fstack-check= and -fstack-clash-protection.
859 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
860 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
861 * toplev.c (process_options): Issue warnings/errors for cases
862 not handled with -fstack-clash-protection.
863 * doc/invoke.texi (-fstack-clash-protection): Document new option.
864 (-fstack-check): Note additional problem with -fstack-check=generic.
865 Note that -fstack-check is primarily for Ada and refer users
866 to -fstack-clash-protection for stack-clash-protection.
867 Document new params for stack clash protection.
869 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
871 * config/i386/i386.c (ix86_split_long_move): Do not handle
872 address used for LEA in a special way.
874 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
876 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
878 2017-09-19 Martin Sebor <msebor@redhat.com>
881 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
882 of incompatible types.
884 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
886 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
887 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
888 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
889 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
891 2017-09-19 Richard Biener <rguenther@suse.de>
893 PR tree-optimization/82244
894 * tree-vrp.c (remove_range_assertions): Do not propagate
895 a constant to abnormals but replace the assert with a copy.
897 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
899 PR rtl-optimization/57878
900 PR rtl-optimization/68988
901 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
902 avoidance test involving non_reload_pseudos. Move frequency test
903 below the general fragmentation avoidance test.
905 2017-09-19 Richard Biener <rguenther@suse.de>
907 PR tree-optimization/69728
908 * graphite-sese-to-poly.c (schedule_error): New global.
909 (add_loop_schedule): Handle empty domain by failing the
911 (build_original_schedule): Handle schedule_error.
913 2017-09-19 Richard Biener <rguenther@suse.de>
915 * graphite-scop-detection.c (scop_detection::can_represent_loop):
916 Do not iterate to sibling loops but only to siblings of inner
919 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
922 * config/m68k/m68k.md (moveq feeding equality comparison): Check
923 that the registers are different.
925 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
927 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
928 to processor_model and "amdfam17h" to arch_names_table.
929 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
931 2017-09-18 Jakub Jelinek <jakub@redhat.com>
934 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
936 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
937 Alan Hayward <alan.hayward@arm.com>
938 David Sherwood <david.sherwood@arm.com>
940 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
942 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
944 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
945 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
946 vect_schedule_slp_instance.
947 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
948 Update call to vect_slp_analyze_node_operations. Simplify return
950 (vect_slp_analyze_bb_1): Update call accordingly.
951 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
952 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
953 (vect_schedule_slp): Update call accordingly.
955 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
956 Alan Hayward <alan.hayward@arm.com>
957 David Sherwood <david.sherwood@arm.com>
959 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
960 with types that aren't in fact scalar.
962 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
964 * tree-vect-slp.c (vect_record_max_nunits): New function,
966 (vect_build_slp_tree_1): ...here.
967 (vect_build_slp_tree_2): Call it for phis too.
969 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
971 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
972 to vect_get_vec_def_for_operand when getting the mask operand.
974 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
975 Alan Hayward <alan.hayward@arm.com>
976 David Sherwood <david.sherwood@arm.com>
978 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
981 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
982 Alan Hayward <alan.hayward@arm.com>
983 David Sherwood <david.sherwood@arm.com>
985 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
986 calculation for vector booleans.
988 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
989 Alan Hayward <alan.hayward@arm.com>
990 David Sherwood <david.sherwood@arm.com>
992 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
994 (vect_transform_stmt): ...here.
995 (vect_analyze_stmt): Use it instead of calling
996 vectorizable_live_operation directly.
998 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
1000 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
1001 non-SIMT targets in acc vector loops.
1003 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
1005 * configure.ac: Add arc and check if assembler supports gdwarf2.
1006 * configure: Regenerate.
1008 2017-09-18 Richard Biener <rguenther@suse.de>
1010 PR tree-optimization/82220
1011 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
1012 epilogue niters from the min_profitable_iters compute.
1014 2017-09-18 Jakub Jelinek <jakub@redhat.com>
1017 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
1018 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
1019 (ix86_init_pic_reg): Revert 2017-09-01 changes.
1021 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
1024 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
1025 switching to a new text section.
1027 2017-09-18 Richard Biener <rguenther@suse.de>
1029 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
1031 (build_alias_set): Reject aliases with no access function.
1033 2017-09-18 Richard Biener <rguenther@suse.de>
1035 PR tree-optimization/79622
1036 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
1038 (build_cross_bb_scalars_use): Likewise.
1040 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
1042 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
1044 2017-09-18 Alan Modra <amodra@gmail.com>
1047 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
1048 stack_pointer_rtx for count 0. Update comments. Break up
1049 large rtl expression.
1051 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
1053 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
1054 Increase to 20 bytes.
1055 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
1056 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
1057 or avx version of the stub.
1059 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
1062 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
1063 compute the minimum stack alignment. Also update preferred stack
1064 boundary for leaf functions.
1066 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
1068 PR tree-optimization/82228
1069 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
1072 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
1074 * common/config/nds32/nds32-common.c
1075 (nds32_option_optimization_table): Refine formatting.
1076 (nds32_option_optimization_table): Use -fsched-pressure and
1077 -fomit-frame-pointer for specific optimization level.
1079 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
1081 * config/nds32/nds32.c: Refine formatting and comments.
1082 * config/nds32/nds32.h: Likewise.
1083 * config/nds32/nds32.md: Likewise.
1084 * config/nds32/nds32-cost.c: Likewise.
1085 * config/nds32/nds32-isr.c: Likewise.
1086 * config/nds32/nds32-md-auxiliary.c: Likewise.
1087 * config/nds32/nds32-multiple.md: Likewise.
1088 * config/nds32/nds32-predicates.c: Likewise.
1090 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
1091 Jakub Jelinek <jakub@redhat.com>
1093 Add support for -std=c++2a.
1094 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
1096 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
1098 2017-09-15 Steve Ellcey <sellcey@cavium.com>
1101 * doc/extend.texi (Common Function Attributes): Add
1102 references to ARM, AArch64, and S/390 specific attributes.
1103 (Function Specific Option Pragmas): Add AArch64 and S/390
1104 to list of back ends that support the target pragma.
1106 2017-09-15 Nathan Sidwell <nathan@acm.org>
1108 * doc/standards.texi: Fix C++17 description. Update URLs for
1111 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
1113 * common.opt (Wcast-align=strict): New warning option.
1114 * doc/invoke.texi: Document -Wcast-align=strict.
1116 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
1118 * cgraph.h (cgraph_thunk_info): Add comments.
1119 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
1120 assert for VIRTUAL_* arguments stricter.
1122 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
1124 PR tree-optimization/71026
1125 * match.pd: Move RDIV patterns from fold-const.c
1126 * fold-const.c (distribute_real_division): Removed.
1127 (fold_binary_loc): Remove calls to distribute_real_divison.
1129 2017-09-15 Jakub Jelinek <jakub@redhat.com>
1131 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
1132 c++1z and gnu++1z as deprecated. Change other references to
1133 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
1134 Change -Wc++1z-compat to -Wc++17-compat.
1135 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
1136 * dwarf2out.c (highest_c_language): Handle C++17.
1137 (gen_compile_unit_die): Likewise.
1139 2017-09-15 Jakub Jelinek <jakub@redhat.com>
1141 PR rtl-optimization/82192
1142 * combine.c (make_extraction): Don't look through non-paradoxical
1143 SUBREGs or TRUNCATE if pos + len is or might be bigger than
1146 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
1147 Alan Hayward <alan.hayward@arm.com>
1148 David Sherwood <david.sherwood@arm.com>
1150 * target.def (function_arg_offset): New hook.
1151 * targhooks.h (default_function_arg_offset): Declare.
1152 * targhooks.c (default_function_arg_offset): New function.
1153 * function.c (locate_and_pad_parm): Use
1154 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
1155 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
1156 (TARGET_FUNCTION_ARG_OFFSET): ...this.
1157 * doc/tm.texi: Regenerate.
1158 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
1159 * config/spu/spu.c (spu_function_arg_offset): New function.
1160 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
1161 * system.h (FUNCTION_ARG_OFFSET): Poison.
1163 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
1164 Alan Hayard <alan.hayward@arm.com>
1165 David Sherwood <david.sherwood@arm.com>
1167 * target.def (truly_noop_truncation): New hook.
1168 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
1169 than TRULY_NOOP_TRUNCATION.
1170 * hooks.h (hook_bool_uint_uint_true): Declare.
1171 * hooks.c (hook_bool_uint_uint_true): New function.
1172 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
1173 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
1174 * doc/tm.texi: Regenerate.
1175 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
1176 rather than TRULY_NOOP_TRUNCATION in comments.
1177 (simplify_comparison): Likewise.
1178 (record_truncated_value): Likewise.
1179 * expmed.c (extract_bit_field_1): Likewise.
1180 (extract_split_bit_field): Likewise.
1181 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
1182 instead of TRULY_NOOP_TRUNCATION.
1183 * function.c (assign_parm_setup_block): Likewise.
1184 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
1185 * rtlhooks.c: Include target.h.
1186 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
1187 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
1188 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
1189 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
1190 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
1191 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
1192 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
1193 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
1194 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
1195 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
1196 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
1197 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
1198 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
1199 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
1200 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
1201 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
1202 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
1203 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
1204 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
1205 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
1206 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
1207 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
1208 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
1209 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
1210 * config/mips/mips.c (mips_truly_noop_truncation): New function.
1211 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
1212 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
1213 rather than TRULY_NOOP_TRUNCATION in comments.
1214 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
1215 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
1216 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
1217 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
1218 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
1219 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
1220 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
1221 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
1222 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
1223 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
1224 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
1225 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
1226 rather than TRULY_NOOP_TRUNCATION in comments.
1227 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
1228 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
1229 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
1230 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
1231 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
1232 TRULY_NOOP_TRUNCATION condition.
1233 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
1234 (TRULY_NOOP_TRUNCATION): Delete.
1235 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
1236 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
1237 * config/spu/spu.c (spu_truly_noop_truncation): New function.
1238 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
1239 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
1240 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
1241 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
1242 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
1243 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
1244 rather than TRULY_NOOP_TRUNCATION in comments.
1245 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
1246 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
1247 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
1248 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
1249 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
1250 * system.h (TRULY_NOOP_TRUNCATION): Poison.
1252 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
1255 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
1256 (*cmp_ior): Likewise.
1257 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
1258 (*ior_scc_scc_cmp): Likewise.
1259 (*and_scc_scc): Likewise.
1260 (*and_scc_scc_cmp): Likewise.
1262 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
1263 Alan Hayard <alan.hayward@arm.com>
1264 David Sherwood <david.sherwood@arm.com>
1266 * target.def (can_change_mode_class): New hook.
1267 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
1268 (hard_regno_nregs): Likewise.
1269 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
1270 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
1271 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
1272 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
1273 (LOAD_EXTEND_OP): Update accordingly.
1274 * doc/tm.texi: Regenerate.
1275 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
1276 CANNOT_CHANGE_MODE_CLASS.
1277 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
1278 (REG_CAN_CHANGE_MODE_P): ...this new macro.
1279 * combine.c (simplify_set): Update accordingly.
1280 * emit-rtl.c (validate_subreg): Likewise.
1281 * recog.c (general_operand): Likewise.
1282 * regcprop.c (mode_change_ok): Likewise.
1283 * reload1.c (choose_reload_regs): Likewise.
1284 (inherit_piecemeal_p): Likewise.
1285 * rtlanal.c (simplify_subreg_regno): Likewise.
1286 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
1287 instead of CANNOT_CHANGE_MODE_CLASS.
1288 (reload_cse_simplify_operands): Likewise.
1289 * reload.c (push_reload): Use targetm.can_change_mode_class
1290 instead of CANNOT_CHANGE_MODE_CLASS.
1291 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
1292 REG_CANNOT_CHANGE_MODE_P.
1293 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1294 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
1295 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1296 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1297 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1298 (arm_can_change_mode_class): New function.
1299 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
1300 than CANNOT_CHANGE_MODE_CLASS in comments.
1301 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1302 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
1303 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
1304 (ix86_can_change_mode_class): ...this new function, inverting the
1305 sense of the return value.
1306 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1307 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1308 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1309 (ia64_can_change_mode_class): New function.
1310 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1311 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
1312 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
1313 (m32c_can_change_mode_class): ...this new function, inverting the
1314 sense of the return value.
1315 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1316 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1317 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
1318 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
1319 (mips_can_change_mode_class): ...this new function, inverting the
1320 sense of the return value.
1321 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1322 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1323 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1324 (msp430_can_change_mode_class): New function.
1325 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1326 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
1327 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1328 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1329 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1330 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
1331 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1332 (pa_cannot_change_mode_class): Replace with...
1333 (pa_can_change_mode_class): ...this new function, inverting the
1334 sense of the return value.
1335 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
1336 than CANNOT_CHANGE_MODE_CLASS in comments.
1337 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1338 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
1339 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1340 (pdp11_cannot_change_mode_class): Replace with...
1341 (pdp11_can_change_mode_class): ...this new function, inverting the
1342 sense of the return value.
1343 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1344 * config/powerpcspe/powerpcspe-protos.h
1345 (rs6000_cannot_change_mode_class_ptr): Delete.
1346 * config/powerpcspe/powerpcspe.c
1347 (rs6000_cannot_change_mode_class_ptr): Delete.
1348 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1349 (rs6000_option_override_internal): Assign to
1350 targetm.can_change_mode_class instead of
1351 rs6000_cannot_change_mode_class_ptr.
1352 (rs6000_cannot_change_mode_class): Replace with...
1353 (rs6000_can_change_mode_class): ...this new function, inverting the
1354 sense of the return value.
1355 (rs6000_debug_cannot_change_mode_class): Replace with...
1356 (rs6000_debug_can_change_mode_class): ...this new function.
1357 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1358 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
1359 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1360 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1361 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
1363 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
1364 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1365 (rs6000_option_override_internal): Assign to
1366 targetm.can_change_mode_class instead of
1367 rs6000_cannot_change_mode_class_ptr.
1368 (rs6000_cannot_change_mode_class): Replace with...
1369 (rs6000_can_change_mode_class): ...this new function, inverting the
1370 sense of the return value.
1371 (rs6000_debug_cannot_change_mode_class): Replace with...
1372 (rs6000_debug_can_change_mode_class): ...this new function.
1373 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1374 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
1375 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
1376 (s390_can_change_mode_class): ...this new function, inverting the
1377 sense of the return value.
1378 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1379 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1380 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
1381 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1382 (sh_cannot_change_mode_class): Replace with...
1383 (sh_can_change_mode_class): ...this new function, inverting the
1384 sense of the return value.
1385 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1386 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1387 (sparc_can_change_mode_class): New function.
1388 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1389 * config/spu/spu.c (spu_can_change_mode_class): New function.
1390 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1391 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
1392 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
1393 (visium_can_change_mode_class): New function.
1394 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
1396 2017-09-15 Richard Biener <rguenther@suse.de>
1398 PR tree-optimization/82217
1399 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
1400 but not undefined case.
1402 2017-09-15 Jakub Jelinek <jakub@redhat.com>
1405 * postreload.c (reload_cse_simplify_operands): Skip
1406 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
1408 2017-09-15 Richard Biener <rguenther@suse.de>
1410 PR tree-optimization/68823
1411 * graphite-scop-detection.c (build_alias_set): If we have a
1412 possible dependence check whether we can handle them by just
1413 looking at the DRs DR_ACCESS_FNs.
1414 (build_scops): If build_alias_set fails, fail the SCOP.
1416 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
1418 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
1419 to support float128 built-in functions that require the ISA 3.0
1421 (BU_FLOAT128_3_HW): Likewise.
1422 (SQRTF128): Add support for the IEEE 128-bit square root and fma
1424 (FMAF128): Likewise.
1426 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
1427 support for built-in functions that need the ISA 3.0 IEEE 128-bit
1428 floating point instructions.
1429 (rs6000_invalid_builtin): Likewise.
1430 (rs6000_builtin_mask_names): Likewise.
1431 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
1432 (RS6000_BTM_FLOAT128_HW): Likewise.
1433 (RS6000_BTM_COMMON): Likewise.
1434 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
1436 * doc/extend.texi (RS/6000 built-in functions): Document the
1437 IEEE 128-bit floating point square root and fused multiply-add
1440 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
1442 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
1443 reg (r2) isn't in the set of registers defined in the prologue.
1445 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
1446 Alan Hayward <alan.hayward@arm.com>
1447 David Sherwood <david.sherwood@arm.com>
1449 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
1450 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
1451 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
1452 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
1453 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
1455 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
1456 max_vectorization_factor.
1457 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
1459 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
1460 Alan Hayward <alan.hayward@arm.com>
1461 David Sherwood <david.sherwood@arm.com>
1463 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
1464 (vect_worthwhile_without_simd_p): Declare.
1465 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
1466 (vectorizable_reduction): Use it.
1467 * tree-vect-stmts.c (vectorizable_shift): Likewise.
1468 (vectorizable_operation): Likewise.
1470 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
1471 Alan Hayward <alan.hayward@arm.com>
1472 David Sherwood <david.sherwood@arm.com>
1474 * tree-vectorizer.h (vect_get_num_copies): New function.
1475 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
1476 * tree-vect-loop.c (vectorizable_reduction): Likewise.
1477 (vectorizable_induction): Likewise.
1478 (vectorizable_live_operation): Likewise.
1479 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
1480 (vectorizable_bswap): Likewise.
1481 (vectorizable_call): Likewise.
1482 (vectorizable_conversion): Likewise.
1483 (vectorizable_assignment): Likewise.
1484 (vectorizable_shift): Likewise.
1485 (vectorizable_operation): Likewise.
1486 (vectorizable_store): Likewise.
1487 (vectorizable_load): Likewise.
1488 (vectorizable_condition): Likewise.
1489 (vectorizable_comparison): Likewise.
1490 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
1492 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
1493 Alan Hayward <alan.hayward@arm.com>
1494 David Sherwood <david.sherwood@arm.com>
1496 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
1497 of vect_init_vector.
1499 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
1500 Alan Hayward <alan.hayward@arm.com>
1501 David Sherwood <david.sherwood@arm.com>
1503 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
1504 an inline wrapper that provides a location.
1505 (gimple_build_vector): Likewise.
1506 * gimple-fold.c (gimple_build_vector_from_val): New function.
1507 (gimple_build_vector): Likewise.
1508 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
1509 functions to build the initial value. Always return a gimple value.
1510 (get_initial_defs_for_reduction): Likewise. Only compute
1512 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
1513 vect_init_vector on the results from get_initial_def(s)_for_reduction.
1514 (vectorizable_induction): Use gimple_build_vector rather than
1517 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
1518 Alan Hayward <alan.hayward@arm.com>
1519 David Sherwood <david.sherwood@arm.com>
1521 * target.h (vec_perm_indices): New typedef.
1522 (auto_vec_perm_indices): Likewise.
1523 * optabs-query.h: Include target.h
1524 (can_vec_perm_p): Take a vec_perm_indices *.
1525 * optabs-query.c (can_vec_perm_p): Likewise.
1526 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
1527 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
1528 * tree-vect-generic.c (lower_vec_perm): Likewise.
1529 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
1530 (vect_grouped_load_supported): Likewise.
1531 (vect_shift_permute_load_chain): Likewise.
1532 (vect_permute_store_chain): Use auto_vec_perm_indices.
1533 (vect_permute_load_chain): Likewise.
1534 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
1535 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
1536 Update uses of can_vec_perm_p.
1537 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
1538 mode with a number of elements. Take a vec_perm_indices *.
1539 (vect_create_epilog_for_reduction): Update accordingly.
1540 Use auto_vec_perm_indices.
1541 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
1542 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
1543 (vect_transform_slp_perm_load): Likewise.
1544 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
1545 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
1546 (vect_gen_perm_mask_checked): Likewise.
1547 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
1548 (vect_gen_perm_mask_checked): Likewise.
1549 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
1550 (vectorizable_store): Likewise.
1551 (vectorizable_load): Likewise.
1552 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
1553 (vectorizable_bswap): Likewise.
1555 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
1556 Alan Hayward <alan.hayward@arm.com>
1557 David Sherwood <david.sherwood@arm.com>
1559 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
1560 * tree.c (build_vector): Likewise.
1561 (build_vector_from_ctor): Update accordingly.
1562 (build_vector_from_val): Likewise.
1563 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
1564 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
1565 * tree-vect-generic.c (add_rshift): Likewise.
1566 (expand_vector_divmod): Likewise.
1567 (optimize_vector_constructor): Likewise.
1568 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
1569 (vect_transform_slp_perm_load): Likewise.
1570 (vect_schedule_slp_instance): Likewise.
1571 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
1572 (vectorizable_call): Likewise.
1573 (vect_gen_perm_mask_any): Likewise. Add elements in order.
1574 * expmed.c (make_tree): Likewise.
1575 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
1576 a vector passed to build_vector.
1577 (fold_convert_const): Likewise.
1578 (exact_inverse): Likewise.
1579 (fold_ternary_loc): Likewise.
1580 (fold_relational_const): Likewise.
1581 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
1582 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
1583 (const_unop): Likewise. Store the reduction accumulator in a
1584 variable rather than an array.
1585 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
1586 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
1587 the new vector, rather than constructing it after the input arrays.
1588 (native_interpret_vector): Use auto_vec<tree> when building
1589 a vector passed to build_vector. Add elements in order.
1590 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
1591 auto_vec<tree> when building a vector passed to build_vector.
1592 (vect_create_epilog_for_reduction): Likewise.
1593 (vectorizable_induction): Likewise.
1594 (get_initial_def_for_reduction): Likewise. Fix indentation of
1596 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
1598 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
1599 passed to build_vector.
1601 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
1602 Alan Hayward <alan.hayward@arm.com>
1603 David Sherwood <david.sherwood@arm.com>
1605 * tree-core.h (tree_base::u): Add an "nelts" field.
1606 (tree_vector): Use VECTOR_CST_NELTS as the length.
1607 * tree.c (tree_size): Likewise.
1608 (make_vector): Initialize VECTOR_CST_NELTS.
1609 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
1610 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
1611 TYPE_VECTOR_SUBPARTS.
1612 * expr.c (const_vector_mask_from_tree): Consistently use "units"
1613 as the number of units, setting it from VECTOR_CST_NELTS.
1614 (const_vector_from_tree): Likewise.
1615 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
1616 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
1617 (fold_negate_expr_1): Likewise.
1618 (fold_convert_const): Likewise.
1619 (const_binop): Likewise. Differentiate the number of output and
1621 (const_unop): Likewise.
1622 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
1623 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
1624 in cases that did the opposite.
1626 2017-09-14 Richard Biener <rguenther@suse.de>
1628 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
1631 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
1633 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
1635 2017-09-14 Jakub Jelinek <jakub@redhat.com>
1638 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
1639 if and where to split a bb, except for splitting before debug insn
1640 sequences followed by non-label real insn. Delete debug insns
1641 in between basic blocks.
1643 * combine.c (make_compound_operation_int): Formatting fixes.
1645 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
1646 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
1647 * config/netbsd.h (LINK_EH_SPEC): Likewise.
1648 * config/sol2.h (LINK_EH_SPEC): Likewise.
1649 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
1650 * config/s390/linux.h (LINK_SPEC): Likewise.
1651 * config/freebsd.h (LINK_EH_SPEC): Likewise.
1652 * config/openbsd.h (LINK_EH_SPEC): Likewise.
1653 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
1654 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
1655 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
1656 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
1657 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
1658 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
1660 2017-09-13 Jakub Jelinek <jakub@redhat.com>
1662 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
1664 (ENDFILE_LINUX_SPEC): Likewise.
1665 (LINK_EH_SPEC): Likewise.
1666 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
1667 (LINK_OS_LINUX_SPEC32): Likewise.
1668 (LINK_OS_LINUX_SPEC64): Likewise.
1669 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
1670 (LINK_OS_LINUX_SPEC): Likewise.
1672 2017-09-13 Martin Liska <mliska@suse.cz>
1675 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
1676 CASE_HIGH is NULL_TREE.
1678 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
1679 Alan Hayward <alan.hayward@arm.com>
1680 David Sherwood <david.sherwood@arm.com>
1682 * target.def (secondary_memory_needed): New hook.
1683 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
1684 instead of SECONDARY_MEMORY_NEEDED.
1685 (secondary_memory_needed_mode): Likewise.
1686 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
1687 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
1688 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
1689 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
1690 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
1691 * doc/tm.texi: Regenerate.
1692 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
1693 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
1694 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1695 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
1696 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
1697 * config/i386/i386.c (inline_secondary_memory_needed): Put the
1698 mode argument first and change the reg_class arguments to reg_class_t.
1699 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
1700 Make static. Update the call to inline_secondary_memory_needed.
1701 (ix86_register_move_cost): Update the call to
1702 inline_secondary_memory_needed.
1703 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1704 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
1706 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
1707 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
1709 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
1710 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
1711 * config/mips/mips.c (mips_secondary_memory_needed): Make static
1712 and match hook interface. Add comment from mips.h.
1713 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1714 * config/mmix/mmix.md (truncdfsf2): Refer to
1715 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
1717 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
1718 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
1719 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1720 (pa_secondary_memory_needed): New function.
1721 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
1722 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
1723 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1724 (pdp11_secondary_memory_needed): Make static and match hook interface.
1725 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
1726 * config/powerpcspe/powerpcspe-protos.h
1727 (rs6000_secondary_memory_needed_ptr): Delete.
1728 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
1730 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1731 (rs6000_option_override_internal): Assign to
1732 targetm.secondary_memory_needed rather than
1733 rs6000_secondary_memory_needed_ptr.
1734 (rs6000_secondary_memory_needed): Match hook interface.
1735 (rs6000_debug_secondary_memory_needed): Likewise.
1736 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
1737 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
1738 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
1739 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1740 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
1741 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
1743 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
1744 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1745 (rs6000_option_override_internal): Assign to
1746 targetm.secondary_memory_needed rather than
1747 rs6000_secondary_memory_needed_ptr.
1748 (rs6000_secondary_memory_needed): Match hook interface.
1749 (rs6000_debug_secondary_memory_needed): Likewise.
1750 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
1751 * config/s390/s390.c (s390_secondary_memory_needed): New function.
1752 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1753 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
1754 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
1755 (sparc_secondary_memory_needed): New function.
1756 * lra-constraints.c (check_and_process_move): Refer to
1757 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
1759 (curr_insn_transform): Likewise.
1760 (process_alt_operands): Use targetm.secondary_memory_needed
1761 instead of TARGET_SECONDARY_MEMORY_NEEDED.
1762 (check_secondary_memory_needed_p): Likewise.
1763 (choose_split_class): Likewise.
1764 * reload.c: Unconditionally include code that was previously
1765 conditional on SECONDARY_MEMORY_NEEDED.
1766 (push_secondary_reload): Use targetm.secondary_memory_needed
1767 instead of TARGET_SECONDARY_MEMORY_NEEDED.
1768 (push_reload): Likewise.
1769 * reload1.c: Unconditionally include code that was previously
1770 conditional on SECONDARY_MEMORY_NEEDED.
1771 (choose_reload_regs): Use targetm.secondary_memory_needed
1772 instead of TARGET_SECONDARY_MEMORY_NEEDED.
1773 (gen_reload): Likewise.
1774 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
1776 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
1777 Alan Hayward <alan.hayward@arm.com>
1778 David Sherwood <david.sherwood@arm.com>
1780 * target.def (secondary_memory_needed_mode): New hook:
1781 * targhooks.c (default_secondary_memory_needed_mode): Declare.
1782 * targhooks.h (default_secondary_memory_needed_mode): New function.
1783 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
1784 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
1785 * doc/tm.texi: Regenerate.
1786 * lra-constraints.c (check_and_process_move): Use
1787 targetm.secondary_memory_needed_mode instead of
1788 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
1789 (curr_insn_transform): Likewise.
1790 * reload.c (get_secondary_mem): Likewise.
1791 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
1792 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
1794 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
1795 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
1796 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
1797 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
1798 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
1800 * config/powerpcspe/powerpcspe-protos.h
1801 (rs6000_secondary_memory_needed_mode): Delete.
1802 * config/powerpcspe/powerpcspe.c
1803 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
1804 (rs6000_secondary_memory_needed_mode): Make static.
1805 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
1806 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
1808 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
1810 (rs6000_secondary_memory_needed_mode): Make static.
1811 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
1812 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
1813 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
1814 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
1815 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
1817 (sparc_secondary_memory_needed_mode): New function.
1818 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
1820 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
1822 * config/aarch64/constraints.md (Umq): New constraint.
1823 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
1825 (mov<mode>): Update condition.
1827 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1829 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
1830 when bitposition is the same.
1832 2017-09-13 Richard Biener <rguenther@suse.de>
1834 * dwarf2out.c (output_die_symbol): Remove.
1835 (output_die): Do not output a DIEs symbol.
1837 2017-09-13 Richard Biener <rguenther@suse.de>
1840 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
1841 default-def to avoid breaking iterator update with the weird
1842 interaction with cgraph_update_edges_for_call_stmt_node.
1844 2017-09-13 Richard Biener <rguenther@suse.de>
1846 * tree-cfg.c (verify_gimple_assign_binary): Add verification
1847 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
1848 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
1849 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
1851 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
1853 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
1854 Disable pc relative literal load irrespective of
1855 TARGET_FIX_ERR_A53_84341 for default.
1857 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
1859 * config/sparc/sparc.c (output_return): Output the source location of
1860 the insn in the delay slot, if any.
1861 (output_sibcall): Likewise.
1863 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
1866 * common.opt (-static-pie): New alias.
1867 (shared): Negate static-pie.
1868 (-no-pie): Update help text.
1870 (static-pie): New option.
1871 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
1872 -static-pie support.
1873 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
1874 (LINK_EH_SPEC): Likewise.
1875 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
1876 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
1877 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
1878 * gcc.c (LINK_COMMAND_SPEC): Likewise.
1879 (init_gcc_specs): Likewise.
1880 (init_spec): Likewise.
1881 (display_help): Update help message for -pie.
1882 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
1885 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
1887 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
1888 (movdi_aarch64): Likewise.
1889 (movti_aarch64): Likewise.
1891 2017-09-12 Simon Wright <simon@pushface.org>
1894 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
1895 calculation of the minor version, always output as 0.
1897 2017-09-12 Jakub Jelinek <jakub@redhat.com>
1900 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
1901 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
1902 on it early, rather than manual conversion late. For
1903 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
1904 instead of performing manual conversion.
1906 2017-09-12 Carl Love <cel@us.ibm.com>
1908 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
1909 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
1911 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
1912 VMULOSW): Add definitions.
1913 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
1914 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
1915 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
1916 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
1917 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
1919 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
1921 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
1923 (movti_aarch64): Likewise.
1924 (movdf_aarch64): Likewise.
1925 (movtf_aarch64): Likewise.
1926 (load_pairdi): Likewise.
1927 (store_pairdi): Likewise.
1928 (load_pairdf): Likewise.
1929 (store_pairdf): Likewise.
1930 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
1931 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
1932 (ldr_got_small_<mode>): Likewise.
1933 (ldr_got_small_28k_<mode>): Likewise.
1934 (ldr_got_tiny): Likewise.
1935 * config/aarch64/iterators.md (ldst_sz): New.
1936 (ldpstp_sz): Likewise.
1937 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
1939 (thunderx_load): Split load_8 to load_16.
1940 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
1942 (thunderx2t99_storepair_basic): Split store_8 to store_16.
1943 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
1944 (xgene1_store_pair): Split store_8 to store_16.
1945 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
1946 (falkor_st_0_st_sd): Split store_8 to store_16.
1948 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
1950 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
1951 and store1/2/3/4 to store_4/8/12/16.
1952 * config/aarch64/aarch64.md: Update for rename.
1953 * config/arm/arm.md: Likewise.: Likewise.
1954 * config/arm/arm.c: Likewise.
1955 * config/arm/thumb1.md: Likewise.
1956 * config/arm/thumb2.md: Likewise.
1957 * config/arm/vfp.md: Likewise.
1958 * config/arm/arm-generic.md: Likewise.
1959 * config/arm/arm1020e.md: Likewise.
1960 * config/arm/arm1026ejs.md: Likewise.
1961 * config/arm/arm1136jfs.md: Likewise.
1962 * config/arm/arm926ejs.md: Likewise.
1963 * config/arm/cortex-a15.md: Likewise.
1964 * config/arm/cortex-a17.md: Likewise.
1965 * config/arm/cortex-a5.md: Likewise.
1966 * config/arm/cortex-a53.md: Likewise.
1967 * config/arm/cortex-a57.md: Likewise.
1968 * config/arm/cortex-a7.md: Likewise.
1969 * config/arm/cortex-a8.md: Likewise.
1970 * config/arm/cortex-a9.md: Likewise.
1971 * config/arm/cortex-m4.md: Likewise.
1972 * config/arm/cortex-m7.md: Likewise.
1973 * config/arm/cortex-r4.md: Likewise.
1974 * config/arm/exynos-m1.md: Likewise.
1975 * config/arm/fa526.md: Likewise.
1976 * config/arm/fa606te.md: Likewise.
1977 * config/arm/fa626te.md: Likewise.
1978 * config/arm/fa726te.md: Likewise.
1979 * config/arm/fmp626.md: Likewise.
1980 * config/arm/iwmmxt.md: Likewise.
1981 * config/arm/ldmstm.md: Likewise.
1982 * config/arm/marvell-pj4.md: Likewise.
1983 * config/arm/xgene1.md: Likewise.
1984 * config/aarch64/thunderx.md: Likewise.
1985 * config/aarch64/thunderx2t99.md: Likewise.
1986 * config/aarch64/falkor.md: Likewise.
1988 2017-09-12 Martin Liska <mliska@suse.cz>
1990 * attribs.c (private_lookup_attribute): New function.
1991 * attribs.h (private_lookup_attribute): Declared here.
1992 (lookup_attribute): Called from this place.
1994 2017-09-12 Richard Biener <rguenther@suse.de>
1996 PR tree-optimization/82157
1997 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
1998 stmts with side-effects.
2000 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2001 Alan Hayward <alan.hayward@arm.com>
2002 David Sherwood <david.sherwood@arm.com>
2004 * target.def (hard_regno_nregs): New hook.
2005 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
2006 * targhooks.h (default_hard_regno_nregs): Declare.
2007 * targhooks.c (default_hard_regno_nregs): New function.
2008 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
2009 (TARGET_HARD_REGNO_NREGS): ...this hook.
2010 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
2011 (CLASS_MAX_NREGS): Likewise.
2012 * doc/tm.texi: Regenerate.
2013 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
2014 instead of HARD_REGNO_NREGS.
2015 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
2016 HARD_REGNO_NREGS in the comment.
2017 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
2018 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
2019 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
2020 Return an unsigned int.
2021 (TARGET_HARD_REGNO_NREGS): Redefine.
2022 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
2023 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
2024 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
2025 (arc_hard_regno_nregs): New function.
2026 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
2027 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
2028 (arm_hard_regno_nregs): New function.
2029 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
2030 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
2031 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
2032 (TARGET_HARD_REGNO_NREGS): Redefine.
2033 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
2034 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
2035 (HARD_REGNO_NREGS): Delete.
2036 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
2037 (cr16_hard_regno_nregs): New function.
2038 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
2039 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
2040 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
2041 (cris_hard_regno_nregs): New function.
2042 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
2043 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
2044 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
2045 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
2046 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
2047 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
2048 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
2049 (frv_hard_regno_nregs): Make static. Take and return an
2051 (frv_class_max_nregs): Remove outdated copy of documentation.
2052 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
2053 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
2054 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
2055 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
2056 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
2057 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
2058 (TARGET_HARD_REGNO_NREGS): Redefine.
2059 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
2060 (CLASS_MAX_NREGS): Update comment.
2061 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
2062 (ia64_hard_regno_nregs): New function.
2063 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
2064 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
2065 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
2066 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
2067 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
2069 (m32c_hard_regno_nregs): Likewise. Make static.
2070 (TARGET_HARD_REGNO_NREGS): Redefine.
2071 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
2072 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
2073 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
2074 (m68k_hard_regno_nregs): New function.
2075 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
2076 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
2077 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
2078 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
2079 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
2080 Take and return an unsigned int.
2081 (TARGET_HARD_REGNO_NREGS): Redefine.
2082 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
2083 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
2084 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
2085 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
2086 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
2087 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
2088 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
2089 (msp430_hard_regno_nregs): Make static. Take and return an
2091 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
2092 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
2093 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
2094 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
2095 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
2096 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
2097 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
2098 (TARGET_HARD_REGNO_NREGS): Redefine.
2099 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
2100 (PA_HARD_REGNO_NREGS): ...this.
2101 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
2102 (PA_HARD_REGNO_NREGS): ...this.
2103 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
2104 (pa_hard_regno_nregs): New function.
2105 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
2106 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
2107 (pdp11_hard_regno_nregs): New function.
2108 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
2109 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
2110 (rs6000_hard_regno_nregs_hook): New function.
2111 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
2112 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
2113 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
2114 Take and return an unsigned int. Move earlier in file.
2115 (TARGET_HARD_REGNO_NREGS): Redefine.
2116 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
2117 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
2118 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
2119 (rl78_hard_regno_nregs): Make static. Take and return an
2121 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
2122 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
2123 (rs6000_hard_regno_nregs_hook): New function.
2124 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
2125 * config/rx/rx.c (rx_hard_regno_nregs): New function.
2126 (TARGET_HARD_REGNO_NREGS): Redefine.
2127 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
2128 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
2129 instead of HARD_REGNO_NREGS.
2130 (s390_hard_regno_nregs): New function.
2131 (s390_hard_regno_mode_ok): Add comment from s390.h.
2132 (TARGET_HARD_REGNO_NREGS): Redefine.
2133 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
2134 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
2135 (sh_hard_regno_nregs): New function.
2136 (sh_pass_in_reg_p): Use it.
2137 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
2138 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
2139 (sparc_hard_regno_nregs): New function.
2140 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
2141 * config/spu/spu.c (spu_hard_regno_nregs): New function.
2142 (spu_function_arg_advance): Use it, supplying a valid register number.
2143 (TARGET_HARD_REGNO_NREGS): Redefine.
2144 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
2145 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
2146 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
2147 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
2148 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
2149 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
2150 (CLASS_MAX_NREGS): Remove copy of old documentation.
2151 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
2152 (visium_hard_regno_nregs): New function.
2153 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
2154 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
2155 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
2156 xtensa_hard_regno_nregs): New function.
2157 * system.h (HARD_REGNO_NREGS): Poison.
2159 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2161 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
2162 hard_regno_nregs instead of HARD_REGNO_NREGS.
2163 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
2164 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
2165 (c6x_expand_epilogue): Likewise.
2166 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
2167 (frv_read_iacc_argument): Likewise.
2168 * config/sh/sh.c: Include regs.h.
2169 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
2170 (regs_used): Likewise.
2171 (output_stack_adjust): Likewise.
2172 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
2173 * expmed.c: Include regs.h.
2174 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
2175 * ree.c: Include regs.h.
2176 (combine_reaching_defs): Use hard_regno_nregs instead of
2178 (add_removable_extension): Likewise.
2180 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2182 * regs.h (hard_regno_nregs): Turn into a function.
2183 (end_hard_regno): Update accordingly.
2184 * caller-save.c (setup_save_areas): Likewise.
2185 (save_call_clobbered_regs): Likewise.
2186 (replace_reg_with_saved_mem): Likewise.
2187 (insert_restore): Likewise.
2188 (insert_save): Likewise.
2189 * combine.c (can_change_dest_mode): Likewise.
2190 (move_deaths): Likewise.
2191 (distribute_notes): Likewise.
2192 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
2193 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
2194 (rs6000_split_multireg_move): Likewise.
2195 (rs6000_register_move_cost): Likewise.
2196 (rs6000_memory_move_cost): Likewise.
2197 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
2198 (rs6000_split_multireg_move): Likewise.
2199 (rs6000_register_move_cost): Likewise.
2200 (rs6000_memory_move_cost): Likewise.
2201 * cselib.c (cselib_reset_table): Likewise.
2202 (cselib_lookup_1): Likewise.
2203 * emit-rtl.c (set_mode_and_regno): Likewise.
2204 * function.c (aggregate_value_p): Likewise.
2205 * ira-color.c (setup_profitable_hard_regs): Likewise.
2206 (check_hard_reg_p): Likewise.
2207 (calculate_saved_nregs): Likewise.
2208 (assign_hard_reg): Likewise.
2209 (improve_allocation): Likewise.
2210 (calculate_spill_cost): Likewise.
2211 * ira-emit.c (modify_move_list): Likewise.
2212 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
2213 (ira_hard_reg_in_set_p): Likewise.
2214 * ira.c (setup_reg_mode_hard_regset): Likewise.
2215 (clarify_prohibited_class_mode_regs): Likewise.
2216 (check_allocation): Likewise.
2217 * lra-assigns.c (find_hard_regno_for_1): Likewise.
2218 (lra_setup_reg_renumber): Likewise.
2219 (setup_try_hard_regno_pseudos): Likewise.
2220 (spill_for): Likewise.
2221 (assign_hard_regno): Likewise.
2222 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
2223 * lra-constraints.c (in_class_p): Likewise.
2224 (lra_constraint_offset): Likewise.
2225 (simplify_operand_subreg): Likewise.
2226 (lra_constraints): Likewise.
2227 (split_reg): Likewise.
2228 (split_if_necessary): Likewise.
2229 (invariant_p): Likewise.
2230 (inherit_in_ebb): Likewise.
2231 * lra-lives.c (process_bb_lives): Likewise.
2232 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
2233 (get_hard_regs): Likewise.
2234 (do_remat): Likewise.
2235 * lra-spills.c (assign_spill_hard_regs): Likewise.
2236 * mode-switching.c (create_pre_exit): Likewise.
2237 * postreload.c (reload_combine_recognize_pattern): Likewise.
2238 * recog.c (peep2_find_free_register): Likewise.
2239 * regcprop.c (kill_value_regno): Likewise.
2240 (set_value_regno): Likewise.
2241 (copy_value): Likewise.
2242 (maybe_mode_change): Likewise.
2243 (find_oldest_value_reg): Likewise.
2244 (copyprop_hardreg_forward_1): Likewise.
2245 * regrename.c (check_new_reg_p): Likewise.
2246 (regrename_do_replace): Likewise.
2247 * reload.c (push_reload): Likewise.
2248 (combine_reloads): Likewise.
2249 (find_dummy_reload): Likewise.
2250 (operands_match_p): Likewise.
2251 (find_reloads): Likewise.
2252 (find_equiv_reg): Likewise.
2253 (reload_adjust_reg_for_mode): Likewise.
2254 * reload1.c (count_pseudo): Likewise.
2255 (count_spilled_pseudo): Likewise.
2256 (find_reg): Likewise.
2257 (clear_reload_reg_in_use): Likewise.
2258 (free_for_value_p): Likewise.
2259 (allocate_reload_reg): Likewise.
2260 (choose_reload_regs): Likewise.
2261 (reload_adjust_reg_for_temp): Likewise.
2262 (emit_reload_insns): Likewise.
2263 (delete_output_reload): Likewise.
2264 * rtlanal.c (subreg_get_info): Likewise.
2265 * sched-deps.c (sched_analyze_reg): Likewise.
2266 * sel-sched.c (init_regs_for_mode): Likewise.
2267 (mark_unavailable_hard_regs): Likewise.
2268 (choose_best_reg_1): Likewise.
2269 (verify_target_availability): Likewise.
2270 * valtrack.c (dead_debug_insert_temp): Likewise.
2271 * var-tracking.c (track_loc_p): Likewise.
2272 (emit_note_insn_var_location): Likewise.
2273 * varasm.c (make_decl_rtl): Likewise.
2274 * reginfo.c (choose_hard_reg_mode): Likewise.
2275 (init_reg_modes_target): Refer directly to
2276 this_target_regs->x_hard_regno_nregs.
2278 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2280 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
2281 instead of hard_regno_nregs.
2283 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2285 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
2286 end_hard_regno instead of hard_regno_nregs.
2287 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
2288 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
2289 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
2290 * ira-color.c (improve_allocation): Likewise.
2291 * lra-assigns.c (find_hard_regno_for_1): Likewise.
2292 * lra-lives.c (mark_regno_live): Likewise.
2293 (mark_regno_dead): Likewise.
2294 * lra-remat.c (operand_to_remat): Likewise.
2295 * lra.c (collect_non_operand_hard_regs): Likewise.
2296 * postreload.c (reload_combine_note_store): Likewise.
2297 (move2add_valid_value_p): Likewise.
2298 * reload.c (regno_clobbered_p): Likewise.
2300 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2302 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
2304 * config/v850/v850.c (v850_reorg): Likewise.
2305 * reload.c (refers_to_regno_for_reload_p): Likewise.
2306 (find_equiv_reg): Likewise.
2307 * reload1.c (reload_reg_reaches_end_p): Likewise.
2309 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2311 * caller-save.c (add_used_regs): Use REG_NREGS instead of
2313 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
2314 * config/arm/arm.c (output_move_neon): Likewise.
2315 (arm_attr_length_move_neon): Likewise.
2316 (neon_split_vcombine): Likewise.
2317 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
2318 (c6x_mark_reg_written): Likewise.
2319 (c6x_dwarf_register_span): Likewise.
2320 * config/i386/i386.c (ix86_save_reg): Likewise.
2321 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
2322 (rws_access_reg): Likewise.
2323 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
2324 * mode-switching.c (create_pre_exit): Likewise.
2325 * ree.c (combine_reaching_defs): Likewise.
2326 (add_removable_extension): Likewise.
2327 * regcprop.c (find_oldest_value_reg): Likewise.
2328 (copyprop_hardreg_forward_1): Likewise.
2329 * reload.c (reload_inner_reg_of_subreg): Likewise.
2330 (push_reload): Likewise.
2331 (combine_reloads): Likewise.
2332 (find_dummy_reload): Likewise.
2333 (reload_adjust_reg_for_mode): Likewise.
2334 * reload1.c (find_reload_regs): Likewise.
2335 (forget_old_reloads_1): Likewise.
2336 (reload_reg_free_for_value_p): Likewise.
2337 (reload_adjust_reg_for_temp): Likewise.
2338 (emit_reload_insns): Likewise.
2339 (delete_output_reload): Likewise.
2340 * sel-sched.c (choose_best_reg_1): Likewise.
2341 (choose_best_pseudo_reg): Likewise.
2343 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2344 Alan Hayward <alan.hayward@arm.com>
2345 David Sherwood <david.sherwood@arm.com>
2347 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
2348 * target.def (slow_unaligned_access): New hook.
2349 * targhooks.h (default_slow_unaligned_access): Declare.
2350 * targhooks.c (default_slow_unaligned_access): New function.
2351 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
2352 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
2353 * doc/tm.texi: Regenerate.
2354 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
2355 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
2356 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
2358 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
2359 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
2361 (rs6000_slow_unaligned_access): New function.
2362 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
2363 (expand_block_compare): Likewise.
2364 (expand_strn_compare): Likewise.
2365 (rs6000_rtx_costs): Likewise.
2366 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
2367 (riscv_slow_unaligned_access): Likewise.
2368 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
2369 (riscv_slow_unaligned_access_p): ...this and make static.
2370 (riscv_option_override): Update accordingly.
2371 (riscv_slow_unaligned_access): New function.
2372 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
2373 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
2374 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
2375 (rs6000_slow_unaligned_access): New function.
2376 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
2377 (rs6000_rtx_costs): Likewise.
2378 * config/rs6000/rs6000-string.c (expand_block_compare)
2379 (expand_strn_compare): Use targetm.slow_unaligned_access instead
2380 of SLOW_UNALIGNED_ACCESS.
2381 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
2382 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
2383 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
2384 of SLOW_UNALIGNED_ACCESS.
2385 * expmed.c (simple_mem_bitfield_p): Likewise.
2386 * expr.c (alignment_for_piecewise_move): Likewise.
2387 (emit_group_load_1): Likewise.
2388 (emit_group_store): Likewise.
2389 (copy_blkmode_from_reg): Likewise.
2390 (emit_push_insn): Likewise.
2391 (expand_assignment): Likewise.
2392 (store_field): Likewise.
2393 (expand_expr_real_1): Likewise.
2394 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
2395 * lra-constraints.c (simplify_operand_subreg): Likewise.
2396 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
2397 * gimple-ssa-store-merging.c: Likewise in block comment at start
2399 * tree-ssa-strlen.c: Include target.h.
2400 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
2401 of SLOW_UNALIGNED_ACCESS.
2402 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
2404 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2406 PR rtl-optimization/82185
2407 * expmed.c (emit_store_flag_int): Only test tem if it has been
2410 2017-09-12 Richard Biener <rguenther@suse.de>
2413 * match.pd ((FTYPE) N CMP CST): Fix typo.
2415 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
2417 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
2419 (mips_near_type_p): Add 'short_call' attribute as a synonym
2421 * doc/extend.texi (short_call): Document new function attribute.
2423 2017-09-12 Jakub Jelinek <jakub@redhat.com>
2426 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
2427 assertion check that in the condition.
2428 (get_atomic_generic_size): Likewise. Before testing if parameter
2429 has pointer type, if it has array type, call for C++
2430 default_conversion to perform array-to-pointer conversion.
2432 2017-09-12 Richard Biener <rguenther@suse.de>
2434 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
2435 for operations we cannot scalarize.
2437 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
2439 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
2440 vectors heap vectors. Clean up comments.
2441 Make visited_bbs a reference.
2442 (profitable_jump_thread_path): Make GC
2443 vectors heap vectors. Clean up comments.
2445 (convert_and_register_jump_thread_path): Make GC vectors heap
2447 (check_subpath_and_update_thread_path): Same. Clean up comments.
2448 Make visited_bbs a reference.
2449 (handle_phi): Abstract common code to to
2450 register_jump_thread_path_if_profitable.
2451 Rename VAR_BB to DEF_BB.
2453 Make GC vectors heap vectors.
2454 Make visited_bbs a reference.
2455 (handle_assignment): Same.
2456 (register_jump_thread_path_if_profitable): New.
2457 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
2459 Make GC vectors heap vectors. Clean up comments.
2460 Make visited_bbs a reference.
2461 (find_jump_threads_backwards): Make visited_bbs live in the stack.
2462 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
2465 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
2468 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
2469 words of E_DImode object are reachable by xtensa_uimm8x4 access.
2471 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
2473 Revert r251800 and r251799.
2475 2017-09-11 Martin Jambor <mjambor@suse.cz>
2478 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
2479 arguments in advance.
2480 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
2481 use it to find predecessor edges.
2482 (naive_outof_ssa): Collect vector of predecessors.
2484 2017-09-08 Jason Merrill <jason@redhat.com>
2486 PR c++/70029 - ICE with ref-qualifier and -flto
2487 * langhooks.h (struct lang_hooks_for_types): Add
2488 copy_lang_qualifiers.
2489 * attribs.c (build_type_attribute_qual_variant): Use it.
2490 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
2492 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
2493 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
2495 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
2498 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
2499 (*mulsi3_sp64): New instruction.
2500 (mulsi3): New expander.
2502 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
2504 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
2506 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2508 * sancov.c: Include memmodel.h.
2510 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
2513 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
2516 2017-09-07 Carl Love <cel@us.ibm.com>
2518 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
2519 the sldi instruction.
2521 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
2523 * sancov.c: Include tm_p.h.
2525 2017-09-07 Jakub Jelinek <jakub@redhat.com>
2528 * output.h (switch_to_other_text_partition): New declaration.
2529 * varasm.c (switch_to_other_text_partition): New function.
2530 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
2531 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
2532 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
2533 to the other text partition before emitting LCL label and switch back
2534 after emitting the word after it.
2536 2017-09-07 Richard Biener <rguenther@suse.de>
2538 * passes.def (pass_split_crit_edges): Remove instance before PRE.
2539 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
2540 critical edges here, after loop init.
2541 (pass_data_pre): Remove PROP_no_crit_edges flags.
2542 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
2543 for valueization of call args to avoid leaking VN_TOP.
2544 (visit_use): Assert we do not visit default defs.
2545 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
2546 Use error_mark_node to more easily detect leaking VN_TOP.
2547 All default-defs are varying, not VN_TOP. Mark them visited.
2548 (run_scc_vn): Make code match comment.
2550 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
2552 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
2553 OPTION_MASK_FLOAT128_KEYWORD.
2554 (POWERPC_MASKS): Likewise.
2555 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
2556 support for the -mfloat128-type option, and make -mfloat128
2557 default on PowerPC Linux systems. Define or undefine
2558 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
2559 Define __float128 to be __ieee128 if IEEE 128-bit support is
2560 enabled, or undefine it.
2561 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
2562 Delete defining __FLOAT128_TYPE__.
2563 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
2564 -mfloat128-type option and make -mfloat128 default on PowerPC
2566 (TARGET_FLOAT128_TYPE): Likewise.
2567 (-mfloat128-type): Likewise.
2568 * config/rs6000/rs6000.c (rs6000_option_override_internal):
2569 Delete the -mfloat128-type option and make -mfloat128 default on
2570 PowerPC Linux systems. Always use __ieee128 to be the keyword for
2571 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
2572 128-bit floating point is enabled. Change tests from using
2573 -mfloat128-type to -mfloat128.
2574 (rs6000_mangle_type): Use the correct mangling for the __float128
2575 type even if normal long double is restricted to 64-bits.
2576 (floatn_mode): Enable the _Float128 type by default on VSX Linux
2578 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
2579 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
2580 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
2582 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
2583 documentation for -mfloat128.
2585 2017-09-06 Olivier Hainque <hainque@adacore.com>
2587 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
2589 2017-09-06 Wish Wu <wishwu007@gmail.com>
2590 Jakub Jelinek <jakub@redhat.com>
2592 * asan.c (initialize_sanitizer_builtins): Add
2593 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
2594 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
2595 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
2596 BT_FN_VOID_UINT64_PTR variables.
2597 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
2598 (BT_FN_VOID_UINT16_UINT16): Likewise.
2599 (BT_FN_VOID_UINT32_UINT32): Likewise.
2600 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
2601 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
2602 (BT_FN_VOID_UINT64_PTR): Likewise.
2603 * common.opt (flag_sanitize_coverage): New variable.
2604 (fsanitize-coverage=trace-pc): Remove.
2605 (fsanitize-coverage=): Add.
2606 * flag-types.h (enum sanitize_coverage_code): New enum.
2607 * fold-const.c (fold_range_test): Disable non-short-circuit
2608 optimization if flag_sanitize_coverage.
2609 (fold_truth_andor): Likewise.
2610 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
2611 * opts.c (COVERAGE_SANITIZER_OPT): Define.
2612 (coverage_sanitizer_opts): New array.
2613 (get_closest_sanitizer_option): Add OPTS argument, handle also
2614 OPT_fsanitize_coverage_.
2615 (parse_sanitizer_options): Adjusted to also handle
2616 OPT_fsanitize_coverage_.
2617 (common_handle_option): Add OPT_fsanitize_coverage_.
2618 * sancov.c (instrument_comparison, instrument_switch): New function.
2619 (sancov_pass): Add trace-cmp support.
2620 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
2621 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
2622 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
2623 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
2624 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
2625 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
2626 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
2627 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
2628 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
2629 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
2631 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
2633 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
2634 error. Only quit immediately if parsing is complete.
2635 (BEGIN): Initialize fatal_err and parse_done.
2636 (begin fpu, end fpu): Check number of arguments.
2637 (begin arch, end arch): Likewise.
2638 (begin cpu, end cpu): Likewise.
2639 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
2640 (optalias): Likewise.
2642 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
2644 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
2645 * config/arm/arm-isa.h: Delete. Move definitions to ...
2646 * arm-cpus.in: ... here. Use new feature and fgroup values.
2647 * config/arm/arm.c (arm_option_override): Use lower case for feature
2649 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
2650 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
2651 * config/arm/parsecpu.awk (END): Add new command 'isa'.
2653 (print_isa_bits_for): New function.
2654 (gen_isa): New function.
2655 (gen_comm_data): Use print_isa_bits_for.
2656 (define feature): New keyword.
2657 (define fgroup): New keyword.
2658 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
2659 (arm-isa.h): Add rule to generate file.
2660 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
2661 case for feature bit names.
2663 2017-09-06 Richard Biener <rguenther@suse.de>
2665 * tree-ssa-pre.c (NECESSARY): Remove.
2666 (create_expression_by_pieces): Do not touch pass-local flags.
2667 (insert_into_preds_of_block): Likewise.
2668 (do_pre_regular_insertion): Likewise.
2669 (eliminate_insert): Likewise.
2670 (eliminate_dom_walker::before_dom_children): Likewise.
2671 (fini_eliminate): Do not look at inserted_exprs.
2672 (mark_operand_necessary): Remove.
2673 (remove_dead_inserted_code): Replace with simple work-list
2674 algorithm based on inserted_exprs and SSA uses.
2675 (pass_pre::execute): Re-order fini_eliminate and
2676 remove_dead_inserted_code.
2678 2017-09-06 Olivier Hainque <hainque@adacore.com>
2680 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
2681 for VxWorks 7. Adjust surrounding comments.
2683 2017-09-06 Richard Biener <rguenther@suse.de>
2685 * gimple-ssa-strength-reduction.c
2686 (find_candidates_dom_walker::before_dom_children): Also allow
2689 2017-09-06 Richard Biener <rguenther@suse.de>
2691 PR tree-optimization/82108
2692 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
2693 for gap in the non-permutation SLP case.
2695 2017-09-06 Martin Jambor <mjambor@suse.cz>
2697 PR tree-optimization/82078
2698 * tree-sra.c (sort_and_splice_var_accesses): Move call to
2699 add_access_to_work_queue...
2700 (build_accesses_from_assign): ...here.
2701 (propagate_all_subaccesses): Make sure racc is the group
2702 representative, if there is one.
2704 2017-09-06 Jakub Jelinek <jakub@redhat.com>
2707 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
2710 2017-09-06 Richard Biener <rguenther@suse.de>
2712 * gimple-ssa-strength-reduction.c
2713 (find_candidates_dom_walker::before_doom_children): Use a
2714 type and not a mode check.
2716 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
2719 * config/arm/predicates.md (arm_general_adddi_operand): Create new
2721 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
2723 2017-09-05 Jeff Law <law@redhat.com>
2725 PR tree-optimization/64910
2726 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
2727 cases where we have 3 or more operands.
2729 2017-09-05 Jakub Jelinek <jakub@redhat.com>
2732 * omp-low.c (lower_omp_for): Recompute tree invariant if
2733 gimple_omp_for_initial/final is ADDR_EXPR.
2736 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
2737 into gimple val before gimplification fo the COND_EXPR.
2739 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
2741 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
2742 REGION_COPY argument.
2743 (thread_through_all_blocks): Remove unused argument to
2744 duplicate_thread_path.
2746 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2747 Alan Hayward <alan.hayward@arm.com>
2748 David Sherwood <david.sherwood@arm.com>
2750 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
2751 Take a scalar_mode rather than a machine_mode.
2752 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
2753 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
2754 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
2755 (aarch64_gen_adjusted_ldpstp): Likewise.
2756 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
2758 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2759 Alan Hayward <alan.hayward@arm.com>
2760 David Sherwood <david.sherwood@arm.com>
2762 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
2763 Take a scalar_int_mode instead of a machine_mode.
2764 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
2765 (aarch64_output_scalar_simd_mov_immediate): Likewise.
2766 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
2767 (aarch64_simd_attr_length_rglist): Delete.
2768 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
2769 a scalar_int_mode instead of a machine_mode.
2770 (aarch64_add_offset): Likewise.
2771 (aarch64_internal_mov_immediate): Likewise
2772 (aarch64_add_constant_internal): Likewise.
2773 (aarch64_add_constant): Likewise.
2774 (aarch64_movw_imm): Likewise.
2775 (aarch64_rtx_arith_op_extract_p): Likewise.
2776 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
2777 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
2778 Remove assert that the mode isn't a vector.
2779 (aarch64_output_scalar_simd_mov_immediate): Likewise.
2780 (aarch64_expand_mov_immediate): Update calls after above changes.
2781 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
2782 (aarch64_and_bitmask_imm): Check for scalar integer modes.
2783 (aarch64_move_imm): Likewise.
2784 (aarch64_can_const_movi_rtx_p): Likewise.
2785 (aarch64_strip_extend): Likewise.
2786 (aarch64_extr_rtx_p): Likewise.
2787 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
2788 a CONST_INT when the mode parameter is VOIDmode.
2789 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
2791 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2793 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
2794 * stor-layout.c (bitwise_mode_for_mode): Likewise.
2795 (bitwise_type_for_mode): Update accordingly.
2797 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2799 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
2800 * stor-layout.c (mode_for_size_tree): Likewise.
2801 (mode_for_array): Update accordingly.
2802 (layout_decl): Likewise.
2803 (compute_record_mode): Likewise. Only set the mode once.
2805 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2807 * target.def (get_mask_mode): Change return type to opt_mode.
2809 * doc/tm.texi: Regenerate.
2810 * targhooks.h (default_get_mask_mode): Return an opt_mode.
2811 * targhooks.c (default_get_mask_mode): Likewise.
2812 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
2813 * optabs-query.c (can_vec_mask_load_store_p): Update use of
2814 targetm.get_mask_mode.
2815 * tree.c (build_truth_vector_type): Likewise.
2817 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2819 * machmode.h (mode_for_vector): Return an opt_mode.
2820 * stor-layout.c (mode_for_vector): Likewise.
2821 (mode_for_int_vector): Update accordingly.
2822 (layout_type): Likewise.
2823 * config/i386/i386.c (emit_memmov): Likewise.
2824 (ix86_expand_set_or_movmem): Likewise.
2825 (ix86_expand_vector_init): Likewise.
2826 (ix86_get_mask_mode): Likewise.
2827 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
2829 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
2830 * expmed.c (extract_bit_field_1): Likewise.
2831 * expr.c (expand_expr_real_2): Likewise.
2832 * optabs-query.c (can_vec_perm_p): Likewise.
2833 (can_vec_mask_load_store_p): Likewise.
2834 * optabs.c (expand_vec_perm): Likewise.
2835 * targhooks.c (default_get_mask_mode): Likewise.
2836 * tree-vect-stmts.c (vectorizable_store): Likewise.
2837 (vectorizable_load): Likewise.
2838 (get_vectype_for_scalar_type_and_size): Likewise.
2840 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2842 * machmode.h (mode_for_int_vector): New function.
2843 * stor-layout.c (mode_for_int_vector): Likewise.
2844 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
2845 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
2846 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
2847 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
2848 (s390_expand_vcond): Likewise.
2850 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2852 * machmode.h (opt_machine_mode): New type.
2853 (opt_mode<T>): Allow construction from anything that can be
2855 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
2856 (mode_for_size): Return an opt_machine_mode.
2857 * stor-layout.c (mode_for_size): Likewise.
2858 (mode_for_size_tree): Update call accordingly.
2859 (bitwise_mode_for_mode): Likewise.
2860 (make_fract_type): Likewise.
2861 (make_accum_type): Likewise.
2862 * caller-save.c (replace_reg_with_saved_mem): Update call
2864 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
2865 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
2866 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
2867 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
2868 * expmed.c (extract_bit_field_1): Likewise.
2869 * reload.c (get_secondary_mem): Likewise.
2870 * varasm.c (assemble_integer): Likewise.
2871 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
2874 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2876 * machmode.h (decimal_float_mode_for_size): New function.
2877 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
2878 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
2879 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
2880 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
2881 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
2882 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
2884 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2886 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
2887 (get_builtin_sync_mode): Likewise.
2888 (expand_ifn_atomic_compare_exchange): Likewise.
2889 (expand_builtin_atomic_clear): Likewise.
2890 (expand_builtin_atomic_test_and_set): Likewise.
2891 (fold_builtin_atomic_always_lock_free): Likewise.
2892 * calls.c (compute_argument_addresses): Likewise.
2893 (emit_library_call_value_1): Likewise.
2894 (store_one_arg): Likewise.
2895 * combine.c (combine_instructions): Likewise.
2896 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
2897 * config/arm/arm.c (arm_function_value): Likewise.
2898 (aapcs_allocate_return_reg): Likewise.
2899 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
2900 * config/i386/i386.c (construct_container): Likewise.
2901 (ix86_gimplify_va_arg): Likewise.
2902 (ix86_expand_sse_cmp): Likewise.
2903 (emit_memmov): Likewise.
2904 (emit_memset): Likewise.
2905 (expand_small_movmem_or_setmem): Likewise.
2906 (ix86_expand_pextr): Likewise.
2907 (ix86_expand_pinsr): Likewise.
2908 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
2909 * config/microblaze/microblaze.c (microblaze_block_move_straight):
2911 * config/mips/mips.c (mips_function_value_1) Likewise.
2912 (mips_block_move_straight): Likewise.
2913 (mips_expand_ins_as_unaligned_store): Likewise.
2914 * config/powerpcspe/powerpcspe.c
2915 (rs6000_darwin64_record_arg_advance_flush): Likewise.
2916 (rs6000_darwin64_record_arg_flush): Likewise.
2917 * config/rs6000/rs6000.c
2918 (rs6000_darwin64_record_arg_advance_flush): Likewise.
2919 (rs6000_darwin64_record_arg_flush): Likewise.
2920 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
2921 (sparc_function_value_1): Likewise.
2922 * config/spu/spu.c (adjust_operand): Likewise.
2923 (spu_emit_branch_or_set): Likewise.
2924 (arith_immediate_p): Likewise.
2925 * emit-rtl.c (gen_lowpart_common): Likewise.
2926 * expr.c (expand_expr_real_1): Likewise.
2927 * function.c (assign_parm_setup_block): Likewise.
2928 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
2929 * reload1.c (alter_reg): Likewise.
2930 * stor-layout.c (mode_for_vector): Likewise.
2931 (layout_type): Likewise.
2933 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2935 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
2936 (spu_convert_move): Likewise.
2937 * lower-subreg.c (resolve_simple_move): Likewise.
2939 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2942 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
2943 define_insn to a define_expand.
2944 (altivec_vsum2sws_direct): New define_insn.
2945 (altivec_vsumsws): Convert from a define_insn to a define_expand.
2947 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
2949 * config/arm/arm.c (arm_option_params_internal): Improve setting of
2952 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
2957 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
2958 realign stack if stack alignment needed is less than incoming
2961 2017-09-05 Marek Polacek <polacek@redhat.com>
2964 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
2967 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
2969 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
2971 2017-09-05 Richard Biener <rguenther@suse.de>
2973 PR tree-optimization/82084
2974 * fold-const.c (can_native_encode_string_p): Handle wide characters.
2976 2017-09-05 Richard Biener <rguenther@suse.de>
2978 PR tree-optimization/82102
2979 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
2981 2017-09-05 Martin Liska <mliska@suse.cz>
2983 PR tree-optimization/82032
2984 * tree-cfg.c (generate_range_test): New function.
2985 * tree-cfg.h (generate_range_test): Declared here.
2986 * tree-cfgcleanup.c (convert_single_case_switch): New function.
2987 (cleanup_control_expr_graph): Use it.
2988 * tree-switch-conversion.c (try_switch_expansion): Remove
2990 (emit_case_nodes): Use generate_range_test.
2992 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
2995 * config/i386/i386.md (*<btsc><mode>_mask): Add
2996 TARGET_USE_BT to insn constraint.
2997 (*btr<mode>_mask): Ditto.
2999 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
3001 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
3002 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
3004 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
3007 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
3008 TARGET_NEON and TARGET_IWMMXT.
3009 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
3010 TARGET_NEON and TARGET_IWMMXT.
3011 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
3013 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
3015 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
3016 (ix86_rewrite_tls_address): Ditto.
3017 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
3018 (ix86_rewrite_tls_address_1): Ditto.
3019 (ix86_rewrite_tls_address): Ditto.
3020 * config/i386/predicates.md (tls_address_pattern): New predicate.
3021 * config/i386/i386.md (TLS address splitter): New splitter.
3023 2017-09-04 Richard Biener <rguenther@suse.de>
3025 PR tree-optimization/82084
3026 * fold-const.h (can_native_encode_string_p): Declare.
3027 * fold-const.c (can_native_encode_string_p): Factor out from ...
3028 (native_encode_string): ... here.
3029 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
3030 vectorizing stores from constants we later cannot handle.
3032 2017-09-04 Marek Polacek <polacek@redhat.com>
3035 * doc/invoke.texi: Update -Wtautological-compare documentation.
3037 2017-09-04 Jeff Law <law@redhat.com>
3039 PR tree-optimization/64910
3040 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
3041 swap the first and last operand if the last is a constant.
3043 2017-09-04 Marek Polacek <polacek@redhat.com>
3046 * convert.c (do_narrow): When sanitizing signed integer overflows,
3047 bail out for signed types.
3048 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
3050 2017-09-04 Richard Biener <rguenther@suse.de>
3052 PR tree-optimization/82060
3053 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
3054 Move devirtualization after stmt folding and before EH/AB/noreturn
3055 cleanup to get the stmt refs canonicalized. Use a bool instead
3056 of gimple_modified_p since that doesn't work for NOPs. Schedule
3057 NOPs generated by folding for removal.
3059 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
3060 Alan Hayward <alan.hayward@arm.com>
3061 David Sherwood <david.sherwood@arm.com>
3063 * coretypes.h (pad_direction): New enum.
3064 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
3065 (FUNCTION_ARG_PADDING): Likewise.
3066 * target.def (function_arg_padding): New hook.
3067 * targhooks.h (default_function_arg_padding): Declare.
3068 * targhooks.c (default_function_arg_padding): New function.
3069 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
3070 (TARGET_FUNCTION_ARG_PADDING): ...this.
3071 * doc/tm.texi: Regenerate.
3072 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
3073 instead of direction.
3074 (compute_argument_addresses): Likewise.
3075 (load_register_parameters): Likewise.
3076 (emit_library_call_value_1): Likewise.
3077 (store_one_arg): Use targetm.calls.function_arg_padding instead
3078 of FUNCTION_ARG_PADDING.
3079 (must_pass_in_stack_var_size_or_pad): Likewise.
3080 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
3081 (emit_group_store): Likewise.
3082 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
3083 instead of FUNCTION_ARG_PADDING.
3084 (emit_push_insn): Likewise, and propagate enum change throughout
3086 * function.h (direction): Delete.
3087 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
3089 * function.c (assign_parm_find_stack_rtl): Likewise.
3090 (assign_parm_setup_block_p): Likewise.
3091 (assign_parm_setup_block): Likewise.
3092 (gimplify_parameters): Likewise.
3093 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
3094 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
3096 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
3097 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
3098 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
3099 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
3100 (aarch64_function_arg_padding): ...this new function.
3101 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
3102 (TARGET_FUNCTION_ARG_PADDING): Redefine.
3103 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
3104 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
3105 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
3106 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3107 (arm_pad_arg_upward): Replace with...
3108 (arm_function_arg_padding): ...this new function.
3109 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
3111 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
3112 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
3113 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3114 (ia64_hpux_function_arg_padding): Replace with...
3115 (ia64_function_arg_padding): ...this new function. Use pad_direction
3116 instead of direction. Check for TARGET_HPUX.
3117 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
3118 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3119 (iq2000_function_arg_padding): New function.
3120 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
3121 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
3122 (mips_function_arg_padding): ...this new function.
3123 (mips_pad_reg_upward): Update accordingly.
3124 (TARGET_FUNCTION_ARG_PADDING): Redefine.
3125 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
3126 targetm.calls.function_arg_padding.
3127 (FUNCTION_ARG_PADDING): Delete.
3128 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
3129 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
3130 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
3131 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
3132 (nios2_block_reg_padding): Return pad_direction instead of direction.
3133 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
3134 instead of direction.
3135 (nios2_function_arg_padding): Likewise. Make static.
3136 (TARGET_FUNCTION_ARG_PADDING): Redefine.
3137 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
3138 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
3139 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
3140 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3141 (pa_function_arg_padding): Make static. Return pad_direction instead
3143 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
3144 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
3145 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
3146 instead of direction. Use targetm.calls.function_arg_padding.
3147 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
3148 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
3149 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
3150 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
3151 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
3153 (function_arg_padding): Rename to...
3154 (rs6000_function_arg_padding): ...this. Make static. Return
3155 pad_direction instead of direction.
3156 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
3157 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
3158 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
3159 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
3160 instead of direction. Use targetm.calls.function_arg_padding.
3161 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
3162 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
3163 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
3164 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
3165 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3166 (function_arg_padding): Rename to...
3167 (rs6000_function_arg_padding): ...this. Make static. Return
3168 pad_direction instead of direction.
3169 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
3170 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
3171 * config/s390/s390.c (s390_function_arg_padding): New function.
3172 (TARGET_FUNCTION_ARG_PADDING): Redefine.
3173 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
3174 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
3175 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
3176 (function_arg_padding): Rename to...
3177 (sparc_function_arg_padding): ...this. Make static. Return
3178 pad_direction instead of direction.
3179 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
3180 * config/spu/spu.c (spu_function_arg_padding): New function.
3181 (TARGET_FUNCTION_ARG_PADDING): Redefine.
3182 * system.h (FUNCTION_ARG_PADDING): Poison.
3184 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
3185 Alan Hayward <alan.hayward@arm.com>
3186 David Sherwood <david.sherwood@arm.com>
3188 * target.def (modes_tieable_p): New hook.
3189 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
3190 (TARGET_MODES_TIEABLE_P): ...this.
3191 * doc/tm.texi.in: Regenerate.
3192 * hooks.h (hook_bool_mode_mode_true): Declare.
3193 * hooks.c (hook_bool_mode_mode_true): New function.
3194 * combine.c (subst): Use targetm.modes_tieable_p instead of
3196 * dse.c (find_shift_sequence): Likewise.
3197 * expmed.c (extract_low_bits): Likewise.
3198 * lower-subreg.c: Include target.h.
3199 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
3201 * rtlanal.c (rtx_cost): Likewise.
3202 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
3203 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
3204 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
3205 (TARGET_MODES_TIEABLE_P): Redefine.
3206 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
3207 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
3208 (TARGET_MODES_TIEABLE_P): Redefine.
3209 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
3210 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
3211 (arc_modes_tieable_p): New function.
3212 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
3213 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
3214 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
3215 (arm_modes_tieable_p): Make static.
3216 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
3217 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
3218 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
3219 (TARGET_MODES_TIEABLE_P): Redefine.
3220 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
3221 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
3222 (TARGET_MODES_TIEABLE_P): Redefine.
3223 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
3224 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
3225 (cr16_modes_tieable_p): New function.
3226 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
3227 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
3228 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
3229 (TRULY_NOOP_TRUNCATION): Update comment.
3230 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
3231 (TRULY_NOOP_TRUNCATION): Update comment.
3232 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
3233 (frv_modes_tieable_p): New function.
3234 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
3235 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
3236 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
3237 (TARGET_MODES_TIEABLE_P): Redefine.
3238 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
3239 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
3240 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
3241 (TARGET_MODES_TIEABLE_P): Redefine.
3242 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
3243 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
3244 (ia64_modes_tieable_p): New function.
3245 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
3246 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
3247 (iq2000_modes_tieable_p): New function.
3248 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
3249 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
3250 (lm32_modes_tieable_p): New function.
3251 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
3252 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
3253 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
3254 (TARGET_MODES_TIEABLE_P): Redefine.
3255 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
3256 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
3257 (m32r_modes_tieable_p): New function.
3258 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
3259 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
3260 (m68k_modes_tieable_p): New function.
3261 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
3262 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
3263 (mcore_modes_tieable_p): New function.
3264 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
3265 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
3267 (TARGET_MODES_TIEABLE_P): Redefine.
3268 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
3269 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
3270 * config/mips/mips.c (mips_modes_tieable_p): Make static.
3271 (TARGET_MODES_TIEABLE_P): Redefine.
3272 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
3273 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
3274 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
3275 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
3276 (mn10300_modes_tieable_p): ...this and make static.
3277 (TARGET_MODES_TIEABLE_P): Redefine.
3278 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
3279 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
3280 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
3281 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
3282 (msp430_modes_tieable_p): Make static.
3283 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
3284 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
3285 (TARGET_MODES_TIEABLE_P): Redefine.
3286 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
3287 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
3288 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
3289 (TARGET_MODES_TIEABLE_P): Redefine.
3290 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
3291 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
3292 * config/pa/pa.c (pa_modes_tieable_p): Make static.
3293 (TARGET_MODES_TIEABLE_P): Redefine.
3294 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
3295 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
3296 (pdp11_modes_tieable_p): New function.
3297 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
3298 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
3299 (rs6000_modes_tieable_p): New function.
3300 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
3301 * config/powerpcspe/powerpcspe.md: Update comment.
3302 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
3303 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
3304 (TARGET_MODES_TIEABLE_P): Redefine.
3305 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
3306 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
3307 (rl78_modes_tieable_p): New function.
3308 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
3309 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
3310 (rs6000_modes_tieable_p): New function.
3311 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
3312 * config/rs6000/rs6000.md: Update comment.
3313 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
3314 * config/rx/rx.c (rx_modes_tieable_p): New function.
3315 (TARGET_MODES_TIEABLE_P): Redefine.
3316 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
3317 * config/s390/s390.c (s390_modes_tieable_p): New function.
3318 (TARGET_MODES_TIEABLE_P): Redefine.
3319 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
3320 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
3321 (sh_modes_tieable_p): New function.
3322 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
3323 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
3324 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
3325 (sparc_modes_tieable_p): Make static.
3326 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
3327 * config/spu/spu.c (spu_modes_tieable_p): New function.
3328 (TARGET_MODES_TIEABLE_P): Redefine.
3329 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
3330 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
3331 (TARGET_MODES_TIEABLE_P): Redefine.
3332 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
3333 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
3334 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
3335 * config/v850/v850.c (v850_modes_tieable_p): New function.
3336 (TARGET_MODES_TIEABLE_P): Redefine.
3337 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
3338 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
3339 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
3340 (visium_modes_tieable_p): New function.
3341 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
3342 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
3343 (xtensa_modes_tieable_p): New function.
3344 * system.h (MODES_TIEABLE_P): Poison.
3346 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
3347 Alan Hayward <alan.hayward@arm.com>
3348 David Sherwood <david.sherwood@arm.com>
3350 * target.def (hard_regno_mode_ok): New hook.
3351 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
3352 (TARGET_HARD_REGNO_MODE_OK): ...this.
3353 * doc/tm.texi.in: Regenerate.
3354 * hooks.h (hook_bool_uint_mode_true): Declare.
3355 * hooks.c (hook_bool_uint_mode_true): New function.
3356 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
3358 * genpreds.c (write_insn_preds_c): Add an include of target.h.
3359 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
3360 instead of HARD_REGNO_MODE_OK.
3361 * caller-save.c: Include target.h.
3362 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
3364 * combine.c (can_combine_p): Likewise.
3365 (combinable_i3pat): Likewise.
3366 (can_change_dest_mode): Likewise.
3367 * expr.c (init_expr_target): Likewise.
3368 (convert_move): Likewise.
3369 (convert_modes): Likewise.
3370 * ira.c (setup_prohibited_class_mode_regs): Likewise.
3371 (setup_prohibited_mode_move_regs): Likewise.
3372 * ira.h (target_ira): Likewise.
3373 * lra-assigns.c (find_hard_regno_for_1): Likewise.
3374 * lra-constraints.c (process_alt_operands): Likewise.
3375 (split_reg): Likewise.
3376 * recog.c (peep2_find_free_register): Likewise.
3377 * ree.c (combine_reaching_defs): Likewise.
3378 * regcprop.c (maybe_mode_change): Likewise.
3379 * reginfo.c (init_reg_sets_1): Likewise.
3380 (choose_hard_reg_mode): Likewise.
3381 (simplifiable_subregs): Likewise.
3382 * regrename.c (check_new_reg_p): Likewise.
3383 * reload.c (find_valid_class): Likewise.
3384 (find_valid_class_1): Likewise.
3385 (reload_inner_reg_of_subreg): Likewise.
3386 (push_reload): Likewise.
3387 (combine_reloads): Likewise.
3388 (find_dummy_reload): Likewise.
3389 (find_reloads): Likewise.
3390 * reload1.c (find_reg): Likewise.
3391 (set_reload_reg): Likewise.
3392 (allocate_reload_reg): Likewise.
3393 (choose_reload_regs): Likewise.
3394 (reload_adjust_reg_for_temp): Likewise.
3395 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
3396 (simplify_subreg_regno): Likewise.
3397 * sel-sched.c (init_regs_for_mode): Likewise.
3398 * varasm.c (make_decl_rtl): Likewise.
3399 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
3400 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
3402 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
3403 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
3404 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3405 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
3406 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
3407 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3408 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
3409 (arc_mode_class): Delete.
3410 (HARD_REGNO_MODE_OK): Delete.
3411 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3412 (arc_hard_regno_mode_ok): Rename old array to...
3413 (arc_hard_regno_mode_ok_modes): ...this.
3414 (arc_conditional_register_usage): Update accordingly.
3415 (arc_mode_class): Make static.
3416 (arc_hard_regno_mode_ok): New function.
3417 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
3418 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
3419 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3420 (arm_hard_regno_mode_ok): Make static.
3421 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
3423 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
3424 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
3425 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
3427 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3428 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
3429 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
3430 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
3431 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
3432 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3433 * config/bfin/predicates.md (valid_reg_operand): Use
3434 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
3435 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
3436 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
3437 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3438 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
3439 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
3440 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3441 (cr16_hard_regno_mode_ok): Make static and return a bool.
3442 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
3443 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3444 (cris_hard_regno_mode_ok): New function.
3445 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
3446 (epiphany_mode_class): Delete.
3447 (HARD_REGNO_MODE_OK): Delete.
3448 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
3449 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3450 (hard_regno_mode_ok): Rename to...
3451 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
3452 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
3453 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
3455 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
3456 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
3457 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3458 (frv_hard_regno_mode_ok): Make static and return a bool.
3459 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
3461 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
3462 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
3463 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
3464 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
3466 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3467 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
3468 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
3469 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
3471 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3472 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
3473 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3474 (ia64_hard_regno_mode_ok): New function.
3475 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
3476 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3477 (iq2000_hard_regno_mode_ok): New function.
3478 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
3479 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3480 (lm32_hard_regno_mode_ok): New function.
3481 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
3482 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
3483 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
3484 instead of HARD_REGNO_MODE_OK.
3485 (m32c_hard_regno_ok): Rename to...
3486 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
3487 (m32c_cannot_change_mode_class): Update accordingly.
3488 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3489 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
3490 (m32r_mode_class): Delete.
3491 (HARD_REGNO_MODE_OK): Delete.
3492 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3493 (m32r_hard_regno_mode_ok): Rename to...
3494 (m32r_hard_regno_modes): ...this.
3495 (m32r_mode_class): Make static.
3496 (m32r_hard_regno_mode_ok): New function.
3497 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
3498 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
3499 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3500 (m68k_hard_regno_mode_ok): Make static.
3501 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
3502 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3503 (mcore_hard_regno_mode_ok): New function.
3504 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
3505 (HARD_REGNO_MODE_OK): Delete.
3506 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
3508 (microblaze_hard_regno_mode_ok_p): ...this and make static.
3509 (microblaze_hard_regno_mode_ok): New function.
3510 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3511 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
3512 (mips_hard_regno_mode_ok): Delete.
3513 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
3514 (mips_hard_regno_mode_ok_p): ...this and make static.
3515 (mips_hard_regno_mode_ok_p): Rename to...
3516 (mips_hard_regno_mode_ok_uncached): ...this.
3517 (mips_hard_regno_mode_ok): New function.
3518 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
3519 of HARD_REGNO_MODE_OK.
3520 (mips_option_override): Update after above name changes.
3521 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3522 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
3523 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
3524 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
3525 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
3526 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3527 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
3528 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
3529 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
3530 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3531 (msp430_hard_regno_mode_ok): Make static and return a bool.
3532 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
3533 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
3534 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
3536 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3537 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
3538 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
3539 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
3540 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
3541 (PA_HARD_REGNO_MODE_OK): ...this
3542 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
3543 (PA_HARD_REGNO_MODE_OK): ...this.
3544 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3545 (pa_hard_regno_mode_ok): New function.
3546 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
3547 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3548 (pdp11_hard_regno_mode_ok): New function.
3549 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
3550 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
3552 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
3554 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3555 (rs6000_hard_regno_mode_ok): Rename to...
3556 (rs6000_hard_regno_mode_ok_uncached): ...this.
3557 (rs6000_init_hard_regno_mode_ok): Update accordingly.
3558 (rs6000_hard_regno_mode_ok): New function.
3559 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
3560 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
3561 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
3562 (riscv_hard_regno_mode_ok): ...this and make static.
3563 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3564 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
3565 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
3566 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3567 (rl78_hard_regno_mode_ok): Make static and return bool.
3568 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
3569 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
3571 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
3572 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3573 (rs6000_hard_regno_mode_ok): Rename to...
3574 (rs6000_hard_regno_mode_ok_uncached): ...this.
3575 (rs6000_init_hard_regno_mode_ok): Update accordingly.
3576 (rs6000_hard_regno_mode_ok): New function.
3577 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
3578 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
3579 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3580 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
3581 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
3582 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
3583 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3584 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
3585 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
3586 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3587 (sh_hard_regno_mode_ok): Make static.
3588 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
3589 instead of HARD_REGNO_MODE_OK.
3590 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
3591 (sparc_mode_class): Delete.
3592 (HARD_REGNO_MODE_OK): Delete.
3593 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3594 (hard_regno_mode_classes): Make static.
3595 (sparc_mode_class): Likewise.
3596 (sparc_hard_regno_mode_ok): New function.
3597 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
3598 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
3599 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
3601 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3602 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
3603 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
3604 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
3605 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
3606 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3607 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
3608 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
3609 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
3610 (visium_hard_regno_mode_ok): New function.
3611 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
3612 instead of HARD_REGNO_MODE_OK.
3613 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
3614 (HARD_REGNO_MODE_OK): Delete.
3615 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
3616 (xtensa_hard_regno_mode_ok_p): ...this and make static.
3617 (xtensa_option_override): Update accordingly.
3618 (TARGET_HARD_REGNO_MODE_OK): Redefine.
3619 (xtensa_hard_regno_mode_ok): New function.
3620 * system.h (HARD_REGNO_MODE_OK): Poison.
3622 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
3623 Alan Hayward <alan.hayward@arm.com>
3624 David Sherwood <david.sherwood@arm.com>
3626 * target.def (hard_regno_call_part_clobbered): New hook.
3627 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
3628 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
3629 * doc/tm.texi: Regenerate.
3630 * hooks.h (hook_bool_uint_mode_false): Declare.
3631 * hooks.c (hook_bool_uint_mode_false): New function.
3632 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3633 * cselib.c (cselib_process_insn): Use
3634 targetm.hard_regno_call_part_clobbered instead of
3635 HARD_REGNO_CALL_PART_CLOBBERED.
3636 * ira-conflicts.c (ira_build_conflicts): Likewise.
3637 * ira-costs.c (ira_tune_allocno_costs): Likewise.
3638 * lra-constraints.c (need_for_call_save_p): Likewise.
3639 * lra-lives.c: Include target.h.
3640 (check_pseudos_live_through_calls): Use
3641 targetm.hard_regno_call_part_clobbered instead of
3642 HARD_REGNO_CALL_PART_CLOBBERED.
3643 * regcprop.c: Include target.h.
3644 (copyprop_hardreg_forward_1): Use
3645 targetm.hard_regno_call_part_clobbered instead of
3646 HARD_REGNO_CALL_PART_CLOBBERED.
3647 * reginfo.c (choose_hard_reg_mode): Likewise.
3648 * regrename.c (check_new_reg_p): Likewise.
3649 * reload.c (find_equiv_reg): Likewise.
3650 * reload1.c (emit_reload_insns): Likewise.
3651 * sched-deps.c (deps_analyze_insn): Likewise.
3652 * sel-sched.c (init_regs_for_mode): Likewise.
3653 (mark_unavailable_hard_regs): Likewise.
3654 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
3655 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3656 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
3658 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3659 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3660 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
3662 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
3664 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3665 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3666 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
3668 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3669 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3670 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
3672 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3673 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
3675 * config/powerpcspe/powerpcspe.c
3676 (rs6000_hard_regno_call_part_clobbered): New function.
3677 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3678 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3679 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
3681 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3682 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3683 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
3685 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
3686 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
3687 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
3689 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
3690 Alan Hayward <alan.hayward@arm.com>
3691 David Sherwood <david.sherwood@arm.com>
3693 * rtl.h (subreg_memory_offset): Declare.
3694 * emit-rtl.c (subreg_memory_offset): New function.
3695 * expmed.c (store_bit_field_1): Use it.
3696 * expr.c (undefined_operand_subword_p): Likewise.
3697 * simplify-rtx.c (simplify_subreg): Likewise.
3699 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
3701 PR rtl-optimization/57448
3704 * optabs.c (expand_atomic_load): Place compiler memory barriers if
3705 using atomic_load pattern.
3706 (expand_atomic_store): Likewise.
3708 2017-09-04 Jakub Jelinek <jakub@redhat.com>
3711 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
3712 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
3713 handling. Use replace_call_with_value with NULL instead of
3714 gsi_replace, unlink_stmt_vdef and release_defs.
3716 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
3719 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
3721 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
3724 * rtl.h (emit_library_call_value_1): Declare.
3725 (emit_library_call): Replace declaration with a series of overloads.
3726 Remove the parameter count argument.
3727 (emit_library_call_value): Likewise.
3728 * calls.c (emit_library_call_value_1): Make global. Replace varargs
3729 with an "rtx_mode_t *".
3730 (emit_library_call_value): Delete.
3731 (emit_library_call): Likewise.
3732 * asan.c (asan_emit_stack_protection): Update calls accordingly.
3733 (asan_emit_allocas_unpoison): Likewise.
3734 * builtins.c (expand_builtin_powi): Likewise.
3735 (expand_asan_emit_allocas_unpoison): Likewise.
3736 * cfgexpand.c (expand_main_function): Likewise.
3737 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
3738 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
3739 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
3740 * config/arm/arm.c (arm_trampoline_init): Likewise.
3741 (arm_call_tls_get_addr): Likewise.
3742 (arm_expand_divmod_libfunc): Likewise.
3743 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
3744 (smulsi3_highpart): Likewise.
3745 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
3746 (c6x_expand_compare): Likewise.
3747 (c6x_expand_movmem): Likewise.
3748 * config/frv/frv.c (frv_trampoline_init): Likewise.
3749 * config/i386/i386.c (ix86_trampoline_init): Likewise.
3750 (ix86_expand_divmod_libfunc): Likewise.
3751 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
3752 (ia64_expand_compare): Likewise.
3753 (ia64_profile_hook): Likewise.
3754 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
3755 (nonlocal_goto): Likewise.
3756 (restore_stack_nonlocal): Likewise.
3757 * config/m32r/m32r.c (block_move_call): Likewise.
3758 (m32r_trampoline_init): Likewise.
3759 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
3760 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
3761 (m68k_call_m68k_read_tp): Likewise.
3762 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
3763 (microblaze_expand_divide): Likewise.
3764 * config/mips/mips.h (mips_args): Likewise.
3765 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
3766 (MIPS_ICACHE_SYNC): Likewise.
3767 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
3768 (nios2_trampoline_init): Likewise.
3769 * config/pa/pa.c (hppa_tls_call): Likewise.
3770 (pa_trampoline_init): Likewise.
3771 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
3772 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
3773 (expand_strn_compare): Likewise.
3774 (rs6000_generate_compare): Likewise.
3775 (rs6000_expand_float128_convert): Likewise.
3776 (output_profile_hook): Likewise.
3777 (rs6000_trampoline_init): Likewise.
3778 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
3779 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
3780 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
3781 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
3782 (rs6000_generate_compare): Likewise.
3783 (rs6000_expand_float128_convert): Likewise.
3784 (output_profile_hook): Likewise.
3785 (rs6000_trampoline_init): Likewise.
3786 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
3787 * config/sh/sh.c (sh_trampoline_init): Likewise.
3788 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
3789 (sparc_emit_float_lib_cmp): Likewise.
3790 (sparc32_initialize_trampoline): Likewise.
3791 (sparc64_initialize_trampoline): Likewise.
3792 (sparc_profile_hook): Likewise.
3793 * config/spu/spu.c (ea_load_store): Likewise.
3794 * config/spu/spu.md (floatunssidf2): Likewise.
3795 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
3796 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
3797 * config/visium/visium.c (expand_block_move_4): Likewise.
3798 (expand_block_move_2): Likewise.
3799 (expand_block_move_1): Likewise.
3800 (expand_block_set_4): Likewise.
3801 (expand_block_set_2): Likewise.
3802 (expand_block_set_1): Likewise.
3803 (visium_trampoline_init): Likewise.
3804 (visium_profile_hook): Likewise.
3805 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
3806 (xtensa_setup_frame_addresses): Likewise.
3807 (xtensa_trampoline_init): Likewise.
3808 * except.c (sjlj_emit_function_enter): Likewise.
3809 (sjlj_emit_function_exit): Likewise.
3810 * explow.c (allocate_dynamic_stack_space): Likewise.
3811 (probe_stack_range): Likewise.
3812 * expr.c (convert_mode_scalar): Likewise.
3813 * optabs.c (expand_binop): Likewise.
3814 (expand_twoval_binop_libfunc): Likewise.
3815 (expand_unop): Likewise.
3816 (prepare_cmp_insn): Likewise.
3817 (prepare_float_lib_cmp): Likewise.
3818 (expand_float): Likewise.
3819 (expand_fix): Likewise.
3820 (expand_fixed_convert): Likewise.
3821 (maybe_emit_sync_lock_test_and_set): Likewise.
3822 (expand_atomic_compare_and_swap): Likewise.
3823 (expand_mem_thread_fence): Likewise.
3824 (expand_atomic_fetch_op): Likewise.
3826 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
3828 * doc/generic.texi (OpenACC): Adjust URL.
3829 * doc/invoke.texi (C Dialect Options): Ditto.
3831 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
3833 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
3834 predicate for operand 1. Add (m,<S>) constraint.
3835 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
3836 Prevent memory operand 1 with register operand 2.
3838 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
3840 PR rtl-optimization/82024
3841 * combine.c (try_combine): If the combination result is a PARALLEL,
3842 and we only need to retain the SET in there that would be placed
3843 at I2, check that we can place that at I3 instead, before doing so.
3845 2017-09-01 Jakub Jelinek <jakub@redhat.com>
3848 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
3850 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
3851 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
3854 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
3855 Jeff Law <law@redhat.com>
3857 * varasm.c (bss_initializer_p): Do not put constants into .bss
3858 (categorize_decl_for_section): Handle bss_initializer_p returning
3859 false when DECL_INITIAL is NULL.
3861 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3864 * config/s390/s390.c (s390_can_inline_p): New function.
3866 2017-09-01 Jeff Law <law@redhat.com>
3868 PR tree-optimization/82052
3869 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
3870 Always initialize the returned slot after a hash table miss
3871 when INSERT is true.
3873 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
3875 * config/s390/s390.md (mem_signal_fence): Remove.
3876 * doc/md.texi (mem_signal_fence): Remove.
3877 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
3879 * target-insns.def (mem_signal_fence): Remove.
3881 2017-09-01 Jakub Jelinek <jakub@redhat.com>
3884 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
3887 * asan.c (create_odr_indicator): Strip name encoding from assembler
3888 name before appending it after __odr_asan_.
3890 2017-09-01 Martin Liska <mliska@suse.cz>
3892 PR tree-optimization/82059
3893 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
3894 frequency only when an edge is redirected.
3896 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
3898 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
3899 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
3900 (arc_conditional_register_usage): Remove ARC600 lp_count
3902 (arc_file_start): Emit Tag_ARC_CPU_variation.
3903 (arc_can_use_doloop_p): New conditions to use ZOLs.
3904 (hwloop_fail): New function.
3905 (hwloop_optimize): Likewise.
3906 (hwloop_pattern_reg): Likewise.
3907 (arc_doloop_hooks): New struct, to be used with reorg_loops.
3908 (arc_reorg_loops): New function, calls reorg_loops.
3909 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
3910 (arc600_corereg_hazard): Remove ZOL checking, case handled by
3912 (arc_loop_hazard): Remove function, functionality moved into
3914 (arc_hazard): Remove arc_loop_hazard call.
3915 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
3916 into hwloop_optimize.
3917 (arc_label_align): Remove ZOL handling.
3918 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
3919 * config/arc/arc.md (doloop_begin): Remove pattern.
3920 (doloop_begin_i): Likewise.
3921 (doloop_end_i): Likewise.
3922 (doloop_fallback): Likewise.
3923 (doloop_fallback_m): Likewise.
3924 (doloop_end): Reimplement expand.
3925 (arc_lp): New pattern for LP instruction.
3926 (loop_end): New pattern.
3927 (loop_fail): Likewise.
3928 (decrement_and_branch_until_zero): Likewise.
3929 * config/arc/arc.opt (mlpc-width): New option.
3930 * doc/invoke.texi (mlpc-width): Document option.
3932 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
3934 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
3935 (arc_ccfsm_advance): Fix checking for delay slots.
3936 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
3938 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
3940 * config/arc/arc.md (movqi_insn): Add stores to save constant long
3942 (movhi_insn): Update store instruction constraint which are saving
3943 6-bit short immediates.
3944 (movsi_insn): Consider also short scaled load operations.
3945 (zero_extendhisi2_i): Use Usd constraint instead of T.
3946 (extendhisi2_i): Add q constraint.
3947 (arc_clzsi2): Add type and length attributes.
3948 (arc_ctzsi2): Likewise.
3949 * config/arc/constraints.md (Usc): Update constraint, the
3950 assembler can parse two relocations for a single instruction.
3952 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
3954 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
3955 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
3957 2017-08-31 Olivier Hainque <hainque@adacore.com>
3959 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
3960 match as powerpc-wrs-vxworks*.
3962 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
3964 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
3965 register constraint for by-element operand.
3966 (aarch64_mls_elt_merge<mode>): Likewise.
3968 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
3970 * config/arc/arc.c (arc_can_follow_jump): Check for short
3973 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
3975 * config.gcc: Use g.opt for arc.
3976 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
3977 functionality moved to ...
3978 (legitimate_scaled_address_p): New function, ...here.
3979 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
3980 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
3981 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
3983 (arc_override_options): Handle G option.
3984 (arc_output_pic_addr_const): Correct function definition.
3985 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
3986 (arc_decl_anon_ns_mem_p): Delete.
3987 (arc_in_small_data_p): Overhaul this function to take into
3988 consideration the value given via G option.
3989 (arc_rewrite_small_data_1): Renamed and corrected old
3990 arc_rewrite_small_data function.
3991 (arc_rewrite_small_data): New function.
3992 (small_data_pattern): Don't use pic_offset_table_rtx.
3993 * config/arc/arc.h (CC1_SPEC): Recognize G option.
3994 * config/arc/simdext.md (movmisalignv2hi): Use
3995 prepare_move_operands function.
3997 (movmisalign*): Likewise.
3998 * doc/invoke.texi (ARC options): Document -G option.
4000 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
4002 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
4004 * config/arc/arc.c (arc_print_operand): Output scalled address for
4005 sdata whenever is possible.
4006 (arc_in_small_data_p): Allow sdata for 64bit datum when double
4007 load/stores are available.
4008 (compact_sda_memory_operand): Check for the alignment required by
4009 code density instructions.
4010 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
4012 * config/arc/constraints.md (Usd): Update constraint.
4013 (Us0): New constraint.
4014 (Usc): Update constraint.
4016 2017-08-31 Richard Biener <rguenther@suse.de>
4019 * dwarf2out.c (dwarf2out_early_global_decl): Process each
4022 2017-08-31 Tamar Christina <tamar.christina@arm.com>
4024 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
4025 Resize type_signature.
4027 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
4028 Alan Hayward <alan.hayward@arm.com>
4029 David Sherwood <david.sherwood@arm.com>
4031 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
4032 subregs whose inner modes can be stored in GPRs.
4033 (aarch64_classify_index): Likewise.
4035 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
4036 Alan Hayward <alan.hayward@arm.com>
4037 David Sherwood <david.sherwood@arm.com>
4039 * config/aarch64/iterators.md (V_cmp_result): Rename to...
4040 (V_INT_EQUIV): ...this.
4041 (v_cmp_result): Rename to...
4042 (v_int_equiv): ...this.
4043 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
4044 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
4045 (copysign<mode>3): Likewise.
4046 (aarch64_simd_bsl<mode>_internal): Likewise.
4047 (aarch64_simd_bsl<mode>): Likewise.
4048 (vec_cmp<mode><mode>): Likewise.
4049 (vcond<mode><mode>): Likewise.
4050 (vcond<v_cmp_mixed><mode>): Likewise.
4051 (vcondu<mode><v_cmp_mixed>): Likewise.
4052 (aarch64_cm<optab><mode>): Likewise.
4053 (aarch64_cmtst<mode>): Likewise.
4054 (aarch64_fac<optab><mode>): Likewise.
4055 (vec_perm_const<mode>): Likewise.
4056 (vcond_mask_<mode><v_cmp_result>): Rename to...
4057 (vcond_mask_<mode><v_int_equiv>): ...this.
4058 (vec_cmp<mode><v_cmp_result>): Rename to...
4059 (vec_cmp<mode><v_int_equiv>): ...this.
4061 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
4062 Alan Hayward <alan.hayward@arm.com>
4063 David Sherwood <david.sherwood@arm.com>
4065 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
4067 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
4068 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
4069 (UNSPEC_LD4_DREG): New unspecs.
4070 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
4071 (aarch64_ld2<mode>_dreg_be): Replace with...
4072 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
4074 (aarch64_ld3<mode>_dreg_le)
4075 (aarch64_ld3<mode>_dreg_be): Replace with...
4076 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
4078 (aarch64_ld4<mode>_dreg_le)
4079 (aarch64_ld4<mode>_dreg_be): Replace with...
4080 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
4083 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4085 PR tree-optimization/81987
4086 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
4087 insert an initializer in a location not dominated by the stride
4090 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
4092 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
4093 on the entire header of the finally block in the fallthru case.
4095 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
4097 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
4099 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
4101 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
4102 rs6000_emit_move_from_cr and call renamed function.
4103 (rs6000_emit_prologue): Call renamed functions.
4104 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
4105 movesi_from_cr, remove volatile CRs.
4107 2017-08-30 Jon Beniston <jon@beniston.com>
4108 Richard Biener <rguenther@suse.de>
4110 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
4111 of VECTOR_MODE_P check.
4112 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
4113 element vector types.
4115 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4117 * df.h (df_read_modify_subreg_p): Remove in favor of...
4118 * rtl.h (read_modify_subreg_p): ...this new function. Take a
4119 const_rtx instead of an rtx.
4120 * cprop.c (local_cprop_find_used_regs): Update accordingly.
4121 * df-problems.c (df_word_lr_mark_ref): Likewise.
4122 * ira-lives.c (mark_pseudo_reg_live): Likewise.
4123 (mark_pseudo_reg_dead): Likewise.
4124 (mark_ref_dead): Likewise.
4125 * reginfo.c (init_subregs_of_mode): Likewise.
4126 * sched-deps.c (sched_analyze_1): Likewise.
4127 * df-scan.c (df_def_record_1): Likewise.
4128 (df_uses_record): Likewise.
4129 (df_read_modify_subreg_p): Remove in favor of...
4130 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
4131 const_rtx instead of an rtx.
4133 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4134 Alan Hayward <alan.hayward@arm.com>
4135 David Sherwood <david.sherwood@arm.com>
4137 * rtl.h (partial_subreg_p): New function.
4138 * caller-save.c (save_call_clobbered_regs): Use it.
4139 * calls.c (expand_call): Likewise.
4140 * combine.c (combinable_i3pat): Likewise.
4141 (simplify_set): Likewise.
4142 (make_extraction): Likewise.
4143 (make_compound_operation_int): Likewise.
4144 (gen_lowpart_or_truncate): Likewise.
4145 (force_to_mode): Likewise.
4146 (make_field_assignment): Likewise.
4147 (reg_truncated_to_mode): Likewise.
4148 (record_truncated_value): Likewise.
4149 (move_deaths): Likewise.
4150 * cse.c (record_jump_cond): Likewise.
4151 (cse_insn): Likewise.
4152 * cselib.c (cselib_lookup_1): Likewise.
4153 * expmed.c (extract_bit_field_using_extv): Likewise.
4154 * function.c (assign_parm_setup_reg): Likewise.
4155 * ifcvt.c (noce_convert_multiple_sets): Likewise.
4156 * ira-build.c (create_insn_allocnos): Likewise.
4157 * lra-coalesce.c (merge_pseudos): Likewise.
4158 * lra-constraints.c (match_reload): Likewise.
4159 (simplify_operand_subreg): Likewise.
4160 (curr_insn_transform): Likewise.
4161 * lra-lives.c (process_bb_lives): Likewise.
4162 * lra.c (new_insn_reg): Likewise.
4163 (lra_substitute_pseudo): Likewise.
4164 * regcprop.c (mode_change_ok): Likewise.
4165 (maybe_mode_change): Likewise.
4166 (copyprop_hardreg_forward_1): Likewise.
4167 * reload.c (push_reload): Likewise.
4168 (find_reloads): Likewise.
4169 (find_reloads_subreg_address): Likewise.
4170 * reload1.c (alter_reg): Likewise.
4171 (eliminate_regs_1): Likewise.
4172 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4174 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
4176 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
4177 back to if statements, including unpack.
4179 2017-08-30 Martin Liska <mliska@suse.cz>
4182 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
4184 (func_checker::compare_asm_inputs_outputs): ... this function.
4185 (func_checker::compare_gimple_asm): Use the function to compare
4186 also ASM constrains.
4187 * ipa-icf-gimple.h: Rename the function.
4189 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4190 Alan Hayward <alan.hayward@arm.com>
4191 David Sherwood <david.sherwood@arm.com>
4193 * coretypes.h (complex_mode): New type.
4194 * gdbhooks.py (build_pretty_printer): Handle it.
4195 * machmode.h (complex_mode): New class.
4196 (complex_mode::includes_p): New function.
4197 (is_complex_int_mode): Likewise.
4198 (is_complex_float_mode): Likewise.
4199 * genmodes.c (get_mode_class): Handle complex mode classes.
4200 * function.c (expand_function_end): Use is_complex_int_mode.
4202 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4203 Alan Hayward <alan.hayward@arm.com>
4204 David Sherwood <david.sherwood@arm.com>
4206 * coretypes.h (scalar_mode_pod): New typedef.
4207 * gdbhooks.py (build_pretty_printer): Handle it.
4208 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
4209 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
4210 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
4211 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
4214 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4215 Alan Hayward <alan.hayward@arm.com>
4216 David Sherwood <david.sherwood@arm.com>
4218 * machmode.h (mode_for_vector): Take a scalar_mode instead
4220 * stor-layout.c (mode_for_vector): Likewise.
4221 * explow.c (promote_mode): Use as_a <scalar_mode>.
4222 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
4224 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4225 Alan Hayward <alan.hayward@arm.com>
4226 David Sherwood <david.sherwood@arm.com>
4228 * target.def (preferred_simd_mode): Take a scalar_mode
4229 instead of a machine_mode.
4230 * targhooks.h (default_preferred_simd_mode): Likewise.
4231 * targhooks.c (default_preferred_simd_mode): Likewise.
4232 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
4233 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
4234 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
4235 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
4236 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
4237 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
4238 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
4239 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
4241 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
4242 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
4243 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
4244 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
4245 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
4246 * doc/tm.texi: Regenerate.
4247 * optabs-query.c (can_vec_mask_load_store_p): Return false for
4250 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4251 Alan Hayward <alan.hayward@arm.com>
4252 David Sherwood <david.sherwood@arm.com>
4254 * target.def (scalar_mode_supported_p): Take a scalar_mode
4255 instead of a machine_mode.
4256 * targhooks.h (default_scalar_mode_supported_p): Likewise.
4257 * targhooks.c (default_scalar_mode_supported_p): Likewise.
4258 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
4259 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
4260 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
4261 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
4262 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
4263 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
4264 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
4265 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
4266 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
4267 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
4268 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
4269 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
4271 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
4272 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
4273 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
4274 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
4275 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
4277 * doc/tm.texi: Regenerate.
4279 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4280 Alan Hayward <alan.hayward@arm.com>
4281 David Sherwood <david.sherwood@arm.com>
4283 * coretypes.h (opt_scalar_mode): New typedef.
4284 * gdbhooks.py (build_pretty_printers): Handle it.
4285 * machmode.h (mode_iterator::get_2xwider): Add overload for
4287 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
4289 * expr.c (convert_mode_scalar): Likewise.
4290 * omp-low.c (omp_clause_aligned_alignment): Likewise.
4291 * optabs.c (expand_float): Likewise.
4292 (expand_fix): Likewise.
4293 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
4295 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4296 Alan Hayward <alan.hayward@arm.com>
4297 David Sherwood <david.sherwood@arm.com>
4299 * optabs.c (expand_float): Explicitly check for scalars before
4300 using a branching expansion.
4301 (expand_fix): Likewise.
4303 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4304 Alan Hayward <alan.hayward@arm.com>
4305 David Sherwood <david.sherwood@arm.com>
4307 * expr.c (convert_mode): Split scalar handling out into...
4308 (convert_mode_scalar): ...this new function. Treat the modes
4311 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4312 Alan Hayward <alan.hayward@arm.com>
4313 David Sherwood <david.sherwood@arm.com>
4315 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
4317 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
4319 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4320 Alan Hayward <alan.hayward@arm.com>
4321 David Sherwood <david.sherwood@arm.com>
4323 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
4324 rather than a machine_mode.
4325 (fixed_from_string): Likewise.
4326 (fixed_convert): Likewise.
4327 (fixed_convert_from_int): Likewise.
4328 (fixed_convert_from_real): Likewise.
4329 (real_convert_from_fixed): Likewise.
4330 * fixed-value.c (fixed_from_double_int): Likewise.
4331 (fixed_from_string): Likewise.
4332 (fixed_convert): Likewise.
4333 (fixed_convert_from_int): Likewise.
4334 (fixed_convert_from_real): Likewise.
4335 (real_convert_from_fixed): Likewise.
4336 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
4338 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4339 Alan Hayward <alan.hayward@arm.com>
4340 David Sherwood <david.sherwood@arm.com>
4342 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
4343 of separate mode class checks. Do not allow vector modes here.
4344 (immed_wide_int_const): Use as_a <scalar_mode>.
4345 * explow.c (trunc_int_for_mode): Likewise.
4346 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
4347 (wi::shwi): Likewise.
4348 (wi::min_value): Likewise.
4349 (wi::max_value): Likewise.
4350 * dwarf2out.c (loc_descriptor): Likewise.
4351 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
4354 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4355 Alan Hayward <alan.hayward@arm.com>
4356 David Sherwood <david.sherwood@arm.com>
4358 * tree.h (SCALAR_TYPE_MODE): New macro.
4359 * expr.c (expand_expr_addr_expr_1): Use it.
4360 (expand_expr_real_2): Likewise.
4361 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
4362 (fold_convert_const_fixed_from_int): Likewise.
4363 (fold_convert_const_fixed_from_real): Likewise.
4364 (native_encode_fixed): Likewise
4365 (native_encode_complex): Likewise
4366 (native_encode_vector): Likewise.
4367 (native_interpret_fixed): Likewise.
4368 (native_interpret_real): Likewise.
4369 (native_interpret_complex): Likewise.
4370 (native_interpret_vector): Likewise.
4371 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
4372 (simd_clone_adjust_argument_types): Likewise.
4373 (simd_clone_init_simd_arrays): Likewise.
4374 (simd_clone_adjust): Likewise.
4375 * stor-layout.c (layout_type): Likewise.
4376 * tree.c (build_minus_one_cst): Likewise.
4377 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
4378 * tree-inline.c (estimate_move_cost): Likewise.
4379 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
4380 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
4381 (vectorizable_reduction): Likewise.
4382 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
4383 (vect_recog_mixed_size_cond_pattern): Likewise.
4384 (check_bool_pattern): Likewise.
4385 (adjust_bool_pattern): Likewise.
4386 (search_type_for_mask_1): Likewise.
4387 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
4388 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
4389 (vectorizable_load): Likewise.
4390 (vectorizable_store): Likewise.
4391 * ubsan.c (ubsan_encode_value): Likewise.
4392 * varasm.c (output_constant): Likewise.
4394 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4395 Alan Hayward <alan.hayward@arm.com>
4396 David Sherwood <david.sherwood@arm.com>
4398 * coretypes.h (scalar_mode): New class.
4399 * machmode.h (scalar_mode): Likewise.
4400 (scalar_mode::includes_p): New function.
4401 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
4402 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
4403 * genmodes.c (get_mode_class): Handle remaining scalar modes.
4404 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
4405 * expmed.c (store_bit_field_1): Likewise.
4406 (extract_bit_field_1): Likewise.
4407 * expr.c (write_complex_part): Likewise.
4408 (read_complex_part): Likewise.
4409 (emit_move_complex_push): Likewise.
4410 (expand_expr_real_2): Likewise.
4411 * function.c (assign_parm_setup_reg): Likewise.
4412 (assign_parms_unsplit_complex): Likewise.
4413 * optabs.c (expand_binop): Likewise.
4414 * rtlanal.c (subreg_get_info): Likewise.
4415 * simplify-rtx.c (simplify_immed_subreg): Likewise.
4416 * varasm.c (output_constant_pool_2): Likewise.
4418 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4419 Alan Hayward <alan.hayward@arm.com>
4420 David Sherwood <david.sherwood@arm.com>
4422 * expmed.c (extract_high_half): Use scalar_int_mode and remove
4424 (expmed_mult_highpart_optab): Likewise.
4425 (expmed_mult_highpart): Likewise.
4427 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4428 Alan Hayward <alan.hayward@arm.com>
4429 David Sherwood <david.sherwood@arm.com>
4431 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
4432 instead of a machine_mode.
4433 (builtin_memset_read_str): Likewise.
4434 * builtins.c (c_readstr): Likewise.
4435 (builtin_memcpy_read_str): Likewise.
4436 (builtin_strncpy_read_str): Likewise.
4437 (builtin_memset_read_str): Likewise.
4438 (builtin_memset_gen_str): Likewise.
4439 (expand_builtin_signbit): Use scalar_int_mode for local variables.
4440 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
4441 instead of a machine_mode.
4442 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
4444 (make_extraction): Likewise.
4445 (try_widen_shift_mode): Take and return scalar_int_modes instead
4447 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
4448 a scalar_int_mode instead of a machine_mode.
4449 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
4450 (avr_addr_space_pointer_mode): Likewise.
4451 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
4452 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
4453 (msp430_unwind_word_mode): Likewise.
4454 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
4455 (spu_addr_space_pointer_mode): Likewise.
4456 (spu_addr_space_address_mode): Likewise.
4457 (spu_libgcc_cmp_return_mode): Likewise.
4458 (spu_libgcc_shift_count_mode): Likewise.
4459 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
4460 (rl78_addr_space_pointer_mode): Likewise.
4461 (fl78_unwind_word_mode): Likewise.
4462 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
4464 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
4465 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
4466 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
4467 (mips_valid_pointer_mode): Likewise.
4468 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
4469 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
4470 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
4472 (ft32_addr_space_address_mode): Likewise.
4473 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
4474 scalar_int_mode instead of a machine_mode.
4475 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
4477 (m32c_addr_space_address_mode): Likewise.
4478 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
4479 (rs6000_eh_return_filter_mode): Likewise.
4480 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
4481 (rs6000_eh_return_filter_mode): Likewise.
4482 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
4483 (s390_libgcc_shift_count_mode): Likewise.
4484 (s390_unwind_word_mode): Likewise.
4485 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
4487 * target.def (mode_rep_extended): Likewise.
4488 (valid_pointer_mode): Likewise.
4489 (addr_space.valid_pointer_mode): Likewise.
4490 (eh_return_filter_mode): Return a scalar_int_mode rather than
4492 (libgcc_cmp_return_mode): Likewise.
4493 (libgcc_shift_count_mode): Likewise.
4494 (unwind_word_mode): Likewise.
4495 (addr_space.pointer_mode): Likewise.
4496 (addr_space.address_mode): Likewise.
4497 * doc/tm.texi: Regenerate.
4498 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
4500 (do_jump): Use scalar_int_mode for local variables.
4501 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
4502 rather than a machine_mode.
4503 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
4504 (scompare_loc_descriptor_wide): Likewise.
4505 (scompare_loc_descriptor_narrow): Likewise.
4506 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
4508 * except.c (sjlj_emit_dispatch_table): Likewise.
4509 (expand_builtin_eh_copy_values): Likewise.
4510 * explow.c (convert_memory_address_addr_space_1): Likewise.
4511 Take a scalar_int_mode rather than a machine_mode.
4512 (convert_memory_address_addr_space): Take a scalar_int_mode rather
4513 than a machine_mode.
4514 (memory_address_addr_space): Use scalar_int_mode for local variables.
4515 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
4516 rather than a machine_mode.
4517 * expmed.c (mask_rtx): Likewise.
4518 (init_expmed_one_conv): Likewise.
4519 (expand_mult_highpart_adjust): Likewise.
4520 (extract_high_half): Likewise.
4521 (expmed_mult_highpart_optab): Likewise.
4522 (expmed_mult_highpart): Likewise.
4523 (expand_smod_pow2): Likewise.
4524 (expand_sdiv_pow2): Likewise.
4525 (emit_store_flag_int): Likewise.
4526 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
4528 (extract_low_bits): Likewise.
4529 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
4531 * expr.c (pieces_addr::adjust): Likewise.
4532 (can_store_by_pieces): Likewise.
4533 (store_by_pieces): Likewise.
4534 (clear_by_pieces_1): Likewise.
4535 (expand_expr_addr_expr_1): Likewise.
4536 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
4537 (expand_expr_real_1): Likewise.
4538 (try_casesi): Likewise.
4539 * final.c (shorten_branches): Likewise.
4540 * fold-const.c (fold_convert_const_int_from_fixed): Change the
4541 type of "mode" to machine_mode.
4542 * internal-fn.c (expand_arith_overflow_result_store): Take a
4543 scalar_int_mode rather than a machine_mode.
4544 (expand_mul_overflow): Use scalar_int_mode for local variables.
4545 * loop-doloop.c (doloop_modify): Likewise.
4546 (doloop_optimize): Likewise.
4547 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
4548 than a machine_mode.
4549 (expand_doubleword_shift_condmove): Likewise.
4550 (expand_doubleword_shift): Likewise.
4551 (expand_doubleword_clz): Likewise.
4552 (expand_doubleword_popcount): Likewise.
4553 (expand_doubleword_parity): Likewise.
4554 (expand_absneg_bit): Use scalar_int_mode for local variables.
4555 (prepare_float_lib_cmp): Likewise.
4556 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
4557 rather than a machine_mode.
4558 (convert_memory_address_addr_space): Likewise.
4559 (get_mode_bounds): Likewise.
4560 (get_address_mode): Return a scalar_int_mode rather than a
4562 * rtlanal.c (get_address_mode): Likewise.
4563 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
4564 than a machine_mode.
4565 * targhooks.c (default_mode_rep_extended): Likewise.
4566 (default_valid_pointer_mode): Likewise.
4567 (default_addr_space_valid_pointer_mode): Likewise.
4568 (default_eh_return_filter_mode): Return a scalar_int_mode rather
4569 than a machine_mode.
4570 (default_libgcc_cmp_return_mode): Likewise.
4571 (default_libgcc_shift_count_mode): Likewise.
4572 (default_unwind_word_mode): Likewise.
4573 (default_addr_space_pointer_mode): Likewise.
4574 (default_addr_space_address_mode): Likewise.
4575 * targhooks.h (default_eh_return_filter_mode): Likewise.
4576 (default_libgcc_cmp_return_mode): Likewise.
4577 (default_libgcc_shift_count_mode): Likewise.
4578 (default_unwind_word_mode): Likewise.
4579 (default_addr_space_pointer_mode): Likewise.
4580 (default_addr_space_address_mode): Likewise.
4581 (default_mode_rep_extended): Take a scalar_int_mode rather than
4583 (default_valid_pointer_mode): Likewise.
4584 (default_addr_space_valid_pointer_mode): Likewise.
4585 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
4587 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
4588 rather than a machine_mode.
4589 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
4590 for local variables.
4591 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
4592 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
4593 than a machine_mode.
4595 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4596 Alan Hayward <alan.hayward@arm.com>
4597 David Sherwood <david.sherwood@arm.com>
4599 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
4600 the mode argument to scalar_int_mode.
4601 (do_jump_by_parts_zero_rtx): Likewise.
4602 (do_jump_by_parts_equality_rtx): Likewise.
4603 (do_jump_by_parts_greater): Take a mode argument.
4604 (do_jump_by_parts_equality): Likewise.
4605 (do_jump_1): Update calls accordingly.
4607 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4608 Alan Hayward <alan.hayward@arm.com>
4609 David Sherwood <david.sherwood@arm.com>
4611 * is-a.h (safe_dyn_cast): New function.
4612 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
4613 (jump_table_for_label): Likewise.
4614 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
4615 instead of an rtx_insn *.
4616 (shorten_branches): Use dyn_cast instead of LABEL_P and
4617 JUMP_TABLE_DATA_P. Use jump_table_for_label and
4618 rtx_jump_table_data::get_data_mode.
4619 (final_scan_insn): Likewise.
4621 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4622 Alan Hayward <alan.hayward@arm.com>
4623 David Sherwood <david.sherwood@arm.com>
4625 * combine.c (try_combine): Use is_a <scalar_int_mode> when
4626 trying to combine a full-register integer set with a subreg
4629 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4630 Alan Hayward <alan.hayward@arm.com>
4631 David Sherwood <david.sherwood@arm.com>
4633 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
4634 that is always either address_mode or pointer_mode.
4636 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4637 Alan Hayward <alan.hayward@arm.com>
4638 David Sherwood <david.sherwood@arm.com>
4640 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
4641 when the two are known to be equal.
4643 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4644 Alan Hayward <alan.hayward@arm.com>
4645 David Sherwood <david.sherwood@arm.com>
4647 * simplify-rtx.c (simplify_const_unary_operation): Use
4648 is_a <scalar_int_mode> instead of checking for a nonzero
4649 precision. Forcibly convert op_mode to a scalar_int_mode
4650 in that case. More clearly differentiate the operand and
4651 result modes and use the former when deciding what the value
4652 of a count-bits operation should be. Use is_int_mode instead
4653 of checking for a MODE_INT. Remove redundant check for whether
4654 this mode has a zero precision.
4656 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4657 Alan Hayward <alan.hayward@arm.com>
4658 David Sherwood <david.sherwood@arm.com>
4660 * optabs.c (widen_leading): Change the type of the mode argument
4661 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
4662 (widen_bswap): Likewise.
4663 (expand_parity): Likewise.
4664 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
4665 (expand_ffs): Likewise.
4666 (epand_unop): Check for scalar integer modes before calling the
4669 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4670 Alan Hayward <alan.hayward@arm.com>
4671 David Sherwood <david.sherwood@arm.com>
4673 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
4675 (expand_expr_real_1): Update call accordingly.
4677 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4678 Alan Hayward <alan.hayward@arm.com>
4679 David Sherwood <david.sherwood@arm.com>
4681 * expmed.c (store_bit_field_using_insv): Add op0_mode and
4682 value_mode arguments. Use scalar_int_mode internally.
4683 (store_bit_field_1): Rename the new integer mode from imode
4684 to op0_mode and use it instead of GET_MODE (op0). Update calls
4685 to store_split_bit_field, store_bit_field_using_insv and
4686 store_fixed_bit_field.
4687 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
4688 Use scalar_int_mode internally. Use a bit count rather than a mode
4689 when calculating the largest bit size for get_best_mode.
4690 Update calls to store_split_bit_field and store_fixed_bit_field_1.
4691 (store_fixed_bit_field_1): Add mode and value_mode arguments.
4692 Remove assertion that OP0 has a scalar integer mode.
4693 (store_split_bit_field): Add op0_mode and value_mode arguments.
4694 Update calls to extract_fixed_bit_field.
4695 (extract_bit_field_using_extv): Add an op0_mode argument.
4696 Use scalar_int_mode internally.
4697 (extract_bit_field_1): Rename the new integer mode from imode to
4698 op0_mode and use it instead of GET_MODE (op0). Update calls to
4699 extract_split_bit_field, extract_bit_field_using_extv and
4700 extract_fixed_bit_field.
4701 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
4702 to extract_split_bit_field and extract_fixed_bit_field_1.
4703 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
4704 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
4706 (extract_split_bit_field): Add an op0_mode argument. Update call
4707 to extract_fixed_bit_field.
4709 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4710 Alan Hayward <alan.hayward@arm.com>
4711 David Sherwood <david.sherwood@arm.com>
4713 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
4714 * explow.c (hard_function_value): Likewise.
4715 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
4716 convert_to_mode call outside the loop.
4717 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
4718 for the mode iterator. Require the mode specified by max_pieces
4720 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
4722 (copy_blkmode_to_reg): Likewise.
4723 (set_storage_via_setmem): Likewise.
4724 * optabs.c (prepare_cmp_insn): Likewise.
4725 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
4726 * stor-layout.c (finish_bitfield_representative): Likewise.
4728 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4729 Alan Hayward <alan.hayward@arm.com>
4730 David Sherwood <david.sherwood@arm.com>
4732 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
4733 * expr.c (convert_move): Use them.
4734 (convert_modes): Likewise.
4735 (store_expr_with_bounds): Likewise.
4737 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4738 Alan Hayward <alan.hayward@arm.com>
4739 David Sherwood <david.sherwood@arm.com>
4741 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
4742 parameter for the mode of "x". Remove the "known_x", "known_mode"
4743 and "known_ret" arguments. Change the type of the mode argument
4745 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
4746 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
4747 (reg_num_sign_bit_copies_for_combine): Likewise.
4748 * rtlanal.c (nonzero_bits1): Likewise.
4749 (num_sign_bit_copies1): Likewise.
4750 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
4751 (reg_num_sign_bit_copies_general): Likewise.
4752 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
4753 (reg_nonzero_bits_general): Likewise.
4755 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4756 Alan Hayward <alan.hayward@arm.com>
4757 David Sherwood <david.sherwood@arm.com>
4759 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
4760 than in subroutines. Return 1 for non-integer modes.
4761 (cached_num_sign_bit_copies): Change the type of the mode parameter
4763 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
4764 classes. Handle CONST_INT_P first and then check whether X also
4765 has a scalar integer mode. Check the same thing for inner registers
4766 of a SUBREG and for values that are being extended or truncated.
4768 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4769 Alan Hayward <alan.hayward@arm.com>
4770 David Sherwood <david.sherwood@arm.com>
4772 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
4773 in subroutines. Return the mode mask for non-integer modes.
4774 (cached_nonzero_bits): Change the type of the mode parameter
4776 (nonzero_bits1): Likewise. Remove early exit for other mode
4777 classes. Handle CONST_INT_P first and then check whether X
4778 also has a scalar integer mode.
4780 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4781 Alan Hayward <alan.hayward@arm.com>
4782 David Sherwood <david.sherwood@arm.com>
4784 * expr.c (widest_int_mode_for_size): Make the comment match the code.
4785 Return a scalar_int_mode and assert that the size is greater than
4787 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
4788 (op_by_pieces_d::op_by_pieces_d): Likewise.
4789 (op_by_pieces_d::run): Likewise.
4790 (can_store_by_pieces): Likewise.
4792 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4793 Alan Hayward <alan.hayward@arm.com>
4794 David Sherwood <david.sherwood@arm.com>
4796 * combine.c (extract_left_shift): Add a mode argument and update
4798 (make_compound_operation_int): Change the type of the mode parameter
4799 to scalar_int_mode and update the call to extract_left_shift.
4801 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4802 Alan Hayward <alan.hayward@arm.com>
4803 David Sherwood <david.sherwood@arm.com>
4805 * combine.c (simplify_and_const_int): Change the type of the mode
4806 parameter to scalar_int_mode.
4807 (simplify_and_const_int_1): Likewise. Update recursive call.
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 (simplify_compare_const): Check that the mode is a
4814 scalar_int_mode (rather than VOIDmode) before testing its
4816 (simplify_comparison): Move COMPARISON_P handling out of the
4817 loop and restrict the latter part of the loop to scalar_int_modes.
4818 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
4819 and when considering SUBREG_REGs. Use is_int_mode instead of
4820 checking GET_MODE_CLASS against MODE_INT.
4822 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4823 Alan Hayward <alan.hayward@arm.com>
4824 David Sherwood <david.sherwood@arm.com>
4826 * combine.c (try_widen_shift_mode): Move check for equal modes to...
4827 (simplify_shift_const_1): ...here. Use scalar_int_mode for
4828 shift_unit_mode and for modes involved in scalar shifts.
4830 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4831 Alan Hayward <alan.hayward@arm.com>
4832 David Sherwood <david.sherwood@arm.com>
4834 * combine.c (force_int_to_mode): New function, split out from...
4835 (force_to_mode): ...here. Keep xmode up-to-date and use it
4836 instead of GET_MODE (x).
4838 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4839 Alan Hayward <alan.hayward@arm.com>
4840 David Sherwood <david.sherwood@arm.com>
4842 * optabs-query.h (extraction_insn::struct_mode): Change type to
4843 opt_scalar_int_mode and update comment.
4844 (extraction_insn::field_mode): Change type to scalar_int_mode.
4845 (extraction_insn::pos_mode): Likewise.
4846 * combine.c (make_extraction): Update accordingly.
4847 * optabs-query.c (get_traditional_extraction_insn): Likewise.
4848 (get_optab_extraction_insn): Likewise.
4849 * recog.c (simplify_while_replacing): Likewise.
4850 * expmed.c (narrow_bit_field_mem): Change the type of the mode
4851 parameter to opt_scalar_int_mode.
4853 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4854 Alan Hayward <alan.hayward@arm.com>
4855 David Sherwood <david.sherwood@arm.com>
4857 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
4858 to a scalar_int_mode instead of a machine_mode.
4859 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
4860 (get_best_mode): Return a boolean and use a pointer argument to store
4861 the selected mode. Replace the limit mode parameter with a bit limit.
4862 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
4863 for the values returned by bit_field_mode_iterator::next_mode.
4864 (store_bit_field): Update call to get_best_mode.
4865 (store_fixed_bit_field): Likewise.
4866 (extract_fixed_bit_field): Likewise.
4867 * expr.c (optimize_bitfield_assignment_op): Likewise.
4868 * fold-const.c (optimize_bit_field_compare): Likewise.
4869 (fold_truth_andor_1): Likewise.
4870 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
4871 Update for new type of m_mode.
4872 (get_best_mode): As above.
4874 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4875 Alan Hayward <alan.hayward@arm.com>
4876 David Sherwood <david.sherwood@arm.com>
4878 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
4879 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
4880 (store_bit_field): Check is_a <scalar_int_mode> before calling
4881 strict_volatile_bitfield_p.
4882 (extract_bit_field): Likewise.
4884 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4885 Alan Hayward <alan.hayward@arm.com>
4886 David Sherwood <david.sherwood@arm.com>
4888 * target.def (cstore_mode): Return a scalar_int_mode.
4889 * doc/tm.texi: Regenerate.
4890 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
4891 * targhooks.h (default_cstore_mode): Likewise.
4892 * targhooks.c (default_cstore_mode): Likewise, using a forced
4894 * expmed.c (emit_cstore): Expect the target of the cstore to be
4897 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4898 Alan Hayward <alan.hayward@arm.com>
4899 David Sherwood <david.sherwood@arm.com>
4901 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
4903 (niter_desc): Likewise mode.
4904 (iv_analyze): Add a mode parameter.
4906 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
4907 and change its type to scalar_int_mode.
4908 * loop-iv.c: Update commentary at head of file.
4909 (iv_constant): Pass the mode paraeter before the rtx it describes
4910 and change its type to scalar_int_mode. Remove VOIDmode handling.
4911 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
4912 (iv_extend): Likewise.
4913 (shorten_into_mode): Likewise.
4914 (iv_add): Use scalar_int_mode.
4915 (iv_mult): Likewise.
4916 (iv_shift): Likewise.
4917 (canonicalize_iv_subregs): Likewise.
4918 (get_biv_step_1): Pass the outer_mode parameter before the rtx
4919 it describes and change its mode to scalar_int_mode. Also change
4920 the type of the returned inner_mode to scalar_int_mode.
4921 (get_biv_step): Likewise, turning outer_mode from a pointer
4922 into a direct parameter. Update call to get_biv_step_1.
4923 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
4924 iv_constant and get_biv_step.
4925 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
4926 and change its type to scalar_int_mode. Don't initialise iv->mode
4927 to VOIDmode and remove later checks for its still being VOIDmode.
4928 Update calls to iv_analyze_op and iv_analyze_expr. Check
4929 is_a <scalar_int_mode> when changing the mode under consideration.
4930 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
4931 Update call to iv_analyze_expr.
4932 (iv_analyze_op): Add a mode parameter. Reject subregs whose
4933 inner register is not also a scalar_int_mode. Update call to
4935 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
4936 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
4937 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
4938 separate mode class checks. Update calls to iv_analyze. Remove
4939 fix-up of VOIDmodes after iv_analyze_biv.
4940 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
4941 don't have a scalar_int_mode. Update call to biv_p.
4943 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4944 Alan Hayward <alan.hayward@arm.com>
4945 David Sherwood <david.sherwood@arm.com>
4947 * cfgexpand.c (convert_debug_memory_address): Use
4948 as_a <scalar_int_mode>.
4949 * combine.c (expand_compound_operation): Likewise.
4950 (make_extraction): Likewise.
4951 (change_zero_ext): Likewise.
4952 (simplify_comparison): Likewise.
4953 * cse.c (cse_insn): Likewise.
4954 * dwarf2out.c (minmax_loc_descriptor): Likewise.
4955 (mem_loc_descriptor): Likewise.
4956 (loc_descriptor): Likewise.
4957 * expmed.c (init_expmed_one_mode): Likewise.
4958 (synth_mult): Likewise.
4959 (emit_store_flag_1): Likewise.
4960 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
4961 of a comparison with size.
4962 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
4963 (reduce_to_bit_field_precision): Likewise.
4964 * function.c (expand_function_end): Likewise.
4965 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
4966 * loop-doloop.c (doloop_modify): Likewise.
4967 * optabs.c (expand_binop): Likewise.
4968 (expand_unop): Likewise.
4969 (expand_copysign_absneg): Likewise.
4970 (prepare_cmp_insn): Likewise.
4971 (maybe_legitimize_operand): Likewise.
4972 * recog.c (const_scalar_int_operand): Likewise.
4973 * rtlanal.c (get_address_mode): Likewise.
4974 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4975 (simplify_cond_clz_ctz): Likewise.
4976 * tree-nested.c (get_nl_goto_field): Likewise.
4977 * tree.c (build_vector_type_for_mode): Likewise.
4978 * var-tracking.c (use_narrower_mode): Likewise.
4980 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4981 Alan Hayward <alan.hayward@arm.com>
4982 David Sherwood <david.sherwood@arm.com>
4984 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
4985 * builtins.c (expand_builtin_signbit): Use it.
4986 * cfgexpand.c (expand_debug_expr): Likewise.
4987 * dojump.c (do_jump): Likewise.
4988 (do_compare_and_jump): Likewise.
4989 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
4990 * expmed.c (make_tree): Likewise.
4991 * expr.c (expand_expr_real_2): Likewise.
4992 (expand_expr_real_1): Likewise.
4993 (try_casesi): Likewise.
4994 * fold-const-call.c (fold_const_call_ss): Likewise.
4995 * fold-const.c (unextend): Likewise.
4996 (extract_muldiv_1): Likewise.
4997 (fold_single_bit_test): Likewise.
4998 (native_encode_int): Likewise.
4999 (native_encode_string): Likewise.
5000 (native_interpret_int): Likewise.
5001 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
5002 * internal-fn.c (expand_addsub_overflow): Likewise.
5003 (expand_neg_overflow): Likewise.
5004 (expand_mul_overflow): Likewise.
5005 (expand_arith_overflow): Likewise.
5006 * match.pd: Likewise.
5007 * stor-layout.c (layout_type): Likewise.
5008 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
5009 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
5010 * tree-ssanames.c (get_range_info): Likewise.
5011 * tree-switch-conversion.c (array_value_type) Likewise.
5012 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
5013 (vect_recog_divmod_pattern): Likewise.
5014 (vect_recog_mixed_size_cond_pattern): Likewise.
5015 * tree-vrp.c (extract_range_basic): Likewise.
5016 (simplify_float_conversion_using_ranges): Likewise.
5017 * tree.c (int_fits_type_p): Likewise.
5018 * ubsan.c (instrument_bool_enum_load): Likewise.
5019 * varasm.c (mergeable_string_section): Likewise.
5020 (narrowing_initializer_constant_valid_p): Likewise.
5021 (output_constant): Likewise.
5023 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5024 Alan Hayward <alan.hayward@arm.com>
5025 David Sherwood <david.sherwood@arm.com>
5027 * machmode.h (NARROWEST_INT_MODE): New macro.
5028 * expr.c (alignment_for_piecewise_move): Use it instead of
5029 GET_CLASS_NARROWEST_MODE (MODE_INT).
5030 (push_block): Likewise.
5031 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
5033 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
5035 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5036 Alan Hayward <alan.hayward@arm.com>
5037 David Sherwood <david.sherwood@arm.com>
5039 * postreload.c (move2add_valid_value_p): Change the type of the
5040 mode parameter to scalar_int_mode.
5041 (move2add_use_add2_insn): Add a mode parameter and use it instead
5043 (move2add_use_add3_insn): Likewise.
5044 (reload_cse_move2add): Update accordingly.
5046 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5047 Alan Hayward <alan.hayward@arm.com>
5048 David Sherwood <david.sherwood@arm.com>
5050 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
5052 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
5053 * optabs.c (expand_unop): Likewise.
5055 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5056 Alan Hayward <alan.hayward@arm.com>
5057 David Sherwood <david.sherwood@arm.com>
5059 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
5060 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
5061 (popcount_loc_descriptor): Likewise.
5062 (bswap_loc_descriptor): Likewise.
5063 (rotate_loc_descriptor): Likewise.
5064 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
5065 calling the functions above.
5067 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5068 Alan Hayward <alan.hayward@arm.com>
5069 David Sherwood <david.sherwood@arm.com>
5071 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
5073 (try_combine): Likewise.
5074 (simplify_if_then_else): Likewise.
5075 * cse.c (cse_insn): Likewise.
5076 * dwarf2out.c (mem_loc_descriptor): Likewise.
5077 * emit-rtl.c (gen_lowpart_common): Likewise.
5078 * simplify-rtx.c (simplify_truncation): Likewise.
5079 (simplify_binary_operation_1): Likewise.
5080 (simplify_const_relational_operation): Likewise.
5081 (simplify_ternary_operation): Likewise.
5082 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
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 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
5089 * reload.c (push_reload): Likewise.
5090 (find_reloads): Likewise.
5092 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5093 Alan Hayward <alan.hayward@arm.com>
5094 David Sherwood <david.sherwood@arm.com>
5096 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
5097 (make_compound_operation_int): Likewise.
5098 (change_zero_ext): Likewise.
5099 * expr.c (convert_move): Likewise.
5100 (convert_modes): Likewise.
5101 * fwprop.c (forward_propagate_subreg): Likewise.
5102 * loop-iv.c (get_biv_step_1): Likewise.
5103 * optabs.c (widen_operand): Likewise.
5104 * postreload.c (move2add_valid_value_p): Likewise.
5105 * recog.c (simplify_while_replacing): Likewise.
5106 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5107 (simplify_binary_operation_1): Likewise. Remove redundant
5108 mode equality check.
5110 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5111 Alan Hayward <alan.hayward@arm.com>
5112 David Sherwood <david.sherwood@arm.com>
5114 * combine.c (combine_simplify_rtx): Add checks for
5115 is_a <scalar_int_mode>.
5116 (simplify_if_then_else): Likewise.
5117 (make_field_assignment): Likewise.
5118 (simplify_comparison): Likewise.
5119 * ifcvt.c (noce_try_bitop): Likewise.
5120 * loop-invariant.c (canonicalize_address_mult): Likewise.
5121 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5123 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5124 Alan Hayward <alan.hayward@arm.com>
5125 David Sherwood <david.sherwood@arm.com>
5127 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
5128 is_a <scalar_int_mode> instead of != BLKmode.
5130 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5131 Alan Hayward <alan.hayward@arm.com>
5132 David Sherwood <david.sherwood@arm.com>
5134 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
5135 instead of != VOIDmode.
5136 * combine.c (if_then_else_cond): Likewise.
5137 (change_zero_ext): Likewise.
5138 * dwarf2out.c (mem_loc_descriptor): Likewise.
5139 (loc_descriptor): Likewise.
5140 * rtlanal.c (canonicalize_condition): Likewise.
5141 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
5143 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5144 Alan Hayward <alan.hayward@arm.com>
5145 David Sherwood <david.sherwood@arm.com>
5147 * simplify-rtx.c (simplify_binary_operation_1): Use
5148 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
5150 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5151 Alan Hayward <alan.hayward@arm.com>
5152 David Sherwood <david.sherwood@arm.com>
5154 * wide-int.h (int_traits<unsigned char>) New class.
5155 (int_traits<unsigned short>) Likewise.
5156 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
5157 Use GET_MODE_UNIT_PRECISION and remove redundant test for
5159 * combine.c (set_nonzero_bits_and_sign_copies): Use
5160 is_a <scalar_int_mode>.
5161 (find_split_point): Likewise.
5162 (combine_simplify_rtx): Likewise.
5163 (simplify_logical): Likewise.
5164 (expand_compound_operation): Likewise.
5165 (expand_field_assignment): Likewise.
5166 (make_compound_operation): Likewise.
5167 (extended_count): Likewise.
5168 (change_zero_ext): Likewise.
5169 (simplify_comparison): Likewise.
5170 * dwarf2out.c (scompare_loc_descriptor): Likewise.
5171 (ucompare_loc_descriptor): Likewise.
5172 (minmax_loc_descriptor): Likewise.
5173 (mem_loc_descriptor): Likewise.
5174 (loc_descriptor): Likewise.
5175 * expmed.c (init_expmed_one_mode): Likewise.
5176 * lra-constraints.c (lra_constraint_offset): Likewise.
5177 * optabs.c (prepare_libcall_arg): Likewise.
5178 * postreload.c (move2add_note_store): Likewise.
5179 * reload.c (operands_match_p): Likewise.
5180 * rtl.h (load_extend_op): Likewise.
5181 * rtlhooks.c (gen_lowpart_general): Likewise.
5182 * simplify-rtx.c (simplify_truncation): Likewise.
5183 (simplify_unary_operation_1): Likewise.
5184 (simplify_binary_operation_1): Likewise.
5185 (simplify_const_binary_operation): Likewise.
5186 (simplify_const_relational_operation): Likewise.
5187 (simplify_subreg): Likewise.
5188 * stor-layout.c (bitwise_mode_for_mode): Likewise.
5189 * var-tracking.c (adjust_mems): Likewise.
5190 (prepare_call_arguments): Likewise.
5192 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5193 Alan Hayward <alan.hayward@arm.com>
5194 David Sherwood <david.sherwood@arm.com>
5196 * machmode.h (is_int_mode): New fuction.
5197 * combine.c (find_split_point): Use it.
5198 (combine_simplify_rtx): Likewise.
5199 (simplify_if_then_else): Likewise.
5200 (simplify_set): Likewise.
5201 (simplify_shift_const_1): Likewise.
5202 (simplify_comparison): Likewise.
5203 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
5204 * cse.c (notreg_cost): Likewise.
5205 (cse_insn): Likewise.
5206 * cselib.c (cselib_lookup_1): Likewise.
5207 * dojump.c (do_jump_1): Likewise.
5208 (do_compare_rtx_and_jump): Likewise.
5209 * dse.c (get_call_args): Likewise.
5210 * dwarf2out.c (rtl_for_decl_init): Likewise.
5211 (native_encode_initializer): Likewise.
5212 * expmed.c (emit_store_flag_1): Likewise.
5213 (emit_store_flag): Likewise.
5214 * expr.c (convert_modes): Likewise.
5215 (store_field): Likewise.
5216 (expand_expr_real_1): Likewise.
5217 * fold-const.c (fold_read_from_constant_string): Likewise.
5218 * gimple-ssa-sprintf.c (get_format_string): Likewise.
5219 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
5220 * optabs.c (expand_binop): Likewise.
5221 (expand_unop): Likewise.
5222 (expand_abs_nojump): Likewise.
5223 (expand_one_cmpl_abs_nojump): Likewise.
5224 * simplify-rtx.c (mode_signbit_p): Likewise.
5225 (val_signbit_p): Likewise.
5226 (val_signbit_known_set_p): Likewise.
5227 (val_signbit_known_clear_p): Likewise.
5228 (simplify_relational_operation_1): Likewise.
5229 * tree.c (vector_type_mode): Likewise.
5231 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5232 Alan Hayward <alan.hayward@arm.com>
5233 David Sherwood <david.sherwood@arm.com>
5235 * machmode.h (smallest_mode_for_size): Fix formatting.
5236 (smallest_int_mode_for_size): New function.
5237 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
5238 instead of smallest_mode_for_size.
5239 * combine.c (make_extraction): Likewise.
5240 * config/arc/arc.c (arc_expand_movmem): Likewise.
5241 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
5242 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
5243 * config/s390/s390.c (s390_expand_insv): Likewise.
5244 * config/sparc/sparc.c (assign_int_registers): Likewise.
5245 * config/spu/spu.c (spu_function_value): Likewise.
5246 (spu_function_arg): Likewise.
5247 * coverage.c (get_gcov_type): Likewise.
5248 (get_gcov_unsigned_t): Likewise.
5249 * dse.c (find_shift_sequence): Likewise.
5250 * expmed.c (store_bit_field_1): Likewise.
5251 * expr.c (convert_move): Likewise.
5252 (store_field): Likewise.
5253 * internal-fn.c (expand_arith_overflow): Likewise.
5254 * optabs-query.c (get_best_extraction_insn): Likewise.
5255 * optabs.c (expand_twoval_binop_libfunc): Likewise.
5256 * stor-layout.c (layout_type): Likewise.
5257 (initialize_sizetypes): Likewise.
5258 * targhooks.c (default_get_mask_mode): Likewise.
5259 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
5261 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5262 Alan Hayward <alan.hayward@arm.com>
5263 David Sherwood <david.sherwood@arm.com>
5265 * machmode.h (opt_mode::else_blk): New function.
5266 (int_mode_for_mode): Declare.
5267 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
5268 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
5270 * cfgexpand.c (expand_debug_expr): Likewise.
5271 * combine.c (gen_lowpart_or_truncate): Likewise.
5272 (gen_lowpart_for_combine): Likewise.
5273 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
5274 * config/avr/avr.c (avr_to_int_mode): Likewise.
5275 (avr_out_plus_1): Likewise.
5276 (avr_out_plus): Likewise.
5277 (avr_out_round): Likewise.
5278 * config/i386/i386.c (ix86_split_to_parts): Likewise.
5279 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
5280 (s390_expand_vcond): Likewise.
5281 * config/spu/spu.c (spu_split_immediate): Likewise.
5282 (spu_expand_mov): Likewise.
5283 * dse.c (get_stored_val): Likewise.
5284 * expmed.c (store_bit_field_1): Likewise.
5285 (convert_extracted_bit_field): Use int_mode_for_mode instead of
5287 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
5288 (extract_low_bits): Likewise.
5289 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
5290 handling rather than repeating the check.
5291 (emit_group_store): Likewise.
5292 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
5293 * optabs.c (expand_absneg_bit): Likewise.
5294 (expand_copysign_absneg): Likewise.
5295 (expand_copysign_bit): Likewise.
5296 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
5297 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
5298 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
5299 * var-tracking.c (prepare_call_arguments): Likewise.
5300 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
5301 int_mode_for_mode instead of mode_for_size.
5302 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
5304 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5305 Alan Hayward <alan.hayward@arm.com>
5306 David Sherwood <david.sherwood@arm.com>
5308 * machmode.h (int_mode_for_size): New function.
5309 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
5310 instead of mode_for_size.
5311 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
5313 * combine.c (expand_field_assignment): Use int_mode_for_size
5314 instead of mode_for_size.
5315 (make_extraction): Likewise.
5316 (simplify_shift_const_1): Likewise.
5317 (simplify_comparison): Likewise.
5318 * dojump.c (do_jump): Likewise.
5319 * dwarf2out.c (mem_loc_descriptor): Likewise.
5320 * emit-rtl.c (init_derived_machine_modes): Likewise.
5321 * expmed.c (flip_storage_order): Likewise.
5322 (convert_extracted_bit_field): Likewise.
5323 * expr.c (copy_blkmode_from_reg): Likewise.
5324 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
5325 * internal-fn.c (expand_mul_overflow): Likewise.
5326 * lower-subreg.c (simple_move): Likewise.
5327 * optabs-libfuncs.c (init_optabs): Likewise.
5328 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5329 * tree.c (vector_type_mode): Likewise.
5330 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
5331 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
5332 * tree-vect-generic.c (expand_vector_parallel): Likewise.
5333 * tree-vect-stmts.c (vectorizable_load): Likewise.
5334 (vectorizable_store): Likewise.
5336 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5337 Alan Hayward <alan.hayward@arm.com>
5338 David Sherwood <david.sherwood@arm.com>
5340 * coretypes.h (pod_mode): New type.
5341 (scalar_int_mode_pod): New typedef.
5342 * machmode.h (pod_mode): New class.
5343 (int_n_data_t::m): Change type to scalar_int_mode_pod.
5344 * genmodes.c (emit_mode_int_n): Update accordingly.
5345 * lower-subreg.h (target_lower_subreg): Change type to
5346 scalar_int_mode_pod.
5347 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
5348 scalar_int_mode_pod.
5350 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5351 Alan Hayward <alan.hayward@arm.com>
5352 David Sherwood <david.sherwood@arm.com>
5354 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
5355 machine_mode to scalar_int_mode.
5356 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
5357 (rs6000_option_override_internal): Remove cast to int.
5358 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
5359 machine_mode to scalar_int_mode.
5360 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
5361 (rs6000_option_override_internal): Remove cast to int.
5362 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
5363 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
5365 * config/s390/s390.c (s390_expand_builtin): Likewise.
5366 * coretypes.h (scalar_int_mode): New type.
5367 (opt_scalar_int_mode): New typedef.
5368 * machmode.h (scalar_int_mode): New class.
5369 (scalar_int_mode::includes_p): New function.
5370 (byte_mode): Change type to scalar_int_mode.
5371 (word_mode): Likewise.
5372 (ptr_mode): Likewise.
5373 * emit-rtl.c (byte_mode): Likewise.
5374 (word_mode): Likewise.
5375 (ptr_mode): Likewise.
5376 (init_derived_machine_modes): Update accordingly.
5377 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
5378 and MODE_PARTIAL_INT.
5379 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
5380 opt_scalar_int_mode.
5382 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5383 Alan Hayward <alan.hayward@arm.com>
5384 David Sherwood <david.sherwood@arm.com>
5386 * target.def (libgcc_floating_mode_supported_p): Take a
5388 * doc/tm.texi: Regenerate.
5389 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
5391 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
5392 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
5395 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5396 Alan Hayward <alan.hayward@arm.com>
5397 David Sherwood <david.sherwood@arm.com>
5399 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
5400 * doc/tm.texi: Regenerate.
5401 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
5402 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
5403 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
5404 * targhooks.h (default_floatn_mode): Likewise.
5405 * targhooks.c (default_floatn_mode): Likewise.
5406 * tree.c (build_common_tree_nodes): Update accordingly.
5408 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5409 Alan Hayward <alan.hayward@arm.com>
5410 David Sherwood <david.sherwood@arm.com>
5412 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
5413 (mode_iterator::iterate_p): Likewise.
5414 (mode_iterator::get_wider): Likewise.
5415 * expr.c (init_expr_target): Use opt_scalar_float_mode.
5417 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5418 Alan Hayward <alan.hayward@arm.com>
5419 David Sherwood <david.sherwood@arm.com>
5421 * coretypes.h (opt_scalar_float_mode): New typedef.
5422 * machmode.h (float_mode_for_size): New function.
5423 * emit-rtl.c (double_mode): Delete.
5424 (init_emit_once): Use float_mode_for_size.
5425 * stor-layout.c (layout_type): Likewise.
5426 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
5428 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5429 Alan Hayward <alan.hayward@arm.com>
5430 David Sherwood <david.sherwood@arm.com>
5432 * output.h (assemble_real): Take a scalar_float_mode.
5433 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
5434 * config/arm/arm.md (consttable_4): Likewise.
5435 (consttable_8): Likewise.
5436 (consttable_16): Likewise.
5437 * config/mips/mips.md (consttable_float): Likewise.
5438 * config/s390/s390.c (s390_output_pool_entry): Likewise.
5439 * varasm.c (assemble_real): Take a scalar_float_mode.
5440 (output_constant_pool_2): Update accordingly.
5441 (output_constant): Likewise.
5443 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5444 Alan Hayward <alan.hayward@arm.com>
5445 David Sherwood <david.sherwood@arm.com>
5447 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
5448 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
5449 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
5450 (native_encode_real): Likewise.
5451 (native_interpret_real): Likewise.
5452 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
5453 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
5455 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5456 Alan Hayward <alan.hayward@arm.com>
5457 David Sherwood <david.sherwood@arm.com>
5459 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
5460 <scalar_float_mode>. Simplify.
5461 (gen_extend_conv_libfunc): Likewise.
5463 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5464 Alan Hayward <alan.hayward@arm.com>
5465 David Sherwood <david.sherwood@arm.com>
5467 * coretypes.h (scalar_float_mode): New type.
5468 * machmode.h (mode_traits::from_int): Use machine_mode if
5469 USE_ENUM_MODES is defined.
5470 (is_a): New function.
5472 (dyn_cast): Likewise.
5473 (scalar_float_mode): New class.
5474 (scalar_float_mode::includes_p): New function.
5475 (is_float_mode): Likewise.
5476 * gdbhooks.py (MachineModePrinter): New class.
5477 (build_pretty_printer): Use it for scalar_float_mode.
5478 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
5479 (format_helper::format_helper): Turn into a template.
5480 * genmodes.c (get_mode_class): New function.
5481 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
5482 or machine_mode if none.
5483 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
5484 as_a <scalar_float_mode>.
5485 * dwarf2out.c (mem_loc_descriptor): Likewise.
5486 (insert_float): Likewise.
5487 (add_const_value_attribute): Likewise.
5488 * simplify-rtx.c (simplify_immed_subreg): Likewise.
5489 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
5490 (expand_unop): Update accordingly.
5491 (expand_abs_nojump): Likewise.
5492 (expand_copysign_absneg): Take a scalar_float_mode.
5493 (expand_copysign_bit): Likewise.
5494 (expand_copysign): Update accordingly.
5496 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5497 Alan Hayward <alan.hayward@arm.com>
5498 David Sherwood <david.sherwood@arm.com>
5500 * coretypes.h (opt_mode): New class.
5501 * machmode.h (opt_mode): Likewise.
5502 (opt_mode::else_void): New function.
5503 (opt_mode::require): Likewise.
5504 (opt_mode::exists): Likewise.
5505 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
5506 (GET_MODE_2XWIDER_MODE): Likewise.
5507 (mode_iterator::get_wider): Update accordingly.
5508 (mode_iterator::get_2xwider): Likewise.
5509 (mode_iterator::get_known_wider): Likewise, turning into a template.
5510 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
5511 forcing a wider mode to exist.
5512 * config/cr16/cr16.h (LONG_REG_P): Likewise.
5513 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
5514 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
5515 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
5516 * lower-subreg.c (init_lower_subreg): Likewise.
5517 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
5518 on the final iteration.
5519 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
5520 a wider mode exists before asking for a move pattern.
5521 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
5522 forcing a wider mode to exist.
5523 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
5524 returning false if no such mode exists.
5525 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
5526 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
5527 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
5528 Avoid checking for a MODE_INT if we already know the mode is not a
5530 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
5531 forcing a wider mode to exist.
5532 (expmed_mult_highpart_optab): Likewise.
5533 (expmed_mult_highpart): Likewise.
5534 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
5536 * lto-streamer-in.c (lto_input_mode_table): Likewise.
5537 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
5538 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
5539 * internal-fn.c (expand_mul_overflow): Update use of
5540 GET_MODE_2XWIDER_MODE.
5541 * omp-low.c (omp_clause_aligned_alignment): Likewise.
5542 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
5543 GET_MODE_WIDER_MODE.
5544 (convert_plusminus_to_widen): Likewise.
5545 * tree-switch-conversion.c (array_value_type): Likewise.
5546 * var-tracking.c (emit_note_insn_var_location): Likewise.
5547 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
5548 Return false inside rather than outside the loop if no wider mode
5550 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
5551 and GET_MODE_2XWIDER_MODE
5552 (can_compare_p): Use else_void.
5553 * gdbhooks.py (OptMachineModePrinter): New class.
5554 (build_pretty_printer): Use it for opt_mode.
5556 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5557 Alan Hayward <alan.hayward@arm.com>
5558 David Sherwood <david.sherwood@arm.com>
5560 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
5561 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
5563 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5564 Alan Hayward <alan.hayward@arm.com>
5565 David Sherwood <david.sherwood@arm.com>
5567 * machmode.h (mode_traits): New structure.
5568 (get_narrowest_mode): New function.
5569 (mode_iterator::start): Likewise.
5570 (mode_iterator::iterate_p): Likewise.
5571 (mode_iterator::get_wider): Likewise.
5572 (mode_iterator::get_known_wider): Likewise.
5573 (mode_iterator::get_2xwider): Likewise.
5574 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
5575 (FOR_EACH_MODE): Likewise.
5576 (FOR_EACH_MODE_FROM): Likewise.
5577 (FOR_EACH_MODE_UNTIL): Likewise.
5578 (FOR_EACH_WIDER_MODE): Likewise.
5579 (FOR_EACH_2XWIDER_MODE): Likewise.
5580 * builtins.c (expand_builtin_strlen): Use new mode iterators.
5581 * combine.c (simplify_comparison): Likewise
5582 * config/i386/i386.c (type_natural_mode): Likewise.
5583 * cse.c (cse_insn): Likewise.
5584 * dse.c (find_shift_sequence): Likewise.
5585 * emit-rtl.c (init_derived_machine_modes): Likewise.
5586 (init_emit_once): Likewise.
5587 * explow.c (hard_function_value): Likewise.
5588 * expmed.c (extract_fixed_bit_field_1): Likewise.
5589 (extract_bit_field_1): Likewise.
5590 (expand_divmod): Likewise.
5591 (emit_store_flag_1): Likewise.
5592 * expr.c (init_expr_target): Likewise.
5593 (convert_move): Likewise.
5594 (alignment_for_piecewise_move): Likewise.
5595 (widest_int_mode_for_size): Likewise.
5596 (emit_block_move_via_movmem): Likewise.
5597 (copy_blkmode_to_reg): Likewise.
5598 (set_storage_via_setmem): Likewise.
5599 (compress_float_constant): Likewise.
5600 * omp-low.c (omp_clause_aligned_alignment): Likewise.
5601 * optabs-query.c (get_best_extraction_insn): Likewise.
5602 * optabs.c (expand_binop): Likewise.
5603 (expand_twoval_unop): Likewise.
5604 (expand_twoval_binop): Likewise.
5605 (widen_leading): Likewise.
5606 (widen_bswap): Likewise.
5607 (expand_parity): Likewise.
5608 (expand_unop): Likewise.
5609 (prepare_cmp_insn): Likewise.
5610 (prepare_float_lib_cmp): Likewise.
5611 (expand_float): Likewise.
5612 (expand_fix): Likewise.
5613 (expand_sfix_optab): Likewise.
5614 * postreload.c (move2add_use_add2_insn): Likewise.
5615 * reg-stack.c (reg_to_stack): Likewise.
5616 * reginfo.c (choose_hard_reg_mode): Likewise.
5617 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
5618 * stor-layout.c (mode_for_size): Likewise.
5619 (smallest_mode_for_size): Likewise.
5620 (mode_for_vector): Likewise.
5621 (finish_bitfield_representative): Likewise.
5622 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
5623 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
5624 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
5625 * var-tracking.c (prepare_call_arguments): Likewise.
5627 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5628 Alan Hayward <alan.hayward@arm.com>
5629 David Sherwood <david.sherwood@arm.com>
5631 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
5632 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
5633 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
5634 * machmode.h (mode_size): Move earlier in file.
5635 (mode_precision): Likewise.
5636 (mode_inner): Likewise.
5637 (mode_nunits): Likewise.
5638 (mode_unit_size): Likewise.
5639 (unit_unit_precision): Likewise.
5640 (mode_wider): Likewise.
5641 (mode_2xwider): Likewise.
5642 (machine_mode): New class.
5643 (mode_to_bytes): New function.
5644 (mode_to_bits): Likewise.
5645 (mode_to_precision): Likewise.
5646 (mode_to_inner): Likewise.
5647 (mode_to_unit_size): Likewise.
5648 (mode_to_unit_precision): Likewise.
5649 (mode_to_nunits): Likewise.
5650 (GET_MODE_SIZE): Use mode_to_bytes.
5651 (GET_MODE_BITSIZE): Use mode_to_bits.
5652 (GET_MODE_PRECISION): Use mode_to_precision.
5653 (GET_MODE_INNER): Use mode_to_inner.
5654 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
5655 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
5656 (GET_MODE_NUNITS): Use mode_to_nunits.
5657 * system.h (ALWAYS_INLINE): New macro.
5658 * config/powerpcspe/powerpcspe-c.c
5659 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
5660 int for arg1_mode and arg2_mode.
5662 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5663 Alan Hayward <alan.hayward@arm.com>
5664 David Sherwood <david.sherwood@arm.com>
5666 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
5667 Prefix mode names with E_ in case statements.
5668 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
5669 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
5670 (aarch64_split_simd_move): Likewise.
5671 (aarch64_gen_storewb_pair): Likewise.
5672 (aarch64_gen_loadwb_pair): Likewise.
5673 (aarch64_gen_store_pair): Likewise.
5674 (aarch64_gen_load_pair): Likewise.
5675 (aarch64_get_condition_code_1): Likewise.
5676 (aarch64_constant_pool_reload_icode): Likewise.
5677 (get_rsqrte_type): Likewise.
5678 (get_rsqrts_type): Likewise.
5679 (get_recpe_type): Likewise.
5680 (get_recps_type): Likewise.
5681 (aarch64_gimplify_va_arg_expr): Likewise.
5682 (aarch64_simd_container_mode): Likewise.
5683 (aarch64_emit_load_exclusive): Likewise.
5684 (aarch64_emit_store_exclusive): Likewise.
5685 (aarch64_expand_compare_and_swap): Likewise.
5686 (aarch64_gen_atomic_cas): Likewise.
5687 (aarch64_emit_bic): Likewise.
5688 (aarch64_emit_atomic_swap): Likewise.
5689 (aarch64_emit_atomic_load_op): Likewise.
5690 (aarch64_evpc_trn): Likewise.
5691 (aarch64_evpc_uzp): Likewise.
5692 (aarch64_evpc_zip): Likewise.
5693 (aarch64_evpc_ext): Likewise.
5694 (aarch64_evpc_rev): Likewise.
5695 (aarch64_evpc_dup): Likewise.
5696 (aarch64_gen_ccmp_first): Likewise.
5697 (aarch64_gen_ccmp_next): Likewise.
5698 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
5699 (alpha_emit_xfloating_libcall): Likewise.
5700 (emit_insxl): Likewise.
5701 (alpha_arg_type): Likewise.
5702 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
5703 (arc_preferred_simd_mode): Likewise.
5704 (arc_secondary_reload): Likewise.
5705 (get_arc_condition_code): Likewise.
5706 (arc_print_operand): Likewise.
5707 (arc_legitimate_constant_p): Likewise.
5708 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
5709 * config/arc/arc.md (casesi_load): Likewise.
5710 (casesi_compact_jump): Likewise.
5711 * config/arc/predicates.md (proper_comparison_operator): Likewise.
5712 (cc_use_register): Likewise.
5713 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
5714 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
5715 (arm_init_iwmmxt_builtins): Likewise.
5716 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
5717 (neon_expand_vector_init): Likewise.
5718 (arm_attr_length_move_neon): Likewise.
5719 (maybe_get_arm_condition_code): Likewise.
5720 (arm_emit_vector_const): Likewise.
5721 (arm_preferred_simd_mode): Likewise.
5722 (arm_output_iwmmxt_tinsr): Likewise.
5723 (thumb1_output_casesi): Likewise.
5724 (thumb2_output_casesi): Likewise.
5725 (arm_emit_load_exclusive): Likewise.
5726 (arm_emit_store_exclusive): Likewise.
5727 (arm_expand_compare_and_swap): Likewise.
5728 (arm_evpc_neon_vuzp): Likewise.
5729 (arm_evpc_neon_vzip): Likewise.
5730 (arm_evpc_neon_vrev): Likewise.
5731 (arm_evpc_neon_vtrn): Likewise.
5732 (arm_evpc_neon_vext): Likewise.
5733 (arm_validize_comparison): Likewise.
5734 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
5735 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
5736 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
5737 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
5738 (c6x_preferred_simd_mode): Likewise.
5739 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
5740 (epiphany_rtx_costs): Likewise.
5741 * config/epiphany/predicates.md (proper_comparison_operator):
5743 * config/frv/frv.c (condexec_memory_operand): Likewise.
5744 (frv_emit_move): Likewise.
5745 (output_move_single): Likewise.
5746 (output_condmove_single): Likewise.
5747 (frv_hard_regno_mode_ok): Likewise.
5748 (frv_matching_accg_mode): Likewise.
5749 * config/h8300/h8300.c (split_adds_subs): Likewise.
5750 (h8300_rtx_costs): Likewise.
5751 (h8300_print_operand): Likewise.
5752 (compute_mov_length): Likewise.
5753 (output_logical_op): Likewise.
5754 (compute_logical_op_length): Likewise.
5755 (compute_logical_op_cc): Likewise.
5756 (h8300_shift_needs_scratch_p): Likewise.
5757 (output_a_shift): Likewise.
5758 (compute_a_shift_length): Likewise.
5759 (compute_a_shift_cc): Likewise.
5760 (expand_a_rotate): Likewise.
5761 (output_a_rotate): Likewise.
5762 * config/i386/i386.c (classify_argument): Likewise.
5763 (function_arg_advance_32): Likewise.
5764 (function_arg_32): Likewise.
5765 (function_arg_64): Likewise.
5766 (function_value_64): Likewise.
5767 (ix86_gimplify_va_arg): Likewise.
5768 (ix86_legitimate_constant_p): Likewise.
5769 (put_condition_code): Likewise.
5770 (split_double_mode): Likewise.
5771 (ix86_avx256_split_vector_move_misalign): Likewise.
5772 (ix86_expand_vector_logical_operator): Likewise.
5773 (ix86_split_idivmod): Likewise.
5774 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
5775 (ix86_build_const_vector): Likewise.
5776 (ix86_build_signbit_mask): Likewise.
5777 (ix86_match_ccmode): Likewise.
5778 (ix86_cc_modes_compatible): Likewise.
5779 (ix86_expand_branch): Likewise.
5780 (ix86_expand_sse_cmp): Likewise.
5781 (ix86_expand_sse_movcc): Likewise.
5782 (ix86_expand_int_sse_cmp): Likewise.
5783 (ix86_expand_vec_perm_vpermi2): Likewise.
5784 (ix86_expand_vec_perm): Likewise.
5785 (ix86_expand_sse_unpack): Likewise.
5786 (ix86_expand_int_addcc): Likewise.
5787 (ix86_split_to_parts): Likewise.
5788 (ix86_vectorize_builtin_gather): Likewise.
5789 (ix86_vectorize_builtin_scatter): Likewise.
5790 (avx_vpermilp_parallel): Likewise.
5791 (inline_memory_move_cost): Likewise.
5792 (ix86_tieable_integer_mode_p): Likewise.
5793 (x86_maybe_negate_const_int): Likewise.
5794 (ix86_expand_vector_init_duplicate): Likewise.
5795 (ix86_expand_vector_init_one_nonzero): Likewise.
5796 (ix86_expand_vector_init_one_var): Likewise.
5797 (ix86_expand_vector_init_concat): Likewise.
5798 (ix86_expand_vector_init_interleave): Likewise.
5799 (ix86_expand_vector_init_general): Likewise.
5800 (ix86_expand_vector_set): Likewise.
5801 (ix86_expand_vector_extract): Likewise.
5802 (emit_reduc_half): Likewise.
5803 (ix86_emit_i387_round): Likewise.
5804 (ix86_mangle_type): Likewise.
5805 (ix86_expand_round_sse4): Likewise.
5806 (expand_vec_perm_blend): Likewise.
5807 (canonicalize_vector_int_perm): Likewise.
5808 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
5809 (expand_vec_perm_1): Likewise.
5810 (expand_vec_perm_interleave3): Likewise.
5811 (expand_vec_perm_even_odd_pack): Likewise.
5812 (expand_vec_perm_even_odd_1): Likewise.
5813 (expand_vec_perm_broadcast_1): Likewise.
5814 (ix86_vectorize_vec_perm_const_ok): Likewise.
5815 (ix86_expand_vecop_qihi): Likewise.
5816 (ix86_expand_mul_widen_hilo): Likewise.
5817 (ix86_expand_sse2_abs): Likewise.
5818 (ix86_expand_pextr): Likewise.
5819 (ix86_expand_pinsr): Likewise.
5820 (ix86_preferred_simd_mode): Likewise.
5821 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
5822 * config/i386/sse.md (*andnot<mode>3): Likewise.
5823 (<mask_codefor><code><mode>3<mask_name>): Likewise.
5824 (*<code><mode>3): Likewise.
5825 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
5826 (ia64_expand_atomic_op): Likewise.
5827 (ia64_arg_type): Likewise.
5828 (ia64_mode_to_int): Likewise.
5829 (ia64_scalar_mode_supported_p): Likewise.
5830 (ia64_vector_mode_supported_p): Likewise.
5831 (expand_vec_perm_broadcast): Likewise.
5832 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
5833 (iq2000_function_arg_advance): Likewise.
5834 (iq2000_function_arg): Likewise.
5835 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
5836 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
5837 (m68k_libcall_value): Likewise.
5838 (m68k_function_value): Likewise.
5839 (sched_attr_op_type): Likewise.
5840 * config/mcore/mcore.c (mcore_output_move): Likewise.
5841 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
5843 (microblaze_function_arg): Likewise.
5844 * config/mips/mips.c (mips16_build_call_stub): Likewise.
5845 (mips_print_operand): Likewise.
5846 (mips_mode_ok_for_mov_fmt_p): Likewise.
5847 (mips_vector_mode_supported_p): Likewise.
5848 (mips_preferred_simd_mode): Likewise.
5849 (mips_expand_vpc_loongson_even_odd): Likewise.
5850 (mips_expand_vec_unpack): Likewise.
5851 (mips_expand_vi_broadcast): Likewise.
5852 (mips_expand_vector_init): Likewise.
5853 (mips_expand_vec_reduc): Likewise.
5854 (mips_expand_msa_cmp): Likewise.
5855 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
5856 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
5857 (cc_flags_for_mode): Likewise.
5858 * config/msp430/msp430.c (msp430_print_operand): Likewise.
5859 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
5860 (nds32_output_casesi_pc_relative): Likewise.
5861 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
5862 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
5863 (nvptx_gen_unpack): Likewise.
5864 (nvptx_gen_pack): Likewise.
5865 (nvptx_gen_shuffle): Likewise.
5866 (nvptx_gen_wcast): Likewise.
5867 (nvptx_preferred_simd_mode): Likewise.
5868 * config/pa/pa.c (pa_secondary_reload): Likewise.
5869 * config/pa/predicates.md (base14_operand): Likewise.
5870 * config/powerpcspe/powerpcspe-c.c
5871 (altivec_resolve_overloaded_builtin): Likewise.
5872 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
5874 (rs6000_preferred_simd_mode): Likewise.
5875 (output_vec_const_move): Likewise.
5876 (rs6000_expand_vector_extract): Likewise.
5877 (rs6000_split_vec_extract_var): Likewise.
5878 (reg_offset_addressing_ok_p): Likewise.
5879 (rs6000_legitimate_offset_address_p): Likewise.
5880 (rs6000_legitimize_address): Likewise.
5881 (rs6000_emit_set_const): Likewise.
5882 (rs6000_const_vec): Likewise.
5883 (rs6000_emit_move): Likewise.
5884 (spe_build_register_parallel): Likewise.
5885 (rs6000_darwin64_record_arg_recurse): Likewise.
5886 (swap_selector_for_mode): Likewise.
5887 (spe_init_builtins): Likewise.
5888 (paired_init_builtins): Likewise.
5889 (altivec_init_builtins): Likewise.
5890 (do_load_for_compare): Likewise.
5891 (rs6000_generate_compare): Likewise.
5892 (rs6000_expand_float128_convert): Likewise.
5893 (emit_load_locked): Likewise.
5894 (emit_store_conditional): Likewise.
5895 (rs6000_output_function_epilogue): Likewise.
5896 (rs6000_handle_altivec_attribute): Likewise.
5897 (rs6000_function_value): Likewise.
5898 (emit_fusion_gpr_load): Likewise.
5899 (emit_fusion_p9_load): Likewise.
5900 (emit_fusion_p9_store): Likewise.
5901 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
5902 (fusion_gpr_mem_load): Likewise.
5903 (fusion_addis_mem_combo_load): Likewise.
5904 (fusion_addis_mem_combo_store): Likewise.
5905 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
5906 (fusion_gpr_mem_load): Likewise.
5907 (fusion_addis_mem_combo_load): Likewise.
5908 (fusion_addis_mem_combo_store): Likewise.
5909 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5911 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
5912 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
5913 (rs6000_preferred_simd_mode): Likewise.
5914 (output_vec_const_move): Likewise.
5915 (rs6000_expand_vector_extract): Likewise.
5916 (rs6000_split_vec_extract_var): Likewise.
5917 (reg_offset_addressing_ok_p): Likewise.
5918 (rs6000_legitimate_offset_address_p): Likewise.
5919 (rs6000_legitimize_address): Likewise.
5920 (rs6000_emit_set_const): Likewise.
5921 (rs6000_const_vec): Likewise.
5922 (rs6000_emit_move): Likewise.
5923 (rs6000_darwin64_record_arg_recurse): Likewise.
5924 (swap_selector_for_mode): Likewise.
5925 (paired_init_builtins): Likewise.
5926 (altivec_init_builtins): Likewise.
5927 (rs6000_expand_float128_convert): Likewise.
5928 (emit_load_locked): Likewise.
5929 (emit_store_conditional): Likewise.
5930 (rs6000_output_function_epilogue): Likewise.
5931 (rs6000_handle_altivec_attribute): Likewise.
5932 (rs6000_function_value): Likewise.
5933 (emit_fusion_gpr_load): Likewise.
5934 (emit_fusion_p9_load): Likewise.
5935 (emit_fusion_p9_store): Likewise.
5936 * config/rx/rx.c (rx_gen_move_template): Likewise.
5937 (flags_from_mode): Likewise.
5938 * config/s390/predicates.md (s390_alc_comparison): Likewise.
5939 (s390_slb_comparison): Likewise.
5940 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
5941 (s390_vector_mode_supported_p): Likewise.
5942 (s390_cc_modes_compatible): Likewise.
5943 (s390_match_ccmode_set): Likewise.
5944 (s390_canonicalize_comparison): Likewise.
5945 (s390_emit_compare_and_swap): Likewise.
5946 (s390_branch_condition_mask): Likewise.
5947 (s390_rtx_costs): Likewise.
5948 (s390_secondary_reload): Likewise.
5949 (__SECONDARY_RELOAD_CASE): Likewise.
5950 (s390_expand_cs): Likewise.
5951 (s390_preferred_simd_mode): Likewise.
5952 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
5953 * config/sh/sh.c (sh_print_operand): Likewise.
5954 (dump_table): Likewise.
5955 (sh_secondary_reload): Likewise.
5956 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
5957 * config/sh/sh.md (casesi_worker_1): Likewise.
5958 (casesi_worker_2): Likewise.
5959 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
5960 (fcc_comparison_operator): Likewise.
5961 * config/sparc/sparc.c (sparc_expand_move): Likewise.
5962 (emit_soft_tfmode_cvt): Likewise.
5963 (sparc_preferred_simd_mode): Likewise.
5964 (output_cbranch): Likewise.
5965 (sparc_print_operand): Likewise.
5966 (sparc_expand_vec_perm_bmask): Likewise.
5967 (vector_init_bshuffle): Likewise.
5968 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
5969 (spu_vector_mode_supported_p): Likewise.
5970 (spu_expand_insv): Likewise.
5971 (spu_emit_branch_or_set): Likewise.
5972 (spu_handle_vector_attribute): Likewise.
5973 (spu_builtin_splats): Likewise.
5974 (spu_builtin_extract): Likewise.
5975 (spu_builtin_promote): Likewise.
5976 (spu_expand_sign_extend): Likewise.
5977 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
5978 (tilegx_simd_int): Likewise.
5979 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
5980 (tilepro_simd_int): Likewise.
5981 * config/v850/v850.c (const_double_split): Likewise.
5982 (v850_print_operand): Likewise.
5983 (ep_memory_offset): Likewise.
5984 * config/vax/vax.c (vax_rtx_costs): Likewise.
5985 (vax_output_int_move): Likewise.
5986 (vax_output_int_add): Likewise.
5987 (vax_output_int_subtract): Likewise.
5988 * config/visium/predicates.md (visium_branch_operator): Likewise.
5989 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
5990 (visium_print_operand_address): Likewise.
5991 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
5992 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
5993 (xtensa_expand_conditional_branch): Likewise.
5994 (xtensa_copy_incoming_a7): Likewise.
5995 (xtensa_output_literal): Likewise.
5996 * dfp.c (decimal_real_maxval): Likewise.
5997 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
5999 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6000 Alan Hayward <alan.hayward@arm.com>
6001 David Sherwood <david.sherwood@arm.com>
6003 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
6004 (mode_nunits_inline): Likewise.
6005 (mode_inner_inline): Likewise.
6006 (mode_unit_size_inline): Likewise.
6007 (mode_unit_precision_inline): Likewise.
6008 (emit_insn_modes_h): Likewise. Also emit a #define of the
6010 (emit_mode_wider): Add an E_ prefix to mode names.
6011 (emit_mode_complex): Likewise.
6012 (emit_mode_inner): Likewise.
6013 (emit_mode_adjustments): Likewise.
6014 (emit_mode_int_n): Likewise.
6015 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
6016 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
6017 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
6018 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
6019 (CRC32_BUILTIN, ENTRY): Likewise.
6020 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
6021 (aarch64_pop_regs): Likewise.
6022 (aarch64_process_components): Likewise.
6023 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
6024 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
6025 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
6026 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
6027 * config/arm/arm.c (arm_init_libfuncs): Likewise.
6028 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
6030 * config/i386/i386-builtin.def (pcmpestr): Likewise.
6031 (pcmpistr): Likewise.
6032 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
6033 * config/mmix/mmix.c (mmix_output_condition): Likewise.
6034 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
6036 * config/rl78/rl78.c (mduc_regs): Likewise.
6037 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
6038 (htm_expand_builtin): Likewise.
6039 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
6040 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
6041 * config/xtensa/xtensa.c (print_operand): Likewise.
6042 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
6043 (NUM_MODE_VECTOR_INT): Likewise.
6044 * genoutput.c (null_operand): Likewise.
6045 (output_operand_data): Likewise.
6046 * genrecog.c (print_parameter_value): Likewise.
6047 * lra.c (debug_operand_data): Likewise.
6049 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6051 * dwarf2out.c (scompare_loc_descriptor_wide)
6052 (scompare_loc_descriptor_narrow): New functions, split out from...
6053 (scompare_loc_descriptor): ...here.
6054 * expmed.c (emit_store_flag_int): New function, split out from...
6055 (emit_store_flag): ...here.
6057 2017-08-30 Richard Biener <rguenther@suse.de>
6059 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
6060 (dwarf2out_early_finish): Move setting of AT_pubnames from
6061 early debug output to early finish.
6063 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
6065 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
6066 and -mdata-region to the assembler.
6068 2017-08-30 Richard Biener <rguenther@suse.de>
6070 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
6072 (gen_subprogram_die): Add DW_AT_object_pointer only early.
6073 (dwarf2out_early_global_decl): Only generate a DIE for the
6074 abstract origin if it doesn't already exist or is a declaration DIE.
6075 (resolve_addr): Do not add the linkage name twice when
6076 generating a stub DIE for the DW_TAG_GNU_call_site target.
6078 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6080 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6081 Use machine_mode rather than int for arg1_mode.
6083 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
6086 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
6087 that the second argument of the built-in functions to unpack
6088 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
6089 switch statement instead a lot of if statements.
6090 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
6091 Allow 64-bit values to be in Altivec registers as well as
6092 traditional floating point registers.
6093 (pack<mode>, FMOVE128_VSX iterator): Likewise.
6095 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
6097 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
6098 MAX_REGS_PER_ADDRESS == 1.
6100 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
6102 * config/i386/i386.opt (flag_fentry): Do not init to -1.
6103 * config/i386/i386.c (ix86_option_override_internal): Simplify
6104 setting of opts->x_flag_entry.
6106 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6107 Jakub Jelinek <jakub@redhat.com>
6108 Richard Biener <rguenther@suse.de>
6110 PR tree-optimization/81503
6111 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
6112 folded constant fits in the target type; reorder tests for clarity.
6114 2017-08-29 Martin Liska <mliska@suse.cz>
6116 * passes.def: Include pass_lower_switch.
6117 * stmt.c (dump_case_nodes): Remove and move to
6118 tree-switch-conversion.
6119 (case_values_threshold): Likewise.
6120 (expand_switch_as_decision_tree_p): Likewise.
6121 (emit_case_decision_tree): Likewise.
6122 (expand_case): Likewise.
6123 (balance_case_nodes): Likewise.
6124 (node_has_low_bound): Likewise.
6125 (node_has_high_bound): Likewise.
6126 (node_is_bounded): Likewise.
6127 (emit_case_nodes): Likewise.
6128 (struct simple_case_node): New struct.
6129 (add_case_node): Remove.
6130 (emit_case_dispatch_table): Use vector instead of case_list.
6131 (reset_out_edges_aux): Remove.
6132 (compute_cases_per_edge): Likewise.
6133 (expand_case): Build list of simple_case_node.
6134 (expand_sjlj_dispatch_table): Use it.
6135 * tree-switch-conversion.c (struct case_node): Moved from
6136 stmt.c and adjusted.
6137 (emit_case_nodes): Likewise.
6138 (node_has_low_bound): Likewise.
6139 (node_has_high_bound): Likewise.
6140 (node_is_bounded): Likewise.
6141 (case_values_threshold): Likewise.
6142 (reset_out_edges_aux): Likewise.
6143 (compute_cases_per_edge): Likewise.
6144 (add_case_node): Likewise.
6145 (dump_case_nodes): Likewise.
6146 (balance_case_nodes): Likewise.
6147 (expand_switch_as_decision_tree_p): Likewise.
6148 (emit_jump): Likewise.
6149 (emit_case_decision_tree): Likewise.
6150 (try_switch_expansion): Likewise.
6151 (do_jump_if_equal): Likewise.
6152 (emit_cmp_and_jump_insns): Likewise.
6153 (fix_phi_operands_for_edge): New function.
6154 (record_phi_operand_mapping): Likewise.
6155 (class pass_lower_switch): New pass.
6156 (pass_lower_switch::execute): New function.
6157 (make_pass_lower_switch): Likewise.
6158 (conditional_probability):
6159 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
6160 * tree-pass.h: Add make_pass_lower_switch.
6162 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
6165 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
6168 2017-08-29 Richard Biener <rguenther@suse.de>
6170 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
6171 we do not add a DW_AT_inline attribute twice.
6172 (gen_subprogram_die): Remove code setting DW_AT_inline on
6173 DECL_ABSTRACT_P nodes.
6175 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
6177 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
6178 calls to internal functions.
6179 (gimplify_modify_expr): Likewise.
6180 * tree-call-cdce.c (use_internal_fn): Likewise.
6181 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
6182 (convert_to_divmod): Set the nothrow flag.
6183 * tree-if-conv.c (predicate_mem_writes): Likewise.
6184 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
6185 (vectorizable_call): Likewise.
6186 (vectorizable_store): Likewise.
6187 (vectorizable_load): Likewise.
6188 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
6189 (vect_recog_mask_conversion_pattern): Likewise.
6191 2017-08-29 Martin Liska <mliska@suse.cz>
6194 * gcc.c (driver_handle_option): Add new argument.
6195 * opts-common.c (handle_option): Pass
6196 target_option_override_hook.
6197 * opts-global.c (lang_handle_option): Add new option.
6198 (set_default_handlers): Add new argument.
6199 (decode_options): Likewise.
6200 * opts.c (target_handle_option): Likewise.
6201 (common_handle_option): Call target_option_override_hook.
6202 * opts.h (struct cl_option_handler_func): Add hook for
6203 target option override.
6204 (struct cl_option_handlers): Likewise.
6205 (set_default_handlers): Add new argument.
6206 (decode_options): Likewise.
6207 (common_handle_option): Likewise.
6208 (target_handle_option): Likewise.
6209 * toplev.c (toplev::main): Pass targetm.target_option.override
6212 2017-08-29 Richard Biener <rguenther@suse.de>
6213 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
6215 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
6216 life to the active subtree.
6218 2017-08-28 Jeff Law <law@redhat.com>
6220 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
6221 derive_equivalences.
6222 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
6224 (edge_info::derive_equivalences): New private member function
6226 * tree-ssa-dom.c (class edge_info): Changed from a struct
6227 to a class. Add ctor/dtor, methods and data members.
6228 (edge_info::edge_info): Renamed from allocate_edge_info.
6229 Initialize additional members.
6230 (edge_info::~edge_info): New.
6231 (free_dom_edge_info): Delete the edge info.
6232 (record_edge_info): Use new class & associated member functions.
6233 Tighten forms for testing for edge equivalences.
6234 (record_temporary_equivalences): Iterate over the simple
6235 equivalences rather than assuming there's only one per edge.
6236 (cprop_into_successor_phis): Iterate over the simple
6237 equivalences rather than assuming there's only one per edge.
6238 (optimize_stmt): Use operand_equal_p rather than pointer
6239 equality for mini-DSE code.
6241 2017-08-28 Nathan Sidwell <nathan@acm.org>
6243 * gcc.c (execute): Fold SIGPIPE handling into switch
6244 statement. Adjust internal error message.
6246 2017-08-28 Richard Biener <rguenther@suse.de>
6249 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
6250 Do nothing for removed DIEs.
6252 2017-08-28 Richard Biener <rguenther@suse.de>
6254 PR tree-optimization/81977
6255 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
6258 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
6261 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
6262 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
6263 using targetm.gen_mem_thread_fence.
6265 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
6268 * config/i386/i386.md (*<btsc><mode>): Change operand 2
6269 predicate to register_operand. Reorder operands.
6270 (*btr<mode>): Ditto.
6271 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
6272 (*btr<mode>_mask): Ditto.
6274 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
6276 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
6277 * config/rs6000/xmmintrin.h: New file.
6278 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
6280 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6283 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
6284 parameter and_insn and return it.
6285 (recombine_lvx_pattern): Insert a copy to ensure availability of
6286 the base register of the copied masking operation at the point of
6287 the instruction replacement.
6288 (recombine_stvx_pattern): Likewise.
6290 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
6292 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
6293 undocumented switches.
6294 (-mpower9-dform-vector): Likewise.
6295 (-mpower9-dform): Likewise.
6296 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
6297 comments to delete references to -mpower9-dform* switches.
6298 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
6299 Delete reference to -mpower9-dform* switches, test for
6300 -mpower9-vector instead.
6301 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
6302 (OTHER_P9_VECTOR_MASKS): Likewise.
6303 (POWERPC_MASKS): Likewise.
6304 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
6305 tests against -mpower9-dform* to -mpower9-vector. Delete code
6306 that checked for -mpower9-dform* consistancy with other options.
6307 Add test for -mpower9-misc to enable other power9 switches.
6308 (rs6000_init_hard_regno_mode_ok): Likewise.
6309 (rs6000_option_override_internal): Likewise.
6310 (rs6000_emit_prologue): Likewise.
6311 (rs6000_emit_epilogue): Likewise.
6312 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
6313 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
6314 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
6316 (emit_fusion_p9_store): Likewise.
6317 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
6318 resetting these macros if the assembler does not support ISA 3.0
6320 (TARGET_P9_DFORM_VECTOR): Likewise.
6321 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
6322 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
6324 2017-08-25 Alan Modra <amodra@gmail.com>
6326 PR rtl-optimization/81747
6327 * cse.c (cse_extended_basic_block): Don't attempt to record
6328 equivalences for degenerate conditional branches that branch
6329 to their fall-through.
6331 2017-08-24 Martin Sebor <msebor@redhat.com>
6334 * gimple-fold.c (size_must_be_zero_p): New function.
6335 (gimple_fold_builtin_memory_op): Call it.
6337 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
6339 * config/rs6000/mm_malloc.h: New file.
6341 2017-08-24 Bin Cheng <bin.cheng@arm.com>
6343 PR tree-optimization/81913
6344 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
6345 analysis when either IVs in condition can wrap.
6347 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
6349 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
6350 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
6352 2017-08-24 Richard Biener <rguenther@suse.de>
6355 * targhooks.c (default_target_can_inline_p): Properly
6356 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
6357 is present and always compare.
6358 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
6359 infer -mfpmath=sse from TARGET_SSE_P.
6360 (ix86_can_inline_p): Properly use target_option_default_node when
6361 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
6363 2017-08-24 Richard Biener <rguenther@suse.de>
6366 * dwarf2out.c (output_die): Handle flag_generate_offload like
6368 (output_comp_unit): Likewise.
6369 (gen_array_type_die): Likewise.
6370 (dwarf2out_early_finish): Likewise.
6371 (note_variable_value_in_expr): Likewise.
6372 (dwarf2out_finish): Likewise. Adjust assert.
6373 * cgraphunit.c (symbol_table::compile): Move setting of
6374 flag_generate_offload earlier ...
6375 (symbol_table::finalize_compilation_unit): ... here, before
6376 early debug finalization.
6378 2017-08-24 Richard Biener <rguenther@suse.de>
6380 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
6381 and ipa-fnsummary.h.
6382 (ix86_can_inline_p): When ix86_fpmath flags do not match
6383 check whether the callee uses FP math at all.
6385 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
6388 * tree-ssanames.c (get_nonzero_bits): Use element_precision
6389 instead of TYPE_PRECISION.
6391 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
6392 Alan Hayward <alan.hayward@arm.com>
6393 David Sherwood <david.sherwood@arm.com>
6395 * combine.c (make_extraction): Use subreg_offset_from_lsb.
6397 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
6399 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
6401 (ix86_frame::stack_realign_allocate): New field.
6402 (struct machine_frame_state): Modify comments.
6403 (machine_frame_state::sp_realigned_fp_end): New field.
6404 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
6406 (sp_valid_at): Add assertion to assure no attempt to access invalid
6407 offset of a realigned stack.
6408 (fp_valid_at): Likewise.
6409 (choose_baseaddr): Modify comments.
6410 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
6411 ix86_expand_prologue.
6412 (ix86_expand_prologue): Modify stack realignment and allocation.
6413 (ix86_expand_epilogue): Modify comments.
6414 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
6415 avx2_runtime, avx512f, and avx512f_runtime.
6417 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
6419 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
6420 (mstackrealign): Do not init to -1.
6421 * config/i386/i386.c (ix86_option_override_internal):
6422 Check opts_set, not opts when setting default value of
6423 opts->x_ix86_force_align_arg_pointer.
6425 2017-08-23 Richard Biener <rguenther@suse.de>
6427 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
6428 lang_hooks.decl_printable_name.
6429 * print-rtl-function.c (print_rtx_function): Likewise.
6430 * tree-pretty-print.c (dump_function_header): Likewise.
6432 2017-08-23 Richard Biener <rguenther@suse.de>
6435 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
6436 -g0 at compile-time.
6438 2017-08-23 Tamar Christina <tamar.christina@arm.com>
6441 * doc/sourcebuild.texi (Other hardware attributes):
6444 2017-08-23 Tamar Christina <tamar.christina@arm.com>
6447 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
6448 Fix single-use check.
6450 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
6452 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
6454 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
6456 * doc/install.texi: Modify to add more details on running selected
6459 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
6461 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
6462 is combined with -mabi=ms.
6463 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
6466 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6468 PR tree-optimization/81488
6469 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
6470 and cached_basis fields.
6471 (MAX_SPREAD): New constant.
6472 (alloc_cand_and_find_basis): Initialize new fields.
6473 (clear_visited): New function.
6474 (create_phi_basis_1): Rename from create_phi_basis, set visited
6475 and cached_basis fields.
6476 (create_phi_basis): New wrapper function.
6477 (phi_add_costs_1): Rename from phi_add_costs, add spread
6478 parameter, set visited field, short-circuit when limits reached.
6479 (phi_add_costs): New wrapper function.
6480 (record_phi_increments_1): Rename from record_phi_increments, set
6482 (record_phi_increments): New wrapper function.
6483 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
6484 (phi_incr_cost): New wrapper function.
6485 (all_phi_incrs_profitable_1): Rename from
6486 all_phi_incrs_profitable, set visited field.
6487 (all_phi_incrs_profitable): New wrapper function.
6489 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
6490 Alan Hayward <alan.hayward@arm.com>
6491 David Sherwood <david.sherwood@arm.com>
6493 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
6494 that takes the outer and inner modes.
6495 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
6496 comparison as the canonical test for a paradoxical subreg.
6497 * combine.c (simplify_set): Use paradoxical_subreg_p.
6498 (make_extraction): Likewise.
6499 (force_to_mode): Likewise.
6500 (rtx_equal_for_field_assignment_p): Likewise.
6501 (gen_lowpart_for_combine): Likewise.
6502 (simplify_comparison): Likewise.
6503 * cse.c (equiv_constant): Likewise.
6504 * expmed.c (store_bit_field_1): Likewise.
6505 * final.c (alter_subreg): Likewise.
6506 * fwprop.c (propagate_rtx): Likewise.
6507 (forward_propagate_subreg): Likewise.
6508 * ira-conflicts.c (ira_build_conflicts): Likewise.
6509 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
6510 * lra-constraints.c (curr_insn_transform): Likewise.
6511 (split_reg): Likewise.
6512 * lra-eliminations.c (move_plus_up): Likewise.
6513 (lra_eliminate_regs_1): Likewise.
6514 * recog.c (general_operand): Likewise.
6515 * ree.c (combine_reaching_defs): Likewise.
6516 * reload.c (push_reload): Likewise.
6517 (find_reloads): Likewise.
6518 * reload1.c (elimination_effects): Likewise.
6519 (compute_reload_subreg_offset): Likewise.
6520 (choose_reload_regs): Likewise.
6521 * rtlanal.c (subreg_lsb_1): Likewise.
6522 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6523 (simplify_subreg): Likewise.
6524 * var-tracking.c (track_loc_p): Likewise.
6525 * emit-rtl.c (byte_lowpart_offset): Likewise.
6526 (paradoxical_subreg_p): Delete out-of-line definition.
6528 2017-08-22 Jeff Law <law@redhat.com>
6530 PR tree-optimization/81741
6531 PR tree-optimization/71947
6532 * tree-ssa-dom.c: Include tree-inline.h.
6533 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
6534 equivalences if one is more expensive to compute than the other.
6535 * tree-ssa-scopedtables.h (class const_or_copies): Make
6536 record_const_or_copy_raw method private.
6537 (class avail_exprs_stack): New method simplify_binary_operation.
6538 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
6539 avail_exprs_stack::simplify_binary_operation as needed.
6540 (avail_exprs_stack::simplify_binary_operation): New function.
6542 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
6544 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
6545 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
6546 (DOT_SYMBOLS): Likewise.
6547 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
6548 (RELOCATABLE_NEEDS_FIXUP): Likewise.
6549 (RS6000_ABI_NAME): Likewise.
6550 (TARGET_CMODEL): Likewise.
6551 (TOC_SECTION_ASM_OP): Likewise.
6552 (SET_CMODEL): New macro.
6553 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
6555 2017-08-22 Richard Biener <rguenther@suse.de>
6557 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
6558 to allow for free-lang-data replacements similar to verify_type_variant.
6560 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
6561 Alan Hayward <alan.hayward@arm.com>
6562 David Sherwood <david.sherwood@arm.com>
6564 * config/aarch64/aarch64.md (casesi): Use DImode rather than
6565 VOIDmode for the LABEL_REF.
6567 2017-08-22 Richard Biener <rguenther@suse.de>
6569 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
6571 2017-08-22 Richard Biener <rguenther@suse.de>
6573 * common.opt (feliminate-dwarf2-dups): Ignore.
6574 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
6575 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
6576 same_die_p_wrap, compute_section_prefix,
6577 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
6578 (comdat_symbol_id, comdat_symbol_number): Likewise.
6579 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
6581 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
6582 (output_die): Mark unreachable path unreachable.
6583 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
6584 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
6585 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
6586 (dwarf2out_early_finish): Likewise.
6588 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
6590 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
6592 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
6595 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
6596 not VAR_P. Filter attribute warnings with OPT_Wattributes.
6597 (avr_attribute_table) <io, io_low, address>: Initialize
6598 .decl_required with true.
6600 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
6602 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
6603 undocumented debugging options.
6604 (-mvsx-scalar-double): Likewise.
6605 (-mallow-df-permute): Likewise.
6606 (-mvectorize-builtins): Likewise.
6607 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
6608 (rs6000_builtin_vectorized_function): Likewise.
6609 (rs6000_builtin_md_vectorized_function): Likewise.
6610 (rs6000_opt_vars): Likewise.
6612 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
6615 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
6616 (*btrq_imm): Rename from *btrq.
6617 (*btcq_imm): Rename from *btcq.
6618 (btsc): New code attribute.
6619 (*<btsc><mode>): New insn pattern.
6620 (*btr<mode>): Ditto.
6621 (*<btsc><mode>_mask): New insn_and_split pattern.
6622 (*btr<mode>_mask): Ditto.
6624 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
6625 Alan Hayward <alan.hayward@arm.com>
6626 David Sherwood <david.sherwood@arm.com>
6628 * function.c (pad_below): Simplify padding calculation.
6630 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
6631 Alan Hayward <alan.hayward@arm.com>
6632 David Sherwood <david.sherwood@arm.com>
6634 * target.def (function_prologue): Remove frame size argument.
6635 (function_epilogue): Likewise.
6636 * doc/tm.texi: Regenerate.
6637 * final.c (final_start_function): Update call to function_prologue.
6638 (final_end_function): Update call to function_epilogue.
6639 (default_function_pro_epilogue): Remove frame size argument.
6640 * output.h (default_function_pro_epilogue): Likewise.
6641 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
6642 (arm_output_function_prologue): Likewise.
6643 * config/frv/frv.c (frv_function_prologue): Likewise.
6644 (frv_function_epilogue): Likewise.
6645 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
6646 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
6647 (ia64_output_function_epilogue): Likewise.
6648 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
6649 (m32r_output_function_epilogue): Likewise.
6650 * config/microblaze/microblaze.c (microblaze_function_prologue)
6651 (microblaze_function_epilogue): Likewise.
6652 * config/mips/mips.c (mips_output_function_prologue): Likewise.
6653 (mips_output_function_epilogue): Likewise.
6654 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
6655 (mmix_target_asm_function_epilogue): Likewise.
6656 * config/msp430/msp430.c (msp430_start_function): Likewise.
6657 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
6658 (nds32_asm_function_epilogue): Likewise.
6659 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
6660 * config/pa/pa.c (pa_output_function_prologue): Likewise.
6661 (pa_output_function_epilogue): Likewise.
6662 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
6663 (rs6000_output_function_epilogue): Likewise.
6664 * config/rl78/rl78.c (rl78_start_function): Likewise.
6665 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
6666 (rs6000_output_function_epilogue): Likewise.
6667 * config/rx/rx.c (rx_output_function_prologue): Likewise.
6668 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
6669 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
6670 (sparc_asm_function_epilogue): Likewise.
6672 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
6674 * tree.h (type_has_mode_precision_p): New function.
6675 * convert.c (convert_to_integer_1): Use it.
6676 * expr.c (expand_expr_real_2): Likewise.
6677 (expand_expr_real_1): Likewise.
6678 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
6679 * match.pd: Likewise.
6680 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
6681 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
6682 * tree-tailcall.c (process_assignment): Likewise.
6683 * tree-vect-loop.c (vectorizable_reduction): Likewise.
6684 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
6685 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
6686 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
6687 (vectorizable_assignment): Likewise.
6688 (vectorizable_shift): Likewise.
6689 (vectorizable_operation): Likewise.
6690 * tree-vrp.c (register_edge_assert_for_2): Likewise.
6692 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
6694 * match.pd: Add pow (C, x) simplification.
6696 2017-08-21 Richard Biener <rguenther@suse.de>
6698 PR tree-optimization/81900
6699 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
6700 for blocks with abnormal predecessors.
6701 (compute_antic): Do not set visited flag prematurely.
6703 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
6706 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
6708 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
6710 * stor-layout.h (vector_type_mode): Move to...
6711 * tree.h (vector_type_mode): ...here.
6712 * stor-layout.c (vector_type_mode): Move to...
6713 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
6715 2017-08-21 Richard Biener <rguenther@suse.de>
6717 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
6718 register_external_die hooks.
6719 (debug_false_tree_charstarstar_uhwistar): Declare.
6720 (debug_nothing_tree_charstar_uhwi): Likewise.
6721 * debug.c (do_nothing_debug_hooks): Adjust.
6722 (debug_false_tree_charstarstar_uhwistar): New do nothing.
6723 (debug_nothing_tree_charstar_uhwi): Likewise.
6724 * dbxout.c (dbx_debug_hooks): Adjust.
6725 (xcoff_debug_hooks): Likewise.
6726 * sdbout.c (sdb_debug_hooks): Likewise.
6727 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
6728 * dwarf2out.c (macinfo_label_base): New global.
6729 (dwarf2out_register_external_die): New function for the
6730 register_external_die hook.
6731 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
6732 (dwarf2_debug_hooks): Use them.
6733 (dwarf2_lineno_debug_hooks): Adjust.
6734 (struct die_struct): Add with_offset flag.
6735 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
6736 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
6737 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
6738 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
6739 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
6740 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
6741 defining section names for the early LTO debug variants.
6742 (reset_indirect_string): New helper.
6743 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
6744 (print_dw_val): Add support for offsetted symbol references.
6745 (get_ultimate_context): Split out from is_cxx.
6746 (is_cxx): Use get_ultimate_context.
6747 (is_fortran): Add decl overload.
6748 (compute_comp_unit_symbol): Split out worker from
6749 compute_section_prefix.
6750 (compute_section_prefix): Call compute_comp_unit_symbol and
6751 set comdat_type_p here.
6752 (output_die): Skip DIE symbol output for the LTO added one.
6753 Handle DIE symbol references with offset.
6754 (output_comp_unit): Guard section name mangling properly.
6755 For LTO debug sections emit a symbol at the section beginning
6756 which we use to refer to its DIEs.
6757 (add_abstract_origin_attribute): For DIEs registered via
6758 dwarf2out_register_external_die directly refer to the early
6759 DIE rather than indirectly through the shadow one we created.
6760 Remove obsolete call to dwarf2out_abstract_function for
6761 non-function/block origins.
6762 (gen_array_type_die): When generating early LTO debug do
6763 not emit DW_AT_string_length.
6764 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
6765 late when in LTO. As suggested place a gcc_unreachable for
6766 the DECL_ABSTRACT_P case.
6767 (gen_subprogram_die): Avoid another specification DIE
6768 for early built declarations/definitions for the late LTO case.
6769 (gen_variable_die): Add type references for late duplicated VLA dies
6771 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
6772 we have the abstract instance already.
6773 (process_scope_var): Adjust decl DIE contexts in LTO which
6774 first puts them in limbo.
6775 (gen_decl_die): Do not generate type DIEs late apart from
6776 types for VLAs or for decls we do not yet have a DIE. Do not
6777 call dwarf2out_abstract_function late.
6778 (dwarf2out_early_global_decl): Make sure to create DIEs
6779 for abstract instances of a decl first.
6780 (dwarf2out_late_global_decl): Adjust comment.
6781 (output_macinfo_op): With multiple macro sections use
6782 macinfo_label_base to distinguish labels.
6783 (output_macinfo): Likewise. Update macinfo_label_base.
6784 Pass in the line info label.
6785 (note_variable_value_in_expr): When generating LTO resolve
6786 all variable values here by generating DIEs as needed.
6787 (init_sections_and_labels): Add early LTO debug flag parameter
6788 and generate different sections and names if set. Add generation
6789 counter for the labels so we can have multiple of them.
6790 (reset_dies): Helper to allow DIEs to be output multiple times.
6791 (dwarf2out_finish): When outputting DIEs to the fat part of an
6792 LTO object first reset DIEs.
6793 (dwarf2out_early_finish): Output early DIEs when generating LTO.
6794 (modified_type_die): Check for decl_ultimate_origin being self
6796 (gen_type_die_with_usage): Likewise.
6797 (gen_typedef_die): Allow decl_ultimate_origin being self.
6798 (set_decl_abstract_flags): Remove.
6799 (set_block_abstract_flags): Likewise.
6800 (dwarf2out_abstract_function): Treat the early generated DIEs
6801 as the abstract copy and only add DW_AT_inline and
6802 DW_AT_artificial here and call set_decl_origin_self.
6803 If the DIE has an abstract origin don't do anything.
6804 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
6805 if we have none yet (Go fails to build one, PR78628).
6806 (variably_modified_type_p): Prevent endless recursion for Ada
6807 cyclic pointer types.
6808 * lto-streamer-in.c: Include debug.h.
6809 (dref_queue): New global.
6810 (lto_read_tree_1): Stream in DIE references.
6811 (lto_input_tree): Register DIE references.
6812 (input_function): Stream DECL_DEBUG_ARGS.
6813 * lto-streamer-out.c: Include debug.h.
6814 (lto_write_tree_1): Output DIE references.
6815 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
6816 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
6817 (output_function): Stream DECL_DEBUG_ARGS.
6818 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
6819 Stream DECL_ABSTRACT_ORIGIN.
6820 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
6821 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
6822 DECL_CONTEXT for file-scope decls.
6823 * lto-streamer.h (struct dref_entry): Declare.
6824 (dref_queue): Likewise.
6825 * cfgexpand.c (pass_expand::execute): Do not call the
6826 outlining_inline_function hook here.
6827 * lto-wrapper.c (debug_obj): New global.
6828 (tool_cleanup): Unlink it if required.
6829 (debug_objcopy): New function.
6830 (run_gcc): Handle early debug sections in the IL files by
6831 extracting them to separate files, partially linkin them and
6832 feeding the result back as result to the linker.
6833 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
6834 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
6835 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
6836 sections into a separate segment.
6837 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
6839 (darwin_asm_dwarf_section): Likewise.
6840 (darwin_asm_output_dwarf_offset): Likewise.
6841 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
6843 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
6844 Alan Hayward <alan.hayward@arm.com>
6845 David Sherwood <david.sherwood@arm.com>
6847 * read-md.h (md_reader::record_potential_iterator_use): Replace
6848 pointer argument with an rtx and an index.
6849 * read-rtl.c (iterator_group::apply_iterator): Likewise.
6850 (apply_mode_iterator): Likewise.
6851 (apply_code_iterator): Likewise.
6852 (apply_int_iterator): Likewise.
6853 (apply_subst_iterator): Likewise.
6854 (record_iterator_use): Likewise.
6855 (record_attribute_use): Likewise.
6856 (md_reader::record_potential_iterator_use): Likewise. Update calls
6857 to record_iterator_use and apply_iterator.
6858 (iterator_use): Replace ptr with x and index.
6859 (attribute_use): Likewise.
6860 (apply_attribute_uses): Update calls to apply_iterator.
6861 (apply_iterators): Likewise. Update initialization of iterator_use.
6862 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
6863 and record_potential_iterator_use.
6864 (rtx_reader::read_rtx_operand): Likewise.
6866 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
6867 Alan Hayward <alan.hayward@arm.com>
6868 David Sherwood <david.sherwood@arm.com>
6870 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
6873 2017-08-21 Richard Biener <rguenther@suse.de>
6876 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
6877 at struct end conservatively when comparing common bases.
6879 2017-08-21 Richard Biener <rguenther@suse.de>
6881 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
6882 (mem_ref_in_stmt): Remove.
6883 (determine_max_movement): Use ref index to get at the reference.
6884 (invariantness_dom_walker::before_dom_children): Deal with
6885 lim data already initialized.
6886 (gather_mem_refs_stmt): Initialize lim data and record ref index.
6888 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
6890 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
6891 (TARGET_ISA_ROUND): Ditto.
6892 (TARGET_ROUND): Ditto.
6893 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
6894 * config/i386/i386.md: Ditto.
6895 * config/i386/sse.md: Ditto.
6896 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
6897 with OPTION_MASK_ISA_SSE4_1.
6899 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
6902 * doc/extend.texi (x86 Built-in Functions): Correct the name of
6903 __builtin_ia32_lzcnt_u16.
6905 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
6908 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
6909 (rs6000_set_current_function): Rewrite function to use it.
6911 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
6914 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
6915 and TYPE_WARN_IF_NOT_ALIGN.
6916 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
6917 (handle_warn_if_not_align): New.
6918 (place_union_field): Call handle_warn_if_not_align.
6919 (place_field): Call handle_warn_if_not_align.
6920 Copy TYPE_WARN_IF_NOT_ALIGN.
6921 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
6922 (layout_type): Likewise.
6923 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
6925 (tree_decl_common): Add warn_if_not_align.
6926 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
6927 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
6928 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
6929 (DECL_WARN_IF_NOT_ALIGN): Likewise.
6930 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
6931 * doc/extend.texi: Document warn_if_not_aligned attribute.
6932 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
6934 2017-08-17 Martin Liska <mliska@suse.cz>
6937 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
6938 (get_data_dependence): Use it as pointer type.
6939 (distribute_loop): Likewise.
6941 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6943 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
6944 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
6945 (p8_vmrgow_<mode>_direct): New define_insn.
6946 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
6947 handle endianness for vmrgew and vmrgow permute patterns.
6949 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
6951 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
6952 * config/rs6000/rs6000-cpus.def: Remove comment.
6953 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
6954 (POWERPC_MASKS): Likewise.
6955 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
6956 use of TARGET_VSX_TIMODE.
6957 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
6958 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
6959 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
6960 (rs6000_option_override_internal): Remove dead code.
6961 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
6962 (rs6000_legitimize_reload_address): Likewise.
6963 (rs6000_legitimate_address_p): Likewise.
6964 (rs6000_opt_masks): Delete "vsx-timode".
6965 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
6966 from function comment.
6967 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
6968 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
6969 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
6971 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
6972 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
6973 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
6976 (define_peephole2): Likewise.
6978 2017-08-17 Martin Sebor <msebor@redhat.com>
6981 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
6982 past the end of an array.
6983 (test_pp_format): Add test cases.
6985 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
6987 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
6988 missing ECF_NOTHROW flags.
6990 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
6993 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
6994 operands residing in integer registers.
6995 (*vsx_le_perm_load_<mode>): Likewise.
6996 (*vsx_le_perm_store_<mode>): Likewise.
6997 (define_peephole2): Add peepholes to optimize the above.
6999 2017-08-17 Marek Polacek <polacek@redhat.com>
7002 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
7003 to mimic what shorten_compare did. Change the return type to bool.
7004 (fold_cond_expr_with_comparison): Update call to
7005 operand_equal_for_comparison_p.
7006 (fold_ternary_loc): Likewise.
7008 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
7010 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
7012 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
7014 2017-08-17 Richard Biener <rguenther@suse.de>
7016 * tree-ssa-structalias.c (solve_graph): When propagating
7017 to successors update the graphs succ edges and avoid duplicate work.
7019 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
7022 * config/i386/i386.c (ix86_option_override_internal): Save target
7023 specific options after ix86_stack_protector_guard_reg was changed.
7025 2017-08-17 Richard Biener <rguenther@suse.de>
7027 PR tree-optimization/81827
7028 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
7029 (new_var_info): Initialize it conservatively.
7030 (get_call_vi): Mark register vars.
7031 (new_scalar_tmp_constraint_exp): Likewise.
7032 (handle_rhs_call): Likewise.
7033 (handle_const_call): Likewise.
7034 (create_function_info_for): Likewise.
7035 (solve_constraints): Sort varinfos to separate register from
7036 non-register vars to pack points-to solution bitmaps during
7039 2017-08-17 Marek Polacek <polacek@redhat.com>
7041 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
7043 2017-08-17 Richard Biener <rguenther@suse.de>
7045 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
7046 to true when overflow is undefined and we saturated the result.
7048 2017-08-17 Alan Modra <amodra@gmail.com>
7051 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
7052 Don't use store multiple if only one reg needs saving.
7053 (interesting_frame_related_regno): New function.
7054 (rs6000_frame_related): Don't emit frame info for regs that
7056 (rs6000_emit_epilogue): Likewise.
7058 2017-08-16 Nathan Sidwell <nathan@acm.org>
7060 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
7061 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
7062 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
7063 * tree.c (free_lang_data_in_type): Use else-if chain. Always
7064 clear TYPE_LANG_1. Remove obsolete member-function stripping.
7065 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
7066 (verify_type): Adjust for TYPE_BINFO move.
7067 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
7068 process TYPE_BINFO directly.
7069 (hash_tree): Likewise.
7070 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
7072 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
7075 2017-08-16 David Malcolm <dmalcolm@redhat.com>
7077 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
7079 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
7082 * config/i386/i386.md (*anddi_1_btr): Change predicates of
7083 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
7084 Add ix86_binary_operator_ok to insn constraint.
7085 (*iordi_1_bts): Ditto.
7086 (*xordi_1_btc): Ditto.
7087 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
7088 Update corresponding peephole2 pattern.
7092 2017-08-16 Bin Cheng <bin.cheng@arm.com>
7094 PR tree-optimization/81832
7095 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
7096 copy loop header which has IFN_LOOP_DIST_ALIAS call.
7098 2017-08-16 Marek Polacek <polacek@redhat.com>
7101 * fold-const.c (fold_indirect_ref_1): Restore original behavior
7102 regarding size_zero_node.
7104 2017-08-16 Martin Liska <mliska@suse.cz>
7107 * config.gcc: Respect previously set extra_objs in case
7110 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
7112 PR tree-optimization/81835
7113 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
7114 the phi SSA_NAME. Check that the condition in a COND_EXPR does
7115 not depend on the phi.
7117 2017-08-16 Alan Modra <amodra@gmail.com>
7119 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
7122 2017-08-16 Alan Modra <amodra@gmail.com>
7124 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
7125 (save_reg_p): ..into this. Update all callers.
7126 (first_reg_to_save): Simplify.
7128 2017-08-16 Alan Modra <amodra@gmail.com>
7130 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
7133 2017-08-15 Joseph Myers <joseph@codesourcery.com>
7137 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
7138 constant count if that count is less than 32.
7140 2017-08-15 Nathan Sidwell <nathan@acm.org>
7142 * gcc.c (execute): Emit friendlier message if inferior is killed
7143 by an external cause.
7145 2017-08-15 Richard Biener <rguenther@suse.de>
7147 PR tree-optimization/81790
7148 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
7149 CONSTRUCTORs from simplifying and VN.
7151 2017-08-14 Martin Sebor <msebor@redhat.com>
7153 * builtin-attrs.def: Add comments.
7155 2017-08-14 Martin Sebor <msebor@redhat.com>
7158 * doc/extend.texi (attribute nonstring): Document new attribute.
7160 2017-08-14 Martin Sebor <msebor@redhat.com>
7163 * tree-diagnostic.c (default_tree_printer): Handle %G.
7164 * gimple-pretty-print.h (percent_G_format): Declare new function.
7165 * gimple-pretty-print.c (percent_G_format): Define.
7166 * tree-pretty-print.c (percent_K_format): Add argument.
7168 2017-08-14 Martin Sebor <msebor@redhat.com>
7170 PR translation/79998
7171 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
7172 Remove a stray space.
7174 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
7177 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
7178 (*iordi_1_bts): Ditto.
7179 (*xordi_1_btc): Ditto.
7181 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7184 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
7185 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7187 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
7188 quoted strings, and make more translator-friendly.
7189 (darwin_rs6000_override_options): Likewise.
7190 (rs6000_option_override_internal): Likewise.
7191 (rs6000_return_in_memory): Fix overlong line.
7192 (init_cmulative_args): Use quoted strings, and make more
7193 translator-friendly.
7194 (rs6000_pass_by_reference): Fix overlong line.
7195 (def_builtin): Use quoted strings.
7196 (altivec_expand_predicate_builtin): Use quoted strings, and make
7197 more translator-friendly.
7198 (htm_expand_builtin): Use quoted strings.
7199 (cpu_expand_builtin): Use quoted strings, and make more
7200 translator-friendly.
7201 (altivec_expand_builtin): Likewise.
7202 (paired_expand_predicate_builtin): Likewise.
7203 (rs6000_invalid_builtin): Likewise.
7204 (builtin_function_type): Use quoted strings.
7205 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
7206 more translator-friendly.
7207 (rs6000_trampoline_init): Likewise.
7208 (rs6000_handle_altivec_attribute): Likewise.
7209 (rs6000_inner_target_options): Use quoted strings.
7210 (rs6000_disable_incompatible_switches): Likewise.
7211 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
7212 strings, and make more translator-friendly.
7213 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
7215 2017-08-14 Bin Cheng <bin.cheng@arm.com>
7217 PR tree-optimization/81799
7218 * tree-loop-distribution.c (version_loop_by_alias_check): Force
7219 cond_expr to simple gimple operand.
7221 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
7224 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
7226 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
7230 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
7231 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
7232 TARGET_GASISR_PROLOGUES.
7233 * config/avr/avr.c (avr_option_override): Same.
7234 (avr_pass_pre_proep::execute): Same.
7236 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
7239 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
7240 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
7242 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
7244 * config/i386/i386.md (*load_tp_<mode>): Redefine as
7245 define_insn_and_split. Split to a memory load from 0 in
7246 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
7247 using PTR mode iterator.
7248 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
7249 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
7250 (*add_tp_<mode>): Redefine as define_insn_and_split.
7251 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
7252 address space. Merge with *add_tp_x32 using PTR mode iterator.
7253 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
7254 Split to an add with a memory load from 0 in
7255 DEFAULT_TLS_SEG_REG address space.
7257 2017-08-12 Andrew Pinski <apinski@cavium.com>
7259 * config/aarch64/aarch64-option-extensions.def (rdma):
7260 Fix feature string to what Linux prints out in /proc/cpuinfo.
7262 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
7265 * dwarf2out.c (modified_type_die): For C typedef types that have
7266 an ultimate origin, process the ultimate origin instead of the
7268 (gen_typedef_die): Assert that input DECLs have no ultimate
7270 (gen_type_die_with_usage): For typedef variants that have an
7271 ultimate origin, just call gen_decl_die on the original DECL.
7272 (process_scope_var): Avoid creating DIEs for local typedefs and
7273 concrete static variables.
7275 2017-08-12 Alan Modra <amodra@gmail.com>
7279 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
7280 match gnu-user.h startfile.
7281 (ENDFILE_LINUX_SPEC): Similarly.
7283 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
7286 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
7288 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
7290 2017-08-11 Tamar Christina <tamar.christina@arm.com>
7291 * config/aarch64/aarch64.md (mov<mode>): Change.
7292 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
7293 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
7294 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
7296 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
7298 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
7299 for storage order barriers.
7301 2017-08-11 Martin Liska <mliska@suse.cz>
7304 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
7305 variables of void type.
7307 2017-08-11 Martin Liska <mliska@suse.cz>
7309 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
7310 TARGET_SUPPORTS_ALIASES.
7311 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
7312 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
7313 (optimize_weakref): Likewise.
7314 * symtab.c (symtab_node::noninterposable_alias): Likewise.
7315 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
7316 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
7318 2017-08-11 Martin Liska <mliska@suse.cz>
7321 * config/i386/i386.c (make_resolver_func): Do complete
7322 refactoring of the function.
7324 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
7327 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
7328 * config/i386/i386.c (ix86_stack_protect_guard): Use
7329 ix86_stack_protect_guard_symbol_str to generate varible declaration.
7330 * doc/invoke.texi (x86 Options): Document
7331 -mstack-protector-guard-symbol= option.
7333 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
7335 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
7336 * config/i386/i386.c (ix86_split_stack_guard): New function.
7337 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
7338 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
7339 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
7340 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
7341 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
7342 (split_stack_space_check): Call ix86_split_stack_guard.
7344 2017-08-10 Martin Sebor <msebor@redhat.com>
7346 * print-tree.c (print_node): Print location using the established
7348 Replace spaces with colons.
7349 (debug_raw, debug): Ditto.
7351 2017-08-10 Martin Sebor <msebor@redhat.com>
7354 * pretty-print.c (pp_format): Correct the handling of %s precision.
7356 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
7359 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
7361 (ix86_finalize_stack_frame_flags): This. Also clear
7362 frame_pointer_needed if -fno-omit-frame-pointer is used without
7364 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
7365 with ix86_finalize_stack_frame_flags.
7366 (ix86_expand_epilogue): Likewise.
7367 (ix86_expand_split_stack_prologue): Likewise.
7368 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
7370 2017-08-10 Martin Liska <mliska@suse.cz>
7373 * c-attribs.c (handle_target_attribute):
7374 Report warning for an empty string argument of target attribute.
7376 2017-08-09 Jakub Jelinek <jakub@redhat.com>
7379 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
7381 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
7382 or DECL_NONLOCAL labels.
7383 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
7384 or DECL_NONLOCAL labels here.
7386 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
7388 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
7389 to indicate when early gimple folding has been disabled.
7390 (rs6000_gimple_fold_builtin): Add debug content.
7391 (rs6000_invalid_builtin): Fix whitespace.
7392 (rs6000_expand_builtin): Fix whitespace.
7393 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
7395 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
7398 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
7399 SAVE_MULTIPLE if not all the registers that saves, should be saved.
7401 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
7403 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
7404 (qdf24xx): Likewise.
7405 * config/aarch64/aarch64.md: Include falkor.md.
7406 * config/aarch64/falkor.md: New.
7408 2017-08-09 Marek Polacek <polacek@redhat.com>
7411 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
7412 * diagnostic.c (emit_diagnostic): Add a comment.
7413 (emit_diagnostic_valist): New function.
7415 2017-08-09 Marek Polacek <polacek@redhat.com>
7418 * input.c (make_location): New overload.
7419 * input.h (make_location): Declare.
7421 2017-08-08 Alan Modra <amodra@gmail.com>
7422 H.J. Lu <hongjiu.lu@intel.com>
7425 * gcc.c (NO_PIE_SPEC): Delete.
7426 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
7428 (LINK_PIE_SPEC): ..to here.
7429 (LINK_COMMAND_SPEC): Support -no-pie.
7430 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
7431 chain of crtbegin*.o selection, update for PIE_SPEC changes and
7433 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
7434 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
7435 (ENDFILE_CRTEND_SPEC): Similarly.
7437 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
7440 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
7441 (mstack-protector-guard-offset=): Ditto.
7442 * config/i386/i386.c (ix86_option_override): Handle
7443 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
7445 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
7446 ix86_stack_protect_guard_offset variables.
7447 (TARGET_STACK_PROTECT_GUARD): Always define.
7448 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
7449 and -mstack-protector-guard-offset= options.
7451 2017-08-08 Bin Cheng <bin.cheng@arm.com>
7453 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
7454 boundary case for the last candidate.
7456 2017-08-08 Bin Cheng <bin.cheng@arm.com>
7458 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
7459 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
7461 2017-08-08 Tamar Christina <tamar.christina@arm.com>
7464 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
7465 * config/aarch64/aarch64-builtins.c
7466 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
7467 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
7468 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
7470 2017-08-08 Tamar Christina <tamar.christina@arm.com>
7471 Andrew Pinski <pinskia@gmail.com>
7474 * internal-fn.def (XORSIGN): New.
7475 * optabs.def (xorsign_optab): New.
7476 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
7477 (convert_expand_mult_copysign): New.
7478 (pass_optimize_widening_mul::execute): Call
7479 convert_expand_mult_copysign.
7481 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7483 PR tree-optimization/81354
7484 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
7485 Insert on edges rather than explicitly creating landing pads.
7486 (analyze_candidates_and_replace): Commit edge inserts.
7488 2017-08-08 Richard Biener <rguenther@suse.de>
7491 * tree-ssa-loop-niter.c: Include tree-dfa.h.
7492 (expand_simple_operations): Also look through ADDR_EXPRs with
7493 MEM_REF bases treating them as POINTER_PLUS_EXPR.
7495 2017-08-08 Richard Biener <rguenther@suse.de>
7497 PR tree-optimization/81723
7498 * tree-vect-slp.c (struct bst_traits): New hash traits.
7499 (bst_fail): New global.
7500 (vect_build_slp_tree_2): New worker, split out from ...
7501 (vect_build_slp_tree): ... this now wrapping it with using
7502 bst_fail set to cache SLP tree build fails. Properly handle
7504 (vect_analyze_slp_instance): Allocate and free bst_fail.
7506 2017-08-08 Martin Liska <mliska@suse.cz>
7509 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
7510 LABEL_DECLs that can be from a different function.
7512 2017-08-08 Bin Cheng <bin.cheng@arm.com>
7514 PR tree-optimization/81744
7515 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
7516 loop's number of iterations.
7518 2017-08-08 Martin Liska <mliska@suse.cz>
7520 * asan.c: Include header files.
7521 * attribs.c (build_decl_attribute_variant): New function moved
7523 (build_type_attribute_qual_variant): Likewise.
7524 (cmp_attrib_identifiers): Likewise.
7525 (simple_cst_list_equal): Likewise.
7526 (omp_declare_simd_clauses_equal): Likewise.
7527 (attribute_value_equal): Likewise.
7528 (comp_type_attributes): Likewise.
7529 (build_type_attribute_variant): Likewise.
7530 (lookup_ident_attribute): Likewise.
7531 (remove_attribute): Likewise.
7532 (merge_attributes): Likewise.
7533 (merge_type_attributes): Likewise.
7534 (merge_decl_attributes): Likewise.
7535 (merge_dllimport_decl_attributes): Likewise.
7536 (handle_dll_attribute): Likewise.
7537 (attribute_list_equal): Likewise.
7538 (attribute_list_contained): Likewise.
7539 * attribs.h (lookup_attribute): New function moved from tree.[ch].
7540 (lookup_attribute_by_prefix): Likewise.
7541 * bb-reorder.c: Include header files.
7542 * builtins.c: Likewise.
7543 * calls.c: Likewise.
7544 * cfgexpand.c: Likewise.
7545 * cgraph.c: Likewise.
7546 * cgraphunit.c: Likewise.
7547 * convert.c: Likewise.
7548 * dwarf2out.c: Likewise.
7549 * final.c: Likewise.
7550 * fold-const.c: Likewise.
7551 * function.c: Likewise.
7552 * gimple-expr.c: Likewise.
7553 * gimple-fold.c: Likewise.
7554 * gimple-pretty-print.c: Likewise.
7555 * gimple.c: Likewise.
7556 * gimplify.c: Likewise.
7557 * hsa-common.c: Likewise.
7558 * hsa-gen.c: Likewise.
7559 * internal-fn.c: Likewise.
7560 * ipa-chkp.c: Likewise.
7561 * ipa-cp.c: Likewise.
7562 * ipa-devirt.c: Likewise.
7563 * ipa-fnsummary.c: Likewise.
7564 * ipa-inline.c: Likewise.
7565 * ipa-visibility.c: Likewise.
7567 * lto-cgraph.c: Likewise.
7568 * omp-expand.c: Likewise.
7569 * omp-general.c: Likewise.
7570 * omp-low.c: Likewise.
7571 * omp-offload.c: Likewise.
7572 * omp-simd-clone.c: Likewise.
7573 * opts-global.c: Likewise.
7574 * passes.c: Likewise.
7575 * predict.c: Likewise.
7576 * sancov.c: Likewise.
7577 * sanopt.c: Likewise.
7578 * symtab.c: Likewise.
7579 * toplev.c: Likewise.
7580 * trans-mem.c: Likewise.
7581 * tree-chkp.c: Likewise.
7582 * tree-eh.c: Likewise.
7583 * tree-into-ssa.c: Likewise.
7584 * tree-object-size.c: Likewise.
7585 * tree-parloops.c: Likewise.
7586 * tree-profile.c: Likewise.
7587 * tree-ssa-ccp.c: Likewise.
7588 * tree-ssa-live.c: Likewise.
7589 * tree-ssa-loop.c: Likewise.
7590 * tree-ssa-sccvn.c: Likewise.
7591 * tree-ssa-structalias.c: Likewise.
7592 * tree-ssa.c: Likewise.
7593 * tree-streamer-in.c: Likewise.
7594 * tree-vectorizer.c: Likewise.
7595 * tree-vrp.c: Likewise.
7597 * ubsan.c: Likewise.
7598 * varasm.c: Likewise.
7599 * varpool.c: Likewise.
7600 * tree.c: Remove functions moved to attribs.[ch].
7602 * config/aarch64/aarch64.c: Add attrs.h header file.
7603 * config/alpha/alpha.c: Likewise.
7604 * config/arc/arc.c: Likewise.
7605 * config/arm/arm.c: Likewise.
7606 * config/avr/avr.c: Likewise.
7607 * config/bfin/bfin.c: Likewise.
7608 * config/c6x/c6x.c: Likewise.
7609 * config/cr16/cr16.c: Likewise.
7610 * config/cris/cris.c: Likewise.
7611 * config/darwin.c: Likewise.
7612 * config/epiphany/epiphany.c: Likewise.
7613 * config/fr30/fr30.c: Likewise.
7614 * config/frv/frv.c: Likewise.
7615 * config/ft32/ft32.c: Likewise.
7616 * config/h8300/h8300.c: Likewise.
7617 * config/i386/winnt.c: Likewise.
7618 * config/ia64/ia64.c: Likewise.
7619 * config/iq2000/iq2000.c: Likewise.
7620 * config/lm32/lm32.c: Likewise.
7621 * config/m32c/m32c.c: Likewise.
7622 * config/m32r/m32r.c: Likewise.
7623 * config/m68k/m68k.c: Likewise.
7624 * config/mcore/mcore.c: Likewise.
7625 * config/microblaze/microblaze.c: Likewise.
7626 * config/mips/mips.c: Likewise.
7627 * config/mmix/mmix.c: Likewise.
7628 * config/mn10300/mn10300.c: Likewise.
7629 * config/moxie/moxie.c: Likewise.
7630 * config/msp430/msp430.c: Likewise.
7631 * config/nds32/nds32-isr.c: Likewise.
7632 * config/nds32/nds32.c: Likewise.
7633 * config/nios2/nios2.c: Likewise.
7634 * config/nvptx/nvptx.c: Likewise.
7635 * config/pa/pa.c: Likewise.
7636 * config/pdp11/pdp11.c: Likewise.
7637 * config/powerpcspe/powerpcspe.c: Likewise.
7638 * config/riscv/riscv.c: Likewise.
7639 * config/rl78/rl78.c: Likewise.
7640 * config/rx/rx.c: Likewise.
7641 * config/s390/s390.c: Likewise.
7642 * config/sh/sh.c: Likewise.
7643 * config/sol2.c: Likewise.
7644 * config/sparc/sparc.c: Likewise.
7645 * config/spu/spu.c: Likewise.
7646 * config/stormy16/stormy16.c: Likewise.
7647 * config/tilegx/tilegx.c: Likewise.
7648 * config/tilepro/tilepro.c: Likewise.
7649 * config/v850/v850.c: Likewise.
7650 * config/vax/vax.c: Likewise.
7651 * config/visium/visium.c: Likewise.
7652 * config/xtensa/xtensa.c: Likewise.
7654 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
7657 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
7658 constraints since the -mupper-regs-* switches have been
7660 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
7661 into a vector from a double word element that was extracted from
7662 another vector, and eliminate extra XXPERMDI instructions.
7663 (vsx_concat_<mode>_2): Likewise.
7664 (vsx_concat_<mode>_3): Likewise.
7665 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
7666 concat to allow optimizing inserts from previous extracts.
7668 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
7670 * config/i386/i386.c (ix86_stack_protect_guard): Generate
7671 memory reference to a SSP offset in TLS address space.
7672 (ix86_print_operand) <case '@'>: Remove.
7673 (ix86_print_operand_punct_valid_p): Remove '@' code.
7674 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
7676 (stack_tls_protect_set_<mode>): Remove.
7677 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
7678 (stack_tls_protect_test_<mode>): Remove.
7679 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
7681 2017-08-07 Olivier Hainque <hainque@adacore.com>
7684 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
7686 2017-08-07 Douglas Rupp <rupp@adacore.com>
7688 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
7689 variable was referenced as multidir in command.
7691 2017-08-07 Jakub Jelinek <jakub@redhat.com>
7694 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
7697 2017-08-07 Martin Liska <mliska@suse.cz>
7699 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
7700 * config/rl78/rl78.c: Add include of attribs.h.
7701 * config/sh/sh.c: Likewise.
7702 * config/v850/v850.c: Likewise.
7704 2017-08-07 Tom de Vries <tom@codesourcery.com>
7707 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
7709 2017-08-07 Martin Liska <mliska@suse.cz>
7711 * config/mips/mips.c: Include attribs.h.
7713 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
7716 * doc/invoke.texi: Document change in behvaior for -Ofast for
7719 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
7721 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
7723 (aarch64_pop_regs): Likewise.
7724 (aarch64_gen_load_pair): Likewise.
7725 (aarch64_save_callee_saves): Likewise.
7726 (aarch64_restore_callee_saves): Likewise.
7728 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
7730 * config/i386/i386.c: Revert the last change.
7732 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
7735 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
7737 (ix86_finalize_stack_frame_flags): This. Also clear
7738 frame_pointer_needed if -fno-omit-frame-pointer is used without
7740 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
7741 with ix86_finalize_stack_frame_flags.
7742 (ix86_expand_epilogue): Likewise.
7743 (ix86_expand_split_stack_prologue): Likewise.
7745 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
7748 * config/i386/i386.c (get_builtin_code_for_version): Set priority
7749 to P_AES for Westmere.
7751 2017-08-07 Jonathan Yong <10walls@gmail.com>
7753 * config/i386/mingw.opt (fset-stack-executable): Removed.
7754 * config/i386/cygming.opt (fset-stack-executable): Moved
7756 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
7758 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
7760 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
7762 2017-08-07 Marek Polacek <polacek@redhat.com>
7765 * fold-const.c (fold_indirect_ref_1): Check type_domain.
7767 2017-08-07 Martin Liska <mliska@suse.cz>
7769 * attribs.h (canonicalize_attr_name): New function.
7770 (cmp_attribs): Move from c-format.c and adjusted.
7771 (is_attribute_p): Moved from tree.h.
7772 * tree-inline.c: Add new includes.
7773 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
7774 (private_is_attribute_p): Remove.
7775 (private_lookup_attribute): Likewise.
7776 (private_lookup_attribute_by_prefix): Simplify.
7777 (remove_attribute): Use is_attribute_p.
7778 * tree.h: Remove removed declarations.
7780 2017-08-07 Jakub Jelinek <jakub@redhat.com>
7783 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
7784 instead of computing it in the function. Formatting fix.
7785 (expand_case): Don't rely on default_edge being the first edge,
7786 clear it if removing it, pass default_edge to
7787 emit_case_dispatch_table.
7788 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
7791 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
7793 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
7794 insn in the function, emit NOP after the insn.
7796 2017-08-06 Tom de Vries <tom@codesourcery.com>
7798 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
7801 2017-08-06 Tom de Vries <tom@codesourcery.com>
7803 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
7806 2017-08-04 Yury Gribov <tetra2005@gmail.com>
7808 PR tree-optimization/57371
7809 * match.pd: New pattern.
7811 2017-08-04 Marek Polacek <polacek@redhat.com>
7814 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
7815 perform the computation in offset_int.
7817 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
7819 PR tree-optimization/81136
7820 * tree-vectorizer.h: Include tree-hash-traits.h.
7821 (vec_base_alignments): New typedef.
7822 (vec_info): Add a base_alignments field.
7823 (vect_record_base_alignments): Declare.
7824 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
7826 (DR_IS_CONDITIONAL_IN_STMT): New macro.
7827 (create_data_ref): Add an is_conditional_in_stmt argument.
7828 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
7829 the is_conditional_in_stmt field.
7830 (data_ref_loc): Add an is_conditional_in_stmt field.
7831 (get_references_in_stmt): Set the is_conditional_in_stmt field.
7832 (find_data_references_in_stmt): Update call to create_data_ref.
7833 (graphite_find_data_references_in_stmt): Likewise.
7834 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
7835 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
7836 (vect_record_base_alignment): New function.
7837 (vect_record_base_alignments): Likewise.
7838 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
7839 for nested statements even if we fail to compute a misalignment.
7840 Use pooled base alignments for unconditional references.
7841 (vect_find_same_alignment_drs): Compare base addresses instead
7843 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
7844 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
7846 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
7848 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
7849 Add an explicit name for the enum. Use auto_vec for slp_instances
7851 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
7853 (_bb_vec_info): Add a constructor and destructor.
7854 (vinfo_for_stmt): Return NULL for uids of -1 as well.
7855 (destroy_loop_vec_info): Delete.
7856 (vect_destroy_datarefs): Likewise.
7857 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
7858 (vec_info::vec_info): New function.
7859 (vec_info::~vec_info): Likewise.
7860 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
7861 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
7862 destroy_loop_vec_info.
7863 * tree-vect-loop.c (new_loop_vec_info): Replace with...
7864 (_loop_vec_info::_loop_vec_info): ...this.
7865 (destroy_loop_vec_info): Replace with...
7866 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
7867 the stmt_vec_infos. Leave handling of vec_info information to its
7868 destructor. Remove explicit vector releases.
7869 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
7870 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
7871 * tree-vect-slp.c (new_bb_vec_info): Replace with...
7872 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
7873 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
7874 (destroy_bb_vec_info): Replace with...
7875 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
7876 information to its destructor.
7877 (vect_slp_analyze_bb_1): Use new and delete instead of
7878 new_bb_vec_info and destroy_bb_vec_info.
7879 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
7882 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
7884 * tree-data-ref.h (subscript): Add access_fn field.
7885 (data_dependence_relation): Add could_be_independent_p.
7886 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
7887 (same_access_functions): Move to tree-data-ref.c.
7888 * tree-data-ref.c (ref_contains_union_access_p): New function.
7889 (access_fn_component_p): Likewise.
7890 (access_fn_components_comparable_p): Likewise.
7891 (dr_analyze_indices): Add a reference to access_fn_component_p.
7892 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
7894 (constant_access_functions): Likewise.
7895 (add_other_self_distances): Likewise.
7896 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
7897 (initialize_data_dependence_relation): Use XCNEW and remove
7898 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
7899 of access functions that have the same type. Allow the
7900 subsequence to end with different bases in some circumstances.
7901 Record the chosen access functions in SUB_ACCESS_FN.
7902 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
7903 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
7904 (subscript_dependence_tester_1): Likewise dra and drb.
7905 (build_classic_dist_vector): Update calls accordingly.
7906 (subscript_dependence_tester): Likewise.
7907 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
7908 DDR_COULD_BE_INDEPENDENT_P.
7909 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
7910 comp_alias_ddrs instead of may_alias_ddrs.
7911 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
7913 (vect_analyze_data_ref_dependence): Use it if
7914 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
7915 distance vectors if that fails.
7916 (dependence_distance_ge_vf): New function.
7917 (vect_prune_runtime_alias_test_list): Use it. Don't clear
7918 LOOP_VINFO_MAY_ALIAS_DDRS.
7920 2017-08-04 Richard Biener <rguenther@suse.de>
7923 * fold-const.c (fold_binary_loc): Properly restrict
7924 minus_var0 && minus_var1 case when associating undefined overflow
7927 2017-08-04 Tom de Vries <tom@codesourcery.com>
7929 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
7931 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7933 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7934 Don't start diagnostic messages with a capital letter.
7935 * config/rs6000/rs6000.c (rs6000_option_override_internal):
7937 (rs6000_invalid_builtin): Likewise.
7938 (rs6000_trampoline_init): Likewise.
7940 2017-08-03 Jakub Jelinek <jakub@redhat.com>
7943 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
7944 after setting changeable df flags.
7946 2017-08-03 Richard Biener <rguenther@suse.de>
7948 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
7949 up if the use is in USE - X.
7951 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
7953 * toplev.c (dumpfile.h): New include.
7954 (internal_error_reentered): New static function. Use it...
7955 (internal_error_function): ...here to handle reentered internal_error.
7957 2017-08-03 Richard Biener <rguenther@suse.de>
7960 * fold-const.c (split_tree): Add minus_var and minus_con
7961 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
7962 here but always use minus_*.
7963 (associate_trees): Assert we never associate with MINUS_EXPR
7964 and NULL first operand. Do not recurse for PLUS_EXPR operands
7965 when associating as MINUS_EXPR either.
7966 (fold_binary_loc): Track minus_var and minus_con.
7968 2017-08-03 Tom de Vries <tom@codesourcery.com>
7971 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
7972 ACCEL_COMPILER, apply finish_options on
7973 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
7975 2017-08-03 Tom de Vries <tom@codesourcery.com>
7978 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
7979 function_entry_patch_area_size > 0.
7981 2017-08-03 Jakub Jelinek <jakub@redhat.com>
7984 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
7985 instead of 10??LU, perform unit multiplication in wide_int,
7986 don't change alloc_object_size_limit if the limit is larger
7989 PR tree-optimization/81655
7990 PR tree-optimization/81588
7991 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
7992 the case when ranges[i].low and high are 1 for unsigned type with
7996 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
7997 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
7999 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
8001 * tree-vrp.h: Add include guard.
8003 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
8006 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
8007 (ud2): New insn pattern.
8008 * config/i386/i386.c (ix86_expand_epilogue):
8009 For naked functions, generate ud2 instead of trap insn.
8011 2017-08-02 Marek Polacek <polacek@redhat.com>
8014 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
8016 2017-08-02 Tom de Vries <tom@codesourcery.com>
8017 Cesar Philippidis <cesar@codesourcery.com>
8019 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
8020 Add missing edge probabilities.
8022 2017-08-02 Tamar Christina <tamar.christina@arm.com>
8024 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
8027 2017-08-02 Jakub Jelinek <jakub@redhat.com>
8030 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
8031 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
8032 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
8033 prologue_seq sequences - if any.
8035 2017-08-02 Richard Biener <rguenther@suse.de>
8037 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
8038 via vectors if supported, integer extracts via punning if supported
8039 or otherwise vector extracts.
8041 2017-08-02 Richard Biener <rguenther@suse.de>
8043 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
8045 (bitmap_insert_into_set): ... this.
8047 2017-08-02 Richard Biener <rguenther@suse.de>
8049 PR tree-optimization/81633
8051 2015-08-17 Alan Hayward <alan.hayward@arm.com>
8053 PR tree-optimization/71752
8054 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
8056 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
8058 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
8059 (machine_function::call_ms2sysv_pad_out): Remove field.
8060 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
8061 (ix86_compute_frame_layout): Likewise.
8063 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
8066 * config/i386/i386.c (ix86_set_func_type): Disallow naked
8067 attribute with interrupt attribute.
8069 2017-08-01 Andrew Pinski <apinski@cavium.com>
8071 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
8072 BIT_INSERT_EXPR's operand 1
8073 to see if the types precision matches.
8075 2017-08-01 Martin Liska <mliska@suse.cz>
8078 * builtins.c (expand_builtin_memcpy_args): Remove.
8079 (expand_builtin_memcpy): Call newly added function
8080 expand_builtin_memory_copy_args.
8081 (expand_builtin_memcpy_with_bounds): Likewise.
8082 (expand_builtin_mempcpy): Remove last argument.
8083 (expand_builtin_mempcpy_with_bounds): Likewise.
8084 (expand_builtin_memory_copy_args): New function created from
8085 expand_builtin_mempcpy_args with small modifications.
8086 (expand_builtin_mempcpy_args): Remove.
8087 (expand_builtin_stpcpy): Remove unused argument.
8088 (expand_builtin): Likewise.
8089 (expand_builtin_with_bounds): Likewise.
8091 2017-08-01 Martin Liska <mliska@suse.cz>
8094 Make mempcpy more optimal (PR middle-end/70140).
8096 2017-08-01 Jakub Jelinek <jakub@redhat.com>
8099 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
8100 __builtin_vec_cmpne verify both arguments are compatible vectors
8101 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
8102 verify arg1_type is a pointer or array type. For __builtin_vec_st,
8103 move computation of aligned to after checking the argument types.
8107 * config/rs6000/vsx.md (vextract_fp_from_shorth,
8108 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
8111 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
8112 Doug Rupp <rupp@adacore.com>
8113 Olivier Hainque <hainque@adacore.com>
8115 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
8116 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
8118 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
8119 for TARGET_OS_CPP_BUILTIN.
8120 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
8121 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
8123 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
8124 passing required abi options to the assembler for EABI configurations.
8125 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
8126 of .text.hot and .text.unlikely sections for kernel modules when
8127 using ARM style exceptions.
8128 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
8129 options. Add EXTRA_CC1_SPEC.
8130 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
8132 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
8134 (ARM_TARGET2_DWARF_FORMAT): Define.
8135 * config/arm/t-vxworks: Adjust multilib control to removal of the
8136 Diab command line options.
8138 2017-08-01 Martin Liska <mliska@suse.cz>
8140 PR gcov-profile/81561
8141 * gcov.c (unblock): Make unblocking safe as we need to preserve
8142 index correspondence of blocks and block_lists.
8144 2017-08-01 Richard Biener <rguenther@suse.de>
8146 PR tree-optimization/81181
8147 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
8148 (compute_antic): ... end of iteration here.
8150 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
8152 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
8153 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
8154 (ftree-slp-vectorize): Likewise.
8155 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
8156 can no longer be set independent of flag_tree_loop_vectorize.
8157 * omp-general.c (emp_max_vf): Likewise.
8158 * opts.c (enable_fdo_optimizations): Remove references to
8159 flag_tree_vectorize, these are now implicit.
8160 (common_handle_option): Remove handling for OPT_ftree_vectorize,
8161 and leave it for the options machinery.
8163 2017-08-01 Martin Liska <mliska@suse.cz>
8166 * builtins.c (expand_builtin_memcpy_args): Remove.
8167 (expand_builtin_memcpy): Call newly added function
8168 expand_builtin_memory_copy_args.
8169 (expand_builtin_memcpy_with_bounds): Likewise.
8170 (expand_builtin_mempcpy): Remove last argument.
8171 (expand_builtin_mempcpy_with_bounds): Likewise.
8172 (expand_builtin_memory_copy_args): New function created from
8173 expand_builtin_mempcpy_args with small modifications.
8174 (expand_builtin_mempcpy_args): Remove.
8175 (expand_builtin_stpcpy): Remove unused argument.
8176 (expand_builtin): Likewise.
8177 (expand_builtin_with_bounds): Likewise.
8179 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
8182 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
8183 print "ds:" only for immediates in generic address space.
8185 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
8188 * config/i386/i386.c (ix86_funciton_naked): New prototype.
8189 (ix86_function_ok_for_sibcall): Return false for naked functions.
8191 2017-08-01 Richard Biener <rguenther@suse.de>
8193 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
8194 (compute_antic): Seed worklist with exit block predecessors.
8195 * cfganal.c (dfs_find_deadend): For a cycle return the source
8196 of the edge closing it.
8198 2017-08-01 Tamar Christina <tamar.christina@arm.com>
8200 * config/aarch64/aarch64.c
8201 (aarch64_can_const_movi_rtx_p): Move 0 check.
8203 2017-08-01 Bin Cheng <bin.cheng@arm.com>
8205 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
8206 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
8208 * match.pd: Ditto in address comparison pattern.
8210 2017-08-01 Bin Cheng <bin.cheng@arm.com>
8212 PR tree-optimization/81627
8213 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
8214 closed ssa form for store-store chain.
8216 2017-08-01 Bin Cheng <bin.cheng@arm.com>
8218 PR tree-optimization/81620
8219 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
8220 for store-store chain.
8222 2017-08-01 Jakub Jelinek <jakub@redhat.com>
8224 PR tree-optimization/81588
8225 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
8226 ranges[i].in_p, invert comparison code ccode. For >/>=,
8227 swap rhs1 and rhs2 and comparison code unconditionally,
8228 for </<= don't do that. Don't swap rhs1/rhs2 again if
8229 ranges[i].in_p, instead invert comparison code ccode if
8230 opcode or oe->rank is BIT_IOR_EXPR.
8233 * optabs.def (vec_extract_optab, vec_init_optab): Change from
8234 a direct optab to conversion optab.
8235 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
8236 with GET_MODE_INNER as last argument instead of optab_handler.
8237 * expmed.c (extract_bit_field_1): Likewise. Use vector from
8238 vector extraction if possible and optab is available.
8239 * expr.c (store_constructor): Use convert_optab_handler instead
8240 of optab_handler. Use vector initialization from smaller
8241 vectors if possible and optab is available.
8242 * tree-vect-stmts.c (vectorizable_load): Likewise.
8243 * doc/md.texi (vec_extract, vec_init): Document that the optabs
8245 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
8246 of vec_init from half-sized vectors with the same element mode.
8247 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
8248 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
8249 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
8250 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
8251 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
8252 after mode in gen_vec_extract* calls.
8253 (vec_extract<mode>): Renamed to ...
8254 (vec_extract<mode><ssescalarmodelower>): ... this.
8255 (vec_extract<mode><ssehalfvecmodelower>): New expander.
8256 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
8257 element mode after mode in gen_vec_init* calls.
8258 (VEC_INIT_HALF_MODE): New mode iterator.
8259 (vec_init<mode>): Renamed to ...
8260 (vec_init<mode><ssescalarmodelower>): ... this.
8261 (vec_init<mode><ssehalfvecmodelower>): New expander.
8262 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
8263 (vec_extractv2sfsf): ... this.
8264 (vec_initv2sf): Renamed to ...
8265 (vec_initv2sfsf): ... this.
8266 (vec_extractv2si): Renamed to ...
8267 (vec_extractv2sisi): ... this.
8268 (vec_initv2si): Renamed to ...
8269 (vec_initv2sisi): ... this.
8270 (vec_extractv4hi): Renamed to ...
8271 (vec_extractv4hihi): ... this.
8272 (vec_initv4hi): Renamed to ...
8273 (vec_initv4hihi): ... this.
8274 (vec_extractv8qi): Renamed to ...
8275 (vec_extractv8qiqi): ... this.
8276 (vec_initv8qi): Renamed to ...
8277 (vec_initv8qiqi): ... this.
8278 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
8279 (vec_init<mode>): Renamed to ...
8280 (vec_init<mode><VEC_base_l>): ... this.
8281 (vec_extract<mode>): Renamed to ...
8282 (vec_extract<mode><VEC_base_l>): ... this.
8283 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
8284 (vec_initv2sfsf): ... this.
8285 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
8286 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
8287 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
8288 element mode after mode in gen_vec_init* calls.
8289 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
8290 (vec_init<mode><Vel>): ... this.
8291 (vec_extract<mode>): Renamed to ...
8292 (vec_extract<mode><Vel>): ... this.
8293 * config/aarch64/iterators.md (Vel): New mode attribute.
8294 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
8295 Add element mode after mode in gen_vec_extract* calls.
8296 * config/s390/vector.md (non_vec_l): New mode attribute.
8297 (vec_extract<mode>): Renamed to ...
8298 (vec_extract<mode><non_vec_l>): ... this.
8299 (vec_init<mode>): Renamed to ...
8300 (vec_init<mode><non_vec_l>): ... this.
8301 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
8302 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
8304 * config/arm/iterators.md (V_elem_l): New mode attribute.
8305 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
8306 (vec_extract<mode><V_elem_l>): ... this.
8307 (vec_extractv2di): Renamed to ...
8308 (vec_extractv2didi): ... this.
8309 (vec_init<mode>): Renamed to ...
8310 (vec_init<mode><V_elem_l>): ... this.
8311 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
8312 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
8313 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
8314 Add element mode after gen_vec_extract* calls.
8315 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
8316 (vec_init<mode><unitmode>): ... this.
8317 (vec_extract<mode>): Renamed to ...
8318 (vec_extract<mode><unitmode>): ... this.
8319 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
8320 (vec_init<mode><unitmode>): ... this.
8321 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
8322 (vec_initv2sfsf): ... this.
8323 (vec_extractv2sf): Renamed to ...
8324 (vec_extractv2sfsf): ... this.
8325 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
8326 Add element mode after gen_vec_extract* calls.
8327 * config/mips/mips.md (unitmode): New mode iterator.
8328 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
8329 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
8330 * config/spu/spu.md (inner_l): New mode attribute.
8331 (vec_init<mode>): Renamed to ...
8332 (vec_init<mode><inner_l>): ... this.
8333 (vec_extract<mode>): Renamed to ...
8334 (vec_extract<mode><inner_l>): ... this.
8335 * config/sparc/sparc.md (veltmode): New mode iterator.
8336 (vec_init<VMALL:mode>): Renamed to ...
8337 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
8338 * config/ia64/vect.md (vec_initv2si): Renamed to ...
8339 (vec_initv2sisi): ... this.
8340 (vec_initv2sf): Renamed to ...
8341 (vec_initv2sfsf): ... this.
8342 (vec_extractv2sf): Renamed to ...
8343 (vec_extractv2sfsf): ... this.
8344 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
8345 (vec_init<mode>): Renamed to ...
8346 (vec_init<mode><VEC_base_l>): ... this.
8347 (vec_extract<mode>): Renamed to ...
8348 (vec_extract<mode><VEC_base_l>): ... this.
8349 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
8350 (vec_initv2sfsf): ... this.
8351 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
8352 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
8353 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
8354 gen_vec_init* calls.
8356 2017-08-01 Richard Biener <rguenther@suse.de>
8358 PR tree-optimization/81297
8359 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
8360 TREE_OVERFLOW from INTEGER_CSTs.
8362 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
8364 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
8366 2017-07-31 Carl Love <cel@us.ibm.com>
8368 * config/rs6000/rs6000-c: Add support for built-in functions
8369 vector signed char vec_xl_be (signed long long, signed char *);
8370 vector unsigned char vec_xl_be (signed long long, unsigned char *);
8371 vector signed int vec_xl_be (signed long long, signed int *);
8372 vector unsigned int vec_xl_be (signed long long, unsigned int *);
8373 vector signed long long vec_xl_be (signed long long, signed long long *);
8374 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
8375 vector signed short vec_xl_be (signed long long, signed short *);
8376 vector unsigned short vec_xl_be (signed long long, unsigned short *);
8377 vector double vec_xl_be (signed long long, double *);
8378 vector float vec_xl_be (signed long long, float *);
8379 * config/rs6000/altivec.h (vec_xl_be): Add #define.
8380 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
8381 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
8383 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
8384 (altivec_expand_builtin): Add switch statement to call
8385 altivec_expand_xl_be for each builtin.
8386 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
8387 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
8388 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
8389 __builtin_vsx_le_be_v16qi.
8390 * doc/extend.texi: Update the built-in documentation file for the
8391 new built-in functions.
8393 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
8396 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
8398 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
8400 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8402 * config.gcc: Add z14.
8403 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
8404 CPU model numbers for z13s and z14.
8405 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
8407 * config/s390/s390-opts.h (enum processor_type): Rename
8408 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
8409 * config/s390/s390.c (processor_table): Add field for CPU name to
8410 be passed to Binutils.
8411 (s390_asm_output_machine_for_arch): Use the new field in
8412 processor_table for Binutils.
8413 (s390_expand_builtin): Replace arch12 with z14.
8414 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
8415 (s390_get_sched_attrmask): Likewise.
8416 (s390_get_unit_mask): Likewise.
8417 * config/s390/s390.opt: Add z14 to processor_type enum.
8419 2017-07-31 Martin Jambor <mjambor@suse.cz>
8422 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
8423 regardless of optimization level.
8425 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
8426 Martin Liska <mliska@suse.cz>
8428 * predict.def: Remove old comment and adjust probability.
8429 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
8432 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
8435 * config/i386/i386.c (ix86_function_naked): New function.
8436 (ix86_can_use_return_insn_p): Return false for naked functions.
8437 (ix86_expand_prologue): Skip prologue for naked functions.
8438 (ix86_expand_epilogue): Skip epilogue for naked functions
8439 and emit trap instruction.
8440 (ix86_warn_func_return): New function.
8441 (ix86_attribute_table): Add "naked" attribute specification.
8442 (TARGET_WARN_FUNC_RETURN): Define.
8443 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
8445 2017-07-31 Martin Liska <mliska@suse.cz>
8447 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
8448 (dump_gimple_bb_header): Always dump BB info.
8449 (pp_cfg_jump): Do not append info about BB when dumping a jump.
8451 2017-07-31 Martin Liska <mliska@suse.cz>
8454 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
8455 also with current_function_decl non-null equality.
8457 2017-07-31 Jakub Jelinek <jakub@redhat.com>
8460 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
8461 change type to the element type, instead add eltype variable and
8462 use it where we are interested in the element type.
8464 PR tree-optimization/81603
8465 * ipa-polymorphic-call.c
8466 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
8467 offset arithmetic in offset_int, bail out if the resulting bit offset
8468 doesn't fit into shwi.
8470 2017-07-31 Martin Liska <mliska@suse.cz>
8472 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
8473 (gimplify_save_expr): Fix comment.
8475 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
8478 * config/i386/i386.c (ix86_function_arg): Update arguments for
8480 (ix86_compute_frame_layout): Set the initial stack offset to
8481 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
8482 INCOMING_FRAME_SP_OFFSET.
8483 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
8484 stack before exception handler returns.
8485 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
8486 the 'ERROR_CODE' for exception handler.
8488 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
8490 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
8491 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
8492 (ASM_OUTPUT_REG_POP): Ditto.
8493 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
8494 instead of asm_fprintf to output pure string.
8496 2017-07-29 Jakub Jelinek <jakub@redhat.com>
8498 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
8499 to imported_module_or_decl hook.
8500 (debug_nothing_tree_tree_tree_bool): Remove.
8501 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
8502 * debug.c (do_nothing_debug_hooks): Use
8503 debug_nothing_tree_tree_tree_bool_bool instead of
8504 debug_nothing_tree_tree_tree_bool.
8505 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
8506 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
8507 * sdbout.c (sdb_debug_hooks): Likewise.
8508 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
8509 (gen_namespace_die): Add DW_AT_export_symbols attribute if
8511 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
8512 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
8513 attribute, don't add anything.
8515 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8517 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
8518 (fold_build2_stat_loc): Likewise.
8519 (fold_build3_stat_loc): Likewise.
8520 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
8521 (fold_build1_loc): Remove macro.
8522 (fold_build2_loc): Likewise.
8523 (fold_build3_loc): Likewise.
8525 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8527 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
8528 (gimple_build_debug_bind_source_stat): Likewise.
8529 * gimple.h (gimple_build_debug_bind): Remove macro.
8530 (gimple_build_debug_bind_source): Likewise.
8532 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8534 * bitmap.c (bitmap_alloc): Adjust.
8535 (bitmap_gc_alloc): Likewise.
8536 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
8538 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8540 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
8541 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
8542 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
8543 (bitmap_gc_alloc_stat): Likewise.
8544 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
8546 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8548 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
8549 * rtl.h (shallow_copy_rtx): Remove macro.
8551 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8553 * emit-rtl.c (gen_raw_REG): Adjust.
8554 * gengenrtl.c (gendef): Likewise.
8555 * rtl.c (rtx_alloc_stat): Remove _stat from name.
8556 * rtl.h (rtx_alloc): Remove macro.
8558 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8560 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
8561 (build_tree_list_stat): Likewise.
8562 * tree.h (build_tree_list): Remove macro.
8563 (build_tree_list_vec): Likewise.
8565 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8567 * tree.c (make_vector_stat): Remove _stat from name.
8568 (build_vector_stat): Likewise.
8569 * tree.h (make_vector_stat): Remove macro.
8570 (build_vector_stat): Likewise.
8572 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8574 * tree.h (build_var_debug_value): Remove prototype.
8576 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8578 * tree.c (tree_cons_stat): Remove _stat from name.
8579 * tree.h (tree_cons): Remove macro.
8581 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8583 * tree.c (build_vl_exp_stat): Remove _stat from name.
8584 * tree.h (build_vl_exp): Remove macro.
8586 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8588 * tree.c (build_decl_stat): Remove _stat from name.
8589 * tree.h (build_decl): Remove macro.
8591 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8593 * gimple.c (gimple_build_with_ops_stat): Adjust.
8594 (gimple_alloc_stat): Remove _stat from name.
8595 * gimple.h (gimple_alloc): Remove macro.
8597 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8599 * tree.c (make_tree_vec_stat): Remove _stat from name.
8600 (grow_tree_vec_stat): Likewise.
8601 * tree.h (make_tree_vec_stat): Adjust prototype.
8602 (grow_tree_vec_stat): Likewise.
8603 (make_tree_vec): Remove macro.
8604 (grow_tree_vec): Likewise.
8606 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8608 * fold-const.c (fold_build1_stat_loc): Adjust.
8609 (fold_build2_stat_loc): Likewise.
8610 (fold_build3_stat_loc): Likewise.
8611 * tree.c (build0_stat): Remove _stat from name.
8612 (build1_stat): Likewise.
8613 (build2_stat): Likewise.
8614 (build3_stat): Likewise.
8615 (build4_stat): Likewise.
8616 (build5_stat): Likewise.
8617 * tree.h (build1_loc): Remove macro, and rename _stat function
8619 (build2_loc): Likewise.
8620 (build3_loc): Likewise.
8621 (build4_loc): Likewise.
8622 (build5_loc): Likewise.
8624 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8626 * tree.c (make_int_cst_stat): Remove _stat from name.
8627 * tree.h (make_int_cst_stat): Adjust prototype.
8628 (make_int_cst): Remove macro.
8630 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8632 * tree.c (make_tre_binfo_stat): Remove _stat from name.
8633 * tree.h (make_tree_binfo_stat): Adjust prototype.
8634 (make_tree_binfo): Remove.
8636 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8638 * tree.c (copy_node_stat): Rename to copy_node.
8639 (build_distinct_type_copy): Adjust.
8640 * tree.h (copy_node_stat): Adjust prototype.
8641 (copy_node): Remove macro.
8643 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8645 * tree.c (make_node_stat): rename to make_node.
8646 (build_tree_list_stat): Adjust.
8647 (build0_stat): Likewise.
8648 (build2_stat): Likewise.
8649 (build3_stat): Likewise.
8650 (build4_stat): Likewise.
8651 (build5_stat): Likewise.
8652 (build_decl_stat): Likewise.
8653 * tree.h (make_node_stat): Adjust prototype.
8654 (make_node): remove macro.
8656 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
8658 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
8659 (PPC_FEATURE2_SCV): Likewise.
8660 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
8662 2017-07-28 Tamar Christina <tamar.christina@arm.com>
8664 * config/aarch64/aarch64.c
8665 (aarch64_internal_mov_immediate): Add new special pattern.
8666 * config/aarch64/aarch64.md (*movdi_aarch64):
8667 Add reg/32bit const mov case.
8669 2017-07-28 Tamar Christina <tamar.christina@arm.com>
8670 Richard Sandiford <richard.sandiford@linaro.org>
8672 * config/aarch64/aarch64.md (mov<mode>): Generalize.
8673 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
8674 Add integer and movi cases.
8675 (movi-split-hf-df-sf split, fp16): New.
8676 (enabled): Added TARGET_FP_F16INST.
8677 * config/aarch64/iterators.md (GPF_HF): New.
8678 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
8680 2017-07-28 Tamar Christina <tamar.christina@arm.com>
8682 * config/aarch64/aarch64.c
8683 (aarch64_simd_container_mode): Add prototype.
8684 (aarch64_expand_mov_immediate): Add HI support.
8685 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
8686 (aarch64_can_const_movi_rtx_p): New.
8687 (aarch64_preferred_reload_class):
8688 Remove restrictions of using FP registers for certain SIMD operations.
8689 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
8690 (aarch64_valid_floating_const): Add integer move validation.
8691 (aarch64_simd_imm_scalar_p): Remove.
8692 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
8693 (aarch64_legitimate_constant_p): Expand list of supported cases.
8694 * config/aarch64/aarch64-protos.h
8695 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
8696 (aarch64_reinterpret_float_as_int): New.
8697 (aarch64_simd_imm_scalar_p): Remove.
8698 * config/aarch64/constraints.md (Uvi): New.
8699 (Dd): Split into Ds and new Dd.
8700 * config/aarch64/aarch64.md (*movsi_aarch64):
8702 (*movdi_aarch64): Add SIMD mov case.
8704 2017-07-28 Bin Cheng <bin.cheng@arm.com>
8706 * tree-predcom.c: (struct chain): Handle store-store chain in which
8707 stores for elimination only store loop invariant values.
8708 (execute_pred_commoning_chain): Ditto.
8709 (prepare_initializers_chain_store_elim): Ditto.
8710 (prepare_finalizers): Ditto.
8711 (is_inv_store_elimination_chain): New function.
8712 (initialize_root_vars_store_elim_1): New function.
8714 2017-07-28 Bin Cheng <bin.cheng@arm.com>
8716 * tree-predcom.c: Revise general description of the pass.
8717 (enum chain_type): New enum type for store elimination.
8718 (struct chain): New field supporting store elimination.
8719 (struct component): Ditto.
8720 (dump_chain): Dump store-stores chain.
8721 (release_chain): Release resources.
8722 (split_data_refs_to_components): Compute and create component
8723 contains only stores for elimination.
8724 (get_chain_last_ref_at): New function.
8725 (make_invariant_chain): Initialization.
8726 (make_rooted_chain): Specify chain type in parameter and record it.
8727 (add_looparound_copies): Skip for store-stores chain.
8728 (determine_roots_comp): Compute type of chain and pass it to
8730 (initialize_root_vars_store_elim_2): New function.
8731 (finalize_eliminated_stores): New function.
8732 (remove_stmt): Handle store for elimination.
8733 (execute_pred_commoning_chain): Execute predictive commoning on
8735 (determine_unroll_factor): Skip unroll for store-stores chain.
8736 (prepare_initializers_chain_store_elim): New function.
8737 (prepare_initializers_chain): Hanlde store-store chain.
8738 (prepare_finalizers_chain, prepare_finalizers): New function.
8739 (tree_predictive_commoning_loop): Return integer value indicating
8740 if loop is unrolled or lcssa form is corrupted.
8741 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
8743 2017-07-28 Bin Cheng <bin.cheng@arm.com>
8745 * tree-predcom.c (initialize_root): Delete.
8746 (execute_pred_commoning_chain): Initialize root vars and replace
8747 reference of non-combined chain directly, rather than call above
8750 2017-07-28 Bin Cheng <bin.cheng@arm.com>
8752 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
8753 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
8755 2017-07-28 Bin Cheng <bin.cheng@arm.com>
8757 * tree-predcom.c (struct chain): New field init_seq.
8758 (release_chain): Release init_seq.
8759 (prepare_initializers_chain): Record intialization stmts in above
8761 (insert_init_seqs): New function.
8762 (tree_predictive_commoning_loop): Call insert_init_seqs.
8764 2017-07-28 Bin Cheng <bin.cheng@arm.com>
8766 * tree-predcom.c (determine_roots_comp): Skip trivial components.
8768 2017-07-28 Richard Biener <rguenther@suse.de>
8770 * match.pd: Remove superfluous :c.
8771 * genmatch.c (simplify::id): Add member.
8772 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
8774 (current_id): New global.
8775 (dt_node::parent): Move from ...
8776 (dt_operand::parent): ... here. Add for_id member.
8777 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
8778 (decision_tree::find_node): Relax order requirement when
8779 merging DT_TRUE nodes to ones inbetween the current simplify
8780 and the one we try to merge with. Add diagnostic whenever
8781 we need to enforce pattern order by not merging.
8782 (decision_tree::insert): Set current_id.
8783 (decision_tree::print_node): Dump parent node and for_id.
8784 (parser::last_id): Add member.
8785 (parser::push_simplify): Assign unique id.
8786 (parser::parser): Initialize last_id.
8788 2017-07-28 Martin Liska <mliska@suse.cz>
8791 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
8792 gimple_build_debug_bind.
8794 2017-07-28 Richard Biener <rguenther@suse.de>
8796 PR tree-optimization/81502
8797 * match.pd: Add pattern combining BIT_INSERT_EXPR with
8799 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
8801 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
8802 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
8803 for BIT_FIELD_REF args.
8804 * fold-const.c (make_bit_field_ref): Likewise.
8805 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
8807 2017-07-28 Jakub Jelinek <jakub@redhat.com>
8810 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
8811 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
8812 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
8813 Or it into SANITIZER_UNDEFINED.
8814 * ubsan.c: Include gimple-fold.h and varasm.h.
8815 (ubsan_expand_ptr_ifn): New function.
8816 (instrument_pointer_overflow): New function.
8817 (maybe_instrument_pointer_overflow): New function.
8818 (instrument_object_size): Formatting fix.
8819 (pass_ubsan::execute): Call instrument_pointer_overflow
8820 and maybe_instrument_pointer_overflow.
8821 * internal-fn.c (expand_UBSAN_PTR): New function.
8822 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
8823 * sanitizer.def (__ubsan_handle_pointer_overflow,
8824 __ubsan_handle_pointer_overflow_abort): New builtins.
8825 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
8826 * internal-fn.def (UBSAN_PTR): New internal function.
8827 * opts.c (sanitizer_opts): Add pointer-overflow.
8828 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
8829 * fold-const.c (build_range_check): Compute pointer range check in
8830 integral type if pointer arithmetics would be needed. Formatting
8833 2017-07-28 Martin Liska <mliska@suse.cz>
8836 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
8837 parameters that are of a variable-length.
8839 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
8841 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
8843 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
8844 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
8845 (CRT_CALL_STATIC_FUNCTION): Likewise.
8846 (ASM_DEFAULT_SPEC): New define.
8847 (ASM_SPEC32): Likewise.
8848 (ASM_SPEC64): Likewise.
8849 (ASM_SPEC_COMMON): Likewise.
8850 (ASM_SPEC): Likewise.
8851 (INVALID_64BIT): Likewise.
8852 (LINK_OS_DEFAULT_SPEC): Likewise.
8853 (LINK_OS_SPEC32): Likewise.
8854 (LINK_OS_SPEC64): Likewise.
8855 (POWERPC_LINUX): Likewise.
8856 (PTRDIFF_TYPE): Likewise.
8857 (RESTORE_FP_PREFIX): Likewise.
8858 (RESTORE_FP_SUFFIX): Likewise.
8859 (SAVE_FP_PREFIX): Likewise.
8860 (SAVE_FP_SUFFIX): Likewise.
8861 (SIZE_TYPE): Likewise.
8862 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
8863 (TARGET_64BIT): Likewise.
8864 (TARGET_64BIT): Likewise.
8865 (TARGET_AIX): Likewise.
8866 (WCHAR_TYPE_SIZE): Likewise.
8867 (WCHAR_TYPE): Undefine.
8868 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
8869 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
8870 (CPP_OS_RTEMS_SPEC): Delete.
8871 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
8872 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
8874 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
8876 2017-07-28 Jakub Jelinek <jakub@redhat.com>
8878 PR tree-optimization/81578
8879 * tree-parloops.c (build_new_reduction): Bail out if
8880 reduction_code isn't one of the standard OpenMP reductions.
8881 Move the details printing after that decision.
8883 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
8885 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
8886 related to reload_in_progress.
8887 (splat_input_operand): Likewise.
8888 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
8890 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
8892 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
8893 (TARGET_INSTANTIATE_DECLS): Likewise.
8894 (legitimate_indexed_address_p): Delete reload_in_progress code.
8895 (rs6000_debug_legitimate_address_p): Likewise.
8896 (rs6000_eliminate_indexed_memrefs): Likewise.
8897 (rs6000_emit_le_vsx_store): Likewise.
8898 (rs6000_emit_move_si_sf_subreg): Likewise.
8899 (rs6000_emit_move): Likewise.
8900 (register_to_reg_type): Likewise.
8901 (rs6000_pre_atomic_barrier): Likewise.
8902 (rs6000_machopic_legitimize_pic_address): Likewise.
8903 (rs6000_allocate_stack_temp): Likewise.
8904 (rs6000_address_for_fpconvert): Likewise.
8905 (rs6000_address_for_altivec): Likewise.
8906 (rs6000_secondary_memory_needed_rtx): Delete function.
8907 (rs6000_check_sdmode): Likewise.
8908 (rs6000_alloc_sdmode_stack_slot): Likewise.
8909 (rs6000_instantiate_decls): Likewise.
8910 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
8911 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
8912 Delete reload_in_progress.
8913 (*vec_reload_and_plus_<mptrsize>): Likewise.
8914 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
8915 (vsx_div_v2di): Likewise.
8916 (vsx_udiv_v2di): Likewise.
8918 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
8920 * config/rs6000/rs6000.opt (mlra): Replace with stub.
8921 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
8922 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
8923 (rs6000_debug_reg_global): Delete print of LRA status.
8924 (rs6000_option_override_internal): Delete dead LRA related code.
8925 (rs6000_lra_p): Delete function.
8926 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
8928 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
8930 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
8931 * config/riscv/rtems.h: New file.
8933 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8934 Sudakshina Das <sudi.das@arm.com>
8936 * config/aarch64/aarch64.md
8937 (define_split for and<mode>3nr_compare): Move
8938 non aarch64_logical_operand to a register.
8939 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
8940 register immediate operand to a register.
8941 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
8943 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
8946 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
8948 2017-07-27 Richard Biener <rguenther@suse.de>
8950 PR tree-optimization/81573
8951 PR tree-optimization/81494
8952 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
8953 multi defuse cycle case.
8955 2017-07-27 Richard Biener <rguenther@suse.de>
8957 PR tree-optimization/81571
8958 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
8961 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
8963 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
8964 earlier and only if MASK_FPU is set. Adjust formatting.
8966 2017-07-27 Martin Liska <mliska@suse.cz>
8968 * opt-functions.awk: Add validation of value of Init.
8969 * optc-gen.awk: Pass new argument.
8971 2017-07-27 Martin Liska <mliska@suse.cz>
8973 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
8974 Fix wrong condition.
8976 2017-07-27 Martin Liska <mliska@suse.cz>
8978 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
8979 BBs and edges seen by autoFDO.
8981 2017-07-27 Richard Biener <rguenther@suse.de>
8983 PR tree-optimization/81502
8984 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
8985 with incompatible but same sized type.
8986 (execute_update_addresses_taken): Likewise.
8988 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
8990 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
8991 flag_tree_loop_vectorize rather than flag_tree_vectorize.
8993 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8996 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
8997 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
8998 Change s_operand to memory_operand.
9000 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
9002 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
9003 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
9004 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
9005 Emit instructions rather than returning an expression. Handle TFmode
9006 and KFmode by casting to TImode.
9007 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
9008 (rs6000_emit_le_vsx_store): Likewise.
9009 * config/rs6000/vsx.md (VSX_TI): New iterator.
9010 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
9011 (*vsx_le_undo_permute_<mode>): Likewise.
9012 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
9013 emit the split sequence.
9014 (*vsx_le_perm_store_<mode>): Likewise.
9016 2017-07-27 Jakub Jelinek <jakub@redhat.com>
9018 PR tree-optimization/81555
9019 PR tree-optimization/81556
9020 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
9021 if true, force CHANGED for the recursive invocation.
9022 (reassociate_bb): Remember original length of ops array, pass
9023 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
9025 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
9026 attributes for noipa attribute. For naked attribute use
9027 lookup_attribute first before lookup_attribute_spec.
9028 * final.c (rest_of_handle_final): Disable IPA RA for functions with
9030 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
9031 for functions with noipa attribute.
9032 (cgraph_externally_visible_p): Return true for functions with noipa
9034 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
9035 for functions with noipa attribute.
9036 * doc/extend.texi: Document noipa function attribute.
9037 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
9038 also for functions with noipa attribute.
9039 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
9041 2017-07-26 Andrew Pinski <apinski@cavium.com>
9043 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
9044 vec_unalign_load_cost and vec_unalign_store_cost.
9046 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
9048 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
9049 -mvsx-small-integer option.
9050 (ISA_3_0_MASKS_IEEE): Likewise.
9051 (OTHER_VSX_VECTOR_MASKS): Likewise.
9052 (POWERPC_MASKS): Likewise.
9053 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
9054 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
9055 code, only testing for DImode being allowed in non-VSX floating
9057 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
9058 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
9060 (rs6000_option_override_internal): Delete -mvsx-small-integer.
9061 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
9062 TARGET_P8_VECTOR test.
9063 (rs6000_secondary_reload_simple_move): Likewise.
9064 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
9065 since TARGET_P9_VECTOR was already tested.
9066 (rs6000_opt_masks): Remove -mvsx-small-integer.
9067 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
9068 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
9070 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
9071 test for TARGET_VEXTRACTUB was used, and that uses
9073 (p9 extract splitter): Likewise.
9074 (vsx_extract_<mode>_di_p9): Likewise.
9075 (vsx_extract_<mode>_store_p9): Likewise.
9076 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
9077 for TARGET_P9_VECTOR was used. Delete code that is now dead with
9078 the elimination of TARGET_VSX_SMALL_INTEGER.
9079 (vsx_extract_<mode>_p8): Likewise.
9080 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
9081 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
9082 (vsx_set_<mode>_p9): Likewise.
9083 (vsx_set_v4sf_p9): Likewise.
9084 (vsx_set_v4sf_p9_zero): Likewise.
9085 (vsx_insert_extract_v4sf_p9): Likewise.
9086 (vsx_insert_extract_v4sf_p9_2): Likewise.
9087 * config/rs6000/rs6000.md (sign extend splitter): Change
9088 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
9089 (floatsi<mode>2_lfiwax_mem): Likewise.
9090 (floatunssi<mode>2_lfiwzx_mem): Likewise.
9091 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
9092 since a test for TARGET_P9_VECTOR was used.
9093 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
9094 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
9095 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
9096 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
9097 TARGET_P8_VECTOR test.
9098 (fix_trunc<mode>si2_stfiwx): Likewise.
9099 (fix_trunc<mode>si2_internal): Likewise.
9100 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
9101 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
9103 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
9104 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
9105 TARGET_P8_VECTOR test.
9106 (fixuns_trunc<mode>si2_stfiwx): Likewise.
9107 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
9108 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
9110 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
9111 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
9112 since a test for TARGET_P9_VECTOR was used.
9113 (splitter for loading small constants): Likewise.
9115 2017-07-26 Andrew Pinski <apinski@cavium.com>
9117 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
9120 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
9123 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
9124 (fp_valid_at): Likewise.
9126 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
9128 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
9129 (qdf24xx_addrcost_table): Likewise.
9130 (cortexa57_tunings): Update to use generic_branch_cost.
9131 (cortexa72_tunings): Likewise.
9132 (cortexa73_tunings): Likewise.
9133 (qdf24xx_tunings): Likewise.
9135 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
9137 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
9138 (thunderx2t99_branch_cost): Likewise.
9139 (cortexa35_tunings): Update to use generic_branch_cost.
9140 (cortexa53_tunings): Likewise.
9141 (cortexa57_tunings): Likewise.
9142 (cortexa72_tunings): Likewise.
9143 (cortexa73_tunings): Likewise.
9144 (thunderx2t99_tunings): Likewise.
9146 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
9148 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
9149 (sparc_option_override): Honour MASK_FSMULD.
9150 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
9151 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
9152 * config/sparc/sparc.opt (mfsmuld): New option.
9153 * doc/invoke.texi (mfsmuld): Document option.
9155 2017-07-26 Marek Polacek <polacek@redhat.com>
9158 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
9160 2017-07-26 Richard Biener <rguenther@suse.de>
9162 * gimple-match-head.c (do_valueize): Return OP if valueize
9164 (get_def): New helper to get at the def stmt of a SSA name
9166 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
9167 do_valueize to get at the def stmt.
9168 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
9170 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
9173 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
9175 2017-07-26 Martin Liska <mliska@suse.cz>
9178 * function.c (expand_function_start): Make expansion of
9179 nonlocal_goto_save_area after parm_birth_insn.
9181 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
9183 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
9184 from all CPU target flags enable members.
9186 2017-07-26 Richard Biener <rguenther@suse.de>
9188 * genmatch.c (dt_simplify::gen): Make iterator vars const.
9189 (decision_tree::gen): Make 'type' const.
9190 (write_predicate): Likewise.
9192 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
9194 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
9195 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
9196 (rs6000_option_override_internal): Likewise.
9197 (rs6000_expand_vector_set): Likewise.
9198 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
9199 (TARGET_UPPER_REGS_SF): Likewise.
9200 (TARGET_UPPER_REGS_DI): Likewise.
9201 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
9202 (TARGET_DIRECT_MOVE_64BIT): Likewise.
9203 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
9204 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
9205 (Splitters for DI constants in Altivec registers): Likewise.
9206 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
9207 (vsx_set_v4sf_p9): Likewise.
9208 (vsx_set_v4sf_p9_zero): Likewise.
9209 (vsx_insert_extract_v4sf_p9): Likewise.
9210 (vsx_insert_extract_v4sf_p9_2): Likewise.
9212 2017-07-25 Carl Love <cel@us.ibm.com>
9214 * doc/extend.texi: Update the built-in documentation file for the
9215 existing built-in functions
9216 vector signed char vec_cnttz (vector signed char);
9217 vector unsigned char vec_cnttz (vector unsigned char);
9218 vector signed short vec_cnttz (vector signed short);
9219 vector unsigned short vec_cnttz (vector unsigned short);
9220 vector signed int vec_cnttz (vector signed int);
9221 vector unsigned int vec_cnttz (vector unsigned int);
9222 vector signed long long vec_cnttz (vector signed long long);
9223 vector unsigned long long vec_cnttz (vector unsigned long long);
9225 2017-07-25 Andrew Pinski <apinski@cavium.com>
9227 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
9228 accesses where the use is for the first operand of a BIT_INSERT.
9230 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
9233 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
9234 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
9236 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
9238 * config/i386/gstabs.h: Delete.
9239 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
9241 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
9243 * config/i386/i386.c (ix86_decompose_address): Do not check for
9244 register RTX when looking at index_reg or base_reg.
9245 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
9247 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
9249 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
9250 to update EH info here.
9252 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
9254 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
9256 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
9258 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
9260 2017-07-25 Torsten Duwe <duwe@suse.de>
9262 * common.opt: Introduce -fpatchable-function-entry
9263 command line option, and its variables function_entry_patch_area_size
9264 and function_entry_patch_area_start.
9265 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
9266 including a two-value parser.
9267 * target.def (print_patchable_function_entry): New target hook.
9268 * targhooks.h (default_print_patchable_function_entry): New function.
9269 * targhooks.c (default_print_patchable_function_entry): Likewise.
9270 * toplev.c (process_options): Switch off IPA-RA if
9271 patchable function entries are being generated.
9272 * varasm.c (assemble_start_function): Look at the
9273 patchable-function-entry command line switch and current
9274 function attributes and maybe generate NOP instructions by
9275 calling the print_patchable_function_entry hook.
9276 * doc/extend.texi: Document patchable_function_entry attribute.
9277 * doc/invoke.texi: Document -fpatchable_function_entry
9278 command line option.
9279 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
9281 * doc/tm.texi: Re-generate.
9283 2017-07-25 Jakub Jelinek <jakub@redhat.com>
9286 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
9287 TARGET_AVX512DQ rather than TARGET_AVX512BW.
9289 2017-07-25 Tamar Christina <tamar.christina@arm.com>
9291 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
9293 2017-07-25 Richard Biener <rguenther@suse.de>
9295 PR tree-optimization/81455
9296 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
9297 not walk in cycles when looking for guards.
9299 2017-07-25 Richard Biener <rguenther@suse.de>
9301 PR tree-optimization/81529
9302 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
9303 when optimizing backedge uses.
9305 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
9307 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
9309 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
9310 to dl_section_ref. On AIX, append an expression to subtract
9311 the size of the section length to dl_section_ref.
9313 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
9315 * configure.ac: If any of the config.* scripts fail, exit 1.
9316 * configure: Regenerate.
9318 2017-07-25 Richard Biener <rguenther@suse.de>
9321 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
9322 of immediate uses, be more verbose on errors.
9324 2017-07-25 Richard Biener <rguenther@suse.de>
9326 PR tree-optimization/81510
9327 * tree-vect-loop.c (vect_is_simple_reduction): When the
9328 reduction stmt is not inside the loop bail out.
9330 2017-07-25 Richard Biener <rguenther@suse.de>
9332 PR tree-optimization/81303
9333 * tree-vect-loop-manip.c (vect_loop_versioning): Build
9334 profitability check against LOOP_VINFO_NITERSM1.
9336 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
9338 * domwalk.c (cmp_bb_postorder): Simplify.
9339 (sort_bbs_postorder): New function. Use it...
9340 (dom_walker::walk): ...here to optimize common cases.
9342 2017-07-25 Martin Liska <mliska@suse.cz>
9345 * ipa-visibility.c (function_and_variable_visibility): Make the
9346 redirection just on target that supports aliasing.
9347 Fix GNU coding style.
9349 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
9352 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
9354 * config/arm/rtems.h: Likewise.
9355 * config/bfin/rtems.h: Likewise.
9356 * config/i386/rtemself.h: Likewise.
9357 * config/lm32/rtems.h: Likewise.
9358 * config/m32c/rtems.h: Likewise.
9359 * config/m68k/rtemself.h: Likewise.
9360 * config/microblaze/rtems.h: Likewise.
9361 * config/mips/rtems.h: Likewise.
9362 * config/moxie/rtems.h: Likewise.
9363 * config/nios2/rtems.h: Likewise.
9364 * config/powerpcspe/rtems.h: Likewise.
9365 * config/rs6000/rtems.h: Likewise.
9366 * config/rtems.h: Likewise.
9367 * config/sh/rtems.h: Likewise.
9368 * config/sh/rtemself.h: Likewise.
9369 * config/sparc/rtemself.h: Likewise.
9371 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
9374 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
9375 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
9376 * tree-ssa-structalias.c (alias_get_name): Same.
9378 2017-07-25 Bin Cheng <bin.cheng@arm.com>
9381 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
9382 instructions if no du chain is found.
9384 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
9386 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
9388 2017-07-25 Richard Biener <rguenther@suse.de>
9391 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
9394 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
9396 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
9398 (ISA_2_7_MASKS_SERVER): Likewise.
9399 (ISA_3_0_MASKS_IEEE): Likewise.
9400 (OTHER_P8_VECTOR_MASKS): Likewise.
9401 (OTHER_VSX_VECTOR_MASKS): Likewise.
9402 (POWERPC_MASKS): Likewise.
9403 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
9404 duplicate list of options.
9405 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
9406 explicit -mupper-regs options.
9407 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
9408 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
9409 alias for -mupper-regs-df.
9410 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
9411 (rs6000_init_hard_regno_mode_ok): Likewise.
9412 (rs6000_option_override_internal): Likewise.
9413 (rs6000_opt_masks): Likewise.
9414 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
9415 options in terms of whether -mvsx or -mpower8-vector was used.
9416 (TARGET_UPPER_REGS_DI): Likewise.
9417 (TARGET_UPPER_REGS_SF): Likewise.
9418 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
9419 -mupper-regs-* options.
9421 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
9423 * passes.c (emergency_dump_function): Print some empty lines and a
9424 header before the RTL dump.
9426 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
9428 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
9430 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
9433 * config/aarch64/aarch64.c (aarch64_classify_symbol):
9434 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
9436 2017-07-24 Carl Love <cel@us.ibm.com>
9438 * config/rs6000/rs6000-c.c: Add support for built-in functions
9439 vector float vec_extract_fp32_from_shorth (vector unsigned short);
9440 vector float vec_extract_fp32_from_shortl (vector unsigned short);
9441 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
9442 vec_extract_fp_from_shortl): Add defines for the two builtins.
9443 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
9444 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
9446 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
9447 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
9448 * doc/extend.texi: Update the built-in documentation file for the
9449 new built-in function.
9451 2017-07-24 Jakub Jelinek <jakub@redhat.com>
9454 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
9456 * doc/generic.texi: Likewise.
9457 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
9458 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
9460 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
9462 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
9463 (aarch64_mls_elt_merge<mode>): Likewise.
9465 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
9467 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
9468 having __cxa_atexit.
9470 2017-07-23 Michael Collison <michael.collison@arm.com>
9472 * config/arm/arm.c (arm_option_override): Deprecate
9473 use of -mstructure-size-boundary.
9474 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
9475 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
9477 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9480 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
9481 Reduce cost estimate for direct moves.
9483 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
9486 * config/i386/i386.c (ix86_option_override_internal): Disable
9487 BMI, BMI2 and TBM instructions for -m16.
9489 2017-07-21 Carl Love <cel@us.ibm.com>
9491 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9492 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
9493 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
9494 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
9495 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
9496 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
9497 VMULOSW): New enum "unspec" values.
9498 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
9499 altivec_vmulosw): New patterns.
9500 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
9501 VMULOSW): Add definitions.
9503 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
9505 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
9506 (qdf24xx): Likewise.
9507 * config/aarch64/aarch64-options-extensions.def (rdma); New.
9508 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
9509 (AARCH64_FL_V8_1): Renumber.
9510 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
9511 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
9512 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
9513 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
9514 rdma to feature modifiers list.
9516 2017-07-21 Yury Gribov <tetra2005@gmail.com>
9519 * ipa-visibility (function_and_variable_visibility): Convert
9520 recursive PLT call to direct call if appropriate.
9522 2017-07-21 Andrew Pinski <apinski@cavium.com>
9524 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
9525 operand 1 to see if the types precision matches.
9526 * fold-const.c (operand_equal_p): Likewise.
9528 2017-07-21 Richard Biener <rguenther@suse.de>
9530 PR tree-optimization/81303
9531 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
9532 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
9533 (vect_peeling_hash_get_lowest_cost): Adjust.
9534 (vect_enhance_data_refs_alignment): Likewise. Use
9535 vect_get_peeling_costs_all_drs to compute the penalty for no
9536 peeling to match up costs.
9538 2017-07-21 Richard Biener <rguenther@suse.de>
9540 PR tree-optimization/81500
9541 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
9542 we didn't identify a reduction path.
9544 2017-07-21 Tom de Vries <tom@codesourcery.com>
9545 Cesar Philippidis <cesar@codesourcery.com>
9547 PR gcov-profile/81442
9548 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
9551 2017-07-21 Tom de Vries <tom@codesourcery.com>
9554 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
9556 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
9557 nvptx_override_options_after_change.
9559 2017-07-21 Ulrich Drepper <drepper@redhat.com>
9561 * dwarf2out.c (output_file_names): Avoid double testing for
9564 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
9566 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
9568 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
9570 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
9572 (try_crossjump_to_edge): Do not punt on partitioned functions.
9574 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
9576 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
9577 Put all BBs reachable only via paths crossing cold region to cold
9579 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
9581 2016-07-21 Richard Biener <rguenther@suse.de>
9583 PR tree-optimization/81303
9584 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
9585 into account prologue and epilogue iterations when raising
9586 min_profitable_iters to sth at least covering one vector iteration.
9588 2017-07-21 Tamar Christina <tamar.christina@arm.com>
9590 * config/arm/arm.c (arm_test_cpu_arch_dat):
9593 2017-07-20 Nathan Sidwell <nathan@acm.org>
9595 Remove TYPE_METHODS.
9596 * tree.h (TYPE_METHODS): Delete.
9597 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
9598 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
9599 (dbxout_type_methods): Scan TYPE_FIELDS.
9600 (dbxout_type): Don't check TYPE_METHODS here.
9601 * function.c (use_register_for_decl): Always ignore register for
9602 class types when not optimizing.
9603 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
9604 * tree.c (free_lang_data_in_type): Stitch out member functions and
9605 templates from TYPE_FIELDS.
9606 (build_distinct_type_copy, verify_type_variant,
9607 verify_type): Member fns are on TYPE_FIELDS.
9608 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
9609 * tree-pretty-print.c (dump_generic_node): Likewise.
9611 2017-07-20 Jakub Jelinek <jakub@redhat.com>
9614 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
9615 V2TImode and V4TImode.
9616 (ix86_expand_vector_extract): Likewise.
9617 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
9618 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
9619 (ssescalarmode): Handle V4TImode and V2TImode.
9620 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
9621 (*vec_extractv2ti, *vec_extractv4ti): New insns.
9622 (VEXTRACTI128_MODE): New mode iterator.
9623 (splitter for *vec_extractv?ti first element): New.
9624 (VEC_INIT_MODE): New mode iterator.
9625 (vec_init<mode>): Consolidate 3 expanders into one using
9626 VEC_INIT_MODE mode iterator.
9628 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
9630 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
9631 non_spilled_static_chain_regno_p.
9633 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
9635 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
9637 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
9639 * bb-reorder.c (connect_traces): Allow copying of blocks within
9642 2017-07-20 Richard Biener <rguenther@suse.de>
9644 * gimple.h (gimple_phi_result): Add gphi * overload.
9645 (gimple_phi_result_ptr): Likewise.
9646 (gimple_phi_arg): Likewise. Adjust index assert to only
9647 allow actual argument accesses rather than all slots available
9649 (gimple_phi_arg_def): Add gphi * overload.
9650 * tree-phinodes.c (make_phi_node): Initialize only actual
9652 (resize_phi_node): Clear memory not covered by old node,
9653 do not initialize excess argument slots.
9654 (reserve_phi_args_for_new_edge): Initialize new argument slot
9657 2017-07-20 Bin Cheng <bin.cheng@arm.com>
9659 PR tree-optimization/81388
9661 2016-07-21 Bin Cheng <bin.cheng@arm.com>
9663 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
9664 by removing computation of may_be_zero.
9666 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
9667 Tom de Vries <tom@codesourcery.com>
9670 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
9671 when gimple level profile disagrees with what RTL expander did.
9673 2017-07-20 Richard Biener <rguenther@suse.de>
9675 PR tree-optimization/61171
9676 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
9677 (vect_analyze_stmt): Add slp instance parameter.
9678 (vectorizable_reduction): Likewise.
9679 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
9680 (vect_is_simple_reduction): Deal with chains not detected
9681 as SLP reduction chain, specifically not properly associated
9682 chains containing a mix of plus/minus.
9683 (get_reduction_op): Remove.
9684 (get_initial_defs_for_reduction): Simplify, pass in whether
9685 this is a reduction chain, pass in the SLP node for the PHIs.
9686 (vect_create_epilog_for_reduction): Get the SLP instance as
9688 (vectorizable_reduction): Get the SLP instance as arg.
9689 During analysis remember the SLP node with the PHIs in the
9690 instance. Simplify getting at the vectorized reduction PHIs.
9691 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
9692 through SLP instance.
9693 (vect_slp_analyze_operations): Likewise.
9694 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
9695 (vect_transform_stmt): Likewise.
9697 2017-07-20 Tom de Vries <tom@codesourcery.com>
9699 PR tree-optimization/81489
9700 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
9701 read of phi arg location to before loop that modifies phi.
9703 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
9705 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
9708 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
9711 * except.c (execute): Fix ordering issue.
9713 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
9715 PR rtl-optimization/81423
9716 * combine.c (make_compound_operation_int): Don't try to optimize
9717 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
9719 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
9721 PR rtl-optimization/81423
9722 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
9723 with a constant that is -1 in the truncated to mode.
9725 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
9727 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
9728 (determine_unlikely_bbs): ... here.
9729 * predict.h (propagate_unlikely_bbs_forward): Declare.
9730 * cfgexpand.c (pass_expand::execute): Use it.
9731 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
9733 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
9734 propagate_unlikely_bbs_forward.
9736 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
9739 * except.c (maybe_add_nop_after_section_switch): New function.
9742 2017-07-19 Tom de Vries <tom@codesourcery.com>
9744 * gimple.h (gimple_phi_set_arg): Make assert more strict.
9746 2017-07-19 Tom de Vries <tom@codesourcery.com>
9748 * gimple.h (gimple_phi_arg): Make assert more strict.
9750 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
9752 * config.gcc (powerpc*-*-*): Add mmintrin.h.
9753 * config/rs6000/mmintrin.h: New file.
9754 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
9756 2017-07-19 Jakub Jelinek <jakub@redhat.com>
9758 PR tree-optimization/81346
9759 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
9761 2017-07-19 Tom de Vries <tom@codesourcery.com>
9763 * config/nvptx/nvptx.md (VECIM): Add V2DI.
9765 2017-07-19 Tom de Vries <tom@codesourcery.com>
9767 * config/nvptx/nvptx-modes.def: Add V2DImode.
9768 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
9769 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
9770 (nvptx_output_mov_insn): Handle lack of mov.b128.
9771 (nvptx_print_operand): Handle 'H' and 'L' codes.
9772 (nvptx_vector_mode_supported): Allow V2DImode.
9773 (nvptx_preferred_simd_mode): New function.
9774 (nvptx_data_alignment): New function.
9775 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
9776 nvptx_preferred_simd_mode.
9777 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
9779 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
9781 2017-07-19 Tom de Vries <tom@codesourcery.com>
9783 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
9784 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
9785 (nvptx_vector_mode_supported): New function. Allow V2SImode.
9786 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
9787 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
9788 (mov<VECIM>_insn): New define_insn.
9789 (define_expand "mov<VECIM>): New define_expand.
9791 2017-07-19 Tom de Vries <tom@codesourcery.com>
9793 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
9795 2017-07-19 Jakub Jelinek <jakub@redhat.com>
9797 PR tree-optimization/81346
9798 * fold-const.h (fold_div_compare, range_check_type): Declare.
9799 * fold-const.c (range_check_type): New function.
9800 (build_range_check): Use range_check_type.
9801 (fold_div_compare): No longer static, rewritten into
9802 a match.pd helper function.
9803 (fold_comparison): Don't call fold_div_compare here.
9804 * match.pd (X / C1 op C2): New optimization using fold_div_compare
9807 2017-07-19 Nathan Sidwell <nathan@acm.org>
9809 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
9810 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
9811 * tree.c (find_decls_types_r, verify_type): Use
9812 TYPE_{MIN,MAX}_VALUE_RAW.
9813 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
9814 (hash_tree): Likewise.
9815 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
9817 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
9820 2017-07-18 Tom de Vries <tom@codesourcery.com>
9823 * omp-expand.c (expand_omp_for_static_chunk): Handle
9824 equal-argument loop exit phi.
9826 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
9829 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
9830 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
9831 operand 2 predicate.
9832 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
9833 operand 2 predicate.
9834 (ror,rol -> rorx splitters): Use const_int_operand as
9835 operand 2 predicate.
9837 2017-06-18 Richard Biener <rguenther@suse.de>
9839 PR tree-optimization/81410
9840 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
9841 the gap in the ! slp_perm SLP case after each group.
9843 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
9846 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
9849 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
9852 * predict.c (set_even_probabilities): Cleanup; do not affect
9853 probabilities that are already known.
9854 (combine_predictions_for_bb): Call even when count is set.
9856 2017-07-18 Nathan Sidwell <nathan@acm.org>
9858 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
9861 2017-07-18 Bin Cheng <bin.cheng@arm.com>
9864 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
9865 optimization for loop niter analysis.
9867 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
9870 * config/avr/avr.c (avr_optimize_casesi): Don't use
9871 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
9873 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
9875 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
9876 body_cost_vec from _vect_peel_extended_info.
9877 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
9878 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
9881 2017-07-18 Bin Cheng <bin.cheng@arm.com>
9883 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
9885 2017-07-18 Richard Biener <rguenther@suse.de>
9887 PR tree-optimization/80620
9888 PR tree-optimization/81403
9889 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
9890 info when re-using a VN table entry.
9892 2017-07-18 Richard Biener <rguenther@suse.de>
9894 PR tree-optimization/81418
9895 * tree-vect-loop.c (vectorizable_reduction): Properly compute
9896 vectype_in. Verify that with lane-reducing reduction operations
9897 we have a single def-use cycle.
9899 2017-07-17 Carl Love <cel@us.ibm.com>
9901 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
9903 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9904 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
9905 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
9906 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
9907 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
9908 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
9909 VMULOSW): New enum "unspec" values.
9910 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
9911 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
9912 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
9913 altivec_vmulosw): New patterns.
9914 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
9915 VMULOSW): Add definitions.
9916 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
9918 * config/alpha/alpha.c: Include predict.h.
9920 2017-07-17 Yury Gribov <tetra2005@gmail.com>
9922 * tree-vrp.c (compare_assert_loc): Fix comparison function
9923 to return predictable results.
9925 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
9927 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
9930 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
9931 * doc/invoke.texi (mexpand-adddi): Update text.
9933 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
9935 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
9936 that also clobbers the CC register. The old expand code is moved
9938 (*arc_clzsi2): ... here.
9939 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
9940 the CC register. The old expand code is moved to ...
9941 (arc_ctzsi2): ... here.
9943 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
9945 * config/arc/arc.opt (mindexed-loads): Use initial value
9946 TARGET_INDEXED_LOADS_DEFAULT.
9947 (mauto-modify-reg): Use initial value
9948 TARGET_AUTO_MODIFY_REG_DEFAULT.
9949 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
9950 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
9951 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
9952 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
9954 2017-07-17 Martin Liska <mliska@suse.cz>
9957 * opts.c (finish_options): Do not allow -fgnu-tm
9958 w/ -fsanitize={kernel-,}address. Say sorry.
9960 2017-07-17 Bin Cheng <bin.cheng@arm.com>
9963 * tree-loop-distribution.c (classify_partition): Only assert on
9964 numer of iterations.
9965 (merge_dep_scc_partitions): Delete prameter. Update function call.
9966 (distribute_loop): Remove code handling loop with unknown niters.
9967 (pass_loop_distribution::execute): Skip loop with unknown niters.
9969 2017-07-17 Bin Cheng <bin.cheng@arm.com>
9972 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
9973 function sort_partitions_by_post_order.
9975 2017-07-17 Bin Cheng <bin.cheng@arm.com>
9977 PR tree-optimization/81374
9978 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
9979 the max index of basic blocks, rather than number of basic blocks.
9981 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
9983 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
9985 (arc_legitimate_pic_operand_p): Likewise.
9986 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
9988 (arc_needs_pcl_p): Likewise.
9989 (arc_legitimate_pc_offset_p): Likewise.
9990 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
9991 function is also used in constrains.md.
9992 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
9993 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
9994 PLUS. Only return true/false in known cases, otherwise assert.
9995 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
9996 is already called in arc_legitimate_constant_p.
9997 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
9999 (LEGITIMATE_PIC_OPERAND_P): Use
10000 arc_raw_symbolic_reference_mentioned_p function.
10001 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
10006 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
10007 Andrew Burgess <andrew.burgess@embecosm.com>
10009 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
10010 (arc_return_address_register): New function.
10011 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
10012 (arc_handle_fndecl_attribute): Add naked attribute.
10013 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
10014 (TARGET_WARN_FUNC_RETURN): Likewise.
10015 (arc_allocate_stack_slots_for_args): New function.
10016 (arc_warn_func_return): Likewise.
10017 (machine_function): Change type fn_type.
10018 (arc_compute_function_type): Consider new naked function type,
10019 change function return type.
10020 (arc_must_save_register): Adapt to handle new
10021 arc_compute_function_type's return type.
10022 (arc_expand_prologue): Likewise.
10023 (arc_expand_epilogue): Likewise.
10024 (arc_return_address_regs): Delete.
10025 (arc_return_address_register): New function.
10026 (arc_epilogue_uses): Use above function.
10027 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
10028 (arc_function_type): Change encoding, add naked type.
10029 (ARC_INTERRUPT_P): Change to handle the new encoding.
10030 (ARC_FAST_INTERRUPT_P): Likewise.
10031 (ARC_NORMAL_P): Define.
10032 (ARC_NAKED_P): Likewise.
10033 (arc_compute_function_type): Delete prototype.
10034 * config/arc/arc.md (in_ret_delay_slot): Use
10035 arc_return_address_register function.
10036 (simple_return): Likewise.
10037 (p_return_i): Likewise.
10039 2017-07-17 Jakub Jelinek <jakub@redhat.com>
10041 PR tree-optimization/81428
10042 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
10043 can't be built for those types.
10045 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
10047 Remove stuff dead since r239246.
10049 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
10050 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
10051 (avr_inform_devices): Remove dead stuff.
10053 2017-07-17 Tamar Christina <tamar.christina@arm.com>
10055 * config/arm/arm_neon.h: Fix softp typo.
10057 2017-07-17 Jakub Jelinek <jakub@redhat.com>
10059 PR tree-optimization/81365
10060 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
10061 aggregate moves onto bb predecessor edges, make sure there are no
10062 loads that could alias the lhs in between the start of bb and the
10065 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
10068 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
10069 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
10070 [LSHIFTRT, outer_code = TRUNCATE]: Same.
10072 2017-07-17 Jakub Jelinek <jakub@redhat.com>
10074 PR tree-optimization/81396
10075 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
10076 (init_symbolic_number): Initialize it to 1.
10077 (perform_symbolic_merge): Add n_ops from both operands into the new
10079 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
10080 without base_addr as useless if they need more than one operation.
10081 (bswap_replace): Handle !bswap case for NULL base_addr.
10083 2017-07-17 Tom de Vries <tom@codesourcery.com>
10086 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
10089 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
10091 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
10092 conditional builtin define __FIX_LEON3FT_B2BST.
10094 2017-07-17 Daniel Cederman <cederman@gaisler.com>
10096 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
10097 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
10100 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
10102 PR rtl-optimization/81424
10103 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
10104 to remove potential trapping from operands if -fnon-call-exceptions.
10106 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10108 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
10109 profile_proability for scalling.
10110 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
10112 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10114 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
10116 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10118 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
10119 fixpoint arithmetics.
10121 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10123 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
10124 fixpoint arithmetics.
10126 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10128 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
10129 fixpoint arithmetics.
10131 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
10133 * profile-count.h (profile_probability::from_reg_br_prob_note,
10134 profile_probability::to_reg_br_prob_note): New functions.
10135 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
10136 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
10137 * predict.c (probability_reliable_p): Update.
10138 (edge_probability_reliable_p): Update.
10139 (br_prob_note_reliable_p): Update.
10140 (invert_br_probabilities): Update.
10141 (add_reg_br_prob_note): New function.
10142 (combine_predictions_for_insn): Update.
10143 * asan.c (asan_clear_shadow): Update.
10144 * cfgbuild.c (compute_outgoing_frequencies): Update.
10145 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
10146 (update_br_prob_note): Update.
10147 (rtl_verify_edges): Update.
10148 (purge_dead_edges): Update.
10149 (fixup_reorder_chain): Update.
10150 * emit-rtl.c (try_split): Update.
10151 * ifcvt.c (cond_exec_process_insns): Update.
10152 (cond_exec_process_if_block): Update.
10153 (dead_or_predicable): Update.
10154 * internal-fn.c (expand_addsub_overflow): Update.
10155 (expand_neg_overflow): Update.
10156 (expand_mul_overflow): Update.
10157 * loop-doloop.c (doloop_modify): Update.
10158 * loop-unroll.c (compare_and_jump_seq): Update.
10159 * optabs.c (emit_cmp_and_jump_insn_1): Update.
10160 * predict.h: Update.
10161 * reorg.c (mostly_true_jump): Update.
10163 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
10164 * config/alpha/alpha.c (emit_unlikely_jump): Update.
10165 * config/arc/arc.c: (emit_unlikely_jump): Update.
10166 * config/arm/arm.c: (emit_unlikely_jump): Update.
10167 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
10168 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
10169 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
10170 (ix86_print_operand): Update.
10171 (ix86_split_fp_branch): Update.
10172 (predict_jump): Update.
10173 * config/ia64/ia64.c (ia64_print_operand): Update.
10174 * config/mmix/mmix.c (mmix_print_operand): Update.
10175 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
10176 (rs6000_expand_split_stack_prologue): Update.
10177 * config/rs6000/rs6000.c: Update.
10178 * config/s390/s390.c (s390_expand_vec_strlen): Update.
10179 (s390_expand_vec_movstr): Update.
10180 (s390_expand_cs_tdsi): Update.
10181 (s390_expand_split_stack_prologue): Update.
10182 * config/sh/sh.c (sh_print_operand): Update.
10183 (expand_cbranchsi4): Update.
10184 (expand_cbranchdi4): Update.
10185 * config/sparc/sparc.c (output_v9branch): Update.
10186 * config/spu/spu.c (get_branch_target): Update.
10187 (ea_load_store_inline): Update.
10188 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
10189 * config/tilepro/tilepro.c: Update.
10191 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
10193 * gimplify.c (mostly_copy_tree_r): Revert latest change.
10194 (gimplify_save_expr): Likewise.
10196 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
10198 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
10200 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
10202 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
10204 * timevar.def (TV_IPA_FNSUMMARY): Define.
10206 2017-07-16 Daniel Cederman <cederman@gaisler.com>
10208 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
10209 to back store errata sensitive sequence from being generated.
10210 (sqrtdf2_fix): Likewise.
10212 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
10214 * tree-ssa-threadupdate.c (compute_path_counts,
10215 update_joiner_offpath_counts): Use profile_probability.
10217 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
10220 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
10222 * config/arm/arm-c.c (arm_cpu_builtins): Define
10223 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
10225 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
10227 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10228 array entries to represent __ieee128 versions of the
10229 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
10230 scalar_extract_sig, and scalar_insert_exp built-in functions.
10231 (altivec_resolve_overloaded_builtin): Add special case handling
10232 for the __builtin_scalar_insert_exp function, as represented by
10233 the P9V_BUILTIN_VEC_VSIEDP constant.
10234 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
10235 exponent support for __ieee128 argument.
10236 (VSESQP): Add scalar extract signature support for __ieee128
10238 (VSTDCNQP): Add scalar test negative support for __ieee128
10240 (VSIEQP): Add scalar insert exponent support for __int128 argument
10241 with __ieee128 result.
10242 (VSIEQPF): Add scalar insert exponent support for __ieee128
10243 argument with __ieee128 result.
10244 (VSTDCQP): Add scalar test data class support for __ieee128
10246 (VSTDCNQP): Add overload support for scalar test negative with
10247 __ieee128 argument.
10248 (VSTDCQP): Add overload support for scalar test data class
10249 __ieee128 argument.
10250 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
10251 UNSPEC_VSX_SXSIGDP.
10252 (UNSPEC_VSX_SIEXPQP): New constant.
10253 (xsxexpqp): New insn for VSX scalar extract exponent quad
10255 (xsxsigqp): New insn for VSX scalar extract significand quad
10257 (xsiexpqpf): New insn for VSX scalar insert exponent quad
10258 precision with floating point argument.
10259 (xststdcqp): New expand for VSX scalar test data class quad
10261 (xststdcnegqp): New expand for VSX scalar test negative quad
10263 (xststdcqp): New insn to match expansions for VSX scalar test data
10264 class quad precision and VSX scalar test negative quad precision.
10265 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
10266 special case operand checking to enforce that second operand of
10267 VSX scalar test data class with quad precision argument is a 7-bit
10269 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
10270 prototypes and descriptions of __ieee128 versions of
10271 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
10272 scalar_test_data_class, and scalar_test_neg built-in functions.
10274 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10276 PR tree-optimization/81162
10277 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
10278 replace a negate with an add.
10280 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
10282 * doc/invoke.texi (arm/-mcpu): Document +crypto.
10284 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
10286 * config/arm/arm-c.c (arm_cpu_builtins): Define
10287 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
10289 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
10291 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
10292 (armv8-r): Set ARM Cortex-R52 as default CPU.
10293 * config/arm/arm-tables.opt: Regenerate.
10294 * config/arm/arm-tune.md: Regenerate.
10295 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
10297 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
10298 extension for -mcpu=cortex-r52.
10300 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
10302 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
10303 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
10304 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
10305 (fp-armv8): Define it as FP_ARMv8 only.
10306 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
10307 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
10309 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
10310 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
10311 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
10312 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
10313 than TARGET_FPU_ARMV8.
10314 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
10315 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
10316 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
10318 * config/arm/neon.md (neon_vrint): Likewise.
10319 (neon_vcvt): Likewise.
10320 (neon_<fmaxmin_op><mode>): Likewise.
10321 (<fmaxmin><mode>3): Likewise.
10322 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
10323 * config/arm/predicates.md (arm_cond_move_operator): Check against
10324 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
10326 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
10328 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
10329 to top of function.
10331 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10333 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
10334 loop in comment with memset.
10336 2017-07-14 Martin Liska <mliska@suse.cz>
10338 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
10339 * dwarf2out.c (is_java): Remove the function.
10340 (output_pubname): Remove usage of the function.
10341 (lower_bound_default): Remove usage of DW_LANG_Java.
10342 (gen_compile_unit_die): Likewise.
10343 * gcc.c: Remove compiler defaults for .java and .zip files.
10344 * gimple-expr.c (remove_suffix): Change as there's no longer
10345 extension than 4-letter one.
10346 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
10347 (gimplify_save_expr): Likewise.
10348 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
10349 as it's possible even for other languages than Java.
10350 * langhooks.h (struct lang_hooks): Remove Java from a comment.
10351 * lto-opts.c (lto_write_options): Remove reference to Java.
10352 * opts.c (strip_off_ending): Update file extension handling.
10353 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
10354 * tree-eh.c (lower_resx): Likewise.
10355 * tree.c (free_lang_data_in_type): Remove dead code.
10356 (find_decls_types_r): Likewise.
10357 (build_common_builtin_nodes): Remove Java from a comment.
10358 (verify_type): Remove dead code.
10359 * varasm.c (assemble_external): Remove Java from a comment.
10361 2017-07-14 Martin Liska <mliska@suse.cz>
10363 * opts.c (finish_options): Add quotes.
10364 (common_handle_option): Likewise.
10366 2017-07-14 Martin Liska <mliska@suse.cz>
10368 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
10369 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
10370 Remove N_SO_PASCAL.
10371 * dwarf2out.c (lower_bound_default): Do not handle
10373 (gen_compile_unit_die): Likewise.
10374 * gcc.c: Remove default extension binding for GNU Pascal.
10375 * stmt.c: Remove Pascal language from a comment.
10376 * xcoffout.c: Likewise.
10378 2017-07-13 David Malcolm <dmalcolm@redhat.com>
10381 * diagnostic-show-locus.c (fixit_cmp): New function.
10382 (layout::layout): Sort m_fixit_hints.
10383 (column_range::column_range): Assert that the values are valid.
10384 (struct char_span): New struct.
10385 (correction::overwrite): New method.
10386 (struct source_line): New struct.
10387 (line_corrections::add_hint): Add assertions. Reimplement memcpy
10388 calls in terms of classes source_line and char_span, and
10389 correction::overwrite.
10390 (selftest::test_overlapped_fixit_printing_2): New function.
10391 (selftest::diagnostic_show_locus_c_tests): Call it.
10393 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
10395 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
10396 early if there is no lhs.
10398 2017-07-13 Martin Liska <mliska@suse.cz>
10400 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
10401 (gen_reference_type_die): Likewise.
10402 * stor-layout.c: Remove Pascal-related comment.
10404 2017-07-13 Martin Liska <mliska@suse.cz>
10406 * opts.c (finish_options): Add quotes to error messages.
10407 (parse_sanitizer_options): Likewise.
10409 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
10411 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
10413 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
10415 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
10417 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
10419 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
10421 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
10423 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
10426 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
10427 provides the hardware capability bits, define the macro
10428 __BUILTIN_CPU_SUPPORTS__.
10429 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
10430 if GLIBC does not provide the hardware capability bits. Add a
10431 gcc_unreachable call if the built-in cpu function is neither
10432 __builtin_cpu_is nor __builtin_cpu_supports.
10433 (rs6000_get_function_versions_dispatcher): Change the warning
10434 that an old GLIBC is used which does not export the capability
10435 bits to be an error.
10436 * doc/extend.texi (target_clones attribute): Document the
10437 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
10438 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
10439 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
10440 the macros defined by GCC if the newer GLIBC is available.
10442 2017-07-12 Jeff Law <law@redhat.com>
10444 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
10445 remaining includes slightly.
10446 * config/riscv/riscv-builtins.c: Include profile-count.h.
10448 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
10451 * config/avr/avr.c (avr_set_current_function): In diagnostic
10452 messages: Quote keywords and (parts of) identifiers.
10453 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
10456 2017-07-12 Carl Love <cel@us.ibm.com>
10458 * config/rs6000/rs6000-c.c: Add support for built-in functions
10459 vector bool char vec_revb (vector bool char);
10460 vector bool short vec_revb (vector short char);
10461 vector bool int vec_revb (vector bool int);
10462 vector bool long long vec_revb (vector bool long long);
10463 * doc/extend.texi: Update the built-in documentation file for the
10464 new built-in functions.
10466 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10468 * config/s390/s390.md: Remove movcc splitter.
10470 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10472 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
10473 load/store on condition.
10475 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
10478 * config/avr/avr.c (avr_encode_section_info)
10479 [progmem && !TREE_READONLY]: Error if progmem object needs
10482 2017-07-11 Michael Collison <michael.collison@arm.com>
10484 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
10487 2017-07-11 Carl Love <cel@us.ibm.com>
10489 * config/rs6000/rs6000-c.c: Add support for builtins
10490 vector unsigned int vec_parity_lsbb (vector signed int);
10491 vector unsigned int vec_parity_lsbb (vector unsigned int);
10492 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
10493 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
10494 vector unsigned long long vec_parity_lsbb (vector signed long long);
10495 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
10496 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
10497 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
10498 * doc/extend.texi: Update the built-in documentation file for the
10499 new built-in functions.
10501 2017-07-11 David Malcolm <dmalcolm@redhat.com>
10503 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
10504 (layout::m_primary_loc): New field.
10505 (layout::layout): Initialize new field. Move location filtering
10506 logic from here to...
10507 (layout::maybe_add_location_range): ...this new method. Add
10508 support for filtering to just the lines already specified by other
10510 (layout::will_show_line_p): New method.
10511 (gcc_rich_location::add_location_if_nearby): New method.
10512 (selftest::test_add_location_if_nearby): New test function.
10513 (selftest::diagnostic_show_locus_c_tests): Call it.
10514 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
10517 2017-07-11 Tom de Vries <tom@codesourcery.com>
10519 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
10520 (bb_first_real_insn): New function.
10521 (nvptx_single): Add extra initialization of broadcasted condition
10524 2017-07-11 Nathan Sidwell <nathan@acm.org>
10526 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
10528 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
10530 * doc/extend.texi (AVR Function Attributes): Remove weblink to
10531 Binutils doc as TEXI will mess them up.
10532 * doc/invoke.texi (AVR Options): Same here.
10534 2017-07-11 Daniel Cederman <cederman@gaisler.com>
10536 * config/sparc/sparc.opt (mfix-ut700): New option.
10537 (mfix-gr712rc): Likewise.
10538 (sparc_fix_b2bst): New variable.
10539 * doc/invoke.texi (SPARC options): Document them.
10540 (ARM options): Fix warnings.
10541 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
10542 instructions to prevent sequences that can trigger the store-store
10543 errata for certain LEON3FT processors.
10544 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
10545 (sparc_option_override): Set sparc_fix_b2bst appropriately.
10546 * config/sparc/sparc.md (fix_b2bst): New attribute.
10547 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
10549 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
10552 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
10554 (*rsqrtsf2_sse): Ditto.
10556 (div<mode>3): Macroize insn from divdf3 and divsf3
10557 using MODEF mode iterator.
10559 2017-07-10 Martin Sebor <msebor@redhat.com>
10561 PR tree-optimization/80397
10562 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
10563 instead of testing for equality to INTEGER_TYPE.
10565 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
10567 * config.gcc: Remove uclibc from arc target spec.
10569 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
10571 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
10573 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
10576 * lto-wrapper.c (remove_option): New function.
10577 (merge_and_complain): Merge PIC/PIE options more realistically.
10579 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
10581 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
10585 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
10586 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
10587 * config.in: Regenerate.
10588 * configure: Regenerate.
10589 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
10590 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
10591 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
10592 (TARGET_GASISR_PROLOGUES): ...target mask.
10593 * common/config/avr/avr-common.c
10594 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
10595 Set -mgas-isr-prologues.
10596 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
10597 INSERT_PASS_BEFORE for it.
10598 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
10599 * config/avr/avr.c (avr_option_override)
10600 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
10601 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
10602 (avr_attribute_table) <no_gccisr>: Add new function attribute.
10603 (avr_set_current_function) <is_no_gccisr>: Init machine field.
10604 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
10606 (make_avr_pass_pre_proep): New function.
10607 (emit_push_sfr) <treg>: Add argument to function and use it
10608 instead of TMP_REG.
10609 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
10610 and set machine->gasisr.yes.
10611 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
10612 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
10613 __gcc_isr.n_pushed to .L__stack_usage.
10614 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
10615 (avr_asm_final_postscan_insn): ...this new static function.
10616 * config/avr/avr.h (machine_function)
10617 <is_no_gccisr, use_L__stack_usage>: New fields.
10618 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
10619 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
10620 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
10621 (gasisr, *gasisr): New expander and insn.
10622 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
10623 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
10624 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
10626 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
10628 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
10631 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
10633 Move jump-tables out of .text again.
10636 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
10637 (ASM_OUTPUT_ADDR_VEC): New function.
10638 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
10639 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
10640 INSN_ADDRESSes as asm comment.
10641 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
10642 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
10643 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
10644 * config/avr/avr.md (*tablejump): Adjust comment.
10645 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
10646 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
10648 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
10649 (avr_output_addr_vec): New proto.
10650 (avr_log_t) <insn_addresses>: New field.
10652 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
10655 * config/i386/i386.c (ix86_function_arg_advance): Set
10656 outgoing_args_on_stack to true if there are outgoing arguments
10658 (ix86_function_arg): Likewise.
10659 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
10660 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
10661 * config/i386/i386.h (machine_function): Add
10662 outgoing_args_on_stack.
10664 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
10666 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
10667 supporting pthreds.
10668 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
10670 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
10672 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
10673 (REAL_H): Remove $(MACHMODE_H).
10674 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
10676 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
10677 $(MACHMODE_H) and double-int.h.
10678 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
10680 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
10683 2017-07-07 Andrew Pinski <apinski@cavium.com>
10685 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
10686 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
10688 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
10690 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
10691 Add warning if GCC was not configured to link against a GLIBC that
10692 exports the hardware capability bits.
10693 (make_resolver_func): Make resolver function private and not a
10694 COMDAT function. Create the name with clone_function_name instead
10695 of make_unique_name.
10698 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
10699 correct operand in doing the split.
10701 2017-07-07 Carl Love <cel@us.ibm.com>
10703 * config/rs6000/rs6000-c: Add support for built-in function
10704 vector unsigned short vec_pack_to_short_fp32 (vector float,
10706 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
10707 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
10708 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
10709 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
10710 (convert_4f32_8i16): Add define_expand.
10711 * doc/extend.texi: Update the built-in documentation file for the
10712 new built-in function.
10714 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10716 * config/sparc/m8.md: New file.
10717 * config/sparc/sparc.md: Include m8.md.
10719 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10721 * config/sparc/sparc.opt: New option -mvis4b.
10722 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
10723 (sparc_option_override): Handle VIS4B.
10724 (enum sparc_builtins): Define
10725 SPARC_BUILTIN_DICTUNPACK{8,16,32},
10726 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
10727 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
10728 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
10729 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
10730 (check_constant_argument): New function.
10731 (sparc_vis_init_builtins): Define builtins
10732 __builtin_vis_dictunpack{8,16,32},
10733 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
10734 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
10735 __builtin_vis_fpcmpde{8,16,32}shl and
10736 __builtin_vis_fpcmpur{8,16,32}shl.
10737 (sparc_expand_builtin): Check that the constant operands to
10738 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
10739 constant and in range.
10740 * config/sparc/sparc-c.c (sparc_target_macros): Handle
10742 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
10743 (SPARC_IMM5_P): Likewise.
10744 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
10745 (enabled): Handle vis4b.
10746 (UNSPEC_DICTUNPACK): New unspec.
10747 (UNSPEC_FPCMPSHL): Likewise.
10748 (UNSPEC_FPUCMPSHL): Likewise.
10749 (UNSPEC_FPCMPDESHL): Likewise.
10750 (UNSPEC_FPCMPURSHL): Likewise.
10751 (cpu_feature): New CPU feature `vis4b'.
10752 (dictunpack{8,16,32}): New insns.
10753 (FPCSMODE): New mode iterator.
10754 (fpcscond): New code iterator.
10755 (fpcsucond): Likewise.
10756 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
10757 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
10758 (fpcmpde{8,16,32}{si,di}shl): Likewise.
10759 (fpcmpur{8,16,32}{si,di}shl): Likewise.
10760 * config/sparc/constraints.md: Define constraints `q' for unsigned
10761 2-bit integer constants and `t' for unsigned 5-bit integer
10763 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
10765 (imm5_operand_dictunpack16): Likewise.
10766 (imm5_operand_dictunpack32): Likewise.
10767 (imm2_operand): Likewise.
10768 * doc/invoke.texi (SPARC Options): Document -mvis4b.
10769 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
10770 ditunpack* and fpcmp*shl builtins.
10772 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10774 * config.gcc: Handle m8 in --with-{cpu,tune} options.
10775 * config.in: Add HAVE_AS_SPARC6 define.
10776 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
10778 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
10780 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
10781 (CPP_CPU_SPEC): Handle m8.
10782 (ASM_CPU_SPEC): Likewise.
10783 * config/sparc/sparc-opts.h (enum processor_type): Add
10785 * config/sparc/sparc.c (m8_costs): New struct.
10786 (sparc_option_override): Handle TARGET_CPU_m8.
10787 (sparc32_initialize_trampoline): Likewise.
10788 (sparc64_initialize_trampoline): Likewise.
10789 (sparc_issue_rate): Likewise.
10790 (sparc_register_move_cost): Likewise.
10791 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
10792 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
10793 (ASM_CPU64_DEFAULT_SPEC): Likewise.
10794 (CPP_CPU_SPEC): Handle M8.
10795 (ASM_CPU_SPEC): Likewise.
10796 (AS_M8_FLAG): Define.
10797 * config/sparc/sparc.md: Add m8 to the cpu attribute.
10798 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
10799 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
10801 * configure: Regenerate.
10802 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
10805 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10807 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
10809 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
10810 ("*movdi_insn_sp32"): Do not set v3pipe.
10811 ("*movsi_insn"): Likewise.
10812 ("*movdi_insn_sp64"): Likewise.
10813 ("*movsf_insn"): Likewise.
10814 ("*movdf_insn_sp32"): Likewise.
10815 ("*movdf_insn_sp64"): Likewise.
10816 ("*zero_extendsidi2_insn_sp64"): Likewise.
10817 ("*sign_extendsidi2_insn"): Likewise.
10818 ("*mov<VM32:mode>_insn"): Likewise.
10819 ("*mov<VM64:mode>_insn_sp64"): Likewise.
10820 ("*mov<VM64:mode>_insn_sp32"): Likewise.
10821 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
10822 ("<vlop:code><VL:mode>3"): Likewise.
10823 ("*not_<vlop:code><VL:mode>3"): Likewise.
10824 ("*nand<VL:mode>_vis"): Likewise.
10825 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
10826 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
10827 ("one_cmpl<VL:mode>2"): Likewise.
10828 ("faligndata<VM64:mode>_vis"): Likewise.
10829 ("alignaddrsi_vis"): Likewise.
10830 ("alignaddrdi_vis"): Likweise.
10831 ("alignaddrlsi_vis"): Likewise.
10832 ("alignaddrldi_vis"): Likewise.
10833 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
10834 ("bmaskdi_vis"): Likewise.
10835 ("bmasksi_vis"): Likewise.
10836 ("bshuffle<VM64:mode>_vis"): Likewise.
10837 ("cmask8<P:mode>_vis"): Likewise.
10838 ("cmask16<P:mode>_vis"): Likewise.
10839 ("cmask32<P:mode>_vis"): Likewise.
10840 ("pdistn<P:mode>_vis"): Likewise.
10841 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
10843 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10845 * config/sparc/sparc.md ("subtype"): New insn attribute.
10846 ("*wrgsr_sp64"): Set insn subtype.
10847 ("*rdgsr_sp64"): Likewise.
10848 ("alignaddrsi_vis"): Likewise.
10849 ("alignaddrdi_vis"): Likewise.
10850 ("alignaddrlsi_vis"): Likewise.
10851 ("alignaddrldi_vis"): Likewise.
10852 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
10853 ("fexpand_vis"): Likewise.
10854 ("fpmerge_vis"): Likewise.
10855 ("faligndata<VM64:mode>_vis"): Likewise.
10856 ("bshuffle<VM64:mode>_vis"): Likewise.
10857 ("cmask8<P:mode>_vis"): Likewise.
10858 ("cmask16<P:mode>_vis"): Likewise.
10859 ("cmask32<P:mode>_vis"): Likewise.
10860 ("fchksm16_vis"): Likewise.
10861 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
10862 ("fmean16_vis"): Likewise.
10863 ("fp<plusminus_insn>64_vis"): Likewise.
10864 ("<plusminus_insn>v8qi3"): Likewise.
10865 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
10866 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
10867 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
10868 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
10869 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
10870 ("*movqi_insn"): Likewise.
10871 ("*movhi_insn"): Likewise.
10872 ("*movsi_insn"): Likewise.
10873 ("movsi_pic_gotdata_op"): Likewise.
10874 ("*movdi_insn_sp32"): Likewise.
10875 ("*movdi_insn_sp64"): Likewise.
10876 ("movdi_pic_gotdata_op"): Likewise.
10877 ("*movsf_insn"): Likewise.
10878 ("*movdf_insn_sp32"): Likewise.
10879 ("*movdf_insn_sp64"): Likewise.
10880 ("*zero_extendhisi2_insn"): Likewise.
10881 ("*zero_extendqihi2_insn"): Likewise.
10882 ("*zero_extendqisi2_insn"): Likewise.
10883 ("*zero_extendqidi2_insn"): Likewise.
10884 ("*zero_extendhidi2_insn"): Likewise.
10885 ("*zero_extendsidi2_insn_sp64"): Likewise.
10886 ("ldfsr"): Likewise.
10887 ("prefetch_64"): Likewise.
10888 ("prefetch_32"): Likewise.
10889 ("tie_ld32"): Likewise.
10890 ("tie_ld64"): Likewise.
10891 ("*tldo_ldub_sp32"): Likewise.
10892 ("*tldo_ldub1_sp32"): Likewise.
10893 ("*tldo_ldub2_sp32"): Likewise.
10894 ("*tldo_ldub_sp64"): Likewise.
10895 ("*tldo_ldub1_sp64"): Likewise.
10896 ("*tldo_ldub2_sp64"): Likewise.
10897 ("*tldo_ldub3_sp64"): Likewise.
10898 ("*tldo_lduh_sp32"): Likewise.
10899 ("*tldo_lduh1_sp32"): Likewise.
10900 ("*tldo_lduh_sp64"): Likewise.
10901 ("*tldo_lduh1_sp64"): Likewise.
10902 ("*tldo_lduh2_sp64"): Likewise.
10903 ("*tldo_lduw_sp32"): Likewise.
10904 ("*tldo_lduw_sp64"): Likewise.
10905 ("*tldo_lduw1_sp64"): Likewise.
10906 ("*tldo_ldx_sp64"): Likewise.
10907 ("*mov<VM32:mode>_insn"): Likewise.
10908 ("*mov<VM64:mode>_insn_sp64"): Likewise.
10909 ("*mov<VM64:mode>_insn_sp32"): Likewise.
10911 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10913 * config/sparc/sparc.md ("type"): New insn type viscmp.
10914 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
10916 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
10917 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
10918 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
10919 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
10921 ("n7_vis_logical_11cycle"): Likewise.
10922 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
10923 * config/sparc/niagara2.md ("niag3_vis": Likewise.
10924 * config/sparc/niagara.md ("niag_vis"): Likewise.
10925 * config/sparc/ultra3.md ("us3_fga"): Likewise.
10926 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
10928 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10930 * config/sparc/sparc.md: New instruction type `bmask'.
10931 (bmaskdi_vis): Use the `bmask' type.
10932 (bmasksi_vis): Likewise.
10933 * config/sparc/ultra3.md (us3_array): Likewise.
10934 * config/sparc/niagara7.md (n7_array): Likewise.
10935 * config/sparc/niagara4.md (n4_array): Likewise.
10936 * config/sparc/niagara2.md (niag2_vis): Likewise.
10937 (niag3_vis): Likewise.
10938 * config/sparc/niagara.md (niag_vis): Likewise.
10940 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
10942 * ipa-comdats.c: Remove optimize check from gate.
10943 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
10944 for functions not optimized.
10945 (ipa_fn_summary_read): Skip optimize check.
10946 (ipa_fn_summary_write): Likewise.
10947 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
10949 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
10951 (can_inline_edge_p): Check flag_pcc_struct_return for match.
10952 (check_callers): Give up on caller which is not optimized.
10953 (inline_small_functions): Likewise.
10954 (ipa_inline): Do not give up when not optimizing.
10955 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
10956 away unoptimizes cdtors.
10957 (whole_program_function_and_variable_visibility): Do
10958 ipa_discover_readonly_nonaddressable_vars in LTO mode.
10959 * ipa.c (process_references): Do not check optimize.
10960 (symbol_table::remove_unreachable_nodes): Update optimize check.
10961 (set_writeonly_bit): Update optimize check.
10962 (pass_ipa_cdtor_merge::gate): Do not check optimize.
10963 (pass_ipa_single_use::gate): Remove.
10965 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
10967 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
10968 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
10969 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
10970 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
10971 permute_load, permute_store, adjust_extract, adjust_splat,
10972 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
10973 replace_swap_with_copy, dump_swap_insn_table,
10974 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
10975 recombine_lvx_pattern, recombine_stvx_pattern,
10976 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
10977 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
10978 to file rs6000-p8swap.c.
10979 * config/rs6000/rs6000-p8swap.c: New file.
10980 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
10981 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
10982 and rs6000*-*-* targets.
10984 2017-07-06 David Malcolm <dmalcolm@redhat.com>
10986 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
10988 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
10990 * lto-wrapper.c (merge_and_complain): Do not merge
10991 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
10992 fsigned_zeros, ftrapping_math, fwrapv.
10993 (append_compiler_options): Do not track these options.
10994 (append_linker_options): Likewie
10996 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
10998 * cgraphunit.c (cgraph_node::finalize_function): When
10999 !flag_toplevel_reorde set no_reorder flag.
11000 (varpool_node::finalize_decl): Likewise.
11001 (symbol_table::compile): Drop no toplevel reorder path.
11003 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
11005 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
11006 edges; zero probability is not better than uninitialized.
11008 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
11010 * asan.h (asan_sanitize_allocas_p): Declare.
11011 * asan.c (asan_sanitize_allocas_p): New function.
11012 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
11013 (handle_builtin_alloca): Likewise.
11014 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
11015 if !asan_sanitize_allocas_p.
11016 * params.def (asan-instrument-allocas): Add new option.
11017 * params.h (ASAN_PROTECT_ALLOCAS): Define.
11018 * opts.c (common_handle_option): Disable allocas sanitization for
11021 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
11023 * asan.c: Include gimple-fold.h.
11024 (get_last_alloca_addr): New function.
11025 (handle_builtin_stackrestore): Likewise.
11026 (handle_builtin_alloca): Likewise.
11027 (asan_emit_allocas_unpoison): Likewise.
11028 (get_mem_refs_of_builtin_call): Add new parameter, remove const
11029 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
11030 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
11031 (instrument_builtin_call): Pass gimple iterator to
11032 get_mem_refs_of_builtin_call.
11033 (last_alloca_addr): New global.
11034 * asan.h (asan_emit_allocas_unpoison): Declare.
11035 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
11036 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
11037 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
11038 if function calls alloca.
11039 * gimple-fold.c (replace_call_with_value): Remove static keyword.
11040 * gimple-fold.h (replace_call_with_value): Declare.
11041 * internal-fn.c: Include asan.h.
11042 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
11043 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
11045 2017-07-06 David Malcolm <dmalcolm@redhat.com>
11047 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
11048 (C_SELFTEST_FLAGS): New.
11049 (CPP_SELFTEST_FLAGS): New.
11050 (SELFTEST_DEPS): New, from deps of s-selftest.
11051 (C_SELFTEST_DEPS): New, from deps of s-selftest.
11052 (CPP_SELFTEST_DEPS): New.
11053 (selftest): Add dependency on s-selftest-c++.
11054 (s-selftest): Rename to...
11055 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
11056 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
11057 than SELFTEST_FLAGS.
11058 (selftest-gdb): Rename to...
11059 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
11061 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
11062 (selftest-valgrind): Rename to...
11063 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
11065 (selftest-valgrind): Reintroduce as an alias for
11066 selftest-c-valgrind.
11067 (s-selftest-c++): New.
11068 (selftest-c++-gdb): New.
11069 (selftest-c++-valgrind): New.
11071 2017-07-06 Olivier Hainque <hainque@adacore.com>
11073 * gcc.c (process_command): When deciding if undefined variables
11074 should be ignored when processing specs, accept "gcc -v" as well.
11076 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
11078 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
11079 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
11081 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
11083 * config/arm/arm-cpus.in (armv8-r): Add new entry.
11084 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
11085 * config/arm/arm-tables.opt: Regenerate.
11086 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
11088 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
11090 2017-07-06 Carl Love <cel@us.ibm.com>
11092 * ChangeLog: Clean up from mid air collision
11094 2017-07-06 Carl Love <cel@us.ibm.com>
11096 * config/rs6000/rs6000-c.c: Add support for built-in functions
11097 vector signed int vec_subc (vector signed int, vector signed int);
11098 vector signed __int128 vec_subc (vector signed __int128,
11099 vector signed __int128);
11100 vector unsigned __int128 vec_subc (vector unsigned __int128,
11101 vector unsigned __int128);
11102 vector signed int vec_sube (vector signed int, vector signed int,
11103 vector signed int);
11104 vector unsigned int vec_sube (vector unsigned int,
11105 vector unsigned int,
11106 vector unsigned int);
11107 vector signed __int128 vec_sube (vector signed __int128,
11108 vector signed __int128,
11109 vector signed__int128);
11110 vector unsigned __int128 vec_sube (vector unsigned __int128,
11111 vector unsigned __int128,
11112 vector unsigned __int128);
11113 vector signed int vec_subec (vector signed int, vector signed int,
11114 vector signed int);
11115 vector unsigned int vec_subec (vector unsigned int,
11116 vector unsigned int,
11117 vector unsigned int);
11118 vector signed __int128 vec_subec (vector signed __int128,
11119 vector signed __int128,
11120 vector signed__int128);
11121 vector unsigned __int128 vec_subec (vector unsigned __int128,
11122 vector unsigned __int128,
11123 vector unsigned __int128);
11124 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
11125 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
11126 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
11127 BU_ALTIVEC_OVERLOAD_X definitions.
11128 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
11129 * doc/extend.texi: Update the built-in documentation file for the new
11130 built-in functions.
11132 2017-07-06 David Malcolm <dmalcolm@redhat.com>
11135 * diagnostic-show-locus.c (layout::layout): Use start and finish
11136 spelling location for the start and finish of each range.
11137 * genmatch.c (linemap_client_expand_location_to_spelling_point):
11138 Add unused aspect param.
11139 * input.c (expand_location_1): Add "aspect" param, and use it
11140 to access the correct part of the location.
11141 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
11143 (expand_location_to_spelling_point): Likewise.
11144 (linemap_client_expand_location_to_spelling_point): Add "aspect"
11145 param, and pass it to expand_location_1.
11147 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
11149 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
11150 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
11151 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
11152 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
11153 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
11154 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
11155 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
11156 _mm_maskz_getmant_ss): New intrinsics.
11157 (__builtin_ia32_getexpss128_mask): Changed to ...
11158 __builtin_ia32_getexpss128_round ... this.
11159 (__builtin_ia32_getexpsd128_mask): Changed to ...
11160 __builtin_ia32_getexpsd128_round ... this.
11161 * config/i386/i386-builtin-types.def
11162 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
11163 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
11164 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
11165 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
11166 __builtin_ia32_getmantss_mask_round): New builtins.
11167 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
11168 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
11169 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
11170 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
11171 * config/i386/sse.md
11172 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
11173 avx512f_sgetexp<mode><mask_scalar_name>
11174 <round_saeonly_scalar_name> ... this.
11175 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
11176 %0, %1, %2<round_saeonly_op3>}): Changed to ...
11177 vgetexp<ssescalarmodesuffix>
11178 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
11179 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
11180 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
11181 avx512f_vgetmant<mode><mask_scalar_name>
11182 <round_saeonly_scalar_name> ... this.
11183 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
11184 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
11185 vgetmant<ssescalarmodesuffix>
11186 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
11187 %0<mask_scalar_operand4>, %1, %2
11188 <round_saeonly_scalar_mask_op4>, %3} ... this.
11189 * config/i386/subst.md (mask_scalar_operand4,
11190 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
11191 round_saeonly_scalar_nimm_predicate): New subst attributes.
11193 2017-07-06 Julia Koval <julia.koval@intel.com>
11195 * config/i386/i386.c (ix86_erase_embedded_rounding):
11196 Remove code for old rounding pattern.
11198 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
11200 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
11202 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
11204 * doc/sourcebuild.texi (Test Directives, Variants of
11205 dg-require-support): Add documentation for dg-require-stack-check.
11207 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
11209 * config/i386/subst.md (mask_scalar, round_scalar,
11210 round_saeonly_scalar): New meta-templates.
11211 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
11212 round_scalar_mask_operand3, round_scalar_mask_op3,
11213 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
11214 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
11215 round_saeonly_scalar_constraint,
11216 round_saeonly_scalar_prefix): New subst attribute.
11217 * config/i386/sse.md
11218 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
11219 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
11220 <round_scalar_name> ... this.
11221 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
11222 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
11223 <round_scalar_name> ... this.
11224 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
11225 <sse>_vm<code><mode>3<mask_scalar_name>
11226 <round_saeonly_scalar_name> ... this.
11227 (v<plusminus_mnemonic><ssescalarmodesuffix>
11228 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
11229 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
11230 v<plusminus_mnemonic><ssescalarmodesuffix>
11231 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
11232 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
11233 (v<multdiv_mnemonic><ssescalarmodesuffix>
11234 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
11235 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
11236 v<multdiv_mnemonic><ssescalarmodesuffix>
11237 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
11238 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
11239 (v<maxmin_float><ssescalarmodesuffix>
11240 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
11241 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
11242 v<maxmin_float><ssescalarmodesuffix>
11243 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
11244 %0<mask_scalar_operand3>, %1, %<iptr>2
11245 <round_saeonly_scalar_mask_op3>} ... this.
11247 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
11249 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
11250 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
11252 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
11253 Alan Hayward <alan.hayward@arm.com>
11254 David Sherwood <david.sherwood@arm.com>
11256 * combine.c (simplify_if_then_else): Remove "enum" before
11258 * compare-elim.c (can_eliminate_compare): Likewise.
11259 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
11261 (aarch64_lookup_simd_builtin_type): Likewise.
11262 (aarch64_simd_builtin_type): Likewise.
11263 (aarch64_init_simd_builtin_types): Likewise.
11264 (aarch64_simd_expand_args): Likewise.
11265 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
11267 (aarch64_reverse_mask): Likewise.
11268 (aarch64_simd_emit_reg_reg_move): Likewise.
11269 (aarch64_gen_adjusted_ldpstp): Likewise.
11270 (aarch64_ccmp_mode_to_code): Likewise.
11271 (aarch64_operands_ok_for_ldpstp): Likewise.
11272 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
11273 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
11275 (aarch64_min_divisions_for_recip_mul): Likewise.
11276 (aarch64_reassociation_width): Likewise.
11277 (aarch64_get_condition_code_1): Likewise.
11278 (aarch64_simd_emit_reg_reg_move): Likewise.
11279 (aarch64_simd_attr_length_rglist): Likewise.
11280 (aarch64_reverse_mask): Likewise.
11281 (aarch64_operands_ok_for_ldpstp): Likewise.
11282 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
11283 (aarch64_gen_adjusted_ldpstp): Likewise.
11284 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
11286 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
11287 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
11288 (arm_lookup_simd_builtin_type): Likewise.
11289 (arm_simd_builtin_type): Likewise.
11290 (arm_init_simd_builtin_types): Likewise.
11291 (arm_expand_builtin_args): Likewise.
11292 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
11293 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
11294 (ft32_setup_incoming_varargs): Likewise.
11295 (ft32_function_arg): Likewise.
11296 (ft32_function_arg_advance): Likewise.
11297 (ft32_pass_by_reference): Likewise.
11298 (ft32_arg_partial_bytes): Likewise.
11299 (ft32_valid_pointer_mode): Likewise.
11300 (ft32_addr_space_pointer_mode): Likewise.
11301 (ft32_addr_space_legitimate_address_p): Likewise.
11302 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
11304 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
11305 (ix86_emit_outlined_ms2sysv_restore): Likewise.
11306 (iamcu_alignment): Likewise.
11307 (canonicalize_vector_int_perm): Likewise.
11308 (ix86_noce_conversion_profitable_p): Likewise.
11309 (ix86_mpx_bound_mode): Likewise.
11310 (ix86_operands_ok_for_move_multiple): Likewise.
11311 * config/microblaze/microblaze-protos.h
11312 (microblaze_expand_conditional_branch_reg): Likewise.
11313 * config/microblaze/microblaze.c
11314 (microblaze_expand_conditional_branch_reg): Likewise.
11315 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
11317 (rs6000_reassociation_width): Likewise.
11318 (rs6000_invalid_binary_op): Likewise.
11319 (fusion_p9_p): Likewise.
11320 (emit_fusion_p9_load): Likewise.
11321 (emit_fusion_p9_store): Likewise.
11322 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
11324 (riscv_hard_regno_mode_ok_p): Likewise.
11325 (riscv_address_insns): Likewise.
11326 (riscv_split_symbol): Likewise.
11327 (riscv_legitimize_move): Likewise.
11328 (riscv_function_value): Likewise.
11329 (riscv_hard_regno_nregs): Likewise.
11330 (riscv_expand_builtin): Likewise.
11331 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
11332 (riscv_build_integer): Likewise.
11333 (riscv_split_integer): Likewise.
11334 (riscv_legitimate_constant_p): Likewise.
11335 (riscv_cannot_force_const_mem): Likewise.
11336 (riscv_regno_mode_ok_for_base_p): Likewise.
11337 (riscv_valid_base_register_p): Likewise.
11338 (riscv_valid_offset_p): Likewise.
11339 (riscv_valid_lo_sum_p): Likewise.
11340 (riscv_classify_address): Likewise.
11341 (riscv_legitimate_address_p): Likewise.
11342 (riscv_address_insns): Likewise.
11343 (riscv_load_store_insns): Likewise.
11344 (riscv_force_binary): Likewise.
11345 (riscv_split_symbol): Likewise.
11346 (riscv_force_address): Likewise.
11347 (riscv_legitimize_address): Likewise.
11348 (riscv_move_integer): Likewise.
11349 (riscv_legitimize_const_move): Likewise.
11350 (riscv_legitimize_move): Likewise.
11351 (riscv_address_cost): Likewise.
11352 (riscv_subword): Likewise.
11353 (riscv_output_move): Likewise.
11354 (riscv_canonicalize_int_order_test): Likewise.
11355 (riscv_emit_int_order_test): Likewise.
11356 (riscv_function_arg_boundary): Likewise.
11357 (riscv_pass_mode_in_fpr_p): Likewise.
11358 (riscv_pass_fpr_single): Likewise.
11359 (riscv_pass_fpr_pair): Likewise.
11360 (riscv_get_arg_info): Likewise.
11361 (riscv_function_arg): Likewise.
11362 (riscv_function_arg_advance): Likewise.
11363 (riscv_arg_partial_bytes): Likewise.
11364 (riscv_function_value): Likewise.
11365 (riscv_pass_by_reference): Likewise.
11366 (riscv_setup_incoming_varargs): Likewise.
11367 (riscv_print_operand): Likewise.
11368 (riscv_elf_select_rtx_section): Likewise.
11369 (riscv_save_restore_reg): Likewise.
11370 (riscv_for_each_saved_reg): Likewise.
11371 (riscv_register_move_cost): Likewise.
11372 (riscv_hard_regno_mode_ok_p): Likewise.
11373 (riscv_hard_regno_nregs): Likewise.
11374 (riscv_class_max_nregs): Likewise.
11375 (riscv_memory_move_cost): Likewise.
11376 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
11377 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
11378 (rl78_addr_space_address_mode): Likewise.
11379 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11381 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
11382 (rs6000_reassociation_width): Likewise.
11383 (rs6000_invalid_binary_op): Likewise.
11384 (fusion_p9_p): Likewise.
11385 (emit_fusion_p9_load): Likewise.
11386 (emit_fusion_p9_store): Likewise.
11387 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
11388 (ok_for_simple_move_operands): Likewise.
11389 (ok_for_simple_move_strict_operands): Likewise.
11390 (ok_for_simple_arith_logic_operands): Likewise.
11391 (visium_legitimize_reload_address): Likewise.
11392 (visium_select_cc_mode): Likewise.
11393 (output_cbranch): Likewise.
11394 (visium_split_double_move): Likewise.
11395 (visium_expand_copysign): Likewise.
11396 (visium_expand_int_cstore): Likewise.
11397 (visium_expand_fp_cstore): Likewise.
11398 * config/visium/visium.c (visium_pass_by_reference): Likewise.
11399 (visium_function_arg): Likewise.
11400 (visium_function_arg_advance): Likewise.
11401 (visium_libcall_value): Likewise.
11402 (visium_setup_incoming_varargs): Likewise.
11403 (visium_legitimate_constant_p): Likewise.
11404 (visium_legitimate_address_p): Likewise.
11405 (visium_legitimize_address): Likewise.
11406 (visium_secondary_reload): Likewise.
11407 (visium_register_move_cost): Likewise.
11408 (visium_memory_move_cost): Likewise.
11409 (prepare_move_operands): Likewise.
11410 (ok_for_simple_move_operands): Likewise.
11411 (ok_for_simple_move_strict_operands): Likewise.
11412 (ok_for_simple_arith_logic_operands): Likewise.
11413 (visium_function_value_1): Likewise.
11414 (rtx_ok_for_offset_p): Likewise.
11415 (visium_legitimize_reload_address): Likewise.
11416 (visium_split_double_move): Likewise.
11417 (visium_expand_copysign): Likewise.
11418 (visium_expand_int_cstore): Likewise.
11419 (visium_expand_fp_cstore): Likewise.
11420 (visium_split_cstore): Likewise.
11421 (visium_select_cc_mode): Likewise.
11422 (visium_split_cbranch): Likewise.
11423 (output_cbranch): Likewise.
11424 (visium_print_operand_address): Likewise.
11425 * expmed.c (flip_storage_order): Likewise.
11426 * expmed.h (emit_cstore): Likewise.
11427 (flip_storage_order): Likewise.
11428 * genrecog.c (validate_pattern): Likewise.
11429 * hsa-gen.c (gen_hsa_addr): Likewise.
11430 * internal-fn.c (expand_arith_overflow): Likewise.
11431 * ira-color.c (allocno_copy_cost_saving): Likewise.
11432 * lra-assigns.c (find_hard_regno_for_1): Likewise.
11433 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
11434 (process_invariant_for_inheritance): Likewise.
11435 * lra-eliminations.c (move_plus_up): Likewise.
11436 * omp-low.c (lower_oacc_reductions): Likewise.
11437 * simplify-rtx.c (simplify_subreg): Likewise.
11438 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
11439 (TARGET_CHKP_BOUND_MODE): Likewise..
11440 * targhooks.c (default_chkp_bound_mode): Likewise.
11441 (default_setup_incoming_vararg_bounds): Likewise.
11442 * targhooks.h (default_chkp_bound_mode): Likewise.
11443 (default_setup_incoming_vararg_bounds): Likewise.
11444 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
11445 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
11446 (have_whole_vector_shift): Likewise.
11447 * tree-vect-stmts.c (vectorizable_load): Likewise.
11448 * doc/tm.texi: Regenerate.
11450 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
11452 Graceful degrade if Binutils PR21472 is not available.
11455 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
11456 .rodata in flash test fails.
11457 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
11458 * confgure: Regenerate.
11459 * config.in: Regenerate.
11460 * config/avr/avr.c (avr_asm_named_section)
11461 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
11462 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
11463 (avr_asm_init_sections): Same.
11465 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
11467 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
11468 (fma<VH:mode>4_intrinsic): Likewise.
11469 (*fmsub<VCVTF:mode>4): Likewise.
11470 (*fmsub<VH:mode>4_intrinsic): Likewise.
11472 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
11475 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
11476 Don't depend on "optimize > 0".
11477 (out_movhi_r_mr, out_movqi_mr_r): Same.
11478 (out_movhi_mr_r, out_movqi_r_mr): Same.
11479 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
11480 io_address_operand on "optimize > 0".
11482 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11484 * tree-loop-distribution.c: Add general explanantion on the pass.
11485 (generate_loops_for_partition): Mark distributed loop.
11486 (pg_add_dependence_edges): New parameter. Handle alias data
11487 dependence specially and record it in the parameter if asked.
11488 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
11489 (init_partition_graph_vertices, add_partition_graph_edge): New.
11490 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
11491 (free_partition_graph_vdata, build_partition_graph): New.
11492 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
11493 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
11494 (data_ref_segment_size, latch_dominated_by_data_ref): New.
11495 (compute_alias_check_pairs, version_loop_by_alias_check): New.
11496 (version_for_distribution_p, finalize_partitions): New.
11497 (distribute_loop): Handle alias data dependence specially. Factor
11498 out loop fusion code as functions and call these functions.
11500 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11502 * tree-loop-distribution.c (classify_partition): New parameter and
11503 better handle reduction statement.
11504 (rdg_build_partitions): Revise comment.
11505 (distribute_loop): Compute statements in all partitions and pass it
11506 to classify_partition.
11508 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11510 * tree-loop-distribution.c (enum partition_type): New.
11511 (struct partition): New field type.
11512 (partition_merge_into): Add parameter. Update partition type.
11513 (data_dep_in_cycle_p, update_type_for_merge): New functions.
11514 (build_rdg_partition_for_vertex): Compute partition type.
11515 (rdg_build_partitions): Dump partition type.
11516 (distribute_loop): Update calls to partition_merge_into.
11518 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11520 * tree-loop-distribution.c (struct ddr_hasher): New.
11521 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
11523 (classify_partition): Call get_data_dependence.
11524 (pg_add_dependence_edges): Ditto.
11525 (distribute_loop): Release data dependence hash table.
11527 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11529 * tree-loop-distribution.c (ref_base_address): Delete.
11530 (similar_memory_accesses): Rename ...
11531 (share_memory_accesses): ... to this. Check if partitions access
11532 the same memory reference.
11533 (distribute_loop): Call share_memory_accesses.
11535 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11537 * tree-loop-distribution.c (struct partition): New field recording
11538 its data reference.
11539 (partition_alloc, partition_free): Init and release data refs.
11540 (partition_merge_into): Merge data refs.
11541 (build_rdg_partition_for_vertex): Collect data refs for partition.
11542 (pg_add_dependence_edges): Change parameters from vector to bitmap.
11544 (distribute_loop): Remve data refs from vertice data of partition
11547 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11549 * tree-loop-distribution.c (params.h): Include header file.
11550 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
11551 (datarefs_vec): New global var.
11552 (create_rdg_vertices): Use datarefs_vec directly.
11553 (free_rdg): Don't free data references.
11554 (build_rdg): Update use. Don't free data references.
11555 (distribute_loop): Compute global variable for data references.
11556 Bail out if there are too many data references.
11558 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11560 * tree-loop-distribution.c (loop_nest): New global var.
11561 (build_rdg): Use loop directly, rather than loop nest.
11562 (pg_add_dependence_edges): Remove loop nest parameter. Use global
11564 (distribute_loop): Compute global variable loop nest. Update use.
11566 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11568 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
11569 (partition_merge_into): New parameter. Dump reason for fusion.
11570 (distribute_loop): Update use of partition_merge_into.
11572 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11574 * tree-loop-distribution.c (bb_top_order_index): New.
11575 (bb_top_order_index_size, bb_top_order_cmp): New.
11576 (stmts_from_loop): Use topological order.
11577 (pass_loop_distribution::execute): Compute and release topological
11578 order for basic blocks.
11580 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11582 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
11585 2017-07-05 Bin Cheng <bin.cheng@arm.com>
11587 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
11588 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
11589 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
11590 * internal-fn.def (LOOP_DIST_ALIAS): New.
11591 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
11592 (fold_loop_internal_call): ... this.
11593 (vect_loop_dist_alias_call): New function.
11594 (set_uid_loop_bbs): Call fold_loop_internal_call.
11595 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
11598 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
11601 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
11602 Require dead FLAGS_REG at the beginning of a peephole.
11604 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
11607 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
11608 arguments in the call to __builtin_ia32_sbb_u32.
11609 (_subborrow_u64): Swap _X and _Y arguments in the call to
11610 __builtin_ia32_sbb_u64.
11612 2017-07-04 Jakub Jelinek <jakub@redhat.com>
11615 * tree-vrp.c (compare_assert_loc): Turn into a function template
11616 with stable template parameter. Only test if a->e is NULL,
11617 !a->e == !b->e has been verified already. Use e == NULL or
11618 e != NULL instead of e or ! e tests. If stable is true, don't use
11619 iterative_hash_expr, on the other side allow a or b or both NULL
11620 and sort the NULLs last.
11621 (process_assert_insertions): Sort using compare_assert_loc<false>
11622 instead of compare_assert_loc, later sort using
11623 compare_assert_loc<true> before calling process_assert_insertions_for
11624 in a loop. Use break instead of continue once seen NULL pointer.
11626 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
11628 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
11629 Cortex-R7 and Cortex-R8 processors.
11631 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
11633 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
11634 uninitialized while src is not.
11636 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
11638 * common/config/arm/arm-common.c: Adjust include path for
11640 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
11641 (arm-cpu.h): Create in build directory. Adjust dependency rules.
11642 (arm-cpu-data.h): Likewise.
11643 (arm-cpu-cdata.h): Likewise.
11644 * config/arm/arm-cpu.h: Delete.
11645 * config/arm/arm-cpu-cdata.h: Delete.
11646 * config/arm/arm-cpu-data.h: Delete.
11648 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
11650 * config/arm/arm-cpus.in (cortex-a55): New.
11651 (cortex-a75): Likewise.
11652 (cortex-a75.cortex-a55): Likewise.
11653 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
11655 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
11656 * config/arm/arm-cpu-cdata.h: Regenerate.
11657 * config/arm/arm-cpu-data.h: Regenerate.
11658 * config/arm/arm-cpu.h: Regenerate.
11659 * config/arm/arm-tables.opt: Regenerate.
11660 * config/arm/arm-tune.md: Regenerate.
11662 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
11664 * haifa-sched.c (sched_create_recovery_edges): Update profile.
11666 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
11668 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
11671 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
11673 PR tree-optimization/81292
11674 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
11675 full_string_p, also call adjust_related_strinfos if the adjustment
11676 is simple, otherwise invalidate related strinfos.
11678 2017-07-04 Martin Liska <mliska@suse.cz>
11681 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
11682 newly created variable as DECL_IGNORED_P.
11684 2017-07-04 Martin Liska <mliska@suse.cz>
11687 * ipa-inline.c (inline_small_functions):
11688 Use xstrdup_for_dump.
11690 2017-07-04 Tom de Vries <tom@codesourcery.com>
11692 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
11694 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
11697 * config/darwin.c (darwin_function_switched_text_sections):
11700 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
11702 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
11704 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
11706 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
11708 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11710 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
11711 min_profitable_iters, and th as inclusive lower bounds.
11712 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
11713 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
11714 for min_profitable_iters and min_profitable_estimate.
11715 (vect_transform_loop): Treat th as an inclusive lower bound.
11716 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
11718 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
11721 * config/darwin.c (darwin_function_switched_text_sections):
11722 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
11723 in two pieces, and suppress the use of buf.
11725 2017-07-03 Nathan Sidwell <nathan@acm.org>
11727 * hash-table.h (hash_table_mod1): Fix indentation.
11729 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11731 PR middle-end/81290
11732 * predict.c (force_edge_cold): Be more careful about propagation
11734 * profile-count.h (profile_probability::guessed,
11735 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
11737 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
11739 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
11741 * doc/invoke.texi (rcpc architecture extension): Document it.
11743 2017-07-03 Richard Biener <rguenther@suse.de>
11745 PR tree-optimization/60510
11746 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
11747 the scalar reduction PHI and use it.
11748 (vectorizable_reduction): Properly guard the single_defuse_cycle
11749 path for non-SLP reduction chains where we cannot use it.
11750 Rework reduc_def/index and vector type deduction. Rework
11751 vector operand gathering during reduction op code-gen.
11752 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
11753 chains dissolve the chain and leave it to non-SLP reduction
11756 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11758 * tree-data-ref.h (dr_alignment): Declare.
11759 * tree-data-ref.c (dr_alignment): New function.
11760 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
11761 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
11763 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
11765 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11767 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
11768 and base_misalignment fields.
11769 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
11770 * tree-data-ref.c: Include builtins.h.
11771 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
11772 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
11773 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
11774 * tree-vect-data-refs.c: Include tree-cfg.h.
11775 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
11776 fields instead of calculating an alignment here.
11777 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
11778 innermost_loop_behavior fields.
11780 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11782 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
11784 (DR_STEP_ALIGNMENT): New macro.
11785 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
11786 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
11787 (create_data_ref): Print it.
11788 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
11789 to tell whether the step preserves vector (mis)alignment.
11790 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
11791 Move the check for an integer step and generalise to all INTEGER_CST.
11792 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
11793 Print the outer step alignment.
11795 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11797 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
11798 with offset_alignment.
11799 (DR_ALIGNED_TO): Delete.
11800 (DR_OFFSET_ALIGNMENT): New macro.
11801 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
11802 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
11803 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
11804 (create_data_ref): Likewise.
11805 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
11806 (vect_analyze_data_refs): Likewise.
11807 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
11808 creating dummy innermost behavior.
11810 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11812 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
11813 with a "innermost_loop_behavior *" and refeence tree.
11814 * tree-data-ref.c (dr_analyze_innermost): Likewise.
11815 (create_data_ref): Update call accordingly.
11816 * tree-predcom.c (find_looparound_phi): Likewise.
11818 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11820 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
11821 fields with dr_wrt_vec_loop.
11822 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
11823 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
11824 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
11825 (vect_dr_behavior): New function.
11826 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
11827 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
11828 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
11829 track whether the step preserves the misalignment.
11830 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
11831 Use vect_dr_behavior.
11832 (vect_setup_realignment): Update call accordingly.
11833 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
11834 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
11835 call to vect_create_addr_base_for_vector_ref.
11836 (vect_create_cond_for_align_checks): Likewise.
11837 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
11838 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
11839 (vect_recog_mask_conversion_pattern): Likewise.
11840 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
11841 (new_stmt_vec_info): Remove redundant zeroing.
11843 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
11845 * common/config/arm/arm-common.c (arm_be8_option): New function.
11846 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
11847 (ISA_ARMv6): Add isa_bit_be8.
11848 * config/arm/arm.h (arm_be8_option): Add prototype.
11849 (BE8_SPEC_FUNCTION): New define.
11850 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
11851 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
11852 (mlittle-endian): Similarly.
11853 (mbe8, mbe32): New options.
11854 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
11855 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
11857 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11859 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
11861 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11863 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
11864 (cleanup_tree_cfg_bb): Use it.
11865 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
11867 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
11869 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11872 * loop-doloop.c (add_test): Update profile.
11874 2017-07-03 Martin Liska <mliska@suse.cz>
11877 * sanopt.c (rewrite_usage_of_param): New function.
11878 (sanitize_rewrite_addressable_params): Likewise.
11879 (pass_sanopt::execute): Call rewrite_usage_of_param.
11881 2017-07-03 Richard Biener <rguenther@suse.de>
11883 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
11884 back to using VIEW_CONVERT_EXPR.
11886 2017-07-03 Martin Liska <mliska@suse.cz>
11889 * doc/extend.texi: Document when a resolver function is
11890 generated for target_clones.
11892 2017-07-03 Martin Liska <mliska@suse.cz>
11894 * asan.c (asan_emit_stack_protection): Unpoison just red zones
11895 and shadow memory of auto variables which are subject of
11896 use-after-scope sanitization.
11897 (asan_expand_mark_ifn): Add do set only when is_poison.
11899 2016-07-03 Richard Biener <rguenther@suse.de>
11901 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
11903 (vect_force_simple_reduction): Record reduction def -> phi mapping.
11904 (vectorizable_reduction): Perform reduction PHI creation when
11905 visiting a reduction PHI and adjust and simplify code generation
11906 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
11907 (vect_transform_loop): Visit reduction PHIs.
11908 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
11909 defs into the SLP tree.
11910 (vect_build_slp_tree): Reduction defs terminate the recursion.
11911 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
11913 (vect_get_vec_defs_for_stmt_copy): Export.
11914 (vect_get_vec_defs): Likewise.
11915 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
11917 (vect_get_vec_defs_for_stmt_copy): Declare.
11918 (vect_get_vec_defs): Likewise.
11920 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11922 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
11923 parameter with a "loop" parameter and use it instead of the
11924 loop containing DR_STMT. Don't check simple_iv when doing
11925 BB analysis. Describe the two analysis modes in the comment.
11927 2017-07-03 Tom de Vries <tom@codesourcery.com>
11929 PR tree-optimization/69468
11930 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
11931 (find_same_succ_bb): Handle ignore_edge_flags.
11933 2017-07-03 Tom de Vries <tom@codesourcery.com>
11935 PR tree-optimization/81192
11936 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
11938 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
11940 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
11942 2017-07-03 Tom de Vries <tom@codesourcery.com>
11944 PR tree-optimization/81192
11945 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
11946 BB_SAME_SUCC (bb) == NULL.
11948 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11950 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
11953 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11955 * dumpfile.c: Include profile-count.h
11956 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
11958 (insert_cond_bb): Update profile.
11959 * tree-cfg.h (insert_cond_bb): Update prototype.
11960 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
11961 * tree-dump.c: Do not include tree-cfg.
11963 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11965 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
11967 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11969 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
11972 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11974 * tree-complex.c (expand_complex_div_wide): update profile.
11976 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
11977 Alan Hayward <alan.hayward@arm.com>
11978 David Sherwood <david.sherwood@arm.com>
11980 * Makefile.in (MACHMODE_H): Remove insn-modes.h
11981 (CORETYPES_H): New define.
11982 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
11983 (insn-modes-inline.h, s-modes-inline-h): New rules.
11984 (generated_files): Add insn-modes-inline.h.
11985 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
11986 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
11987 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
11988 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
11989 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
11990 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
11991 (build/gencodes.o, build/genconditions.o): Likewise.
11992 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
11993 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
11994 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
11995 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
11996 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
11997 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
11998 * coretypes.h: Include everything up to real.h for generators.
11999 Include insn-modes.h first. Include wide-int-print.h after
12000 wide-int.h. Include insn-modes-inline.h and then machmode.h.
12001 * machmode.h: Don't include insn-modes.h here.
12002 * function-tests.c: Remove includes of signop.h, machmode.h,
12003 double-int.h and wide-int.h.
12005 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
12007 * optc-save-gen.awk: Likewise.
12008 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
12009 * godump.c: Remove include of wide-int-print.h.
12010 * pretty-print.h: Likewise.
12011 * wide-int-print.cc: Likewise.
12012 * wide-int.cc: Likewise.
12013 * hash-map-tests.c: Remove include of signop.h.
12014 * hash-set-tests.c: Likewise.
12015 * rtl-tests.c: Likewise.
12016 * mkconfig.sh: Remove include of machmode.h.
12017 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
12019 (emit_insn_modes_inline_h): ...this new function. Emit the code
12020 into an insn-modes-inline.h header file, adding appropriate
12021 include guards and end comments.
12022 (emit_insn_modes_c_header): Remove include of machmode.h.
12023 (emit_min_insn_modes_c_header): Include coretypes.h rather than
12025 (main): Handle -i flag and call emit_insn_modes_inline_h when
12028 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
12030 * tree-ssa-strlen.c (strinfo): Rename the length field to
12031 nonzero_chars. Add a full_string_p field.
12032 (compare_nonzero_chars, zero_length_string_p): New functions.
12033 (get_addr_stridx): Add an offset_out parameter.
12034 Use compare_nonzero_chars.
12035 (get_stridx): Update accordingly. Use compare_nonzero_chars.
12036 (new_strinfo): Update after above changes to strinfo.
12037 (set_endptr_and_length): Set full_string_p.
12038 (get_string_length): Update after above changes to strinfo.
12039 (unshare_strinfo): Update call to new_strinfo.
12040 (maybe_invalidate): Likewise.
12041 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
12042 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
12043 as a uhwi instead of an shwi. Update after above changes to
12044 strinfo and new_strinfo.
12045 (zero_length_string): Assert that chainsi contains full strings.
12046 Use zero_length_string_p. Update call to new_strinfo.
12047 (adjust_related_strinfos): Update after above changes to strinfo.
12048 Copy full_string_p from origsi.
12049 (adjust_last_stmt): Use zero_length_string_p.
12050 (handle_builtin_strlen): Update after above changes to strinfo and
12051 new_strinfo. Install the lhs as the string length if the previous
12052 entry didn't describe a full string.
12053 (handle_builtin_strchr): Update after above changes to strinfo
12055 (handle_builtin_strcpy): Likewise.
12056 (handle_builtin_strcat): Likewise.
12057 (handle_builtin_malloc): Likewise.
12058 (handle_pointer_plus): Likewise.
12059 (handle_builtin_memcpy): Likewise. Track nonzero characters
12060 that aren't necessarily followed by a nul terminator.
12061 (handle_char_store): Likewise.
12063 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
12065 PR tree-optimization/80769
12066 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
12067 for malloc and calloc. Document the new invariant that all related
12068 strinfos have delayed lengths or none do.
12069 (verify_related_strinfos): Move earlier in file.
12070 (set_endptr_and_length): New function, split out from...
12071 (get_string_length): ...here. Also set the lengths of related
12073 (zero_length_string): Assert that chainsi has known (rather than
12075 (adjust_related_strinfos): Likewise.
12077 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
12079 PR tree-optimization/81136
12080 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
12081 assert that two references with the same misalignment have the same
12082 compile-time misalignment if those compile-time misalignments
12085 2017-07-01 Andi Kleen <ak@linux.intel.com>
12087 * print-tree.c (print_node): Print all attributes.
12089 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12091 * cfg.c (scale_bbs_frequencies): New function.
12092 * cfg.h (scale_bbs_frequencies): Declare it.
12093 * cfgloopanal.c (single_likely_exit): Cleanup.
12094 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
12096 (scale_loop_profile): Likewise.
12097 (loop_version): Likewise.
12098 (create_empty_loop_on_edge): Update.
12099 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
12100 scale_loop_frequencies, scale_loop_profile, loopify,
12101 loop_version): Update prototypes.
12102 * modulo-sched.c (sms_schedule): Update.
12103 * predict.c (unlikely_executed_edge_p): Also check probability.
12104 (probably_never_executed_edge_p): Fix typo.
12105 * tree-if-conv.c (version_loop_for_if_conversion): Update.
12106 * tree-parloops.c (gen_parallel_loop): Update.
12107 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
12108 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
12109 * tree-ssa-loop-split.c (split_loop): Update.
12110 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
12111 * tree-vect-loop-manip.c (vect_do_peeling): Update.
12112 (vect_loop_versioning): Update.
12113 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
12115 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12117 * trans-mem.c (split_bb_make_tm_edge): Update profile.
12119 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12121 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
12122 to keep profile consistent.
12124 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12126 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
12127 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
12128 * profile-count.h (max_safe_multiplier): Make unsigned.
12129 (profile_count::guessed_zero): New.
12131 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12133 * bb-reorder.c (fix_up_crossing_landing_pad,
12134 fix_crossing_conditional_branches): Use make_single_succ_edge
12135 to keep profile consistent.
12137 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
12139 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
12142 2017-07-01 Jakub Jelinek <jakub@redhat.com>
12145 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
12146 the right scopes, make sure cond_jump isn't preserved between multiple
12147 iterations. Search for fallthru edge whenever there are 3+ edges and
12148 use find_fallthru_edge for it.
12150 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
12152 Patch by Alexander Monakov <amonakov@ispras.ru>
12153 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
12154 probabilities consistently.
12156 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
12158 * pa.c (pa_expand_compare_and_swap_loop): Update call of
12159 emit_cmp_and_jump_insns.
12161 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
12164 * tree-inline.c (expand_call_inline): Combine profile statuses.
12166 2017-06-30 Andrew Pinski <apinski@cavium.com>
12168 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
12169 fold_stmt returned true.
12171 2017-06-30 Nathan Sidwell <nathan@acm.org>
12173 * ggc.h (empty_string): Delete.
12174 * cfgexpand.c (expand_asm_stmt): Use plain "".
12175 * optabs.c (expand_asm_memory_barrier): Likewise.
12176 * stringpool.c (empty_string): Delete.
12177 (digit_vector, digit_string): Delete.
12178 (ggc_alloc_string): Use plain "", don't optimize single digit
12179 strings. Use ggc_alloc_atomic.
12181 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
12183 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
12184 comparison set and one other set, use the cost of the non-comparison
12187 2017-06-30 Nathan Sidwell <nathan@acm.org>
12189 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
12192 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
12194 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
12195 loops. Remove now unneeded calls to gimple_switch_set_label() that
12196 just set removed labels to NULL_TREE.
12198 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
12200 * tree-ssanames.c (set_range_info_raw): Abstract from ...
12201 (set_range_info): ...here. Only call set_range_info_raw if domain
12203 (set_nonzero_bits): Call set_range_info_raw.
12204 * tree-ssanames.h (set_range_info_raw): New.
12206 2017-06-30 Jakub Jelinek <jakub@redhat.com>
12209 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
12210 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
12211 of nonimmediate_operand and <store_mask_constraint> instead of m
12212 for the input operand. For V8FI iterator, always split if input
12213 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
12214 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
12215 <store_mask_predicate> instead of register_operand and
12216 <store_mask_constraint> instead of v for the input operand. Make
12217 sure both operands aren't MEMs for if not <mask_applied>.
12219 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
12221 * lto-wrapper.c (copy_file) Close both file descriptors before
12224 2017-06-30 Martin Liska <mliska@suse.cz>
12227 * multiple_target.c (create_dispatcher_calls): Make ifunc
12228 also for function that don't have calls or are not referenced.
12230 2017-06-30 Richard Biener <rguenther@suse.de>
12232 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
12233 analyze the first scalar stmt. Move vector type computation
12234 for the BB case here from ...
12235 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
12236 live operation processing in the SLP case properly.
12238 2017-06-30 Richard Biener <rguenther@suse.de>
12240 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
12242 2017-06-30 Martin Liska <mliska@suse.cz>
12245 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
12246 before BUILT_IN_UNWIND_RESUME when ASAN is used.
12248 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
12250 * doc/invoke.texi (AArch64): Add missing options and remove redundant
12253 2017-06-30 Richard Biener <rguenther@suse.de>
12255 PR tree-optimization/81249
12256 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
12257 condition reduction result to original scalar type.
12259 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12261 * profile-count.h (enum profile_quality): Fix typos and whitespace
12264 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12266 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
12267 type for branch probabilities.
12269 2017-06-29 Julian Brown <julian@codesourcery.com>
12270 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
12272 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
12273 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
12274 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
12275 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
12277 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
12279 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
12280 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
12281 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
12282 CC usage from generic code to here.
12283 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
12284 CC usage into the target macros.
12286 2017-06-29 Maya Rashish <coypu@sdf.org>
12288 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
12291 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
12293 * arm/arm-builtins.c: Include profile-count.h
12294 * except.c (sjlj_emit_function_enter): Use
12295 profile_probability::unlikely.
12297 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
12299 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
12300 and tocrel_offset be pointer args rather than implicitly using
12302 (legitimate_constant_pool_address_p, rs6000_emit_move,
12303 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
12304 tocrel_offset and use in toc_relative_expr_p call.
12305 (print_operand, print_operand_address): Use static tocrel_base_oac
12306 and tocrel_offset_oac.
12307 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
12310 2017-06-29 Maya Rashish <coypu@sdf.org>
12312 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
12314 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
12316 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
12317 objects, take into account only the alignment of 'op0' and 'mode1' if
12320 2017-06-29 Steve Ellcey <sellcey@cavium.com>
12322 * ccmp.c (ccmp_tree_comparison_p): New function.
12323 (ccmp_candidate_p): Update to use above function.
12324 (get_compare_parts): New function.
12325 (expand_ccmp_next): Update to use new functions.
12326 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
12328 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
12329 take mode as argument.
12330 * ccmp.h (expand_ccmp_expr): Add mode as argument.
12331 * expr.c (expand_expr_real_1): Pass mode as argument.
12333 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
12335 * combine.c (combine_instructions): Print insns to dump_file, together
12338 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
12340 * asan.c (asan_emit_stack_protection): Update.
12341 (create_cond_insert_point): Update.
12342 * auto-profile.c (afdo_propagate_circuit): Update.
12343 * basic-block.h (struct edge_def): Turn probability to
12344 profile_probability.
12345 (EDGE_FREQUENCY): Update.
12346 * bb-reorder.c (find_traces_1_round): Update.
12347 (better_edge_p): Update.
12348 (sanitize_hot_paths): Update.
12349 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
12350 (make_single_succ_edge): Update.
12351 (check_bb_profile): Update.
12352 (dump_edge_info): Update.
12353 (update_bb_profile_for_threading): Update.
12354 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
12355 probabilitycount to 0.
12356 * cfgbuild.c (compute_outgoing_frequencies): Update.
12357 * cfgcleanup.c (try_forward_edges): Update.
12358 (outgoing_edges_match): Update.
12359 (try_crossjump_to_edge): Update.
12360 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
12361 (expand_gimple_tailcall): Update.
12362 (construct_init_block): Use make_single_succ_edge.
12363 (construct_exit_block): Use make_single_succ_edge.
12364 * cfghooks.c (verify_flow_info): Update.
12365 (redirect_edge_succ_nodup): Update.
12366 (split_edge): Update.
12367 (account_profile_record): Update.
12368 * cfgloopanal.c (single_likely_exit): Update.
12369 * cfgloopmanip.c (scale_loop_profile): Update.
12370 (set_zero_probability): Remove.
12371 (duplicate_loop_to_header_edge): Update.
12372 * cfgloopmanip.h (loop_version): Update prototype.
12373 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
12374 (force_nonfallthru_and_redirect): Update.
12375 (update_br_prob_note): Update.
12376 (rtl_verify_edges): Update.
12377 (purge_dead_edges): Update.
12378 (rtl_lv_add_condition_to_bb): Update.
12379 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
12380 * cgraphunit.c (init_lowered_empty_function): Update.
12381 (cgraph_node::expand_thunk): Update.
12382 * cilk-common.c: Include profile-count.h
12383 * dojump.c (inv): Remove.
12384 (jumpifnot): Update.
12385 (jumpifnot_1): Update.
12386 (do_jump_1): Update.
12388 (do_jump_by_parts_greater_rtx): Update.
12389 (do_compare_rtx_and_jump): Update.
12390 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
12391 do_jump_1. do_compare_rtx_and_jump): Update prototype.
12392 * dwarf2cfi.c: Include profile-count.h
12393 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
12394 (sjlj_emit_dispatch_table): Likewise.
12395 * explow.c: Include profile-count.h
12396 * expmed.c (emit_store_flag_force): Update.
12397 (do_cmp_and_jump): Update.
12398 * expr.c (compare_by_pieces_d::generate): Update.
12399 (compare_by_pieces_d::finish_mode): Update.
12400 (emit_block_move_via_loop): Update.
12401 (store_expr_with_bounds): Update.
12402 (store_constructor): Update.
12403 (expand_expr_real_2): Update.
12404 (expand_expr_real_1): Update.
12405 * expr.h (try_casesi, try_tablejump): Update prototypes.
12406 * gimple-pretty-print.c (dump_probability): Update.
12407 (dump_profile): New.
12408 (dump_gimple_label): Update.
12409 (dump_gimple_bb_header): Update.
12410 * graph.c (draw_cfg_node_succ_edges): Update.
12411 * hsa-gen.c (convert_switch_statements): Update.
12412 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
12413 (find_if_case_1): Update.
12414 (find_if_case_2): Update.
12415 * internal-fn.c (expand_arith_overflow_result_store): Update.
12416 (expand_addsub_overflow): Update.
12417 (expand_neg_overflow): Update.
12418 (expand_mul_overflow): Update.
12419 (expand_vector_ubsan_overflow): Update.
12420 * ipa-cp.c (good_cloning_opportunity_p): Update.
12421 * ipa-split.c (split_function): Use make_single_succ_edge.
12422 * ipa-utils.c (ipa_merge_profiles): Update.
12423 * loop-doloop.c (add_test): Update.
12424 (doloop_modify): Update.
12425 * loop-unroll.c (compare_and_jump_seq): Update.
12426 (unroll_loop_runtime_iterations): Update.
12427 * lra-constraints.c (lra_inheritance): Update.
12428 * lto-streamer-in.c (input_cfg): Update.
12429 * lto-streamer-out.c (output_cfg): Update.
12430 * mcf.c (adjust_cfg_counts): Update.
12431 * modulo-sched.c (sms_schedule): Update.
12432 * omp-expand.c (expand_omp_for_init_counts): Update.
12433 (extract_omp_for_update_vars): Update.
12434 (expand_omp_ordered_sink): Update.
12435 (expand_omp_for_ordered_loops): Update.
12436 (expand_omp_for_generic): Update.
12437 (expand_omp_for_static_nochunk): Update.
12438 (expand_omp_for_static_chunk): Update.
12439 (expand_cilk_for): Update.
12440 (expand_omp_simd): Update.
12441 (expand_omp_taskloop_for_outer): Update.
12442 (expand_omp_taskloop_for_inner): Update.
12443 * omp-simd-clone.c (simd_clone_adjust): Update.
12444 * optabs.c (expand_doubleword_shift): Update.
12445 (expand_abs): Update.
12446 (emit_cmp_and_jump_insn_1): Update.
12447 (expand_compare_and_swap_loop): Update.
12448 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
12449 * predict.c (predictable_edge_p): Update.
12450 (edge_probability_reliable_p): Update.
12451 (set_even_probabilities): Update.
12452 (combine_predictions_for_insn): Update.
12453 (combine_predictions_for_bb): Update.
12454 (propagate_freq): Update.
12455 (estimate_bb_frequencies): Update.
12456 (force_edge_cold): Update.
12457 * profile-count.c (profile_count::dump): Add missing space into dump.
12458 (profile_count::debug): Add newline.
12459 (profile_count::differs_from_p): Explicitly convert to unsigned.
12460 (profile_count::stream_in): Update.
12461 (profile_probability::dump): New member function.
12462 (profile_probability::debug): New member function.
12463 (profile_probability::differs_from_p): New member function.
12464 (profile_probability::differs_lot_from_p): New member function.
12465 (profile_probability::stream_in): New member function.
12466 (profile_probability::stream_out): New member function.
12467 * profile-count.h (profile_count_quality): Rename to ...
12468 (profile_quality): ... this one.
12469 (profile_probability): New.
12470 (profile_count): Update.
12471 * profile.c (compute_branch_probabilities): Update.
12472 * recog.c (peep2_attempt): Update.
12473 * sched-ebb.c (schedule_ebbs): Update.
12474 * sched-rgn.c (find_single_block_region): Update.
12475 (compute_dom_prob_ps): Update.
12476 (schedule_region): Update.
12477 * sel-sched-ir.c (compute_succs_info): Update.
12478 * stmt.c (struct case_node): Update.
12479 (do_jump_if_equal): Update.
12480 (get_outgoing_edge_probs): Update.
12481 (conditional_probability): Update.
12482 (emit_case_dispatch_table): Update.
12483 (expand_case): Update.
12484 (expand_sjlj_dispatch_table): Update.
12485 (emit_case_nodes): Update.
12486 * targhooks.c: Update.
12487 * tracer.c (better_p): Update.
12488 (find_best_successor): Update.
12489 * trans-mem.c (expand_transaction): Update.
12490 * tree-call-cdce.c: Update.
12491 * tree-cfg.c (gimple_split_edge): Upate.
12492 (move_sese_region_to_fn): Upate.
12493 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
12494 * tree-eh.c (lower_resx): Upate.
12495 (cleanup_empty_eh_move_lp): Upate.
12496 * tree-if-conv.c (version_loop_for_if_conversion): Update.
12497 * tree-inline.c (copy_edges_for_bb): Update.
12498 (copy_cfg_body): Update.
12499 * tree-parloops.c (gen_parallel_loop): Update.
12500 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
12501 (gimple_gen_time_profiler): Update.
12502 * tree-ssa-dce.c (remove_dead_stmt): Update.
12503 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
12504 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
12505 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
12506 (unloop_loops): Update.
12507 (try_peel_loop): Update.
12508 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
12509 * tree-ssa-loop-split.c (connect_loops): Update.
12510 (split_loop): Update.
12511 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
12512 (hoist_guard): Update.
12513 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
12514 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
12515 (value_replacement): Update.
12516 * tree-ssa-reassoc.c (branch_fixup): Update.
12517 * tree-ssa-tail-merge.c (replace_block_by): Update.
12518 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
12519 (create_edge_and_update_destination_phis): Update.
12520 (compute_path_counts): Update.
12521 (recompute_probabilities): Update.
12522 (update_joiner_offpath_counts): Update.
12523 (freqs_to_counts_path): Update.
12524 (duplicate_thread_path): Update.
12525 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
12526 (struct switch_conv_info): Update.
12527 (gen_inbound_check): Update.
12528 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
12529 (vect_do_peeling): Update.
12530 (vect_loop_versioning): Update.
12531 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
12532 (optimize_mask_stores): Update.
12533 * ubsan.c (ubsan_expand_null_ifn): Update.
12534 * value-prof.c (gimple_divmod_fixed_value): Update.
12535 (gimple_divmod_fixed_value_transform): Update.
12536 (gimple_mod_pow2): Update.
12537 (gimple_mod_pow2_value_transform): Update.
12538 (gimple_mod_subtract): Update.
12539 (gimple_mod_subtract_transform): Update.
12540 (gimple_ic): Update.
12541 (gimple_stringop_fixed_value): Update.
12542 (gimple_stringops_transform): Update.
12543 * value-prof.h: Update.
12545 2017-06-29 Carl Love <cel@us.ibm.com>
12547 * config/rs6000/rs6000-c.c: Add support for built-in functions
12548 vector signed int vec_signed (vector float);
12549 vector signed long long vec_signed (vector double);
12550 vector signed int vec_signed2 (vector double, vector double);
12551 vector signed int vec_signede (vector double);
12552 vector signed int vec_signedo (vector double);
12553 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
12554 instruction generator.
12555 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
12556 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
12557 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
12559 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
12560 vunsignede_v2df): Add define_expands.
12561 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
12562 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
12563 VEC_UNSIGNEDO): Add definitions.
12564 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
12565 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
12566 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
12567 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
12568 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
12569 * config/rs6000/altivec.h (vec_signed, vec_signed2,
12570 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
12571 vec_unsignede, vec_unsignedo): Add builtin defines.
12572 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
12574 * doc/extend.texi: Update the built-in documentation file for the
12575 new built-in functions.
12577 2017-06-29 Richard Biener <rguenther@suse.de>
12579 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
12580 reduction chains to LOOP_VINFO_REDUCTIONS.
12581 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
12582 SLP reductions after processing reduction chains.
12584 2017-06-29 Nathan Sidwell <nathan@acm.org>
12586 * builtins.c (fold_builtin_FUNCTION): Use
12587 lang_hooks.decl_printable_name.
12589 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
12591 PR middle-end/81194
12592 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
12593 with only one label.
12594 * stmt.c (expand_case): Assert NCASES is greater than one.
12596 2017-06-29 Richard Biener <rguenther@suse.de>
12598 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
12600 (group_case_labels): Likewise.
12601 (find_taken_edge): Push sanity checking on val to workers...
12602 (find_taken_edge_cond_expr): ... here
12603 (find_taken_edge_switch_expr): ... and here, handle cases
12604 with just a default label.
12605 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
12606 (group_case_labels): Likewise.
12607 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
12608 group_case_labels does anything cleanup the CFG again.
12610 2017-06-29 Bin Cheng <bin.cheng@arm.com>
12612 PR tree-optimization/81196
12613 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
12614 exit condition comparing two IVs.
12616 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
12618 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
12619 profile to the dummy entry at the end of the list of architectures.
12620 * config/arm/arm-cpu-cdata.h: Regenerated.
12622 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12623 Michael Collison <michael.collison@arm.com>
12626 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
12628 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
12629 (*aarch64_reg_<mode>3_minus_mask): New pattern.
12630 (*aarch64_<optab>_reg_di3_mask2): New pattern.
12631 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
12632 of shift when the shift amount is masked with constant equal to
12633 the size of the mode.
12634 * config/aarch64/predicates.md (subreg_lowpart_operator): New
12637 2017-06-29 Martin Liska <mliska@suse.cz>
12639 * config/i386/i386.opt: Change range from [1,5] to [0,5].
12641 2017-06-29 Yury Gribov <tetra2005@gmail.com>
12644 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
12647 (edge_growth_cache_entry::edge_growth_cache_entry): New
12649 (reset_edge_growth_cache): Update to use constructor.
12651 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
12653 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
12654 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
12655 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
12657 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
12659 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
12660 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
12662 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
12664 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
12665 (*-linux-uclibc*): Add t-uclibc tmake_file.
12666 * config/t-musl: New.
12667 * config/t-uclibc: New.
12669 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
12671 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
12673 (gen_comm_data): Emit architectural setting of arch_prof.
12674 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
12676 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
12677 (armv8-m.base, armv8-m.main): Likewise.
12678 * arm-protos.h (arm_build_target): Add profile field.
12679 (arch_option): Likewise.
12680 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
12682 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
12683 arm_active_target.profile.
12685 2017-06-28 Richard Biener <rguenther@suse.de>
12687 PR middle-end/81227
12688 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
12689 TYPE_OVERFLOW_WRAPS.
12690 * match.pd (negate_expr_p): Likewise.
12691 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
12692 fold_build2, not fold_binary.
12694 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
12696 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
12697 Convert memory address to Pmode.
12698 (aarch64_print_operand): Assert MEM operands are always Pmode.
12700 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
12703 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
12704 Remove redundant if.
12705 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
12706 * config/arm/aarch-common-protos.h
12707 (aarch_forward_to_shift_is_not_shifted_re): Remove.
12708 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
12710 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
12713 * multiple_target.c (create_dispatcher_calls): Set the default
12714 clone to be static, not public.
12716 2017-06-28 Richard Biener <rguenther@suse.de>
12718 * tree-vect-loop.c (vectorizable_reduction): Move special
12719 cond reduction IV var creation ...
12720 (vect_create_epilog_for_reduction): ... here. Remove induction_index
12721 parameter. Use STMT_VINFO_VECTYPE.
12722 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
12725 2017-06-28 Martin Liska <mliska@suse.cz>
12728 * ipa-visibility.c (non_local_p): Handle visibility.
12730 2017-06-28 Martin Liska <mliska@suse.cz>
12733 * common.opt: Add IntegerRange to various options.
12734 * opt-functions.awk (integer_range_info): New function.
12735 * optc-gen.awk: Add integer_range_info to cl_options struct.
12736 * opts-common.c (decode_cmdline_option): Handle
12737 CL_ERR_INT_RANGE_ARG.
12738 (cmdline_handle_error): Likewise.
12739 * opts.c (print_filtered_help): Show valid interval in
12740 when --help is provided.
12741 * opts.h (struct cl_option): Add range_min and range_max fields.
12742 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
12744 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
12746 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
12747 (x * C EQ/NE y * C): New transformation.
12749 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
12751 * genmultilib (combination_space): Accept '+' in option names.
12753 2017-06-28 Martin Liska <mliska@suse.cz>
12756 * asan.c (instrument_derefs): Bail out inner references
12757 that are hard register variables.
12759 2017-06-28 Jakub Jelinek <jakub@redhat.com>
12762 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
12763 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
12765 2017-06-28 Richard Biener <rguenther@suse.de>
12767 * tree-vectorizer.h (vect_get_vec_defs): Remove.
12768 (vect_get_slp_defs): Adjust.
12769 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
12771 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
12773 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
12774 get_initial_defs_for_reduction instead of vect_get_vec_defs.
12775 (vectorizable_reduction): Adjust.
12776 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
12778 (vect_get_slp_defs): Likewise.
12779 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
12780 (vectorizable_bswap): Adjust.
12781 (vectorizable_call): Likewise.
12782 (vectorizable_conversion): Likewise.
12783 (vectorizable_assignment): Likewise.
12784 (vectorizable_shift): Likewise.
12785 (vectorizable_operation): Likewise.
12786 (vectorizable_store): Likewise.
12787 (vectorizable_condition): Likewise.
12788 (vectorizable_comparison): Likewise.
12790 2017-06-28 Michael Collison <michael.collison@arm.com>
12793 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
12795 (arm_gen_movmemqi): Removed unused variable 'i'.
12796 Convert 'for' loop into 'while' loop.
12797 (arm_expand_prologue): Remove last unnecessary set of insn.
12798 (thumb_pop): Remove unused variable 'pushed_words'.
12799 (thumb_exit): Remove last unnecessary set of regs_to_pop.
12801 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12803 * config/s390/predicates.md: Use s390_rel_address_ok_p.
12804 * config/s390/s390-protos.h: Add prototype of
12805 s390_rel_address_ok_p.
12806 * config/s390/s390.c (s390_got_symbol): New function.
12807 (s390_rel_address_ok_p): New function.
12808 (legitimize_pic_address): Use s390_rel_address_ok_p.
12809 (s390_load_got): Use s390_got_symbol.
12810 (s390_option_override): Issue error if
12811 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
12812 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
12814 * config/s390/s390.opt: New option mpic-data-is-text-relative.
12816 2017-06-27 Andrew Pinski <apinski@cavium.com>
12818 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
12819 (X * copysign (1.0, X)): New pattern.
12820 (X * copysign (1.0, -X)): New pattern.
12821 (copysign (-1.0, CST)): New pattern.
12823 2017-06-27 Joseph Myers <joseph@codesourcery.com>
12825 * genmultilib (combination_space): Remove variable.
12826 Validate reuse rules against regular expression for any sequence
12827 of multilib options in any order.
12829 2017-06-27 Michael Collison <michael.collison@arm.com>
12831 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
12832 call aarch64_split_simd_combine.
12833 * (aarch64_combine_internal<mode>): Delete pattern.
12834 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
12835 Allow register and subreg operands.
12837 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
12839 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
12840 specific need, just fallback on defaults.
12841 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
12843 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
12844 Olivier Hainque <hainque@adacore.com>
12846 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
12848 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
12849 targets. Pick a default if no particular attempt applied.
12850 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
12853 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
12855 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
12856 (x86_64-wrs-vxworks7): Likewise.
12858 2017-06-27 Marek Polacek <polacek@redhat.com>
12861 * ubsan.c (instrument_null): Check get_base_address's result for null.
12863 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
12865 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
12867 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
12869 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
12870 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
12871 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
12872 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
12873 New function types.
12874 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
12875 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
12876 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
12877 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
12878 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
12879 BUILT_IN_FEUPDATEENV): New builtins.
12880 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
12881 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
12882 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
12883 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
12885 (builtin_structptr_types): Adjust size.
12886 * tree.c (builtin_structptr_types): Add four entries.
12888 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
12889 Olivier Hainque <hainque@adacore.com>
12891 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
12892 (TLS_SYM): New local macro, forcing reference to __tls__ on
12893 link command lines for VxWorks 7 RTPs, triggering initialization
12895 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
12896 OS features TLS support, true for RTPs on VxWorks 7.
12897 * config/vxworks.c (vxworks_override_options): Setup emutls
12900 2017-06-27 Jakub Jelinek <jakub@redhat.com>
12902 * predict.c (test_prediction_value_range): Use -1U instead of -1
12903 to avoid narrowing conversion warning.
12904 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
12905 to avoid narrowing conversion warning.
12906 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
12908 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
12910 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
12912 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
12913 64bit configurations.
12914 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
12915 (SIZE_TYPE): Likewise.
12916 * config/vxworks.c (vxworks_emutls_var_fields): Use
12917 long_unsigned_type_node instead of unsigned_type_node as the offset
12918 field type, which is "pointer" mode in emutls.c.
12920 2017-06-27 Jakub Jelinek <jakub@redhat.com>
12923 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
12925 PR middle-end/81207
12926 * gimple-fold.c (replace_call_with_call_and_fold): Handle
12927 gimple_vuse copying separately from gimple_vdef copying.
12929 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
12931 * value-prof.c (free_hist): Remove call to memset and the enclosing if
12934 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
12935 Olivier Hainque <hainque@adacore.com>
12937 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
12938 for all vxworks7 targets.
12939 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
12940 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
12941 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
12942 variations for VX6/VX7 and 32/64bits later on in ...
12943 (VXWORKS_LIB_SPEC): Leverage new macros.
12944 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
12945 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
12947 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
12949 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
12950 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
12952 2017-06-26 Carl Love <cel@us.ibm.com>
12954 * config/rs6000/rs6000-c.c: Add support for built-in functions
12955 vector bool char vec_reve (vector bool char);
12956 vector signed char vec_reve (vector signed char);
12957 vector unsigned char vec_reve (vector unsigned char);
12958 vector bool int vec_reve (vector bool int);
12959 vector signed int vec_reve (vector signed int);
12960 vector unsigned int vec_reve (vector unsigned int);
12961 vector bool long long vec_reve (vector bool long long);
12962 vector signed long long vec_reve (vector signed long long);
12963 vector unsigned long long vec_reve (vector unsigned long long);
12964 vector bool short vec_reve (vector bool short);
12965 vector signed short vec_reve (vector signed short);
12966 vector double vec_reve (vector double);
12967 vector float vec_reve (vector float);
12968 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
12969 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
12970 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
12971 (altivec_vreve): New pattern.
12972 * config/rs6000/altivec.h (vec_reve): New define.
12973 * doc/extend.texi (vec_rev): Update the built-in documentation file
12974 for the new built-in functions.
12976 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12978 PR tree-optimization/71815
12979 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
12981 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
12983 (slsr_process_phi): Likewise.
12984 (replace_uncond_cands_and_profitable_phis): Don't replace a
12985 multiply candidate with a stride of 1 (copy or cast).
12986 (phi_incr_cost): Call uses_consumed_by_stmt rather than
12988 (lowest_cost_path): Likewise.
12989 (total_savings): Likewise.
12991 2017-06-26 Richard Biener <rguenther@suse.de>
12994 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
12995 Use def_builtin_pure for all gather builtins.
12997 2017-06-26 Richard Biener <rguenther@suse.de>
12999 PR tree-optimization/81203
13000 * tree-tailcall.c (find_tail_calls): Do not move stmts into
13001 non-dominating BBs.
13003 2017-06-26 Marek Polacek <polacek@redhat.com>
13006 * doc/invoke.texi: Document -Wmultistatement-macros.
13008 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
13010 * doc/sourcebuild.texi (ARM-specific attributes): Document new
13011 arm_neon_ok_no_float_abi effective target.
13013 2017-06-26 Richard Biener <rguenther@suse.de>
13015 PR tree-optimization/80928
13016 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
13017 (copy_bbs): Set BB_DUPLICATED flag early.
13018 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
13020 (execute_on_shrinking_pred): Likewise.
13021 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
13022 BB_DUPLICATED blocks.
13023 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
13024 iterate over all PHIs considering removal of *gsi.
13026 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
13028 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
13031 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
13033 * config/rs6000/rs6000-string.c: (expand_block_clear,
13034 do_load_for_compare, select_block_compare_mode,
13035 compute_current_alignment, expand_block_compare,
13036 expand_strncmp_align_check, expand_strn_compare,
13037 expand_block_move, rs6000_output_load_multiple)
13038 Move functions related to string/block move/compare
13039 to a separate file.
13040 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
13041 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
13042 for this function which is now used in two files.
13043 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
13044 * config.gcc: Add rs6000-string.o to extra_objs for
13045 targets powerpc*-*-* and rs6000*-*-*.
13047 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
13050 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
13051 32-bit, since indexed is not valid for DImode.
13052 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
13053 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
13054 (define_peephole2 for Altivec d-form load): Add 32-bit support.
13055 (define_peephole2 for Altivec d-form store): Likewise.
13058 * multiple_target.c (create_dispatcher_calls): Only create the
13059 dispatcher call if the function is the default clone of a
13060 versioned function.
13062 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
13064 PR middle-end/80902
13065 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
13066 a call, force the call to not be a tail call.
13068 2017-06-23 Jeff Law <law@redhat.com>
13070 * doc/contrib.texi: Add entry for Steven Pemberton's work on
13073 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
13075 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
13076 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
13077 handling for early expansion of vector shifts (sl,sr,sra,rl).
13078 (builtin_function_type): Add vector shift right instructions
13079 to the unsigned argument list.
13081 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
13083 rtl-optimizatoin/79286
13084 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
13085 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
13086 trap. PIC register plus a const unspec without offset can never trap.
13088 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
13090 * tree.h (builtin_structptr_type): New type.
13091 (builtin_structptr_types): Declare new array.
13092 * tree.c (builtin_structptr_types): New array.
13093 (free_lang_data, build_common_tree_nodes): Use it.
13095 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
13098 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
13101 2017-06-22 Matt Turner <mattst88@gmail.com>
13103 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
13104 Lake models to skylake case. Assume skylake for unknown
13105 models with clflushopt.
13107 2017-06-22 Jeff Law <law@redhat.com>
13109 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
13110 frame sizes that do not satisfy aarch64_uimm12_shift.
13112 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
13114 * profile-count.h (apply_probability,
13115 apply_scale, probability_in): Fix checks for zero.
13117 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
13119 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
13120 * doc/cppdiropts.texi (-I @var{dir}): Document it.
13122 2016-06-22 Richard Biener <rguenther@suse.de>
13124 * tree-vect-loop.c (vect_model_reduction_cost): Handle
13125 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
13126 REDUC_MAX_EXPR support.
13127 (vectorizable_reduction): Likewise.
13128 (vect_create_epilog_for_reduction): Likewise.
13130 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
13132 * match.pd (A / (1 << B) -> A >> B): New.
13133 * generic-match-head.c: Include optabs-tree.h.
13134 * gimple-match-head.c: Likewise.
13135 * optabs-tree.h (target_supports_op_p): New.
13136 * optabs-tree.c (target_supports_op_p): New.
13138 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
13140 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
13141 $gcc_cv_ld --help output.
13142 (gcc_cv_ld_demangle): Likewise.
13143 (gcc_cv_ld_eh_frame_hdr): Likewise.
13144 (gcc_cv_ld_pie): Likewise.
13145 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
13146 (gcc_cv_ld_buildid): Likewise.
13147 (gcc_cv_ld_sysroot): Likewise.
13148 (ld_bndplt_support): Likewise.
13149 (ld_pushpopstate_support): Likewise.
13150 * configure: Regenerate.
13151 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
13153 2017-06-21 Jakub Jelinek <jakub@redhat.com>
13156 * config/i386/sse.md (round<mode>2): Renumber match_dup and
13157 operands indexes to avoid gap between operands and match_dups.
13159 2017-06-21 Andrew Pinski <apinski@cavium.com>
13161 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
13162 Increment Arith_shift and Arith_shift_reg by 1.
13163 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
13165 * config/aarch64/aarch64.c (thunderx_tunings): Enable
13166 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
13167 (aarch64_strip_extend): Add new argument and test for it.
13168 (aarch64_cheap_mult_shift_p): New function.
13169 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
13170 add a cost if it is true.
13171 Update calls to aarch64_strip_extend.
13172 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
13174 2017-06-21 Andrew Pinski <apinski@cavium.com>
13176 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
13178 (thunderxt88): Likewise.
13179 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
13180 (thunderx_prefetch_tune): New variable.
13181 (thunderx2t99_prefetch_tune): Update for the correct values.
13182 (thunderxt88_tunings): New variable.
13183 (thunderx_tunings): Use thunderx_prefetch_tune instead of
13184 generic_prefetch_tune.
13185 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
13187 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13189 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
13190 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
13191 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
13192 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
13193 (aarch64_atomic_cas<mode>, GPI): Likewise.
13195 2017-06-21 Martin Liska <mliska@suse.cz>
13197 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
13198 statements on cold and hot labels.
13199 * predict.c (tree_estimate_probability_bb): Remove the
13200 prediction from this place.
13202 2017-06-21 Martin Liska <mliska@suse.cz>
13204 PR tree-optimization/79489
13205 * gimplify.c (maybe_add_early_return_predict_stmt): New
13207 (gimplify_return_expr): Call the function.
13208 * predict.c (tree_estimate_probability_bb): Remove handling
13210 * predict.def: Update comment about early return predictor.
13211 * gimple-predict.h (is_gimple_predict): New function.
13212 * predict.def: Change default value of early return to 66.
13213 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
13215 * passes.def: Put pass_strip_predict_hints to the beginning of
13218 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
13220 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
13221 FUNCTION_DECL declarations.
13222 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
13224 (dwaf2out_decl): Likewise.
13225 * godump.c (go_early_global_decl): Skip call to the real debug hook
13226 for FUNCTION_DECL declarations.
13227 * passes.c (rest_of_decl_compilation): Skip call to the
13228 early_global_decl debug hook for FUNCTION_DECL declarations, unless
13229 -fdump-go-spec is passed.
13231 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
13233 * config/i386/i386.c (struct builtin_isa): New field pure_p.
13234 Reorder for compactness.
13235 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
13236 (def_builtin_pure, def_builtin_pure2): New functions.
13237 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
13239 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
13241 * match.pd (nop_convert): New predicate.
13242 ((A +- CST1) +- CST2): Allow some NOP conversions.
13244 2017-06-21 Jakub Jelinek <jakub@redhat.com>
13247 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
13248 with ctors/dtors if GOVD_SHARED is set.
13250 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
13252 * config/aarch64/aarch64.md (movti_aarch64):
13253 Emit mov rather than orr.
13254 (movtf_aarch64): Likewise.
13255 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
13256 Emit mov rather than orr.
13258 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
13260 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
13261 Swap alternatives, make integer dup more expensive.
13263 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
13265 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
13266 Return true for non-tls symbols.
13268 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
13270 * config/aarch64/aarch64-cores.def (cortex-a55): New.
13271 (cortex-a75): Likewise.
13272 (cortex-a75.cortex-a55): Likewise.
13273 * config/aarch64/aarch64-tune.md: Regenerate.
13274 * doc/invoke.texi (-mtune): Document new values for -mtune.
13276 2017-06-21 Tom de Vries <tom@codesourcery.com>
13278 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
13279 stack_size feature.
13280 (Effective-Target Keywords, Other attributes): Suggest using
13281 dg-add-options stack_size feature to get stack limit in stack_size
13282 effective target documentation.
13284 2017-06-21 Julian Brown <julian@codesourcery.com>
13285 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
13287 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
13288 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
13289 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
13291 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
13292 attribute type list for neon_multiply.
13293 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
13294 attribute type list for neon_multiply.
13295 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
13296 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
13297 attribute type list for neon_multiply.
13298 * config/arm/types.md (crypto_pmull): Add.
13299 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
13300 attribute type list.
13302 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
13304 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
13307 2017-06-20 Jakub Jelinek <jakub@redhat.com>
13309 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
13310 to make sure not to dereference a NULL cost_classes_ptr pointer.
13312 2017-06-20 Carl Love <cel@us.ibm.com>
13314 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13315 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
13316 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
13317 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
13318 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
13319 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
13320 VMULOSW): New enum "unspec" values.
13321 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
13322 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
13323 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
13324 altivec_vmulosw): New patterns.
13325 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
13326 VMULOSW): Add definitions.
13328 2017-06-20 Julia Koval <julia.koval@intel.com>
13330 * config/i386/i386.c: Fix rounding expand for new pattern.
13331 * config/i386/subst.md: Fix pattern (parallel -> unspec).
13333 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
13335 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
13336 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
13338 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
13340 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
13343 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
13345 * config/aarch64/aarch64-cores.def: Rearrange to sort by
13346 architecture, then by implementer ID.
13347 * config/aarch64/aarch64-tune.md: Regenerate.
13349 2017-06-20 Richard Biener <rguenther@suse.de>
13351 PR middle-end/81097
13352 * fold-const.c (split_tree): Fold to type before negating.
13354 2017-06-20 David Malcolm <dmalcolm@redhat.com>
13356 * diagnostic-show-locus.c
13357 (selftest::test_fixit_deletion_affecting_newline): New function.
13358 (selftest::diagnostic_show_locus_c_tests): Call it.
13360 2017-06-20 Andreas Schwab <schwab@suse.de>
13363 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
13366 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
13368 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
13369 __ARM_FEATURE_COPROC according to support.
13371 2017-06-20 Jakub Jelinek <jakub@redhat.com>
13373 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
13374 Rewritten to avoid overflow for > 32-bit pointers.
13377 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
13378 by removing enum keyword.
13379 (ubsan_type_descriptor): Likewise. Formatting fix.
13382 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
13383 splitter): Require TARGET_SSE2 in the condition.
13385 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
13388 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
13389 for doing vector set of SFmode on ISA 3.0.
13390 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
13391 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
13393 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
13394 SFmode value into a V4SF variable that was extracted from another
13395 V4SF variable without converting the element to double precision
13396 and back to single precision vector format.
13397 (vsx_insert_extract_v4sf_p9_2): Likewise.
13399 2017-06-19 Jakub Jelinek <jakub@redhat.com>
13401 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
13402 in UWHI to avoid undefined overflow.
13405 * ubsan.h (enum ubsan_encode_value_phase): New.
13406 (ubsan_encode_value): Change second argument to
13407 enum ubsan_encode_value_phase with default value of
13408 UBSAN_ENCODE_VALUE_GENERIC.
13409 * ubsan.c (ubsan_encode_value): Change second argument to
13410 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
13411 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
13412 create_tmp_var_raw instead of create_tmp_var and use a
13414 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
13415 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
13416 ubsan_encode_value callers.
13419 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
13420 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
13421 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
13423 2017-06-19 Richard Biener <rguenther@suse.de>
13425 PR middle-end/81118
13426 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
13427 estimates if we changed anything.
13429 2017-06-19 Richard Biener <rguenther@suse.de>
13431 PR tree-optimization/80887
13432 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
13433 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
13434 simplified lookups, then reset mprts_hook.
13435 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
13437 (try_to_simplify): Likewise.
13439 2017-06-19 Martin Liska <mliska@suse.cz>
13442 * gimplify.c (gimplify_switch_expr):
13443 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
13445 2017-06-19 Martin Liska <mliska@suse.cz>
13447 * doc/install.texi: Document that PGO runs in 4 stages.
13449 2017-06-19 Martin Liska <mliska@suse.cz>
13452 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
13453 to dispatcher function name.
13454 * multiple_target.c (replace_function_decl): New function.
13455 (create_dispatcher_calls): Redirect both edges and references.
13457 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
13459 * profile-count.c (profile_count::dump): Dump quality.
13460 (profile_count::differs_from_p): Update for unsigned val.
13461 * profile-count.h (profile_count_quality): New enum.
13462 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
13464 2017-06-19 Richard Biener <rguenther@suse.de>
13466 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
13467 struct function as arg.
13468 (estimate_numbers_of_iterations): Export overload with loop arg.
13469 (free_numbers_of_iterations_estimates_loop): Use an overload of
13470 free_numbers_of_iterations_estimates instead.
13471 * tree-cfg.c (remove_bb): Adjust.
13472 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
13473 * tree-parloops.c (gen_parallel_loop): Likewise.
13474 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
13476 (tree_unroll_loops_completely): Likewise.
13477 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
13478 Use an overload instead and export.
13479 (estimated_loop_iterations): Adjust.
13480 (max_loop_iterations): Likewise.
13481 (likely_max_loop_iterations): Likewise.
13482 (estimate_numbers_of_iterations): Take struct function as arg
13484 (loop_exits_before_overflow): Adjust.
13485 (free_numbers_of_iterations_estimates_loop): Use an overload.
13486 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
13487 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
13489 2017-06-19 Richard Biener <rguenther@suse.de>
13492 * ipa-prop.c (find_constructor_constant_at_offset): Handle
13493 RANGE_EXPR conservatively.
13495 2017-06-16 Carl Love <cel@us.ibm.com>
13497 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13498 definitions for vec_float, vec_float2, vec_floato,
13499 vec_floate built-ins.
13500 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
13501 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
13503 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
13504 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
13505 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
13506 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
13507 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
13508 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
13509 vec_floato): Add builtin defines.
13510 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
13511 Update the built-in documentation file for the new built-in
13514 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13516 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
13517 (mthumb): Mark as the negative of -marm.
13519 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13521 * doc/invoke.texi (ARM Options, -mcpu): Document supported
13523 (ARM Options, -mtune): Document that this accepts the same
13524 extension options as -mcpu.
13525 (ARM Options, -mfpu): Document addition of -mfpu=auto.
13527 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13529 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
13530 permitted extensions.
13532 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13534 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
13535 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
13536 (armv8-m.main): Add option +nodsp.
13537 * config/arm/arm-cpu-cdata.h: Regenerated.
13539 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13541 * config/arm/t-fuchsia: New file.
13542 * config.gcc (arm*-*-fuchsia*): Use it.
13544 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13546 * config/arm/t-symbian: Rewrite for new option infrastructure.
13548 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13550 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
13551 (MULTILIB_REQUIRED): Likewise.
13553 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13555 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
13556 (MULTILIB_RESUE): Likewise.
13557 (MULTILIB_MATCHES): Likewise.
13558 (MULTLIB_REQUIRED): Likewise.
13560 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13562 * config/arm/t-rtems: Rewrite for new option framework.
13564 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13566 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
13567 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
13568 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
13569 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
13570 * config/arm/t-multilib: ... here.
13571 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
13572 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
13573 armv7-a and armv8*-a when A-profile libraries have not been built.
13574 * config/arm/t-rmprofile: Rewrite.
13576 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13578 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
13579 with a backslash. Remove the backslash after substituting unescaped
13581 * doc/fragments.texi (MULTILIB_REUSE): Document it.
13583 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13585 * config.gcc: (arm*-*-*): When building a-profile libraries, force
13586 the driver to pass through the default setting of -mfloat-abi.
13587 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
13589 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
13590 (all_feat_combs): New rule.
13591 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
13593 * config/arm/t-aprofile: Rewrite.
13595 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13597 * config/arm/arm.h (FPUTYPE_AUTO): Define.
13598 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
13599 fpu is not specified by the user/command-line.
13600 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
13601 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
13602 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
13603 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
13604 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
13605 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
13607 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13609 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
13610 * config/arm/t-arm-elf: Rewritten.
13612 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13614 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
13615 have some floating-point instructions.
13616 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
13617 (TARGET_MAYBE_HARD_FLOAT): New macro.
13618 * config/arm/arm-builtins.c (arm_init_builtins): Use
13619 TARGET_MAYBE_HARD_FLOAT.
13620 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
13622 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13624 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
13625 (configargs.h): Include it.
13626 (arm_print_hint_for_fpu_option): New function.
13627 (arm_parse_fpu_option): New function.
13628 (candidate_extension): New class.
13629 (arm_canon_for_multilib): New function.
13630 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
13631 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
13632 (ARCH_CANONICAL_SPECS): New macro.
13633 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
13635 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13637 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
13638 are set after handling multilib fragments. Set target_cpu_default2
13641 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13643 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
13645 (arm*-*-*): Set target_cpu_default2 to a quoted string.
13646 * config/arm/parsecpu.awk (check_cpu): Validate any extension
13648 (check_arch): Likewise.
13649 * config/arm/arm.c (arm_configure_build_target): Handle
13650 TARGET_CPU_DEFAULT being a string constant. Scan any feature
13651 options in the default.
13653 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13655 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
13656 when an option is an alias of another.
13657 * config/arm/parsecpu.awk (optalias): New parser token.
13658 (gen_comm_data): Mark non-alias options as such. Emit entries
13659 for extension aliases.
13660 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
13661 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
13662 (armv6kz, armv6zk, armv6t2): Likewise.
13663 (armv7): Make vfpv3-d16 an alias.
13664 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
13666 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
13667 Sort in canonical order.
13668 (armv8-a): Sort in canonical order.
13669 (armv8.1-a, armv8.2-a): Likewise.
13670 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
13672 (cortex-a9): Sort in canonical order.
13673 * config/arm/arm.c (selftests.h): Include it.
13674 (arm_test_cpu_arch_data): New function.
13675 (arm_run_self_tests): New function.
13676 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
13677 (targetm): Move declaration to the end of the file.
13678 * arm-cpu-cdata.h: Regenerated.
13680 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13682 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
13683 call to target_mode_check describing the type of option passed.
13684 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
13685 (arm_target_thumb_only): Use arm_parse_arch_option_name or
13686 arm_parse_cpu_option_name to match parameters against list of
13688 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
13689 arm_arch_core_flags data structure.
13690 * config/arm/arm-cpu_cdata.h: Regenerated.
13692 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13694 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
13696 (arm_print_hint_for_cpu_option): Likewise.
13697 (arm_print_hint_for_arch_option): Likewise.
13698 (arm_parse_cpu_option_name): Likewise.
13699 (arm_parse_arch_option_name): Likewise.
13700 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
13701 of entries in the all_fpus list.
13702 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
13703 (arm_parse_cpu_option_name): Declare.
13704 (arm_parse_arch_option_name): Declare.
13705 (arm_parse_option_features): Declare.
13706 (arm_intialize_isa): Declare.
13707 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
13709 (gen_comm_data): ... here. Make definitions non-static.
13710 * config/arm/arm-cpu-data.h: Regenerated.
13711 * config/arm/arm-cpu-cdata.h: Regenerated.
13713 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13715 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
13716 (cpu_arch_extension): New structure.
13717 (cpu_arch_option, arch_option, cpu_option): New structures.
13718 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
13719 architecture types.
13720 (gen_data): Generate new format data tables.
13721 * config/arm/arm.c (cpu_tune): New structure.
13722 (cpu_option, processors): Delete.
13723 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
13724 (arm_print_hint_for_cpu_option): ... this and ...
13725 (arm_print_hint_for_arch_option): ... this.
13726 (arm_parse_arch_cpu_name): Delete. Replace with ...
13727 (arm_parse_cpu_option_name): ... this and ...
13728 (arm_parse_arch_option_name): ... this.
13729 (arm_unrecognized_feature): Change type of target parameter to
13731 (arm_parse_arch_cpu_features): Delete. Replace with ...
13732 (arm_parse_option_features): ... this.
13733 (arm_configure_build_target): Rework to use new configuration data
13735 (arm_print_tune_info): Rework for new configuration data tables.
13736 * config/arm/arm-cpu-data.h: Regenerated.
13737 * config/arm/arm-cpu.h: Regenerated.
13739 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13741 * Makefile.in (OBJS): Move sbitmap.o from here ...
13742 (OBJS-libcommon): ... to here.
13744 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13746 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
13747 (ISA_ALL_CRYPTO): New macro.
13748 (ISA_ALL_SIMD): New macro
13749 (ISA_ALL_FP): New macro.
13750 * config/arm/arm.c (fpu_bitlist): Update initializer.
13751 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
13753 (arm9e): Add fpu. Add option for nofp
13754 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
13755 (arm926ej-s, arm1026ej-s): Likewise.
13756 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
13757 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
13758 neon-fp16, neon-vfpv4, nofp and nosimd.
13759 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
13760 (cortex-a8): Add fpu. Add option for nofp.
13761 (cortex-a9): Add fpu. Add options for nosimd and nofp.
13762 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
13763 (cortex-r4f): Add fpu.
13764 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
13765 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
13767 (cortex-r8): Likewise.
13768 (cortex-m4): Add fpu. Add option for nofp.
13769 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
13770 (cortex-a17.cortex-a7): Likewise.
13771 (cortex-a32): Add fpu. Add options for crypto and nofp.
13772 (cortex-a35, cortex-a53): Likewise.
13773 (cortex-a57): Add fpu. Add option for crypto.
13774 (cortex-a72, cortex-a73): Likewise.
13775 (exynos-m1): Likewise.
13776 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
13777 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
13778 (cortex-m33): Add fpu. Add option for nofp.
13779 * config/arm/arm-cpu-cdata.h: Regenerated
13780 * config/arm/arm-cpu-data.h: Regenerated.
13782 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13784 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
13785 (armv5te, armv5tej): Likewise.
13786 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
13787 (armv7): Add options fp and vfpv3-d16.
13788 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
13789 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
13791 (armv7ve): Likewise.
13792 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
13793 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
13794 (armv8-a): Add nocrypto option.
13795 (armv8.1-a, armv8.2-a): Likewise.
13796 (armv8-m.main): add options fp, fp.dp and nofp.
13798 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13800 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
13802 (armv8-a+crc): Delete.
13803 (armv8.1-a): Add options simd, crypto and nofp.
13804 (armv8.2-a): Add options fp16, simd, crypto and nofp.
13805 (armv8.2-a+fp16): Delete.
13806 (armv8-m.main): Add option dsp.
13807 (armv8-m.main+dsp): Delete.
13808 (cortex-a8): Add fpu. Add nofp option.
13809 (cortex-a9): Add fpu. Add nofp and nosimd options.
13810 * config/arm/parsecpu.awk (gen_data): Generate option tables and
13811 link to main cpu and architecture data structures.
13812 (gen_comm_data): Only put isa attributes from the main architecture
13814 (option): New statement for architecture and CPU entries.
13815 * arm.c (struct cpu_option): New structure.
13816 (struct processors): Add entry for options.
13817 (arm_unrecognized_feature): New function.
13818 (arm_parse_arch_cpu_name): Ignore any characters after the first
13820 (arm_parse_arch_cpu_feature): New function.
13821 (arm_configure_build_target): Separate out any CPU and architecture
13822 features and parse separately. Don't error out if -mfpu=auto is
13823 used with only an architecture string.
13824 (arm_print_asm_arch_directives): New function.
13825 (arm_file_start): Call it.
13826 * config/arm/arm-cpu-cdata.h: Regenerated.
13827 * config/arm/arm-cpu-data.h: Likewise.
13828 * config/arm/arm-tables.opt: Likewise.
13830 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13832 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
13833 assembler when it is not -mfpu=auto.
13835 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13837 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
13838 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
13839 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
13840 (ASM_CPU_SPEC): Rewrite.
13841 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
13842 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
13843 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
13844 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
13845 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
13846 copied string is NUL-terminated. Also strip any characters prefixed
13848 (arm_rewrite_selected_arch): New function.
13849 (arm_rewrite_march): New function.
13851 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13853 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
13854 (x_arm_cpu_string, x_arm_tune_string): Likewise.
13855 (march, mcpu, mtune): Convert to string-based options.
13856 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
13857 (arm_parse_arch_cpu_name): New function.
13858 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
13859 identify selected architecture or CPU.
13860 (arm_option_save): New function.
13861 (TARGET_OPTION_SAVE): Redefine.
13862 (arm_option_restore): Restore string options.
13863 (arm_option_print): Print string options.
13865 2017-06-16 Martin Sebor <msebor@redhat.com>
13867 PR tree-optimization/80933
13868 PR tree-optimization/80934
13869 * builtins.c (fold_builtin_3): Do not handle bcmp here.
13870 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
13871 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
13872 (gimple_fold_builtin): Call them.
13874 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13876 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
13877 as unlikely; update profile.
13879 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13881 * predict.c (force_edge_cold): Handle declaring edges impossible
13884 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13886 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
13888 (try_unroll_loop_completely): Fix reporting.
13890 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13892 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
13894 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
13897 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
13898 if given a non-constant argument for an intrinsic which requires a
13901 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13903 * profile.c (compare_freqs): New function.
13904 (branch_prob): Sort edge list.
13905 (find_spanning_tree): Assume that the list is priority sorted.
13907 2017-06-16 Richard Biener <rguenther@suse.de>
13909 PR tree-optimization/81090
13910 * passes.def (pass_record_bounds): Remove.
13911 * tree-pass.h (make_pass_record_bounds): Likewise.
13912 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
13913 make_pass_record_bounds): Likewise.
13914 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
13915 not free niter estimates at the beginning but at the end.
13916 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
13918 2017-06-16 Richard Biener <rguenther@suse.de>
13920 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
13921 initializer to workaround ICE in host GCC 4.8.
13923 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13925 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
13927 (clone_inlined_nodes): Update.
13929 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
13931 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
13932 prefetch settings, and enable prefetching by default at -O3.
13934 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
13936 * config/aarch64/aarch64.c (aarch64_override_options_internal):
13937 Set flag_prefetch_loop_arrays according to tuning data.
13939 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
13941 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
13942 New tune structure.
13943 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
13944 [Unrelated to main purpose of the patch] Place the pointer field last
13945 to enable type checking errors when tune structure are wrongly merged.
13946 * config/aarch64/aarch64.c (generic_prefetch_tune,)
13947 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
13948 (thunderx2t99_prefetch_tune): New tune constants.
13949 (tune_params *_tunings): Update all tunings (no functional change).
13950 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
13951 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
13952 from tunings structures.
13954 2017-06-16 Jakub Jelinek <jakub@redhat.com>
13957 * ubsan.c (instrument_null): Add T argument, use it instead
13958 of computing it based on IS_LHS.
13959 (instrument_object_size): Likewise.
13960 (pass_ubsan::execute): Adjust instrument_null and
13961 instrument_object_size callers to pass gimple_get_lhs or
13962 gimple_assign_rhs1 result to it. Use instrument_null instead of
13963 calling get_base_address and instrument_mem_ref. Handle
13964 aggregate call arguments for object-size sanitization.
13966 2017-06-16 Yury Gribov <tetra2005@gmail.com>
13968 PR tree-optimization/81089
13969 * tree-vrp.c (is_masked_range_test): Validate operands of
13972 2017-06-15 Martin Sebor <msebor@redhat.com>
13975 * dumpfile.c (dump_register): Avoid calling memset to initialize
13976 a class with a default ctor.
13977 * gcc.c (struct compiler): Remove const qualification.
13978 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
13979 * hash-table.h: Ditto.
13980 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
13982 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
13983 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
13985 * params.h (struct param_info): Make struct members non-const.
13986 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
13987 with default initialization.
13988 * vec.h (vec_copy_construct, vec_default_construct): New helper
13990 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
13991 with vec_copy_construct.
13992 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
13993 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
13994 * doc/invoke.texi (-Wclass-memaccess): Document.
13996 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
13998 * emit-rtl.h (is_leaf): Update comment about local
13999 register allocator.
14001 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
14004 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
14005 for a variable to have a section before checking if the section has a
14007 Set section to.persistent if persistent attribute is set.
14008 Warn if .persistent attribute is used on an automatic variable.
14010 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
14012 PR rtl-optimization/80474
14013 * reorg.c (update_block): Do not ignore instructions in a delay slot.
14015 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
14017 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
14020 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
14022 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
14023 (casesi): Emit bounds checking as RTL.
14024 (casesi_internal_mips16_<mode>): Remove bounds checking.
14026 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
14028 * config/xtensa/xtensa.c (xtensa_option_override): Append
14029 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
14030 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
14031 xtensa_doloop_hooks): Define unconditionally.
14032 (xtensa_reorg_loops): Only call reorg_loops in the presence of
14034 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
14035 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
14036 for it in xtensa_option_override.
14037 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
14038 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
14040 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
14042 * doc/cppopts.texi: Document '-' special value to -MF.
14044 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
14046 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
14047 (cortex_a53_fconst): Likewise.
14048 (cortex_a53_fpmul): Likewise.
14049 (cortex_a53_f_load_64): Likewise.
14050 (cortex_a53_f_load_many): Likewise.
14051 (cortex_a53_advsimd_alu): Likewise.
14052 (cortex_a53_advsimd_alu_q): Likewise.
14053 (cortex_a53_advsimd_mul): Likewise.
14054 (cortex_a53_advsimd_mul_q): Likewise.
14055 (fpmac bypass): Add new bypass for fpmac-fpmac case.
14056 Add missing fmul, r2f_cvt and fconst cases.
14058 2017-06-14 Richard Biener <rguenther@suse.de>
14060 PR middle-end/81088
14061 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
14063 (fold_binary_loc): When associating do not treat pre-existing
14064 TREE_OVERFLOW on literal constants as a reason to allow
14065 TREE_OVERFLOW on associated literal constants.
14067 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
14069 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
14070 (MASK_FEATURES): New macro.
14071 * config/sparc/sparc.c (sparc_option_override): Remove the special
14072 handling of -mfpu and generalize it to all MASK_FEATURES switches.
14074 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
14076 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
14077 a division of 0 if non-call exceptions are enabled.
14079 2017-06-14 Andrew Pinski <apinski@cavium.com>
14080 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
14083 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
14084 Improve vector initialization code gen for only variable case.
14086 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
14088 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
14090 2017-06-14 Richard Biener <rguenther@suse.de>
14092 PR tree-optimization/81083
14093 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
14096 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14098 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
14099 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
14100 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
14101 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
14102 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
14103 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
14105 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14107 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
14108 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
14110 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14112 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
14114 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14116 * config/rs6000/t-rtems: Don't handle SPE.
14118 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14120 * config/rs6000/t-linux: Don't handle SPE.
14122 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14124 * config/rs6000/eabispe.h: Delete file.
14126 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14128 * config/rs6000/t-spe: Delete file.
14130 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14132 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
14133 (rs6000_legitimate_offset_address_p): Return false for anything in
14134 V2SImode or V2SFmode.
14136 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
14138 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
14139 except V2SF and V2SI. Rearrange the vector modes, and add comments.
14140 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
14142 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
14143 (rs6000_legitimate_offset_address_p): Ditto.
14144 (rs6000_emit_move): Ditto.
14145 (rs6000_init_builtins): Remove V4HI_type_node.
14147 2017-06-13 Martin Liska <mliska@suse.cz>
14150 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
14151 (gate_asan): Likewise.
14152 * asan.h (asan_no_sanitize_address_p): Remove the function.
14153 (sanitize_flags_p): New function.
14154 * builtins.def: Fix coding style.
14155 * common.opt: Use renamed enum value.
14156 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
14157 * doc/extend.texi: Document no_sanitize attribute.
14158 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
14159 to SANITIZE_UNDEFINED_NONDEFAULT.
14160 * gcc.c (sanitize_spec_function): Use the renamed enum value.
14161 * gimple-fold.c (optimize_atomic_compare_exchange_p):
14162 Use sanitize_flags_p.
14163 * gimplify.c (gimplify_function_tree): Likewise.
14164 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
14165 * opts.c (parse_no_sanitize_attribute): New function.
14166 (common_handle_option): Use renamed enum value.
14167 * opts.h (parse_no_sanitize_attribute): Declare.
14168 * tree.c (sanitize_flags_p): New function.
14169 * tree.h: Declared here.
14170 * tsan.c: Use sanitize_flags_p.
14171 * ubsan.c (ubsan_expand_null_ifn): Likewise.
14172 (instrument_mem_ref): Likewise.
14173 (instrument_bool_enum_load): Likewise.
14174 (do_ubsan_in_current_function): Remove the function.
14175 (pass_ubsan::execute): Use sanitize_flags_p.
14176 * ubsan.h: Remove do_ubsan_in_current_function
14177 * tree-cfg.c (print_no_sanitize_attr_value): New function.
14178 (dump_function_to_file): Use it here.
14180 2017-06-13 Martin Jambor <mjambor@suse.cz>
14182 PR tree-optimization/80803
14183 PR tree-optimization/81063
14184 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
14185 (propagate_subaccesses_across_link): Enqueue subtree whenever
14186 necessary instead of relying on the caller.
14188 2017-06-13 Martin Jambor <mjambor@suse.cz>
14190 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
14191 that have a first_link.
14192 (sort_and_splice_var_accesses): Do not check first_link before
14194 (subtree_mark_written_and_enqueue): Likewise.
14195 (propagate_all_subaccesses): Likewise and do not stop at first
14196 parent with a first_link.
14198 2017-06-13 Martin Jambor <mjambor@suse.cz>
14200 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
14203 2017-06-13 Yury Gribov <tetra2005@gmail.com>
14205 * match.pd: New pattern.
14207 2017-06-13 Yury Gribov <tetra2005@gmail.com>
14209 * tree-vrp.c (is_masked_range_test): New function.
14210 (register_edge_assert_for): Determine ranges for
14213 2017-06-13 Yury Gribov <tetra2005@gmail.com>
14215 PR tree-optimization/67328
14216 * fold-const.c (maskable_range_p): New function.
14217 (build_range_check): Generate bittests if possible.
14219 2017-06-13 Martin Liska <mliska@suse.cz>
14221 * gimple-pretty-print.c (dump_probability): Add new argument.
14222 (dump_edge_probability): Dump both probability and count.
14223 (dump_gimple_label): Likewise.
14224 (dump_gimple_bb_header): Likewise.
14226 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
14229 * config/avr/avr-devices.c: Fix indentation.
14230 * config/avr/gen-avr-mmcu-specs.c: Dito.
14232 2017-06-13 Richard Biener <rguenther@suse.de>
14234 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
14235 instead get vector type from stmt_info.
14236 (vectorizable_reduction): Adjust. Remove dead code.
14238 2017-06-13 Richard Biener <rguenther@suse.de>
14240 PR middle-end/81065
14241 * fold-const.c (extract_muldiv_1): Remove bogus distribution
14242 case of C * (x * C2 + C3).
14243 (fold_addr_of_array_ref_difference): Properly fold index difference.
14245 2017-06-12 David S. Miller <davem@davemloft.net>
14248 * config/sparc/sparc.md (return expander): Emit frame blockage if
14249 function uses alloca.
14251 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
14253 * combine.c (make_field_assignment): Check len rather than the mode
14254 precision when calling force_to_mode.
14256 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
14258 Support multilibs and devices that see flash in RAM address range.
14261 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
14262 (avr_mcu_t) <flash_pm_offset>: New field.
14263 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
14264 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
14265 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
14266 (AVR_TINY_PM_OFFSET): Remove macro.
14267 * config/avr/avr.opt (-mshort-calls): New option.
14268 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
14269 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
14270 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
14271 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
14272 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
14273 instead of avr_arch->have_jmp_call.
14274 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
14275 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
14276 avr_arch->flash_pm_offset to define.
14277 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
14278 new field flash_pm_offset. Add entry for avrxmega3.
14279 (avr_texinfo): Add entry for avrxmega3.
14280 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
14281 attiny212, attiny214,
14282 attiny412, attiny414, attiny416, attiny417,
14283 attiny814, attiny816, attiny817,
14284 attiny1614, attiny1616, attiny1617,
14285 attiny3214, attiny3216, attiny3217.
14286 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
14287 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
14288 (avr_print_operand_address) [AVR_TINY]: Same.
14289 (avr_asm_init_sections) <readonly_data_section>: Only patch
14290 callback if avr_arch->flash_pm_offset = 0.
14291 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
14292 for rodata if avr_arch->flash_pm_offset != 0.
14293 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
14294 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
14295 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
14296 (m_options): Append opt_rcall.
14297 (m_dirnames): Append dir_rcall.
14298 * config/avr/t-multilib: Regenerate.
14300 * configure.ac [target=avr]: Check whether avrxmega3 default
14301 linker description file works as needed.
14302 * configure: Regenerate.
14303 * doc/avr-mmcu.texi: Regenerate.
14304 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
14305 <__AVR_ARCH__>: Document avrxmega3 and 103.
14306 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
14307 <__AVR_SHORT_CALLS__>: Document it.
14308 <__AVR_PM_BASE_ADDRESS__>: Document it.
14309 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
14310 (AVR Variable Attributes) <progmem>: Document this is
14311 not needed for avrxmega3.
14312 (AVR Named Address Spaces) <__flash>: Dito.
14314 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
14316 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
14318 2017-06-12 Doug Rupp <rupp@adacore.com>
14320 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
14321 Append vxworks-stdint.h to the tm_file list.
14322 * config/vxworks-stdint.h: New file.
14324 2017-06-12 Martin Liska <mliska@suse.cz>
14326 PR tree-optimization/81041
14327 * tree-profile.c (gimple_gen_ic_func_profiler):
14328 Create an extra BB in profile-generate
14329 (gimple_gen_time_profiler): Likewise.
14331 2017-06-12 Jakub Jelinek <jakub@redhat.com>
14333 PR tree-optimization/81003
14334 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
14335 (update_range_test): Use it instead of force_gimple_operand_gsi.
14337 2017-06-12 Richard Biener <rguenther@suse.de>
14339 PR tree-optimization/81053
14340 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
14341 with backedge value not defined in loop. Simplify def stmt
14344 2017-06-11 Tom de Vries <tom@codesourcery.com>
14347 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
14349 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
14350 nvptx_cannot_force_const_mem.
14352 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
14354 * opts.c (finish_options): Move test for flag_split_stack after
14355 it has been initialized.
14357 2017-06-11 Jason Merrill <jason@redhat.com>
14359 * tree.h (id_equal): New.
14360 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
14361 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
14362 instead of strcmp of IDENTIFIER_POINTER.
14364 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
14366 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
14367 (mark_all_inlined_calls_cdtor): Fix formating.
14368 (inline_transform): Rescale profile before inlining.
14370 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
14372 * cgraph.h (cgraph_edge::clone): Update prototype.
14373 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
14374 (cgraph_node::create_clone): Update.
14375 (cgraph_node::create_version_clone): Update.
14376 * tree-inline.c (copy_bb): Update.
14377 (expand_call_inline): Update.
14379 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
14381 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
14382 factored out from ...
14383 (rs6000_emit_prologue): ... here.
14385 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
14387 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
14388 factored out from ...
14389 (rs6000_emit_prologue): ... here.
14391 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
14393 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
14395 (handle_missing_profiles): Fix computation of tp_first_run.
14396 (counts_to_freqs): Do not touch freqs when count is 0.
14398 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
14400 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
14403 2017-06-10 Tom de Vries <tom@codesourcery.com>
14405 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
14406 attributes): Document signal effective target.
14408 2017-06-10 Tom de Vries <tom@codesourcery.com>
14410 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
14411 Document effective target stack_size.
14413 2017-06-09 David Malcolm <dmalcolm@redhat.com>
14415 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
14416 to the edit_context if they can be auto-applied.
14418 2017-06-9 Ian Lance Taylor <iant@golang.org>
14420 * opts.c (finish_options): If -fsplit-stack, disable implicit
14421 -forder-blocks-and-partition.
14422 * doc/invoke.texi (Optimize Options): Document that when using
14423 -fsplit-stack -forder-blocks-and-partition is not implicitly
14426 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
14428 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
14429 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
14430 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
14431 * builtins.def (abort, trap, unreachable): Declare cold.
14432 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
14433 * tree-core.h (ECF_COLD): New.
14434 * tree.c (set_call_expr_flags): Handle ECF_COLD.
14435 (build_common_builtin_nodes): Mark unreachable and abort as cold.
14437 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
14439 * predict.c (unlikely_executed_stmt_p): Cleanup.
14441 2017-06-09 Richard Biener <rguenther@suse.de>
14443 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
14444 model if the ref is always written to.
14446 2017-06-09 Tamar Christina <tamar.christina@arm.com>
14448 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
14450 2017-06-09 Tamar Christina <tamar.christina@arm.com>
14452 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
14455 2017-06-09 Tom de Vries <tom@codesourcery.com>
14458 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
14459 "target cannot support label values" when encountering LABEL_REF.
14461 2017-06-09 Martin Liska <mliska@suse.cz>
14463 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
14464 (gimple_gen_ic_func_profiler): Emit direct comparison
14465 of __gcov_indirect_call_callee with NULL.
14466 (gimple_gen_time_profiler): Change probability from
14467 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
14469 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
14471 * profile.c (edge_gcov_counts): Turn to pointer.
14472 (compute_branch_probabilities, compute_branch_probabilities): Update.
14473 (branch_prob): Do not clear edge_gcov_count.
14474 * profile.h (edge_gcov_counts): Turn to pointer.
14475 (edge_gcov_count): Update.
14477 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
14479 * gimple.h (gimple_check_failed): Mark cold.
14481 2017-06-09 Richard Biener <rguenther@suse.de>
14483 PR tree-optimization/66623
14484 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
14485 refactor check_reduction into two parts, properly computing
14486 whether we have to check reduction validity for outer loop
14489 2017-06-09 Richard Biener <rguenther@suse.de>
14491 PR tree-optimization/79483
14492 * graphite-scop-detection.c (order): New global.
14493 (get_order): Compute bb to order mapping that satisfies code
14494 generation constraints.
14495 (cmp_pbbs): New helper.
14496 (build_scops): Start domwalk at entry block, sort generated
14499 2017-06-09 Richard Biener <rguenther@suse.de>
14501 PR middle-end/81007
14502 * ipa-polymorphic-call.c
14503 (ipa_polymorphic_call_context::restrict_to_inner_class):
14504 Skip FIELD_DECLs with error_mark_node type.
14505 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
14508 2017-06-09 Martin Liska <mliska@suse.cz>
14510 * predict.c (struct branch_predictor): New struct.
14511 (test_prediction_value_range): New test.
14512 (predict_c_tests): New function.
14513 * selftest-run-tests.c (selftest::run_tests): Run the function.
14514 * selftest.h: Declare new tests.
14516 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
14519 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
14520 gen_add3_insn did not fail.
14521 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
14522 r0, construct that number in a temporary reg and add that reg to r0.
14523 If asked to put the result in r0 as well, fail.
14525 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
14527 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14528 for early expansion of vec_eqv.
14530 2017-06-08 Jakub Jelinek <jakub@redhat.com>
14532 PR middle-end/81005
14533 * ubsan.c (instrument_null): Avoid pointless code temporary.
14534 (pass_ubsan::execute): Instrument aggregate arguments of calls.
14536 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
14540 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
14543 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
14546 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
14548 * predict.c (unlikely_executed_edge_p): Move ahead.
14549 (probably_never_executed_edge_p): Use it.
14551 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
14553 PR middle-end/79988
14554 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
14555 gimple_call_builtin_p call.
14557 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
14559 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
14560 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
14561 rtl_check_failed_type2, rtl_check_failed_code1,
14562 rtl_check_failed_code2, rtl_check_failed_code_mode,
14563 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
14564 rtvec_check_failed_bounds, rtl_check_failed_flag,
14565 _fatal_insn_not_found, _fatal_insn): Likewise.
14566 * tree.h (tree_contains_struct_check_failed,
14567 tree_check_failed, tree_not_check_failed,
14568 tree_class_check_failed, tree_range_check_failed,
14569 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
14570 tree_vec_elt_check_failed, phi_node_elt_check_failed,
14571 tree_operand_check_failed, omp_clause_check_failed,
14572 omp_clause_operand_check_failed, omp_clause_range_check_failed):
14575 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
14577 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
14578 flag_branch_probabilities.
14579 * ipa-inline.c (edge_badness): Likewise.
14580 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
14581 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
14582 * predict.c (maybe_hot_frequency_p): Likewise.
14583 (probably_never_executed): Likewise.
14584 * sched-ebb.c (schedule_ebbs): Likewise.
14585 * sched-rgn.c (find_single_block_region): Likewise.
14586 * tracer.c (tail_duplicate): Likewise.
14588 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
14590 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
14591 longer requires x_flag_profile_use.
14593 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
14595 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
14596 instead of flag_reorder_blocks_and_partition.
14597 * dbxout.c (dbxout_function_end): Likewise.
14598 * dwarf2out.c (gen_subprogram_die): Likewise.
14599 * haifa-sched.c (sched_create_recovery_edges): Likewise.
14600 * hw-doloop.c (reorg_loops): Likewise.
14601 * varasm.c (assemble_start_function,
14602 assemble_end_function): Likewise.
14603 (decide_function_section): Do not check for
14604 flag_reorder_blocks_and_partition.
14606 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
14608 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
14610 (chkp_get_hard_register_fake_addr_expr): Ditto.
14611 (chkp_build_addr_expr): Add check for hard reg case.
14612 (chkp_parse_array_and_component_ref): Ditto.
14613 (chkp_find_bounds_1): Ditto.
14614 (chkp_process_stmt): Don't generate bounds store for
14617 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
14619 * predict.c (maybe_hot_bb_p): Do not check profile status.
14620 (maybe_hot_edge_p): Likewise.
14621 (probably_never_executed): Check for zero counts even if profile
14623 (unlikely_executed_edge_p): New function.
14624 (unlikely_executed_stmt_p): New function.
14625 (unlikely_executed_bb_p): New function.
14626 (set_even_probabilities): Use unlikely predicates.
14627 (combine_predictions_for_bb): Likewise.
14628 (predict_paths_for_bb): Likewise.
14629 (predict_paths_leading_to_edge): Likewise.
14630 (determine_unlikely_bbs): New function.
14631 (estimate_bb_frequencies): Use it.
14632 (compute_function_frequency): Use zero counts even if profile is
14634 * profile-count.h: Fix typo.
14636 2017-08-08 Julia Koval <julia.koval@intel.com>
14638 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
14639 _mm512_mask_cvtsepi16_storeu_epi8,
14640 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
14641 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
14642 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
14643 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
14644 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
14645 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
14646 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
14647 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
14648 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
14649 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
14650 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
14651 __builtin_ia32_pmovuswb256mem_mask,
14652 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
14653 __builtin_ia32_pmovwb512mem_mask): New builtins.
14655 2017-08-08 Julia Koval <julia.koval@intel.com>
14657 PR target/73350,80862
14658 * config/i386/subst.md (round): Fix round pattern.
14659 * config/i386/i386.c (ix86_erase_embedded_rounding):
14660 Fix erasing rounding for the fixed pattern.
14662 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
14664 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
14666 2017-06-08 Martin Liska <mliska@suse.cz>
14668 PR gcov-profile/80911
14669 * gcov.c (block_info::block_info): New constructor.
14671 2017-06-07 Carl Love <cel@us.ibm.com>
14673 * config/rs6000/rs6000-c: The return type of the following
14674 built-in functions was implemented as int not long long. Fix sign
14675 of return value for the unsigned version of vec_mulo and vec_mule.
14676 vector unsigned long long vec_bperm (vector unsigned long long,
14677 vector unsigned char)
14678 vector signed long long vec_mule (vector signed int,
14680 vector unsigned long long vec_mule (vector unsigned int,
14681 vector unsigned int)
14682 vector signed long long vec_mulo (vector signed int,
14684 vector unsigned long long vec_mulo (vector unsigned int,
14685 vector unsigned int)
14686 * doc/extend.texi: Fix the documentation for the built-in
14689 2017-06-07 Carl Love <cel@us.ibm.com>
14692 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
14695 2017-06-07 Carl Love <cel@us.ibm.com>
14697 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
14698 support, Generate doublehv for signed int/float for BE case only.
14700 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
14702 * doc/invoke.texi (mcx16): Rewrite.
14704 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14706 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
14707 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
14708 *mov<mode>_softfloat, and an anonymous splitter): Use
14709 nonimmediate_operand instead of rs6000_nonimmediate_operand.
14711 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14713 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
14715 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
14716 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
14717 (rs6000_debug_reg_global): Adjust.
14718 (rs6000_init_hard_regno_mode_ok): Adjust.
14719 (rs6000_dbx_register_number): Adjust.
14720 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
14721 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
14722 Remove SPE_ACC and SPEFSCR.
14723 (REG_ALLOC_ORDER): Ditto.
14724 (FRAME_POINTER_REGNUM): Change to 111.
14725 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
14726 (REG_CLASS_NAMES): Ditto.
14727 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
14728 (REGISTER_NAMES): Ditto.
14729 (ADDITIONAL_REG_NAMES): Ditto.
14730 (rs6000_reg_names): Ditto.
14731 * config/rs6000/rs6000.md: Renumber some register number
14734 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14736 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
14738 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
14739 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
14741 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
14742 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
14743 Delete the SPE high registers.
14744 (REG_ALLOC_ORDER): Ditto.
14745 (enum reg_class): Remove SPE_HIGH_REGS.
14746 (REG_CLASS_NAMES): Ditto.
14747 (REG_CLASS_CONTENTS): Delete the SPE high registers.
14748 (REGISTER_NAMES): Ditto.
14749 (rs6000_reg_names): Ditto.
14750 * doc/tm.texi.in: Remove SPE as example.
14751 * doc/tm.texi: Regenerate.
14753 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14755 * config/rs6000/8540.md (ppc8540_brinc): Delete.
14756 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
14757 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
14758 * config/rs6000/rs6000.md (type): Remove "brinc".
14760 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14762 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
14763 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
14764 * config/rs6000/linuxspe.h: Delete file.
14765 * config/rs6000/rs6000.md: Don't include spe.md.
14766 * config/rs6000/spe.h: Delete file.
14767 * config/rs6000/spe.md: Delete file.
14768 * config/rs6000/t-rs6000: Remove spe.md.
14770 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14772 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
14773 (reg_or_none500mem_operand): Delete.
14774 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
14775 instead of reg_or_none500mem_operand.
14777 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14779 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
14780 handling of SPE flags.
14781 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
14783 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14785 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
14787 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
14788 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
14789 paired_divv2sf3): Similar.
14790 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
14791 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
14792 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
14794 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
14795 Rename the paired_* instruction patterns.
14796 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
14798 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
14799 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
14800 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
14801 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
14802 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
14803 PAIRED_VECTOR_MODE.
14804 (struct machine_function): Delete field spe_insn_chain_scanned_p.
14805 (spe_func_has_64bit_regs_p): Delete.
14806 (spe_expand_predicate_builtin): Delete.
14807 (spe_expand_evsel_builtin): Delete.
14808 (TARGET_DWARF_REGISTER_SPAN): Do not define.
14809 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
14810 (invalid_e500_subreg): Delete.
14811 (rs6000_legitimize_address): Always force_reg op2 as well, for
14812 paired single memory accesses.
14813 (rs6000_member_type_forces_blk): Delete.
14814 (rs6000_spe_function_arg): Delete.
14815 (rs6000_expand_unop_builtin): Delete SPE handling.
14816 (rs6000_expand_binop_builtin): Ditto.
14817 (spe_expand_stv_builtin): Delete.
14818 (bdesc_2arg_spe): Delete.
14819 (spe_expand_builtin): Delete.
14820 (spe_expand_predicate_builtin): Delete.
14821 (spe_expand_evsel_builtin): Delete.
14822 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
14823 (spe_init_builtins): Delete.
14824 (spe_func_has_64bit_regs_p): Delete.
14825 (savres_routine_name): Delete "info" parameter. Adjust callers.
14826 (rs6000_emit_stack_reset): Ditto.
14827 (rs6000_dwarf_register_span): Delete.
14828 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
14829 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
14830 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
14832 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
14834 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
14835 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
14836 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
14837 mulv2sf3, divv2sf3): Delete expanders.
14839 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14841 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
14843 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14845 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
14846 * config/rs6000/rs6000.c: Ditto.
14848 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14850 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
14851 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
14852 comparison_operator.
14854 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14856 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
14857 * config/rs6000/rs6000.opt: Ditto.
14858 * config/rs6000/t-rtems: Ditto.
14860 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14862 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
14863 TARGET_E500_SINGLE by 0, simplify.
14864 * config/rs6000/rs6000.c: Ditto.
14865 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
14866 (spe_build_register_parallel): Delete.
14867 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
14868 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
14869 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
14870 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
14871 (E500_CONVERT): Delete.
14872 * config/rs6000/spe.md: Remove many patterns and all define_constants.
14874 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14876 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
14877 * config/rs6000/dfp.md: Ditto.
14878 (negdd2, *negdd2_fpr): Merge.
14879 (absdd2, *absdd2_fpr): Merge.
14880 (negtd2, *negtd2_fpr): Merge.
14881 (abstd2, *abstd2_fpr): Merge.
14882 * config/rs6000/e500.h: Delete file.
14883 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
14884 TARGET_FPRS by 1 and simplify.
14885 * config/rs6000/rs6000-c.c: Ditto.
14886 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
14887 TARGET_DF_SPE by 0.
14888 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
14890 * config/rs6000/rs6000.md: Ditto.
14891 (floatdidf2, *floatdidf2_fpr): Merge.
14892 (move_from_CR_gt_bit): Delete.
14893 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
14894 (E500_CR_IOR_COMPARE): Delete.
14895 (All patterns that require !TARGET_FPRS): Delete.
14896 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
14898 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14900 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
14902 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14904 * graphds.c (add_edge): Intitialize edge's attached data.
14905 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
14906 pointer parameter. Call pointed function on each edge during
14907 graph traversing. Skip traversing the edge when the function
14909 (graphds_dfs, graphds_scc): Ditto.
14910 (for_each_edge): New parameter. Pass the new parameter to callback
14912 * graphds.h (skip_edge_callback): New function pointer type.
14913 (graphds_dfs, graphds_scc): New function pointer parameter.
14914 (graphds_edge_callback, for_each_edge): New parameter.
14916 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14918 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
14919 out code checking if runtime alias check is possible to below ...
14920 Call the new function.
14921 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
14922 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
14924 2017-06-07 Marek Polacek <polacek@redhat.com>
14927 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
14928 TYPE_OVERFLOW_WRAPS check.
14930 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14932 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
14933 if versioning is required.
14934 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
14935 peeling with the check for versioning.
14937 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14939 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
14940 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
14941 Set true to new parameter if new ssa variable is defined.
14942 (vect_gen_vector_loop_niters): Refactor. Set range information
14943 for the new vector loop bound variable.
14944 (vect_do_peeling): Ditto.
14946 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14948 * tree-affine.c (ssa.h): Include header file.
14949 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
14950 has wrapping overflow behavior.
14952 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14954 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
14956 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14958 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
14959 (tree_to_aff_combination): ... here.
14961 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14963 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
14964 reg_pressure model function.
14965 (ivopts_global_cost_for_size): Delete.
14966 (determine_set_costs, iv_ca_recount_cost): Call new model function
14967 ivopts_estimate_reg_pressure.
14969 2017-06-07 Tamar Christina <tamar.christina@arm.com>
14971 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
14972 expensive than udiv. Remove floating point cases from mod.
14974 2017-06-07 Tamar Christina <tamar.christina@arm.com>
14976 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
14977 Increase idiv cost.
14979 2017-06-07 Tamar Christina <tamar.christina@arm.com>
14981 * config/aarch64/aarch64.md
14982 (copysignsf3): Fix mask generation.
14984 2017-06-07 Jakub Jelinek <jakub@redhat.com>
14986 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
14988 (class dump_manager): Add register_dumps method.
14989 * dumpfile.c: Include langhooks.h.
14990 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
14991 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
14992 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
14993 (dump_manager::dump_register): Start with 512 entries instead of 32.
14994 (dump_manager::register_dumps): New method.
14995 * toplev.c (general_init): Instead of invoking register_dumps
14996 langhook, invoke register_dumps method on the dump manager.
14997 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
15000 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
15002 * doc/md.texi: Clarify the restrictions on a define_insn condition.
15003 Say that # requires an associated define_split to exist, and that
15004 the define_split must be suitable for use after register allocation.
15006 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15008 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
15009 (compute_outgoing_frequencies): Also initialize zero counts.
15010 (find_many_sub_basic_blocks): Do not produce uninitialized profile
15011 around loops; preserve more of profile when nothing changes.
15013 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
15015 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
15017 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
15018 * config/arm/arm-cpu-cdata.h: Regenerate.
15019 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
15020 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
15021 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
15022 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
15023 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
15025 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
15027 * config/arm/t-rmprofile: Likewise.
15028 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
15030 2017-06-06 David S. Miller <davem@davemloft.net>
15033 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
15034 blockage if function uses alloca.
15036 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
15038 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
15039 New "uid" fields to hold pretty-print IDs of group and ref.
15040 Memory references are now identified as <group_id>:<ref_id>
15041 instead of using [random] addresses.
15042 (dump_mem_details): Simplify, no functional change.
15043 (dump_mem_ref): Simplify and make output more concise.
15044 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
15045 (find_or_create_group): Initialize group uid.
15046 (record_ref): Initialize ref uid. Improve debug output.
15047 (prune_group_by_reuse, should_issue_prefetch_p,)
15048 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
15049 (mark_nontemporal_store, determine_loop_nest_reuse):
15050 Improve debug output.
15052 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
15054 * dbgcnt.def (prefetch): New debug counter.
15055 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
15056 (schedule_prefetches): Stop issueing prefetches if debug counter
15059 2017-06-06 Tom de Vries <tom@codesourcery.com>
15061 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
15062 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
15064 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15066 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
15067 Use aarch64_reg_or_zero predicate for operand 4.
15068 (aarch64_compare_and_swap<mode> define_insn_and_split):
15069 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
15070 (aarch64_store_exclusive<mode>): Likewise for operand 2.
15072 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
15074 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
15075 (arm_compute_save_core_reg_mask): This.
15076 (thumb1_compute_save_reg_mask): Rename into ...
15077 (thumb1_compute_save_core_reg_mask): This.
15078 (arm_compute_save_reg0_reg12_mask): Adapt comment.
15079 (arm_compute_frame_layout): Likewise.
15081 2017-06-06 Richard Biener <rguenther@suse.de>
15083 PR tree-optimization/80974
15084 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
15085 keep or clear leaders SSA info.
15087 2017-06-06 Tom de Vries <tom@codesourcery.com>
15089 * config/nvptx/nvptx.c (split_mode_p): New function.
15090 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
15092 2017-06-06 Tom de Vries <tom@codesourcery.com>
15094 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
15096 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15099 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
15102 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15104 * shrink-wrap.c (handle_simple_exit): Update profile.
15105 (try_shrink_wrapping): Upate profile.
15107 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15109 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
15110 (tree_guess_outgoing_edge_probabilities): New.
15111 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
15112 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
15114 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15116 * ipa-split.c (split_function): Initialize return bb profile.
15118 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15120 * profile.c (compute_branch_probabilities): Also initialize
15121 EXIT_BLOCK profile.
15123 2017-06-06 Richard Biener <rguenther@suse.de>
15125 PR tree-optimization/80928
15126 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
15127 (vect_analyze_loop_operations): Properly guard analysis for
15129 (vect_transform_loop): Likewise.
15130 (vect_analyze_loop_2): Also reset SLP type on PHIs.
15131 (vect_model_induction_cost): Do not cost for pure SLP.
15132 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
15133 of induction in inner loop vectorization.
15134 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
15135 (vect_get_and_check_slp_defs): Handle vect_induction_def.
15136 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
15138 (vect_analyze_slp_cost_1): Cost induction.
15139 (vect_detect_hybrid_slp_stmts): Handle PHIs.
15140 (vect_get_slp_vect_defs): Likewise.
15141 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
15142 (vect_transform_stmt): Handle SLP reductions.
15143 * tree-vectorizer.h (vectorizable_induction): Adjust.
15145 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
15147 * config/rs6000/rs6000.c (make_resolver_func): Update
15148 init_lowered_empty_function call.
15150 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
15152 * doc/invoke.texi: Document the -fprofile-abs-path option.
15153 * common.opt (fprofile-abs-path): New option.
15154 * gcov-io.h (gcov_write_filename): Declare.
15155 * gcov-io.c (gcov_write_filename): New function.
15156 * coverage.c (coverage_begin_function): Use gcov_write_filename.
15157 * profile.c (output_location): Likewise.
15159 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
15161 * shring-wrap.c: Revert accidental commit.
15163 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
15165 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
15167 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
15169 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
15171 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
15172 profile in callgraph edge.
15173 * profile-count.h (apply_probability): If THIS is 0, then result is 0
15174 (apply_scale): Likewise.
15175 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
15176 Also scale profile when inlining function with zero profile.
15177 (initialize_cfun): Update exit block profile even when it is zero.
15178 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
15179 when profile is read.
15181 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
15183 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
15184 (CLONE_*): New constants to define the processors we can generate
15185 code for with the target_clone attribute.
15186 (rs6000_clone_map): New array to identify which clone processors
15187 the current program is running on.
15188 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
15189 target_clone attribute.
15190 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
15191 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
15192 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
15193 (cpu_expand_builtin): Add support for target_clone attribute.
15194 (rs6000_valid_attribute_p): Allow "default" attribute.
15195 (get_decl_name): New debug function to simplify printing the
15196 current function name in debugging statements.
15197 (rs6000_clone_priority): New functions to support the target_clone
15198 attribute, and be able to generate code to switch between ISA 2.05
15199 through ISA 3.0 (power6 through power9).
15200 (rs6000_compare_version_priority): Likewise.
15201 (rs6000_get_function_versions_dispatcher): Likewise.
15202 (make_resolver_func): Likewise.
15203 (add_condition_to_bb): Likewise.
15204 (dispatch_function_versions): Likewise.
15205 (rs6000_generate_version_dispatcher_body): Likewise.
15206 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
15207 (fusion_gpr_load_p): Fix a spacing issue.
15208 * doc/extend.texi (Common Function Attributes): Document that the
15209 PowerPC supports the target_clone attribute.
15211 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
15213 * config/arm/arm.h: explain F symbol found in description of ARM
15214 register allocation in its legend.
15216 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
15218 * config/mips/frame-header-opt.c: Include profile-count.h.
15219 * config/riscv/riscv.c: Include profile-count.h
15221 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
15223 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
15225 (sm_set_flag_if_changed): Add bbs field.
15226 (execute_sm_if_changed_flag_set): Pass BBS.
15227 (execute_sm): Update.
15229 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15231 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
15234 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15236 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
15237 (peephole2): New peephole2 to emit the above.
15238 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
15240 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15242 * config/aarch64/aarch64.c (define_peephole2 above
15243 *sub_<shift>_<mode>): New peephole.
15245 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
15247 * config/i386/i386.c (make_resolver_func): Update.
15248 * Makefile.in: Add profile-count.h and profile-count.o
15249 * auto-profile.c (afdo_indirect_call): Update to new API.
15250 (afdo_set_bb_count): Update.
15251 (afdo_propagate_edge): Update.
15252 (afdo_propagate_circuit): Update.
15253 (afdo_calculate_branch_prob): Update.
15254 (afdo_annotate_cfg): Update.
15255 * basic-block.h: Include profile-count.h
15256 (struct edge_def): Turn count to profile_count.
15257 (struct basic_block_def): Likewie.
15258 (REG_BR_PROB_BASE): Move to profile-count.h
15259 (RDIV): Move to profile-count.h
15260 * bb-reorder.c (max_entry_count): Turn to profile_count.
15261 (find_traces): Update.
15262 (rotate_loop):Update.
15263 (connect_traces):Update.
15264 (sanitize_hot_paths):Update.
15265 * bt-load.c (migrate_btr_defs): Update.
15266 * cfg.c (RDIV): Remove.
15267 (init_flow): Use alloc_block.
15268 (alloc_block): Uninitialize count.
15269 (unchecked_make_edge): Uninitialize count.
15270 (check_bb_profile): Update.
15271 (dump_edge_info): Update.
15272 (dump_bb_info): Update.
15273 (update_bb_profile_for_threading): Update.
15274 (scale_bbs_frequencies_int): Update.
15275 (scale_bbs_frequencies_gcov_type): Update.
15276 (scale_bbs_frequencies_profile_count): New.
15277 * cfg.h (update_bb_profile_for_threading): Update.
15278 (scale_bbs_frequencies_profile_count): Declare.
15279 * cfgbuild.c (compute_outgoing_frequencies): Update.
15280 (find_many_sub_basic_blocks): Update.
15281 * cfgcleanup.c (try_forward_edges): Update.
15282 (try_crossjump_to_edge): Update.
15283 * cfgexpand.c (expand_gimple_tailcall): Update.
15284 (construct_exit_block): Update.
15285 * cfghooks.c (verify_flow_info): Update.
15286 (dump_bb_for_graph): Update.
15287 (split_edge): Update.
15288 (make_forwarder_block): Update.
15289 (duplicate_block): Update.
15290 (account_profile_record): Update.
15291 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
15292 (get_estimated_loop_iterations): Update.
15293 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
15294 (single_likely_exit): Update.
15295 * cfgloopmanip.c (scale_loop_profile): Update.
15297 (set_zero_probability): Update.
15298 (lv_adjust_loop_entry_edge): Update.
15299 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
15300 (purge_dead_edges): Update.
15301 (rtl_account_profile_record): Update.
15302 * cgraph.c (cgraph_node::create): Uninitialize count.
15303 (symbol_table::create_edge): Uninitialize count.
15304 (cgraph_update_edges_for_call_stmt_node): Update.
15305 (cgraph_edge::dump_edge_flags): Update.
15306 (cgraph_node::dump): Update.
15307 (cgraph_edge::maybe_hot_p): Update.
15308 * cgraph.h: Include profile-count.h
15309 (create_clone), create_edge, create_indirect_edge): Update.
15310 (cgraph_node): Turn count to profile_count.
15311 (cgraph_edge0: Likewise.
15312 (make_speculative, clone): Update.
15313 (create_edge): Update.
15314 (init_lowered_empty_function): Update.
15315 * cgraphclones.c (cgraph_edge::clone): Update.
15316 (duplicate_thunk_for_node): Update.
15317 (cgraph_node::create_clone): Update.
15318 * cgraphunit.c (cgraph_node::analyze): Update.
15319 (cgraph_node::expand_thunk): Update.
15320 * final.c (dump_basic_block_info): Update.
15321 * gimple-streamer-in.c (input_bb): Update.
15322 * gimple-streamer-out.c (output_bb): Update.
15323 * graphite.c (print_global_statistics): Update.
15324 (print_graphite_scop_statistics): Update.
15325 * hsa-brig.c: Include basic-block.h.
15326 * hsa-dump.c: Include basic-block.h.
15327 * hsa-gen.c (T sum_slice): Update.
15328 (convert_switch_statements):Update.
15329 * hsa-regalloc.c: Include basic-block.h.
15330 * ipa-chkp.c (chkp_produce_thunks): Update.
15331 * ipa-cp.c (struct caller_statistics): Update.
15332 (init_caller_stats): Update.
15333 (gather_caller_stats): Update.
15334 (ipcp_cloning_candidate_p): Update.
15335 (good_cloning_opportunity_p): Update.
15336 (get_info_about_necessary_edges): Update.
15337 (dump_profile_updates): Update.
15338 (update_profiling_info): Update.
15339 (update_specialized_profile): Update.
15340 (perhaps_add_new_callers): Update.
15341 (decide_about_value): Update.
15342 (ipa_cp_c_finalize): Update.
15343 * ipa-devirt.c (struct odr_type_warn_count): Update.
15344 (struct decl_warn_count): Update.
15345 (struct final_warning_record): Update.
15346 (possible_polymorphic_call_targets): Update.
15347 (ipa_devirt): Update.
15348 * ipa-fnsummary.c (redirect_to_unreachable): Update.
15349 * ipa-icf.c (sem_function::merge): Update.
15350 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
15351 * ipa-inline.c (compute_uninlined_call_time): Update.
15352 (compute_inlined_call_time): Update.
15353 (want_inline_small_function_p): Update.
15354 (want_inline_self_recursive_call_p): Update.
15355 (edge_badness): Update.
15356 (lookup_recursive_calls): Update.
15357 (recursive_inlining): Update.
15358 (inline_small_functions): Update.
15359 (dump_overall_stats): Update.
15360 (dump_inline_stats): Update.
15361 * ipa-profile.c (ipa_profile_generate_summary): Update.
15362 (ipa_propagate_frequency): Update.
15363 (ipa_profile): Update.
15364 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
15365 * ipa-utils.c (ipa_merge_profiles): Update.
15366 * loop-doloop.c (doloop_modify): Update.
15367 * loop-unroll.c (report_unroll): Update.
15368 (unroll_loop_runtime_iterations): Update.
15369 * lto-cgraph.c (lto_output_edge): Update.
15370 (lto_output_node): Update.
15371 (input_node): Update.
15372 (input_edge): Update.
15373 (merge_profile_summaries): Update.
15374 * lto-streamer-in.c (input_cfg): Update.
15375 * lto-streamer-out.c (output_cfg): Update.
15376 * mcf.c (create_fixup_graph): Update.
15377 (adjust_cfg_counts): Update.
15378 (sum_edge_counts): Update.
15379 * modulo-sched.c (sms_schedule): Update.
15380 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
15381 * predict.c (maybe_hot_count_p): Update.
15382 (probably_never_executed): Update.
15383 (dump_prediction): Update.
15384 (combine_predictions_for_bb): Update.
15385 (propagate_freq): Update.
15386 (handle_missing_profiles): Update.
15387 (counts_to_freqs): Update.
15388 (rebuild_frequencies): Update.
15389 (force_edge_cold): Update.
15390 * predict.h: Include profile-count.h
15391 (maybe_hot_count_p, counts_to_freqs): UPdate.
15392 * print-rtl-function.c: Do not include cfg.h
15393 * print-rtl.c: Include basic-block.h
15394 * profile-count.c: New file.
15395 * profile-count.h: New file.
15396 * profile.c (is_edge_inconsistent): Update.
15397 (correct_negative_edge_counts): Update.
15398 (is_inconsistent): Update.
15399 (set_bb_counts): Update.
15400 (read_profile_edge_counts): Update.
15401 (compute_frequency_overlap): Update.
15402 (compute_branch_probabilities): Update; Initialize and deinitialize
15404 (branch_prob): Update.
15405 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
15406 (edge_gcov_count): New.
15407 (bb_gcov_count): New.
15408 * shrink-wrap.c (try_shrink_wrapping): Update.
15409 * tracer.c (better_p): Update.
15410 * trans-mem.c (expand_transaction): Update.
15411 (ipa_tm_insert_irr_call): Update.
15412 (ipa_tm_insert_gettmclone_call): Update.
15413 * tree-call-cdce.c: Update.
15414 * tree-cfg.c (gimple_duplicate_sese_region): Update.
15415 (gimple_duplicate_sese_tail): Update.
15416 (gimple_account_profile_record): Update.
15417 (execute_fixup_cfg): Update.
15418 * tree-inline.c (copy_bb): Update.
15419 (copy_edges_for_bb): Update.
15420 (initialize_cfun): Update.
15421 (freqs_to_counts): Update.
15422 (copy_cfg_body): Update.
15423 (expand_call_inline): Update.
15424 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
15425 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
15426 (try_unroll_loop_completely): Update.
15427 (try_peel_loop): Update.
15428 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
15429 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
15430 * tree-ssa-loop-split.c (connect_loops): Update.
15431 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
15432 * tree-ssa-reassoc.c (branch_fixup): Update.
15433 * tree-ssa-tail-merge.c (replace_block_by): Update.
15434 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
15435 (compute_path_counts): Update.
15436 (update_profile): Update.
15437 (recompute_probabilities): Update.
15438 (update_joiner_offpath_counts): Update.
15439 (estimated_freqs_path): Update.
15440 (freqs_to_counts_path): Update.
15441 (clear_counts_path): Update.
15442 (ssa_fix_duplicate_block_edges): Update.
15443 (duplicate_thread_path): Update.
15444 * tree-switch-conversion.c (case_bit_test_cmp): Update.
15445 (struct switch_conv_info): Update.
15446 * tree-tailcall.c (decrease_profile): Update.
15447 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
15448 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
15449 * value-prof.c (check_counter): Update.
15450 (gimple_divmod_fixed_value): Update.
15451 (gimple_mod_pow2): Update.
15452 (gimple_mod_subtract): Update.
15453 (gimple_ic_transform): Update.
15454 (gimple_stringop_fixed_value): Update.
15455 * value-prof.h (gimple_ic): Update.
15457 2017-06-02 Carl Love <cel@us.ibm.com>
15459 * config/rs6000/rs6000-c: Add support for built-in functions
15460 vector double vec_doublee (vector signed int);
15461 vector double vec_doublee (vector unsigned int);
15462 vector double vec_doublee (vector float);
15463 vector double vec_doubleh (vector signed int);
15464 vector double vec_doubleh (vector unsigned int);
15465 vector double vec_doubleh (vector float);
15466 vector double vec_doublel (vector signed int);
15467 vector double vec_doublel (vector unsigned int);
15468 vector double vec_doublel (vector float);
15469 vector double vec_doubleo (vector signed int);
15470 vector double vec_doubleo (vector unsigned int);
15471 vector double vec_doubleo (vector float);.
15472 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
15473 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
15475 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
15476 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
15477 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
15479 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
15480 vec_doublel, vec_doubleh.
15481 * doc/extend.texi: Update the built-in documentation file for the
15482 new built-in functions.
15484 2017-06-02 David Malcolm <dmalcolm@redhat.com>
15487 * ipa-inline-analysis.c (free_growth_caches): Set
15488 edge_removal_hook_holder to NULL after removing it.
15490 2017-06-02 Sudakshina Das <sudi.das@arm.com>
15492 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
15493 comparision with zero.
15495 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
15496 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
15497 for early expansion of vec_min and vec_max builtins.
15498 (builtin_function_type): Add min/max unsigned variants to those
15499 identified as having unsigned arguments.
15501 2017-06-02 Olivier Hainque <hainque@adacore.com>
15503 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
15505 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15507 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
15508 Use VALL_F16 iterator rather than VALL.
15510 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15512 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
15513 Emit CBNZ inside loop when doing a strong exchange and comparing
15514 against zero. Generate the CC flags after the loop.
15516 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
15518 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
15519 (dl_section_ref): New.
15520 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
15521 On AIX, append an expression to subtract the size of the
15522 section length to dl_section_ref.
15524 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
15526 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
15527 for early expansion of vector absolute builtins.
15529 2017-06-02 Richard Biener <rguenther@suse.de>
15531 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
15532 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
15534 2017-06-02 Richard Biener <rguenther@suse.de>
15536 PR tree-optimization/80948
15537 * tree-tailcall.c (find_tail_calls): Track stmts to move in
15538 stmt order as well.
15540 2017-06-02 Richard Biener <rguenther@suse.de>
15542 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
15544 * tree-vect-stmts.c (process_use): Do not mark backedge defs
15545 for inductions as relevant.
15547 2017-06-02 Richard Biener <rguenther@suse.de>
15549 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
15550 (vectorizable_induction): ... this. Remove dead code.
15552 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
15554 * builtins. (expand_builtin_alloca): Remove second parameter and
15555 infer its value from the first parameter instead.
15556 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
15558 2017-06-02 Jakub Jelinek <jakub@redhat.com>
15560 PR rtl-optimization/80903
15561 * loop-doloop.c (add_test): Unshare sequence.
15563 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
15565 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
15567 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
15569 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
15571 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
15572 xlogue_layout::get_instance, logue_layout::xlogue_layout,
15573 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
15574 (xlogue_layout::get_stub_rtx): Make static.
15575 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
15576 (xlogue_layout::compute_stub_managed_regs): Rename to...
15577 (xlogue_layout::count_stub_managed_regs): ...this.
15578 (xlogue_layout::is_stub_managed_reg): New function.
15579 (xlogue_layout::m_stub_names): Rename to...
15580 (xlogue_layout::s_stub_names): ...this, make static.
15581 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
15582 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
15583 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
15584 xlogue_layout::s_stub_names): Instantiate statics.
15585 (stub_managed_regs): Remove.
15586 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
15587 (disable_call_ms2sysv_xlogues): Rename to...
15588 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
15589 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
15591 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
15592 change after reload_completed.
15593 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
15595 (ix86_expand_prologue): Likewise.
15596 (ix86_expand_epilogue): Likewise.
15597 (ix86_expand_split_stack_prologue): Likewise.
15598 (ix86_compute_frame_layout): Remove frame parameter ...
15599 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
15600 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
15602 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
15603 (ix86_frame): Move from here ...
15604 * config/i386/i386.h (ix86_frame): ... to here.
15605 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
15606 complete ix86_frame data structure instead. Remove some_ld_name.
15608 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
15610 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
15611 symbols that hold a DECL_VALUE_EXPR.
15613 2017-06-01 Martin Jambor <mjambor@suse.cz>
15615 PR tree-optimization/80898
15616 * tree-sra.c (process_subtree_disqualification): Removed.
15617 (disqualify_candidate): Do not acll
15618 process_subtree_disqualification.
15619 (subtree_mark_written_and_enqueue): New function.
15620 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
15621 RHS has been disqualified and re-queue LHS if necessary. Apart
15622 from that, ignore disqualified RHS.
15624 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15626 * config/s390/s390.c (s390_emit_epilogue): Disable early return
15627 address fetch for z10 or later.
15629 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15631 * config/arc/arc.md (tst_movb): Add guard when splitting.
15633 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15635 * config/arc/arc.c (arc_can_eliminate): Test against
15636 arc_frame_pointer_needed.
15638 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15640 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
15641 to prevent store reordering.
15642 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
15643 (type): Add block type.
15644 (stack_tie): Define special instruction to be used in
15647 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15649 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
15650 constraint. It is not valid for the pattern.
15651 (noncommutative_binary_comparison): Likewise.
15653 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15655 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
15658 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15660 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
15661 be used by the reg-alloc.
15663 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15665 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
15666 reg-alloc when having mul64 or mul32x16 instructions.
15667 (mulsidi3): Likewise.
15668 (umulsidi3): Likewise.
15669 (mulsi32x16): New pattern.
15670 (mulsi64): Likewise.
15671 (mulsidi64): Likewise.
15672 (umulsidi64): Likewise.
15673 (MUL32x16_REG): Define.
15674 (mul64_600): Use MUL32x16_REG.
15675 (mac64_600): Likewise.
15676 (umul64_600): Likewise.
15677 (umac64_600): Likewise.
15679 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
15681 * config/arc/arc.md (mulsi3_700): Make it commutative.
15683 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
15685 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
15687 (*sign_extendsidi2_insn): Likewise for movstosw.
15689 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
15691 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
15692 the type of the input discriminant value. Convert the
15693 discriminant value of signedness vary.
15695 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
15697 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
15698 Add to -Wall section.
15700 2017-06-01 Richard Biener <rguenther@suse.de>
15702 PR middle-end/66313
15703 * fold-const.c (fold_plusminus_mult_expr): If the factored
15704 factor may be zero use a wrapping type for the inner operation.
15705 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
15706 and handle moved defs.
15707 (process_assignment): Properly guard the unary op case. Return a
15708 tri-state indicating that moving the stmt before the call may allow
15709 to continue. Pass through to_move.
15710 (find_tail_calls): Handle moving unrelated defs before
15713 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
15716 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
15717 splitter result in the canonical way.
15719 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
15721 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
15722 also for 32bit target. Update insn attributes.
15723 (zero-extendsidi2 splitter): Allow all registers for operand 1.
15725 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
15727 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
15728 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
15729 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
15730 (_mm_maskz_min_ss): New intrinsics.
15732 2017-05-31 Martin Liska <mliska@suse.cz>
15734 * tree-vect-loop.c (vect_create_epilog_for_reduction):
15735 Change comment style to one we normally use.
15736 (vectorizable_reduction): Likewise.
15737 (vectorizable_induction): Likewise.
15738 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
15739 (vectorizable_call): Likewise.
15740 (vectorizable_simd_clone_call): Likewise.
15741 (vectorizable_conversion): Likewise.
15742 (vectorizable_assignment): Likewise.
15743 (vectorizable_shift): Likewise.
15744 (vectorizable_operation): Likewise.
15745 (vectorizable_store): Likewise.
15746 (vectorizable_load): Likewise.
15747 * tree-vectorizer.h: Likewise.
15749 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
15751 * passes.c (emergency_dump_function): New.
15752 * tree-pass.h (emergency_dump_function): Declare.
15753 * plugin.c (plugins_internal_error_function): Remove.
15754 * plugin.h (plugins_internal_error_function): Remove declaration.
15755 * toplev.c (internal_error_function): New static function. Use it...
15756 (general_init): ...here.
15758 2017-05-31 Graham Markall <graham.markall@embecosm.com>
15760 * config/arc/arc.c (arc_print_operand): Handle constant operands.
15761 (arc_rtx_costs): Add costs for new patterns.
15762 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
15763 * config/arc/predicates.md: Add _1_2_3_operand predicate.
15765 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
15767 * tree-ssa-strlen.c (get_next_strinfo): New function.
15768 (get_stridx_plus_constant): Use it.
15769 (zero_length_string): Likewise.
15770 (adjust_related_strinfos): Likewise.
15771 (adjust_last_stmt): Likewise.
15773 2017-05-31 Richard Biener <rguenther@suse.de>
15776 * config/i386/i386.c (ix86_expand_builtin): Remove assert
15777 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
15779 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
15781 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
15782 loop_vinfo argument and use of dependence distance vectors.
15783 Check instead whether the two references differ only in their
15784 initial value and assume that they have the same alignment if the
15785 difference is a multiple of the vector alignment.
15786 (vect_analyze_data_refs_alignment): Update call accordingly.
15788 2017-05-31 Martin Liska <mliska@suse.cz>
15791 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
15793 2017-05-31 Bin Cheng <bin.cheng@arm.com>
15795 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
15796 (create_intersect_range_checks): Move from ...
15797 * tree-data-ref.c (create_intersect_range_checks_index)
15798 (create_intersect_range_checks): ... to here.
15799 (create_runtime_alias_checks): New function factored from ...
15800 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
15801 here. Call above function.
15802 * tree-data-ref.h (create_runtime_alias_checks): New function.
15804 2017-05-31 Bin Cheng <bin.cheng@arm.com>
15806 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
15807 segment length for dr_b and compute it in wide_int.
15809 2017-05-31 Richard Biener <rguenther@suse.de>
15811 PR tree-optimization/80906
15812 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
15813 and pass through iv_map.
15814 (copy_bb_and_scalar_dependences): Adjust.
15815 (translate_pending_phi_nodes): Likewise.
15816 (copy_loop_close_phi_args): Handle code-generating IVs instead
15819 2017-05-30 David Malcolm <dmalcolm@redhat.com>
15821 * diagnostic-color.c (color_dict): Add "type-diff".
15822 (parse_gcc_colors): Update comment.
15823 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
15824 -fdiagnostics-show-template-tree and -fno-elide-type.
15825 (GCC_COLORS): Add type-diff to example.
15827 (-fdiagnostics-show-template-tree): New.
15828 (-fno-elide-type): New.
15829 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
15830 the pp_format_decoder callback. Call any m_format_postprocessor's
15832 (pretty_printer::pretty_printer): Initialize
15833 m_format_postprocessor.
15834 (pretty_printer::~pretty_printer): Delete any
15835 m_format_postprocessor.
15836 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
15837 (class format_postprocessor): New class.
15838 (struct pretty_printer::format_decoder): Document the new parameters.
15839 (struct pretty_printer::m_format_postprocessor): New field.
15840 * tree-diagnostic.c (default_tree_printer): Update for new
15841 bool and const char ** params.
15842 * tree-diagnostic.h (default_tree_printer): Likewise.
15844 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
15846 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
15847 (lwa_operand): Delete rs6000_gen_cell_microcode test.
15848 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
15849 rs6000_gen_cell_microcode code.
15850 (rs6000_final_prescan_insn): Delete.
15851 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
15852 "warn-cell-microcode" entries.
15853 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
15854 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
15855 throughout. Change cc_reg_not_micro_cr0_operand to
15856 cc_reg_not_cr0_operand throughout.
15857 (*extendhi<mode>2_noload): Delete.
15858 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
15859 (mwarn-cell-microcode): Delete.
15860 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
15861 -mgen-cell-microcode and -mwarn-cell-microcode.
15863 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
15866 * config/i386/constraints.md (Yd): New constraint.
15868 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
15869 and (?Yd, r) alternatives. Update insn attributes.
15870 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
15871 and (?*Yd, r) alternatives. Update insn attributes.
15872 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
15874 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
15876 * gimplify.c (gimplify_modify_expr): Don't create a
15877 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
15880 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
15882 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
15884 2017-05-30 Richard Biener <rguenther@suse.de>
15886 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
15887 and reduc_def fields.
15888 (STMT_VINFO_REDUC_TYPE): New define.
15889 (STMT_VINFO_REDUC_DEF): Likewise.
15890 (vect_force_simple_reduction): Adjust prototype.
15891 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
15892 (vect_is_simple_reduction): Remove check_reduction argument.
15893 (vect_force_simple_reduction): Adjust and set
15894 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
15895 (vectorizable_reduction): Do not re-do reduction analysis
15896 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
15897 * tree-parloops.c (gather_scalar_reductions): Adjust.
15899 2017-05-30 Richard Biener <rguenther@suse.de>
15901 PR middle-end/80901
15902 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
15905 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
15907 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
15908 Introduce unknown_misalignment parameter and remove vf.
15909 (vect_peeling_hash_get_lowest_cost):
15910 Pass unknown_misalignment parameter.
15911 (vect_enhance_data_refs_alignment):
15912 Fix unsupportable data ref treatment.
15914 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
15916 * tree-vect-data-refs.c (vect_get_data_access_cost):
15917 Workaround for SLP handling.
15918 (vect_enhance_data_refs_alignment):
15919 Compute costs for doing no peeling at all, compare to the best
15920 peeling costs so far and avoid peeling if cheaper.
15922 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
15924 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
15925 Return peeling info and set costs to zero for unlimited cost
15927 (vect_enhance_data_refs_alignment): Also inspect all datarefs
15928 with unknown misalignment. Compute and costs for unknown
15929 misalignment, compare them to the costs for known misalignment
15930 and choose the cheapest for peeling.
15932 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
15934 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
15935 (vect_get_peeling_costs_all_drs): Create function.
15936 (vect_peeling_hash_get_lowest_cost):
15937 Use vect_get_peeling_costs_all_drs.
15938 (vect_peeling_supportable): Create function.
15939 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
15941 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
15943 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
15944 DR_HAS_NEGATIVE_STEP.
15945 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
15946 (vect_enhance_data_refs_alignment): Use.
15947 (vect_duplicate_ssa_name_ptr_info): Use.
15948 * tree-vectorizer.h (dr_misalignment): Use.
15949 (known_alignment_for_access_p): Use.
15951 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
15954 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
15956 (has_section_name): New function.
15958 2017-05-30 Martin Liska <mliska@suse.cz>
15961 * auto-profile.c (get_function_decl_from_block): Fix
15964 2017-05-30 Richard Biener <rguenther@suse.de>
15966 PR middle-end/80876
15967 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
15969 2017-05-30 Martin Liska <mliska@suse.cz>
15971 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
15972 * dumpfile.h (struct dump_file_info): Remove ctors.
15974 2017-05-30 Martin Liska <mliska@suse.cz>
15976 * predict.def: Fix GNU coding style.
15978 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
15980 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
15981 Mark 'to' argument with ATTRIBUTE_UNUSED.
15983 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
15985 * config/xtensa/xtensa.c (xtensa_emit_call): Use
15986 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
15987 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
15990 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
15992 * doc/install.texi (Options specification): Restore entry of
15993 --enable-sjlj-exceptions.
15995 2017-05-27 Michael Eager <eager@eagercon.com>
15998 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
16000 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
16002 * config/microblaze/microblaze.h
16003 (FIXED_REGISTERS): Update in macro.
16004 (CALL_USED_REGISTERS): Update in macro.
16006 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
16008 * doc/install.texi: Add links to macOS binary distributions.
16010 2017-05-27 Jakub Jelinek <jakub@redhat.com>
16014 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
16016 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
16018 2017-05-26 Martin Liska <mliska@suse.cz>
16020 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
16022 2017-05-26 Martin Liska <mliska@suse.cz>
16024 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
16025 always leading ';; '.
16026 (dump_bb_info): Likewise.
16027 (brief_dump_cfg): Likewise.
16028 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
16029 * dumpfile.c: Remove usage of TDF_VERBOSE.
16030 * dumpfile.h (enum dump_kind): Likewise.
16031 (dump_gimple_bb_header): Do not use TDF_COMMENT.
16032 * print-tree.c (debug_verbose): Remove.
16033 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
16034 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
16035 * tree-diagnostic.c (default_tree_printer): Replace
16036 TDF_DIAGNOSTIC with TDF_SLIM.
16038 2017-05-26 Bin Cheng <bin.cheng@arm.com>
16040 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
16041 in parameter loop, rather than loop_vinfo.
16042 (create_intersect_range_checks): Ditto.
16043 (vect_create_cond_for_alias_checks): Update call to above functions.
16045 2017-05-26 Bin Cheng <bin.cheng@arm.com>
16047 PR tree-optimization/80815
16048 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
16049 for merging runtime alias checks. Handle negative DR_STEPs.
16051 2017-05-26 Bin Cheng <bin.cheng@arm.com>
16053 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
16055 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
16056 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
16057 out code pruning runtime alias checks.
16058 * tree-data-ref.c (prune_runtime_alias_test_list): New function
16059 factored out from above.
16060 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
16062 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
16064 (prune_runtime_alias_test_list): New decalaration.
16066 2017-05-26 Bin Cheng <bin.cheng@arm.com>
16068 * tree-vect-data-refs.c (compare_tree): Rename and move ...
16069 * tree-data-ref.c (data_ref_compare_tree): ... to here.
16070 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
16071 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
16072 (operator==, comp_dr_with_seg_len_pair): Ditto.
16073 (vect_prune_runtime_alias_test_list): Ditto.
16075 2017-05-26 Martin Liska <mliska@suse.cz>
16078 * params.def: Bound partial-inlining-entry-probability param.
16080 2017-05-26 Marek Polacek <polacek@redhat.com>
16083 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
16086 2017-05-26 Richard Biener <rguenther@suse.de>
16088 PR tree-optimization/80842
16089 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
16092 2017-05-26 Richard Biener <rguenther@suse.de>
16094 PR tree-optimization/80844
16095 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
16097 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
16099 * doc/md.texi (Machine Constraints): Update x86 family
16100 machine constraints section to match 'config/i386/constraints.md'.
16102 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
16104 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
16106 2017-05-25 Nathan Sidwell <nathan@acm.org>
16108 * doc/invoke.texi (--enable-languages): Update documentation.
16110 2017-05-25 Martin Liska <mliska@suse.cz>
16112 * dumpfile.c: Add TDF_FOLDING.
16113 * dumpfile.h (enum dump_kind): Likewise.
16114 * genmatch.c (dt_simplify::gen_1): Use it.
16116 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
16118 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
16120 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
16122 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
16123 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
16125 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
16127 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
16128 * match.pd (X == C): Rewrite it here.
16129 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
16130 with_certain_nonzero_bits2): New predicates.
16131 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
16133 2017-05-24 Nathan Sidwell <nathan@acm.org>
16135 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
16138 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
16141 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
16143 * config/powerpcspe: New port. Files are copied from the rs6000
16144 port, with "rs6000" in filenames replaced by "powerpcspe".
16146 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
16148 PR rtl-optimization/80754
16149 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
16151 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
16153 * config/sparc/sparc.md (length): Return the correct value for -mflat
16154 sibcalls to match output_sibcall.
16156 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
16160 * config/rs6000/rs6000.c (struct machine_function): Add new field
16162 (rs6000_get_separate_components): Init that field, use it.
16163 (rs6000_components_for_bb): Use the field.
16165 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
16167 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
16169 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
16171 PR middle-end/80823
16172 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
16174 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16177 * config/s390/s390.c (s390_check_qrst_address): Check incoming
16178 address against address_operand predicate.
16179 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
16181 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
16183 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
16184 parameters passed indirectly.
16186 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
16188 * config/i386/i386.md (*movdi_internal): Remove SSE4
16189 alternative 18 (?r, *v). Update insn attributes.
16190 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
16191 Update insn attributes.
16192 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
16193 Update insn attributes.
16194 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
16195 alternative 1 (r, v). Remove isa attribute.
16196 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
16197 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
16198 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
16200 2017-05-23 Tom de Vries <tom@codesourcery.com>
16202 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
16205 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
16207 * cgraphunit.c (symbol_table::process_new_functions): Update.
16208 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
16209 (inline_generate_summary): Rename to ...
16210 (ipa_fn_summary_generate): ... this one.
16211 (inline_read_summary): Rename to ...
16212 (ipa_fn_summary_read): ... this one.
16213 (inline_write_summary): Rename to ...
16214 (ipa_fn_summary_write): ... this one.
16215 (inline_free_summary): Rename to ...
16216 (ipa_free_fn_summary): ... this one.
16217 (pass_data_local_fn_summary, pass_local_fn_summary,
16218 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
16219 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
16220 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
16221 make_pass_ipa_fn_summary): New.
16222 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
16223 inline_write_summary, inline_free_summary): Remove.
16224 (ipa_free_fn_summary) : New.
16225 * ipa-inline.c (ipa_inline): Update.
16226 (pass_ipa_inline): Do not generate summaries.
16227 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
16229 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
16230 and add pass_ipa_fn_summary.
16231 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
16233 (make_pass_inline_parameters): Remove.
16235 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
16237 * omp-low.c (struct omp_context): Remove "default_kind" member.
16240 * omp-offload.c (execute_oacc_device_lower): Remove the
16241 parallelism dimensions function attributes for unparallelized
16242 OpenACC kernels constructs.
16244 2017-05-23 Martin Liska <mliska@suse.cz>
16246 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
16248 (cgraph_edge::make_speculative): Likewise.
16249 (cgraph_edge::resolve_speculation): Likewise.
16250 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
16251 (cgraph_node::dump): Likewise.
16252 * cgraph.h: Likewise.
16253 * cgraphunit.c (analyze_functions): Likewise.
16254 (symbol_table::compile): Likewise.
16255 * ipa-cp.c (print_all_lattices): Likewise.
16256 (determine_versionability): Likewise.
16257 (initialize_node_lattices): Likewise.
16258 (ipcp_verify_propagated_values): Likewise.
16259 (estimate_local_effects): Likewise.
16260 (update_profiling_info): Likewise.
16261 (create_specialized_node): Likewise.
16262 (perhaps_add_new_callers): Likewise.
16263 (decide_about_value): Likewise.
16264 (decide_whether_version_node): Likewise.
16265 (identify_dead_nodes): Likewise.
16266 (ipcp_store_bits_results): Likewise.
16267 * ipa-devirt.c (dump_targets): Likewise.
16268 (ipa_devirt): Likewise.
16269 * ipa-icf.c (sem_item::dump): Likewise.
16270 (sem_function::equals): Likewise.
16271 (sem_variable::equals): Likewise.
16272 (sem_item_optimizer::read_section): Likewise.
16273 (sem_item_optimizer::execute): Likewise.
16274 (congruence_class::dump): Likewise.
16275 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
16276 (dump_inline_summary): Likewise.
16277 (estimate_node_size_and_time): Likewise.
16278 (inline_analyze_function): Likewise.
16279 * ipa-inline-transform.c (inline_call): Likewise.
16280 * ipa-inline.c (report_inline_failed_reason): Likewise.
16281 (want_early_inline_function_p): Likewise.
16282 (edge_badness): Likewise.
16283 (update_edge_key): Likewise.
16284 (inline_small_functions): Likewise.
16285 * ipa-profile.c (ipa_profile): Likewise.
16286 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
16287 (ipa_make_edge_direct_to_target): Likewise.
16288 (remove_described_reference): Likewise.
16289 (ipa_impossible_devirt_target): Likewise.
16290 (propagate_controlled_uses): Likewise.
16291 (ipa_print_node_params): Likewise.
16292 (ipcp_transform_function): Likewise.
16293 * ipa-pure-const.c (pure_const_read_summary): Likewise.
16294 (propagate_pure_const): Likewise.
16295 * ipa-reference.c (generate_summary): Likewise.
16296 (read_write_all_from_decl): Likewise.
16297 (propagate): Likewise.
16298 (ipa_reference_read_optimization_summary): Likewise.
16299 * ipa-utils.c (ipa_merge_profiles): Likewise.
16300 * ipa.c (walk_polymorphic_call_targets): Likewise.
16301 (symbol_table::remove_unreachable_nodes): Likewise.
16302 (ipa_single_use): Likewise.
16303 * passes.c (execute_todo): Likewise.
16304 * predict.c (drop_profile): Likewise.
16305 * symtab.c (symtab_node::get_dump_name): New function.
16306 (symtab_node::dump_name): Likewise.
16307 (symtab_node::dump_asm_name): Likewise.
16308 (symtab_node::dump_references): Likewise.
16309 (symtab_node::dump_referring): Likewise.
16310 (symtab_node::dump_base): Likewise.
16311 (symtab_node::debug_symtab): Likewise.
16312 * tree-sra.c (convert_callers_for_node): Likewise.
16313 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
16314 * value-prof.c (init_node_map): Likewise.
16316 2017-05-23 Martin Liska <mliska@suse.cz>
16318 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
16319 and symtab_node::debug_symtab to symbol_table::debug.
16320 * cgraphunit.c (analyze_functions): Use the renamed function.
16321 (symbol_table::compile): Likewise.
16322 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
16323 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
16324 * passes.c (execute_todo): Likewise.
16325 * symtab.c (symbol_table::dump): New function.
16326 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
16328 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
16330 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
16331 that nonconst implies exec.
16333 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
16335 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
16336 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
16337 (inline_edge_summary_vec): Turn into ...
16338 (ipa_call_summaries): ... this one.
16339 (redirect_to_unreachable, edge_set_predicate,
16340 evaluate_properties_for_edge, inline_summary_alloc,
16341 reset_ipa_call_summary, reset_inline_summary,
16342 inline_summary_t::duplicate): Update.
16343 (inline_edge_duplication_hook): Turn to ...
16344 (ipa_call_summary_t::duplicate): ... this one.
16345 (inline_edge_removal_hook): Turn to ...
16346 (ipa_call_summary_t::remove): ... this one.
16347 (dump_inline_edge_summary): Turn to ...
16348 (dump_ipa_call_summary): ... this one.
16349 (estimate_function_body_sizes): Update.
16350 (inline_update_callee_summaries): Update.
16351 (remap_edge_change_prob): Update.
16352 (remap_edge_summaries): Update.
16353 (inline_merge_summary): Update.
16354 (do_estimate_edge_time): Update.
16355 (inline_generate_summary): Update.
16356 (inline_read_section): Update.
16357 (inline_read_summary): Update.
16358 (inline_free_summary): Update.
16359 * ipa-inline.c (can_inline_edge_p): Update.
16360 (compute_inlined_call_time): Update.
16361 (want_inline_small_function_p): Update.
16362 (edge_badness): Update.
16363 (early_inliner): Update.
16364 * ipa-inline.h (inline_edge_summary): Turn to ...
16365 (ipa_call_summary): ... this one.
16366 (ipa_call_summary_t): New class.
16367 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
16368 (ipa_call_summaries): New.
16369 (inline_edge_summary): Remove.
16370 (estimate_edge_growth): Update.
16371 * ipa-profile.c (ipa_propagate_frequency_1): Update.
16372 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
16373 * ipa-split.c (execute_split_functions): Update.
16374 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
16376 2017-05-23 Tom de Vries <tom@codesourcery.com>
16378 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
16379 attributes): Document rdrand effective target.
16381 2017-05-23 Tom de Vries <tom@codesourcery.com>
16383 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
16384 attributes): Sort alphabetically.
16386 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
16388 * config/avr/genmultilib.awk: Use gsub instead of gensub.
16390 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
16393 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
16394 V2DF/V2DI splat into two separate patterns, one that handles
16395 registers, and the other that only handles memory. Drop support
16396 for splatting from a GPR on ISA 2.07 and then splitting the
16397 splat into direct move and splat.
16398 (vsx_splat_<mode>_reg): Likewise.
16399 (vsx_splat_<mode>_mem): Likewise.
16401 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
16403 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
16405 2017-05-22 Jakub Jelinek <jakub@redhat.com>
16407 PR middle-end/80809
16408 * omp-low.c (finish_taskreg_remap): New function.
16409 (finish_taskreg_scan): If unit size of ctx->record_type
16410 is non-constant, unshare the size expression and replace
16411 decls in it with possible outer var refs.
16413 PR middle-end/80809
16414 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
16415 GOVD_SHARED rather than GOVD_PRIVATE with it.
16416 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
16417 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
16419 PR middle-end/80853
16420 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
16421 as last argument to build_outer_var_ref for pointer bases of array
16422 section reductions.
16424 2017-05-19 Martin Sebor <msebor@redhat.com>
16426 * print-tree.c (print_node): Print DECL_READ_P flag.
16428 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
16430 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
16431 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
16432 * cgraph.c: Likewise.
16433 * cgraphunit.c: Likewise.
16434 * gengtype.c: Likewise.
16435 * ipa-cp.c: Likewise.
16436 * ipa-devirt.c: Likewise.
16437 * ipa-icf.c: Likewise.
16438 * ipa-predicate.c: Likewise.
16439 * ipa-profile.c: Likewise.
16440 * ipa-prop.c: Likewise.
16441 * ipa-split.c: Likewise.
16443 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
16444 edge_predicate_pool, dump_inline_hints,
16445 inline_summary::account_size_time, redirect_to_unreachable,
16446 edge_set_predicate, set_hint_predicate,
16447 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
16448 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
16449 inline_summary_t::remove, remap_hint_predicate_after_duplication,
16450 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
16451 ipa_call_summary_t::remove, initialize_growth_caches,
16452 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
16453 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
16454 mark_modified, unmodified_parm_1, unmodified_parm,
16455 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
16456 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
16457 compute_bb_predicates, will_be_nonconstant_expr_predicate,
16458 will_be_nonconstant_predicate, record_modified_bb_info,
16459 get_minimal_bb, record_modified, param_change_prob,
16460 phi_result_unknown_predicate, predicate_for_phi_result,
16461 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
16462 estimate_function_body_sizes, compute_inline_parameters,
16463 compute_inline_parameters_for_curren, pass_data_inline_parameters,
16464 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
16465 inline_update_callee_summaries, remap_edge_change_prob,
16466 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
16467 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
16468 inline_analyze_function, inline_summary_t::insert,
16469 inline_generate_summary, read_ipa_call_summary, inline_read_section,
16470 inline_read_summary, write_ipa_call_summary, inline_write_summary,
16471 inline_free_summary): Move to ipa-fnsummary.h
16472 (predicate_t): Remove.
16473 * ipa-fnsummary.c: New file.
16474 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
16475 (enum inline_hints_vals, inline_hints, agg_position_info,
16476 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
16477 inline_summaries, ipa_call_summary, ipa_call_summary_t,
16478 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
16479 dump_inline_summary, dump_inline_hints, inline_generate_summary,
16480 inline_read_summary, inline_write_summary, inline_free_summary,
16481 inline_analyze_function, initialize_inline_failed,
16482 inline_merge_summary, inline_update_overall_summary,
16483 compute_inline_parameters): Move to ipa-fnsummary.h
16484 * ipa-fnsummary.h: New file.
16485 * ipa-inline-transform.h: Include ipa-inline.h.
16486 * ipa-inline.c: LIkewise.
16488 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
16490 * ipa-inline.c (edge_badness): Use inlined_time instead of
16491 inline_summaries->get.
16493 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
16495 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
16497 2017-05-22 Nathan Sidwell <nathan@acm.org>
16499 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
16500 (fdump-lang): Document 'raw' option.
16501 * dumpfile.h (TDI_tu): Delete.
16502 * dumpfile.c (dump_files): Remove translation-unit.
16503 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
16505 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
16507 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
16508 command option from $(AWK) call.
16509 * config/avr/genmultilib.awk: Simplify and rewrite so that it
16510 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
16511 [FORMAT]: Remove handling of variable.
16512 * config/avr/t-multilib: Regenerate.
16514 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
16516 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
16518 (dump_inline_summary): Do not print self_time.
16519 (estimate_function_body_sizes): Do not set self_time.
16520 (compute_inline_parameters): Likewise.
16521 (inline_read_section, inline_write_summary): Do not stream self_time.
16522 * ipa-inline.h (inline_summary): Drop self_time.
16524 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
16526 * ipa-inline-analysis.c (account_size_time): Rename to ...
16527 (inline_summary::account_size_time): ... this one.
16528 (reset_ipa_call_summary): Turn to ...
16529 (ipa_call_summary::reset): ... this one.
16530 (reset_inline_summary): Turn to ...
16531 (inline_summary::reset): ... this one.
16532 (inline_summary_t::remove): Update.
16533 (inline_summary_t::duplicate): Update.
16534 (ipa_call_summary_t::remove): Update.
16535 (dump_inline_summary): Update.
16536 (estimate_function_body_sizes): Update.
16537 (compute_inline_parameters): Update.
16538 (estimate_node_size_and_time): Update.
16539 (inline_merge_summary): Update.
16540 (inline_update_overall_summary): Update.
16541 (inline_read_section): Update.
16542 (inline_write_summary): Update.
16543 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
16544 add account_size_time and reset member functions.
16545 (ipa_call_summary): Add reset function.
16546 * ipa-predicate.h (predicate::operator &): Constify.
16548 2017-05-22 Richard Biener <rguenther@suse.de>
16550 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
16552 2017-05-19 Jason Merrill <jason@redhat.com>
16554 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
16556 2017-05-19 Marek Polacek <polacek@redhat.com>
16559 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
16560 TYPE_OVERFLOW_WRAPS checks.
16562 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
16564 * tree-core.h (enum omp_clause_default_kind): Add
16565 "OMP_CLAUSE_DEFAULT_PRESENT".
16566 * tree-pretty-print.c (dump_omp_clause): Handle it.
16567 * gimplify.c (enum gimplify_omp_var_data): Add
16568 "GOVD_MAP_FORCE_PRESENT".
16569 (gimplify_adjust_omp_clauses_1): Map it to
16570 "GOMP_MAP_FORCE_PRESENT".
16571 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
16573 * gimplify.c (oacc_default_clause): Clarify.
16575 2017-05-19 Nathan Sidwell <nathan@acm.org>
16577 LANG_HOOK_REGISTER_DUMPS
16578 * toplev.c (general_init): Call register dump lang hook.
16579 * doc/invoke.texi: Document -fdump-lang option family.
16580 * dumpfile.c (dump_files): Remove class dump here.
16581 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
16582 * dumpfile.h (tree_dump_index): Remove TDI_class.
16583 * langhooks-def.h (lhd_register_dumps): Declare.
16584 (LANG_HOOKS_REGISTER_DUMPS): Define.
16585 (LANG_HOOKS_INITIALIZER): Add it.
16586 * langhooks.c (lhd_register_dumps): Define.
16587 * langhooks.h (struct lang_hooks): Add register_dumps.
16589 2017-05-19 Nathan Sidwell <nathan@acm.org>
16591 * context.h (context::set_passes): New.
16592 * context.c (context::context): Do not create pass manager.
16593 * toplev.c (general_init): Create pass manager here.
16595 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
16597 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
16598 use this splitter if two add or or instructions would also work for
16599 the constant we want to generate.
16601 2017-05-19 Richard Biener <rguenther@suse.de>
16604 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
16605 predicate evaluation.
16607 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
16609 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
16611 * ipa-inline.c (want_inline_small_function_p): Do not cast to
16614 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
16616 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
16617 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
16618 (inline_edge_summary_vec): Turn into ...
16619 (ipa_call_summaries): ... this one.
16620 (redirect_to_unreachable, edge_set_predicate,
16621 evaluate_properties_for_edge, inline_summary_alloc,
16622 reset_ipa_call_summary, reset_inline_summary,
16623 inline_summary_t::duplicate): Update.
16624 (inline_edge_duplication_hook): Turn to ...
16625 (ipa_call_summary_t::duplicate): ... this one.
16626 (inline_edge_removal_hook): Turn to ...
16627 (ipa_call_summary_t::remove): ... this one.
16628 (dump_inline_edge_summary): Turn to ...
16629 (dump_ipa_call_summary): ... this one.
16630 (estimate_function_body_sizes): Update.
16631 (inline_update_callee_summaries): Update.
16632 (remap_edge_change_prob): Update.
16633 (remap_edge_summaries): Update.
16634 (inline_merge_summary): Update.
16635 (do_estimate_edge_time): Update.
16636 (inline_generate_summary): Update.
16637 (inline_read_section): Update.
16638 (inline_read_summary): Update.
16639 (inline_free_summary): Update.
16640 * ipa-inline.c (can_inline_edge_p): Update.
16641 (compute_inlined_call_time): Update.
16642 (want_inline_small_function_p): Update.
16643 (edge_badness): Update.
16644 (early_inliner): Update.
16645 * ipa-inline.h (inline_edge_summary): Turn to ...
16646 (ipa_call_summary): ... this one.
16647 (ipa_call_summary_t): New class.
16648 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
16649 (ipa_call_summaries): New.
16650 (inline_edge_summary): Remove.
16651 (estimate_edge_growth): Update.
16652 * ipa-profile.c (ipa_propagate_frequency_1): Update.
16653 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
16654 * ipa-split.c (execute_split_functions): Update.
16655 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
16657 2017-05-19 Richard Biener <rguenther@suse.de>
16659 PR middle-end/80764
16660 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
16662 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
16664 * config/rs6000/rs6000.c (struct machine_function): Add field
16665 fpr_is_wrapped_separately.
16666 (rs6000_get_separate_components): Use 64 components. Handle the
16667 new FPR components.
16668 (rs6000_components_for_bb): Handle the FPR components.
16669 (rs6000_emit_prologue_components): Handle the FPR components.
16670 (rs6000_emit_epilogue_components): Handle the FPR components.
16671 (rs6000_set_handled_components): Handle the FPR components.
16672 (rs6000_emit_prologue): Don't output prologue code for those FPRs
16673 that are already separately shrink-wrapped.
16674 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
16675 that are already separately shrink-wrapped.
16677 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
16680 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
16683 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
16684 (define_peephole2 for Altivec d-form load): Add peepholes to catch
16685 cases where the register allocator uses a move and an offsettable
16686 memory operation to/from a FPR register on ISA 2.06/2.07.
16687 (define_peephole2 for Altivec d-form store): Likewise.
16689 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
16692 * config/i386/mmx.md (*mov<mode>_internal): Enable
16693 alternatives 11, 12, 13 and 14 also for 32bit targets.
16694 Remove alternatives 15, 16, 17 and 18.
16695 * config/i386/sse.md (vec_concatv2di): Change
16696 alternative (!x, *y) to (x, ?!*Yn).
16698 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
16700 * dumpfile.h (enum dump_kind): Remove stray comma.
16702 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
16704 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
16705 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
16706 predicate::num_conditions
16707 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
16708 (CHANGED): turn into predicate::changed.
16709 (agg_position_info): Move to ipa-predicate.h
16710 (add_condition, predicate::add_clause, predicate::operator &=,
16711 predicate::or_with, predicate::evaluate, predicate::probability,
16712 dump_condition, dump_clause, predicate::dump,
16713 predicate::remap_after_duplication, predicate::remap_after_inlining,
16714 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
16715 (evaluate_conditions_for_known_args): Update.
16716 (set_cond_stmt_execution_predicate): Update.
16717 * ipa-inline.h: Include ipa-predicate.h
16718 (condition, inline_param_summary, conditions, agg_position_info,
16719 predicate): Move to ipa-predicate.h
16720 * ipa-predicate.c: New file.
16721 * ipa-predicate.h: New file.
16723 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
16725 * final.c (leaf_function_p): Check we are not in a sequence.
16727 2017-05-18 Martin Liska <mliska@suse.cz>
16729 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
16730 * dumpfile.c (dump_register): Use new enum dump_kind.
16731 (get_dump_file_name): Likewise.
16732 (dump_enable_all): Likewise.
16733 (dump_switch_p_1): Likewise.
16734 (enable_rtl_dump_file): Remove usage of TDF_RTL.
16735 * dumpfile.h (enum dump_kind): New enum type.
16736 (struct dump_file_info): Create constructor and
16737 format fields and comments.
16738 * passes.c (pass_manager::register_one_dump_file):
16740 * statistics.c (statistics_early_init): Likewise.
16741 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
16742 TDF_TREE with TDF_SLIM.
16743 (gather_memory_references_ref): Likewise.
16745 2017-05-18 Martin Liska <mliska@suse.cz>
16747 * vec.h (struct vnull): Use it.
16749 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
16751 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
16752 (true_predicate, false_predicate, true_predicate_p,
16753 false_predicate_p): Remove.
16754 (single_cond_predicate, not_inlined_predicate): Turn to member function
16756 (add_condition): Update.
16757 (add_clause): Turn to...
16758 (predicate::add_clause): ... this one; update; allow passing NULL
16760 (and_predicates): Turn to ...
16761 (predicate::operator &=): ... this one.
16762 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
16763 (or_predicates): Turn to ...
16764 (predicate::or_with): ... this one.
16765 (evaluate_predicate): Turn to ...
16766 (predicate::evaluate): ... this one.
16767 (predicate_probability): Turn to ...
16768 (predicate::probability): ... this one.
16769 (dump_condition): Update.
16770 (dump_predicate): Turn to ...
16771 (predicate::dump): ... this one.
16772 (account_size_time): Update.
16773 (edge_set_predicate): Update.
16774 (set_hint_predicate): UPdate.
16775 (evaluate_conditions_for_known_args): Update.
16776 (evaluate_properties_for_edge): Update.
16777 (remap_predicate_after_duplication): Turn to...
16778 (predicate::remap_after_duplication): ... this one.
16779 (remap_hint_predicate_after_duplication): Update.
16780 (inline_summary_t::duplicate): UPdate.
16781 (dump_inline_edge_summary): Update.
16782 (dump_inline_summary): Update.
16783 (set_cond_stmt_execution_predicate): Update.
16784 (set_switch_stmt_execution_predicate): Update.
16785 (compute_bb_predicates): Update.
16786 (will_be_nonconstant_expr_predicate): Update.
16787 (will_be_nonconstant_predicate): Update.
16788 (phi_result_unknown_predicate): Update.
16789 (predicate_for_phi_result): Update.
16790 (array_index_predicate): Update.
16791 (estimate_function_body_sizes): Update.
16792 (estimate_node_size_and_time): Update.
16793 (estimate_ipcp_clone_size_and_time): Update.
16794 (remap_predicate): Rename to ...
16795 (predicate::remap_after_inlining): ... this one.
16796 (remap_hint_predicate): Update.
16797 (inline_merge_summary): Update.
16798 (inline_update_overall_summary): Update.
16799 (estimate_size_after_inlining): Update.
16800 (read_predicate): Rename to ...
16801 (predicate::stream_in): ... this one.
16802 (read_inline_edge_summary): Update.
16803 (write_predicate): Rename to ...
16804 (predicate::stream_out): ... this one.
16805 (write_inline_edge_summary): Update.
16806 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
16807 (clause_t): Turn to uint32_t
16808 (predicate): Turn to class; implement constructor and operators
16810 (size_time_entry): Update.
16811 (inline_summary): Update.
16812 (inline_edge_summary): Update.
16814 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
16816 * fold-const.c (fold_binary_loc): Move transformation...
16817 * match.pd (C - X CMP X): ... here.
16819 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
16821 * config/sparc/sparc.c (sparc_option_override): Set function
16822 alignment for -mcpu=niagara7 to 64 to match the I$ line.
16823 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
16825 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
16827 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
16829 2017-05-18 Marek Polacek <polacek@redhat.com>
16832 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
16833 (pass_ubsan::execute): Call gimple_assign_single_p instead of
16834 gimple_assign_load_p.
16836 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
16838 PR middle-end/80692
16839 * real.c (do_compare): Give decimal_do_compare preference over
16840 comparing just the signs.
16842 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
16844 * doc/md.texi (Canonicalization of Instructions): Describe the
16845 canonical form of instructions that inherently set a condition
16848 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
16850 PR middle-end/80775
16851 * tree-cfg.c: Move deletion of unreachable case statements to after
16852 the merging of consecutive case labels.
16854 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
16856 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
16857 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
16858 restoring of callee-saved registers.
16860 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
16862 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
16863 * config/visium/visium.c (single_set_and_flags): Likewise.
16864 * config/visium/visium.md (Substitutions): Likewise.
16866 2017-05-17 Martin Liska <mliska@suse.cz>
16868 * cfg.c: Introduce dump_flags_t type and
16869 use it instead of int type.
16871 * cfghooks.c: Likewise.
16872 * cfghooks.h (struct cfg_hooks): Likewise.
16873 * cfgrtl.c: Likewise.
16874 * cfgrtl.h: Likewise.
16875 * cgraph.c (cgraph_node::get_body): Likewise.
16876 * coretypes.h: Likewise.
16877 * domwalk.c: Likewise.
16878 * domwalk.h: Likewise.
16879 * dumpfile.c (struct dump_option_value_info): Likewise.
16880 (dump_enable_all): Likewise.
16881 (dump_switch_p_1): Likewise.
16882 (opt_info_switch_p): Likewise.
16883 * dumpfile.h (enum tree_dump_index): Likewise.
16884 (struct dump_file_info): Likewise.
16885 * genemit.c: Likewise.
16886 * generic-match-head.c: Likewise.
16887 * gengtype.c (open_base_files): Likewise.
16888 * gimple-pretty-print.c: Likewise.
16889 * gimple-pretty-print.h: Likewise.
16890 * graph.c (print_graph_cfg): Likewise.
16891 * graphite-scop-detection.c (dot_all_sese): Likewise.
16892 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
16893 * loop-unroll.c (report_unroll): Likewise.
16894 * passes.c (pass_manager::register_one_dump_file): Likewise.
16895 * print-tree.c: Likewise.
16896 * statistics.c: Likewise.
16897 * tree-cfg.c: Likewise.
16898 * tree-cfg.h: Likewise.
16899 * tree-dfa.c: Likewise.
16900 * tree-dfa.h: Likewise.
16901 * tree-dump.c (dump_function): Likewise.
16902 * tree-dump.h (struct dump_info): Likewise.
16903 * tree-pretty-print.c: Likewise.
16904 * tree-pretty-print.h: Likewise.
16905 * tree-ssa-live.c: Likewise.
16906 * tree-ssa-live.h: Likewise.
16907 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
16908 * tree-vect-loop.c: Likewise.
16909 * tree-vect-slp.c: Likewise.
16911 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
16912 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16914 PR tree-optimization/80457
16915 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
16916 of all arguments to a statement as scalar_to_vec operations.
16917 (vectorizable_call): Adjust call to vect_model_simple_cost for
16919 (vectorizable_conversion): Likewise.
16920 (vectorizable_assignment): Likewise.
16921 (vectorizable_shift): Likewise.
16922 (vectorizable_operation): Likewise.
16923 (vectorizable_comparison): Likewise.
16924 (vect_is_simple_cond): Record the def types for operands.
16925 (vectorizable_condition): Likewise, call vect_model_simple_cost.
16926 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
16927 for statement argument count.
16929 2017-05-16 Carl Love <cel@us.ibm.com>
16931 * config/rs6000/rs6000-c: Add support for built-in functions
16932 vector unsigned long long vec_bperm (vector unsigned long long,
16933 vector unsigned char)
16934 vector signed long long vec_mule (vector signed int,
16936 vector unsigned long long vec_mule (vector unsigned int,
16937 vector unsigned int)
16938 vector signed long long vec_mulo (vector signed int,
16940 vector unsigned long long vec_mulo (vector unsigned int,
16941 vector unsigned int)
16942 vector signed char vec_sldw (vector signed char,
16943 vector signed char,
16945 vector unsigned char vec_sldw (vector unsigned char,
16946 vector unsigned char,
16948 vector signed short vec_sldw (vector signed short,
16949 vector signed short,
16951 vector unsigned short vec_sldw (vector unsigned short,
16952 vector unsigned short,
16954 vector signed int vec_sldw (vector signed int,
16957 vector unsigned int vec_sldw (vector unsigned int,
16958 vector unsigned int,
16960 vector signed long long vec_sldw (vector signed long long,
16961 vector signed long long,
16963 vector unsigned long long vec_sldw (vector unsigned long long,
16964 vector unsigned long long,
16966 * config/rs6000/rs6000-c: Add support for built-in functions
16967 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
16968 * config/rs6000/altivec.h: Add defintion for vec_sldw.
16969 * doc/extend.texi: Update the built-in documentation for the
16970 new built-in functions.
16972 2017-05-16 Marek Polacek <polacek@redhat.com>
16976 * tree.c (save_expr): Don't fold the expression.
16978 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
16980 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
16981 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
16982 and (?*y,m). Update insn attributes.
16984 2017-05-16 Martin Liska <mliska@suse.cz>
16986 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
16987 flags argument of print_gimple_stmt, print_gimple_expr,
16988 print_generic_stmt and print_generic_expr.
16989 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
16990 * coretypes.h: Likewise.
16991 * except.c (dump_eh_tree): Likewise.
16992 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
16993 * gimple-pretty-print.h: Likewise.
16994 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
16995 (backprop::push_to_worklist): Likewise.
16996 (backprop::pop_from_worklist): Likewise.
16997 (backprop::process_use): Likewise.
16998 (backprop::intersect_uses): Likewise.
16999 (note_replacement): Likewise.
17000 * gimple-ssa-store-merging.c
17001 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
17002 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
17003 (pass_store_merging::execute): Likewise.
17004 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
17005 (ssa_base_cand_dump_callback): Likewise.
17006 (dump_incr_vec): Likewise.
17007 (replace_refs): Likewise.
17008 (replace_mult_candidate): Likewise.
17009 (create_add_on_incoming_edge): Likewise.
17010 (create_phi_basis): Likewise.
17011 (insert_initializers): Likewise.
17012 (all_phi_incrs_profitable): Likewise.
17013 (introduce_cast_before_cand): Likewise.
17014 (replace_one_candidate): Likewise.
17015 * gimplify.c (gimplify_expr): Likewise.
17016 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
17017 (set_rename): Likewise.
17018 (rename_uses): Likewise.
17019 (copy_loop_phi_nodes): Likewise.
17020 (add_close_phis_to_merge_points): Likewise.
17021 (copy_loop_close_phi_args): Likewise.
17022 (copy_cond_phi_args): Likewise.
17023 (graphite_copy_stmts_from_block): Likewise.
17024 (translate_pending_phi_nodes): Likewise.
17025 * graphite-poly.c (print_pdr): Likewise.
17026 (dump_gbb_cases): Likewise.
17027 (dump_gbb_conditions): Likewise.
17028 (print_scop_params): Likewise.
17029 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
17030 (build_cross_bb_scalars_use): Likewise.
17031 (gather_bbs::before_dom_children): Likewise.
17032 * hsa-dump.c (dump_hsa_immed): Likewise.
17033 * ipa-cp.c (print_ipcp_constant_value): Likewise.
17034 (get_replacement_map): Likewise.
17035 * ipa-inline-analysis.c (dump_condition): Likewise.
17036 (estimate_function_body_sizes): Likewise.
17037 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
17038 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
17039 * ipa-prop.c (ipa_dump_param): Likewise.
17040 (ipa_print_node_jump_functions_for_edge): Likewise.
17041 (ipa_modify_call_arguments): Likewise.
17042 (ipa_modify_expr): Likewise.
17043 (ipa_dump_param_adjustments): Likewise.
17044 (ipa_dump_agg_replacement_values): Likewise.
17045 (ipcp_modif_dom_walker::before_dom_children): Likewise.
17046 * ipa-pure-const.c (check_stmt): Likewise.
17047 (pass_nothrow::execute): Likewise.
17048 * ipa-split.c (execute_split_functions): Likewise.
17049 * omp-offload.c (dump_oacc_loop_part): Likewise.
17050 (dump_oacc_loop): Likewise.
17051 * trans-mem.c (tm_log_emit): Likewise.
17052 (tm_memopt_accumulate_memops): Likewise.
17053 (dump_tm_memopt_set): Likewise.
17054 (dump_tm_memopt_transform): Likewise.
17055 * tree-cfg.c (gimple_verify_flow_info): Likewise.
17056 (print_loop): Likewise.
17057 * tree-chkp-opt.c (chkp_print_addr): Likewise.
17058 (chkp_gather_checks_info): Likewise.
17059 (chkp_get_check_result): Likewise.
17060 (chkp_remove_check_if_pass): Likewise.
17061 (chkp_use_outer_bounds_if_possible): Likewise.
17062 (chkp_reduce_bounds_lifetime): Likewise.
17063 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
17064 (chkp_mark_completed_bounds): Likewise.
17065 (chkp_register_incomplete_bounds): Likewise.
17066 (chkp_mark_invalid_bounds): Likewise.
17067 (chkp_maybe_copy_and_register_bounds): Likewise.
17068 (chkp_build_returned_bound): Likewise.
17069 (chkp_get_bound_for_parm): Likewise.
17070 (chkp_build_bndldx): Likewise.
17071 (chkp_get_bounds_by_definition): Likewise.
17072 (chkp_generate_extern_var_bounds): Likewise.
17073 (chkp_get_bounds_for_decl_addr): Likewise.
17074 * tree-chrec.c (chrec_apply): Likewise.
17075 * tree-data-ref.c (dump_data_reference): Likewise.
17076 (dump_subscript): Likewise.
17077 (dump_data_dependence_relation): Likewise.
17078 (analyze_overlapping_iterations): Likewise.
17079 * tree-inline.c (expand_call_inline): Likewise.
17080 (tree_function_versioning): Likewise.
17081 * tree-into-ssa.c (dump_defs_stack): Likewise.
17082 (dump_currdefs): Likewise.
17083 (dump_names_replaced_by): Likewise.
17084 (dump_update_ssa): Likewise.
17085 (update_ssa): Likewise.
17086 * tree-object-size.c (pass_object_sizes::execute): Likewise.
17087 * tree-parloops.c (build_new_reduction): Likewise.
17088 (try_create_reduction_list): Likewise.
17089 (ref_conflicts_with_region): Likewise.
17090 (oacc_entry_exit_ok_1): Likewise.
17091 (oacc_entry_exit_single_gang): Likewise.
17092 * tree-pretty-print.h: Likewise.
17093 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
17094 (get_scalar_evolution): Likewise.
17095 (add_to_evolution): Likewise.
17096 (get_loop_exit_condition): Likewise.
17097 (analyze_evolution_in_loop): Likewise.
17098 (analyze_initial_condition): Likewise.
17099 (analyze_scalar_evolution): Likewise.
17100 (instantiate_scev): Likewise.
17101 (number_of_latch_executions): Likewise.
17102 (gather_chrec_stats): Likewise.
17103 (final_value_replacement_loop): Likewise.
17104 (scev_const_prop): Likewise.
17105 * tree-sra.c (dump_access): Likewise.
17106 (disqualify_candidate): Likewise.
17107 (create_access): Likewise.
17108 (reject): Likewise.
17109 (maybe_add_sra_candidate): Likewise.
17110 (create_access_replacement): Likewise.
17111 (analyze_access_subtree): Likewise.
17112 (analyze_all_variable_accesses): Likewise.
17113 (sra_modify_assign): Likewise.
17114 (initialize_constant_pool_replacements): Likewise.
17115 (find_param_candidates): Likewise.
17116 (decide_one_param_reduction): Likewise.
17117 (replace_removed_params_ssa_names): Likewise.
17118 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
17119 * tree-ssa-copy.c (dump_copy_of): Likewise.
17120 (copy_prop_visit_cond_stmt): Likewise.
17121 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
17122 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
17123 (record_equivalences_from_stmt): Likewise.
17124 * tree-ssa-dse.c (compute_trims): Likewise.
17125 (delete_dead_call): Likewise.
17126 (delete_dead_assignment): Likewise.
17127 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
17128 (forward_propagate_into_cond): Likewise.
17129 (pass_forwprop::execute): Likewise.
17130 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
17131 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
17133 (move_computations_worker): Likewise.
17134 (execute_sm): Likewise.
17135 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
17136 (remove_exits_and_undefined_stmts): Likewise.
17137 (remove_redundant_iv_tests): Likewise.
17138 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
17139 (adjust_iv_update_pos): Likewise.
17140 * tree-ssa-math-opts.c (bswap_replace): Likewise.
17141 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
17142 (value_replacement): Likewise.
17143 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
17144 * tree-ssa-pre.c (print_pre_expr): Likewise.
17145 (get_representative_for): Likewise.
17146 (create_expression_by_pieces): Likewise.
17147 (insert_into_preds_of_block): Likewise.
17148 (eliminate_insert): Likewise.
17149 (eliminate_dom_walker::before_dom_children): Likewise.
17150 (eliminate): Likewise.
17151 (remove_dead_inserted_code): Likewise.
17152 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
17153 * tree-ssa-reassoc.c (get_rank): Likewise.
17154 (eliminate_duplicate_pair): Likewise.
17155 (eliminate_plus_minus_pair): Likewise.
17156 (eliminate_not_pairs): Likewise.
17157 (undistribute_ops_list): Likewise.
17158 (eliminate_redundant_comparison): Likewise.
17159 (update_range_test): Likewise.
17160 (optimize_range_tests_var_bound): Likewise.
17161 (optimize_vec_cond_expr): Likewise.
17162 (rewrite_expr_tree): Likewise.
17163 (rewrite_expr_tree_parallel): Likewise.
17164 (linearize_expr): Likewise.
17165 (break_up_subtract): Likewise.
17166 (linearize_expr_tree): Likewise.
17167 (attempt_builtin_powi): Likewise.
17168 (attempt_builtin_copysign): Likewise.
17169 (transform_stmt_to_copy): Likewise.
17170 (transform_stmt_to_multiply): Likewise.
17171 (dump_ops_vector): Likewise.
17172 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
17173 (print_scc): Likewise.
17174 (set_ssa_val_to): Likewise.
17175 (visit_reference_op_store): Likewise.
17176 (visit_use): Likewise.
17177 (sccvn_dom_walker::before_dom_children): Likewise.
17178 (run_scc_vn): Likewise.
17179 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
17181 (expr_hash_elt::print): Likewise.
17182 (const_and_copies::pop_to_marker): Likewise.
17183 (const_and_copies::record_const_or_copy_raw): Likewise.
17184 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
17185 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
17186 (dump_predicates): Likewise.
17187 (find_uninit_use): Likewise.
17188 (warn_uninitialized_phi): Likewise.
17189 (pass_late_warn_uninitialized::execute): Likewise.
17190 * tree-ssa.c (verify_vssa): Likewise.
17191 (verify_ssa): Likewise.
17192 (maybe_optimize_var): Likewise.
17193 * tree-vrp.c (dump_value_range): Likewise.
17194 (dump_all_value_ranges): Likewise.
17195 (dump_asserts_for): Likewise.
17196 (register_edge_assert_for_2): Likewise.
17197 (vrp_visit_cond_stmt): Likewise.
17198 (vrp_visit_switch_stmt): Likewise.
17199 (vrp_visit_stmt): Likewise.
17200 (vrp_visit_phi_node): Likewise.
17201 (simplify_cond_using_ranges_1): Likewise.
17202 (fold_predicate_in): Likewise.
17203 (evrp_dom_walker::before_dom_children): Likewise.
17204 (evrp_dom_walker::push_value_range): Likewise.
17205 (evrp_dom_walker::pop_value_range): Likewise.
17206 (execute_early_vrp): Likewise.
17208 2017-05-16 Richard Biener <rguenther@suse.de>
17210 * dwarf2out.c (loc_list_from_tree_1): Do not create
17211 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
17213 2017-05-16 Richard Biener <rguenther@suse.de>
17215 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
17217 (note_variable_value_in_expr): If we resolved the decl ref
17218 do not push to the stack.
17220 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
17222 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
17223 operations in fast-math mode.
17224 (vaddq_f16): Likewise.
17225 (vmul_f16): Likewise.
17226 (vmulq_f16): Likewise.
17227 (vsub_f16): Likewise.
17228 (vsubq_f16): Likewise.
17229 * config/arm/neon.md (add<mode>3): New.
17231 (fma:<VH:mode>3): New. Also remove outdated comment.
17234 2017-05-16 Martin Liska <mliska@suse.cz>
17238 * ipa-devirt.c (warn_types_mismatch): Fix typo.
17239 (odr_types_equivalent_p): Likewise.
17241 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
17243 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
17245 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
17248 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
17249 non-interunit SSE move alternatives with '?'.
17250 (zero-extendsidi peephole2): New peephole to skip intermediate
17251 general register in SSE zero-extend sequence.
17253 2017-05-15 Jeff Law <law@redhat.com>
17255 * reorg.c (relax_delay_slots): Create a new variable to hold
17256 the temporary target rather than clobbering TARGET_LABEL.
17258 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
17259 missing argument to extract_bit_field call.
17260 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
17262 2017-05-15 Martin Liska <mliska@suse.cz>
17265 * gcc.c (process_command): Do not allow empty argument of -o option.
17267 2017-05-15 Renlin Li <renlin.li@arm.com>
17269 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
17270 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
17271 * config/aarch64/constraints.md (Usf): Add long call check.
17272 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
17273 (call_value): Likewise.
17274 (sibcall): Likewise.
17275 (sibcall_value): Likewise.
17277 (call_value_insn): New.
17278 (sibcall_insn): Update rtx pattern.
17279 (sibcall_value_insn): Likewise.
17280 (call_internal): Remove.
17281 (call_value_internal): Likewise.
17282 (sibcall_internal): Likewise.
17283 (sibcall_value_internal): Likewise.
17284 (call_reg): Likewise.
17285 (call_symbol): Likewise.
17286 (call_value_reg): Likewise.
17287 (call_value_symbol): Likewise.
17289 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
17292 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
17294 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
17296 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
17297 compatible with CCGOCmode and with CCZmode.
17299 2017-05-14 Martin Sebor <msebor@redhat.com>
17301 PR middle-end/77671
17302 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
17303 (gimple_fold_builtin_snprintf): Same.
17304 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
17305 (gimple_fold_builtin_snprintf): Same.
17306 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
17307 of character types.
17308 (is_call_safe): New function.
17309 (try_substitute_return_value): Call it.
17310 (try_simplify_call): New function.
17311 (pass_sprintf_length::handle_gimple_call): Call it.
17313 2017-05-14 Martin Sebor <msebor@redhat.com>
17315 PR middle-end/80669
17316 * builtins.c (expand_builtin_stpncpy): Simplify.
17318 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
17320 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
17321 * config/i386/i386.h
17322 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
17323 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
17324 (struct machine_function): Add new members call_ms2sysv,
17325 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
17326 (struct machine_frame_state): New fields sp_realigned and
17327 sp_realigned_offset.
17328 * config/i386/i386.c
17329 (enum xlogue_stub): New enum.
17330 (enum xlogue_stub_sets): New enum.
17331 (class xlogue_layout): New class.
17332 (struct ix86_frame): New fields stack_realign_allocate_offset,
17333 stack_realign_offset and outlined_save_offset. Modify comments to
17334 detail stack layout when using out-of-line stubs.
17335 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
17336 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
17337 -mcall-ms2sysv-xlogues.
17338 (stub_managed_regs): New static variable.
17339 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
17340 registers managed by out-of-line stub.
17341 (disable_call_ms2sysv_xlogues): New function.
17342 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
17343 m->call_ms2sysv when appropriate and compute frame layout for
17345 (sp_valid_at, fp_valid_at): New inline functions.
17346 (choose_basereg): New function.
17347 (choose_baseaddr): Add align parameter, use choose_basereg and modify
17349 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
17350 Use align parameter of choose_baseaddr to generated aligned SSE movs
17352 (pro_epilogue_adjust_stack): Modify to track
17353 machine_frame_state::sp_realigned.
17354 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
17355 (ix86_nsaved_sseregs): Likewise.
17356 (ix86_emit_save_regs): Likewise.
17357 (ix86_emit_save_regs_using_mov): Likewise.
17358 (ix86_emit_save_sse_regs_using_mov): Likewise.
17359 (get_scratch_register_on_entry): Likewise.
17360 (gen_frame_set): New function.
17361 (gen_frame_load): Likewise.
17362 (gen_frame_store): Likewise.
17363 (emit_outlined_ms2sysv_save): Likewise.
17364 (emit_outlined_ms2sysv_restore): Likewise.
17365 (ix86_expand_prologue): Modify stack re-alignment code and call
17366 emit_outlined_ms2sysv_save when appropriate.
17367 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
17368 parameter rtx_insn *insn, which allows the function to be used to only
17369 generate the notes.
17370 (ix86_expand_epilogue): Modify validity checks of frame and stack
17371 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
17372 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
17373 * config/i386/predicates.md
17374 (save_multiple): New predicate.
17375 (restore_multiple): Likewise.
17376 * config/i386/sse.md
17377 (save_multiple<mode>): New pattern.
17378 (save_multiple_realign<mode>): Likewise.
17379 (restore_multiple<mode>): Likewise.
17380 (restore_multiple_and_return<mode>): Likewise.
17381 (restore_multiple_leave_return<mode>): Likewise.
17382 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
17384 2017-05-14 Julia Koval <julia.koval@intel.com>
17386 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
17387 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
17388 (__builtin_ia32_xsetbv): New builtins.
17389 * config/i386/i386.c (ix86_expand_special_args_builtin):
17391 (ix86_expand_builtin): Special expand for new intrinsics.
17392 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
17393 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
17394 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
17396 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17398 * cfganal.c (inverted_post_order_compute): Change argument type
17400 * cfganal.h (inverted_post_order_compute): Adjust prototype.
17401 * df-core.c (rest_of_handle_df_initialize): Adjust.
17402 (rest_of_handle_df_finish): Likewise.
17403 (df_analyze_1): Likewise.
17404 (df_analyze): Likewise.
17405 (loop_inverted_post_order_compute): Change argument to be a vec *.
17406 (df_analyze_loop): Adjust.
17407 (df_get_n_blocks): Likewise.
17408 (df_get_postorder): Likewise.
17409 * df.h (struct df_d): Change field to be a vec.
17410 * lcm.c (compute_laterin): Adjust.
17411 (compute_available): Likewise.
17412 * lra-lives.c (lra_create_live_ranges_1): Likewise.
17413 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
17414 * tree-ssa-pre.c (compute_antic): Likewise.
17416 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17418 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
17419 (depth_first_search::depth_first_search): Change structure init
17420 function to this constructor.
17421 (depth_first_search::add_bb): Rename function to this member.
17422 (depth_first_search::execute): Likewise.
17423 (flow_dfs_compute_reverse_finish): Adjust.
17425 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17427 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
17428 (longest_simple_path): Likewise.
17429 * shrink-wrap.c (spread_components): Likewise.
17430 (disqualify_problematic_components): Likewise.
17431 (emit_common_heads_for_components): Likewise.
17432 (emit_common_tails_for_components): Likewise.
17433 (insert_prologue_epilogue_for_components): Likewise.
17435 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17437 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
17440 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17442 * df-core.c (df_set_blocks): Start using auto_bitmap.
17443 (df_compact_blocks): Likewise.
17444 * df-problems.c (df_rd_confluence_n): Likewise.
17445 * df-scan.c (df_insn_rescan_all): Likewise.
17446 (df_process_deferred_rescans): Likewise.
17447 (df_update_entry_block_defs): Likewise.
17448 (df_update_exit_block_uses): Likewise.
17449 (df_entry_block_bitmap_verify): Likewise.
17450 (df_exit_block_bitmap_verify): Likewise.
17451 (df_scan_verify): Likewise.
17452 * lra-constraints.c (lra_constraints): Likewise.
17453 (undo_optional_reloads): Likewise.
17454 (lra_undo_inheritance): Likewise.
17455 * lra-remat.c (calculate_gen_cands): Likewise.
17456 (do_remat): Likewise.
17457 * lra-spills.c (assign_spill_hard_regs): Likewise.
17458 (spill_pseudos): Likewise.
17459 * tree-ssa-pre.c (bitmap_set_and): Likewise.
17460 (bitmap_set_subtract_values): Likewise.
17462 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17464 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
17465 management with auto_bitmap.
17466 (fix_inter_tick): Likewise.
17467 (fix_recovery_deps): Likewise.
17468 * ira.c (add_store_equivs): Likewise.
17469 (find_moveable_pseudos): Likewise.
17470 (split_live_ranges_for_shrink_wrap): Likewise.
17471 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
17472 (rtx_reuse_manager::seen_def_p): Likewise.
17473 (rtx_reuse_manager::set_seen_def): Likewise.
17474 * print-rtl.h (class rtx_reuse_manager): Likewise.
17476 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17478 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
17480 (migrate_btr_def): Likewise.
17481 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
17482 * df-core.c (loop_post_order_compute): Likewise.
17483 (loop_inverted_post_order_compute): Likewise.
17484 * hsa-common.h: Likewise.
17485 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
17486 * init-regs.c (initialize_uninitialized_regs): Likewise.
17487 * ipa-inline.c (resolve_noninline_speculation): Likewise.
17488 (inline_small_functions): Likewise.
17489 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
17490 * ira.c (combine_and_move_insns): Likewise.
17491 (build_insn_chain): Likewise.
17492 * loop-invariant.c (find_invariants): Likewise.
17493 * lower-subreg.c (propagate_pseudo_copies): Likewise.
17494 * predict.c (tree_predict_by_opcode): Likewise.
17495 (predict_paths_leading_to): Likewise.
17496 (predict_paths_leading_to_edge): Likewise.
17497 (estimate_loops_at_level): Likewise.
17498 (estimate_loops): Likewise.
17499 * shrink-wrap.c (try_shrink_wrapping): Likewise.
17500 (spread_components): Likewise.
17501 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
17502 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
17503 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
17504 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
17505 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
17506 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
17507 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
17508 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
17509 (mark_threaded_blocks): Likewise.
17510 (thread_through_all_blocks): Likewise.
17511 * tree-ssa.c (verify_ssa): Likewise.
17512 (execute_update_addresses_taken): Likewise.
17513 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
17515 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17517 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
17519 (post_order_compute): Likewise.
17520 (inverted_post_order_compute): Likewise.
17521 (pre_and_rev_post_order_compute_fn): Likewise.
17523 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17525 * genrecog.c (int_set::int_set): Explicitly construct our
17526 auto_vec base class.
17527 * vec.h (auto_vec::auto_vec): New constructor.
17529 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17531 * bitmap.h (class auto_bitmap): New constructor taking
17532 bitmap_obstack * argument.
17534 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17536 * bitmap.h (class auto_bitmap): Change type of m_bits to
17537 bitmap_head, and adjust ctor / dtor and member operators.
17539 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
17541 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
17542 when returned register mode doesn't match original mode.
17544 2017-05-12 Jeff Law <law@redhat.com>
17545 Jakub Jelinek <jakub@redhat.com>
17547 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
17548 we look for cc setter after the compare-elim changes.
17549 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
17550 within the vector to match what compare-elim now expects.
17551 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
17552 (xorsi3_flags, one_cmplsi2_flags): Likewise.
17554 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
17555 after the compare-elim changes.
17556 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
17557 the vector to match what compare-elim now expects.
17558 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
17559 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
17560 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
17561 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
17562 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
17564 * config/visium/visium.c (single_set_and_flags): Fix where
17565 we look for cc setter after the compare-elim changes.
17566 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
17567 with the vector to match what compare-elim now expects.
17568 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
17569 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
17570 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
17571 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
17572 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
17573 (neg<mode>2_insn_set_overflow): Likewise.
17575 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
17577 PR middle-end/79794
17578 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
17579 maybe_expand_insn call, set ops[0].target. If still set after call,
17580 set alt_rtl. Add extra arg to recursive calls.
17581 (extract_bit_field): Add alt_rtl argument. Pass to
17583 * expmed.h (extract_bit_field): Fix prototype.
17584 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
17585 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
17586 to extract_bit_field_calls.
17587 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
17588 Pass alt_rtl to extract_bit_field calls.
17589 * calls.c (store_unaligned_arguments_into_psuedos)
17590 load_register_parameters): Pass extra NULL to extract_bit_field calls.
17591 * optabs.c (maybe_legitimize_operand): Clear op->target when call
17593 * optabs.h (struct expand_operand): Add target bitfield.
17595 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
17597 * compare-elim.c (try_eliminate_compare): Canonicalize
17598 operation with embedded compare to
17599 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
17600 (set (reg) (operation)].
17602 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
17604 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
17607 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
17608 cost of adding a carry flag for ADC instruction.
17609 [case MINUS]: Ignore the cost of subtracting a carry flag
17610 for SBB instruction.
17612 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
17614 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
17616 * config/rs6000/bmiintrin.h: New file.
17617 * config/rs6000/bmi2intrin.h: New file.
17618 * config/rs6000/x86intrin.h: New file.
17620 2017-05-12 Jeff Law <law@redhat.com>
17622 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
17625 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
17627 PR middle-end/80707
17628 * tree-cfg.c: Remove cfg edges of unreachable case statements.
17630 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
17632 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
17633 early expansion of vector divide builtins.
17634 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
17635 builtins identified as having unsigned arguments.
17637 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
17639 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
17640 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
17641 expansion of vector logical operations (and, andc, or, xor,
17644 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
17646 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
17647 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
17649 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
17651 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
17652 early GIMPLE expansion of vector multiplies.
17654 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
17656 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
17657 TARGET_HAVE_MOVT conditional.
17658 (movt splitter): Likewise.
17660 2017-05-12 Richard Biener <rguenther@suse.de>
17662 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
17663 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
17664 Fold all stmts not inplace.
17666 2017-05-12 Richard Biener <rguenther@suse.de>
17668 PR tree-optimization/80713
17669 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
17670 inserted_exprs bit for not removed stmts.
17672 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
17674 PR middle-end/69921
17675 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
17676 parallelized" attribute for parallelized OpenACC kernels.
17677 * omp-offload.c (execute_oacc_device_lower): Use it.
17679 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
17680 Set "oacc kernels" attribute.
17681 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
17682 parameter. Adjust all users.
17683 (oacc_fn_attrib_kernels_p): Remove function.
17684 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
17685 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
17686 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
17687 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
17688 assert "oacc kernels" attribute is set.
17690 2017-05-11 Carl Love <cel@us.ibm.com>
17692 * config/rs6000/rs6000-c: Add support for built-in functions
17693 vector unsigned char vec_popcnt (vector signed char)
17694 vector unsigned char vec_popcnt (vector unsigned char)
17695 vector unsigned short vec_popcnt (vector signed short)
17696 vector unsigned short vec_popcnt (vector unsigned short)
17697 vector unsigned int vec_popcnt (vector signed int)
17698 vector unsigned int vec_popcnt (vector unsigned int)
17699 vector unsigned long long vec_popcnt (vector signed long long)
17700 vector unsigned long long vec_popcnt (vector unsigned long long)
17701 vector signed long long vec_slo (vector signed long long,
17702 vector signed char)
17703 vector signed long long vec_slo (vector signed long long,
17704 vector unsigned char)
17705 vector unsigned long long vec_slo (vector unsigned long long,
17706 vector signed char)
17707 vector unsigned long long vec_slo (vector unsigned long long,
17708 vector unsigned char)
17709 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
17710 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
17711 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
17712 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
17713 * doc/extend.texi: Update the built-in documentation file for the
17714 new built-in functions.
17716 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
17718 * attribs.h (sorted_attr_string): Move machine independent
17719 functions for target clone support from the i386 port to common
17720 code. Rename ix86_function_versions to common_function_versions.
17721 Rename make_name to make_unique_name.
17722 (common_function_versions): Likewise.
17723 (make_unique_name): Likewise.
17724 (make_dispatcher_decl): Likewise.
17725 (is_function_default_version): Likewise.
17726 * attribs.c (attr_strcmp): Likewise.
17727 (sorted_attr_string): Likewise.
17728 (common_function_versions): Likewise.
17729 (make_unique_name): Likewise.
17730 (make_dispatcher_decl): Likewise.
17731 (is_function_default_version): Likewise.
17732 * config/i386/i386.c (attr_strcmp): Likewise.
17733 (sorted_attr_string): Likewise.
17734 (ix86_function_versions): Likewise.
17735 (make_name): Likewise.
17736 (make_dispatcher_decl): Likewise.
17737 (is_function_default_version): Likewise.
17738 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
17740 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17743 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
17744 Account for direct move costs for vec_construct of integer
17747 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
17750 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
17751 (UNSPEC_STX_ATOMIC): Ditto.
17752 (loaddi_via_sse): New insn.
17753 (storedi_via_sse): Ditto.
17754 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
17755 Update corresponding peephole2 patterns.
17756 (atomic_storedi_fpu): Ditto.
17758 2017-05-11 Julia Koval <julia.koval@intel.com>
17760 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
17761 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
17763 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
17764 (__builtin_ia32_rsqrt14ss_mask): New builtins.
17765 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
17767 2017-05-11 Nathan Sidwell <nathan@acm.org>
17769 * graphite-poly.c: Include dumpfile.h.
17771 * dumpfle.h (dump_function): Declare here ...
17772 * tree-dump.h (dump_function): ... not here.
17773 * dumpfile.c: #include tree-cfg.h.
17774 (dump_function): Move here from ...
17775 * tree-dump.c (dump_function): ... here.
17776 * gimplify.c: #include splay-tree.h, not tree-dump.h.
17777 * graphite-poly.c: Don't include tree-dump.h.
17778 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
17779 * print-tree.c: Likewise.
17780 * stor-layout.c: Likewise.
17781 * tree-nested.c: Likewise.
17783 * dumpfile.c (dump_start): Use TDF_FLAGS.
17784 (dump_enable_all): Fix TDF_KIND check thinko.
17786 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
17788 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17789 array entries to represent two legal parameterizations of the
17790 overloaded __builtin_cmpb function, as represented by the
17791 P6_OV_BUILTIN_CMPB constant.
17792 (altivec_resolve_overloaded_builtin): Add special case handling
17793 for the __builtin_cmpb function, as represented by the
17794 P6_OV_BUILTIN_CMPB constant.
17795 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
17796 (BU_P6_64BIT_2): New macro.
17797 (BU_P6_OVERLOAD_2): New macro
17798 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
17799 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
17800 (CMPB): Add overload support to represent both 32-bit and 64-bit
17801 compare-bytes function.
17802 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
17803 support for TARGET_CMPB.
17804 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
17805 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
17806 documentation of the __builtin_cmpb overloaded built-in function.
17808 2017-05-11 Richard Biener <rguenther@suse.de>
17810 PR tree-optimization/80705
17811 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
17812 bases are not vectorizable.
17814 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17816 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
17817 when counting register pressure.
17819 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17821 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
17822 (iv_ca_more_deps): Renamed to ...
17823 (iv_ca_compare_deps): ... this.
17824 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
17826 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17828 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
17830 (determine_group_iv_costs): ... here.
17831 (find_inv_vars_cb): Record inv var if it's not recorded before.
17833 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17835 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
17836 (get_shiftadd_cost): Ditto.
17838 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17840 * tree-ssa-address.c: Include header file.
17841 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
17843 (add_to_parts): Refactor.
17844 (addr_to_parts): New parameter. Update use of move_hint_to_base.
17845 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
17848 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17850 PR tree-optimization/53090
17851 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
17853 (extract_cond_operands): Detect condition with IV on both sides
17854 and return COMP_IV_EXPR_2.
17855 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
17856 (rewrite_use_compare): Simplify by removing call to function
17857 extract_cond_operands.
17859 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17861 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
17862 (extract_cond_operands): Detect condition comparing against non-
17863 invariant bound and return appropriate enum value.
17864 (find_interesting_uses_cond): Update use of extract_cond_operands.
17865 Handle its return value accordingly.
17866 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
17868 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17870 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
17871 nonlinear iv_use computation in loop invariant sensitive way.
17873 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17875 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
17876 (find_iv_candidates): Call relate_compare_use_with_all_cands.
17878 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17880 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
17881 (dump_cand): Support iv_cand.inv_exprs.
17882 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
17884 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
17887 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17889 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
17891 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
17892 as local function. Include necessary header files.
17893 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
17895 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17897 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
17899 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17901 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
17902 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
17903 RSHIFT_EXPR and BIT_NOT_EXPR.
17905 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17907 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
17908 (adjust_setup_cost): New parameter supporting round up adjustment.
17909 (struct address_cost_data): Delete.
17910 (force_expr_to_var_cost): Don't bound cost with spill_cost.
17911 (split_address_cost, ptr_difference_cost): Delete.
17912 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
17913 (struct ainc_cost_data): New struct.
17914 (get_address_cost_ainc): New function.
17915 (get_address_cost, get_computation_cost): Reimplement.
17916 (determine_group_iv_cost_address): Record inv_expr for all uses of
17918 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
17919 (iv_ca_has_deps): Reimplemented to ...
17920 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
17922 (iv_ca_extend): Call iv_ca_more_deps.
17924 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17926 * tree-ssa-address.c (struct mem_address): Move to header file.
17927 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
17928 * tree-ssa-address.h (struct mem_address): Move from C file.
17929 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
17931 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17933 * tree-affine.h (aff_combination_type): New interface.
17934 (aff_combination_zero_p): Remove static.
17935 (aff_combination_const_p): New interface.
17936 (aff_combination_singleton_var_p): New interfaces.
17938 2017-05-11 Richard Biener <rguenther@suse.de>
17940 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
17941 Skip unreachable blocks and destinations.
17942 (eliminate): Move stmt removal and fixup ...
17943 (fini_eliminate): ... here. Skip inserted exprs.
17944 (pass_pre::execute): Move fini_pre after fini_eliminate.
17945 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
17946 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
17947 PRE to get rid of dead code that has invalid SSA form and
17948 split critical edges again.
17950 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17952 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
17954 2017-05-11 Richard Biener <rguenther@suse.de>
17956 * passes.c (execute_function_todo): Verify loops if they are
17957 said to be up-to-date.
17958 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
17959 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
17961 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
17964 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
17965 handle calling assemble_external ourself.
17968 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
17969 modes with zero size. Enhance comment.
17971 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17973 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
17974 built-ins for vec_xl and vec_xst with short and char pointer
17977 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
17979 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
17980 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
17981 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
17982 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
17983 (_mm_maskz_min_round_ss): New intrinsics.
17984 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
17985 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
17986 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
17987 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
17988 (__builtin_ia32_minss_mask_round): New builtins.
17989 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
17990 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
17991 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
17993 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
17994 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
17996 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
17999 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
18001 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
18002 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
18003 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
18004 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
18005 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
18006 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
18007 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
18008 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
18009 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
18010 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
18011 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
18012 (__builtin_ia32_mulss_mask_round): New builtins.
18013 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
18014 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
18015 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
18017 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
18018 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
18020 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
18023 2017-05-10 Julia Koval <julia.koval@intel.com>
18025 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
18026 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
18027 (_mm256_setr_m128i): New intrinsics.
18029 2017-05-10 Julia Koval <julia.koval@intel.com>
18031 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
18032 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
18033 (_mm_maskz_rcp14_ss): New intrinsics.
18034 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
18035 (__builtin_ia32_rcp14ss_mask): New builtins.
18036 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
18038 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
18040 PR tree-optimization/51513
18041 * tree-cfg.c (gimple_seq_unreachable_p): New function.
18042 (assert_unreachable_fallthru_edge_p): Use it.
18043 (group_case_labels_stmt): Likewise.
18044 * tree-cfg.h: Prototype it.
18045 * stmt.c: Include cfghooks.h and tree-cfg.h.
18046 (emit_case_dispatch_table) <gap_label>: New local variable.
18047 Use it to fill dispatch table gaps.
18048 Test for default_label before updating probabilities.
18049 (expand_case) <default_label>: Remove unneeded initialization.
18050 Test for unreachable default case statement and remove its edge.
18051 Set default_label accordingly.
18052 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
18054 2017-05-10 Carl Love <cel@us.ibm.com>
18056 * config/rs6000/rs6000-c: Add support for built-in functions
18057 vector signed char vec_neg (vector signed char)
18058 vector signed short int vec_neg (vector short int)
18059 vector signed int vec_neg (vector signed int)
18060 vector signed long long vec_neg (vector signed long long)
18061 vector float vec_neg (vector float)
18062 vector double vec_neg (vector double)
18063 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
18065 * config/rs6000/altivec.h: Add define for vec_neg
18066 * doc/extend.texi: Update the built-in documentation for the
18067 new built-in functions.
18069 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18071 PR tree-optimization/77644
18072 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
18074 2017-05-10 Nathan Sidwell <nathan@acm.org>
18076 * dumpfile.h (TDI_lang_all): New.
18077 (TDF_KIND): New. Renumber others
18078 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
18080 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
18082 (get_dump_file_name): Adjust suffix generation.
18083 (dump_enable_all): Use TDF_KIND.
18084 * doc/invoke.texi (-fdump-lang-all): Document.
18086 * dumpfile.h: Tabify.
18088 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
18091 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
18092 Move member access before delete.
18094 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
18096 * tree-inline.c (expand_call_inline): Split block at stmt
18099 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
18102 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
18103 are now unused after splitting mov{sf,sd}_hardfloat.
18104 (f32_lr2): Likewise.
18105 (f32_lm): Likewise.
18106 (f32_lm2): Likewise.
18107 (f32_li): Likewise.
18108 (f32_li2): Likewise.
18109 (f32_lv): Likewise.
18110 (f32_sr): Likewise.
18111 (f32_sr2): Likewise.
18112 (f32_sm): Likewise.
18113 (f32_sm2): Likewise.
18114 (f32_si): Likewise.
18115 (f32_si2): Likewise.
18116 (f32_sv): Likewise.
18117 (f32_dm): Likewise.
18118 (f32_vsx): Likewise.
18119 (f32_av): Likewise.
18120 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
18121 For movsf, order stores so the VSX stores occur before the GPR
18122 store which encourages the register allocator to use a traditional
18123 FPR instead of a GPR. For movsd, order the stores so that the GPR
18124 store comes before the VSX stores to allow the power6 to work.
18125 This is due to the power6 not having a 32-bit integer store
18126 instruction from a FPR.
18127 (movsf_hardfloat): Likewise.
18128 (movsd_hardfloat): Likewise.
18130 2017-05-09 Martin Sebor <msebor@redhat.com>
18132 PR translation/80280
18133 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
18136 PR translation/80280
18137 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
18138 data member added in r247778.
18139 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
18141 2017-05-09 Nathan Sidwell <nathan@acm.org>
18143 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
18145 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
18148 2017-05-09 Marek Polacek <polacek@redhat.com>
18150 * doc/invoke.texi: Fix typo.
18152 2017-05-09 Richard Biener <rguenther@suse.de>
18154 * tree-vrp.c (vrp_val_is_max): Adjust comment.
18155 (vrp_val_is_min): Likewise.
18156 (set_value_range_to_value): Likewise.
18157 (set_value_range_to_nonnegative): Likewise.
18158 (gimple_assign_nonzero_p): Likewise.
18159 (gimple_stmt_nonzero_p): Likewise.
18160 (vrp_int_const_binop): Likewise. Remove unreachable case.
18161 (adjust_range_with_scev): Adjust comments.
18162 (compare_range_with_value): Likewise.
18163 (extract_range_from_phi_node): Likewise.
18164 (test_for_singularity): Likewise.
18166 2017-05-09 Richard Biener <rguenther@suse.de>
18168 * tree-vrp.c (get_single_symbol): Add assert that we don't
18169 get overflowed constants as invariant part.
18170 (compare_values_warnv): Add comment before the TREE_NO_WARNING
18171 checks. Use wi::cmp instead of recursing for integer constants.
18172 (compare_values): Just ignore whether we assumed undefined
18173 overflow instead of failing the compare.
18174 (extract_range_for_var_from_comparison_expr): Add comment before the
18175 TREE_NO_WARNING sets.
18176 (test_for_singularity): Likewise.
18177 (extract_range_from_comparison): Do not disable optimization
18178 when we assumed undefined overflow.
18179 (extract_range_basic): Remove init of unused var.
18181 2017-05-09 Richard Biener <rguenther@suse.de>
18183 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
18184 (extract_range_from_multiplicative_op_1): Adjust.
18185 (extract_range_from_binary_expr_1): Use int_const_binop.
18187 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
18190 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
18191 rs6000_store_data_bypass_p in seven define_bypass directives and
18192 in several comments.
18193 * config/rs6000/rs6000-protos.h: Add prototype for
18194 rs6000_store_data_bypass_p function.
18195 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
18196 function implements slightly different (rs6000-specific) semantics
18197 than store_data_bypass_p, returning false rather than aborting
18198 with assertion error when arguments do not satisfy the
18199 requirements of store data bypass.
18200 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
18201 rs6000_store_data_bypass_p.
18203 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
18205 * config/xtensa/xtensa-protos.h
18206 (xtensa_initial_elimination_offset): New declaration.
18207 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
18208 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
18209 macro definition, add case for FRAME_POINTER_REGNUM when
18210 FRAME_GROWS_DOWNWARD.
18211 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
18212 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
18213 xtensa_initial_elimination_offset.
18215 2017-05-08 Nathan Sidwell <nathan@acm.org>
18217 * doc/invoke.texi: Alphabetize -fdump options.
18219 2017-05-08 Martin Sebor <msebor@redhat.com>
18221 PR translation/80280
18222 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
18224 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
18226 * target.def (compute_frame_layout): New optional target hook.
18227 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
18228 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
18229 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
18231 * reload1.c (verify_initial_elim_offsets): Likewise.
18232 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
18233 (use_simple_return_p): Call arm_compute_frame_layout if needed.
18234 (arm_get_frame_offsets): Split up into this ...
18235 (arm_compute_frame_layout): ... and this function.
18237 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
18239 * config/aarch64/constraints.md (Usa): New constraint.
18240 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
18242 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
18244 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
18245 with_multilib_list after it has been checked.
18247 2017-05-08 Richard Biener <rguenther@suse.de>
18249 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
18250 (bitmap_set_subtract_values): Likewise.
18252 2017-05-08 Richard Biener <rguenther@suse.de>
18254 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
18255 (gimple_assign_nonzero): ... this and remove strict_overflow_p
18257 (gimple_stmt_nonzero_warnv_p): Rename to ...
18258 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
18260 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
18261 (extract_range_basic): Adjust, do not disable propagation on
18262 strict overflow sensitive simplification.
18263 (vrp_visit_cond_stmt): Likewise.
18265 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
18267 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
18268 body size unconditionally.
18270 2017-05-07 Jeff Law <law@redhat.com>
18273 2017-05-06 Jeff Law <law@redhat.com>
18274 PR tree-optimization/78496
18275 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
18278 PR tree-optimization/78496
18279 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
18280 (simplify_stmt_using_ranges): Call it.
18281 (vrp_dom_walker::before_dom_children): Extract equivalences
18282 from an ASSERT_EXPR with an equality comparison against a
18285 2017-05-06 Jeff Law <law@redhat.com>
18287 PR tree-optimization/78496
18288 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
18291 PR tree-optimization/78496
18292 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
18293 (simplify_stmt_using_ranges): Call it.
18294 (vrp_dom_walker::before_dom_children): Extract equivalences
18295 from an ASSERT_EXPR with an equality comparison against a
18298 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
18300 * lra-constraints.c (lra_copy_reg_equiv): New function.
18301 (split_reg): Use it to copy equivalence information from the
18302 original register to the spill register.
18304 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
18306 PR rtl-optimization/75964
18307 * simplify-rtx.c (simplify_const_relational_operation): Remove
18308 invalid handling of comparisons of integer ABS.
18310 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
18312 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
18313 initialize to zero.
18314 (init_regs): Remove declaration.
18315 (function_arg_advance_32): Initialize error_p as boolean variable.
18317 2017-05-05 Nathan Sidwell <nathan@acm.org>
18319 * store-motion.c (remove_reachable_equiv_notes): Reformat long
18320 lines. Use for (;;).
18322 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18324 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
18325 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
18326 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
18327 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
18328 VF=2 that require versioning.
18330 2017-05-05 David Malcolm <dmalcolm@redhat.com>
18332 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
18335 2017-05-05 David Malcolm <dmalcolm@redhat.com>
18337 * diagnostic.h (diagnostic_override_option_index): Convert from
18338 macro to inline function.
18340 2017-05-05 David Malcolm <dmalcolm@redhat.com>
18342 * diagnostic.c (last_module_changed_p): New function.
18343 (set_last_module): New function.
18344 (diagnostic_report_current_module): Convert macro usage to
18345 the above functions.
18346 * diagnostic.h (diagnostic_context::last_module): Strengthen
18347 from const line_map * to const line_map_ordinary *.
18348 (diagnostic_last_module_changed): Delete macro.
18349 (diagnostic_set_last_module): Delete macro.
18351 2017-05-05 David Malcolm <dmalcolm@redhat.com>
18353 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
18354 with diagnostic_report_diagnostic.
18355 (diagnostic_n_impl_richloc): Likewise.
18356 * diagnostic.h (report_diagnostic): Delete macro.
18357 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
18358 with diagnostic_report_diagnostic.
18359 * substring-locations.c (format_warning_va): Likewise.
18361 2017-05-05 David Malcolm <dmalcolm@redhat.com>
18363 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
18364 save/restor of format_spec. Move option-printing code to...
18365 (print_option_information): ...this new function, and
18366 reimplement by simply printing to the pretty_printer,
18367 rather than appending to the format string.
18369 2017-05-05 David Malcolm <dmalcolm@redhat.com>
18371 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
18372 handling logic into...
18373 (update_effective_level_from_pragmas): ...this new function.
18375 2017-05-04 Andrew Waterman <andrew@sifive.com>
18377 * config/riscv/riscv.opt (mstrict-align): New option.
18378 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
18379 (SLOW_UNALIGNED_ACCESS): Define.
18380 (riscv_slow_unaligned_access): Declare.
18381 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
18383 (riscv_slow_unaligned_access): New variable.
18384 (rocket_tune_info): Set slow_unaligned_access to true.
18385 (optimize_size_tune_info): Set slow_unaligned_access to false.
18386 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
18387 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
18388 (riscv_option_override): Set riscv_slow_unaligned_access.
18389 * doc/invoke.texi: Add -mstrict-align to RISC-V.
18391 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
18393 * config/riscv/riscv.md: Unify indentation.
18395 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
18400 * config/rs6000/rs6000.md (u code attribute): Add FIX and
18402 (extendsi<mode>2): Add support for doing sign extension via
18403 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
18404 don't have ISA 3.0 instructions.
18405 (extendsi<mode>2 splitter): Likewise.
18406 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
18407 generate the normal insns since SImode can now go in vector
18408 registers. Disallow the special UNSPECs needed for previous
18409 machines to hide SImode being used. Add new insns
18410 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
18411 (fix_trunc<mode>si2_stfiwx): Likewise.
18412 (fix_trunc<mode>si2_internal): Likewise.
18413 (fixuns_trunc<mode>si2): Likewise.
18414 (fixuns_trunc<mode>si2_stfiwx): Likewise.
18415 (fctiw<u>z_<mode>_smallint): Likewise.
18416 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
18417 of floating point to 32-bit integer from doing a direct move to
18418 the GPR registers to do a store.
18419 (fctiwz_<mode>): Break long line.
18421 2017-05-05 Bin Cheng <bin.cheng@arm.com>
18423 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
18424 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
18425 (addr_list, addr_offset_valid_p): New.
18426 (split_address_groups): Check offset validity with above function.
18427 (gt-tree-ssa-loop-ivopts.h): Include header file.
18429 2017-05-05 Nathan Sidwell <nathan@acm.org>
18431 * config.gcc (arm*-*-*): Add missing 'fi'.
18433 2017-05-05 Steve Ellcey <sellcey@cavium.com>
18435 * doc/invoke.texi (-fopt-info): Explicitly say order of options
18436 included in -fopt-info does not matter.
18437 * doc/optinfo.texi (-fopt-info): Fix description of default
18438 behavour. Explicitly say order of options included in -fopt-info
18441 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
18443 * config.gcc: Allow combinations of aprofile and rmprofile values for
18444 --with-multilib-list.
18445 * config/arm/t-multilib: New file.
18446 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
18447 variables. Remove setting of ISA and floating-point ABI in
18448 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
18449 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
18450 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
18451 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
18453 * config/arm/t-rmprofile: Likewise except for the matches changes.
18454 * doc/install.texi (--with-multilib-list): Document the combination of
18455 aprofile and rmprofile values and warn about pitfalls in doing that.
18457 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
18459 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
18460 (movdi_aarch64): Likewise.
18462 2017-05-05 Jakub Jelinek <jakub@redhat.com>
18464 PR tree-optimization/80632
18465 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
18467 (build_arrays): Initialize it for virtual phis.
18468 (fix_phi_nodes): Use it for virtual phis.
18470 PR tree-optimization/80558
18471 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
18472 [x, y] op z into [x op, y op z] for op & or | if conditions
18475 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
18476 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
18479 * config/arm/arm.md (use_literal_pool): Remove.
18480 (64-bit immediate split): No longer takes cost into consideration
18481 if arm_disable_literal_pool is enabled.
18482 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
18483 used when arm_disable_literal_pool is enabled.
18484 (arm_max_const_double_inline_cost): Remove use of
18485 arm_disable_literal_pool.
18486 (push_minipool_fix): Add assert.
18487 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
18488 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
18489 (no_literal_pool_sf_immediate): New.
18491 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18493 PR tree-optimization/80613
18494 * tree-ssa-dce.c (propagate_necessity): Remove cases for
18495 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
18497 2017-05-05 Richard Biener <rguenther@suse.de>
18499 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
18501 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
18503 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
18504 of this flag from insn conditions due to removal from r247495.
18506 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
18508 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
18510 (arm_early_store_addr_dep_ptr): Likewise.
18511 * config/arm/aarch-common-protos.h
18512 (arm_early_load_addr_dep_ptr): Add prototype.
18513 (arm_early_store_addr_dep_ptr): Likewise.
18514 * config/arm/cortex-a53.md: Add new bypasses.
18516 2017-05-05 Jakub Jelinek <jakub@redhat.com>
18518 * tree.c (next_type_uid): Change type to unsigned.
18519 (type_hash_canon): Decrement back next_type_uid if
18520 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
18521 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
18524 2017-05-04 Martin Sebor <msebor@redhat.com>
18526 * builtins.c: Fix a trivial typo in a comment.
18528 PR middle-end/79234
18529 * builtins.c (check_sizes): Adjust to handle reading past the end.
18530 Avoid printing excessive upper bound of ranges. Use %E to print
18531 tree nodes instead of converting them to %wu.
18532 (expand_builtin_memchr): New function.
18533 (compute_dest_size): Rename...
18534 (compute_objsize): ...to this.
18535 (expand_builtin_memcpy): Adjust.
18536 (expand_builtin_mempcpy): Adjust.
18537 (expand_builtin_strcat): Adjust.
18538 (expand_builtin_strcpy): Adjust.
18539 (check_strncat_sizes): Adjust.
18540 (expand_builtin_strncat): Adjust.
18541 (expand_builtin_strncpy): Adjust and simplify.
18542 (expand_builtin_memset): Adjust.
18543 (expand_builtin_bzero): Adjust.
18544 (expand_builtin_memcmp): Adjust.
18545 (expand_builtin): Handle memcmp.
18546 (maybe_emit_chk_warning): Check strncat just once.
18548 2017-05-04 Martin Sebor <msebor@redhat.com>
18550 PR preprocessor/79214
18551 PR middle-end/79222
18552 PR middle-end/79223
18553 * builtins.c (check_sizes): Add inlining context and issue
18554 warnings even when -Wno-system-headers is set.
18555 (check_strncat_sizes): Same.
18556 (expand_builtin_strncat): Same.
18557 (expand_builtin_memmove): New function.
18558 (expand_builtin_stpncpy): Same.
18559 (expand_builtin): Handle memmove and stpncpy.
18561 2017-05-04 Bin Cheng <bin.cheng@arm.com>
18563 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
18564 which is not used any more.
18566 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
18568 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
18570 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
18572 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
18573 (cortexa53_tunings): Likewise.
18574 (cortexa57_tunings): Likewise.
18575 (cortexa72_tunings): Likewise.
18576 (cortexa73_tunings): Likewise.
18578 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
18580 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
18581 Set loop alignment to 8.
18583 2017-05-04 Martin Sebor <msebor@redhat.com>
18585 PR translation/80280
18586 * builtins.c (expand_builtin_object_size): Add missing quoting to
18587 %D and like directives.
18588 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
18589 (hsa_type_for_tree_type): Same.
18590 (verify_function_arguments): Same.
18591 * symtab.c (symbol_table::change_decl_assembler_name): Same.
18592 * varasm.c (get_section): Same.
18595 2017-05-04 Martin Sebor <msebor@redhat.com>
18597 PR translation/80280
18598 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
18600 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
18602 * config/aarch64/aarch64.c (generic_addrcost_table):
18603 Change HI/TI mode setting.
18605 2017-05-04 Martin Jambor <mjambor@suse.cz>
18607 PR tree-optimization/80622
18608 * tree-sra.c (comes_initialized_p): New function.
18609 (build_accesses_from_assign): Only set write lazily when
18610 comes_initialized_p is false.
18611 (analyze_access_subtree): Use comes_initialized_p.
18612 (propagate_subaccesses_across_link): Assert !comes_initialized_p
18613 instead of testing for PARM_DECL.
18615 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18617 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
18618 constraint on operand 0 to allow more general addressing modes.
18619 Adjust output template.
18620 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
18622 * config/aarch64/aarch64-protos.h
18623 (aarch64_address_valid_for_prefetch_p): Declare prototype.
18624 * config/aarch64/constraints.md (Dp): New address constraint.
18625 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
18628 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
18630 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
18631 update use of estimate_ipcp_clone_size_and_time.
18632 (estimate_local_effects): Update use of
18633 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
18634 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
18635 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
18636 Return nonspecialized time.
18638 2017-05-04 Richard Biener <rguenther@suse.de>
18640 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
18641 for the last VUSE which def dominates the PHI. Directly call
18643 (get_continuation_for_phi_1): Remove.
18645 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
18647 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
18648 to explain the use of truncating division. Cap the number of
18649 iterations to the maximum given by nb_iterations_upper_bound,
18652 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
18654 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
18655 * configure: Regenerate.
18656 * config.in: Regenerate.
18657 * config/i386/driver-mingw32.c: new file.
18658 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
18659 * config.host: Link driver-mingw32.o on MinGW host.
18660 * doc/install.texi: Document new --enable-mingw-wildcard configure
18663 2017-05-04 Marek Polacek <polacek@redhat.com>
18665 PR tree-optimization/80612
18666 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
18668 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
18669 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
18671 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
18672 (movt splitter): Likewise.
18673 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
18674 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
18675 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
18676 block for Thumb-1 with MOVT.
18677 (thumb2_legitimate_address_p): Move code block ...
18678 (can_avoid_literal_pool_for_label_p): ... into this new function.
18679 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
18681 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
18682 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
18683 "M-profile targets with the MOVT instruction".
18685 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
18687 * config/arm/arm-builtins.c (arm_init_builtins): Rename
18688 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
18689 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
18691 2017-05-04 Martin Liska <mliska@suse.cz>
18693 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
18694 variable cond_code.
18696 2017-05-04 Richard Biener <rguenther@suse.de>
18698 * tree.c (array_at_struct_end_p): Handle arrays at struct
18699 end with flexarrays more conservatively. Refactor and treat
18700 arrays of arrays or aggregates more strict. Fix
18701 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
18702 * tree.c (array_at_struct_end_p): Adjust prototype.
18703 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
18704 * gimple-fold.c (get_range_strlen): Likewise.
18705 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
18707 2017-05-04 Richard Biener <rguenther@suse.de>
18709 PR tree-optimization/31130
18710 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
18712 (supports_overflow_infinity): Likewise.
18713 (is_negative_overflow_infinity): Likewise.
18714 (is_positive_overflow_infinity): Likewise.
18715 (is_overflow_infinity): Likewise.
18716 (stmt_overflow_infinity): Likewise.
18717 (overflow_infinity_range_p): Likewise.
18718 (usable_range_p): Remove as always returning true.
18719 (make_overflow_infinity): Remove.
18720 (negative_overflow_infinity): Likewise.
18721 (positive_overflow_infinity): Likewise.
18722 (avoid_overflow_infinity): Likewise.
18723 (set_value_range): Adjust accordingly.
18724 (set_value_range_to_nonnegative): Likewise, remove now unused
18725 overflow_infinity arg.
18726 (vrp_operand_equal_p): Adjust.
18727 (update_value_range): Likewise.
18728 (range_int_cst_singleton_p): Likewise.
18729 (operand_less_p): Likewise.
18730 (compare_values_warnv): Likewise.
18731 (extract_range_for_var_from_comparison_expr): Likewise.
18732 (vrp_int_const_binop): Likewise.
18733 (zero_nonzero_bits_from_vr): Likewise.
18734 (extract_range_from_multiplicative_op_1): Likewise.
18735 (extract_range_from_binary_expr_1): Likewise.
18736 (extract_range_from_unary_expr): Likewise.
18737 (extract_range_from_comparison): Likewise.
18738 (extract_range_basic): Likewise.
18739 (adjust_range_with_scev): Likewise.
18740 (compare_ranges): Likewise.
18741 (compare_range_with_value): Likewise.
18742 (dump_value_range): Likewise.
18743 (test_for_singularity): Likewise, remove strict_overflow_p parameter
18745 (simplify_cond_using_ranges): Adjust.
18747 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
18749 * brig-builtins.def: Added a builtin for class_f64.
18750 * builtin-types.def: Added a builtin type needed by class_f64.
18752 2017-05-03 Jason Merrill <jason@redhat.com>
18754 * timevar.def: Add TV_CONSTEXPR.
18756 2017-05-03 David Malcolm <dmalcolm@redhat.com>
18758 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
18760 2017-05-03 Martin Jambor <mjambor@suse.cz>
18762 * ipa-prop.c (ipa_update_after_lto_read): Removed.
18763 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
18764 * ipa-cp.c (ipcp_propagate_stage): Do not call
18765 ipa_update_after_lto_read.
18766 * ipa-inline.c (ipa_inline): Likewise.
18768 2017-05-03 Martin Jambor <mjambor@suse.cz>
18770 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
18771 tag. Added a default constructor and a destructor.
18772 (ipa_edge_args_sum_t): New class;
18773 (ipa_edge_args_sum): Declare.
18774 (ipa_edge_args_vector): Remove declaration.
18775 (IPA_EDGE_REF): Use ipa_edge_args_sum.
18776 (ipa_free_edge_args_substructures): Remove declaration.
18777 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
18778 (ipa_edge_args_info_available_for_edge_p): Likewise.
18779 * ipa-prop.c (ipa_edge_args_vector): Removed.
18780 (edge_removal_hook_holder): Likewise.
18781 (edge_duplication_hook_holder): Likewise.
18782 (ipa_edge_args_sum): New variable.
18783 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
18784 ipa_edge_args_vector.
18785 (ipa_free_edge_args_substructures): Likewise.
18786 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
18787 ipa_edge_args_vector.
18788 (ipa_edge_removal_hook): Turned into method
18789 ipa_edge_args_sum_t::remove.
18790 (ipa_edge_duplication_hook): Turned into method
18791 ipa_edge_args_sum_t::duplicate.
18792 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
18793 registering edge hooks.
18794 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
18795 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
18796 ipa_edge_args_sum instead of ipa_edge_args_vector.
18797 * ipa-profile.c (ipa_profile): Likewise.
18799 2017-05-03 Martin Jambor <mjambor@suse.cz>
18801 * symbol-summary.h (function_summary): New method exists.
18802 (function_summary::symtab_removal): Deallocate through release.
18803 (call_summary): New class.
18804 (gt_ggc_mx): New overload.
18805 (gt_pch_nx): Likewise.
18806 (gt_pch_nx): Likewise.
18808 2017-05-03 Jeff Law <law@redhat.com>
18810 PR tree-optimization/78496
18811 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
18812 from simplify_cond_using_ranges. Split off code to walk
18813 backwards through casts into ...
18814 (simplify_cond_using_ranges_2): New function.
18815 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
18816 (execute_vrp): After identifying jump threads, call
18817 simplify_cond_using_ranges_2.
18819 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
18822 * ipa-inline.h (inline_summary): Add ctor.
18823 (create_ggc): Do not use ggc_cleared_alloc.
18825 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
18826 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18828 * gcc.c (handle_braces): Support escaping in switch matching
18830 * doc/invoke.texi (Spec Files): Document it.
18831 Remove superfluous @code markup in items.
18833 2017-05-03 David Malcolm <dmalcolm@redhat.com>
18835 * diagnostic-show-locus.c (struct column_range): New struct.
18836 (get_affected_columns): New function.
18837 (get_printed_columns): New function.
18838 (struct correction): New struct.
18839 (correction::ensure_capacity): New function.
18840 (correction::ensure_terminated): New function.
18841 (struct line_corrections): New struct.
18842 (line_corrections::~line_corrections): New dtor.
18843 (line_corrections::add_hint): New function.
18844 (layout::print_trailing_fixits): Reimplement in terms of the new
18846 (selftest::test_overlapped_fixit_printing): New function.
18847 (selftest::diagnostic_show_locus_c_tests): Call it.
18849 2017-05-03 Nathan Sidwell <nathan@acm.org>
18851 Canonicalize canonical type hashing
18852 * tree.h (type_hash_canon_hash): Declare.
18853 * tree.c (type_hash_list, attribute_hash_list): Move into
18854 type_hash_canon_hash.
18855 (build_type_attribute_qual_variant): Break out hash code calc into
18856 type_hash_canon_hash.
18857 (type_hash_canon_hash): New. Generic type hash computation.
18858 (build_range_type_1, build_array_type_1, build_function_type,
18859 build_method_type_directly, build_offset_type, build_complex_type,
18860 make_vector_type): Call it.
18862 2017-05-03 Richard Biener <rguenther@suse.de>
18864 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
18865 When all DRs have unknown misaligned do not always peel
18866 when there is a store but apply the same costing model as if
18867 there were only loads.
18869 2017-05-03 Richard Biener <rguenther@suse.de>
18872 PR tree-optimization/80492
18873 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
18874 compare_base_decls returning dont-know properly.
18876 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
18878 * config/arm/iterators.md (CCSI): New mode iterator.
18879 (arch): New mode attribute.
18880 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
18881 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
18882 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
18883 code iterator for success result mode.
18884 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
18885 the corresponding new insn generators.
18887 2017-05-03 Bin Cheng <bin.cheng@arm.com>
18890 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18891 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
18893 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
18895 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
18896 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
18897 (DDR_A): Wrap DDR argument in brackets.
18898 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
18899 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
18900 (DDR_REVERSED_P): Likewise.
18902 2017-05-03 Jakub Jelinek <jakub@redhat.com>
18904 PR tree-optimization/79472
18905 * tree-switch-conversion.c (struct switch_conv_info): Add
18906 contiguous_range and default_case_nonstandard fields.
18907 (collect_switch_conv_info): Compute contiguous_range and
18908 default_case_nonstandard fields, don't clear final_bb if
18909 contiguous_range and only the default case doesn't have the required
18911 (check_all_empty_except_final): Set default_case_nonstandard instead
18912 of failing if contiguous_range and the default case doesn't have empty
18914 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
18915 and only the default case doesn't have the required constants. Skip
18917 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
18918 if default_case_nonstandard.
18919 (build_constructors): Build constant 1 just once. Assert that default
18920 values aren't inserted in between cases if contiguous_range. Skip
18922 (build_arrays): Skip virtual phis.
18923 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
18924 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
18925 Handle virtual phis.
18926 (gen_inbound_check): Handle default_case_nonstandard case.
18927 (process_switch): Adjust check_final_bb caller. Call
18928 gather_default_values with the first non-default case instead of
18929 default case if default_case_nonstandard.
18931 2017-05-02 Nathan Sidwell <nathan@acm.org>
18933 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
18934 check. Fix formatting.
18936 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
18938 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
18939 errors when comparing specialized and unspecialized times.
18941 2017-05-02 David Malcolm <dmalcolm@redhat.com>
18943 * diagnostic-show-locus.c
18944 (layout::should_print_annotation_line_p): Make private.
18945 (layout::print_annotation_line): Make private.
18946 (layout::annotation_line_showed_range_p): Make private.
18947 (layout::show_ruler): Make private.
18948 (layout::print_source_line): Make private. Pass in line and
18949 line_width, rather than calling location_get_source_line. Drop
18951 (layout::print_leading_fixits): New method.
18952 (layout::print_any_fixits): Rename to...
18953 (layout::print_trailing_fixits): ...this, and make private.
18954 Don't print newline fixits.
18955 (diagnostic_show_locus): Move logic for printing one row into...
18956 (layout::print_line): ...this new function. Move the
18957 location_get_source_line call and error-handling from
18958 print_source_line to here. Call print_leading_fixits, and rename
18959 print_any_fixits to print_trailing_fixits.
18960 (selftest::test_fixit_insert_containing_newline): Update now that
18961 newlines are partially supported.
18962 (selftest::test_fixit_insert_containing_newline_2): New test.
18963 (selftest::test_fixit_replace_containing_newline): Update comments.
18964 (selftest::diagnostic_show_locus_c_tests): Call the new test.
18965 * edit-context.c (class added_line): New class.
18966 (class edited_line): Describe newline handling in comment.
18967 (edited_line::actually_edited_p): New method.
18968 (edited_line::print_content): Delete redundant decl.
18969 (edited_line::m_predecessors): New field.
18970 (edited_file::print_content): Call edited_line::print_content.
18971 (edited_file::print_diff): Update to support newlines.
18972 (edited_file::print_diff_hunk): Likewise.
18973 (edited_file::print_run_of_changed_lines): New function.
18974 (edited_file::print_diff_line): Convert to...
18975 (print_diff_line): ...this.
18976 (edited_file::get_effective_line_count): New function.
18977 (edited_line::edited_line): Initialize new field m_predecessors.
18978 (edited_line::~edited_line): Clean up m_predecessors.
18979 (edited_line::apply_fixit): Handle newlines.
18980 (edited_line::get_effective_line_count): New function.
18981 (edited_line::print_content): New function.
18982 (edited_line::print_diff_lines): New function.
18983 (selftest::test_applying_fixits_insert_containing_newline): New
18985 (selftest::test_applying_fixits_replace_containing_newline): New
18987 (selftest::insert_line): New function.
18988 (selftest::test_applying_fixits_multiple_lines): Add example of
18990 (selftest::edit_context_c_tests): Call the new tests.
18992 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18994 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
18995 parameter cand. Update dump information.
18996 (get_computation_cost): Update uses.
18998 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19000 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
19001 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
19002 (get_computation_at, rewrite_use_address): Update use of
19003 get_computation_aff.
19005 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19007 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
19008 (get_computation): Delete.
19009 (get_computation_cost): Implement like get_computation_cost_at.
19010 Use get_computation_at.
19011 (get_computation_cost_at): Delete.
19012 (rewrite_use_nonlinear_expr): Use get_computation_at.
19013 (rewrite_use_compare, remove_unused_ivs): Ditto.
19015 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19017 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
19019 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19021 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
19022 (ivopts_global_cost_for_size): Rename parameter and update uses.
19023 (iv_ca_recount_cost): Update uses.
19024 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
19025 candidates seperately in n_invs and n_cands.
19026 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
19028 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19030 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
19031 (find_inv_vars_cb): New.
19032 (find_depends): Renamed to ...
19033 (find_inv_vars): ... this.
19034 (add_candidate_1, force_var_cost): Call find_inv_vars.
19035 (split_address_cost, determine_group_iv_cost_cond): Ditto.
19037 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19039 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
19040 inv_vars. Add inv_exprs.
19041 (struct iv_cand): Rename depends_on to inv_vars.
19042 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
19043 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
19044 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
19045 (dump_cand): Dump inv_vars.
19046 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
19047 (record_invariant, find_depends, add_candidate_1): Ditto.
19048 (set_group_iv_cost, force_var_cost): Ditto.
19049 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
19050 (get_computation_cost_at, get_computation_cost): Ditto.
19051 (determine_group_iv_cost_generic): Ditto.
19052 (determine_group_iv_cost_address): Ditto.
19053 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
19054 (determine_group_iv_costs): Ditto.
19055 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
19056 (iv_ca_set_remove_invariants): Renamed to ...
19057 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
19058 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
19059 (iv_ca_set_add_invariants): Renamed to ...
19060 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
19061 (iv_ca_set_cp): Use iv_ca_set_add_invs.
19062 (iv_ca_has_deps): Support inv_vars and inv_exprs.
19063 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
19064 (create_new_ivs): Remove useless dump.
19066 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19068 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
19070 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
19071 (iv_ca_set_no_cp, create_new_iv): Ditto.
19073 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19075 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
19077 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
19079 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
19080 function tree_check2.
19082 2017-05-02 Martin Liska <mliska@suse.cz>
19084 * doc/gcov.texi: Add missing preposition.
19085 * gcov.c (function_info::function_info): Properly fill up
19086 all member variables.
19088 2017-05-02 Tamar Christina <tamar.christina@arm.com>
19090 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
19092 2017-05-02 Tamar Christina <tamar.christina@arm.com>
19094 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
19096 2017-05-02 Martin Liska <mliska@suse.cz>
19099 * lto-streamer-in.c (lto_read_tree_1): Remove
19100 LTO_STREAMER_DEBUG.
19101 * lto-streamer.c (struct tree_hash_entry): Likewise.
19102 (struct tree_entry_hasher): Likewise.
19103 (tree_entry_hasher::hash): Likewise.
19104 (tree_entry_hasher::equal): Likewise.
19105 (lto_streamer_init): Likewise.
19106 (lto_orig_address_map): Likewise.
19107 (lto_orig_address_get): Likewise.
19108 (lto_orig_address_remove): Likewise.
19109 * lto-streamer.h: Likewise.
19110 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
19111 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
19113 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
19115 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
19116 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
19117 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
19118 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
19119 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
19120 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
19121 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
19122 (mm_maskz_sub_ss): New intrinsics.
19123 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
19124 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
19125 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
19126 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
19127 (__builtin_ia32_subss_mask_round): New builtins.
19128 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
19129 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
19130 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
19132 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
19133 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
19135 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
19138 2017-05-02 Martin Jambor <mjambor@suse.cz>
19140 PR tree-optimization/78687
19141 * tree-sra.c (access): New field parent.
19142 (process_subtree_disqualification): New function.
19143 (disqualify_candidate): Call it.
19144 (build_accesses_from_assign): Reset write flag if creating an
19146 (build_access_subtree): Fill in parent field and also prpagate
19147 down grp_write flag.
19148 (create_artificial_child_access): New parameter set_grp_write, set
19149 grp_write to its value.
19150 (propagate_subaccesses_across_link): Also propagate grp_write flag
19152 (propagate_all_subaccesses): Push the closest parent back to work
19153 queue if add_access_to_work_queue returned true.
19155 2017-05-02 Richard Biener <rguenther@suse.de>
19157 * common.opt (fstrict-overflow): Alias negative to fwrapv.
19158 * doc/invoke.texi (fstrict-overflow): Remove all traces of
19159 -fstrict-overflow documentation.
19160 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
19161 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
19162 flag_strict_overflow.
19163 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
19164 * lto-opts.c (lto_write_options): Do not stream it.
19165 * lto-wrapper.c (merge_and_complain): Do not handle it.
19166 * opts.c (default_options_table): Do not set -fstrict-overflow.
19167 (finish_options): Likewise do not clear it when sanitizing.
19168 * simplify-rtx.c (simplify_const_relational_operation): Do not
19169 test flag_strict_overflow.
19171 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
19173 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
19174 using enabled attribute.
19175 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
19176 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
19177 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
19178 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
19179 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
19180 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
19181 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
19182 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
19183 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
19184 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
19186 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
19188 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
19190 2017-05-02 Richard Biener <rguenther@suse.de>
19192 PR tree-optimization/80591
19194 2017-04-10 Richard Biener <rguenther@suse.de>
19196 * tree-ssa-structalias.c (find_func_aliases): Properly handle
19199 2017-05-02 Richard Biener <rguenther@suse.de>
19201 PR tree-optimization/80549
19202 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
19203 (cleanup_tree_cfg_noloop): Create forwarders to known loop
19204 headers if they do not have a preheader.
19206 2017-05-02 Martin Liska <mliska@suse.cz>
19209 * common.opt: Fix typo.
19210 * doc/invoke.texi: Likewise.
19212 2017-05-01 Jan Beulich <jbeulich@suse.com>
19214 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
19215 swapping, add (x,x,m,x,n) alternative.
19217 2017-05-01 Nathan Sidwell <nathan@acm.org>
19219 * calls.c (combine_pending_stack_adjustment_and_call): Remove
19220 unnecessary unadjusted_alignment check.
19222 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
19225 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
19227 * gimplify.c (gimplify_cilk_detach): New function.
19228 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
19229 * tree-core.h: Document EXPR_CILK_SPAWN.
19230 * tree.h (EXPR_CILK_SPAWN): Define.
19232 2017-05-01 David Malcolm <dmalcolm@redhat.com>
19234 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
19235 to use new fixit_hint representation, using the "replace" logic.
19236 (get_line_span_for_fixit_hint): Likewise.
19237 (layout::print_any_fixits): Likewise.
19238 (selftest::test_one_liner_many_fixits): Rename to...
19239 (selftest::test_one_liner_many_fixits_1): ...this, and update
19240 comment and expected output to reflect that the multiple fix-it
19241 hints are now consolidated into one insertion.
19242 (selftest::test_one_liner_many_fixits_2): New test.
19243 (selftest::test_diagnostic_show_locus_one_liner): Update for
19245 (selftest::test_fixit_consolidation): Update for fix-it API
19247 * diagnostic.c (print_parseable_fixits): Likewise.
19248 * edit-context.c (edited_line::m_line_events): Convert from
19249 auto_vec <line_event *> to auto_vec <line_event>.
19250 (class line_event): Convert from abstract base class to a concrete
19251 class, taking over the role of replace_event.
19252 (class insert_event): Delete.
19253 (class replace_event): Rename to class line_event. Convert to
19255 (edit_context::add_fixits): Reimplement.
19256 (edit_context::apply_insert): Delete.
19257 (edit_context::apply_replace): Rename to...
19258 (edit_context::apply_fixit): ...this. Convert to half-open range.
19259 (edited_file::apply_insert): Delete.
19260 (edited_file::apply_replace): Rename to...
19261 (edited_file::apply_fixit): ...this.
19262 (edited_line::~edited_line): Drop deletion of events.
19263 (edited_line::apply_insert): Delete.
19264 (edited_line::apply_replace): Rename to...
19265 (edited_line::apply_fixit): ...this. Convert to half-open range.
19266 Update for change to type of m_line_events.
19267 * edit-context.h (edit_context::apply_insert): Delete.
19268 (edit_context::apply_replace): Rename to...
19269 (edit_context::apply_fixit): ...this.
19271 2017-05-01 Martin Sebor <msebor@redhat.com>
19273 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
19276 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
19279 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
19280 __get_cpuid_max returns 0.
19281 (__get_cpuid_count): Ditto.
19283 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
19285 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
19286 replacement expression is another instance of one of its arguments.
19288 2017-05-01 Jakub Jelinek <jakub@redhat.com>
19291 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
19292 check for stack push/pop autoinc.
19293 * config/i386/i386.c (ix86_agi_dependent): Return false
19294 if the only reason why modified_in_p returned true is that
19295 addr is SP based and set_insn is a push or pop.
19297 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
19299 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
19302 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
19305 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
19306 (account_size_time): Use two predicates - exec_pred and
19308 (evaluate_conditions_for_known_args): Compute both clause and
19310 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
19311 (inline_summary_t::duplicate): Update.
19312 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
19314 (compute_inline_parameters): Likewise.
19315 (estimate_edge_size_and_time): Update caluclation of time.
19316 (estimate_node_size_and_time): Compute both time and nonspecialized
19318 (estimate_ipcp_clone_size_and_time): Update.
19319 (inline_merge_summary): Update.
19320 (do_estimate_edge_time): Update.
19321 (do_estimate_edge_size): Update.
19322 (do_estimate_edge_hints): Update.
19323 (inline_read_section, inline_write_summary): Stream both new predicates.
19324 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
19326 (compute_inlined_call_time): Cleanup.
19327 (big_speedup_p): Update.
19328 (edge_badness): Update.
19329 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
19330 (size_time_entry): Replace predicate by exec_predicate and
19331 nonconst_predicate.
19332 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
19333 (estimate_edge_time): Return also nonspec_time.
19334 (reset_edge_growth_cache): Update.
19336 2017-04-29 Jakub Jelinek <jakub@redhat.com>
19338 PR rtl-optimization/80491
19339 * ifcvt.c (noce_process_if_block): When looking for x setter
19340 with missing else_bb, don't check only the insn right before
19341 cond_earliest, but look for the last insn that x is modified in
19342 within the same bb.
19344 PR rtl-optimization/80491
19345 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
19347 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
19349 PR tree-optimization/80487
19350 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
19352 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19354 PR tree-optimization/79697
19355 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
19356 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
19357 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
19359 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
19360 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
19362 2017-04-28 Martin Sebor <msebor@redhat.com>
19364 PR tree-optimization/80523
19365 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
19366 (init_target_to_host_charmap, target_to_host, target_strtol10): New
19368 (maybe_warn, format_directive, parse_directive): Use new functions.
19369 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
19371 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
19373 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
19375 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
19377 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
19378 target_header_dir): Set correctly.
19379 * configure: Regenerated.
19380 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
19381 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
19382 instead of SYSTEM_HEADER_DIR.
19384 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
19386 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
19387 (estimate_local_effects): Likewise.
19388 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
19389 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
19390 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
19391 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
19392 do_estimate_edge_time, estimate_edge_time): Likewise.
19393 * ipa-inline-analysis.c (estimate_node_size_and_time,
19394 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
19395 (estimate_time_after_inlining): Remove.
19397 2017-04-28 Martin Liska <mliska@suse.cz>
19399 * doc/gcov.texi: Enhance documentation of gcov.
19401 2017-04-28 Martin Liska <mliska@suse.cz>
19403 * doc/gcov.texi: Sort options in alphabetic order.
19404 * doc/gcov-dump.texi: Likewise.
19405 * doc/gcov-tool.texi: Likewise.
19406 * gcov.c (print_usage): Likewise.
19407 * gcov-dump.c (print_usage): Likewise.
19408 * gcov-tool.c (print_merge_usage_message): Likewise.
19409 (print_rewrite_usage_message): Likewise.
19410 (print_overlap_usage_message): Likewise.
19412 2017-04-28 Martin Liska <mliska@suse.cz>
19414 PR gcov-profile/53915
19415 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
19417 2017-04-28 Martin Liska <mliska@suse.cz>
19419 PR gcov-profile/79891
19420 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
19421 is marked by compiler as living on a line.
19422 (get_cycles_count): Remove usage of the union.
19423 (output_intermediate_file): Likewise.
19424 (find_source): Fix GNU coding style.
19425 (accumulate_line_counts): Remove old non-all block mode.
19426 (output_lines): Remove usage of the union.
19427 * profile.c (output_location): Include all BBs, even if
19428 belonging to a same line (and file) as a previous BB.
19430 2017-04-28 Martin Liska <mliska@suse.cz>
19432 * gcov.c (process_args): Handle new argument 'w'.
19433 (read_graph_file): Assign ID to BBs.
19434 (output_branch_count): Display BB # if verbose flag is set.
19435 (output_lines): Likewise for arcs.
19436 (print_usage): Add '--verbose' option help.
19437 * doc/gcov.texi: Document --verbose (-w) option.
19439 2017-04-28 Martin Liska <mliska@suse.cz>
19441 * gcov.c (struct block_location_info): New struct.
19442 (process_file): Fill up the new structure.
19443 (read_graph_file): Replace usage of encoding by the newly added
19445 (add_line_counts): Likewise.
19446 (accumulate_line_counts): Remove usage of the union.
19447 (function_info::function_info): New function.
19448 (function_info::~function_info): Likewise.
19449 (process_file): Call delete instead of release_function.
19450 (release_function): Release the function.
19451 (release_structures): Call delete instead of release_function.
19452 (solve_flow_graph): Replace usage of num_blocks.
19453 (find_exception_blocks): Likewise.
19454 (output_lines): Fix GNU coding style.
19456 2017-04-28 Martin Liska <mliska@suse.cz>
19459 * coverage.c (coverage_remove_note_file): New function.
19460 * coverage.h: Declare the function.
19461 * toplev.c (finalize): Clean if an error has been seen.
19463 2017-04-28 Martin Liska <mliska@suse.cz>
19465 PR gcov-profile/80031
19466 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
19467 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
19468 * gcov.c (read_graph_file): Read just number of blocks.
19469 * profile.c (branch_prob): Do not stream 0 flags per a basic
19472 2017-04-28 Martin Liska <mliska@suse.cz>
19474 * gcov-dump.c (tag_*): Add new argument to declarations.
19475 (dump_gcov_file): Likewise.
19476 (tag_blocks): Add and use new argument depth.
19477 (tag_arcs): Likewise.
19478 (tag_lines): Likewise.
19479 (tag_counters): Likewise.
19480 (tag_summary): Likewise.
19481 (dump_working_sets): Use depth to do a proper indentation.
19483 2017-04-28 Jakub Jelinek <jakub@redhat.com>
19486 * cgraph.h (symtab_node::debug_symtab): No longer inline.
19487 * symtab.c (symtab_node::debug_symtab): Move definition here.
19489 2017-04-28 Richard Biener <rguenther@suse.de>
19491 * lto-streamer.h (LTO_major_version): Bump to 7.
19493 2017-04-28 Richard Biener <rguenther@suse.de>
19495 * tree-vrp.c (assert_info): New struct.
19496 (add_assert_info): New helper.
19497 (register_edge_assert_for_2): Refactor to add asserts to a vector
19499 (register_edge_assert_for_1): Likewise.
19500 (register_edge_assert_for): Likewise.
19501 (finish_register_edge_assert_for): New helper actually registering
19502 asserts where live on edge.
19503 (find_conditional_asserts): Adjust.
19504 (find_switch_asserts): Likewise.
19505 (evrp_dom_walker::try_find_new_range): Generalize.
19506 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
19508 2017-04-27 Marek Polacek <polacek@redhat.com>
19511 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
19512 arg10 and arg11 to itype.
19514 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
19516 * doc/extend.texi (Object Size Checking): Improve grammar.
19518 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
19521 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
19522 that the logic for permitting reciprocal estimates matches that
19525 2017-04-27 Jakub Jelinek <jakub@redhat.com>
19528 * tree.c (type_cache_hasher::equal): Only compare
19529 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
19530 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
19531 non-aggregate element types.
19532 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
19533 about the flag on ARRAY_TYPEs in the comment, formatting fix.
19535 2017-04-27 Richard Biener <rguenther@suse.de>
19537 PR middle-end/80533
19538 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
19539 stripping ARRAY_REFs from MEM_EXPR make sure we're not
19540 keeping a reference to a trailing array.
19542 2017-04-27 Richard Biener <rguenther@suse.de>
19544 PR middle-end/80539
19545 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
19546 being in loop-closed SSA form conservatively.
19547 (chrec_fold_multiply_poly_poly): Likewise.
19549 2017-04-27 Tamar Christina <tamar.christina@arm.com>
19551 PR middle-end/79665
19552 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
19553 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
19555 2017-04-27 Jakub Jelinek <jakub@redhat.com>
19558 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
19559 (aarch64_function_arg_alignment): Return unsigned int again, but still
19560 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
19561 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
19562 Don't emit -Wpsabi note.
19563 (aarch64_function_arg_boundary): Likewise.
19564 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
19567 2017-04-26 Nathan Sidwell <nathan@acm.org>
19569 * tree.h (crc32_unsigned_n): Declare.
19570 (crc32_unsigned, crc32_unsigned): Make inline.
19571 * tree.c (crc32_unsigned_bits): Replace with ...
19572 (crc32_unsigned_n): ... this.
19573 (crc32_unsigned, crc32_byte): Remove.
19574 (crc32_string): Remove unnecessary braces.
19576 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
19578 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
19579 * ipa-inline-analysis.c (MAX_TIME): Remove.
19580 (account_size_time): Use sreal for time.
19581 (dump_inline_summary): Update.
19582 (estimate_function_body_sizes): Update.
19583 (estimate_edge_size_and_time): Update.
19584 (estimate_calls_size_and_time): Update.
19585 (estimate_node_size_and_time): Update.
19586 (inline_merge_summary): Update.
19587 (inline_update_overall_summary): Update.
19588 (estimate_time_after_inlining): Update.
19589 (inline_read_section): Update.
19590 (inline_write_summary): Update.
19591 * ipa-inline.c (compute_uninlined_call_time): Update.
19592 (compute_inlined_call_time): Update.
19593 (recursive_inlining): Update.
19594 (inline_small_functions): Update.
19595 (dump_overall_stats): Update.
19596 * ipa-inline.h: Include sreal.h.
19597 (size_time_entry): Turn time to sreal.
19598 (inline_summary): Turn self_time nad time to sreal.
19600 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
19602 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
19604 (sreal::stream_out, sreal::stream_in): New.
19605 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
19607 2017-04-25 Jakub Jelinek <jakub@redhat.com>
19609 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
19612 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
19615 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
19617 (dimode_scalar_chain::compute_convert_gain): Ditto.
19618 (dimode_scalar_chain::make_vector_copies): Ditto.
19619 (dimode_scalar_chain::convert_reg): Ditto.
19620 (dimode_scalar_chain::convert_insn): Ditto.
19621 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
19622 (VI248_AVX512BW_1): New mode iterator.
19623 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
19624 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
19627 2017-04-25 Martin Sebor <msebor@redhat.com>
19629 PR tree-optimization/80497
19630 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
19631 constants are representable in HOST_WIDE_INT.
19632 (parse_directive): Ditto.
19634 2017-04-25 Martin Sebor <msebor@redhat.com>
19637 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
19638 (new_zero_array): Adjust signature.
19639 (dom_info::dom_init): Used unsigned rather that size_t.
19640 (dom_info::dom_info): Same.
19642 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19643 Jakub Jelinek <jakub@redhat.com>
19646 * config/arm/arm.c: Include gimple.h.
19647 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
19648 returns negative, increment ncrn only if it returned positive.
19649 (arm_needs_doubleword_align): Return int instead of bool,
19650 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
19651 members, but if there is any such non-FIELD_DECL
19652 > PARM_BOUNDARY aligned decl, return -1 instead of false.
19653 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
19654 returns negative, increment nregs only if it returned positive.
19655 (arm_setup_incoming_varargs): Likewise.
19656 (arm_function_arg_boundary): Emit -Wpsabi note if
19657 arm_needs_doubleword_align returns negative, return
19658 DOUBLEWORD_ALIGNMENT only if it returned positive.
19660 2017-04-25 Marek Polacek <polacek@redhat.com>
19663 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
19664 first argument to type.
19666 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
19669 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
19670 type checks to test for compatibility instead of equality.
19672 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19673 Jakub Jelinek <jakub@redhat.com>
19676 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
19678 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
19679 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
19680 the alignment computation, but return their maximum in warn_alignment.
19681 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
19682 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
19684 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
19685 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
19688 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19690 * config/arc/simdext.md (dmpyh): Fix typo.
19692 2017-04-25 Richard Biener <rguenther@suse.de>
19694 PR tree-optimization/80492
19695 * alias.c (compare_base_decls): Handle registers with asm
19696 specification conservatively.
19697 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
19698 compare_base_decls returning dont-know properly.
19700 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19702 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
19703 (legitimate_offset_address_p): New function.
19704 (arc_legitimate_address_p): Use above function.
19706 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19708 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
19710 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19712 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
19713 ACCH registers whenever they are available.
19715 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19717 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
19718 double regs fix when not used.
19720 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19722 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
19724 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
19725 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
19727 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19729 * config/arc/arc.c (arc_output_addsi): Check for h-register class
19730 when emitting short ADD instructions.
19732 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
19734 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
19736 (cmpsi_cc_c_insn): Likewise.
19737 (cbranchsi4_scratch): Compute proper instruction length using
19738 compact_hreg_operand.
19739 * config/arc/predicates.md (compact_hreg_operand): New predicate.
19741 2017-04-25 Richard Biener <rguenther@suse.de>
19743 PR middle-end/80509
19744 * passes.c (pass_manager::pass_manager): Initialize
19745 m_name_to_pass_map.
19747 2017-04-25 Richard Biener <rguenther@suse.de>
19749 PR tree-optimization/79201
19750 * tree-ssa-sink.c (statement_sink_location): Handle calls.
19752 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19755 * config/s390/vector.md: Split MEM->GPR vector moves for
19756 non-s_operand addresses.
19758 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19761 * config/s390/predicates.md (reload_const_wide_int_operand): New
19763 * config/s390/s390.md ("movti"): Remove d/P alternative.
19764 ("movti_bigconst"): New pattern definition.
19766 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
19769 * s390-protos.h (s390_expand_cs_hqi): Removed.
19770 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
19771 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
19772 modes as well as CCZ1mode and CCZmode.
19773 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
19774 signature of s390_emit_compare_and_swap.
19775 (s390_expand_cs_hqi): Likewise, make static.
19776 (s390_expand_cs_tdsi): Generate an explicit compare before trying
19777 compare-and-swap, in some cases.
19778 (s390_expand_cs): Wrapper function.
19779 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
19781 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
19782 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
19783 patterns for small and large integers. Forbid symref memory operands.
19784 Move expander to s390.c. Require cc register.
19785 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
19786 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
19787 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
19788 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
19789 symref memory operands. Remove CC mode and call s390_match_ccmode
19791 ("atomic_exchange<mode>"): Allow and implement all integer modes.
19793 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
19795 * config/s390/s390.md (define_peephole2): New peephole to help
19796 combining the load-and-test pattern with volatile memory.
19798 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
19800 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
19801 with CCZmode for TARGET_Z196.
19803 2017-04-25 Jakub Jelinek <jakub@redhat.com>
19805 PR rtl-optimization/80501
19806 * combine.c (make_compound_operation_int): Set subreg_code to SET
19807 even for AND with mask of the sign bit of mode.
19809 PR rtl-optimization/80500
19810 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
19811 sum's initial value.
19813 2017-04-25 Julian Brown <julian@codesourcery.com>
19814 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19816 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
19818 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
19820 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
19822 2017-04-25 Julian Brown <julian@codesourcery.com>
19823 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19825 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
19826 (thunderx2t99_sha): New Reservation.
19828 2017-04-25 Julian Brown <julian@codesourcery.com>
19829 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19831 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
19832 type for 1-element load.
19834 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
19836 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
19838 2017-04-24 Martin Jambor <mjambor@suse.cz>
19840 PR tree-optimization/80293
19841 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
19842 char arrays not totally scalarizable if it is false.
19843 (analyze_all_variable_accesses): Pass correct value in the new
19844 parameter. Add a statistics counter.
19846 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
19848 PR middle-end/79931
19849 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
19851 2017-04-24 Richard Biener <rguenther@suse.de>
19853 PR tree-optimization/80494
19854 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
19855 out for complex types.
19857 2017-04-24 Richard Biener <rguenther@suse.de>
19859 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
19860 * tree-ssa-sccvn.c (print_scc): Print SCC size.
19861 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
19862 (DFS): Adjust and never fail.
19863 (sccvn_dom_walker::fail): Remove.
19864 (sccvn_dom_walker::before_dom_children): Adjust.
19865 (run_scc_vn): Likewise and never fail.
19866 * tree-ssa-pre.c (pass_pre::execute): Adjust.
19867 (pass_fre::execute): Likewise.
19869 2017-04-24 Richard Biener <rguenther@suse.de>
19871 PR tree-optimization/79725
19872 * tree-ssa-sink.c (statement_sink_location): Return whether
19873 failure reason was zero uses. Move that check later.
19874 (sink_code_in_bb): Deal with zero uses by removing the stmt
19877 2017-04-24 Richard Biener <rguenther@suse.de>
19880 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
19881 pointer-based references.
19883 2017-04-24 Richard Biener <rguenther@suse.de>
19886 * pass_manager.h (pass_manager::operator new): Remove.
19887 (pass_manager::operator delete): Likewise.
19888 * passes.c (pass_manager::operator new): Remove.
19889 (pass_manager::operator delete): Likewise.
19890 (pass_manager::pass_manager): Zero individual pass members.
19892 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
19895 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
19896 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
19897 Check "XEXP (src, 1)" operand here.
19898 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
19899 Check "XEXP (src, 1)" operand here.
19900 (dimode_scalar_chain::make_vector_copies): Detect count register
19901 of a shift instruction. Zero extend count register from QImode
19902 to DImode to satisfy vector shift pattern count operand predicate.
19903 Substitute vector shift count operand with a DImode copy.
19904 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
19907 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
19909 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
19910 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
19911 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
19912 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
19913 (UNSPEC_NOREX_MEM): Remove definition.
19915 2017-04-21 Richard Biener <rguenther@suse.de>
19917 PR tree-optimization/79547
19918 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
19919 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
19920 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
19921 without any constraints.
19923 2017-04-21 Richard Biener <rguenther@suse.de>
19925 PR tree-optimization/78847
19926 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
19928 2017-04-21 Richard Biener <rguenther@suse.de>
19930 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
19931 (build_distinct_type_copy): Likewise.
19932 (build_variant_type_copy): Likewise.
19933 * tree.c (build_qualified_type): Pass down mem-stat info.
19934 (build_distinct_type_copy): Likewise.
19935 (build_variant_type_copy): Likewise.
19937 2017-04-21 Richard Biener <rguenther@suse.de>
19939 PR tree-optimization/80237
19940 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
19942 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
19943 for a simplified result.
19945 2016-04-21 Richard Biener <rguenther@suse.de>
19947 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
19948 sth as strict as a simple_iv but a chrec without symbols and an
19949 operand defined in the loop we are peeling (and not some subloop).
19950 (propagate_constants_for_unrolling): Propagate all constants.
19952 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
19955 * config/i386/i386.c (print_reg): Remove assert for disalowed
19956 regno values, call output_operand_lossage instead.
19958 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
19961 * config/i386/constraints.md (Yc): New register constraint.
19962 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
19963 Use Yc constraint for alternative 2 of operand 0. Remove
19964 preferred_for_speed attribute.
19966 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
19968 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
19969 lastprivate clauses in SIMT case.
19971 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
19973 * doc/invoke.texi (-Wextra-semi): Document new warning option.
19975 2017-04-20 Richard Biener <rguenther@suse.de>
19977 PR tree-optimization/57796
19978 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
19979 as N scalar stores.
19980 (vect_model_load_cost): Cost gathers as N scalar loads.
19982 2017-04-20 Richard Biener <rguenther@suse.de>
19984 * ggc-page.c (ggc_allocated_p): Rename to ...
19985 (safe_lookup_page_table_entry): ... this and return the lookup
19987 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
19989 2017-04-20 Richard Biener <rguenther@suse.de>
19991 PR tree-optimization/80453
19992 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
19993 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
19994 from the conditions.
19995 (vn_phi_eq): Pass them down.
19996 (vn_phi_lookup): Record them.
19997 (vn_phi_insert): Likewise.
19999 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
20001 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
20002 uninitialized variable warning to avoid buffer overrun.
20004 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
20007 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
20008 is suppressed for '{ 0 }' in C.
20010 2017-04-20 Jakub Jelinek <jakub@redhat.com>
20012 * BASE-VER: Set to 8.0.0.
20014 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
20016 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
20017 priority .init_array and .fini_array section with SECTION_NOTYPE
20020 2017-04-20 Jakub Jelinek <jakub@redhat.com>
20022 PR middle-end/80423
20023 * tree.h (build_array_type): Add typeless_storage default argument.
20024 * tree.c (type_cache_hasher::equal): Also compare
20025 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
20026 (build_array_type): Add typeless_storage argument, set
20027 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
20029 (build_nonshared_array_type): Adjust build_array_type_1 caller.
20030 (build_array_type): Likewise. Add typeless_storage argument.
20032 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
20033 Jakub Jelinek <jakub@redhat.com>
20035 PR tree-optimization/80426
20036 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
20037 operation on symbolic operands, also compute the overflow for the
20038 invariant part when the operation degenerates into a negation.
20040 2017-04-19 Jakub Jelinek <jakub@redhat.com>
20043 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
20044 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
20047 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
20049 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
20052 * config/avr/avr.c (tree.h): Include it.
20053 (cgraph.h): Include it.
20054 (avr_encode_section_info): Don't warn for uninitialized progmem
20055 variable if it's just an alias.
20057 2017-04-19 Richard Biener <rguenther@suse.de>
20060 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
20061 when needed by AutoPGO.
20063 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
20066 * doc/lto.texi: Remove an extra 'that'.
20068 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
20070 PR rtl-optimization/80429
20071 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
20072 are only used in debug insns.
20074 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
20075 Vladimir Makarov <vmakarov@redhat.com>
20077 * config/sparc/predicates.md (input_operand): Add comment. Return
20078 true for any memory operand when LRA is in progress.
20079 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
20081 2017-04-18 Jeff Law <law@redhat.com>
20084 * mips.md ({return,simple_return}_internal): Do not overwrite
20087 2017-04-18 Jakub Jelinek <jakub@redhat.com>
20089 PR tree-optimization/80443
20090 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
20091 instead of adding 1, subtract -1 and similarly instead of subtracting
20094 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
20096 PR rtl-optimization/80357
20097 * haifa-sched.c (tmp_bitmap): New variable.
20098 (model_recompute): Handle duplicate use records.
20099 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
20100 (free_global_sched_pressure_data): Free it.
20102 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
20105 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
20106 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
20107 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
20108 instead of SYSTEM_HEADER_DIR.
20110 2017-04-18 Jeff Law <law@redhat.com>
20112 PR middle-end/80422
20113 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
20114 predecessors after walking up the insn chain.
20116 2017-04-18 Jakub Jelinek <jakub@redhat.com>
20119 * dwarf2out.c (modified_type_die): Try harder not to emit internal
20120 sizetype type into debug info.
20122 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
20125 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
20126 unneeded test for TARGET_UPPER_REGS_SF.
20127 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
20129 2017-04-18 Jakub Jelinek <jakub@redhat.com>
20132 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
20133 instead of gsi_after_labels.
20135 2017-04-18 Jeff Law <law@redhat.com>
20137 * regcprop.c (maybe_mode_change): Avoid creating copies of the
20141 2017-04-13 Jeff Law <law@redhat.com>
20142 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
20143 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
20145 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
20148 * config/avr/avr.c (intl.h): Include it.
20149 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
20151 2017-04-18 Martin Liska <mliska@suse.cz>
20153 PR gcov-profile/78783
20154 * gcov-tool.c (gcov_output_files): Validate that destination
20155 file is either removed by the tool or by a user.
20157 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
20158 Guy Benyei <guybe@mellanox.com>
20160 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
20161 block, and do not negate it, the stored id is already negative.
20163 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
20165 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
20167 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
20170 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
20171 masks of options that should be turned off if the VSX vector
20172 options are turned off.
20173 (OTHER_P8_VECTOR_MASKS): Likewise.
20174 (OTHER_VSX_VECTOR_MASKS): Likewise.
20175 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
20176 rs6000_disable_incompatible_switches to validate no type switches
20178 (rs6000_incompatible_switch): New function to disallow turning on
20179 other vector options if -mno-vsx, -mno-power8-vector, or
20180 -mno-power9-vector are specified.
20182 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
20184 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
20186 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
20188 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
20189 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
20190 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
20191 (ARG_POINTER_CFA_OFFSET): Likewise.
20193 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
20195 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
20196 conditions to take advantage of various optimizations.
20198 2017-04-13 Jeff Law <law@redhat.com>
20200 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
20201 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
20202 (zero_extendsidi2_dext): Likewise.
20204 2017-04-13 Jakub Jelinek <jakub@redhat.com>
20207 * fold-const.c (fold_ternary_loc): Revert
20208 use op0 instead of fold_convert_loc (loc, type, arg0) part of
20211 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
20213 PR rtl-optimization/80343
20214 * lra-remat.c (update_scratch_ops): Assign original hard reg to
20215 new scratch pseudo.
20217 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
20220 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
20221 to ubsan_encode_value.
20223 2017-04-13 Jeff Law <law@redhat.com>
20225 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
20226 appearing in DEBUG_INSNs.
20228 2017-04-13 Martin Liska <mliska@suse.cz>
20230 PR gcov-profile/80413
20231 * gcov-io.c (gcov_write_string): Copy to buffer just when
20232 allocated size is greater than zero.
20234 2017-04-13 Jakub Jelinek <jakub@redhat.com>
20237 * dwarf2out.c (decls_for_scope): Ignore declarations of
20238 current_function_decl in BLOCK_NONLOCALIZED_VARS.
20240 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
20243 * ipa-visibility.c (non_local_p): Fix typos.
20244 (localize_node): When localizing symbol in same comdat group,
20245 dissolve the group only when we know external symbols are going
20247 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
20249 2017-04-12 Jakub Jelinek <jakub@redhat.com>
20251 PR tree-optimization/79390
20252 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
20253 order does not result in usable sequence, retry with reversed operand
20259 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
20260 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
20261 op0 instead of fold_convert_loc (loc, type, arg0).
20263 2017-04-12 Jeff Law <law@redhat.com>
20265 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
20266 has a delay slot in the generated code.
20268 * config/cris/cris.md (cris_preferred_reload_class): Return
20269 GENNONACR_REGS rather than GENERAL_REGS.
20271 2017-04-12 Jakub Jelinek <jakub@redhat.com>
20274 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
20275 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
20276 signedness of the result type.
20278 2017-04-12 Richard Biener <rguenther@suse.de>
20279 Jeff Law <law@redhat.com>
20281 PR tree-optimization/80359
20282 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
20283 trim stores to TARGET_MEM_REFs.
20285 2017-04-12 Richard Biener <rguenther@suse.de>
20287 PR tree-optimization/79390
20288 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
20289 threading case even more.
20291 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
20294 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
20295 for quad_address_p for TImode, instead of just not indexed_address.
20297 2017-04-12 Richard Biener <rguenther@suse.de>
20298 Bernd Edlinger <bernd.edlinger@hotmail.de>
20300 PR middle-end/79671
20301 * alias.c (component_uses_parent_alias_set_from): Handle
20302 TYPE_TYPELESS_STORAGE.
20303 (get_alias_set): Likewise.
20304 * tree-core.h (tree_type_common): Add typeless_storage flag.
20305 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
20306 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
20307 for types containing members with TYPE_TYPELESS_STORAGE.
20308 (place_field): Likewise.
20309 (layout_type): Likewise for ARRAY_TYPE.
20310 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
20311 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
20312 TYPE_TYPELESS_STORAGE.
20313 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
20315 2017-04-12 Jakub Jelinek <jakub@redhat.com>
20318 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
20319 first argument to type.
20321 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20325 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
20326 CONST0_RTX (mode) rather than const0_rtx where appropriate.
20327 (rs6000_expand_binop_builtin): Likewise.
20328 (rs6000_expand_ternop_builtin): Likewise; also add missing
20329 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
20330 vshasigma built-ins.
20331 * doc/extend.texi: Document that vec_xxpermdi's third argument
20332 must be a constant.
20334 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
20336 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
20337 Use shift_const cost parameter when calculating gain of STV shifts.
20339 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
20341 PR rtl-optimization/70478
20342 * lra-constraints.c (process_alt_operands): Check memory for
20343 disfavoring memory insn operand.
20345 2017-04-11 Jakub Jelinek <jakub@redhat.com>
20347 PR middle-end/80100
20348 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
20349 left shift in unsigned HOST_WIDE_INT type.
20351 PR rtl-optimization/80385
20352 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
20353 (not (neg X)) into (plus X -1) for complex or non-integral modes.
20356 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
20357 if they have any depend clauses.
20359 2017-04-11 Martin Liska <mliska@suse.cz>
20362 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
20363 * ipa-split.c (split_function): Create a local comdat symbol
20364 if caller is in a comdat group.
20366 2017-04-11 Martin Liska <mliska@suse.cz>
20369 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
20372 2017-04-11 Martin Sebor <msebor@redhat.com>
20374 PR middle-end/80364
20375 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
20376 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
20378 (directive::set_width, directive::set_precision, format_character):
20380 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
20383 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
20386 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
20387 conflict, set target->arch_name instead of target->cpu_name.
20389 2017-04-11 Richard Biener <rguenther@suse.de>
20391 PR tree-optimization/80374
20392 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
20393 build_zero_cst, remove fold_convertible_p check again.
20395 2017-04-11 Martin Liska <mliska@suse.cz>
20398 * ubsan.c (instrument_object_size): Do not instrument register
20401 2017-04-11 Jakub Jelinek <jakub@redhat.com>
20404 * config/i386/i386-builtin-types.def
20405 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
20406 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
20407 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
20408 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
20409 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
20410 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
20411 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
20412 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
20413 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
20414 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
20415 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
20416 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
20417 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
20418 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
20419 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
20420 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
20421 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
20422 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
20423 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
20424 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
20425 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
20426 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
20427 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
20428 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
20429 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
20430 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
20431 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
20432 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
20433 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
20434 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
20435 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
20436 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
20437 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
20438 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
20439 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
20440 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
20441 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
20442 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
20443 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
20444 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
20445 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
20446 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
20447 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
20448 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
20449 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
20450 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
20452 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
20453 flag to second_arg_count, handle 4 argument function type _COUNT
20454 aliases, handle second_arg_count on second argument rather than last.
20456 2017-04-10 Jeff Law <law@redhat.com>
20458 PR tree-optimization/80374
20459 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
20460 record anything if we can not convert integer_zero_node to the
20463 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
20466 * config/rs6000/rs6000.c (rs6000_option_override_internal):
20467 Enhance special handling given to the TARGET_P9_MINMAX option in
20468 relation to certain other options.
20470 2017-04-10 Bin Cheng <bin.cheng@arm.com>
20472 PR tree-optimization/80153
20473 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
20474 remove POINTER_PLUS_EXPR's base part directly, rather than through
20477 2017-04-10 Richard Biener <rguenther@suse.de>
20478 Bin Cheng <bin.cheng@arm.com>
20480 PR tree-optimization/80153
20481 * tree-affine.c (aff_combination_to_tree): Get base pointer from
20482 the first element of pointer type aff_tree. Build result expr in
20484 (add_elt_to_tree): Convert to type unconditionally. Remove other
20485 fold_convert calls.
20486 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
20487 (rewrite_use_nonlinear_expr): Check invariant using iv information.
20489 2017-04-10 Richard Biener <rguenther@suse.de>
20491 * tree-ssa-structalias.c (find_func_aliases): Properly handle
20494 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
20496 PR rtl-optimization/70478
20497 * lra-constraints.c (curr_small_class_check): New.
20498 (update_and_check_small_class_inputs): New.
20499 (process_alt_operands): Update curr_small_class_check. Disfavor
20500 alternative insn memory operands. Check available regs for small
20503 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
20506 * config/mips/mips.opt (-mvirt): Update description.
20507 * doc/invoke.texi (-mvirt): Likewise.
20509 2017-04-10 Richard Biener <rguenther@suse.de>
20511 PR middle-end/80362
20512 * fold-const.c (fold_binary_loc): Look at unstripped ops when
20513 looking for NEGATE_EXPR in -A / -B to A / B folding.
20515 2017-04-10 Martin Liska <mliska@suse.cz>
20517 PR gcov-profile/80224
20518 * gcov.c (print_usage): Fix usage string.
20519 (get_gcov_intermediate_filename): Remove.
20520 (output_gcov_file): Use both for normal and intermediate format.
20521 (generate_results): Do not initialize special file for
20522 intermediate format.
20524 2017-04-10 Richard Biener <rguenther@suse.de>
20526 PR tree-optimization/80304
20527 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
20530 2017-04-10 Nathan Sidwell <nathan@acm.org>
20533 * config/rs6000/rs6000.c (rs6000_vector_type): New.
20534 (rs6000_init_builtins): Use it.
20536 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20538 * config/arm/arm.md (<mrc>): Add mode to SET source.
20539 (<mrrc>): Likewise.
20541 2017-04-10 Richard Biener <rguenther@suse.de>
20543 PR middle-end/80344
20544 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
20546 2017-04-10 Jakub Jelinek <jakub@redhat.com>
20549 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
20550 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
20551 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
20552 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
20553 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
20554 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
20555 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
20556 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
20557 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
20558 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
20559 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
20560 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
20561 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
20562 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
20563 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
20564 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
20565 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
20566 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
20567 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
20568 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
20569 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
20570 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
20571 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
20573 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
20575 PR rtl-optimization/70478
20576 * lra-constraints.c: Reverse the last patch.
20578 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
20580 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
20581 Add comment for WCHAR_T.
20583 2017-04-08 Martin Liska <mliska@suse.cz>
20586 2017-04-07 Martin Liska <mliska@suse.cz>
20589 * ipa-split.c (split_function): Add function part to a same comdat
20592 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20595 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
20597 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
20599 * rs6000/rs6000.c (vec_load_pendulum): Rename...
20600 (vec_pairing): ...to this.
20601 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
20602 (rs6000_sched_init): Adjust for name change.
20603 (struct rs6000_sched_context): Likewise.
20604 (rs6000_init_sched_context): Likewise.
20605 (rs6000_set_sched_context): Likewise.
20607 2017-04-07 Jakub Jelinek <jakub@redhat.com>
20613 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
20615 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
20616 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
20617 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
20619 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
20621 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
20623 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
20625 PR rtl-optimization/70703
20626 * ira-color.c (update_conflict_hard_regno_costs): Use
20627 int64_t instead of HOST_WIDE_INT.
20629 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
20631 PR rtl-optimization/70478
20632 * lra-constraints.c (process_alt_operands): Disfavor alternative
20633 insn memory operands.
20635 2017-04-07 Jeff Law <law@redhat.com>
20637 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
20638 CALL and NOTE_INSN_CALL_ARG_LOCATION.
20640 2017-04-07 Martin Liska <mliska@suse.cz>
20643 * config/aarch64/aarch64.c (aarch64_process_target_attr):
20644 Show error message instead of an ICE.
20646 2017-04-07 Martin Liska <mliska@suse.cz>
20649 * ipa-split.c (split_function): Add function part to a same comdat
20652 2017-04-07 Richard Biener <rguenther@suse.de>
20654 PR middle-end/80341
20655 * tree.c (get_unwidened): Also handle ! for_type case for
20657 * convert.c (do_narrow): Split out from ...
20658 (convert_to_integer_1): ... here. Do not pass final truncation
20659 type to get_unwidened for TRUNC_DIV_EXPR.
20661 2017-04-07 Richard Biener <rguenther@suse.de>
20663 * tree-affine.c (wide_int_ext_for_comb): Take type rather
20665 (aff_combination_const): Adjust.
20666 (aff_combination_scale): Likewise.
20667 (aff_combination_add_elt): Likewise.
20668 (aff_combination_add_cst): Likewise.
20669 (aff_combination_convert): Likewise.
20670 (add_elt_to_tree): Likewise. Remove unused argument.
20671 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
20673 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
20675 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
20677 * config/arm/arm.c (arm_default_short_enums): Use
20678 ARM_DEFAULT_SHORT_ENUMS.
20679 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
20681 2017-04-06 Jakub Jelinek <jakub@redhat.com>
20684 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
20685 members with redundant out-of-class redeclaration.
20687 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
20690 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
20691 * config/i386/i386.md (*zero_extendsidi2):
20692 Add (?*x,*x) and (?*v,*v) alternatives.
20694 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
20697 * config/i386/i386.c (ix86_expand_builtin)
20698 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
20699 mode from insn data. Convert operands to insn operand mode.
20700 Copy operands that don't satisfy insn predicate to a register.
20702 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
20704 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
20707 2017-04-06 Richard Biener <rguenther@suse.de>
20709 PR tree-optimization/80334
20710 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
20711 preserve alignment of accesses.
20713 2017-04-06 Richard Biener <rguenther@suse.de>
20715 PR tree-optimization/80262
20716 * tree-sra.c (build_ref_for_offset): Preserve address-space
20718 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
20719 Drop useless address-space information on MEM_REF offsets.
20721 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
20723 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
20725 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
20727 PR rtl-optimization/70703
20728 * ira-color.c (update_conflict_hard_regno_costs): Use
20729 HOST_WIDE_INT instead of long.
20731 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
20734 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
20735 not defined for x86_64 target. Add -mmmx target option when __SSE2__
20737 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
20738 for x86_64 target. Handle -m3dnowa option.
20740 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
20742 PR rtl-optimization/70703
20743 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
20744 (update_conflict_hard_regno_costs): Use long instead of unsigned
20745 arithmetic for cost calculation.
20747 2017-04-05 Jakub Jelinek <jakub@redhat.com>
20748 Bernd Edlinger <bernd.edlinger@hotmail.de>
20751 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
20754 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
20757 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
20758 ptr_mode with Pmode throughout.
20759 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
20760 into probe_stack_range and use DImode.
20762 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
20765 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
20766 call_eh_return is true.
20768 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20770 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
20771 Initialize last_match_fntype_index.
20773 2017-04-05 Jakub Jelinek <jakub@redhat.com>
20776 * tree-nvr.c: Include internal-fn.h.
20777 (pass_return_slot::execute): Ignore internal calls without
20780 2017-04-04 Jakub Jelinek <jakub@redhat.com>
20781 Richard Biener <rguenther@suse.de>
20784 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
20785 captures used multiple times, except for the last use.
20786 * generic-match-head.c: Include gimplify.h.
20788 2017-04-04 Jakub Jelinek <jakub@redhat.com>
20790 PR tree-optimization/79390
20791 * target.h (struct noce_if_info): Declare.
20792 * targhooks.h (default_noce_conversion_profitable_p): Declare.
20793 * target.def (noce_conversion_profitable_p): New target hook.
20794 * ifcvt.h (struct noce_if_info): New type, moved from ...
20795 * ifcvt.c (struct noce_if_info): ... here.
20796 (noce_conversion_profitable_p): Renamed to ...
20797 (default_noce_conversion_profitable_p): ... this. No longer
20799 (noce_try_store_flag_constants, noce_try_addcc,
20800 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
20801 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
20802 instead of noce_conversion_profitable_p.
20803 * config/i386/i386.c: Include ifcvt.h.
20804 (ix86_option_override_internal): Don't override
20805 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
20806 (ix86_noce_conversion_profitable_p): New function.
20807 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
20808 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
20809 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
20810 * doc/tm.texi: Regenerated.
20812 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20814 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
20817 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
20820 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
20821 instructions for small multiply cores.
20823 2017-04-04 Jeff Law <law@redhat.com>
20825 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
20827 (mips_expand_vec_perm_const): Initialize elements in orig_perm
20828 that are not set by the loop over the elements.
20830 2017-04-04 Jakub Jelinek <jakub@redhat.com>
20833 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
20834 int mode, convert_modes it to mode as unsigned, otherwise use
20835 lowpart_subreg to mode rather than SImode.
20836 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
20837 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
20838 Use DImode instead of SImode for the shift count operand.
20839 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
20842 2017-04-04 Richard Biener <rguenther@suse.de>
20844 PR middle-end/80281
20845 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
20846 arithmetic done for the negate or the plus. Simplify.
20847 (A - (-B) -> A + B): Likewise.
20848 * fold-const.c (split_tree): Make sure to not negate pointers.
20850 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
20852 PR rtl-optimization/60818
20853 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
20854 a compare of comparisons with the thing compared if this results
20855 in a different machine mode.
20857 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
20859 * alias.c (base_alias_check): Fix typo in comment.
20860 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
20861 * cgraphunit.c (symbol_table::compile): Likewise.
20862 * collect2.c (maybe_run_lto_and_relink): Likewise.
20863 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
20864 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
20865 * config/avr/avr.c (avr_map_op_t): Likewise.
20866 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
20867 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
20868 * config/epiphany/epiphany.md (movcc): Likewise.
20869 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
20870 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
20872 * config/mips/mips.c (mips_save_restore_reg): Likewise.
20873 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
20874 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
20875 * config/sh/sh.c (sh_rtx_costs): Likewise.
20876 * fold-const.c (fold_truth_andor): Likewise.
20877 * genautomata.c (collapse_flag): Likewise.
20878 * gengtype.h (struct type::u::s): Likewise.
20879 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
20880 * input.c (FORMAT_AMOUNT): Likewise.
20881 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
20882 (known_aggs_to_agg_replacement_list): Likewise.
20883 * ipa-inline-analysis.c: Likewise.
20884 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
20885 * ipa-polymorphic-call.c
20886 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
20887 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
20888 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
20890 * modulo-sched.c (apply_reg_moves): Likewise.
20891 * omp-expand.c (build_omp_regions_1): Likewise.
20892 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
20893 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
20894 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
20895 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
20896 * value-prof.c: Likewise.
20897 * var-tracking.c (val_reset): Likewise.
20899 2017-04-03 Richard Biener <rguenther@suse.de>
20901 PR tree-optimization/80275
20902 * fold-const.c (split_address_to_core_and_offset): Handle
20905 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
20907 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
20908 descriptors is at least equal to that of functions.
20910 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
20912 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
20914 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
20917 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
20918 (mov<IMOD4:mode>): New expander.
20919 (*mov<IMOD4:mode>_internal): New insn and split pattern.
20921 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
20923 PR rtl-optimization/79405
20924 * fwprop.c (propagations_left): New variable.
20925 (forward_propagate_into): Decrement it.
20926 (fwprop_init): Initialize it.
20927 (fw_prop): If the variable has reached zero, stop propagating.
20928 (fwprop_addr): Ditto.
20930 2017-03-31 Jakub Jelinek <jakub@redhat.com>
20933 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
20934 a FUNCTION_DECL, pass it as decl instead of origin to
20937 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
20939 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
20942 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
20945 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
20946 TARGET_VSX_SMALL_INTEGER.
20948 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20950 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
20951 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
20953 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
20955 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
20956 extraction from odd-numbered MSA register.
20958 2017-03-31 Jakub Jelinek <jakub@redhat.com>
20960 PR middle-end/80173
20961 * expmed.c (store_bit_field_1): Don't attempt to create
20962 a word subreg out of hard registers wider than word if they
20963 have HARD_REGNO_NREGS of 1 for their mode.
20965 PR middle-end/80163
20966 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
20967 conversions to integer types wider than word and pointer.
20970 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
20971 (rtx_equal_for_cselib_p): Pass 0 to it.
20972 * cselib.c (cselib_hasher::equal): Likewise.
20973 (rtx_equal_for_cselib_1): Add depth argument. If depth
20974 is 128, don't look up VALUE locs and punt. Increment
20975 depth in recursive calls when walking VALUE locs.
20977 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
20979 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
20980 (make_gcov_file_name): Use the canonical path name for generating
20982 (read_line): Fix handling of files with ascii null bytes.
20984 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
20986 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
20987 to initialise a vector register instead
20988 of using a const_int.
20990 2017-03-30 Jakub Jelinek <jakub@redhat.com>
20992 PR translation/80189
20993 * gimplify.c (omp_default_clause): Use %qs instead of %s in
20994 diagnostic messages.
20996 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
20999 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
21000 (dfp_diex_<mode>): Update mode of operand 1.
21001 * doc/extend.texi (dxex, dxexq): Document change to return type.
21002 (diex, diexq): Document change to argument type.
21004 2017-03-30 Martin Jambor <mjambor@suse.cz>
21007 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
21008 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
21009 it reflects the signature changes performed at the callee side.
21010 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
21011 to cgraph_build_function_type_skip_args.
21012 (build_function_decl_skip_args): Adjust call to the above function.
21014 2017-03-30 Jakub Jelinek <jakub@redhat.com>
21017 * config/i386/sse.md
21018 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
21019 register as dest whenever it is a MEM not rtx_equal_p to the
21020 corresponding dup operand, and when forcing into reg move the
21021 reg into the memory afterwards.
21022 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
21023 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
21024 for the force_reg mode.
21025 (avx512vl_vextractf128<mode>): Use register as dest either
21026 always when a MEM, or when it is a MEM not rtx_equal_p to the
21027 corresponding dup operand, or even not when it is a CONST_VECTOR
21028 depending on the mode and lo vs. hi.
21029 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
21031 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
21032 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
21033 Likewise. Require that operands[2] is even.
21034 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
21035 Remove extraneous parens. Require that operands[2] is a multiple
21037 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
21038 operands[0] is a MEM if <mask_applied>, the predicates/constraints
21039 disallow memory then.
21041 2017-03-30 Richard Biener <rguenther@suse.de>
21043 PR tree-optimization/77498
21044 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
21045 to non-constants over backedges.
21047 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
21049 PR rtl-optimization/80233
21050 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
21051 as last_combined_insn. Do not test for BARRIER_P separately.
21053 2017-03-29 Andreas Schwab <schwab@suse.de>
21056 * calls.c (prepare_call_address): Convert funexp to Pmode before
21057 copying to temp reg.
21059 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21061 PR tree-optimization/80158
21062 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
21063 Handle possible future case of more than one alternate
21065 (replace_rhs_if_not_dup): Likewise.
21066 (replace_one_candidate): Likewise.
21068 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
21070 PR rtl-optimization/80193
21071 * ira.c (ira): Do not check allocation for LRA.
21073 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
21075 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
21076 (nvptx_output_simt_exit): Declare.
21077 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
21078 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
21079 (init_softstack_frame): Move initialization of crtl->is_leaf to...
21080 (nvptx_declare_function_name): ...here. Emit declaration of local
21081 memory space buffer for omp_simt_enter insn.
21082 (nvptx_output_unisimt_switch): New.
21083 (nvptx_output_softstack_switch): New.
21084 (nvptx_output_simt_enter): New.
21085 (nvptx_output_simt_exit): New.
21086 * config/nvptx/nvptx.h (struct machine_function): New fields
21087 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
21088 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
21089 (UNSPECV_SIMT_EXIT): Ditto.
21090 (omp_simt_enter_insn): New insn.
21091 (omp_simt_enter): New expansion.
21092 (omp_simt_exit): New insn.
21093 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
21095 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
21096 (expand_GOMP_SIMT_ENTER_ALLOC): New.
21097 (expand_GOMP_SIMT_EXIT): New.
21098 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
21099 (GOMP_SIMT_ENTER_ALLOC): Ditto.
21100 (GOMP_SIMT_EXIT): Ditto.
21101 * target-insns.def (omp_simt_enter): New insn.
21102 (omp_simt_exit): Ditto.
21103 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
21105 (lower_rec_simd_input_clauses): Implement SIMT privatization.
21106 (lower_rec_input_clauses): Likewise.
21107 (lower_lastprivate_clauses): Handle SIMT privatization.
21109 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
21110 (ompdevlow_adjust_simt_enter): New.
21111 (find_simtpriv_var_op): New.
21112 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
21113 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
21115 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
21116 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
21117 (copy_decl_for_dup_finish): Ditto.
21119 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
21121 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
21124 * config/i386/i386.c (ix86_option_override_internal): Always
21125 allow -mpreferred-stack-boundary=3 for 64-bit targets.
21127 2017-03-28 Bin Cheng <bin.cheng@arm.com>
21129 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
21131 2017-03-28 Bin Cheng <bin.cheng@arm.com>
21133 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
21134 mark new edge's irreducible flag accordign to it.
21135 (vect_do_peeling): Check loop preheader edge's irreducible flag
21136 and pass it to function slpeel_add_loop_guard.
21138 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
21140 PR tree-optimization/80218
21141 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
21142 Update block frequencies and counts.
21144 2017-03-28 Richard Biener <rguenther@suse.de>
21146 PR tree-optimization/78644
21147 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
21148 of a simplification result we may not use it at all.
21150 2017-03-28 Richard Biener <rguenther@suse.de>
21153 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
21154 without arguments, generate default definition of a SSA name.
21156 2017-03-28 Richard Biener <rguenther@suse.de>
21158 PR middle-end/80222
21159 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
21160 TYPE_REF_CAN_ALIAS_ALL references.
21161 * fold-const.c (fold_indirect_ref_1): Likewise.
21163 2017-03-28 Martin Liska <mliska@suse.cz>
21166 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
21167 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
21169 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
21170 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
21172 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
21173 (EXTRA_SPECS): Define.
21174 (SUBTARGET_EXTRA_SPECS): Likewise.
21175 (SUBTARGET_CPP_SPEC): Likewise.
21176 * config/arc/elf.h (EXTRA_SPECS): Renamed to
21177 SUBTARGET_EXTRA_SPECS.
21178 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
21180 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
21182 * config/arc/simdext.md (vst64_insn): Update pattern.
21183 (vld32wh_insn): Likewise.
21184 (vld32wl_insn): Likewise.
21185 (vld64_insn): Likewise.
21186 (vld32_insn): Likewise.
21188 2017-03-28 Marek Polacek <polacek@redhat.com>
21191 * fold-const.c (fold_comparison): Use protected_set_expr_location
21192 instead of SET_EXPR_LOCATION.
21194 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
21196 * tree.c (add_expr): Avoid name lookup warning.
21198 2017-03-27 Jeff Law <law@redhat.com>
21200 PR tree-optimization/80216
21201 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
21202 function name. Limit recursion depth.
21203 (record_temporary_equivalences): Corresponding changes.
21205 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
21207 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
21210 2017-03-27 Jakub Jelinek <jakub@redhat.com>
21213 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
21215 * cfgcleanup.c (reg_note_cfa_p): New array.
21216 (insns_have_identical_cfa_notes): New function.
21217 (old_insns_match_p): Don't cross-jump in between /f
21218 and non-/f instructions. If both i1 and i2 are frame related,
21219 verify all CFA notes, their order and content.
21221 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
21224 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
21225 HImode and SImode with zero extend to DImode to one insn.
21226 (bswap<mode>2_extenddi): Likewise.
21227 (bswapsi2_extenddi): Likewise.
21228 (bswaphi2_extendsi): Likewise.
21229 (bswaphi2): Combine bswap HImode and SImode into one insn.
21230 Separate memory insns from swapping register.
21231 (bswapsi2): Likewise.
21232 (bswap<mode>2): Likewise.
21233 (bswaphi2_internal): Delete, no longer used.
21234 (bswapsi2_internal): Likewise.
21235 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
21236 store, and gpr<-gpr swap insns.
21237 (bswap<mode>2_store): Likewise.
21238 (bswaphi2_reg): Register only splitter, combine with the splitter.
21239 (bswaphi2 splitter): Likewise.
21240 (bswapsi2_reg): Likewise.
21241 (bswapsi2 splitter): Likewise.
21242 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
21243 the insns into load, store, and register/register insns.
21244 (bswapdi2_ldbrx): Likewise.
21245 (bswapdi2_load): Likewise.
21246 (bswapdi2_store): Likewise.
21247 (bswapdi2_reg): Likewise.
21249 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
21251 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
21252 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
21254 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
21257 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
21259 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
21260 special handling for target option conflicts between dform
21261 options (-mpower9-dform, -mpower9-dform-vector,
21262 -mpower9-dform-scalar) and -mno-direct-move.
21264 2017-03-27 Richard Biener <rguenther@suse.de>
21266 PR tree-optimization/80181
21267 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
21269 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
21271 * config/arc/predicates.md (move_double_src_operand): Replace the
21272 call to move_double_src_operand with a call to address_operand.
21274 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
21276 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
21277 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
21278 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
21280 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
21282 * config/arc/predicates.md (long_immediate_loadstore_operand):
21283 Consider scaled addresses cases.
21285 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
21287 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
21288 restored when in interrupt.
21289 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
21290 doesn't have delay slot.
21292 2017-03-27 Richard Biener <rguenther@suse.de>
21295 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
21296 inlined thunk clones.
21298 2017-03-27 Jakub Jelinek <jakub@redhat.com>
21301 * asan.c (instrument_derefs): Copy over last operand from
21302 original COMPONENT_REF to the new COMPONENT_REF with
21303 DECL_BIT_FIELD_REPRESENTATIVE.
21304 * ubsan.c (instrument_object_size): Likewise.
21306 2017-03-27 Richard Biener <rguenther@suse.de>
21308 PR tree-optimization/80170
21309 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
21310 sure DR/SCEV didnt fold in constants we do not see when looking
21311 at the reference base alignment.
21313 2017-03-27 Richard Biener <rguenther@suse.de>
21315 PR middle-end/80171
21316 * gimple-fold.c (fold_ctor_reference): Properly guard against
21317 NULL return value from canonicalize_constructor_val.
21319 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
21322 * config/i386/i386.c (ix86_expand_builtin)
21323 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
21324 flags reg setting and flags reg using instructions.
21325 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
21326 clobbering instructions to zero extend op2.
21328 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
21330 * doc/install.texi (Configuration) <--with-aix-soname>:
21331 Update link to AIX ld.
21333 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
21335 PR rtl-optimization/80160
21336 PR rtl-optimization/80159
21337 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
21338 reg_alternate_class into account.
21340 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
21343 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
21344 to consider in curr_insn_transform.
21346 2017-03-24 Jakub Jelinek <jakub@redhat.com>
21348 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
21349 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
21350 and emit_mode_inner.
21352 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21354 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
21355 argument to the overloaded builtin variants. Use the new flag to
21356 deprecate certain builtin variants.
21357 * config/s390/s390-builtin-types.def: Add new builtin types.
21358 * config/s390/s390-builtins.h: Support new flags field for
21359 overloaded builtins.
21360 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
21361 (s390_macro_to_expand): Enable vector float data type.
21362 (s390_cpu_cpp_builtins_internal): Indicate support of the new
21363 builtins by incrementing the __VEC__ version number.
21364 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
21366 (s390_resolve_overloaded_builtin): Emit error messages depending
21367 on the builtin flags.
21368 * config/s390/s390.c (s390_expand_builtin): Support additional
21369 flags argument. Change error message to match the messages
21370 emitted in s390-c.c.
21371 * config/s390/s390.md: New UNSPEC_* constants.
21372 (op_type): Add new instruction types.
21373 * config/s390/vecintrin.h: Add new builtins and test data class
21375 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
21376 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
21377 (VEC_INEXACT, VEC_NOINEXACT): New constants.
21378 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
21379 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
21380 ("vec_mergel<mode>"): V_HW -> VEC_HW.
21382 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
21383 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
21384 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
21385 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
21387 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
21388 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
21389 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
21390 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
21392 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
21393 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
21394 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
21395 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
21396 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
21397 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
21398 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
21400 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
21401 ("vec_scatter_element<V_HW_4:mode>_DI")
21402 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
21403 ("vec_fpint<mode>", "vflls")
21404 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
21405 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
21406 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
21407 ("*vec_cmphe<mode>_cc"): ... these.
21409 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
21410 mode constant instead of magic value.
21412 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21414 * config/s390/s390.c (s390_expand_vec_compare): Support other
21415 vector floating point modes than just V2DF.
21416 (s390_expand_vcond): Likewise.
21417 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
21418 (s390_cannot_change_mode_class): Prevent mode changes between TF
21419 and V1TF in vector registers.
21420 * config/s390/s390.md (DF, SF): New mode attributes.
21421 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
21422 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
21423 SFmode support for VRs.
21424 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
21426 (VFT, VF_HW): New mode iterators.
21427 (vw, sdx): New mode attributes.
21428 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
21429 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
21430 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
21431 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
21432 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
21433 also the new vector floating point modes. Renaming to ...
21435 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
21436 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
21437 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
21438 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
21439 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
21440 ("vec_unordered<mode>"): ... these.
21442 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
21443 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
21444 ("*vec_extendv2df"): New insn definitions.
21446 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21448 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
21449 ("mulditi3_2", "*muldi3_sign"): New patterns.
21450 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
21451 rename the pattern definition.
21453 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21455 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
21457 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
21459 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21461 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
21462 instruction if possible.
21463 * config/s390/vector.md (vec_halfnumelts): New mode
21465 ("*vec_vllezlf<mode>"): New pattern.
21467 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21469 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
21470 ("popcountv4si2", "popcountv2di2"): Rename to ...
21471 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
21472 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
21474 ("popcount<mode>2_vxe"): New pattern.
21476 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21478 * common/config/s390/s390-common.c (processor_flags_table): Add
21480 * config.gcc: Add arch12.
21481 * config/s390/driver-native.c (s390_host_detect_local_cpu):
21482 Default to arch12 for unknown CPU model numbers.
21483 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
21484 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
21485 PROCESSOR_max sanity check.
21486 * config/s390/s390-opts.h (enum processor_type): Add
21488 * config/s390/s390.c (processor_table): Add arch12.
21489 (s390_expand_builtin): Add check for B_VXE flag.
21490 (s390_issue_rate): Add PROCESSOR_ARCH12.
21491 (s390_get_sched_attrmask): Likewise.
21492 (s390_get_unit_mask): Likewise.
21493 (s390_sched_score): Enable z13 scheduling for arch12.
21494 (s390_sched_reorder): Likewise.
21495 (s390_sched_variable_issue): Likewise.
21496 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
21498 (s390_tune_attr): Use z13 scheduling also for arch12.
21499 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
21500 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
21501 (TARGET_VXE_P): New macros.
21502 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
21503 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
21504 * config/s390/s390.opt: Add arch12 as processor_type.
21506 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21508 * config/s390/s390.md
21509 ("fixuns_truncdddi2", "fixuns_trunctddi2")
21510 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
21511 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
21513 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
21514 Rename expanders to ...
21516 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
21517 ("fixuns_truncdddi2_emu"): ... these.
21519 ("fixuns_trunc<mode>si2_emu"): New expander.
21521 ("*fixuns_truncdfdi2_z13"): Rename to ...
21522 ("*fixuns_truncdfdi2_vx"): ... this.
21524 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21526 * config/s390/2964.md: Remove the single element vector compare
21527 instructions which are no longer used.
21528 * config/s390/s390.c (s390_select_ccmode): Remove handling of
21530 (s390_canonicalize_comparison): Remove handling of DFmode
21532 (s390_expand_vec_compare_scalar): Remove function.
21533 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
21534 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
21536 ("*cmp<mode>_ccs"): Add wfcdb instruction.
21538 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21540 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
21542 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
21543 will anyway by matched by mov<mode>_64dfp.
21545 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21547 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
21548 vlef/vstef. Add missing operand to vleif.
21550 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21552 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
21553 pair for all vector types with 64 bit elements.
21554 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
21555 * config/s390/vector.md (V_HW_64): ... here.
21556 (V_128_NOSINGLE): New mode iterator.
21557 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
21558 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
21559 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
21560 ("*vec_load_pairv2di"): Change to ...
21561 ("*vec_load_pair<mode>"): ... this one.
21563 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21565 * config/s390/constraints.md: Add comments.
21566 (jKK): Reject element sizes > 8 bytes.
21567 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
21569 * config/s390/s390.md: Add the s_operand checks formerly in
21570 s390_split_ok_p to various splitters where they are still
21572 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
21573 for 128 bit vectors. Plus two splitters.
21575 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21577 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
21580 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21583 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
21584 error if the boundary argument is not constant.
21586 2017-03-24 Jakub Jelinek <jakub@redhat.com>
21588 PR rtl-optimization/80112
21589 * loop-doloop.c (doloop_condition_get): Don't check condition
21590 if cmp isn't SET with IF_THEN_ELSE src.
21592 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21594 PR tree-optimization/80158
21595 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
21596 replacing a candidate statement, also replace it for the
21597 candidate's alternate interpretation.
21598 (replace_rhs_if_not_dup): Likewise.
21599 (replace_one_candidate): Likewise.
21601 2017-03-24 Richard Biener <rguenther@suse.de>
21603 PR tree-optimization/80167
21604 * graphite-isl-ast-to-gimple.c
21605 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
21607 (translate_isl_ast_to_gimple::get_rename): Likewise.
21609 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
21611 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
21612 handling of certain combinations of target options, including the
21613 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
21614 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
21616 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21619 * config/arm/arm.md (*load_multiple): Add reload_completed to
21620 matching condition.
21622 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21623 Richard Biener <rguenth@suse.de>
21625 PR tree-optimization/79908
21626 PR tree-optimization/80136
21627 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
21628 been cast away, gimplify_and_add suffices.
21630 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
21632 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
21634 2017-03-23 Richard Biener <rguenther@suse.de>
21636 PR tree-optimization/80032
21637 * gimplify.c (gimple_push_cleanup): Forced unconditional
21638 cleanups still have to go to the conditional_cleanups
21641 2017-03-22 Jakub Jelinek <jakub@redhat.com>
21643 PR tree-optimization/80072
21644 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
21646 (next_operand_entry_id): Change type to unsigned int.
21647 (sort_by_operand_rank): Make sure to return the right return value
21648 even if unsigned fields are bigger than INT_MAX.
21649 (struct oecount): Change cnt and id type to unsigned int.
21650 (oecount_hasher::equal): Formatting fix.
21651 (oecount_cmp): Make sure to return the right return value
21652 even if unsigned fields are bigger than INT_MAX.
21653 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
21656 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
21657 TREE_READONLY on result if writing it more than once.
21660 * doc/invoke.texi (-fsanitize=thread): Document that with
21661 -fnon-call-exceptions atomics are not able to throw
21665 * tsan.c: Include tree-eh.h.
21666 (instrument_builtin_call): Call maybe_clean_eh_stmt or
21667 maybe_clean_or_replace_eh_stmt where needed.
21668 (instrument_memory_accesses): Add cfg_changed argument.
21669 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
21670 if it returned true.
21671 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
21673 PR rtl-optimization/63191
21674 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
21675 wrapper function, moved the whole old content into ...
21676 (ix86_delegitimize_address_1): ... this. New inline function.
21677 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
21678 true as last argument instead of ix86_delegitimize_address.
21680 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
21682 * config/aarch64/aarch64.c (generic_branch_cost): Copy
21683 cortexa57_branch_cost.
21685 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
21687 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
21689 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21692 * doc/md.texi (Constraints): Document wA constraint.
21693 * config/rs6000/constraints.md (wA): New.
21694 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
21695 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
21696 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
21697 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
21699 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
21702 * gimplify.c (is_oacc_declared): New function.
21703 (oacc_default_clause): Use it to set default flags for acc declared
21704 variables inside parallel regions.
21705 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
21706 declared variables.
21707 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
21708 declare attribute to any decl as necessary.
21710 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
21713 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
21714 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
21715 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
21716 (arm_arch_lpae): This.
21717 * config/arm/arm.c (arm_arch7ve): Rename into ...
21718 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
21719 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
21722 2017-03-22 Martin Liska <mliska@suse.cz>
21725 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
21726 error message instead of an ICE.
21728 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21730 * doc/extend.texi (6.11 Additional Floating Types): Revise.
21732 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
21734 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
21736 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
21739 2017-03-21 Martin Sebor <msebor@redhat.com>
21741 * doc/extend.texi: Use "cannot" instead of "can't."
21742 * doc/hostconfig.texi: Same.
21743 * doc/install.texi: Same.
21744 * doc/invoke.texi: Same.
21745 * doc/loop.texi: Same.
21746 * doc/md.texi: Same.
21747 * doc/objc.texi: Same.
21748 * doc/rtl.texi: Same.
21749 * doc/tm.texi: Same.
21750 * doc/tm.texi.in: Same.
21751 * doc/trouble.texi: Same.
21753 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
21756 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
21757 (collect_checksum_attributes): Set it.
21758 (die_checksum_ordered): Use it.
21760 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21762 PR tree-optimization/79908
21763 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
21764 change: For a VA_ARG whose LHS has been cast away, use
21765 force_gimple_operand to construct the side effects.
21767 2017-03-21 David Malcolm <dmalcolm@redhat.com>
21769 PR translation/80001
21770 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
21771 more amenable to translation.
21772 (oacc_loop_auto_partitions): Likewise.
21774 2017-03-21 Marek Polacek <polacek@redhat.com>
21775 Martin Sebor <msebor@redhat.com>
21777 PR tree-optimization/80109
21778 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
21779 on INTEGRAL_TYPE_P.
21781 2017-03-21 Jakub Jelinek <jakub@redhat.com>
21782 Segher Boessenkool <segher@kernel.crashing.org>
21785 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
21786 check reg_used_between_p between insn and one of succ or succ2
21787 depending on if succ is artificial insn not inserted into insn
21790 2017-03-21 Martin Liska <mliska@suse.cz>
21792 PR gcov-profile/80081
21793 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
21794 * doc/gcc.texi: Include gcov-dump stuff.
21795 * doc/gcov-dump.texi: New file.
21797 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
21799 PR rtl-optimization/79150
21800 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
21801 conditional jump, if the jump is the last insn of the loop.
21803 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21804 Richard Biener <rguenth@suse.de>
21806 PR tree-optimization/79908
21807 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
21808 been cast away, use force_gimple_operand to construct the side
21811 2017-03-21 Martin Liska <mliska@suse.cz>
21813 PR libfortran/79956
21814 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
21817 2017-03-21 Brad Spengler <spender@grsecurity.net>
21820 * plugin.c (htab_hash_plugin): New function.
21821 (add_new_plugin): Use it and adjust.
21822 (parse_plugin_arg_opt): Adjust.
21823 (init_one_plugin): Likewise.
21825 2017-03-21 Richard Biener <rguenther@suse.de>
21827 PR tree-optimization/80032
21828 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
21829 if set force the cleanup to happen unconditionally.
21830 (gimplify_target_expr): Push inserted clobbers with force_uncond
21831 to avoid them being removed by control-dependent DCE.
21833 2017-03-21 Richard Biener <rguenther@suse.de>
21835 PR tree-optimization/80122
21836 * tree-inline.c (copy_bb): Do not expans va-arg packs or
21837 va_arg_pack_len when the inlined call stmt requires pack
21839 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
21841 2017-03-21 Jakub Jelinek <jakub@redhat.com>
21844 * tsan.c (instrument_builtin_call): If the memory model argument
21845 is not a constant, assume it is valid.
21848 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
21851 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
21853 PR rtl-optimization/79910
21854 * combine.c (can_combine_p): Do not allow combining an I0 or I1
21855 if its dest is used by an insn before I2 (other than the combined
21856 insns themselves, which are properly handled already).
21858 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
21861 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
21863 * combine.c (record_used_regs): New static function.
21864 (try_combine): Handle situations where there is an additional
21865 instruction between I2 and I3 which needs to have a LOG_LINK
21869 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
21871 * combine.c (try_combine): Delete redundant i1 test. Call
21872 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
21874 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21877 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
21878 alternatives 13/14.
21880 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21882 PR tree-optimization/80054
21883 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
21884 the optimization if a PHI or any of its arguments is not dominated
21885 by the candidate's basis. Use gphi* rather than gimple* as
21887 (replace_profitable_candidates): Clean up a gimple* variable that
21888 should be a gphi* variable.
21890 2017-03-20 Martin Sebor <msebor@redhat.com>
21893 * doc/extend.texi (attribute constructor): Document present limitation.
21895 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
21898 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
21899 __POWER9_VECTOR__ #ifdef control, change template definition to
21900 use Power9-specific built-in function.
21901 (vec_any_eq): Likewise.
21902 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
21903 to control outcomes from this test.
21904 (vector_ae_<mode>p): For VEC_F modes, likewise.
21906 2017-03-20 Ian Lance Taylor <iant@google.com>
21908 * config/i386/i386.c (ix86_function_regparm): Save an extra
21909 register for -fsplit-stack with DECL_STATIC_CHAIN.
21911 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
21914 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
21915 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
21917 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
21919 * config/riscv/riscv.c (riscv_print_operand): Use "fence
21921 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
21924 2017-03-20 Marek Polacek <polacek@redhat.com>
21927 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
21929 2017-03-20 Richard Biener <rguenther@suse.de>
21931 PR tree-optimization/80113
21932 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
21933 allocate extra SSA name for PHI def.
21934 (add_close_phis_to_outer_loops): Likewise.
21935 (add_close_phis_to_merge_points): Likewise.
21936 (copy_loop_close_phi_args): Likewise.
21937 (copy_cond_phi_nodes): Likewise.
21939 2017-03-20 Martin Liska <mliska@suse.cz>
21941 PR middle-end/79753
21942 * tree-chkp.c (chkp_build_returned_bound): Do not build
21943 returned bounds for a LHS that's not a BOUNDED_P type.
21945 2017-03-20 Martin Liska <mliska@suse.cz>
21949 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
21950 COMPLEX_CST and VECTOR_CST.
21952 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21955 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
21956 target operand. A new splitter adds the clobber statement in case
21957 the target operand is dead anyway.
21959 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
21961 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
21962 to age-old versions of binutils and glibc.
21964 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
21966 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
21968 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
21970 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
21972 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
21974 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
21975 requirement for binutils 2.13.
21977 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
21979 * combine.c (try_combine): Delete redundant i1 test. Call
21980 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
21982 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
21984 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
21985 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
21987 <riscv64-*-elf>: Re-arrange section
21988 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
21989 <riscv32-*-linux>: Likewise.
21990 <riscv64-*-elf>: Likewise
21991 <riscv64-*-linux>: Likewise.
21993 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
21996 * aarch64.opt(verbose-cost-dump): Fix typo.
21998 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
22001 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
22002 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
22004 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
22006 * reload.c (find_reloads): When reloading a nonoffsettable address,
22007 use RELOAD_OTHER for it and its address reloads.
22009 PR rtl-optimization/79910
22010 * combine.c (record_used_regs): New static function.
22011 (try_combine): Handle situations where there is an additional
22012 instruction between I2 and I3 which needs to have a LOG_LINK
22015 2017-03-17 Jeff Law <law@redhat.com>
22017 PR tree-optimization/71437
22018 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
22019 conditional in the hash table first.
22020 (vrp_dom_walker::before_dom_children): Extract condition from
22021 ASSERT_EXPR. Record condition, its inverion and any implied
22022 conditions as well.
22024 2017-03-17 Marek Polacek <polacek@redhat.com>
22025 Markus Trippelsdorf <markus@trippelsdorf.de>
22027 PR tree-optimization/80079
22028 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
22031 2017-03-17 Richard Biener <rguenther@suse.de>
22033 PR middle-end/80075
22034 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
22035 Properly verify the LHS before the RHS possibly claims to be
22037 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
22040 2017-03-17 Martin Jambor <mjambor@suse.cz>
22042 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
22043 (List of -O2 options): Likewise.
22044 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
22047 2017-03-17 Tom de Vries <tom@codesourcery.com>
22049 * gcov-dump.c (print_usage): Print bug_report_url.
22051 2017-03-17 Richard Biener <rguenther@suse.de>
22053 PR middle-end/80050
22054 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
22055 (parser::peek): Likewise.
22057 2017-03-17 Richard Biener <rguenther@suse.de>
22059 PR tree-optimization/80048
22060 * sese.c (free_sese_info): Properly release rename_map and
22061 copied_bb_map elements.
22063 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
22065 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
22066 Add linked-list forward and backlinks. Insert on
22067 construction, remove on destruction.
22068 (class pass_store_merging): Add m_stores_head field.
22069 (pass_store_merging::terminate_and_process_all_chains):
22070 Iterate over m_stores_head list.
22071 (pass_store_merging::terminate_all_aliasing_chains):
22073 (pass_store_merging::execute): Check for debug stmts first.
22074 Push new chains onto the m_stores_head stack.
22076 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
22079 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
22080 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
22081 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
22083 2017-03-16 Jeff Law <law@redhat.com>
22085 PR tree-optimization/71437
22086 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
22087 member function. Implementation moved into after_dom_children
22088 member function and into the threader's thread_outgoing_edges
22090 (dom_opt_dom_walker::after_dom_children): Simplify by moving
22091 some code into new thread_outgoing_edges.
22092 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
22093 definition. Simplify marker handling (do it here). Assume we always
22094 have the available expression and the const/copies tables.
22095 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
22097 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
22098 * tree-vrp.c (equiv_stack): No longer file scoped.
22099 (vrp_dom_walker): New class.
22100 (vrp_dom_walker::before_dom_children): New member function.
22101 (vrp_dom_walker::after_dom_children): Likewise.
22102 (identify_jump_threads): Setup domwalker. Use it rather than
22103 walking edges in a random order by hand. Simplify setup/finalization.
22104 (finalize_jump_threads): Remove.
22105 (vrp_finalize): Do not call identify_jump_threads here.
22106 (execute_vrp): Do it here instead and call thread_through_all_blocks
22109 PR tree-optimization/71437
22110 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
22112 (simplify_stmt_for_jump_threading): Add basic_block argument. All
22114 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
22115 (dom_opt_dom_walker::thread_across_edge): Remove
22116 handle_dominating_asserts argument. All callers changed.
22117 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
22118 changes. Remove calls to lhs_of_dominating_assert. Other
22119 uses of handle_dominating_asserts turn into unconditional code
22120 (simplify_control_stmt_condition_1): Likewise.
22121 (simplify_control_stmt_condition): Likewise.
22122 (thread_through_normal_block, thread_across_edge): Likewise.
22123 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
22124 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
22125 object if it is not an SSA_NAME.
22126 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
22127 before calling into the VRP specific simplifiers.
22128 (identify_jump_threads): Remove handle_dominating_asserts
22131 2017-03-16 Jakub Jelinek <jakub@redhat.com>
22134 * tree-diagnostic.c (default_tree_printer): No longer static.
22135 * tree-diagnostic.h (default_tree_printer): New prototype.
22137 2017-03-16 Tamar Christina <tamar.christina@arm.com>
22139 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
22140 Change ins into fmov.
22142 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22144 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
22145 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
22146 Use h_con constraint for operand 1.
22147 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
22148 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
22150 2017-03-15 Jeff Law <law@redhat.com>
22152 PR tree-optimization/71437
22153 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
22154 (record_temporary_equivalences): Use it.
22156 PR tree-optimization/71437
22157 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
22158 tree-ssa-scopedtables.
22159 (lookup_avail_expr, build_and_record_new_cond): Likewise.
22160 (record_conditions, record_cond, vuse_eq): Likewise.
22161 (record_edge_info): Adjust to API tweak of record_conditions.
22162 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
22163 (record_temporary_equivalences, optimize_stmt): Likewise.
22164 (eliminate_redundant_computations): Likewise.
22165 (record_equivalences_from_stmt): Likewise.
22166 * tree-ssa-scopedtables.c: Include options.h and params.h.
22167 (vuse_eq): New function, moved from tree-ssa-dom.c
22168 (build_and_record_new_cond): Likewise.
22169 (record_conditions): Likewise. Accept vector of conditions rather
22170 than edge_equivalence structure for first argument.
22171 for the first argument.
22172 (avail_exprs_stack::lookup_avail_expr): New member function, moved
22173 from tree-ssa-dom.c.
22174 (avail_exprs_stack::record_cond): Likewise.
22175 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
22176 from tree-ssa-dom.c.
22177 (avail_exprs_stack): Add new member functions lookup_avail_expr
22179 (record_conditions): Declare.
22181 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
22184 * lra-constraints.c (process_alt_operands): Increase reject for
22185 reloading an input/output operand.
22187 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
22190 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
22191 insns to convert from signed/unsigned char/short to IEEE 128-bit
22193 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
22195 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
22198 * config/i386/i386.c (ix86_vector_duplicate_value): Create
22199 subreg of inner mode for values already in registers.
22201 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
22203 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
22204 iteration reg is used after the loop.
22206 2017-03-14 Martin Sebor <msebor@redhat.com>
22208 PR tree-optimization/79800
22209 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
22210 precision in negative-positive range.
22211 (format_floating): Call non-const overload with adjusted precision.
22213 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
22216 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
22219 2017-03-14 Martin Sebor <msebor@redhat.com>
22221 PR middle-end/80020
22222 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
22223 * builtins.def (aligned_alloc): Use it.
22226 * Makefile.in (GTFILES): Add calls.c.
22227 * calls.c: Include "gt-calls.h".
22229 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
22231 PR rtl-optimization/79728
22232 * regs.h (struct target_regs): New field
22233 x_contains_allocatable_regs_of_mode.
22234 (contains_allocatable_regs_of_mode): New macro.
22235 * reginfo.c (init_reg_sets_1): Initialize it, and change
22236 contains_reg_of_mode so it includes global regs as well.
22237 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
22238 rather than contains_regs_of_mode.
22240 2017-03-14 Martin Liska <mliska@suse.cz>
22242 * doc/invoke.texi: Document options that can't be combined with
22243 -fcheck-pointer-bounds.
22245 2017-03-14 Martin Liska <mliska@suse.cz>
22247 PR middle-end/79831
22248 * doc/invoke.texi (-Wchkp): Document the option.
22250 2017-03-14 Martin Liska <mliska@suse.cz>
22252 * Makefile.in: Install gcov-dump.
22254 2017-03-14 Martin Liska <mliska@suse.cz>
22256 * multiple_target.c (expand_target_clones): Bail out for
22257 an invalid attribute.
22259 2017-03-14 Richard Biener <rguenther@suse.de>
22261 * alias.c (struct alias_set_entry): Pack properly.
22262 * cfgloop.h (struct loop): Likewise.
22263 * cse.c (struct set): Likewise.
22264 * ipa-utils.c (struct searchc_env): Likewise.
22265 * loop-invariant.c (struct invariant): Likewise.
22266 * lra-remat.c (struct cand): Likewise.
22267 * recog.c (struct change_t): Likewise.
22268 * rtl.h (struct address_info): Likewise.
22269 * symbol-summary.h (function_summary): Likewise.
22270 * tree-loop-distribution.c (struct partition): Likewise.
22271 * tree-object-size.c (struct object_size_info): Likewise.
22272 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
22273 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
22274 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
22275 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
22276 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
22277 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
22278 (struct _stmt_vec_info): Likewise.
22280 2017-03-14 Martin Liska <mliska@suse.cz>
22283 * multiple_target.c (create_dispatcher_calls): Check that
22284 a target can create a function dispatcher.
22286 2017-03-14 Martin Liska <mliska@suse.cz>
22289 * multiple_target.c (expand_target_clones): Drop local.local
22290 flag for default implementation.
22292 2017-03-14 Richard Biener <rguenther@suse.de>
22294 PR tree-optimization/80030
22295 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
22297 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
22299 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
22300 gcc_fallthrough() instead of __attribute__((fallthrough));
22302 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
22304 * doc/gcc.texi: Remove "up" link to (DIR).
22305 * doc/gccint.texi: Ditto.
22307 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
22309 * doc/install.texi (Specific) <avr>: Remove reference to
22312 2017-03-13 Jeff Law <law@redhat.com>
22314 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
22315 attribute rather than comments.
22317 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
22318 match_scratch operand is highest.
22320 2017-03-13 Martin Liska <mliska@suse.cz>
22322 PR middle-end/78339
22323 * ipa-pure-const.c (warn_function_noreturn): If the declarations
22324 is a CHKP clone, use original declaration.
22326 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
22328 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
22329 (arc_conditional_register_usage): Use a different allocation order
22330 when optimizing for size.
22331 * common/config/arc/arc-common.c (arc_option_optimization_table):
22332 Section anchors default on when optimizing for size.
22334 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
22336 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
22338 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
22340 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
22341 * config/arc/arc.md (cpu_facility): Add cd variant.
22342 (*movqi_insn): Add code density variant.
22343 (*movhi_insn): Likewise.
22344 (*movqi_insn): Likewise.
22345 (*addsi3_mixed): Likewise.
22346 (subsi3_insn): Likewise.
22348 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
22350 * config/arc/arc.md (movsi_cond_exec): Update constraint.
22352 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
22354 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
22355 expressions with MINUS and UNARY ops.
22357 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22360 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
22362 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
22363 between vec_select and vector argument.
22364 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
22365 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
22366 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
22367 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
22368 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
22369 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
22371 2017-03-13 Richard Biener <rguenther@suse.de>
22374 * params.def (vect-max-peeling-for-alignment): Fix typo.
22376 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
22378 * doc/install.texi (Specific) <mips-*-*>: Remove description of
22379 issue that only occurred with binutils below 2.18.
22381 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
22383 * doc/install.texi (Specific) <cris-axis-elf>: No longer
22384 refer to binutils 2.11/2.12 minimum.
22386 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
22388 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
22389 ftp.kernel.org and simplify binutils requirement.
22391 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
22393 * doc/invoke.texi (Warning Options): Fix spelling of link-time
22395 (Optimize Options): Ditto. Also remove redundancy.
22397 2017-03-10 David Malcolm <dmalcolm@redhat.com>
22399 PR translation/79848
22400 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
22402 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
22403 to G_ to avoid double translation.
22405 2017-03-10 David Malcolm <dmalcolm@redhat.com>
22407 PR translation/79923
22408 * auto-profile.c (get_combined_location): Convert leading
22409 character of diagnostics to lower case and remove trailing period.
22410 (read_profile): Likewise for various diagnostics.
22411 * config/arm/arm.c (arm_option_override): Remove trailing period
22412 from various diagnostics.
22413 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
22414 (msp430_expand_delay_cycles): Likewise.
22416 2017-03-10 David Malcolm <dmalcolm@redhat.com>
22419 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
22420 full command-line argument, rather than just "str".
22421 (aarch64_validate_march): Likewise.
22422 (aarch64_validate_mtune): Likewise.
22424 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
22426 PR rtl-optimization/78911
22427 * lra-assigns.c (must_not_spill_p): New function.
22428 (spill_for): Use it.
22430 2017-03-10 Jakub Jelinek <jakub@redhat.com>
22432 PR tree-optimization/79981
22433 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
22434 ATOMIC_COMPARE_EXCHANGE ifn result.
22435 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
22436 IFN_ATOMIC_COMPARE_EXCHANGE.
22438 2017-03-10 David Malcolm <dmalcolm@redhat.com>
22441 * opts.c (parse_sanitizer_options): Add missing question mark to
22442 "did you mean" message.
22444 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22446 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
22448 (VMULEUH_UNS): Likewise.
22449 (VMULOUB_UNS): Likewise.
22450 (VMULOUH_UNS): Likewise.
22451 * config/rs6000/rs6000.c (builtin_function_type): Remove
22452 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
22454 2017-03-10 David Malcolm <dmalcolm@redhat.com>
22457 * read-rtl-function.c (function_reader::read_rtx_operand): Update
22458 x with result of extra_parsing_for_operand_code_0.
22459 (function_reader::extra_parsing_for_operand_code_0): Convert
22460 return type from void to rtx, returning x. When reading
22461 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
22462 larger size containing struct block_symbol.
22464 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
22466 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
22467 -mfloat128-hardware without -m64.
22469 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
22472 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
22473 entries to the case statement that marks unsigned arguments to
22474 overloaded functions.
22476 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
22478 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
22479 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
22481 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
22484 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
22485 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
22487 2017-03-10 Martin Liska <mliska@suse.cz>
22491 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
22492 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
22495 2017-03-10 Olivier Hainque <hainque@adacore.com>
22497 * tree-switch-conversion (array_value_type): Start by resetting
22498 candidate type to it's main variant.
22500 2017-03-10 Jakub Jelinek <jakub@redhat.com>
22502 PR rtl-optimization/79909
22503 * combine.c (try_combine): Use simplify_replace_rtx on individual
22504 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
22505 of the whole CALL_INSN_FUNCTION_USAGE.
22507 PR tree-optimization/79972
22508 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
22509 get_range_info on SSA_NAMEs. Formatting fixes.
22511 2017-03-10 Richard Biener <rguenther@suse.de>
22512 Jakub Jelinek <jakub@redhat.com>
22514 PR tree-optimization/77975
22515 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
22516 edge to be constant.
22517 (get_val_for): For constant x return it. Formatting fix.
22518 (loop_niter_by_eval): Avoid pointless looping if the next iteration
22519 would use the same bases as the current one.
22521 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22523 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
22524 instead of vec_select for V1TImode.
22525 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
22527 (VSX_LE_128): Add V1TI to this mode iterator.
22528 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
22529 (*vsx_le_perm_store_<mode>): Likewise.
22530 (pre-reload splitter for VSX stores): Likewise.
22531 (post-reload splitter for VSX stores): Likewise.
22532 (*vsx_xxpermdi2_le_<mode>): Likewise.
22533 (*vsx_lxvd2x2_le_<mode>): Likewise.
22534 (*vsx_stxvd2x2_le_<mode>): Likewise.
22536 2017-03-09 Michael Eager <eager@eagercon.com>
22538 Correct failures with --enable-checking=yes,rtl.
22540 * config/microblaze/microblaze.c (microblaze_expand_shift):
22541 Replace GET_CODE test with CONST_INT_P and INTVAL test with
22542 test for const0_rtx.
22543 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
22544 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
22546 2017-03-09 Richard Biener <rguenther@suse.de>
22548 PR tree-optimization/79977
22549 * graphite-scop-detection.c (scop_detection::merge_sese):
22550 Handle the case of extra exits to blocks dominating the entry.
22552 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
22554 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
22557 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
22559 PR rtl-optimization/79949
22560 * lra-constraints.c (process_alt_operands): Check memory when
22561 trying to predict a cycle. Print about the overall increase.
22563 2017-03-09 Richard Biener <rguenther@suse.de>
22565 PR middle-end/79971
22566 * gimple-expr.c (useless_type_conversion_p): Preserve
22567 TYPE_SATURATING for fixed-point types.
22569 2017-03-09 Richard Biener <rguenther@suse.de>
22572 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
22573 alignment of BLKmode params.
22575 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22578 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
22579 (VALL_NO_V2Q): Likewise.
22581 * config/aarch64/aarch64-simd.md
22582 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
22584 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
22585 VALL_NO_V2Q mode iterator.
22586 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
22588 2017-03-09 Martin Liska <mliska@suse.cz>
22590 PR tree-optimization/79631
22591 * tree-chkp-opt.c (chkp_is_constant_addr): Call
22592 tree_int_cst_sign_bit just for INTEGER constants.
22594 2017-03-09 Martin Liska <mliska@suse.cz>
22598 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
22601 2017-03-09 Marek Polacek <polacek@redhat.com>
22604 * tree.c (inchash::add_expr): Handle TREE_VEC.
22606 2017-03-09 Martin Liska <mliska@suse.cz>
22609 (chkp_narrow_size_and_offset): New function.
22610 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
22611 (void chkp_parse_bit_field_ref): New function.
22612 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
22613 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
22615 2017-03-09 Martin Liska <mliska@suse.cz>
22618 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
22619 (chkp_find_bounds_1): Remove gcc_unreachable.
22621 2017-03-09 Jakub Jelinek <jakub@redhat.com>
22624 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
22625 BUILT_IN_SYNC*, determine the access type from the size suffix and
22626 always build a MEM_REF with that type. Handle forgotten
22627 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
22630 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
22631 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
22632 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
22633 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
22634 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
22635 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
22636 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
22637 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
22638 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
22639 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
22640 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
22641 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
22642 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
22643 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
22644 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
22645 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
22646 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
22647 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
22648 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
22649 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
22650 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
22651 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
22652 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
22653 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
22654 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
22655 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
22656 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
22657 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
22658 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
22659 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
22660 definitions outside of __OPTIMIZE__ guarded section.
22663 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
22664 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
22665 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
22666 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
22669 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22671 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
22672 ("vfenez<mode>"): Add missing constraints.
22674 2017-03-08 Martin Sebor <msebor@redhat.com>
22677 * config/nds32/nds32.c (nds32_option_override):
22678 Fix misspelled diagnostic.
22680 2017-03-08 Jakub Jelinek <jakub@redhat.com>
22683 * gimplify.c (gimplify_omp_for): Replace index var in outer
22684 taskloop statement with an artificial variable and add
22685 OMP_CLAUSE_PRIVATE clause for it.
22687 2017-03-08 Richard Biener <rguenther@suse.de>
22689 PR tree-optimization/79955
22690 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
22691 for accesses that are completely outside of the variable.
22693 2017-03-08 Andrew Haley <aph@redhat.com>
22695 PR tree-optimization/79943
22696 * tree-ssa-loop-split.c (compute_new_first_bound): When
22697 calculating the new upper bound, (END-BEG) should be added, not
22700 2017-03-08 Jakub Jelinek <jakub@redhat.com>
22702 * config/avr/avr.md (setmemhi): Make sure match_dup
22703 operand number comes before match_scratch.
22705 2017-03-08 Richard Biener <rguenther@suse.de>
22707 PR tree-optimization/79920
22708 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
22709 with ncopies == 1 to ...
22710 (vect_transform_slp_perm_load): ... here. Properly compute
22711 all element loads by iterating VF times over the group. Do
22712 not handle ncopies (computed in a broken way) in
22713 vect_create_mask_and_perm.
22715 2017-03-08 Jakub Jelinek <jakub@redhat.com>
22718 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
22719 is a uniform vector, use uniform_vector_p return value instead of
22720 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
22722 2017-03-07 Marek Polacek <polacek@redhat.com>
22724 PR middle-end/79809
22725 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
22726 (alloca_call_type): Likewise.
22728 2017-03-07 Martin Liska <mliska@suse.cz>
22730 * gcov.c (process_args): Put comment to correct location.
22732 2017-03-07 Martin Liska <mliska@suse.cz>
22734 PR middle-end/68270
22735 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
22736 Use array_at_struct_end_p instead of DECL_CHAIN (field).
22737 (chkp_narrow_bounds_for_field): Likewise.
22738 (chkp_parse_array_and_component_ref): Pass one more argument to
22741 2017-03-07 Richard Biener <rguenther@suse.de>
22743 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
22746 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
22748 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
22749 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
22751 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22754 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
22755 to end of description.
22756 (PARAM_MAX_STORES_TO_MERGE): Likewise.
22758 2017-03-07 Jakub Jelinek <jakub@redhat.com>
22760 PR rtl-optimization/79901
22761 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
22763 (*avx512f_<code><mode>3<mask_name>): ... this.
22764 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
22765 iterator instead of VI8_AVX2_AVX512BW.
22767 PR rtl-optimization/79901
22768 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
22769 min/max expander, expand it using expand_vec_cond_expr.
22772 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
22775 2017-03-06 Jakub Jelinek <jakub@redhat.com>
22778 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
22779 to void * for PCH reasons.
22780 * dwarf2out.c (output_loc_operands, output_die): Cast
22781 v.val_vec.array to unsigned char *.
22783 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
22786 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
22789 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
22791 PR rtl-optimization/79571
22792 * lra-constraints.c (process_alt_operands): Calculate static
22793 reject and subtract it from overall when only addresses will be
22796 2017-03-06 Julia Koval <julia.koval@intel.com>
22799 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
22800 incoming stack boundary to 128 for 64-bit targets.
22802 2017-03-06 Richard Biener <rguenther@suse.de>
22804 PR tree-optimization/79894
22805 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
22806 to NULL after folding it.
22808 2017-03-06 Richard Biener <rguenther@suse.de>
22810 PR tree-optimization/79824
22811 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
22812 check disabling peeling for gaps.
22814 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
22816 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
22817 attributes): Document gettimeofday.
22819 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
22821 * config/s390/s390.c (s390_option_override_internal): Set
22822 PARAM_MIN_VECT_LOOP_BOUND
22824 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
22826 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
22827 * config/s390/s390.md: Likewise.
22829 2017-03-06 Jakub Jelinek <jakub@redhat.com>
22832 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
22833 (<avx2_avx512>_perm<mode>): Rename to ...
22834 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
22836 (<avx512>_perm<mode>_mask): Rename to ...
22837 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
22839 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
22840 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
22841 instead of VI8F_256_512.
22842 (avx512f_perm<mode>): New define_expand.
22843 (avx512f_perm<mode>_mask): Likewise.
22844 (avx512f_perm<mode>_1<mask_name>): New define_insn.
22845 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
22847 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
22849 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
22850 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
22852 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
22854 2017-03-06 Martin Liska <mliska@suse.cz>
22857 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
22858 when having a SSA NAME w/o VAR_DECL assigned to it.
22860 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
22862 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
22863 msa_dpsub_<su>_d): Fix MODE for vec_select.
22865 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
22867 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
22869 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
22871 2017-03-06 Richard Biener <rguenther@suse.de>
22873 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
22874 * plugin.c (register_plugin_info): Likewise.
22875 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
22877 2017-03-05 Jakub Jelinek <jakub@redhat.com>
22879 * config/i386/sse.md (sse_storehps, sse_storelps,
22880 avx_<castmode><avxsizesuffix>_<castmode>,
22881 avx512f_<castmode><avxsizesuffix>_<castmode>,
22882 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
22883 in condition that at least one operand is not a MEM.
22885 2017-03-03 Jakub Jelinek <jakub@redhat.com>
22887 PR middle-end/79805
22888 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
22889 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
22891 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
22892 gimple_call_nothrow_p flag based on whether original builtin can throw.
22893 If it can, emit following stmts on the fallthrough edge.
22894 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
22895 don't create new bb if inserting just debug stmts on the edge, try to
22896 insert them on the fallthru bb or just reset debug stmts.
22898 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
22901 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
22902 restore recog_data (including the operand rtxes inside it) around
22903 the call to get_insn_template.
22905 2017-03-03 Martin Sebor <msebor@redhat.com>
22907 PR tree-optimization/79699
22908 * context.c (context::~context): Free MPFR caches to avoid
22909 a memory leak on program exit.
22911 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22913 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
22914 Use wide_int::ulow () instead of .elt (0).
22916 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
22918 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
22919 (*pushxf): Limit oF constraint to 32bit targets and add oC
22920 constraint for 64bit targets.
22921 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
22922 (*pushdf): Change rmF constraint to rmC.
22924 2017-03-03 Martin Liska <mliska@suse.cz>
22926 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
22927 Remove unused variable.
22929 2017-03-03 Jakub Jelinek <jakub@redhat.com>
22932 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
22933 is a memory operand, increase num_memory.
22934 (ix86_expand_args_builtin): Likewise.
22936 2017-03-03 Jan Hubicka <jh@suse.cz>
22939 * ipa-devirt.c (maybe_record_node): Properly handle
22940 __cxa_pure_virtual visibility.
22942 2017-03-03 Martin Liska <mliska@suse.cz>
22944 PR tree-optimization/79803
22945 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
22947 (pass_loop_prefetch::execute): Disabled optimization if an
22948 assumption about L1 cache size is not met.
22950 2017-03-03 Martin Liska <mliska@suse.cz>
22952 PR rtl-optimization/79574
22953 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
22954 (hash_scan_set): Likewise.
22955 (dump_hash_table): Likewise.
22956 (hoist_code): Likewise.
22958 2017-03-03 Richard Biener <rguenther@suse.de>
22960 * fixed-value.c (fixed_from_string): Restore use of elt (1)
22961 in place of uhigh ().
22962 (fixed_convert_from_real): Likewise.
22964 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
22967 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
22969 2017-03-03 Richard Biener <rguenther@suse.de>
22971 PR middle-end/79818
22972 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
22973 TYPE_OVERFLOW_UNDEFINED check.
22975 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22977 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
22979 (vector_ae_<mode>_p): Likewise.
22980 (vector_nez_<mode>_p): Likewise.
22981 (vector_ne_v2di_p): Likewise.
22982 (vector_ae_v2di_p): Likewise.
22983 (vector_ne_<mode>_p): Likewise.
22984 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
22986 (vsx_tsqrt<mode>2_fe): Likewise.
22988 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
22991 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
22993 2017-03-02 Jakub Jelinek <jakub@redhat.com>
22995 PR rtl-optimization/79780
22996 * cprop.c (one_cprop_pass): When second and further conditional trap
22997 in a single basic block is turned into an unconditional trap, turn it
22998 into a deleted note to avoid RTL verification failures.
23000 2017-03-02 Richard Biener <rguenther@suse.de>
23002 * fold-const.c (const_binop): Use ulow () instead of elt (0).
23004 2017-03-02 Richard Biener <rguenther@suse.de>
23006 PR tree-optimization/79345
23008 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
23009 param and abort the walk, returning -1 if it is hit.
23010 (walk_aliased_vdefs): Take a limit param and pass it on.
23011 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
23012 defaulting to 0 and return a signed int.
23013 * tree-ssa-uninit.c (struct check_defs_data): New struct.
23014 (check_defs): New helper.
23015 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
23016 about uninitialized memory.
23017 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
23018 bogus uninitialized warning.
23019 (fixed_convert_from_real): Likewise.
23021 2017-03-02 Bin Cheng <bin.cheng@arm.com>
23023 PR tree-optimization/66768
23024 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
23025 iv_use if base object can't be determined.
23027 2017-03-02 Jakub Jelinek <jakub@redhat.com>
23029 PR tree-optimization/79345
23030 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
23031 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
23032 (get_pattern_stats): Initialize it.
23033 * genemit.c (gen_expand): Verify match_scratch numbers come after
23034 match_operand/match_dup numbers.
23035 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
23036 match_scratch numbers.
23037 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
23039 * config/s390/s390.md (trunctdsd2): Likewise.
23041 2017-03-02 Richard Biener <rguenther@suse.de>
23043 * wide-int.h (wide_int_storage::operator=): Implement in terms
23046 2017-03-02 Richard Biener <rguenther@suse.de>
23048 PR tree-optimization/79777
23049 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
23050 the to insert expression to sth existing.
23052 2017-03-01 Martin Sebor <msebor@redhat.com>
23054 PR middle-end/79692
23055 * gimple-ssa-sprintf.c
23056 (directive::known_width_and_precision): New function.
23057 (format_integer): Use it.
23058 (get_mpfr_format_length): Consider the full range of precision
23059 when computing %g output with the # flag. Set the likely byte
23060 count to 3 rather than 1 when precision is indeterminate.
23061 (format_floating): Correct the lower bound of precision.
23063 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23065 * doc/invoke.texi: Document default code model for 64-bit Linux.
23067 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23070 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
23071 udiv rather than div since input pattern is unsigned.
23073 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
23075 * config/i386/i386.c (print_reg): Warn for values of
23076 unsupported size in integer register.
23078 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
23081 * config/rs6000/predicates.md (current_file_function_operand): Do
23082 not allow self calls to be local if the function is replaceable.
23084 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
23087 * config/rs6000/altivec.h (vec_ctz and others): Change the
23088 preprocessor macro that controls conditional compilation from
23089 _ARCH_PWR9 to __POWER9_VECTOR__.
23090 (vec_all_ne): Change parameterization of __altivec_scalar_pred
23091 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
23092 control (instead of _ARCH_PWR9 control) so that template
23093 definition uses power9-specific function.
23094 (vec_any_eq): Likewise.
23095 (vec_all_ne): Change macro definition to use a power9-specific
23096 expansion under #ifdef __POWER9_VECTOR__ control (instead of
23097 _ARCH_PWR9 control).
23098 (vec_any_eq) Likewise.
23099 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
23100 expansion for CMPNEF to remove support for xvcmpnesp instruction.
23101 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
23102 support for xvcmpnedp instruction.
23103 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
23104 macro expansion so that Power9 implementation of vec_all_ne does
23105 not use the AltiVec predicate framework.
23106 (VCMPNEH_P): Likewise.
23107 (VCMPNEW_P): Likewise.
23108 (VCMPNED_P): Likewise.
23109 (VCMPNEFP_P): Likewise.
23110 (VCMPNEDP_P): Likewise.
23111 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
23112 implementation of vec_any_eq to not use AltiVec predicate
23114 (VCMPAEH_P): Likewise.
23115 (VCMPAEW_P): Likewise.
23116 (VCMPAED_P): Likewise.
23117 (VCMPAEFP_P): Likewise.
23118 (VCMPAEDP_P): Likewise.
23119 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
23120 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
23121 not use the AltiVec predicate framework.
23122 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
23123 of vec_any_eq to not use AltiVec predicate framework.
23124 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
23125 support for predefined __POWER9_VECTOR__ macro to indicate that
23126 Power9 instruction selection is enabled.
23127 (altivec_overloaded_builtins): Remove extraneous
23128 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
23129 function argument types RS6000_BTI_bool_V16QI and
23130 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
23131 entry for overloaded function argument types RS6000_BTI_bool_V4SI
23132 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
23133 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
23134 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
23135 Power9 for implementations of vec_cmpne. Change the signature for
23136 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
23137 (representing vec_all_ne) to remove the previously described first
23138 argument of type RS6000_BTI_INTSI, as this was an artifact of
23139 reliance on the AltiVec predicate framework, which is no longer
23140 used in the implementation of these functions. Add
23141 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
23142 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
23143 since, unlike the AltiVec predicate framework implementation, we
23144 do not share function descriptors between vec_alle and vec_anyeq.
23145 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
23146 set of modes that receive special treatment even when
23147 TARGET_P9_VECTOR is true. The special treatment emits code that
23148 does not depend on Power9 instructions.
23149 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
23150 define_expand to not rely on AltiVec predicate framework.
23151 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
23153 (vector_ne_v2di_p): Change this define_expand to not rely on
23154 AltiVec predicate framework.
23155 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
23157 (vector_ne_<mode>_p): Change this define_expand to not rely on
23158 AltiVec predicate framework.
23159 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
23161 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
23162 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
23163 define_insn pattern.
23164 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
23165 define_insn pattern because the xvcmpne<VSs>. instruction is not
23167 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
23168 instruction is not supported.
23170 2017-03-01 Jakub Jelinek <jakub@redhat.com>
23172 * config/nvptx/nvptx.c: Include intl.h.
23174 2017-03-01 Martin Jambor <mjambor@suse.cz>
23177 * ipa-prop.h (ipa_bits): Removed field known.
23178 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
23179 to pointers. Adjusted their comments to warn about their sharing.
23180 (ipcp_transformation_summary): Change bits to a vector of pointers.
23181 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
23182 (ipa_get_ipa_bits_for_value): Declare.
23183 * tree-vrp.h (value_range): Mark as GTY((for_user)).
23184 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
23185 (ipa_bits_hash_table): Likewise.
23186 (ipa_vr_ggc_hash_traits): Likewise.
23187 (ipa_vr_hash_table): Likewise.
23188 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
23189 being pointers and vr_known being removed.
23190 (ipa_set_jf_unknown): Likewise.
23191 (ipa_get_ipa_bits_for_value): New function.
23192 (ipa_set_jfunc_bits): Likewise.
23193 (ipa_get_value_range): New overloaded functions.
23194 (ipa_set_jfunc_vr): Likewise.
23195 (ipa_compute_jump_functions_for_edge): Use the above functions to
23196 construct bits and vr parts of jump functions.
23197 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
23198 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
23200 (ipcp_grow_transformations_if_necessary): Also allocate
23201 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
23203 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
23204 them. Fix too long lines.
23205 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
23206 vr_known being removed.
23207 (ipa_read_jump_function): Use new setter functions to construct bits
23208 and vr parts of jump functions or set them to NULL.
23209 (write_ipcp_transformation_info): Adjust for bits being pointers.
23210 (read_ipcp_transformation_info): Likewise.
23211 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
23213 Include gt-ipa-prop.h.
23214 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
23216 (ipcp_store_bits_results): Likewise.
23217 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
23218 Do not write to existing jump functions but use a temporary instead.
23220 2017-03-01 Jakub Jelinek <jakub@redhat.com>
23223 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
23224 attempt to use its first operand as BIT_FIELD_REF base.
23226 2017-03-01 Richard Biener <rguenther@suse.de>
23228 PR middle-end/79721
23229 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
23230 interpolating formula in wrapping arithmetic.
23231 (chrec_apply): Convert chrec_evaluate return value to wanted type.
23233 2017-03-01 Jakub Jelinek <jakub@redhat.com>
23235 PR tree-optimization/79734
23236 * tree-vect-generic.c (expand_vector_condition): Optimize
23237 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
23238 Handle VEC_COND_EXPR where comparison has different inner width from
23239 type's inner width.
23241 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
23243 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
23244 markup, and similar issues. Remove @opindex entries for things
23245 that aren't options. Add missing -mmpy-option entries.
23247 2017-02-28 Jakub Jelinek <jakub@redhat.com>
23249 PR tree-optimization/79737
23250 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
23251 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
23252 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
23253 instead of byte_size. Formatting fix.
23254 (shift_bytes_in_array_right): Formatting fix.
23256 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
23259 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
23260 condition on optimize for the leaf function test.
23262 2017-02-28 Martin Liska <mliska@suse.cz>
23265 * read-rtl-function.c (function_reader::handle_unknown_directive):
23266 Bail out when one uses -flto.
23268 2017-02-28 Martin Liska <mliska@suse.cz>
23270 * common.opt: Replace space with tabular for options of <number>
23272 * config/i386/i386.opt: Show <number> value for
23273 -mlarge-data-threshold.
23274 * opts.c (print_filtered_help): Do not display number in hexadecimal
23277 2017-02-28 Martin Liska <mliska@suse.cz>
23279 * common.opt: Fix --help=option -Q for options which are of
23282 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
23284 * config/i386/i386.c (print_reg): Error out for values
23285 of 8-bit size in invalid integer register.
23287 2017-02-28 Martin Sebor <msebor@redhat.com>
23289 PR tree-optimization/79691
23290 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
23292 2017-02-28 Jakub Jelinek <jakub@redhat.com>
23295 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
23296 gcc_unreachable with output_operand_lossage.
23298 2017-02-28 Richard Biener <rguenther@suse.de>
23300 PR tree-optimization/79740
23301 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
23303 (visit_nary_op): Insert the nary into the hashtable if we
23304 pattern-matched sth.
23305 * tree-ssa-pre.c (eliminate_insert): Robustify.
23307 2017-02-28 Richard Biener <rguenther@suse.de>
23309 PR middle-end/79731
23310 * fold-const.c (decode_field_reference): Reject out-of-bound
23313 2017-02-28 Jakub Jelinek <jakub@redhat.com>
23315 * config/i386/i386.c: Include intl.h.
23316 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
23317 instead of just cond ? "..." : "...".
23318 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
23319 * coverage.c (read_counts_file): Likewise.
23320 * omp-offload.c: Include intl.h.
23321 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
23322 of just cond ? "..." : "...".
23323 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
23324 of just cond ? "..." : "...".
23326 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
23329 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
23331 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
23332 'tune for' CPU name.
23333 * config/arm/arm-cpu-data.h: Regenerated.
23335 2017-02-28 Richard Biener <rguenther@suse.de>
23337 PR tree-optimization/79732
23338 * tree-inline.c (expand_call_inline): Do not shadow var.
23340 2017-02-28 Richard Biener <rguenther@suse.de>
23342 PR tree-optimization/79723
23343 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
23344 address-space properly.
23346 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
23348 * doc/optinfo.texi (Optimization groups): Fix option used for
23350 * doc/invoke.texi (-fopt-info): Document "omp".
23351 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
23352 (OPTGROUP_ALL): Add OPTGROUP_OMP.
23353 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
23354 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
23355 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
23357 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
23359 * dumpfile.c (optgroup_options): Instead of "openmp", associate
23360 OPTGROUP_OMP with "omp".
23362 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
23365 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
23366 for arithmetic shift of unsigned V2DI.
23368 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
23370 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
23371 arc/linux.h headers.
23372 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
23373 (LINK_SPEC): Likewise.
23374 (ARC_TLS_EXTRA_START_SPEC): Likewise.
23375 (EXTRA_SPECS): Likewise.
23376 (STARTFILE_SPEC): Likewise.
23377 (ENDFILE_SPEC): Likewise.
23378 (LIB_SPEC): Likewise.
23379 (TARGET_SDATA_DEFAULT): Likewise.
23380 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
23381 (MULTILIB_DEFAULTS): Likewise.
23382 (DWARF2_UNWIND_INFO): Likewise.
23383 * config/arc/big.h: New file.
23384 * config/arc/elf.h: Likewise.
23385 * config/arc/linux.h: Likewise.
23386 * config/arc/t-uClibc: Remove.
23388 2017-02-27 Bin Cheng <bin.cheng@arm.com>
23390 PR tree-optimization/77536
23391 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
23392 (tree_transform_and_unroll_loop): Use above function to compute the
23393 estimated niter of unrolled loop and use it when scaling profile.
23394 Also use count info rather than frequency if it's non-zero.
23395 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
23396 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
23397 (vect_transform_loop): Call above function.
23399 2017-02-27 Richard Biener <rguenther@suse.de>
23401 PR tree-optimization/45397
23402 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
23403 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
23404 (visit_nary_op): Add pattern matching for CSEing sign-changed
23405 or truncated operations with wider ones.
23407 2017-02-27 Richard Biener <rguenther@suse.de>
23409 PR tree-optimization/79690
23410 * tree-vect-stmts.c (vectorizable_store): Use vector type
23411 built from the DR with address-space.
23413 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
23415 * doc/invoke.texi (Optimize Options): Refine the description
23416 of asan-use-after-return.
23418 2017-02-25 Alan Modra <amodra@gmail.com>
23420 PR rtl-optimization/79584
23421 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
23422 base, not ad->base_term, the reg within base. Remove assertion
23423 that ad->base == ad->base_term. Replace gen_int_mode using
23424 bogus mode with const0_rtx.
23426 2017-02-25 Jakub Jelinek <jakub@redhat.com>
23428 PR middle-end/79396
23429 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
23430 FMA_EXPR like tcc_binary or tcc_unary.
23432 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
23435 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
23437 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
23438 (output_loc_operands): Handle DW_OP_call_ref and
23439 DW_OP_GNU_variable_value.
23440 (struct variable_value_struct): New type.
23441 (struct variable_value_hasher): Likewise.
23442 (variable_value_hash): New variable.
23443 (string_types): Remove.
23444 (copy_loc_descr): New function.
23445 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
23446 (prepend_loc_descr_to_each): New function.
23447 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
23448 instead of add_loc_descr_to_each if the first argument is single
23449 location list and the second has multiple.
23450 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
23451 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
23452 when looking for variable value which doesn't have other location info.
23453 (loc_list_from_tree): Formatting fix.
23454 (gen_array_type_die): Simplify DW_AT_string_length handling.
23455 (adjust_string_types): Remove.
23456 (gen_subprogram_die): Don't call adjust_string_types nor test/set
23457 string_types. Call resolve_variable_values.
23458 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
23459 (resolve_addr_in_expr): Likewise. Add A argument.
23460 (copy_deref_exprloc): Remove deref argument. Adjust for the
23461 original expression being DW_OP_GNU_variable_value with optionally
23462 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
23463 optionally after it.
23464 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
23465 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
23466 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
23467 (variable_value_hasher::hash, variable_value_hasher::equal): New
23469 (resolve_variable_value_in_expr, resolve_variable_value,
23470 resolve_variable_values, note_variable_value_in_expr,
23471 note_variable_value): New functions.
23472 (dwarf2out_early_finish): Call note_variable_value on all toplevel
23475 2017-02-24 Jakub Jelinek <jakub@redhat.com>
23478 * opts.h (handle_generated_option): Add GENERATED_P argument.
23479 * opts-common.c (handle_option): Adjust function comment.
23480 (handle_generated_option): Add GENERATED_P argument, pass it to
23482 (control_warning_option): Pass false to handle_generated_option
23484 * opts.c (maybe_default_option): Pass true to handle_generated_option
23486 * optc-gen.awk: Likewise.
23488 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
23490 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
23491 a REG, look at the REG it is a SUBREG of.
23492 (splitter for cmpeqsi_t): Ditto.
23494 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
23496 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
23497 the special USEs with the pattern of the insn, not the insn itself.
23499 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
23502 * doc/invoke.texi: Document -mload-store-pairs.
23504 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
23505 Sandra Loosemore <sandra@codesourcery.com>
23507 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
23508 argument isn't a CONST_INT.
23509 (nios2_alternate_compare_const): Assert op is a CONST_INT.
23510 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
23511 (nios2_validate_compare): Bypass alternate compare logic if *op2
23512 is not a CONST_INT.
23513 (ldstwm_operation_p): Return false if first_base is not a REG or
23514 if first_offset is not a CONST_INT.
23516 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
23518 * config/cris/cris.md: Use correct operand in a define_peephole2.
23520 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
23522 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
23524 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
23526 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
23527 this_insn if it is an INSN or JUMP_INSN.
23528 (force_offsettable): Look at base, not at addr.
23529 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
23530 on things that aren't necessarily CONST_INTs.
23532 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
23534 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
23535 -mfpmath=sse is the default also for x86-32 targets with SSE2
23536 instruction set when @option{-ffast-math} is enabled
23538 2017-02-24 Jeff Law <law@redhat.com>
23540 PR rtl-optimizatoin/79286
23541 * ira.c (update_equiv_regs): Drop may_trap_p exception to
23544 2017-02-24 Richard Biener <rguenther@suse.de>
23546 PR tree-optimization/79389
23547 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
23550 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
23552 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
23553 function comment to reflect reality.
23554 (loop_exits_before_overflow): Fix typo in function description.
23556 2017-02-24 Richard Biener <rguenther@suse.de>
23558 PR tree-optimization/79389
23559 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
23560 properly that a threading opportunity exists. Detect conditional
23561 copy/constant propagation opportunities.
23563 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
23565 * config/visium/visium.md (type): Add trap.
23566 (b): New mode attribute.
23567 (*btst): Rename into...
23568 (*btst<mode>): ...this and adjust.
23569 (*cbranchsi4_btst_insn): Rename into...
23570 (*cbranch<mode>4_btst_insn): ...this and adjust.
23571 (trap): New define_insn.
23573 2017-02-23 Jakub Jelinek <jakub@redhat.com>
23575 PR tree-optimization/79389
23576 * ifcvt.c (struct noce_if_info): Add rev_cond field.
23577 (noce_reversed_cond_code): New function.
23578 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
23579 reversed_comparison_code. Formatting fix.
23580 (noce_try_store_flag): Test rev_cond != NULL in addition to
23581 reversed_comparison_code.
23582 (noce_try_store_flag_constants): Likewise.
23583 (noce_try_store_flag_mask): Likewise.
23584 (noce_try_addcc): Use rev_cond if non-NULL instead of
23585 reversed_comparison_code.
23586 (noce_try_cmove_arith): Likewise. Formatting fixes.
23587 (noce_try_minmax, noce_try_abs): Clear rev_cond.
23588 (noce_find_if_block): Initialize rev_cond.
23589 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
23590 instead of false as last argument never attempt to reverse it
23593 2017-02-23 Bin Cheng <bin.cheng@arm.com>
23595 PR tree-optimization/79663
23596 * tree-predcom.c (combine_chains): Process refs in reverse order
23597 only for ZERO length chains, and add explaining comment.
23599 2017-02-23 Jeff Law <law@redhat.com>
23601 PR tree-optimization/79578
23602 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
23603 in call to operand_equal_p.
23605 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
23608 * config/i386/cpuid.h: Fix another undefined behavior.
23610 2017-02-23 Richard Biener <rguenther@suse.de>
23612 PR tree-optimization/79683
23613 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
23614 vector types for data-refs.
23616 2017-02-23 Martin Liska <mliska@suse.cz>
23618 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
23620 2017-02-23 Jakub Jelinek <jakub@redhat.com>
23622 PR middle-end/79665
23623 * internal-fn.c (get_range_pos_neg): Moved to ...
23624 * tree.c (get_range_pos_neg): ... here. No longer static.
23625 * tree.h (get_range_pos_neg): New prototype.
23626 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
23627 are known to be in between 0 and signed maximum inclusive, try to
23628 expand both unsigned and signed divmod and use the cheaper one from
23631 2017-02-22 Jeff Law <law@redhat.com>
23633 PR tree-optimization/79578
23634 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
23635 to compare base operands.
23637 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
23640 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
23641 gpc_reg_operand instead of fpr_reg_operand.
23643 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
23645 * config/mips/mips.c (mips_return_in_memory): Force FP
23646 vector types to be returned in memory for o32 ABI.
23648 2017-02-22 Jakub Jelinek <jakub@redhat.com>
23650 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
23651 instead of DW_TAG_member for static data member declarations and don't
23652 set no_linkage_name for static inline data members.
23653 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
23656 2017-02-22 Martin Liska <mliska@suse.cz>
23658 * doc/invoke.texi: Replace inequality signs with square brackets
23661 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23663 PR tree-optimization/68644
23664 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
23666 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
23669 * lra-constraints.c (simplify_operand_subreg): Handle
23670 WORD_REGISTER_OPERATIONS targets.
23672 2017-02-22 Jakub Jelinek <jakub@redhat.com>
23675 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
23676 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
23677 elimination by swapping fld*.
23679 2017-02-22 Richard Biener <rguenther@suse.de>
23681 PR tree-optimization/79673
23682 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
23683 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
23684 irrelevant address-space qualifiers and avoiding a
23685 ADDR_SPACE_CONVERT_EXPR from fold_convert.
23687 2017-02-22 Richard Biener <rguenther@suse.de>
23689 PR tree-optimization/79666
23690 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
23691 to not symbolically negate if that may introduce undefined
23694 2017-02-22 Martin Liska <mliska@suse.cz>
23697 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
23698 * data-streamer-out.c (streamer_write_gcov_count_stream):
23700 * value-prof.c (stream_out_histogram_value): Make assert more
23701 precise based on type of counter.
23703 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
23706 * config/i386/i386.md (standard_x87sse_constant_load splitter):
23707 Use nonimmediate_operand instead of memory_operand for operand 1.
23708 (float-extend standard_x87sse_constant_load splitter): Ditto.
23710 2017-02-21 Jeff Law <law@redhat.com>
23712 PR tree-optimization/79621
23713 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
23714 blocks with edges to themselves.
23716 2017-02-21 Jakub Jelinek <jakub@redhat.com>
23719 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
23720 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
23721 Use gimple_call_builtin_p.
23724 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
23725 on temporarily removed DEBUG_INSNs.
23727 PR tree-optimization/79649
23728 * tree-loop-distribution.c (classify_partition): Give up on
23729 non-generic address space loads/stores.
23731 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
23733 * doc/loop.texi (Loop manipulation): Remove nonexistent
23734 tree_ssa_loop_version from the documentation.
23735 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
23737 2017-02-21 Jakub Jelinek <jakub@redhat.com>
23740 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
23741 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
23742 * config/rs6000/rs6000.c: Include except.h.
23743 (rs6000_expand_split_stack_prologue): Call
23744 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
23746 2017-02-21 Martin Jambor <mjambor@suse.cz>
23749 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
23750 have been analyzed.
23752 2017-02-21 Martin Jambor <mjambor@suse.cz>
23754 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
23755 for backward compatibility only.
23756 * doc/invoke.texi (Option Summary): Remove all references to
23757 -fipa-cp-alignment.
23759 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
23763 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
23765 * lra-constraints.c (curr_insn_transform): Handle
23766 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
23768 2017-02-21 Martin Liska <mliska@suse.cz>
23770 * config/i386/i386.opt: Replace -masm-dialect with -masm.
23772 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
23774 PR translation/79638
23775 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
23777 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
23780 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
23781 (arm_function_ok_for_sibcall): Return false for an indirect call by
23782 descriptor if all the argument registers are used.
23783 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
23784 alignment of the function.
23786 2017-02-21 Jakub Jelinek <jakub@redhat.com>
23788 PR tree-optimization/61441
23789 * simplify-rtx.c (simplify_const_unary_operation): For
23790 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
23791 the sNaN unmodified.
23793 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
23795 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
23796 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
23797 instead of SYSTEM_HEADER_DIR.
23799 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
23800 Martin Liška <mliska@suse.cz>
23802 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
23803 Fix typos and grammar, use active voice, and clarify.
23805 2017-02-20 Marek Polacek <polacek@redhat.com>
23807 PR middle-end/79537
23808 * gimplify.c (gimplify_expr): Handle unused *&&L;.
23811 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
23813 2017-02-20 Jakub Jelinek <jakub@redhat.com>
23816 * config/i386/i386.c (ix86_expand_builtin): Handle
23817 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
23818 ix86_builtins_isa[fcode].isa as a requirement of those
23819 flags and any other flag in the bitmask.
23820 (ix86_init_mmx_sse_builtins): Use 0 instead of
23821 ~OPTION_MASK_ISA_64BIT as mask.
23822 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
23823 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
23824 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
23825 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
23827 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
23830 * lra-constraints.c (split_reg): Check requested split mode
23831 is supported by the register.
23833 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
23835 * lra-constraints.c (simplify_operand_subreg): Remove early
23838 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
23841 * lra-constraints.c (curr_insn_transform): Tighten condition
23842 for converting SUBREG reloads from OP_OUT to OP_INOUT.
23844 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
23847 * lra-constraints.c (curr_insn_transform): Handle
23848 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
23850 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
23853 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
23855 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
23857 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
23860 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
23863 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23865 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
23868 2017-02-18 Jakub Jelinek <jakub@redhat.com>
23870 * final.c (last_columnnum, override_columnnum): New variables.
23871 (final_start_function): Set last_columnnum, pass it to begin_prologue
23872 hook and pass 0 to dwarf2out_begin_prologue.
23873 (final_scan_insn): Update override_columnnum. Pass last_columnnum
23874 to source_line debug hook.
23875 (notice_source_line): Compute last_columnnum and for debug_column_info
23876 return true on column changes.
23877 * debug.h (struct gcc_debug_hooks): Add column argument to
23878 source_line and begin_prologue hooks.
23879 (debug_nothing_int_charstar_int_bool): Remove prototype.
23880 (debug_nothing_int_int_charstar,
23881 debug_nothing_int_int_charstar_int_bool): New prototypes.
23882 (dwarf2out_begin_prologue): Add column argument.
23883 * debug.c (do_nothing_debug_hooks): Adjust source_line and
23884 begin_prologue hooks.
23885 (debug_nothing_int_charstar_int_bool): Remove.
23886 (debug_nothing_int_int_charstar,
23887 debug_nothing_int_int_charstar_int_bool): New functions.
23888 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
23889 through to dwarf2out_source_line.
23890 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
23891 (dwarf2out_source_line): Add column argument, emit it if requested.
23892 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
23894 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
23895 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
23896 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
23897 through to dwarf2out_begin_prologue.
23898 (vmsdbgout_source_line): Add column argument, pass it through to
23899 dwarf2out_source_line.
23900 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
23901 dbxout_source_line caller.
23902 (dbxout_source_line): Add column argument.
23904 * common.opt (gno-column-info, gcolumn-info): New options.
23905 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
23906 (check_die): Also test for multiple DW_AT_decl_column attributes.
23907 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
23908 DW_AT_decl_column if requested.
23909 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
23911 (gen_variable_die): Likewise.
23912 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
23913 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
23916 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
23917 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
23918 (ix86_handle_option): Handle OPT_m3dnowa.
23919 * doc/invoke.texi (-m3dnowa): Document.
23920 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
23921 -m3dnowa instead of -m3dnow -march=athlon.
23924 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
23925 instead of gcc_assert for K, r and R code checks. Formatting fixes.
23927 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23930 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
23931 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
23932 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
23933 generator for vsx_xxpermdi_<mode>_be.
23934 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
23935 force big-endian semantics.
23936 (vsx_xxpermdi_<mode>_be): New define_expand with same
23937 implementation as previous version of vsx_xxpermdi_<mode>.
23939 2017-02-17 Jakub Jelinek <jakub@redhat.com>
23941 PR tree-optimization/79327
23942 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
23943 variable, its initialization and use.
23945 2017-02-17 Julia Koval <julia.koval@intel.com>
23947 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
23948 (OPTION_MASK_ISA_PKU_UNSET): New.
23949 (ix86_handle_option): Handle -mrdpid.
23950 * config/i386/cpuid.h (bit_RDPID): New.
23951 * config/i386/driver-i386.c (host_detect_local_cpu):
23952 Detect RDPID feature.
23953 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
23954 * config/i386/i386-c.c (ix86_target_macros_internal):
23956 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
23957 (ix86_valid_target_attribute_inner_p): Add "rdpid".
23958 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
23959 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
23960 * config/i386/i386.md (define_insn "rdpid"): New.
23961 * config/i386/i386.opt Add -mrdpid.
23962 * config/i386/immintrin.h (_rdpid_u32): New.
23964 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
23966 PR rtl-optimization/79541
23967 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
23968 instead of transforming it into USE.
23970 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
23972 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
23973 If HONOR_SNANS (SFmode) force the input to a register.
23974 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
23975 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
23976 an frsp or similar insn.
23978 2017-02-17 Martin Liska <mliska@suse.cz>
23980 PR rtl-optimization/79577
23981 * params.def (selsched-max-sched-times): Increase minimum to 1.
23983 2017-02-17 Martin Liska <mliska@suse.cz>
23985 PR rtl-optimization/79574
23986 * gcse.c (want_to_gcse_p): Prevent integer overflow.
23988 2017-02-17 Martin Liska <mliska@suse.cz>
23990 PR tree-optimization/79529
23991 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
23992 ssa_defined_default_def_p to handle cases which are implicitly
23994 * tree-ssa.c (ssa_defined_default_def_p): New function.
23995 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
23996 which are implicitly defined.
23997 * tree-ssa.h (ssa_defined_default_def_p): Declare.
23999 2017-02-17 Richard Biener <rguenther@suse.de>
24001 PR middle-end/79576
24002 * params.def (max-ssa-name-query-depth): Limit to 10.
24004 2017-02-17 Richard Biener <rguenther@suse.de>
24006 PR tree-optimization/79552
24007 * tree-ssa-structalias.c (visit_loadstore): Properly verify
24010 2017-02-17 Richard Biener <rguenther@suse.de>
24013 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
24015 2017-02-17 Marek Polacek <polacek@redhat.com>
24017 PR middle-end/79536
24018 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
24019 (fold_negate_expr): New wrapper.
24021 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
24023 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
24024 Correct terminology and de-emphasize pre-standard behavior.
24026 2017-02-16 Alan Modra <amodra@gmail.com>
24028 PR rtl-optimization/79286
24029 * ira.c (def_dominates_uses): New function.
24030 (update_equiv_regs): Don't create an equivalence for insns that
24031 may trap where the register def does not dominate the use.
24033 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
24035 PR rtl-optimization/78127
24036 * lra.c (lra): Call lra_eliminate before finish the loop after
24039 2017-02-16 Richard Biener <rguenther@suse.de>
24041 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
24043 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
24044 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
24045 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
24046 (isl_val_int_from_wi): New function.
24047 (extract_affine_gmp): Rename to ...
24048 (extract_affine_wi): ... this, take a widest_int.
24049 (extract_affine_int): Just wrap extract_affine_wi.
24050 (add_param_constraints): Use isl_val_int_from_wi.
24051 (add_loop_constraints): Likewise, and extract_affine_wi.
24053 2017-02-15 Jeff Law <law@redhat.com>
24055 PR middle-end/79521
24056 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
24057 ira_init_register_move_cost_if_necessary.
24059 2017-02-15 Martin Sebor <msebor@redhat.com>
24061 PR middle-end/32003
24062 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
24063 removed in a prior commit.
24065 2017-02-15 Bin Cheng <bin.cheng@arm.com>
24067 PR tree-optimization/79347
24068 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
24069 counters during peeling.
24071 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
24073 * Makefile.in (site.exp): Remove "set ISLVER".
24075 2017-02-15 Jakub Jelinek <jakub@redhat.com>
24078 * real.c (real_from_integer): Call real_convert even for decimal.
24080 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
24083 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
24085 2017-02-14 Andrew Pinski <apinski@cavium.com>
24087 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
24088 cores and change the partno/implementer to be correct.
24089 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
24090 the 'B" as the implementer.
24091 * config/aarch64/aarch64-tune.md: Regenerate.
24093 2017-02-14 Carl Love <cel@us.ibm.com>
24095 * config/rs6000/rs6000.c: Add case statement entry to make the
24096 xvcvuxdsp built-in argument unsigned.
24097 * config/rs6000/vsx.md: Fix the source and return operand types so they
24098 match the instruction definitions from the ISA document. Fix typo
24099 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
24102 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
24105 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
24106 member early_clobber_alts.
24107 * lra-lives.c (reg_early_clobber_p): New.
24108 (process_bb_lives): Use it.
24109 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
24110 (debug_operand_data): Initialize early_clobber_alts.
24111 (setup_operand_alternative): Set up early_clobber_alts.
24112 (collect_non_operand_hard_regs): Ditto. Pass early clobber
24113 alternatives to new_insn_reg.
24114 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
24116 (lra_update_insn_regno_info): Pass the new arg.
24118 2017-02-14 Jakub Jelinek <jakub@redhat.com>
24120 PR middle-end/79505
24121 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
24122 (new_oacc_loop_raw): Don't clear already cleared fields.
24125 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
24126 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
24127 _mm512_prefetch_i64gather_ps): New inline functions and macros.
24129 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
24132 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
24134 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
24137 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
24138 the extra instruction to the right place to store 128-bit constant
24141 2017-02-14 Martin Sebor <msebor@redhat.com>
24143 PR middle-end/79448
24144 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
24145 warning for strings of unknown length.
24147 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
24149 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
24151 2017-02-14 Jeff Law <law@redhat.com>
24154 * ira-costs.c (scan_one_insn): Initialize register move costs
24155 for pseudos seen in USE/CLOBBER insns.
24157 PR tree-optimization/79095
24158 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
24159 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
24160 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
24161 if the operands are known to be not equal, then the resulting range
24163 (intersect_ranges): If the new range is ~[0,0] and the old range is
24164 wide, then prefer ~[0,0].
24165 * tree-vrp.c (overflow_comparison_p_1): New function.
24166 (overflow_comparison_p): New function.
24167 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
24168 if NAME is used in an overflow test.
24169 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
24170 overflow check that can be expressed as an equality test, then adjust
24171 ops to be that equality test.
24173 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24175 * config/s390/s390-builtin-types.def: Remove flags argument.
24176 * config/s390/s390.c (s390_init_builtins): Likewise.
24178 2017-02-14 Martin Liska <mliska@suse.cz>
24180 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
24181 vector. Fix trailing white spaces.
24183 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
24185 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
24188 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24190 PR rtl-optimization/68664
24191 * config/arm/arm.c (arm_sched_can_speculate_insn):
24192 New function. Declare prototype.
24193 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
24195 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24197 PR rtl-optimization/68664
24198 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
24200 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
24202 2017-02-14 Amit Pawar <amit.pawar@amd.com>
24204 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
24205 max skip bytes for function, loop and jump.
24207 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24209 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
24210 ABS_EXPR for gimple dump.
24212 2017-02-14 Jakub Jelinek <jakub@redhat.com>
24215 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
24217 PR tree-optimization/79408
24218 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
24219 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
24220 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
24221 also if rhs1 is INTEGER_CST.
24223 2017-02-14 Richard Biener <rguenther@suse.de>
24225 PR middle-end/79432
24226 * tree-into-ssa.c (insert_phi_nodes): When the function can
24227 have abnormal edges rewrite SSA names with broken use-def
24228 dominance out of SSA and register them for PHI insertion.
24230 2017-02-13 Martin Sebor <msebor@redhat.com>
24232 PR middle-end/79496
24233 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
24234 clearing info.nowrite flag when snprintf size argument is a range.
24236 2017-02-13 Jakub Jelinek <jakub@redhat.com>
24238 * cprop.c (cprop_jump): Add missing space in string literal.
24239 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
24240 (get_constraint_for_component_ref): Likewise.
24241 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
24242 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
24243 * lra-constraints.c (process_alt_operands): Likewise.
24244 * ipa-inline.c (inline_small_functions): Likewise.
24245 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
24246 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
24247 * trans-mem.c (diagnose_tm_1_op): Likewise.
24248 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
24249 (grid_parallel_clauses_gridifiable): Likewise.
24251 * config/nvptx/mkoffload.c (process): Add space in between
24254 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
24255 "MOD4_SSE_REGS" and "ALL_REGS".
24257 * spellcheck.c (test_data): Add , in between "foo" and "food".
24259 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24262 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
24263 boundary crossing check and subsequent code generation agree.
24265 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24267 * config/aarch64/aarch64.c (has_memory_op): Delete.
24268 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
24271 2017-02-13 Jakub Jelinek <jakub@redhat.com>
24273 PR rtl-optimization/79388
24274 PR rtl-optimization/79450
24275 * combine.c (distribute_notes): When removing TEM_INSN for which
24276 corresponding dest has last value recorded, invalidate that last
24279 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24281 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
24282 of explicit '@'. Add missing assembly comment marker on branch costs
24285 2017-02-13 Nathan Sidwell <nathan@acm.org>
24287 * gengtype-lex.l (<in_struct>): Add '/'.
24289 2017-02-13 Martin Liska <mliska@suse.cz>
24292 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
24294 2017-02-13 Richard Biener <rguenther@suse.de>
24296 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
24298 * configure: Re-generate.
24299 * config.in: Likewise.
24300 * graphite-dependences.c: Simplify as if
24301 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
24302 * graphite-isl-ast-to-gimple.c: Likewise.
24303 * graphite-optimize-isl.c: Likewise.
24304 * graphite-poly.c: Likewise.
24305 * graphite-sese-to-poly.c: Likewise.
24306 * graphite.h: Likewise.
24307 * toplev.c: Include isl/version.h and use isl_version () for
24308 printing the ISL version.
24309 * doc/install.texi: Update ISL requirement.
24311 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
24313 * doc/standards.texi (Standards): Update reference to
24316 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
24318 * doc/extend.texi (Named Address Spaces): sourceware.org now
24320 * doc/install.texi (Binaries): Ditto.
24323 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
24325 * doc/cpp.texi: Replace "stringify"/"stringification" with C
24326 standard terminology "stringize"/"stringizing" throughout.
24327 * doc/cppinternals.texi: Likewise.
24329 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
24331 * doc/extend.texi: Fix some spelling mistakes and typos.
24332 * doc/invoke.texi: Likewise.
24334 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
24337 * params.def (inline-min-speedup) Change from 10 to 8.
24339 2017-02-11 Jakub Jelinek <jakub@redhat.com>
24341 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
24344 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
24347 * ipa-inline-analysis.c (get_minimal_bb): New function.
24348 (record_modified): Use it.
24349 (remap_edge_change_prob): Handle also ancestor functions.
24351 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
24353 * doc/contrib.texi (Contributors): Remove broken link into
24354 the Mauve CVS repository.
24356 2017-02-11 Jakub Jelinek <jakub@redhat.com>
24358 PR middle-end/79454
24359 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
24360 result computation whenever lhs doesn't have vector mode, not
24361 just when it has BLKmode.
24363 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
24365 * doc/makefile.texi (profiledbootstrap): Refer to the
24366 installation instructions only in textual form.
24368 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24371 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
24373 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
24375 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
24376 (Specific): Update mingw-w64 reference.
24378 (Specific): Remove broken link to Renesas RX processor.
24380 2017-02-10 Richard Biener <rguenther@suse.de>
24382 * toplev.c (process_options): Do not mention obsolete graphite
24383 options when printing sorry message about missing graphite support.
24384 Mention -floop-nest-optimize.
24386 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
24388 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
24389 (vtst_p16): Likewise.
24390 (vtstq_p8): Likewise.
24391 (vtstq_p16): Likewise.
24393 (vtstq_p64): Likewise.
24394 * config/arm/arm_neon.h (vgetq_lane_p64): New.
24395 (vset_lane_p64): New.
24396 (vsetq_lane_p64): New.
24398 2017-02-10 Jakub Jelinek <jakub@redhat.com>
24400 PR tree-optimization/79411
24401 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
24402 stmt operands are SSA_NAMEs used in abnormal phis.
24403 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
24406 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
24409 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
24412 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
24414 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
24416 2017-02-09 Jakub Jelinek <jakub@redhat.com>
24418 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
24422 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
24423 not arbitrary TREE_CONSTANT.
24426 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
24427 "omp declare target link" attribute unless is_global_var.
24428 * omp-offload.c (find_link_var_op): Likewise.
24430 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
24431 Chung-Lin Tang <cltang@codesourcery.com>
24433 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
24435 (gimplify_adjust_omp_clauses): Don't delete TILE.
24436 (gimplify_omp_for): Deal with TILE.
24437 * internal-fn.c (expand_GOACC_TILE): New function.
24438 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
24440 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
24441 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
24443 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
24444 avoid DIV for outermost collapse var.
24445 (expand_oacc_for): Insert tile element loop as needed. Adjust.
24446 Remove out of date comments, fix whitespace.
24447 * omp-general.c (omp_extract_for_data): Deal with tiling.
24448 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
24449 adjust OLF_DIM_BASE value.
24450 (struct omp_for_data): Add tiling field.
24451 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
24452 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
24453 for auto loops. Remove default auto determining, moved to
24454 oacc_loop_fixed_partitions.
24455 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
24456 stmts, add e_mask field.
24457 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
24458 (oacc_thread_numbers): Use oacc_dim_call.
24459 (oacc_xform_tile): New.
24460 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
24461 (finish_oacc_loop): Adjust for ifns vector.
24462 (oacc_loop_discover_walk): Append loop abstraction sites to list,
24463 add case for GOACC_TILE fns.
24464 (oacc_loop_xform_loop): Delete.
24465 (oacc_loop_process): Iterate over call list directly, and add
24466 handling for GOACC_TILE fns.
24467 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
24469 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
24470 vector partitioning to outer loops. Assign 2 partitions to loops
24471 when available. Add TILE handling.
24472 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
24473 (execite_oacc_device_lower): Process GOACC_TILE fns,
24474 ignore unknown specs.
24475 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
24476 * tree.c (omp_clause_num_ops): Adjust TILE ops.
24477 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
24479 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
24481 * configure.ac (ACX_BUGURL): Update.
24482 * configure: Regenerate.
24484 2017-02-09 Richard Biener <rguenther@suse.de>
24486 PR tree-optimization/69823
24487 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
24488 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
24490 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
24492 * config/arc/arc-c.def: Add __NPS400__ definition.
24493 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
24494 (TARGET_NPS400): Define.
24496 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
24498 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
24500 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
24501 pointer, arch_info.
24502 (arc_cpu_types): Fill the arch_info field with a pointer into the
24503 arc_arch_types table.
24504 (arc_selected_cpu): Declare.
24505 * config/arc/arc.c (arc_selected_cpu): Make global.
24506 (arc_selected_arch): Delete.
24507 (arc_base_cpu): Delete.
24508 (arc_override_options): Remove references to deleted variables,
24509 update access to arch information.
24510 (ARC_OPT): Update access to arch information.
24511 (ARC_OPTX): Likewise.
24512 * config/arc/arc.h (arc_base_cpu): Remove declaration.
24513 (TARGET_ARC600): Update access to arch information.
24514 (TARGET_ARC601): Likewise.
24515 (TARGET_ARC700): Likewise.
24516 (TARGET_EM): Likewise.
24517 (TARGET_HS): Likewise.
24518 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
24521 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
24524 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
24525 condition/operands for integer GE/LE/GEU/LEU operations.
24527 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
24529 PR translation/79397
24530 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
24533 2017-02-08 Martin Jambor <mjambor@suse.cz>
24536 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
24537 whether allocation happened.
24538 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
24539 nothing was allocated.
24541 2017-02-08 Jakub Jelinek <jakub@redhat.com>
24543 PR tree-optimization/79408
24544 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
24545 constant, but SSA_NAME with a known integer range, use the minimum
24546 of that range instead of op1 to determine if modulo can be replaced
24547 with its first operand.
24549 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24551 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
24553 2017-02-08 Richard Biener <rguenther@suse.de>
24555 PR tree-optimization/71824
24556 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
24557 Check all loops contained in the merged region.
24559 2017-02-07 Andrew Pinski <apinski@cavium.com>
24561 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
24563 2017-02-07 Andrew Pinski <apinski@cavium.com>
24565 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
24566 (thunderxt88): Likewise.
24567 (thunderxt81): Disable LSE and change v8.1 to v8.
24568 (thunderxt83): Likewise.
24570 2017-02-07 Jakub Jelinek <jakub@redhat.com>
24571 Richard Biener <rguenther@suse.de>
24573 PR middle-end/79399
24574 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
24575 type from int to size_t.
24576 * ira-costs.c (struct_costs_size): Change type from int to size_t.
24578 2017-02-07 Jakub Jelinek <jakub@redhat.com>
24580 PR rtl-optimization/79386
24581 * cprop.c (bypass_conditional_jumps): Initialize
24582 bypass_last_basic_block already before splitting bbs after
24583 unconditional traps...
24584 (bypass_conditional_jumps): ... rather than here.
24587 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
24588 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
24589 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
24590 fix -masm=intel patterns.
24592 2017-02-07 Richard Biener <rguenther@suse.de>
24594 PR tree-optimization/79256
24595 PR middle-end/79278
24596 * builtins.c (get_object_alignment_2): Use min_align_of_type
24597 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
24598 and ADJUST_FIELD_ALIGN.
24600 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
24602 * doc/tm.texi: Regenerate.
24603 * stor-layout.c (layout_decl): Adjust.
24604 (update_alignment_for_field): Likewise.
24605 (place_field): Likewise.
24606 (min_align_of_type): Likewise.
24607 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
24608 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
24609 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
24610 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
24611 * config/frv/frv.c (frv_adjust_field_align): Likewise.
24612 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
24613 * config/i386/i386.c (x86_field_alignment): Likewise.
24614 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
24615 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
24616 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
24617 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
24618 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
24619 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
24623 2017-01-30 Richard Biener <rguenther@suse.de>
24625 PR tree-optimization/79256
24626 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
24627 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
24630 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
24632 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
24633 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
24634 builtins to SImode and emit a zero-extend, if necessary.
24636 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
24638 * docs/invoke.texi (RISC-V Options): Alphabetize.
24640 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
24642 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
24645 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
24647 * config/riscv/riscv.c: New file.
24648 * common/config/riscv/riscv-common.c: Likewise.
24649 * config.gcc: Likewise.
24650 * config/riscv/constraints.md: Likewise.
24651 * config/riscv/elf.h: Likewise.
24652 * config/riscv/generic.md: Likewise.
24653 * config/riscv/linux.h: Likewise.
24654 * config/riscv/multilib-generator: Likewise.
24655 * config/riscv/peephole.md: Likewise.
24656 * config/riscv/pic.md: Likewise.
24657 * config/riscv/predicates.md: Likewise.
24658 * config/riscv/riscv-builtins.c: Likewise.
24659 * config/riscv/riscv-c.c: Likewise.
24660 * config/riscv/riscv-ftypes.def: Likewise.
24661 * config/riscv/riscv-modes.def: Likewise.
24662 * config/riscv/riscv-opts.h: Likewise.
24663 * config/riscv/riscv-protos.h: Likewise.
24664 * config/riscv/riscv.h: Likewise.
24665 * config/riscv/riscv.md: Likewise.
24666 * config/riscv/riscv.opt: Likewise.
24667 * config/riscv/sync.md: Likewise.
24668 * config/riscv/t-elf-multilib: Likewise.
24669 * config/riscv/t-linux: Likewise.
24670 * config/riscv/t-linux-multilib: Likewise.
24671 * config/riscv/t-riscv: Likewise.
24672 * configure.ac: Likewise.
24673 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
24674 Waterman as RISC-V maintainers.
24675 * doc/install.texi: Add RISC-V entries.
24676 * doc/invoke.texi: Add RISC-V options section.
24677 * doc/md.texi: Add RISC-V constraints section.
24678 * configure: Regenerated.
24680 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
24683 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
24684 false values to be constant vectors with all 0 or all 1 bits set.
24685 (vcondu<mode><mode>): Likewise.
24686 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
24688 (fpmask_comparison_operator): Update comment.
24689 (vecint_comparison_operator): New predicate.
24690 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
24691 vector conditionals when the true and false values are constant
24692 vectors with all 0 bits or all 1 bits set.
24694 2017-02-06 Martin Sebor <msebor@redhat.com>
24696 PR tree-optimization/79376
24697 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
24699 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
24701 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
24702 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
24703 to simplify split condition.
24705 2017-02-06 Jakub Jelinek <jakub@redhat.com>
24707 * omp-expand.c (oxpand_omp_atomic_fetch_op,
24708 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
24711 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
24713 PR rtl-optimization/68664
24714 * target.def (can_speculate_insn): New hook.
24715 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
24716 * doc/tm.texi: Regenerate.
24717 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
24718 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
24719 (rs6000_sched_can_speculate_insn): New function.
24721 2017-02-06 Jakub Jelinek <jakub@redhat.com>
24723 PR tree-optimization/79284
24724 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
24725 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
24726 vectorizable_mask_load_store, vectorizable_operation,
24727 vect_is_simple_cond, get_same_sized_vectype): Use it instead
24728 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
24729 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
24730 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
24731 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
24732 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
24733 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
24734 is_gimple_assign (stmt). Replace another such test with
24735 is_gimple_assign (stmt).
24737 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
24740 * config/avr/avr.c (rtl-iter.h): Include it.
24741 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
24742 (avr_legitimate_combined_insn): ...and implementation.
24744 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
24746 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
24747 * config/s390/s390.c (s390_const_operand_ok)
24748 (s390_canonicalize_comparison, s390_extract_part)
24749 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
24750 (s390_contiguous_bitmask_p, s390_rtx_costs)
24751 (legitimize_pic_address): Likewise.
24752 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
24753 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
24754 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
24755 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
24756 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
24758 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
24760 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
24761 REGNO($0) == REGNO($1).
24763 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24765 * config/s390/linux.h(SIZE_TYPE): Add comment.
24767 2017-02-06 Julian Brown <julian@codesourcery.com>
24768 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24769 Virendra Pathak <virendra.pathak@broadcom.com>
24771 * config/aarch64/aarch64-cores.def: Change the scheduler
24773 * config/aarch64/aarch64.md: Include thunderx2t99.md.
24774 * config/aarch64/thunderx2t99.md: New file.
24776 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
24778 * doc/standards.texi (Go Language): Update link to language
24781 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
24783 * tree-eh.c (lower_resx): Sanitize profile.
24784 (cleanup_empty_eh_move_lp): Likewise.
24786 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
24789 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
24791 * cfgloopmanip.h (loop_version): Update prototype.
24792 * modulo-sched.c (sms_schedule): Update call of loop_version.
24793 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
24794 * tree-parloops.c (gen_parallel_loop): Likewise.
24795 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
24796 * tree-ssa-loop-split.c (split_loop): Likewise.
24797 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
24798 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
24800 2017-02-05 Martin Liska <mliska@suse.cz>
24803 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
24805 (print_operand_address): Initialize a struct to zero.
24807 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
24809 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
24810 garbage collector only in textual form.
24812 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
24814 * doc/extend.texi (x86 specific memory model extensions for
24815 transactional memory): Simplify a phrase.
24817 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
24820 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
24821 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
24822 (atomic_storedi_1): Likewise.
24824 2017-02-04 Jakub Jelinek <jakub@redhat.com>
24826 PR tree-optimization/79338
24827 * tree-parloops.c (gather_scalar_reductions): Don't call
24828 vect_analyze_loop_form for loop->inner before destroying loop's
24831 2017-02-03 Martin Sebor <msebor@redhat.com>
24833 PR tree-optimization/79327
24834 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
24835 when precision has resulted in leading zeros.
24836 (format_integer): Adjust the likely counter to assume an unknown
24837 argument that may be zero is non-zero.
24839 2017-02-03 Jason Merrill <jason@redhat.com>
24842 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
24843 avoid copying non-taken branch.
24845 2017-02-03 Jakub Jelinek <jakub@redhat.com>
24847 PR tree-optimization/79340
24848 * tree-vect-loop.c (vectorizable_reduction): Release
24849 vec_defs elements after safe_splicing them into other vectors.
24852 PR tree-optimization/79327
24853 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
24854 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
24856 (format_integer): Use wide_int_to_tree instead of build_int_cst
24857 + to_?hwi. If argmin is NULL, just set argmin and argmax to
24858 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
24859 of shortest and longest sequence.
24861 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
24863 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
24864 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
24866 2017-02-03 Walter Lee <walt@tilera.com>
24869 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
24870 after initial stackframe link reg save.
24871 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
24873 2017-02-03 Jakub Jelinek <jakub@redhat.com>
24876 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
24877 wu for stxssp alternative.
24879 2017-02-03 Martin Sebor <msebor@redhat.com>
24881 PR tree-optimization/79352
24882 * gimple-fold.c (get_range_strlen): Add argument.
24883 (get_range_strlen): Change return type to bool.
24884 (get_maxval_strlen): Pass in a dummy argument.
24885 * gimple-fold.h (get_range_strlen): Change return type to bool.
24886 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
24887 * tree.h (array_at_struct_end_p): Add argument.
24888 * tree.c (array_at_struct_end_p): Handle it.
24890 2017-02-03 Martin Liska <mliska@suse.cz>
24893 * multiple_target.c (create_dispatcher_calls): Redirect edge
24894 from a caller of a dispatcher.
24895 (expand_target_clones): Make the clones local.
24896 (ipa_target_clone): Do both target clones and resolvers.
24897 (ipa_dispatcher_calls): Remove the pass.
24898 (pass_dispatcher_calls::gate): Likewise.
24899 (make_pass_dispatcher_calls): Likewise.
24900 * passes.def (pass_target_clone): Put as very first IPA early
24903 2017-02-03 Martin Liska <mliska@suse.cz>
24905 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
24906 in case of a function with ifunc attribute.
24908 2017-02-03 Martin Liska <mliska@suse.cz>
24910 * cgraph.c (cgraph_node::dump): Dump function version info.
24911 * symtab.c (symtab_node::dump_base): Add missing new line.
24913 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
24915 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
24916 (ifcombine_ifandif): Use it.
24918 2017-02-03 Martin Liska <mliska@suse.cz>
24920 * doc/invoke.texi: Document default value for
24921 use-after-scope-direct-emission-threshold.
24923 2017-02-03 Martin Liska <mliska@suse.cz>
24925 PR tree-optimization/79339
24926 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
24927 (format_floating): Likewise.
24929 2017-02-03 Martin Liska <mliska@suse.cz>
24932 * ipa-prop.c (ipa_node_params_t::insert): Remove current
24934 (ipa_node_params_t::remove): Likewise.
24935 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
24936 initialization from removed ipa_node_params_t::insert.
24937 (ipa_node_params::~ipa_node_params): Move from removed
24938 ipa_node_params_t::release.
24939 * symbol-summary.h (symbol_summary::m_released): New member.
24940 Do not release a summary twice. Do not allow to call finalizer
24941 for types of a summary that live in GGC memory.
24943 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24945 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
24948 2017-02-02 Martin Sebor <msebor@redhat.com>
24950 PR middle-end/79275
24951 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
24952 (format_string): Tighten up the range of output for non-constant
24953 strings and correct the expected range for wide non-constant strings.
24955 2017-02-02 Martin Sebor <msebor@redhat.com>
24957 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
24959 PR middle-end/32003
24960 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
24962 (-fdump-tree-@var): Add to index and document how to come up
24963 with pass-specific option and dump file names.
24964 (-fdump-passes): Clarify where to look for output.
24966 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
24968 PR middle-end/77445
24969 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
24970 statistics of the analyzed path; allow threading for speed when
24971 any of BBs along the path are optimized for speed.
24973 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
24975 PR middle-end/78468
24976 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
24977 settings of the virtual registers.
24980 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
24982 * explow.c (get_dynamic_stack_size): Take known alignment of stack
24983 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
24986 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24988 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
24989 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
24991 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24993 * config/s390/s390.md: Add missing comments with the expanded
24995 * config/s390/vector.md: Likewise.
24996 * config/s390/vx-builtins.md: Likewise.
24998 2017-02-02 Jakub Jelinek <jakub@redhat.com>
25001 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
25002 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
25003 conditions on a single line.
25005 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25007 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
25008 __S390_VX__ to __VX__.
25010 2017-02-01 Andrew Pinski <apinski@cavium.com>
25012 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
25013 stmt_info to record_stmt_cost.
25014 (vect_get_known_peeling_cost): Pass stmt_info if known to
25016 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
25017 cpu_vector_cost field into
25018 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
25019 field into vec_int_stmt_cost and vec_fp_stmt_cost.
25020 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
25021 splitting of scalar_stmt_cost and vec_stmt_cost.
25022 (thunderx_vector_cost): Likewise.
25023 (cortexa57_vector_cost): LIkewise.
25024 (exynosm1_vector_cost): Likewise.
25025 (xgene1_vector_cost): Likewise.
25026 (thunderx2t99_vector_cost): Improve after the splitting of the two
25028 (aarch64_builtin_vectorization_cost): Update for the splitting of
25029 scalar_stmt_cost and vec_stmt_cost.
25031 2017-02-01 Torvald Riegel <triegel@redhat.com>
25032 Richard Henderson <rth@redhat.com>
25034 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
25035 conditional on existance of a fast atomic load.
25036 * optabs-query.c (can_atomic_load_p): New function.
25037 * optabs-query.h (can_atomic_load_p): Declare it.
25038 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
25039 no fast atomic load is available for the particular size of access.
25040 (expand_atomic_compare_and_swap): Likewise.
25041 (expand_atomic_load): Likewise.
25042 (expand_atomic_store): Likewise.
25043 (expand_atomic_fetch_op): Likewise.
25044 * testsuite/lib/target-supports.exp
25045 (check_effective_target_sync_int_128): Remove x86 because it provides
25046 no fast atomic load.
25047 (check_effective_target_sync_int_128_runtime): Likewise.
25049 2017-02-01 Richard Biener <rguenther@suse.de>
25051 * graphite.c: Include tree-vectorizer.h for find_loop_location.
25052 (graphite_transform_loops): Provide opt-info for optimized nests.
25053 * tree-parloop.c (parallelize_loops): Provide opt-info for
25054 parallelized loops.
25056 2017-02-01 Richard Biener <rguenther@suse.de>
25058 PR middle-end/79315
25059 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
25060 was not set before.
25062 2017-02-01 Richard Biener <rguenther@suse.de>
25064 PR tree-optimization/71824
25065 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
25066 Verify the loops are valid in the merged SESE region.
25067 (scop_detection::can_represent_loop_1): Check analyzing the
25068 evolution of the number of iterations in the region succeeds.
25070 2017-01-31 Ian Lance Taylor <iant@golang.org>
25072 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
25073 REG_ARGS_SIZE note to 32-bit push insns and call insn.
25075 2017-01-31 David Malcolm <dmalcolm@redhat.com>
25077 PR preprocessor/79210
25078 * input.c (get_substring_ranges_for_loc): Replace line_width
25079 assertion with error-handling.
25081 2017-01-31 Richard Biener <rguenther@suse.de>
25083 PR tree-optimization/77318
25084 * graphite-sese-to-poly.c (extract_affine): Fix assert.
25085 (create_pw_aff_from_tree): Take loop parameter.
25086 (add_condition_to_pbb): Pass loop of the condition to
25087 create_pw_aff_from_tree.
25089 2017-01-31 Jakub Jelinek <jakub@redhat.com>
25091 * config/s390/s390.c (s390_asan_shadow_offset): New function.
25092 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
25094 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
25098 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
25100 (convert_int_to_float128): Likewise.
25101 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
25102 (convert_int_to_float128): Likewise.
25103 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
25104 (UNSPEC_IEEE128_CONVERT): Likewise.
25105 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
25106 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
25107 Use local variables for IBM extended format.
25108 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
25109 (fix_trunc<mode>si2_fprs): Likewise.
25110 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
25111 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
25112 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
25113 to know that we can now have integers of all sizes in vector
25115 (fix<uns>_<mode>di2_hw): Likewise.
25116 (float<uns>_<mode>si2_hw): Likewise.
25117 (fix_<mode>si2_hw): Likewise.
25118 (fixuns_<mode>si2_hw): Likewise.
25119 (float<uns>_<mode>di2_hw): Likewise.
25120 (float_<mode>di2_hw): Likewise.
25121 (float_<mode>si2_hw): Likewise.
25122 (floatuns_<mode>di2_hw): Likewise.
25123 (floatuns_<mode>si2_hw): Likewise.
25124 (xscvqp<su>wz_<mode>): Delete, no longer used.
25125 (xscvqp<su>dz_<mode>): Likewise.
25126 (xscv<su>dqp_<mode>): Likewise.
25127 (ieee128_mfvsrd_64bit): Likewise.
25128 (ieee128_mfvsrd_32bit): Likewise.
25129 (ieee128_mfvsrwz): Likewise.
25130 (ieee128_mtvsrw): Likewise.
25131 (ieee128_mtvsrd_64bit): Likewise.
25132 (ieee128_mtvsrd_32bit): Likewise.
25134 2017-01-31 Martin Liska <mliska@suse.cz>
25137 * ipa-prop.c (ipa_free_all_node_params): Call release method
25138 instead of ~sumbol_summary to not to trigger double times
25141 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
25143 PR tree-optimization/71691
25144 * bitmap.h (class auto_bitmap): New.
25145 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
25146 is_maybe_undefined instead of ssa_undefined_value_p.
25148 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25150 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
25151 __S390_ARCH_LEVEL__ to __ARCH__.
25153 2017-01-31 Jakub Jelinek <jakub@redhat.com>
25155 PR tree-optimization/79267
25156 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
25157 if should_remove_lhs_p is true.
25159 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
25162 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
25163 (add_alignment_attribute): New.
25164 (base_type_die): Add alignment attribute.
25165 (subrange_type_die): Likewise.
25166 (modified_type_die): Likewise.
25167 (gen_array_type_die): Likewise.
25168 (gen_descr_array_type_die: Likewise.
25169 (gen_enumeration_type_die): Likewise.
25170 (gen_subprogram_die): Likewise.
25171 (gen_variable_die): Likewise.
25172 (gen_field_die): Likewise.
25173 (gen_ptr_to_mbr_type_die): Likewise.
25174 (gen_struct_or_union_type_die): Likewise.
25175 (gen_subroutine_type_die): Likewise.
25176 (gen_typedef_die): Likewise.
25177 (base_type_cmp): Compare alignment attribute.
25179 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25182 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
25183 (setb_unsigned) New pattern for setb with CCUNS.
25184 * config/rs6000/rs6000.c (expand_block_compare): Use a different
25185 subfc./subfe sequence to avoid overflow problems. Generate a
25186 shorter sequence with cmpld/setb for power9.
25187 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
25188 for generating subfc. instruction.
25189 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
25190 now uses this instruction.
25192 2017-01-30 Ian Lance Taylor <iant@google.com>
25195 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
25196 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
25198 2017-01-30 Martin Sebor <msebor@redhat.com>
25200 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
25201 Move constant to the right of a relational operator.
25202 (get_mpfr_format_length, format_character, format_string): Ditto.
25203 (should_warn_p, maybe_warn): Same.
25205 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
25207 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
25210 * asan.c (get_translation_unit_decl): Remove function.
25211 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
25213 2017-01-30 Martin Liska <mliska@suse.cz>
25215 PR gcov-profile/79259
25216 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
25217 -fprofile-generate.
25219 2017-01-30 Martin Liska <mliska@suse.cz>
25222 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
25223 Initialize variables with NULL value.
25225 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
25228 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
25230 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
25232 2017-01-30 Richard Biener <rguenther@suse.de>
25234 PR tree-optimization/79276
25235 * tree-vrp.c (process_assert_insertions): Properly adjust common
25236 when removing a duplicate.
25238 * gcc.dg/torture/pr79276.c: New testcase.
25240 2017-01-30 Richard Biener <rguenther@suse.de>
25242 PR tree-optimization/79256
25243 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
25244 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
25246 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
25248 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
25251 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
25252 ("*r<noxa>sbg_<mode>_sll_bitmask")
25253 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
25254 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
25255 Use contiguous_bitmask_nowrap_operand.
25257 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25260 * config/rs6000/altivec.h (vec_xl): Revise #define.
25261 (vec_xst): Likewise.
25263 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
25265 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
25267 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
25269 PR rtl-optimization/79194
25270 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
25271 traps before call to bypass_conditional_jumps.
25273 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
25275 PR tree-optimization/71374
25276 * lra-constraints.c (check_conflict_input_operands): New.
25277 (match_reload): Use it.
25279 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
25282 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
25283 account to calculate conflict_set.
25285 2017-01-27 Bin Cheng <bin.cheng@arm.com>
25287 PR rtl-optimization/78559
25288 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
25289 other_insn in combine.
25291 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
25293 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
25294 uint16_type_node for BT_UINT16.
25296 2017-01-27 David Malcolm <dmalcolm@redhat.com>
25298 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
25299 "RTL Tests" to menu.
25300 (GIMPLE Tests): New node.
25301 (RTL Tests): New node.
25303 2017-01-27 Richard Biener <rguenther@suse.de>
25305 PR tree-optimization/79245
25306 * tree-loop-distribution.c (distribute_loop): Apply cost
25307 modeling also to detected patterns.
25309 2017-01-27 Richard Biener <rguenther@suse.de>
25311 PR tree-optimization/71433
25312 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
25313 (compare_assert_loc): New function.
25314 (process_assert_insertions): Sort and optimize assert locations
25315 to remove duplicates and push down identical assertions on
25316 edges to their destination block.
25318 2017-01-27 Richard Biener <rguenther@suse.de>
25320 PR tree-optimization/79244
25321 * tree-vrp.c (remove_range_assertions): Forcefully propagate
25322 out SSA names even if abnormal.
25324 2017-01-27 Jakub Jelinek <jakub@redhat.com>
25326 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
25327 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
25328 instead of MPFR_RNDN.
25330 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
25333 * arm.c (arm_option_override): Don't call build_target_option_node
25334 until after doing all option overrides.
25335 (arm_valid_target_attribute_tree): Likewise.
25337 2017-01-27 Martin Liska <mliska@suse.cz>
25339 * doc/invoke.texi (-fprofile-arcs): Document profiling support
25340 for {cd}tors and C++ {cd}tors.
25342 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
25344 * config/s390/s390.md ("*setmem_long_and")
25345 ("*setmem_long_and_31z"): Use zero_extend instead of and.
25347 2017-01-26 Martin Sebor <msebor@redhat.com>
25349 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
25352 2017-01-26 Martin Sebor <msebor@redhat.com>
25354 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
25355 HAVE_DFmode before using XFmode or DFmode.
25356 (parse_directive): Avoid using the z length modifier to avoid
25357 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
25359 PR middle-end/78703
25360 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
25361 to accept adjustment as an array.
25362 (get_int_range): New function.
25363 (struct directive): Make width and prec arrays.
25364 (directive::set_width, directive::set_precision): Call get_int_range.
25365 (format_integer, format_floating): Handle width and precision ranges.
25366 (format_string, parse_directive): Same.
25368 2017-01-26 Jakub Jelinek <jakub@redhat.com>
25371 * dwarf2out.c (generate_skeleton_bottom_up): For children with
25372 comdat_type_p set, just clone them, but keep the children in the
25376 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
25377 which have direct callers with -fvar-tracking-assignments enabled
25379 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
25380 inside of type units.
25382 2017-01-26 Martin Sebor <msebor@redhat.com>
25384 PR middle-end/78703
25385 * gimple-ssa-sprintf.c (struct result_range): Add likely and
25387 (struct format_result): Replace number_chars, number_chars_min,
25388 and number_chars_max with a single member of struct result_range.
25390 (format_result::operator+=): Adjust.
25391 (struct fmtresult): Remove bounded. Handle likely and unlikely
25393 (fmtresult::adjust_for_width_or_precision): New function.
25394 (fmtresult:type_max_digits): New function.
25395 (bytes_remaining): Handle likely and unlikely counters.
25396 (min_bytes_remaining): Remove.
25397 (format_percent): Simplify.
25398 (format_integer, format_floating): Set likely and unlikely counters.
25399 (get_string_length, format_character, format_string): Same.
25400 (format_plain, should_warn_p): New function.
25401 (maybe_warn): Call should_warn_p. Update diagnostic messages
25402 and handle those for all directives, including plain strings.
25403 (format_directive): Handle likely and unlikely counters.
25404 Remove unnecessary quoting from diagnostics. Add an informational
25406 (add_bytes): Remove.
25407 (pass_sprintf_length::compute_format_length): Simplify.
25408 (try_substitute_return_value): Handle likely and unlikely counters.
25410 2017-01-26 Carl Love <cel@us.ibm.com>
25412 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
25413 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
25415 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
25418 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
25419 endianess for subregs into account.
25420 * lra-constraints.c (lra_constraints): Do risky transformations
25421 always on the first iteration.
25422 * lra-lives.c (check_pseudos_live_through_calls): Add arg
25423 last_call_used_reg_set.
25424 (process_bb_lives): Define and use last_call_used_reg_set.
25425 * lra.c (lra): Always continue after lra_constraints on the first
25428 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
25430 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
25432 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
25434 2017-01-26 Jakub Jelinek <jakub@redhat.com>
25436 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
25437 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
25438 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
25439 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
25440 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
25441 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
25442 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
25443 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
25444 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
25446 2017-01-26 Marek Polacek <polacek@redhat.com>
25449 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
25450 for the third operand.
25452 2017-01-26 Jakub Jelinek <jakub@redhat.com>
25454 PR middle-end/79236
25455 * omp-low.c (struct omp_context): Add simt_stmt field.
25456 (scan_omp_for): Return omp_context *.
25457 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
25458 context to the _simt_ SIMD stmt.
25459 (lower_omp_for): For combined SIMD with sibling _simt_
25460 SIMD, make sure to use the same decls in _looptemp_
25461 clauses as in the sibling.
25463 2017-01-26 David Sherwood <david.sherwood@arm.com>
25465 PR middle-end/79212
25466 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
25469 2017-01-26 Jakub Jelinek <jakub@redhat.com>
25472 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
25473 emit fld b; fld a; if possible.
25475 * brig-builtins.def: Update copyright years.
25476 * config/arm/arm_acle_builtins.def: Update copyright years.
25478 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
25481 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
25482 constraint instead of o for the stxsd instruction.
25484 2017-01-25 Carl Love <cel@us.ibm.com>
25486 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
25487 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
25489 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
25491 * doc/invoke.texi (C++ Dialect Options): Fix typo.
25493 2017-01-25 Richard Biener <rguenther@suse.de>
25495 PR tree-optimization/69264
25496 * target.def (vector_alignment_reachable): Improve documentation.
25497 * doc/tm.texi: Regenerate.
25498 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
25500 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
25501 earlier changes with respect to TYPE_USER_ALIGN.
25502 (vector_alignment_reachable_p): Likewise. Improve dumping.
25504 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25507 * config/arm/arm.md (xordi3): Force constant operand into a register
25510 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25512 * doc/invoke.texi (-fstore-merging): Correct default optimization
25513 levels at which it is enabled.
25514 (-O): Move -fstore-merging from list to...
25517 2017-01-25 Richard Biener <rguenther@suse.de>
25520 * omp-expand.c: Include debug.h.
25521 (expand_omp_taskreg): Make sure to generate early debug before
25522 outlining anything from a function.
25523 (expand_omp_target): Likewise.
25524 (grid_expand_target_grid_body): Likewise.
25526 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
25529 * asan.c (get_translation_unit_decl): New function.
25530 (asan_add_global): Extract modules file name from globals
25531 TRANSLATION_UNIT_DECL name.
25533 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
25536 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
25537 for long calls with APCS frame and VFP.
25539 2017-01-24 David Malcolm <dmalcolm@redhat.com>
25541 * cfg.c (original_copy_tables_initialized_p): New function.
25542 * cfg.h (original_copy_tables_initialized_p): New decl.
25543 * cfgrtl.c (relink_block_chain): Guard the call to
25544 free_original_copy_tables with a call to
25545 original_copy_tables_initialized_p.
25546 * cgraph.h (symtab_node::native_rtl_p): New decl.
25547 * cgraphunit.c (symtab_node::native_rtl_p): New function.
25548 (symtab_node::needed_p): Don't assert for early assembly output
25549 for __RTL functions.
25550 (cgraph_node::finalize_function): Set "force_output" for __RTL
25552 (cgraph_node::analyze): Bail out early for __RTL functions.
25553 (analyze_functions): Update assertion to support __RTL functions.
25554 (cgraph_node::expand): Bail out early for __RTL functions.
25555 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
25557 * function.h (struct function): Update comment for field
25559 * gimple-expr.c: Include "tree-pass.h".
25560 (gimple_has_body_p): Return false for __RTL functions.
25561 * Makefile.in (OBJS): Add run-rtl-passes.o.
25562 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
25564 (gcc::pass_manager::get_clean_slate): New accessor.
25565 * passes.c: Include "insn-addr.h".
25566 (should_skip_pass_p): Add logging. Update logic for running
25567 "expand" to be compatible with both __GIMPLE and __RTL. Guard
25568 property-provider override so it is only done for gimple passes.
25570 (skip_pass): New function.
25571 (execute_one_pass): Call skip_pass when skipping passes.
25572 * read-md.c (md_reader::read_char): Support filtering
25573 the input to a subset of line numbers.
25574 (md_reader::md_reader): Initialize fields
25575 m_first_line and m_last_line.
25576 (md_reader::read_file_fragment): New function.
25577 * read-md.h (md_reader::read_file_fragment): New decl.
25578 (md_reader::m_first_line): New field.
25579 (md_reader::m_last_line): New field.
25580 * read-rtl-function.c (function_reader::create_function): Only
25581 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
25582 curr_properties. Set DECL_INITIAL to a dummy block.
25583 (read_rtl_function_body_from_file_range): New function.
25584 * read-rtl-function.h (read_rtl_function_body_from_file_range):
25586 * run-rtl-passes.c: New file.
25587 * run-rtl-passes.h: New file.
25589 2017-01-24 Jeff Law <law@redhat.com>
25591 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
25594 2017-01-24 Bin Cheng <bin.cheng@arm.com>
25596 PR tree-optimization/79159
25597 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
25598 (record_nonwrapping_iv): Improve boundary using above function if no
25599 value range information.
25601 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
25602 Martin Jambor <mjambor@suse.cz>
25604 * brig-builtins.def: New file.
25605 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
25606 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
25607 (DEF_HSAIL_SAT_BUILTIN): Likewise.
25608 (DEF_HSAIL_INTR_BUILTIN): Likewise.
25609 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
25610 * builtin-types.def (BT_INT8): New.
25611 (BT_INT16): Likewise.
25612 (BT_UINT8): Likewise.
25613 (BT_UINT16): Likewise.
25614 (BT_FN_ULONG): Likewise.
25615 (BT_FN_UINT_INT): Likewise.
25616 (BT_FN_UINT_ULONG): Likewise.
25617 (BT_FN_UINT_LONG): Likewise.
25618 (BT_FN_UINT_PTR): Likewise.
25619 (BT_FN_ULONG_PTR): Likewise.
25620 (BT_FN_INT8_FLOAT): Likewise.
25621 (BT_FN_INT16_FLOAT): Likewise.
25622 (BT_FN_UINT32_FLOAT): Likewise.
25623 (BT_FN_UINT16_FLOAT): Likewise.
25624 (BT_FN_UINT8_FLOAT): Likewise.
25625 (BT_FN_UINT64_FLOAT): Likewise.
25626 (BT_FN_UINT16_UINT32): Likewise.
25627 (BT_FN_UINT32_UINT16): Likewise.
25628 (BT_FN_UINT16_UINT16_UINT16): Likewise.
25629 (BT_FN_INT_PTR_INT): Likewise.
25630 (BT_FN_UINT_PTR_UINT): Likewise.
25631 (BT_FN_LONG_PTR_LONG): Likewise.
25632 (BT_FN_ULONG_PTR_ULONG): Likewise.
25633 (BT_FN_VOID_UINT64_UINT64): Likewise.
25634 (BT_FN_UINT8_UINT8_UINT8): Likewise.
25635 (BT_FN_INT8_INT8_INT8): Likewise.
25636 (BT_FN_INT16_INT16_INT16): Likewise.
25637 (BT_FN_INT_INT_INT): Likewise.
25638 (BT_FN_UINT_FLOAT_UINT): Likewise.
25639 (BT_FN_FLOAT_UINT_UINT): Likewise.
25640 (BT_FN_ULONG_UINT_UINT): Likewise.
25641 (BT_FN_ULONG_UINT_PTR): Likewise.
25642 (BT_FN_ULONG_ULONG_ULONG): Likewise.
25643 (BT_FN_UINT_UINT_UINT): Likewise.
25644 (BT_FN_VOID_UINT_PTR): Likewise.
25645 (BT_FN_UINT_UINT_PTR: Likewise.
25646 (BT_FN_UINT32_UINT64_PTR): Likewise.
25647 (BT_FN_INT_INT_UINT_UINT): Likewise.
25648 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
25649 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
25650 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
25651 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
25652 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
25653 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
25654 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
25655 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
25656 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
25657 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
25658 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
25659 * doc/frontends.texi: List BRIG FE.
25660 * doc/install.texi (Testing): Add BRIG tesring requirements.
25661 * doc/invoke.texi (Overall Options): Mention BRIG.
25662 * doc/standards.texi (Standards): Doucment BRIG HSA version.
25664 2017-01-24 Richard Biener <rguenther@suse.de>
25666 PR translation/79208
25667 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
25669 2017-01-24 Martin Jambor <mjambor@suse.cz>
25672 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
25673 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
25674 and known_contexts.
25676 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
25678 PR middle-end/79123
25679 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
25680 casts from signed to unsigned really don't have a range.
25682 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
25684 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
25685 GMP_RNDx for compatiblity.
25687 2017-01-24 Martin Liska <mliska@suse.cz>
25690 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
25691 that would prevent us to call alloca with -1 as argument.
25693 2017-01-24 Jakub Jelinek <jakub@redhat.com>
25695 * dwarf2out.c (output_compilation_unit_header, output_file_names):
25696 Avoid -Wformat-security warning.
25698 2017-01-23 Andrew Pinski <apinski@cavium.com>
25700 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
25703 2017-01-23 Martin Sebor <msebor@redhat.com>
25705 PR middle-end/78703
25706 * gimple-ssa-sprintf.c (warn_level): New global.
25707 (format_integer): Use it here and throughout the rest of the file.
25708 Use the same switch to compute sign as base.
25709 (maybe_warn): New function.
25710 (format_directive): Factor out warnings into maybe_warn.
25711 Add debugging output. Use warn_level.
25712 (add_bytes): Use warn_level.
25713 (pass_sprintf_length::compute_format_length): Add debugging output.
25714 (try_substitute_return_value): Same.
25715 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
25717 PR middle-end/78703
25718 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
25719 (struct fmtresult, format_integer, format_floating): Adjust.
25720 (fmtresult::fmtresult): Set max correctly in two argument ctor.
25721 (get_string_length, format_string,format_directive): Same.
25722 (pass_sprintf_length::compute_format_length): Same.
25723 (try_substitute_return_value): Simplify slightly.
25725 PR middle-end/78703
25726 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
25727 (fmtresult::operator+=): Outlined.
25728 (struct fmtresult): Add ctors.
25729 (struct conversion_spec): Rename...
25730 (struct directive): ...to this. Add and remove data members.
25731 (directive::set_width, directive::set_precision): New functions.
25732 (format_percent): Use fmtresult ctor.
25733 (get_width_and_precision): Remove.
25734 (format_integer): Make naming changes. Avoid computing width and
25736 (format_floating): Same. Adjust indentation.
25737 (format_character, format_none): New functions.
25738 (format_string): Moved character handling to format_character.
25739 (format_directive): Remove arguments, change return type.
25740 (parse_directive): New function.
25741 (pass_sprintf_length::compute_format_length): Move directive
25742 parsing to parse_directive.
25744 2017-01-23 Jakub Jelinek <jakub@redhat.com>
25746 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
25747 (assign_assembler_name_if_needed): ... this.
25748 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
25749 (assign_assembler_name_if_needed): ... this.
25750 (free_lang_data_in_cgraph): Adjust callers.
25751 * cgraphunit.c (cgraph_node::analyze): Likewise.
25752 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
25755 2017-01-23 Richard Biener <rguenther@suse.de>
25757 PR tree-optimization/79088
25758 PR tree-optimization/79188
25759 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
25760 resetting loop bounds after last path deletion. Reset loop
25761 bounds of the target loop, make code match the comments.
25762 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
25763 Make sure loops need no fixups.
25765 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
25767 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
25768 exponent support with double type for first argument.
25769 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
25770 type returned by __builtin_vec_extract_sig,
25771 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
25772 functions from "vector int" to "vector unsigned int" or from
25773 "vector long long int" to "vector unsigned long long int".
25774 Changed type returned by __builtin_vec_extract_exp,
25775 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
25776 functions from "vector int" to "vector unsigned int" or from
25777 "vector long long int" to "vector unsigned long long int".
25778 Changed return type of __builtin_vec_test_data_class,
25779 __builtin_vec_test_data_class_sp, and
25780 __builtin_vec_test_data_class_dp from "vector int" to
25781 "vector bool int" or from "vector long long int" to "vector bool
25782 long long int" and changed second argument type from "unsigned
25783 int" to "int". Added new overloaded function forms "vector float
25784 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
25785 "vector float __builtin_vec_insert_exp_sp (vector float, vector
25786 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
25787 double, vector unsigned long long int)" and "vector double
25788 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
25789 long int)". Changed return type of
25790 __builtin_scalar_test_data_class and
25791 __builtin_scalar_test_data_class_sp and
25792 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
25793 int" and changed second argument from "unsigned int" to "int".
25794 Changed type returned by __builtin_scalar_test_neg,
25795 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
25796 from "int" to "bool int". Added new overloaded function form
25797 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
25798 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
25799 exponent double-precision with floating point first argument.
25800 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
25801 documentation of scalar_test_data_class, scalar_test_neg,
25802 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
25803 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
25804 vec_test_data_class built-in functions to reflect refinements in
25805 their type signatures.
25807 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
25809 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
25811 (aarch64_elf_asm_destructor): Likewise.
25813 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
25815 PR rtl-optimization/78634
25816 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
25817 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
25818 * ifcvt.c (noce_try_cmove): Add missing cost check.
25820 PR rtl-optimization/71724
25821 * combine.c (if_then_else_cond): Look for situations where it is
25822 beneficial to undo the work of one of the recursive calls.
25824 2017-01-23 Bin Cheng <bin.cheng@arm.com>
25826 PR tree-optimization/70754
25827 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
25828 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
25829 combined stmt before it if not NULL.
25830 (combine_chains): Process refs reversely and compute dominance point
25833 2017-01-23 Martin Liska <mliska@suse.cz>
25835 PR tree-optimization/79196
25836 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
25837 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
25839 (strlen_optimize_stmt): Call the renamed function.
25841 2017-01-23 Michael Matz <matz@suse.de>
25843 PR tree-optimization/78384
25844 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
25846 2017-01-23 Richard Biener <rguenther@suse.de>
25848 PR tree-optimization/79186
25849 * tree-vrp.c (register_new_assert_for): Make sure we've seen
25850 both incoming edges before moving an assert.
25852 2017-01-23 Martin Jambor <mjambor@suse.cz>
25854 * ipa-prop.c (load_from_param_1): Removed.
25855 (load_from_unmodified_param): Bits from load_from_param_1 put back
25857 (load_from_param): Removed.
25858 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
25859 with stmt. Reverted back to use of load_from_unmodified_param.
25861 2017-01-23 Martin Jambor <mjambor@suse.cz>
25864 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
25865 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
25866 field a pointer to garbage collected vector, mark lattices and
25867 ipcp_orig_node with GTY((skip)).
25868 (ipa_get_param_count): Adjust to descriptors being a pointer.
25869 (ipa_get_param): Likewise.
25870 (ipa_get_type): Likewise.
25871 (ipa_get_param_move_cost): Likewise.
25872 (ipa_set_param_used): Likewise.
25873 (ipa_get_controlled_uses): Likewise.
25874 (ipa_set_controlled_uses): Likewise.
25875 (ipa_is_param_used): Likewise.
25876 (ipa_node_params_t): Move into garbage collector. New methods insert
25878 (ipa_node_params_sum): Annotate wth GTY(()).
25879 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
25881 (ipa_load_from_parm_agg): Adjust declaration.
25882 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
25883 * ipa-profile.c (ipa_profile): Likewise.
25884 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
25885 (ipa_populate_param_decls): Make descriptors parameter garbage
25887 (ipa_dump_param): Adjust to descriptors being a pointer.
25888 (ipa_alloc_node_params): Likewise.
25889 (ipa_initialize_node_params): Likewise.
25890 (load_from_param_1): Make descriptors parameter garbage collected.
25891 (load_from_unmodified_param): Likewise.
25892 (load_from_param): Likewise.
25893 (ipa_load_from_parm_agg): Likewise.
25894 (ipa_node_params::~ipa_node_params): Removed.
25895 (ipa_free_all_node_params): Remove call to delete operator.
25896 (ipa_node_params_t::insert): New.
25897 (ipa_node_params_t::remove): Likewise.
25898 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
25899 copy known_csts and known_contexts vectors.
25900 (ipa_read_node_info): Adjust to descriptors being a pointer.
25901 (ipcp_modif_dom_walker): Make m_descriptors field garbage
25903 (ipcp_transform_function): Make descriptors variable garbage
25906 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
25908 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
25909 * config/i386/avx512dqintrin.h: Ditto.
25910 * config/i386/avx512fintrin.h: Ditto.
25911 * config/i386/i386.c: Handle new builtins.
25912 * config/i386/i386-builtin.def: Add new builtins.
25913 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
25914 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
25916 2017-01-23 Jakub Jelinek <jakub@redhat.com>
25917 Martin Liska <mliska@suse.cz>
25919 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
25920 * asan.c (asan_expand_poison_ifn): Support stores and use
25921 appropriate ASAN report function.
25922 * internal-fn.c (expand_ASAN_POISON_USE): New function.
25923 * internal-fn.def (ASAN_POISON_USE): Declare.
25924 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
25925 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
25926 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
25927 ASAN_POISON calls w/o LHS.
25928 * tree-ssa.c (execute_update_addresses_taken): Create clobber
25929 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
25930 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
25931 * gimplify.c (asan_poison_variables): Add attribute
25932 use_after_scope_memory to variables that really needs to live
25934 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
25935 having the attribute.
25937 2017-01-23 Martin Liska <mliska@suse.cz>
25939 * asan.c (create_asan_shadow_var): New function.
25940 (asan_expand_poison_ifn): Likewise.
25941 * asan.h (asan_expand_poison_ifn): New declaration.
25942 * internal-fn.c (expand_ASAN_POISON): Likewise.
25943 * internal-fn.def (ASAN_POISON): New builtin.
25944 * sanopt.c (pass_sanopt::execute): Expand
25945 asan_expand_poison_ifn.
25946 * tree-inline.c (copy_decl_for_dup_finish): Make function
25948 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
25949 * tree-ssa.c (is_asan_mark_p): New function.
25950 (execute_update_addresses_taken): Rewrite local variables
25951 (identified just by use-after-scope as addressable) into SSA.
25953 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
25955 * doc/install.texi (Specific): opensource.apple.com uses https
25956 now. Remove trailing slash.
25958 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
25960 * README.Portability: Remove note on an Irix compatibility issue.
25962 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
25964 * gcov.c (INCLUDE_ALGORITHM): Define.
25965 (INCLUDE_VECTOR): Define.
25966 No longer include <vector> and <algorithm> directly.
25968 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
25970 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
25972 * doc/invoke.texi (Code Gen Options): Ditto.
25974 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
25977 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
25979 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
25981 rtl-optimization/79125
25982 * cprop.c (local_cprop_pass): Handle cases where we make an
25983 unconditional trap.
25985 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
25989 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
25990 read from, for big endian.
25992 2017-01-20 Jiong Wang <jiong.wang@arm.com>
25994 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
25995 register pauth builtins for LP64 only.
25997 2017-01-20 Marek Polacek <polacek@redhat.com>
26000 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
26003 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
26005 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
26007 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
26008 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
26009 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
26011 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26014 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
26015 in big-endian mode when they are not a single duplicated value.
26017 2017-01-20 Richard Biener <rguenther@suse.de>
26019 * BASE-VER: Bump to 7.0.1.
26021 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
26023 * omp-low.c (omplow_simd_context): New struct. Use it...
26024 (lower_rec_simd_input_clauses): ...here and...
26025 (lower_rec_input_clauses): ...here to hold common data. Adjust all
26026 references to idx, lane, max_vf, is_simt.
26028 2017-01-20 Graham Markall <graham.markall@embecosm.com>
26030 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
26033 2017-01-20 Martin Jambor <mjambor@suse.cz>
26035 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
26036 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
26038 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
26039 (GTFILES): Rename hsa.c to hsa-common.c.
26040 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
26041 * hsa-dump.c: Likewise.
26042 * hsa-gen.c: Likewise.
26043 * hsa-regalloc.c: Likewise.
26044 * ipa-hsa.c: Likewise.
26045 * omp-expand.c: Likewise.
26046 * omp-low.c: Likewise.
26047 * toplev.c: Likewise.
26049 2017-01-20 Marek Polacek <polacek@redhat.com>
26052 * doc/invoke.texi: Document -Wduplicated-branches.
26053 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
26054 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
26055 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
26056 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
26057 return 0 only when not OEP_LEXICOGRAPHIC.
26058 (fold_build_cleanup_point_expr): Use the expression
26059 location when building CLEANUP_POINT_EXPR.
26060 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
26061 * tree.c (add_expr): Handle error_mark_node.
26063 2017-01-20 Martin Liska <mliska@suse.cz>
26066 * tree-profile.c (init_ic_make_global_vars): Do not call
26068 (gimple_init_gcov_profiler): Likewise.
26070 2017-01-20 Martin Liska <mliska@suse.cz>
26073 * cgraph.h (maybe_create_reference): Remove argument and
26075 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
26077 * ipa-cp.c (create_specialized_node): Likewise.
26078 * symtab.c (symtab_node::maybe_create_reference): Handle
26079 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
26081 2017-01-20 Martin Liska <mliska@suse.cz>
26083 * read-rtl-function.c (function_reader::create_function): Use
26084 build_decl instread of build_decl_stat.
26086 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
26088 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
26089 * config/i386/avx512dqintrin.h: Ditto.
26090 * config/i386/avx512fintrin.h: Ditto.
26091 * config/i386/i386-builtin-types.def: Add new types.
26092 * config/i386/i386.c: Handle new types.
26093 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
26094 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
26095 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
26096 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
26097 (__builtin_ia32_kshiftridi): New.
26098 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
26100 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
26104 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
26105 define to rs6000_init_stack_protect_guard.
26106 (rs6000_init_stack_protect_guard): New function.
26108 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
26109 Yunqiang Su <yunqiang.su@imgtec.com>
26111 * config.gcc (supported_defaults): Add madd4.
26112 (with_madd4): Add validation.
26113 (all_defaults): Add madd4.
26114 * config/mips/mips.opt (mmadd4): New option.
26115 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
26117 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
26119 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
26120 (ISA_HAS_FUSED_MADD4): Likewise.
26121 * doc/invoke.texi (-mmadd4): Document the new option.
26122 * doc/install.texi (--with-madd4): Document the new option.
26124 2017-01-19 Jiong Wang <jiong.wang@arm.com>
26126 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
26127 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
26128 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
26129 (aarch64_init_pauth_hint_builtins): New.
26130 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
26131 (aarch64_expand_builtin): Expand new builtins.
26133 2017-01-19 Jiong Wang <jiong.wang@arm.com>
26135 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
26136 * combine-stack-adj.c (no_unhandled_cfa): Handle
26137 REG_CFA_TOGGLE_RA_MANGLE.
26138 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
26139 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
26140 info for return address signing.
26141 (aarch64_expand_epilogue): Likewise.
26143 2017-01-19 Jiong Wang <jiong.wang@arm.com>
26145 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
26146 * config/aarch64/aarch64-protos.h
26147 (aarch64_return_address_signing_enabled): New declaration.
26148 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
26150 (aarch64_expand_prologue): Sign return address before it's pushed onto
26152 (aarch64_expand_epilogue): Authenticate return address fetched from
26154 (aarch64_override_options): Sanity check for ILP32 and ISA level.
26155 (aarch64_attributes): New function attributes for "sign-return-address".
26156 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
26157 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
26158 ("*do_return"): Generate combined instructions according to key index.
26159 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
26160 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
26162 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
26163 * config/aarch64/aarch64.opt (msign-return-address=): New.
26164 * doc/extend.texi (AArch64 Function Attributes): Documents
26165 "sign-return-address=".
26166 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
26168 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
26170 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
26171 overall option summary.
26173 2017-01-19 Jiong Wang <jiong.wang@arm.com>
26175 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
26176 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
26177 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
26178 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
26180 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
26182 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
26183 -mpower9-minmax by default for -mcpu=power9.
26184 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
26185 128-bit floating point.
26187 2017-01-20 Alan Modra <amodra@gmail.com>
26189 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
26190 optimizing for size.
26192 2017-01-20 Alan Modra <amodra@gmail.com>
26195 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
26196 for strcmp and strncmp from corresponding builtin decl.
26198 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
26200 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
26201 instead of i386/rtems-64.h.
26202 * config/i386/rtems-64.h: Remove.
26204 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
26208 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
26210 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
26212 2017-01-19 Tamar Christina <tamar.christina@arm.com>
26214 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
26215 Change int to HOST_WIDE_INT.
26216 * config/aarch64/aarch64-protos.h
26217 (aarch64_simd_gen_const_vector_dup): Likewise.
26218 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
26220 2017-01-19 David Malcolm <dmalcolm@redhat.com>
26222 * langhooks-def.h (lhd_type_for_size): New decl.
26223 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
26224 * langhooks.c (lhd_type_for_size): New function, taken from
26227 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
26229 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
26230 define_bypass for CR latency.
26231 (power9-cracked-alu): Update bypass latency and remove power9-branch.
26232 (power9-alu2): Add define_bypass for CR latency.
26234 (power9-mul): Update insn latency.
26235 (power9-mul-compare): Update insn latency, bypass latency and remove
26238 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26240 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
26242 * config/aarch64/aarch64.md
26243 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
26244 aarch64_nopcrelative_literal_loads.
26245 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
26247 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
26249 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
26250 TARGET_LOONGSON_3A.
26251 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
26253 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
26256 * config.gcc (supported_defaults): Add lxc1-sxc1.
26257 (with_lxc1_sxc1): Add validation.
26258 (all_defaults): Add lxc1-sxc1.
26259 * config/mips/mips.opt (mlxc1-sxc1): New option.
26260 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
26262 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
26263 __mips_no_lxc1_sxc1.
26264 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
26265 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
26266 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
26268 2017-01-19 Richard Biener <rguenther@suse.de>
26270 PR tree-optimization/72488
26271 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
26272 sure to restore SSA info.
26273 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
26275 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
26277 PR rtl-optimization/79121
26278 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
26279 of the inner type when shifting an extended value.
26281 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
26284 * symtab.c (symtab_node::equal_address_to): Fix comparing of
26285 interposable aliases.
26287 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
26290 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
26291 Use the evmergelohi instruction.
26292 (mov_si<mode>_e500_subreg4_2_le): Likewise.
26293 (mov_sitf_e500_subreg8_2_be): Likewise.
26294 (mov_sitf_e500_subreg12_2_le): Likewise.
26295 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
26296 (mov_si<mode>_e500_subreg4_2_be): Likewise.
26297 (mov_sitf_e500_subreg8_2_le): Likewise.
26298 (mov_sitf_e500_subreg12_2_be): Likewise.
26300 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26302 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
26303 attribute from vecsimple to vecperm.
26304 (altivec_vbpermq2): Likewise.
26306 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26309 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
26311 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26312 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
26313 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
26314 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
26315 case where N arg is SIZE_MAX.
26316 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
26317 (cmpstrsi): Add pattern.
26319 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
26321 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26322 __builtin_vec_revb builtins.
26323 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
26324 built-in functions to support generation of the ISA 3.0 XXBR<x>
26325 vector byte reverse instructions.
26326 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
26327 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
26328 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
26329 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
26330 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
26331 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
26332 (P9V_BUILTIN_VEC_REVB): Likewise.
26333 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
26334 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
26335 (p9_xxbrq_v16qi): Likewise.
26336 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
26337 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
26338 (p9_xxbrh_v8hi): Likewise.
26339 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
26340 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
26341 vec_revb built-in functions.
26343 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
26345 PR rtl-optimization/78952
26346 * config/i386/i386.md (any_extract): New code iterator.
26347 (*insvqi_2): Use any_extract for source operand.
26348 (*insvqi_3): Use any_shiftrt for source operand.
26350 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
26352 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
26354 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
26356 2017-01-18 Matthias Klose <doko@ubuntu.com>
26358 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
26360 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26362 * config/rs6000/altivec.h (vec_bperm): Change #define.
26363 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
26364 (altivec_vbpermq2): New define_insn.
26365 (altivec_vbpermd): Likewise.
26366 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
26367 function interface.
26368 (VBPERMD): Likewise.
26369 (VBPERM): New polymorphic function interface.
26370 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
26371 Add entries for P9V_BUILTIN_VEC_VBPERM.
26372 * doc/extend.texi: Add interfaces for vec_bperm.
26374 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26376 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
26377 first letter of error messages.
26378 (s390_resolve_overloaded_builtin): Likewise.
26379 * config/s390/s390.c (s390_expand_builtin): Likewise.
26380 (s390_invalid_arg_for_unprototyped_fn): Likewise.
26381 (s390_valid_target_attribute_inner_p): Likewise.
26382 * config/s390/s390.md ("tabort"): Likewise.
26384 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
26386 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
26387 (ISA_AVOID_DIV_HILO): New macro.
26388 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
26389 (ISA_HAS_DDIV): Likewise.
26391 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
26393 * doc/invoke.texi (fabi-version): Correct number of occurrences.
26395 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
26397 * doc/invoke.texi (fabi-version): Spelling fix.
26399 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
26402 * doc/invoke.texi (fabi-version): Mention mangling fix for
26405 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
26408 * doc/invoke.texi (fabi-version): Document discriminator mangling.
26410 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
26413 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
26414 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
26416 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
26418 (stack_protect_test): Ditto.
26419 * config/rs6000/rs6000.opt (mstack-protector-guard=,
26420 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
26422 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
26423 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
26424 -mstack-protector-guard-offset=.
26425 (RS/6000 and PowerPC Options): Ditto.
26427 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
26429 * config/i386/i386.h (MASK_CLASS_P): New define.
26430 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
26431 there are no registers from different register sets also when
26432 mask registers are used. Update function comment.
26433 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
26434 to (*k/*r) and (*k/*km) alternatives.
26436 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
26438 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
26439 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
26440 (EH_RETURN_HANDLER_RTX): New define.
26441 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
26442 Force frame pointer in EH return functions.
26443 (aarch64_expand_epilogue): Add barrier for eh_return.
26444 (aarch64_final_eh_return_addr): Remove.
26445 (aarch64_eh_return_handler_rtx): New function.
26446 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
26448 (aarch64_eh_return_handler_rtx): New prototype.
26450 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26452 * config/rs6000/altivec.h (vec_rlmi): New #define.
26453 (vec_vrlnm): Likewise.
26454 (vec_rlnm): Likewise.
26455 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
26456 (UNSPEC_VRLNM): Likewise.
26457 (VIlong): New mode iterator.
26458 (altivec_vrl<VI_char>mi): New define_insn.
26459 (altivec_vrl<VI_char>nm): Likewise.
26460 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
26462 (VRLDNM): Likewise.
26463 (RLNM): New polymorphic function entry.
26464 (VRLWMI): New monomorphic function entry.
26465 (VRLDMI): Likewise.
26466 (RLMI): New polymorphic function entry.
26467 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
26468 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
26469 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
26472 2017-01-17 Jakub Jelinek <jakub@redhat.com>
26475 * dwarf2out.c (field_byte_offset): Restore the
26476 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
26477 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
26478 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
26481 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
26484 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
26486 2017-01-17 Jakub Jelinek <jakub@redhat.com>
26489 * dwarf2out.c (add_data_member_location_attribute): For constant
26490 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
26491 instead of DW_AT_data_member_location, DW_AT_bit_offset and
26492 DW_AT_byte_size attributes.
26494 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
26496 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
26497 after forcing to constant memory when the code model is medium.
26499 2017-01-17 Julia Koval <julia.koval@intel.com>
26502 * config/i386/avx512fintrin.h
26503 (_mm512_i32gather_ps): Change __addr type to void const*.
26504 (_mm512_mask_i32gather_ps): Ditto.
26505 (_mm512_i32gather_pd): Ditto.
26506 (_mm512_mask_i32gather_pd): Ditto.
26507 (_mm512_i64gather_ps): Ditto.
26508 (_mm512_mask_i64gather_ps): Ditto.
26509 (_mm512_i64gather_pd): Ditto.
26510 (_mm512_mask_i64gather_pd): Ditto.
26511 (_mm512_i32gather_epi32): Ditto.
26512 (_mm512_mask_i32gather_epi32): Ditto.
26513 (_mm512_i32gather_epi64): Ditto.
26514 (_mm512_mask_i32gather_epi64): Ditto.
26515 (_mm512_i64gather_epi32): Ditto.
26516 (_mm512_mask_i64gather_epi32): Ditto.
26517 (_mm512_i64gather_epi64): Ditto.
26518 (_mm512_mask_i64gather_epi64): Ditto.
26519 (_mm512_i32scatter_ps): Change __addr type to void*.
26520 (_mm512_mask_i32scatter_ps): Ditto.
26521 (_mm512_i32scatter_pd): Ditto.
26522 (_mm512_mask_i32scatter_pd): Ditto.
26523 (_mm512_i64scatter_ps): Ditto.
26524 (_mm512_mask_i64scatter_ps): Ditto.
26525 (_mm512_i64scatter_pd): Ditto.
26526 (_mm512_mask_i64scatter_pd): Ditto.
26527 (_mm512_i32scatter_epi32): Ditto.
26528 (_mm512_mask_i32scatter_epi32): Ditto.
26529 (_mm512_i32scatter_epi64): Ditto.
26530 (_mm512_mask_i32scatter_epi64): Ditto.
26531 (_mm512_i64scatter_epi32): Ditto.
26532 (_mm512_mask_i64scatter_epi32): Ditto.
26533 (_mm512_i64scatter_epi64): Ditto.
26534 (_mm512_mask_i64scatter_epi64): Ditto.
26535 * config/i386/avx512pfintrin.h
26536 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
26537 (_mm512_mask_prefetch_i32gather_ps): Ditto.
26538 (_mm512_mask_prefetch_i64gather_pd): Ditto.
26539 (_mm512_mask_prefetch_i64gather_ps): Ditto.
26540 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
26541 (_mm512_prefetch_i32scatter_ps): Ditto.
26542 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
26543 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
26544 (_mm512_prefetch_i64scatter_pd): Ditto.
26545 (_mm512_prefetch_i64scatter_ps): Ditto.
26546 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
26547 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
26548 * config/i386/avx512vlintrin.h
26549 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
26550 (_mm_mmask_i32gather_ps): Ditto.
26551 (_mm256_mmask_i32gather_pd): Ditto.
26552 (_mm_mmask_i32gather_pd): Ditto.
26553 (_mm256_mmask_i64gather_ps): Ditto.
26554 (_mm_mmask_i64gather_ps): Ditto.
26555 (_mm256_mmask_i64gather_pd): Ditto.
26556 (_mm_mmask_i64gather_pd): Ditto.
26557 (_mm256_mmask_i32gather_epi32): Ditto.
26558 (_mm_mmask_i32gather_epi32): Ditto.
26559 (_mm256_mmask_i32gather_epi64): Ditto.
26560 (_mm_mmask_i32gather_epi64): Ditto.
26561 (_mm256_mmask_i64gather_epi32): Ditto.
26562 (_mm_mmask_i64gather_epi32): Ditto.
26563 (_mm256_mmask_i64gather_epi64): Ditto.
26564 (_mm_mmask_i64gather_epi64): Ditto.
26565 (_mm256_i32scatter_ps): Change __addr type to void*.
26566 (_mm256_mask_i32scatter_ps): Ditto.
26567 (_mm_i32scatter_ps): Ditto.
26568 (_mm_mask_i32scatter_ps): Ditto.
26569 (_mm256_i32scatter_pd): Ditto.
26570 (_mm256_mask_i32scatter_pd): Ditto.
26571 (_mm_i32scatter_pd): Ditto.
26572 (_mm_mask_i32scatter_pd): Ditto.
26573 (_mm256_i64scatter_ps): Ditto.
26574 (_mm256_mask_i64scatter_ps): Ditto.
26575 (_mm_i64scatter_ps): Ditto.
26576 (_mm_mask_i64scatter_ps): Ditto.
26577 (_mm256_i64scatter_pd): Ditto.
26578 (_mm256_mask_i64scatter_pd): Ditto.
26579 (_mm_i64scatter_pd): Ditto.
26580 (_mm_mask_i64scatter_pd): Ditto.
26581 (_mm256_i32scatter_epi32): Ditto.
26582 (_mm256_mask_i32scatter_epi32): Ditto.
26583 (_mm_i32scatter_epi32): Ditto.
26584 (_mm_mask_i32scatter_epi32): Ditto.
26585 (_mm256_i32scatter_epi64): Ditto.
26586 (_mm256_mask_i32scatter_epi64): Ditto.
26587 (_mm_i32scatter_epi64): Ditto.
26588 (_mm_mask_i32scatter_epi64): Ditto.
26589 (_mm256_i64scatter_epi32): Ditto.
26590 (_mm256_mask_i64scatter_epi32): Ditto.
26591 (_mm_i64scatter_epi32): Ditto.
26592 (_mm_mask_i64scatter_epi32): Ditto.
26593 (_mm256_i64scatter_epi64): Ditto.
26594 (_mm256_mask_i64scatter_epi64): Ditto.
26595 (_mm_i64scatter_epi64): Ditto.
26596 (_mm_mask_i64scatter_epi64): Ditto.
26597 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
26598 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
26599 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
26600 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
26601 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
26602 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
26603 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
26604 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
26605 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
26606 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
26607 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
26608 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
26609 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
26610 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
26611 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
26612 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
26613 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
26614 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
26615 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
26616 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
26617 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
26618 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
26619 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
26620 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
26621 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
26622 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
26623 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
26624 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
26625 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
26626 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
26627 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
26628 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
26629 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
26630 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
26631 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
26632 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
26633 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
26634 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
26635 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
26636 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
26637 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
26638 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
26639 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
26640 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
26641 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
26642 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
26643 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
26644 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
26645 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
26646 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
26647 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
26648 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
26649 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
26650 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
26651 definitions accordingly.
26653 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
26654 Kuan-Lin Chen <kuanlinchentw@gmail.com>
26657 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
26660 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
26663 * ira-conflicts.c (ira_build_conflicts): Update total conflict
26664 hard regs for inner regno.
26666 2017-01-17 Martin Liska <mliska@suse.cz>
26669 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
26670 assumption and add comment.
26672 2017-01-17 Nathan Sidwell <nathan@acm.org>
26674 * ipa-visibility.c (localize_node): New function, broken out of ...
26675 (function_and_variable_visibility): ... here. Call it.
26677 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
26679 PR middle-end/77445
26680 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
26681 correctly set frequency of oudgoing edge.
26682 (duplicate_thread_path): Fix profile updating.
26684 2017-01-17 Jakub Jelinek <jakub@redhat.com>
26687 * configure.ac: Add GCC_BASE_VER.
26688 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
26689 version from BASE-VER file.
26690 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
26691 (gcc.o): Depend on $(BASEVER).
26692 * common.opt (dumpfullversion): New option.
26693 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
26694 * doc/invoke.texi: Document -dumpfullversion.
26695 * doc/install.texi: Document --with-gcc-major-version-only.
26696 * configure: Regenerated.
26698 2017-01-17 Richard Biener <rguenther@suse.de>
26700 PR tree-optimization/71433
26701 * tree-vrp.c (register_new_assert_for): Merge same asserts
26702 on all incoming edges.
26703 (process_assert_insertions_for): Handle insertions at the
26706 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
26708 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
26709 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
26711 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
26714 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
26717 2017-01-17 Alan Modra <amodra@gmail.com>
26720 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
26721 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
26722 symbolic stack limit when pic.
26724 2017-01-16 Martin Sebor <msebor@redhat.com>
26726 PR tree-optimization/78608
26727 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
26729 2017-01-16 Jeff Law <law@redhat.com>
26732 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
26733 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
26734 for several include directories that may be relative to sysroot.
26735 * config/i386/x-mingw32 (gplus_includedir): Define.
26736 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
26737 (native_system_includedir): Likewise.
26738 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
26739 override if TARGET_SYSTEM_ROOT is defined.
26740 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
26742 PR tree-optimization/79090
26743 PR tree-optimization/33562
26744 PR tree-optimization/61912
26745 PR tree-optimization/77485
26746 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
26747 and computed trims into the dump file.
26749 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
26751 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
26753 2017-01-16 Jakub Jelinek <jakub@redhat.com>
26756 * gimplify.c (gimplify_init_constructor): If want_value and
26757 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
26761 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
26762 sequence. Formatting fixes.
26763 (doloop_optimize): Formatting fixes.
26766 * gcc.c (debug_level_greater_than_spec_func): New function.
26767 (static_spec_functions): Add debug-level-gt spec function.
26768 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
26770 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
26771 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
26772 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
26773 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
26774 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
26775 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
26777 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
26779 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
26780 QImode fixups to general and mask registers only.
26782 2017-01-16 Carl Love <cel@us.ibm.com>
26784 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
26785 for built-in functions
26786 vector signed char vec_nabs (vector signed char)
26787 vector signed short vec_nabs (vector signed short)
26788 vector signed int vec_nabs (vector signed int)
26789 vector signed long long vec_nabs (vector signed long long)
26790 vector float vec_nabs (vector float)
26791 vector double vec_nabs (vector double)
26792 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
26794 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
26795 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
26796 * doc/extend.texi: Update the documentation file for the new built-in
26799 2017-01-16 Martin Sebor <msebor@redhat.com>
26801 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
26804 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26806 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
26807 UNSPEC_VSX__XXSPLTD to require special splat handling.
26809 2017-01-16 David Malcolm <dmalcolm@redhat.com>
26812 * system.h: Poison strndup.
26814 2017-01-16 Alan Modra <amodra@gmail.com>
26817 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
26820 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
26822 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
26824 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
26826 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
26827 call recog here. Assert that INSN_CODE (insn) is non-negative.
26829 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
26832 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
26834 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
26835 in the currently scheduled RTL fragment.
26837 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
26839 PR rtl-optimization/78751
26840 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
26843 2017-01-14 Jeff Law <law@redhat.com>
26845 PR tree-optimization/79090
26846 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
26847 variable length stores.
26848 (compute_trims): Delete dead assignment to *trim_tail.
26849 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
26852 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
26854 PR rtl-optimization/78626
26855 PR rtl-optimization/78727
26856 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
26857 of a block, and split such blocks after everything else is finished.
26859 2017-01-14 Alan Modra <amodra@gmail.com>
26862 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
26863 target legitimate_combined_insn.
26864 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
26865 (rs6000_legitimate_combined_insn): New function.
26866 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
26868 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
26869 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
26870 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
26872 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
26874 * doc/frontends.texi (G++ and GCC): Remove references to Java.
26876 2017-01-13 Jeff Law <law@redhat.com>
26878 PR tree-optimization/33562
26879 PR tree-optimization/61912
26880 PR tree-optimization/77485
26881 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
26883 (delete_dead_assignment): Likewise.
26884 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
26885 statement to delete_dead_call and delete_dead_assignment.
26887 2017-01-13 David Malcolm <dmalcolm@redhat.com>
26890 * substring-locations.c (format_warning_va): Strengthen case 1 so
26891 that both endpoints of the substring must be within the format
26892 range for just the substring to be printed.
26894 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
26896 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
26897 * config/i386/i386.c (ix86_target_string): Add missing options
26898 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
26899 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
26900 flags_other and ix86_target_other to flags2_other. Display unknown
26902 (ix86_valid_target_attribute_inner_p): Add missing options and
26903 reorder options by implied ISAs, as in ix86_target_string.
26905 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
26907 * hash-table.h (hash_table::too_empty_p): New function.
26908 (hash_table::expand): Use it.
26909 (hash_table::traverse): Likewise.
26910 (hash_table::empty_slot): Use sizeof (value_type) instead of
26911 sizeof (PTR) to convert bytes to elements. Shrink the table
26912 if the current size is excessive for the current number of
26915 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
26917 * ira-costs.c (record_reg_classes): Break from the inner loop
26918 early once alt_fail is known to be true. Update outer loop
26919 handling accordingly.
26921 2017-01-13 Jeff Law <law@redhat.com>
26923 * tree-ssa-dse.c (decrement_count): New function.
26924 (increment_start_addr, maybe_trim_memstar_call): Likewise.
26925 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
26926 when we know the partially dead statement is a mem* function.
26928 PR tree-optimization/61912
26929 PR tree-optimization/77485
26930 * tree-ssa-dse.c: Include expr.h.
26931 (maybe_trim_constructor_store): New function.
26932 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
26934 PR tree-optimization/33562
26935 PR tree-optimization/61912
26936 PR tree-optimization/77485
26937 * doc/invoke.texi: Document new dse-max-object-size param.
26938 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
26939 * tree-ssa-dse.c: Include params.h.
26940 (dse_store_status): New enum.
26941 (initialize_ao_ref_for_dse): New, partially extracted from
26943 (valid_ao_ref_for_dse, normalize_ref): New.
26944 (setup_live_bytes_from_ref, compute_trims): Likewise.
26945 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
26946 (maybe_trim_partially_dead_store): Likewise.
26947 (maybe_trim_complex_store): Likewise.
26948 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
26949 Track what bytes live from the original store. Return tri-state
26950 for dead, partially dead or live.
26951 (dse_dom_walker): Add constructor, destructor and new private members.
26952 (delete_dead_call, delete_dead_assignment): New extracted from
26954 (dse_optimize_stmt): Make a member of dse_dom_walker.
26955 Use initialize_ao_ref_for_dse.
26957 PR tree-optimization/33562
26958 PR tree-optimization/61912
26959 PR tree-optimization/77485
26960 * sbitmap.h (bitmap_count_bits): Prototype.
26961 (bitmap_clear_range, bitmap_set_range): Likewise.
26962 * sbitmap.c (bitmap_clear_range): New function.
26963 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
26965 2017-01-13 Martin Liska <mliska@suse.cz>
26968 * function.c (set_cfun): Add new argument force.
26969 * function.h (set_cfun): Likewise.
26970 * ipa-inline-transform.c (inline_call): Use the function when
26971 strict alising from is dropped for function we inline to.
26973 2017-01-13 Richard Biener <rguenther@suse.de>
26975 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
26976 for dumping GIMPLE INTEGER_CSTs.
26978 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26980 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
26981 to 201112L since C++17.
26983 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
26986 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
26987 if -fsanitize=kernel-address is present.
26989 2017-01-13 Richard Biener <rguenther@suse.de>
26991 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
26992 as _Literal ( type ) number in case usual suffixes do not
26993 preserve all information.
26995 2017-01-13 Richard Biener <rguenther@suse.de>
26997 PR tree-optimization/77283
26998 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
26999 and ssa-iterators.h.
27000 (is_feasible_trace): Implement a cost model based on joiner
27003 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
27006 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
27007 char or short to __float128/_Float128 directly.
27009 2017-01-12 Martin Sebor <msebor@redhat.com>
27011 to -Wformat-overflow.
27012 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
27013 (min_bytes_remaining): Same.
27014 (get_string_length): Same.
27015 (format_string): Same.
27016 (format_directive): Same.
27018 (pass_sprintf_length::handle_gimple_call): Same.
27020 2017-01-12 Jakub Jelinek <jakub@redhat.com>
27022 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
27023 info.nowrite calls with no lhs that can't throw. Return bool
27024 whether gsi_remove has been called or not.
27025 (pass_sprintf_length::handle_gimple_call): Return bool whether
27026 try_substitute_return_value called gsi_remove. Formatting fix.
27027 (pass_sprintf_length::execute): Don't use gsi_remove if
27028 handle_gimple_call returned true.
27031 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
27032 be removed due to side-effects, don't remove following barrier nor
27033 turn the successor edge into fallthru edge.
27035 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27038 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
27039 element-reversing loads and stores as not swappable.
27041 2017-01-12 Nathan Sidwell <nathan@acm.org>
27042 Nicolai Stange <nicstange@gmail.com>
27044 * combine.c (try_combine): Don't ignore result of overlap checking
27045 loop. Combine overlap & asm check into single loop.
27047 2017-01-12 Richard Biener <rguenther@suse.de>
27049 * tree-pretty-print.c (dump_generic_node): Provide -gimple
27050 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
27052 2017-01-12 Richard Biener <rguenther@suse.de>
27054 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
27055 and TS_TARGET_OPTION directly derive from TS_BASE.
27056 * tree-core.h (tree_optimization_option): Derive from tree_base.
27057 (tree_target_option): Likewise.
27059 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
27061 * config/i386/i386.c (memory_address_length): Increase len
27062 only when rip_relative_addr_p returns false.
27064 2017-01-11 Julia Koval <julia.koval@intel.com>
27066 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
27067 (OPTION_MASK_ISA_SGX_SET): New.
27068 (ix86_handle_option): Handle OPT_msgx.
27069 * config.gcc: Added sgxintrin.h.
27070 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
27071 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
27072 * config/i386/i386.c (ix86_target_string): Add -msgx.
27074 (ix86_option_override_internal): Handle new options.
27075 (ix86_valid_target_attribute_inner_p): Add sgx.
27076 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
27077 * config/i386/i386.opt: Add msgx.
27078 * config/i386/sgxintrin.h: New file.
27079 * config/i386/x86intrin.h: Add sgxintrin.h.
27081 2017-01-11 Jakub Jelinek <jakub@redhat.com>
27084 * fold-const.c (maybe_nonzero_address): Return 1 for function
27086 (tree_single_nonzero_warnv_p): Don't handle function local objects
27090 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
27093 2017-01-11 David Malcolm <dmalcolm@redhat.com>
27096 * opts.c: Include "spellcheck.h"
27097 (struct string_fragment): New struct.
27098 (struct edit_distance_traits<const string_fragment &>): New
27100 (get_closest_sanitizer_option): New function.
27101 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
27103 2017-01-11 Jakub Jelinek <jakub@redhat.com>
27105 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
27107 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
27108 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
27109 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
27110 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
27111 for initial die_offset if dwarf_split_debug_info.
27112 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
27113 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
27115 (output_skeleton_debug_sections): Formatting fix. Use
27116 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
27117 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
27119 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
27121 * config/arm/cortex-a53.md: Add bypasses for
27122 cortex_a53_r2f_cvt.
27123 (cortex_a53_r2f): Only use for transfers.
27124 (cortex_a53_f2r): Likewise.
27125 (cortex_a53_r2f_cvt): Add reservation for conversions.
27126 (cortex_a53_f2r_cvt): Likewise.
27128 2017-01-11 Tamar Christina <tamar.christina@arm.com>
27130 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
27131 to all inlined functions, change static to extern.
27133 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
27136 * config/arm/arm.c (legitimize_pic_address): Handle reference to
27138 (arm_assemble_integer): Likewise.
27140 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
27142 * config.gcc: Use new awk script to check CPU, FPU and architecture
27143 parameters for --with-... options.
27144 * config/arm/parsecpu.awk: New file
27145 * config/arm/arm-cpus.in: New file.
27146 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
27148 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
27150 * config/arm/t-arm: Update dependency rules.
27151 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
27152 of processing .def files.
27153 * config/arm/genopt.sh: Deleted.
27154 * config/arm/gentune.sh: Deleted.
27155 * config/arm/arm-cores.def: Deleted.
27156 * config/arm/arm-arches.def: Deleted.
27157 * config/arm/arm-fpus.def: Deleted.
27158 * config/arm/arm-tune.md: Regenerated.
27159 * config/arm/arm-tables.opt: Regenerated.
27160 * config/arm/arm-cpu.h: New generated file.
27161 * config/arm/arm-cpu-data.h: New generated file.
27162 * config/arm/arm-cpu-cdata.h: New generated file.
27164 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
27167 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
27169 (input_varpool_node): Unpack dynamically_initialized bit.
27171 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
27173 PR rtl-optimization/79032
27174 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
27175 the alignment of the adjusted memory reference against that of MODE,
27176 instead of the alignment of the original memory reference.
27178 2017-01-11 Martin Jambor <mjambor@suse.cz>
27180 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
27182 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
27183 decorated functions.
27185 2017-01-11 Richard Biener <rguenther@suse.de>
27187 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
27188 set range/nonnull info for PHI results. Do not set it on
27189 stmts marked for removal.
27191 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
27193 * expr.c (store_field): In the bitfield case, fetch the return value
27194 from the registers before applying a single big-endian adjustment.
27195 Always do a final load for a BLKmode value not larger than a word.
27197 2017-01-10 David Malcolm <dmalcolm@redhat.com>
27200 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
27201 that we correctly handle column numbers greater than
27202 LINE_MAP_MAX_COLUMN_NUMBER.
27204 2017-01-10 Martin Sebor <msebor@redhat.com>
27206 PR middle-end/78245
27207 * gimple-ssa-sprintf.c (get_destination_size): Call
27208 {init,fini}object_sizes.
27209 * tree-object-size.c (addr_object_size): Adjust.
27210 (pass_through_call): Adjust.
27211 (pass_object_sizes::execute): Adjust.
27212 * tree-object-size.h (fini_object_sizes): Declare.
27214 2017-01-10 Martin Sebor <msebor@redhat.com>
27216 PR tree-optimization/78775
27217 * builtins.c (get_size_range): Move...
27218 * calls.c: ...to here.
27219 (alloc_max_size): Accept zero argument.
27220 (operand_signed_p): Remove.
27221 (maybe_warn_alloc_args_overflow): Call get_size_range.
27222 * calls.h (get_size_range): Declare.
27224 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
27226 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
27227 from TI's devices.csv file as of September 2016.
27228 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
27230 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
27232 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
27233 * doc/invoke.texi: Likewise.
27234 * doc/md.texi: Likewise.
27235 * doc/objc.texi: Likewise.
27237 2017-01-10 Joshua Conner <joshconner@google.com>
27239 * config/arm/fuchsia-elf.h: New file.
27240 * config/fuchsia.h: New file.
27241 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
27242 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
27244 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
27246 2016-01-10 Richard Biener <rguenther@suse.de>
27248 PR tree-optimization/79034
27249 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
27250 Propagate out degenerate PHIs in the joiner.
27252 2017-01-10 Martin Liska <mliska@suse.cz>
27254 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
27255 (sort_congruence_classes_by_decl_uid): Likewise.
27256 (sort_congruence_class_groups_by_decl_uid): Likewise.
27257 (sem_item_optimizer::merge_classes): Sort class, groups in these
27258 classes and members in the groups by DECL_UID of declarations.
27259 This would make merge operations stable.
27261 2017-01-10 Martin Liska <mliska@suse.cz>
27263 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
27264 usage of m_classes_vec.
27265 (sem_item_optimizer::~sem_item_optimizer): Likewise.
27266 (sem_item_optimizer::get_group_by_hash): Likewise.
27267 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
27268 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
27269 (sem_item_optimizer::verify_classes): Likewise.
27270 (sem_item_optimizer::process_cong_reduction): Likewise.
27271 (sem_item_optimizer::dump_cong_classes): Likewise.
27272 (sem_item_optimizer::merge_classes): Likewise.
27273 * ipa-icf.h (congruence_class_hash): Rename from
27274 congruence_class_group_hash. Remove declaration of m_classes_vec.
27276 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
27278 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
27279 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
27280 * config.gcc: Add avx512vpopcntdqintrin.h.
27281 * config/i386/avx512vpopcntdqintrin.h: New.
27282 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
27283 * config/i386/i386-builtin-types.def: Add new types.
27284 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
27285 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
27286 __builtin_ia32_vpopcountq_v8di_mask): New.
27287 * config/i386/i386-c.c (ix86_target_macros_internal): Define
27288 __AVX512VPOPCNTDQ__.
27289 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
27290 (PTA_AVX512VPOPCNTDQ): Define.
27291 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
27292 TARGET_AVX512VPOPCNTDQ_P): Define.
27293 * config/i386/i386.opt: Add mavx512vpopcntdq.
27294 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
27295 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
27297 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
27299 PR middle-end/77484
27300 * predict.def (PRED_CALL): Set to 67.
27302 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
27304 * expr.c (store_field): In the bitfield case, if the value comes from
27305 a function call and is of an aggregate type returned in registers, do
27306 not modify the field mode; extract the value in all cases if the mode
27307 is BLKmode and the size is not larger than a word.
27309 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
27312 * config/i386/cpuid.h: Fix undefined behavior.
27314 2017-01-04 Jeff Law <law@redhat.com>
27316 PR tree-optimization/79007
27317 PR tree-optimization/67955
27318 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
27319 conservative for pt.null when flag_non_call_exceptions is on.
27321 2017-01-09 Jakub Jelinek <jakub@redhat.com>
27323 PR translation/79019
27324 PR translation/79020
27325 * params.def (PARAM_INLINE_MIN_SPEEDUP,
27326 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
27327 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
27329 * config/avr/avr.opt (maccumulate-args): Likewise.
27330 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
27331 * common.opt (freport-bug): Likewise.
27332 * cif-code.def (CIF_FINAL_ERROR): Likewise.
27333 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
27334 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
27335 translatable string.
27336 * config/i386/i386.c (function_value_32): Likewise.
27337 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
27338 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
27340 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
27341 * common/config/msp430/msp430-common.c (msp430_handle_option):
27343 * symtab.c (symtab_node::verify_base): Likewise.
27344 * opts.c (set_debug_level): Likewise.
27345 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
27346 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
27347 missing whitespace to translatable strings.
27348 * config/avr/avr.md (bswapsi2): Fix typo in comment.
27349 * config/sh/superh.h: Likewise.
27350 * config/i386/xopintrin.h: Likewise.
27351 * config/i386/znver1.md: Likewise.
27352 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
27353 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
27354 * double-int.h (struct double_int): Likewise.
27355 * double-int.c (div_and_round_double): Likewise.
27356 * wide-int.cc: Likewise.
27357 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
27358 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
27359 * cfgcleanup.c (crossjumps_occured): Renamed to ...
27360 (crossjumps_occurred): ... this.
27361 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
27364 PR tree-optimization/78899
27365 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
27366 returning bool return struct loop *, NULL for failure and the new
27368 (versionable_outer_loop_p): Don't version outer loop if it has
27369 dont_vectorized bit set.
27370 (tree_if_conversion): When versioning outer loop, ensure
27371 tree_if_conversion is performed also on the inner loop of the
27372 non-vectorizable outer loop copy.
27373 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
27374 LOOP_VECTORIZED in inner loop of the scalar outer loop and
27375 prevent vectorization of it.
27376 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
27377 the outer loop vectorization of the non-scalar version is attempted
27378 before vectorization of the inner loop in scalar version. If
27379 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
27380 vectorization of its inner loop.
27381 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
27382 has 2 inner loops, rename also on edges from bb whose single pred
27383 is outer_loop->header. Fix typo in function comment.
27385 2017-01-09 Martin Sebor <msebor@redhat.com>
27388 * asan.c (asan_emit_stack_protection): Increase local buffer size
27389 to avoid snprintf truncation warning.
27391 2017-01-09 Andrew Pinski <apinski@cavium.com>
27393 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
27394 to reference thunderx2t99 for the tuning structure
27395 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
27397 (thunderx2t99_extra_costs): This.
27398 * config/aarch64/aarch64-tune.md: Regenerate.
27399 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
27400 (vulcan_addrcost_table): This.
27401 (vulcan_regmove_cost): Rename to ...
27402 (thunderx2t99_regmove_cost): This.
27403 (vulcan_vector_cost): Rename to ...
27404 (thunderx2t99_vector_cost): this.
27405 (vulcan_branch_cost): Rename to ...
27406 (thunderx2t99_branch_cost): This.
27407 (vulcan_tunings): Rename to ...
27408 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
27409 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
27411 2017-01-09 Martin Jambor <mjambor@suse.cz>
27415 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
27416 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
27417 (propagate_vr_accross_jump_function): Use the above function for all
27418 value range computations for pass-through jump functions and type
27419 converasion from explicit value range values.
27420 (ipcp_propagate_stage): Do not attempt to deduce types of formal
27421 parameters from TYPE_ARG_TYPES.
27422 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
27423 (ipa_write_node_info): Stream type of the actual argument.
27424 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
27426 2017-01-09 Martin Liska <mliska@suse.cz>
27429 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
27430 (lookup_compiler): Do not show error message with have_E.
27432 2017-01-09 Jakub Jelinek <jakub@redhat.com>
27434 PR tree-optimization/78938
27435 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
27436 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
27437 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
27438 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
27441 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27443 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
27446 2017-01-09 Richard Biener <rguenther@suse.de>
27448 PR tree-optimization/78997
27449 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
27450 name condition properly.
27452 2017-01-09 Richard Biener <rguenther@suse.de>
27455 * dwarf2out.c (is_cxx): New overload with context.
27456 (is_naming_typedef_decl): Use it.
27458 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
27460 * invoke.texi (Option Summary): Correct spacing in option lists
27461 and add line breaks to fix over-long lines.
27463 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
27465 PR middle-end/17660
27467 * extend.texi (Common Variable Attributes): Add xref to GCC
27468 Internals manual to explain mode attribute keywords.
27470 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
27473 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
27474 and Preprocessor Options.
27475 (Options for Linking): Document -pthread here....
27476 (RS/6000 and PowerPC Options): ...not here.
27477 (Solaris 2 Options): ...or here.
27478 * doc/cppopts.texi: Document -pthread.
27480 2017-01-08 Martin Sebor <msebor@redhat.com>
27482 PR middle-end/77708
27483 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
27484 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
27485 New member functions.
27486 (format_directive): Used them.
27488 (pass_sprintf_length::handle_gimple_call): Same.
27489 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
27490 to avoid truncation for any argument.
27491 (extract_affine_mul): Same.
27492 * tree.c (get_file_function_name): Same.
27494 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
27496 PR middle-end/77484
27497 * predict.def (PRED_INDIR_CALL): Set to 86.
27499 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
27501 PR preprocessor/54124
27502 * doc/cppopts.texi: Reformat -d subtable to list the full name
27503 of the options. Add cross-reference to the docs for the general
27504 compiler -d options.
27505 * doc/invoke.texi (Developer Options): Add cross-reference to the
27506 preprocessor-specific -d option documentation.
27508 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
27510 PR preprocessor/13498
27511 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
27512 redudant material, and reflect new command-line options.
27513 (System Headers): Likewise.
27515 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
27517 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
27518 -isystem, and -idirafter. Copy-edit.
27519 * doc/cppopts.texi: Copy-edit. Remove contradiction about
27520 default for -ftrack-macro-expansion. Delete obsolete and
27521 badly-formatted implementation details about -fdebug-cpp output.
27522 * doc/cppwarnopts.texi: Copy-edit.
27524 2017-01-07 David Malcolm <dmalcolm@redhat.com>
27527 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
27528 that the transition from a max line width >= 1<<10 to narrower
27529 lines works correctly.
27531 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
27533 * doc/options.texi (PerFunction): New.
27534 * opt-functions.awk (switch_flags): Map both Optimization and
27535 PerFunction to CL_OPTIMIZATION.
27536 * opth-gen.awk: Test for PerFunction flag along with
27538 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
27539 it only when the latter is present. Skip those that don't in
27540 the hash function generator.
27541 * common.opt (fvar-tracking): Mark as PerFunction instead of
27543 (fvar-tracking-assignments): Likewise.
27544 (fvar-tracking-assignments-toggle): Likewise.
27545 (fvar-tracking-uninit): Likewise.
27547 2017-01-07 Jakub Jelinek <jakub@redhat.com>
27549 PR translation/79018
27550 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
27553 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
27556 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
27557 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
27558 TARGET_LONG_JUMP_TABLE_OFFSETS.
27559 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
27560 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
27561 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
27562 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
27563 * config/m68k/m68k.md (tablejump expander): Likewise.
27564 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
27565 TARGET_LONG_JUMP_TABLE_OFFSETS.
27566 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
27567 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
27569 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
27570 David Holsgrove <david.holsgrove@xilinx.com>
27572 * common/config/microblaze/microblaze-common.c
27573 (TARGET_EXCEPT_UNWIND_INFO): Remove.
27574 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
27576 * config/microblaze/microblaze.c (microblaze_must_save_register)
27577 (microblaze_expand_epilogue, microblaze_return_addr): Handle
27579 (microblaze_eh_return): New function.
27580 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
27581 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
27582 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
27583 * config/microblaze/microblaze.md (eh_return): New pattern.
27585 2017-01-06 Jakub Jelinek <jakub@redhat.com>
27587 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
27588 GCC_DIAGNOSTIC_STRINGIFY): Define.
27590 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
27592 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
27594 * config/arm/arm.md (<mcrr>): New.
27596 * config/arm/arm.c (arm_arch5te): New.
27597 (arm_option_override): Set arm_arch5te.
27598 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
27600 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
27601 (arm_mcrr_qualifiers): ... this. New.
27602 (MRRC_QUALIFIERS): Define to...
27603 (arm_mrrc_qualifiers): ... this. New.
27604 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
27606 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
27607 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
27608 (MRRCI, mrrc, MRRC): New.
27609 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
27610 VUNSPEC_MRRC2): New.
27612 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
27614 * config/arm/arm.md (<mcr>): New.
27616 * config/arm/arm.c (arm_coproc_builtin_available): Add
27617 support for mcr, mrc, mcr2 and mrc2.
27618 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
27619 (arm_mcr_qualifiers): ... this. New.
27620 (MRC_QUALIFIERS): Define to ...
27621 (arm_mrc_qualifiers): ... this. New.
27622 (MCR_QUALIFIERS): Define to ...
27623 (arm_mcr_qualifiers): ... this. New.
27624 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
27626 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
27627 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
27628 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
27629 VUNSPEC_MRC2): New.
27631 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
27633 * config/arm/arm.md (*ldc): New.
27637 * config/arm/arm.c (arm_coproc_builtin_available): Add
27638 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
27639 (arm_coproc_ldc_stc_legitimate_address): New.
27640 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
27641 'qualifier_const_pointer'.
27642 (LDC_QUALIFIERS): Define to...
27643 (arm_ldc_qualifiers): ... this. New.
27644 (STC_QUALIFIERS): Define to...
27645 (arm_stc_qualifiers): ... this. New.
27646 * config/arm/arm-protos.h
27647 (arm_coproc_ldc_stc_legitimate_address): New.
27648 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
27649 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
27650 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
27651 stc2, stcl, stc2l): New.
27652 * config/arm/constraints.md (Uz): New.
27653 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
27654 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
27655 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
27656 VUNSPEC_STC2L): New.
27658 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
27660 * config/arm/arm.md (<cdp>): New.
27661 * config/arm/arm.c (neon_const_bounds): Rename this ...
27662 (arm_const_bounds): ... this.
27663 (arm_coproc_builtin_available): New.
27664 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
27665 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
27666 (CDP_QUALIFIERS): Define to...
27667 (arm_cdp_qualifiers): ... this. New.
27669 (arm_expand_builtin_args): Add case for 6 arguments.
27670 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
27671 (arm_const_bounds): ... this.
27672 (arm_coproc_builtin_available): New.
27673 * config/arm/arm_acle.h (__arm_cdp): New.
27675 * config/arm/arm_acle_builtins.def (cdp): New.
27677 * config/arm/iterators.md (CDPI,CDP,cdp): New.
27678 * config/arm/neon.md: Rename all 'neon_const_bounds' to
27679 'arm_const_bounds'.
27680 * config/arm/types.md (coproc): New.
27681 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
27682 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
27683 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
27684 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
27686 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
27688 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
27689 (UBINOP_QUALIFIERS): New.
27691 (acle_builtin_data): New. Change comment.
27692 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
27693 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
27694 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
27695 arm_acle_builtins.def.
27696 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
27697 (arm_init_acle_builtins): New.
27698 (CRC32_BUILTIN): Remove.
27699 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
27700 crc32cb, crc32ch and crc32cw.
27701 (arm_init_crc32_builtins): Remove.
27702 (arm_init_builtins): Use arm_init_acle_builtins rather
27703 than arm_init_crc32_builtins.
27704 (arm_expand_acle_builtin): New.
27705 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
27706 * config/arm/arm_acle_builtins.def: New.
27708 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
27710 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
27711 (arm_builtin_datum): ... this.
27712 (arm_init_neon_builtin): Rename to ...
27713 (arm_init_builtin): ... this. Add a new parameters PREFIX
27714 and USE_SIG_IN_NAME.
27715 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
27716 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
27717 'arm_builtin_datum'.
27718 (arm_init_vfp_builtins): Likewise.
27719 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
27720 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
27721 (arm_expand_neon_args): Rename to ...
27722 (arm_expand_builtin_args): ... this. Rename builtin_arg
27723 enum values and differentiate between ARG_BUILTIN_MEMORY
27724 and ARG_BUILTIN_NEON_MEMORY.
27725 (arm_expand_neon_builtin_1): Rename to ...
27726 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
27727 values, arm_expand_builtin_args and add bool parameter NEON.
27728 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
27729 (arm_expand_vfp_builtin): Likewise.
27730 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
27732 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
27734 PR middle-end/77484
27735 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
27736 * predict.c (tree_estimate_probability_bb): Reverse direction of
27737 polymorphic call predictor.
27739 2017-01-06 David Malcolm <dmalcolm@redhat.com>
27741 * passes.c (execute_one_pass): Split out pass-skipping logic into...
27742 (determine_pass_name_match): ...this new function and...
27743 (should_skip_pass_p): ...this new function.
27745 2017-01-06 Nathan Sidwell <nathan@acm.org>
27747 * ipa-visibility.c (function_and_variable_visibility): Reformat
27748 comments and long lines. Remove extrneous if.
27749 * symtab.c (symtab_node::make_decl_local): Fix code format.
27750 (symtab_node::set_section_for_node): Fix comment typo.
27752 2017-01-06 Martin Liska <mliska@suse.cz>
27755 * lra-constraints.c: Rename invariant to lra_invariant.
27756 * predict.c (set_even_probabilities): Initialize e to NULL.
27758 2017-01-05 Martin Sebor <msebor@redhat.com>
27760 PR tree-optimization/78910
27761 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
27762 (format_integer): Correct off-by-one error in the handling
27763 of precision with negative numbers in signed conversions..
27765 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
27767 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
27769 2017-01-05 Jakub Jelinek <jakub@redhat.com>
27771 PR tree-optimization/71016
27772 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
27773 factor_out_conditional_conversion. Formatting fix.
27774 (factor_out_conditional_conversion): Add cond_stmt argument.
27775 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
27776 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
27779 2017-01-05 David Malcolm <dmalcolm@redhat.com>
27781 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
27782 read-rtl-function.o, and selftest-rtl.o.
27783 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
27784 (selftest::aarch64_test_loading_full_dump): New function.
27785 (selftest::aarch64_run_selftests): New function.
27786 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
27787 selftest::aarch64_run_selftests.
27788 * config/i386/i386.c
27789 (selftest::ix86_test_loading_dump_fragment_1): New function.
27790 (selftest::ix86_test_loading_call_insn): New function.
27791 (selftest::ix86_test_loading_full_dump): New function.
27792 (selftest::ix86_test_loading_unspec): New function.
27793 (selftest::ix86_run_selftests): Call the new functions.
27794 * emit-rtl.c (maybe_set_max_label_num): New function.
27795 * emit-rtl.h (maybe_set_max_label_num): New decl.
27796 * function.c (instantiate_decls): Guard call to
27797 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
27798 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
27800 * gensupport.c (gen_reader::gen_reader): Pass "false"
27801 for new "compact" param of rtx_reader.
27802 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
27803 rather than an empty string for NULL strings.
27804 * read-md.c: Potentially include config.h rather than bconfig.h.
27805 Wrap include of errors.h with #ifdef GENERATOR_FILE.
27806 (have_error): New global, copied from errors.c.
27807 (md_reader::read_name): Rename to...
27808 (md_reader::read_name_1): ...this, adding "out_loc" param,
27809 and converting "missing name or number" to returning false, rather
27811 (md_reader::read_name): Reimplement in terms of read_name_1.
27812 (md_reader::read_name_or_nil): New function.
27813 (md_reader::read_string): Handle "(nil)" by returning NULL.
27814 (md_reader::md_reader): Add new param "compact".
27815 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
27816 (md_reader::read_file): New method.
27817 * read-md.h (md_reader::md_reader): Add new param "compact".
27818 (md_reader::read_file): New method.
27819 (md_reader::is_compact): New accessor.
27820 (md_reader::read_name): Convert return type from void to file_location.
27821 (md_reader::read_name_or_nil): New decl.
27822 (md_reader::read_name_1): New decl.
27823 (md_reader::m_compact): New field.
27824 (noop_reader::noop_reader): Pass "false" for new "compact" param
27826 (rtx_reader::rtx_reader): Add new "compact" param.
27827 (rtx_reader::read_rtx_operand): Make virtual and convert return
27828 type from void to rtx.
27829 (rtx_reader::read_until): New decl.
27830 (rtx_reader::handle_any_trailing_information): New virtual function.
27831 (rtx_reader::postprocess): New virtual function.
27832 (rtx_reader::finalize_string): New virtual function.
27833 (rtx_reader::m_in_call_function_usage): New field.
27834 (rtx_reader::m_reuse_rtx_by_id): New field.
27835 * read-rtl-function.c: New file.
27836 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
27837 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
27838 (selftest::verify_three_block_rtl_cfg): New decl.
27839 * read-rtl-function.h: New file.
27840 * read-rtl.c: Potentially include config.h rather than bconfig.h.
27841 For host, include function.h, memmodel.h, and emit-rtl.h.
27842 (one_time_initialization): New function.
27843 (struct compact_insn_name): New struct.
27844 (compact_insn_names): New array.
27845 (find_code): Handle insn codes in compact dumps.
27846 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
27847 (bind_subst_iter_and_attr): Likewise.
27848 (add_condition_to_string): Likewise.
27849 (add_condition_to_rtx): Likewise.
27850 (apply_attribute_uses): Likewise.
27851 (add_current_iterators): Likewise.
27852 (apply_iterators): Likewise.
27853 (initialize_iterators): Guard usage of apply_subst_iterator with
27854 #ifdef GENERATOR_FILE.
27855 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
27856 (md_reader::read_mapping): Likewise.
27857 (add_define_attr_for_define_subst): Likewise.
27858 (add_define_subst_attr): Likewise.
27859 (read_subst_mapping): Likewise.
27860 (check_code_iterator): Likewise.
27861 (rtx_reader::read_rtx): Likewise. Move one-time initialization
27863 (one_time_initialization): New function.
27864 (rtx_reader::read_until): New method.
27865 (read_flags): New function.
27866 (parse_reg_note_name): New function.
27867 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
27868 Handle reuse_rtx ids.
27869 Wrap iterator lookup within #ifdef GENERATOR_FILE.
27870 Add parsing support for RTL dumps, mirroring the special-cases in
27871 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
27872 values, and calling handle_any_trailing_information.
27873 (rtx_reader::read_rtx_operand): Convert return type from void
27874 to rtx, returning return_rtx. Handle case 'e'. Call
27875 finalize_string on XSTR and XTMPL fields.
27876 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
27877 "(nil)" values were omitted. Call the postprocess vfunc on the
27879 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
27880 class ctor. Initialize m_in_call_function_usage. Call
27881 one_time_initialization.
27882 * rtl-tests.c (selftest::test_uncond_jump): Call
27883 set_new_first_and_last_insn.
27884 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
27885 * selftest-rtl.c: New file.
27886 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
27887 (selftest::get_insn_by_uid): New decl.
27888 * selftest-run-tests.c (selftest::run_tests): Call
27889 read_rtl_function_c_tests.
27890 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
27891 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
27894 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
27896 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
27897 operands in a special way. Assert that pos+len <= mode precision.
27899 2017-01-05 Jakub Jelinek <jakub@redhat.com>
27901 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
27902 3 argument Alias with unlimited for the negative form.
27903 (fno-vect-cost-model): Removed.
27905 2017-01-05 Martin Liska <mliska@suse.cz>
27907 * hsa-gen.c (gen_hsa_divmod): New function.
27908 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
27910 2017-01-05 Martin Liska <mliska@suse.cz>
27913 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
27916 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27918 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
27919 small constant length operands.
27921 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27923 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
27924 between loop iterations.
27926 2017-01-05 Martin Liska <mliska@suse.cz>
27929 * gimplify.c (gimplify_decl_expr): Compare to
27930 asan_poisoned_variables instread of checking flags.
27931 (gimplify_target_expr): Likewise.
27932 (gimplify_expr): Likewise.
27933 (gimplify_function_tree): Conditionally initialize
27934 asan_poisoned_variables.
27936 2017-01-04 Jeff Law <law@redhat.com>
27938 PR tree-optimizatin/78812
27939 * rtl.h (contains_mem_rtx_p): Prototype.
27940 * ifcvt.c (containts_mem_rtx_p): Move from here to...
27941 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
27942 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
27943 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
27944 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
27946 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
27948 * input.c (assert_char_at_range): Default-initialize actual_range.
27950 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
27952 * df-scan.c (df_ref_create_structure): Make regno unsigned,
27953 to match the caller.
27955 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
27957 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
27958 insns after final jump in test to emit dummy move.
27960 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
27962 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
27963 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
27965 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
27967 * multiple_target.c (create_dispatcher_calls): Init e_next.
27968 * tree-ssa-loop-split.c (split_loop): Init border.
27969 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
27972 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
27977 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
27978 (altivec_register_operand): Do not return true if the operand
27979 contains a SUBREG mixing SImode and SFmode.
27980 (vsx_register_operand): Likewise.
27981 (vsx_reg_sfsubreg_ok): New predicate.
27982 (vfloat_operand): Do not return true if the operand contains a
27983 SUBREG mixing SImode and SFmode.
27984 (vint_operand): Likewise.
27985 (vlogical_operand): Likewise.
27986 (gpc_reg_operand): Likewise.
27987 (int_reg_operand): Likewise.
27988 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
27989 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
27990 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
27992 (rs6000_emit_move_si_sf_subreg): New helper function.
27993 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
27994 fixup SUBREGs involving SImode and SFmode.
27995 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
27996 numbers for the new peephole2 optimization.
27997 (peephole2 for SFmode unions): New peephole2 to optimize cases in
27998 the GLIBC math library that do AND/IOR/XOR operations on single
27999 precision floating point.
28000 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
28001 target macros to say whether we need to avoid SUBREGs mixing
28003 (TARGET_ALLOW_SF_SUBREG): Likewise.
28004 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
28005 (UNSPEC_SI_FROM_SF): Likewise.
28006 (iorxor): Change spacing.
28007 (and_ior_xor): New iterator for AND, IOR, and XOR.
28008 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
28009 (movdi_from_sf_zero_ext): Likewise.
28010 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
28011 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
28012 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
28013 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
28014 (fms<mode>4): Likewise.
28015 (fnma<mode>4): Likewise.
28016 (fnms<mode>4): Likewise.
28017 (nfma<mode>4): Likewise.
28018 (nfms<mode>4): Likewise.
28020 2017-01-04 Marek Polacek <polacek@redhat.com>
28023 * doc/invoke.texi: Document -Wpointer-compare.
28025 2017-01-04 Jakub Jelinek <jakub@redhat.com>
28027 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
28030 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
28031 descriptions for -gdwarf-5 and emit them as uleb128 instead of
28034 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
28037 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
28038 documentation of the powerpc_popcntb_ok attribute.
28039 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
28040 code to issue warning messages if a requested CPU configuration is
28041 not supported by the binary (assembler and loader) toolchain.
28042 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
28043 made to define a built-in function that has been disabled.
28044 (paired_init_builtins): Add assertion to prevent ICE if attempt is
28045 made to define a built-in function that has been disabled.
28046 (altivec_init_builtins): Add comment explaining why definition
28047 of the DST built-in functions is not preceded by an assertion
28048 check. Add assertions to prevent ICE if attempts are made to
28049 define an altivec predicate or an abs* built-in function that has
28051 (htm_init_builtins): Add comment explaining why definition of the
28052 htm built-in functions is not preceded by an assertion check.
28054 2017-01-04 Jeff Law <law@redhat.com>
28056 PR tree-optimizatin/67955
28057 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
28058 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
28059 the points-to solution does not include pt_null. Use DECL_PT_UID
28062 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
28064 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
28065 Use gen_int_mode instead of gen_lopwart for const_int operands.
28067 2017-01-04 Jakub Jelinek <jakub@redhat.com>
28069 PR tree-optimization/71563
28070 * match.pd: Simplify X << Y into X if Y is known to be 0 or
28071 out of range value - has low bits known to be zero.
28073 2017-01-04 Alan Modra <amodra@gmail.com>
28075 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
28076 * configure: Regenerate.
28077 * config.in: Regenerate.
28079 2017-01-04 Jakub Jelinek <jakub@redhat.com>
28082 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
28083 a substring of the message, but strcmp with the whole message. Ifdef
28084 ENABLE_NLS, translate the message first using dgettext.
28086 2017-01-03 Jeff Law <law@redhat.com>
28088 PR tree-optimizatin/78856
28089 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
28090 (mark_threaded_blocks): Remove code to truncate thread paths that
28091 cross multiple loop headers. Instead invalidate the cached loop
28092 iteration information and handle case of a thread path walking
28093 into an irreducible region.
28095 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
28098 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
28099 assertions. Add support for doing the signbit if the IEEE 128-bit
28100 floating point value is in a GPR.
28101 * config/rs6000/rs6000.md (Fsignbit): Delete.
28102 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
28103 Update the length attribute if the value is in a GPR.
28104 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
28105 the sign or zero extension instruction, since the value is always 0/1.
28106 (signbit<mode>2_dm2): Delete using <Fsignbit>.
28109 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
28110 extracting SImode to a GPR register so that we can generate a
28111 store, limit the vector to be in a traditional Altivec register
28112 for the vextuwrx instruction.
28114 2017-01-03 Ian Lance Taylor <iant@google.com>
28116 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
28118 2017-01-03 Martin Sebor <msebor@redhat.com>
28120 PR tree-optimization/78696
28121 * gimple-ssa-sprintf.c (format_floating): Correct handling of
28122 precision. Use MPFR for %f for greater fidelity. Correct handling
28124 (pass_sprintf_length::compute_format_length): Set width and precision
28125 specified by asrerisk to void_node for vararg functions.
28126 (try_substitute_return_value): Adjust dump output.
28128 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
28130 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
28132 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
28134 * doc/invoke.texi (SPARC options): Document -mlra as the default.
28135 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
28136 -mlra/-mno-lra was passed to the compiler.
28138 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
28140 PR rtl-optimization/65618
28141 * emit-rtl.c (try_split): Move initialization of "before" and
28142 "after" to just before the call to emit_insn_after_setloc.
28144 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
28146 * doc/md.texi (Standard Names): Remove reference to Java frontend.
28148 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
28150 * dwarf2out.c (gen_enumeration_type_die): When
28151 -gno-strict-dwarf, add a DW_AT_encoding attribute.
28153 2017-01-03 Jakub Jelinek <jakub@redhat.com>
28155 PR tree-optimization/78965
28156 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
28157 Change first argument from const call_info & to call_info &. For %n
28158 set info.nowrite to false.
28160 PR middle-end/78901
28161 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
28162 possibly throwing calls.
28164 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
28165 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
28166 and fns handling, rather than in a separate case SSA_NAME.
28168 2017-01-02 Jeff Law <law@redhat.com>
28170 * config/darwin-driver.c (darwin_driver_init): Const-correctness
28171 fixes for first_period and second_period variables.
28173 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
28176 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
28177 (*insvqi_1): New insn pattern.
28178 (*insvqi_1_mem_rex64): Ditto.
28179 (*insvqi_2): Ditto.
28180 (*insvqi_3): Rename from *insvqi.
28182 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
28184 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
28186 * doc/cfg.texi (Edges): Remove reference to Java.
28187 (Maintaining the CFG): Ditto.
28189 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
28191 PR middle-end/77674
28192 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
28193 transparent aliases.
28195 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
28197 PR middle-end/77484
28198 * predict.def (PRED_CALL): Update hitrate.
28199 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
28200 * predict.c (tree_estimate_probability_bb): Split CALL predictor
28201 into direct/indirect/polymorphic variants.
28203 2017-01-01 Jakub Jelinek <jakub@redhat.com>
28205 Update copyright years.
28207 * gcc.c (process_command): Update copyright notice dates.
28208 * gcov-dump.c (print_version): Ditto.
28209 * gcov.c (print_version): Ditto.
28210 * gcov-tool.c (print_version): Ditto.
28211 * gengtype.c (create_file): Ditto.
28212 * doc/cpp.texi: Bump @copying's copyright year.
28213 * doc/cppinternals.texi: Ditto.
28214 * doc/gcc.texi: Ditto.
28215 * doc/gccint.texi: Ditto.
28216 * doc/gcov.texi: Ditto.
28217 * doc/install.texi: Ditto.
28218 * doc/invoke.texi: Ditto.
28220 Copyright (C) 2017 Free Software Foundation, Inc.
28222 Copying and distribution of this file, with or without modification,
28223 are permitted in any medium without royalty provided the copyright
28224 notice and this notice are preserved.