1 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
4 disturbin profile of entry block.
6 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
8 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
11 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
13 * ipa-cp.c (update_profiling_info): Handle conversion to local
15 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
17 2017-11-17 Jeff Law <law@redhat.com>
19 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
20 from evrp_dom_walker class. Various methods moved into new class.
21 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
22 (evrp_range_analyzer::enter): New method.
23 (evrp_range_analyzer::leave): New method.
24 (evrp_dom_walker): Remove delegators no longer needed by this class.
25 Replace vr_values data member with evrp_range_analyzer
27 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
28 method extracted from evrp_dom_walker::before_dom_children.
29 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
30 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
32 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
33 Add private copy constructor and move assignment operators.
34 Privatize methods and class data where trivially possible.
35 (evrp_dom_walker::cleanup): New function, extracted from
36 execute_early_vrp. Simplify access to class data.
38 * vr-values.h (get_output_for_vrp): Prototype.
39 * vr-values.c (get_output_for_vrp): New function extracted from
40 vrp_visit_assignment_or_call and extract_range_from_stmt.
41 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
43 2017-11-17 Luis Machado <luis.machado@linaro.org>
45 * config/aarch64/aarch64.c
46 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
47 (qdf24xx_tunings) <autoprefetcher_model>: Set to
48 tune_params::AUTOPREFETCHER_WEAK.
50 2017-11-17 Tamar Christina <tamar.christina@arm.com>
53 * config/arm/arm.c (arm_valid_target_attribute_rec):
54 Parse "arch=" and "+<ext>".
55 (arm_valid_target_attribute_tree): Re-init global options.
56 (arm_option_override): Make non-static.
57 (arm_options_perform_arch_sanity_checks): Make errors fatal.
58 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
59 (__ARM_FEATURE_CRC32): Support undef.
60 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
61 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
63 2017-11-17 David Malcolm <dmalcolm@redhat.com>
65 * gdbinit.in (break-on-diagnostic): New command.
67 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
69 * config/i386/i386.c (ix86_expand_epilogue): Change simple
70 return to indirect jump for EH return if control-flow
71 protection is enabled. Change explicit 'false' argument in
72 pro_epilogue_adjust_stack with a value of
74 * config/i386/i386.md (simple_return_indirect_internal):
75 Remove SImode restriction to support 64-bit.
77 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
79 * combine.c (added_notes_insn): New.
80 (try_combine): Handle added_notes_insn like added_links_insn.
81 Rewrite return value code.
82 (distribute_notes): Set added_notes_insn to the earliest insn we added
85 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
87 PR rtl-optimization/82621
88 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
89 dest of one of those SETs is unused.
91 2017-11-17 Richard Biener <rguenther@suse.de>
94 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
95 * tree-pretty-print.c (dump_generic_node): Handle
96 annot_expr_parallel_kind.
97 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
98 * gimplify.c (gimple_boolify): Likewise.
100 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
102 * config.gcc (extra_headers): Add cet.h for x86 targets.
103 * config/i386/cet.h: New file.
104 * doc/install.texi: Add --enable-cet/--disable-cet.
106 2017-11-17 Richard Biener <rguenther@suse.de>
108 PR tree-optimization/83017
109 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
110 (gen_parallel_loop): Properly count iterations.
111 (parallelize_loops): Handle loop->can_be_parallel independent
112 of flag_loop_parallelize_all. Make static profitability test match
114 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
115 * invoke.texi (parloops-min-per-thread): Document.
117 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
119 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
120 upstreaming review comments.
122 2017-11-17 Tamar Christina <tamar.christina@arm.com>
124 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
125 with fast unaligned access.
126 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
128 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
130 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
131 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
133 2017-11-17 Richard Biener <rguenther@suse.de>
135 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
136 folding of references.
138 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
141 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
142 of replacing call to strncmp with corresponding call to strcmp when
145 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
147 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
148 option prefer-avx256 for skylake-avx512 configuration.
149 * config/i386/i386.c (ix86_option_override_internal): Ditto.
150 (get_builtin_code_for_version): Ditto.
152 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
153 Monk Chiang <sh.chiang04@gmail.com>
155 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
156 (FIXED_REGISTERS): Reserve more register numbers.
157 (CALL_USED_REGISTERS): Likewise.
158 (REG_ALLOC_ORDER): Likewise.
159 (REG_CLASS_CONTENTS): Likewise.
160 (REGISTER_NAMES): Likewise.
162 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
163 Kito Cheng <kito.cheng@gmail.com>
165 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
167 * config/nds32/iterators.md: Add vector mode iterators and attributes.
169 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
171 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
172 parameter list for vec_splats.
174 2017-11-16 Joseph Myers <joseph@codesourcery.com>
176 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
178 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option
181 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
184 * doc/install.texi: Document incompatibility of Perl >=5.6.26
185 with the required version of automake 1.11.6.
187 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
189 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
191 (DU_C2_3_power9): Correct reservation combinations.
192 (FP_DIV_power9, VEC_DIV_power9): New.
193 (power9-alu): Split out rotate/shift...
194 (power9-rot): ...to here, correct dispatch resource.
195 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
197 (power9-fp): Correct latency.
198 (power9-sdiv): Add div/sqrt resource.
199 (power9-ddiv): Correct latency, add div/sqrt resource.
200 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
201 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
203 (power9-qpdiv, power9-qpmul): Adjust resource usage.
205 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
207 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
208 switch statement mapping KF built-ins to TF built-ins if we don't
209 have the proper ISA 3.0 assembler support.
211 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
213 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
214 (lower_emutls_function_body): Do not compute it.
216 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
218 * ipa-split.c (split_bb_info): Turn time to sreal.
219 (split_point): Likewise.
220 (dump_split_point): Likewise.
221 (fine_split_points): Likewise.
222 (execute_split_functions): Only zero split_bbs; turn time to sreals.
224 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
226 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
229 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
231 * predict.c (combine_predictions_for_bb): Preserve zero predicted
233 (expensive_function_p): Remove useless assert.
234 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
236 2017-11-16 Martin Sebor <msebor@redhat.com>
238 PR tree-optimization/82588
239 PR tree-optimization/82583
240 * tree-vrp.c (check_array_ref): Handle flexible array members,
241 string literals, and inner indices.
242 (search_for_addr_array): Add detail to diagnostics.
244 2017-11-16 Nathan Sidwell <nathan@acm.org>
248 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
249 (SET_DECL_ASSEMBLER_NAME): Forward to
250 overwrite_decl_assembler_name.
251 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
252 (overwrite_decl_assembler_name): Declare.
253 * tree.c (overwrite_decl_assembler_name): New.
254 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
255 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
256 (LANG_HOOKS_INITIALIZER): Add it.
257 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
258 * langhooks.c (lhd_set_decl_assembler_name): Use
259 SET_DECL_ASSEMBLER_NAME.
260 (lhd_overwrite_decl_assembler_name): Default implementation.
262 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
263 Jackson Woodruff <jackson.woodruff@arm.com>
265 PR tree-optimization/71026
266 * match.pd: Canonicalize constant multiplies in division.
268 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
270 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
271 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
272 Remove OPT_fomit_frame_pointer entry.
273 * common/config/arc/arc-common.c: Likewise.
274 * common/config/arm/arm-common.c: Likewise.
275 * common/config/avr/avr-common.c: Likewise.
276 * common/config/c6x/c6x-common.c: Likewise.
277 * common/config/cr16/cr16-common.c: Likewise.
278 * common/config/cris/cris-common.c: Likewise.
279 * common/config/epiphany/epiphany-common.c: Likewise.
280 * common/config/fr30/fr30-common.c: Likewise.
281 * common/config/frv/frv-common.c: Likewise.
282 * common/config/ia64/ia64-common.c: Likewise.
283 * common/config/iq2000/iq2000-common.c: Likewise.
284 * common/config/lm32/lm32-common.c: Likewise.
285 * common/config/m32r/m32r-common.c: Likewise.
286 * common/config/mcore/mcore-common.c: Likewise.
287 * common/config/microblaze/microblaze-common.c: Likewise.
288 * common/config/mips/mips-common.c: Likewise.
289 * common/config/mmix/mmix-common.c: Likewise.
290 * common/config/mn10300/mn10300-common.c: Likewise.
291 * common/config/nios2/nios2-common.c: Likewise.
292 * common/config/pa/pa-common.c: Likewise.
293 * common/config/pdp11/pdp11-common.c: Likewise.
294 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
295 * common/config/riscv/riscv-common.c: Likewise.
296 * common/config/rs6000/rs6000-common.c: Likewise.
297 * common/config/rx/rx-common.c: Likewise.
298 * common/config/s390/s390-common.c: Likewise.
299 * common/config/sh/sh-common.c: Likewise.
300 * common/config/sparc/sparc-common.c: Likewise.
301 * common/config/tilegx/tilegx-common.c: Likewise.
302 * common/config/tilepro/tilepro-common.c: Likewise.
303 * common/config/v850/v850-common.c: Likewise.
304 * common/config/visium/visium-common.c: Likewise.
305 * common/config/xstormy16/xstormy16-common.c: Likewise.
306 * common/config/xtensa/xtensa-common.c: Likewise.
307 * invoke.texi (-fomit-frame-pointer): Update documentation.
309 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
311 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
313 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
315 * cfg.c (scale_bbs_frequencies_int,
316 cale_bbs_frequencies_gcov_type): Remove.
317 * cfg.h (scale_bbs_frequencies_int,
318 cale_bbs_frequencies_gcov_type): Remove.
320 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
322 * tree-ssa-loop-manip.c
323 (scale_dominated_blocks_in_loop): Update to profile counts.
324 (tree_transform_and_unroll_loop): Likewise.
326 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
328 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
329 scale_bbs_frequencies_int.
331 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
333 * final.c (compute_alignments): Use counts rather than frequencies.
335 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
337 * cfgloopanal.c: Include sreal.h
338 (average_num_loop_insns): Use counts and sreal for accounting.
340 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
342 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
345 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
346 Kito Cheng <kito.cheng@gmail.com>
348 * config/nds32/constraints.md: Provide more constraints.
349 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
350 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
351 support constraints usage.
353 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
355 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
357 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
358 Kito Cheng <kito.cheng@gmail.com>
360 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
361 * config/nds32/nds32.opt: Refine the layout.
362 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
363 TARGET_EXT_STRING): Support new options.
364 * config/nds32/nds32.h: Likewise.
365 * config/nds32/nds32.md: Likewise.
366 * config/nds32/nds32-predicates.c: Likewise.
367 * config/nds32/constraints.md: Likewise.
368 * common/config/nds32/nds32-common.c: Likewise.
370 2017-11-16 Julia Koval <julia.koval@intel.com>
373 * config/i386/gfniintrin.h: Add sse check.
374 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
376 2017-11-16 Julia Koval <julia.koval@intel.com>
378 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
379 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
380 (ix86_handle_option): Handle -mavx512vbmi2.
381 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
382 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
383 * config/i386/i386-c.c (__AVX512VBMI2__): New.
384 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
385 (ix86_valid_target_attribute_inner_p): Ditto.
386 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
387 * config/i386/i386.opt (mavx512vbmi2): New option.
388 * doc/invoke.texi: Add new option.
390 [2017-11-16 Julia Koval <julia.koval@intel.com>
392 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
393 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
394 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
395 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
396 _mm512_gf2p8mul_epi8): New intrinsics.
397 * config/i386/i386-builtin-types.def
398 (V64QI_FTYPE_V64QI_V64QI): New type.
399 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
400 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
401 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
402 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
403 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
404 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
406 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
408 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
409 explicitly as a stream of bytes.
411 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
413 * config/rs6000/altivec.h (vec_xst_be): New #define.
414 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
415 and externalize from *altivec_vperm_<mode>_internal.
416 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
418 (XL_BE_V8HI): Likewise.
419 (XL_BE_V4SI): Likewise.
420 (XL_BE_V4SI): Likewise.
421 (XL_BE_V2DI): Likewise.
422 (XL_BE_V4SF): Likewise.
423 (XL_BE_V2DF): Likewise.
424 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
425 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
426 all array entries with these keys: VSX_BUILTIN_VEC_XL,
427 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
428 VSX_BUILTIN_VEC_XST_BE.
429 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
430 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
432 (altivec_init_builtins): Replace conditional calls to def_builtin
433 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
434 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
435 with unconditional calls. Remove calls to def_builtin for
436 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
437 __builtin_vec_xst_be.
438 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
439 to define_expand, and add alternate RTL generation for P8.
440 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
442 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
443 add alternate RTL generation for P8.
444 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
445 vsx_ld_elemrev_v16qi.
446 (vsx_st_elemrev_v8hi): Convert define_insn
447 to define_expand, and add alternate RTL generation for P8.
448 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
450 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
451 add alternate RTL generation for P8.
452 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
453 vsx_st_elemrev_v16qi.
455 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
458 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
459 TARGET_AVX512ER check.
460 (ix86_option_override_internal): Set MASK_VZEROUPPER if
461 neither -mzeroupper nor -mno-zeroupper is used and
462 TARGET_EMIT_VZEROUPPER is set.
463 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
464 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
466 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
468 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
469 folding of vector compares.
470 (fold_build_vec_cmp): New helper function.
471 (fold_compare_helper): New helper function.
472 (builtin_function_type): Add compare builtins to the list of functions
473 having unsigned arguments. Cosmetic updates to comment indentation.
474 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
475 the not+eq combination.
477 2017-11-15 Bin Cheng <bin.cheng@arm.com>
479 PR tree-optimization/82726
480 PR tree-optimization/70754
481 * tree-predcom.c (order_drefs_by_pos): New function.
482 (combine_chains): Move code setting has_max_use_after to...
483 (try_combine_chains): ...here. New parameter. Sort combined chains
484 according to position information.
485 (tree_predictive_commoning_loop): Update call to above function.
486 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
488 2017-11-15 Bin Cheng <bin.cheng@arm.com>
490 PR tree-optimization/82726
492 2017-01-23 Bin Cheng <bin.cheng@arm.com>
494 PR tree-optimization/70754
495 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
496 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
497 combined stmt before it if not NULL.
498 (combine_chains): Process refs reversely and compute dominance point
502 2017-02-23 Bin Cheng <bin.cheng@arm.com>
504 PR tree-optimization/79663
505 * tree-predcom.c (combine_chains): Process refs in reverse order
506 only for ZERO length chains, and add explaining comment.
508 2017-11-15 Tamar Christina <tamar.christina@arm.com>
510 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
511 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
512 * doc/invoke.texi (ARM Options): Add armv8.3-a.
514 2017-11-15 Tamar Christina <tamar.christina@arm.com>
516 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
518 2017-11-15 Martin Liska <mliska@suse.cz>
520 * tree-cfg.c (pass_warn_function_return::execute):
521 Compare warn_return_type for greater than zero.
523 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
527 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
528 to return true on Xeon and not on Xeon Phi.
529 (ix86_check_avx256_register): Changed to ...
530 (ix86_check_avx_upper_register): ... this. Add extra check for
531 VALID_AVX512F_REG_OR_XI_MODE.
532 (ix86_avx_u128_mode_needed): Changed
533 ix86_check_avx256_register to ix86_check_avx_upper_register.
534 (ix86_check_avx256_stores): Changed to ...
535 (ix86_check_avx_upper_stores): ... this. Changed
536 ix86_check_avx256_register to ix86_check_avx_upper_register.
537 (ix86_avx_u128_mode_after): Changed
538 avx_reg256_found to avx_upper_reg_found. Changed
539 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
540 (ix86_avx_u128_mode_entry): Changed
541 ix86_check_avx256_register to ix86_check_avx_upper_register.
542 (ix86_avx_u128_mode_exit): Ditto.
543 * config/i386/i386.h: (host_detect_local_cpu): New define.
545 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
547 * config/arm/xgene1.md (xgene1): Split into automatons
548 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
549 (xgene1_f_load): Adjust reservations and/or types.
550 (xgene1_f_store): Likewise.
551 (xgene1_load_pair): Likewise.
552 (xgene1_store_pair): Likewise.
553 (xgene1_fp_load1): Likewise.
554 (xgene1_load1): Likewise.
555 (xgene1_store1): Likewise.
556 (xgene1_move): Likewise.
557 (xgene1_alu): Likewise.
558 (xgene1_simd): Likewise.
559 (xgene1_bfm): Likewise.
560 (xgene1_neon_load1): Likewise.
561 (xgene1_neon_store1): Likewise.
562 (xgene1_neon_logic): Likewise.
563 (xgene1_neon_st1): Likewise.
564 (xgene1_neon_ld1r): Likewise.
565 (xgene1_alu_cond): Added.
566 (xgene1_shift_reg): Likwise.
567 (xgene1_bfx): Likewise.
568 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
570 2017-11-15 Jakub Jelinek <jakub@redhat.com>
573 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
575 (can_widen_mult_without_libcall): New function.
576 (expand_mul_overflow): If only checking unsigned mul overflow,
577 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
578 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
579 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
580 (expand_DIVMOD): Formatting fix.
581 * expmed.h (expand_mult): Add NO_LIBCALL argument.
582 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
583 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
585 PR tree-optimization/82977
586 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
587 constructed temporary to strlen_to_stridx.put.
589 2017-11-15 Martin Liska <mliska@suse.cz>
591 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
592 * configure: Regenerate.
594 2017-11-15 Martin Liska <mliska@suse.cz>
597 * config/sh/sh-mem.cc: Use proper probability for
600 2017-11-14 Jeff Law <law@redhat.com>
602 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
603 the red zone for stack_clash_protection_final_dynamic_probe targets
604 when the total dynamic stack size is zero bytes.
606 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
607 blocks is post order.
609 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
611 * dumpfile.h (TDF_COMPARE_DEBUG): New.
612 * final.c (rest_of_clean_state): Set it for the
613 -fcompare-debug dump.
614 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
615 class when TDF_COMPARE_DEBUG is set.
617 * dwarf2out.c (gen_producer_string): Discard
620 2017-11-15 Joseph Myers <joseph@codesourcery.com>
623 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
624 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
625 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
626 (__TGMATH_REAL_2_3): Remove macros.
627 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
628 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
630 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
632 (remquo): Define using __TGMATH_REAL_3.
634 2017-11-14 Jeff Law <law@redhat.com>
636 * vr-values.c: New file with contents extracted from tree-vrp.c.
637 * Makefile.in (OBJS): Add vr-values.o
638 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
639 (set_value_range, set_and_canonicalize_value_range): Likewise.
640 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
641 (value_range_constant_singleton, symbolic_range_p): Likewise.
642 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
643 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
644 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
645 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
646 (find_case_label_range, find_case_label_index): Likewise.
647 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
648 (range_int_cst_singleton_p, value_inside_range): Likewise.
649 (get_single_symbol): Likewise.
650 (switch_update): Move structure definition here.
651 (to_remove_edges, to_update_switch_stmts): Provide externs.
652 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
653 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
654 (vrp_val_is_min, set_value_range): Likewise.
655 (set_and_canonicalize_value_range, copy_value_range): Likewise.
656 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
657 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
658 (range_is_nonnull, range_int_cst_p): Likewwise.
659 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
660 (get_single_symbol, operand_less_p): Likewise
661 (compare_values_warnv, compare_values): Likewise.
662 (value_inside_range, value_range_constant_singleton): Likewise.
663 (zero_nonzero_bitgs_from_vr): Likewise.
664 (extract_range_from_binary_expr_1): Likewise.
665 (overflow_comparison_p): Likewise.
666 (to_remove_edges, to_update_switch_stmts): Likewise.
667 (find_case_label-index, find_case_label_range): Likewise.
668 (switch_update, set_value_range_to_nonnegative): Remove.
669 (set_value_range_to_truthvalue): Likewise.
670 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
671 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
672 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
673 (find_case_label_ranges, test_for_singularity): Likewise.
674 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
675 (x_vr_values): Move to its remaining use site.
677 2017-11-10 Jeff Law <law@redhat.com>
679 * vr-values.h (VR_INITIALIZER): Move #define here.
680 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
681 * Makefile.in (OBJS): Add tree-evrp.o
682 * tree-vrp.h (assert_info): Move structure definition here.
683 (set_value_range_to_varying): Prototype.
684 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
685 (infer_value_range, register_edge_assert_for): Likewise.
686 (stmt_interesting_for_vrp): Likewise.
687 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
688 (set_value_range_to_varying): No longer static.
689 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
690 (infer_value_range, register_edge_assert_for): Likewise.
692 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
694 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
695 generate the XXBRD instruction.
697 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
698 (rs6000_builtin_type_compatible): Treat _Float128 and long double
699 as being compatible if -mabi=ieeelongdouble.
700 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
701 to setup float128 built-ins with hardware support.
702 (BU_FLOAT128_HW_2): Likewise.
703 (BU_FLOAT128_HW_3): Likewise.
704 (BU_FLOAT128_HW_VSX_1): Likewise.
705 (BU_FLOAT128_HW_VSX_2): Likewise.
706 (scalar_extract_expq): Change float128 built-in functions to
707 accommodate having both KFmode and TFmode functions. Use the
708 KFmode variant as the default.
709 (scalar_extract_sigq): Likewise.
710 (scalar_test_neg_qp): Likewise.
711 (scalar_insert_exp_q): Likewise.
712 (scalar_insert_exp_qp): Likewise.
713 (scalar_test_data_class_qp): Likewise.
714 (sqrtf128_round_to_odd): Delete processing the round to odd
715 built-in functions as special built-in functions, and define them
716 as float128 built-ins. Use the KFmode variant as the default.
717 (truncf128_round_to_odd): Likewise.
718 (addf128_round_to_odd): Likewise.
719 (subf128_round_to_odd): Likewise.
720 (mulf128_round_to_odd): Likewise.
721 (divf128_round_to_odd): Likewise.
722 (fmaf128_round_to_odd): Likewise.
723 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
724 support for KFmode and TFmode xststdcqp calls.
725 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
726 point, switch the built-in handlers for the get/set float128
727 exponent, get float128 mantissa, float128 test built-ins, and the
728 float128 round to odd built-in functions. Eliminate creating the
729 float128 round to odd built-in functions as special built-ins.
730 (rs6000_init_builtins): Eliminate special creation of the float128
731 round to odd built-in functions.
732 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
733 function insns to support both TFmode and KFmode variants.
734 (xsxsigqp_<mode>): Likewise.
735 (xsiexpqpf_<mode>): Likewise.
736 (xsiexpqp_<mode>): Likewise.
737 (xststdcqp_<mode>): Likewise.
738 (xststdcnegqp_<mode>): Likewise.
739 (xststdcqp_<mode>): Likewise.
741 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
743 * tree-ssa-threadupdate.c (compute_path_counts): Remove
744 unused path_in_freq_ptr parameter.
745 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
747 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
749 * ipa-inline.c (edge_badness): Dump sreal frequency.
750 (compute_inlined_call_time): Match natural implementaiton ...
751 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
752 forgotten division by CGRAPH_FREQ_BASE.
754 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
756 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
757 Solaris 11. Update comment.
758 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
760 * config/sol2.h (STARTFILE_SPEC): Likewise.
761 * configure: Regenerate.
763 2017-11-14 Carl Love <cel@us.ibm.com>
765 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
766 le_ and be_ prefixes to swap* variables. Remove
767 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
770 2017-11-14 Jason Merrill <jason@redhat.com>
772 Support GTY((cache)) on hash_map.
773 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
774 (ggc_cache_remove): Override it instead of ggc_mx.
775 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
776 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
777 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
778 (simple_cache_map_traits): Override maybe_mx.
779 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
780 (hash_map): Friend gt_cleare_cache.
781 (gt_cleare_cache): New.
782 * tree.h (tree_cache_traits): New hash_map traits class.
783 (tree_cache_map): New typedef.
785 2017-11-14 Richard Biener <rguenther@suse.de>
787 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
788 paramter and handling.
789 (cleanup_control_flow_bb): Likewise.
790 (cleanup_control_flow_pre): New helper performing a DFS walk
791 to call cleanup_control_flow_bb in PRE order.
792 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
793 via cleanup_control_flow_pre.
795 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
797 * config/aarch64/aarch64-simd.md
798 (aarch64_simd_bsl<mode>_internal): Remove DImode.
799 (*aarch64_simd_bsl<mode>_alt): Likewise.
800 (aarch64_simd_bsldi_internal): New.
801 (aarch64_simd_bsldi_alt): Likewise.
803 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
805 * tracer.c (better_p): Do not compare frequencies.
806 * reg-stack.c (better_edge): Likewise.
807 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
810 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
812 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
813 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
814 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
815 * lto-streamer-in.c (input_function): Use update_max_bb_count.
816 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
817 * predict.c (maybe_hot_frequency_p): Inline to ...
818 (maybe_hot_count_p): ... here; rewrite to counts.
819 (counts_to_freqs): Rename to ...
820 (update_max_bb_count): ... this one.
821 (expensive_function_p): Use counts.
822 (estimate_bb_frequencies): Update.
823 (rebuild_frequencies): Update.
824 * predict.h (counts_to_freqs): Rename to ...
825 (update_max_bb_count): ... this one.
826 * profile.c (compute_branch_probabilities): Add debug info
827 * tree-inline.c (expand_call_inline): Update debug info.
828 (optimize_inline_calls): Use update_max_bb_count..
829 (tree_function_versioning): Use update_max_bb_count..
830 * value-prof.c (gimple_value_profile_transformations):
831 Do not use update_max_bb_count.
833 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
835 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
836 always use frequencies.
838 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
840 * bb-reorder.c: Remove frequencies from comments.
841 (better_edge_p): Use profile counts.
842 (find_traces): Dump profile counts.
843 (rotate_loop): Use profile counts.
844 (find_traces_1_round): Likewise.
845 (connect_better_edge_p): Use counts instead of probabilities for
847 (copy_bb_p): Drop early check for non-0 frequency.
848 (sanitize_hot_paths): Update comments.
850 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
852 * ipa-split.c (struct split_point): Add count.
853 (consider_split): Do not compute incoming frequency; compute incoming
854 count and store it to split_point.
855 (split_function): Set count of the call to split part correctly.
857 2017-11-13 Carl Love <cel@us.ibm.com>
859 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
861 2017-11-13 Tom Tromey <tom@tromey.com>
863 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
865 2017-11-13 Carl Love <cel@us.ibm.com>
867 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
868 Add support for builtins:
869 unsigned int vec_first_{,miss}_match_{,or_eos}index,
870 vector {un,}signed {char,int,short},
871 vector {un,}signed {char,int,short}) arguments.
872 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
873 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
874 Add BU_P9V_AV_2 expansions for the builtins.
875 * config/rs6000/altivec.h (vec_first_match_index,
876 vec_first_mismatch_index, vec_first_match_or_eos_index,
877 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
878 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
879 new extern declaration.
880 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
881 * config/rs6000/vsx.md (first_match_index_<mode>,
882 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
883 first_mismatch_or_eos_index_<mode>): Add define expand.
884 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
885 * doc/extend.texi: Update the built-in documenation file for the new
888 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
890 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
891 into the min/max operations for _Float<N> and _Float<N>X types.
893 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
896 * dwarf2out.c (do_eh_frame): New static variable.
897 (dwarf2out_begin_prologue): Set it.
898 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
900 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
902 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
904 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
906 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
909 * debug.h (dwarf2out_do_eh_frame): Declare.
910 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
911 (dwarf2out_do_frame): Use it.
912 (dwarf2out_do_cfi_asm): Likewise.
913 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
914 (dwarf2out_assembly_start): Likewise.
915 (dwarf2out_begin_prologue): Fix comment.
916 * toplev.c (compile_file): Always call dwarf2out_frame_finish
917 if the target needs either debug or unwind DWARF2 info.
918 * lto-opts.c (lto_write_options): Do not save -fexceptions,
919 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
920 -ftrapping-math, -ftrapv and -fwrapv.
922 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
924 * cgraph.c (cgraph_edge::sreal_frequency): New function.
925 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
926 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
927 (estimate_edge_size_and_time): Likewise.
928 (ipa_merge_fn_summary_after_inlining): Likewise.
929 * ipa-inline.c (cgraph_freq_base_rec): Remove.
930 (compute_uninlined_call_time): Use sreal_frequency.
931 (compute_inlined_call_time): Likewise.
932 (ipa_inline): Do not initialize cgraph_freq_base_rec.
933 * profile-count.c: Include sreal.h.
934 (profile_count::to_sreal_scale): New.
935 * profile-count.h: Forward declare sreal.
936 (profile_count::to_sreal_scale): Declare.
938 2017-11-13 Nathan Sidwell <nathan@acm.org>
940 * diagnostic.c (maybe_line_and_column): New.
941 (diagnostic_get_location_text): Use it.
942 (diagnostic_report_current_module): Likewise.
943 (test_diagnostic_get_location_text): Add tests.
945 2017-11-13 Luis Machado <luis.machado@linaro.org>
947 * doc/md.texi (Specifying processor pipeline description): Fix
948 incorrect latency for the div instruction example.
950 2017-11-13 Jakub Jelinek <jakub@redhat.com>
952 PR tree-optimization/78821
953 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
954 that bit_not_p is the same.
955 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
956 (split_group): Count precisely bit_not_p bits in each statement.
957 (invert_op): New function.
958 (imm_store_chain_info::output_merged_store): Use invert_op to
959 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
960 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
962 2017-11-13 Martin Liska <mliska@suse.cz>
964 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
965 (struct source_info): Likewise.
966 (add_branch_counts): Likewise.
967 (add_line_counts): Likewise.
968 (function_summary): Likewise.
969 (output_intermediate_line): Likewise.
970 (generate_results): Likewise.
972 2017-11-13 Martin Liska <mliska@suse.cz>
974 * gcov.c (struct block_info): Remove typedef for block_t.
975 (struct line_info): Likewise.
976 (line_info::has_block): Likewise.
977 (EXIT_BLOCK): Likewise.
980 (get_cycles_count): Likewise.
981 (process_file): Likewise.
982 (read_graph_file): Likewise.
983 (solve_flow_graph): Likewise.
984 (find_exception_blocks): Likewise.
985 (add_line_counts): Likewise.
986 (accumulate_line_info): Likewise.
987 (output_line_details): Likewise.
989 2017-11-13 Martin Liska <mliska@suse.cz>
991 * gcov.c (struct arc_info): Remove typedef for arc_t.
992 (struct line_info): Likewise.
993 (add_branch_counts): Likewise.
994 (output_branch_count): Likewise.
995 (function_info::~function_info): Likewise.
997 (output_intermediate_line): Likewise.
998 (read_graph_file): Likewise.
999 (solve_flow_graph): Likewise.
1000 (find_exception_blocks): Likewise.
1001 (add_line_counts): Likewise.
1002 (accumulate_line_info): Likewise.
1003 (output_line_details): Likewise.
1004 (output_function_details): Likewise.
1006 2017-11-13 Martin Liska <mliska@suse.cz>
1008 * gcov.c (struct function_info): Remove typedef for function_t.
1009 (struct source_info): Likewise.
1010 (source_info::get_functions_at_location): Likewise.
1011 (solve_flow_graph): Likewise.
1012 (find_exception_blocks): Likewise.
1013 (add_line_counts): Likewise.
1014 (output_intermediate_file): Likewise.
1015 (process_file): Likewise.
1016 (generate_results): Likewise.
1017 (release_structures): Likewise.
1018 (read_graph_file): Likewise.
1019 (read_count_file): Likewise.
1020 (accumulate_line_counts): Likewise.
1021 (output_lines): Likewise.
1023 2017-11-13 Martin Liska <mliska@suse.cz>
1025 * gcov.c (function_info::function_info): Remove num_counts
1026 and add vector<gcov_type>.
1027 (function_info::~function_info): Use the vector.
1028 (process_file): Likewise.
1029 (read_graph_file): Likewise.
1030 (read_count_file): Likewise.
1031 (solve_flow_graph): Likewise.
1033 2017-11-13 Martin Liska <mliska@suse.cz>
1035 * gcov.c (function_info::is_artificial): New function.
1036 (process_file): Erase all artificial early.
1037 (generate_results): Skip as all artificial are already
1040 2017-11-13 Martin Liska <mliska@suse.cz>
1042 * gcov.c (read_graph_file): Store to global vector of functions.
1043 (read_count_file): Iterate the vector.
1044 (process_file): Likewise.
1045 (generate_results): Likewise.
1046 (release_structures): Likewise.
1048 2017-11-13 Jakub Jelinek <jakub@redhat.com>
1050 PR tree-optimization/82954
1051 * gimple-ssa-store-merging.c
1052 (imm_store_chain_info::coalesce_immediate_stores): If
1053 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
1055 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
1057 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
1058 Upddate call to ENDIAN_LANE_N.
1059 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
1060 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
1061 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
1062 and use aarch64_endian_lane_rtx.
1063 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
1065 2017-11-12 Tom de Vries <tom@codesourcery.com>
1067 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
1069 2017-11-12 Tom de Vries <tom@codesourcery.com>
1071 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
1073 2017-11-12 Tom de Vries <tom@codesourcery.com>
1075 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
1077 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
1078 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
1079 * defaults.h (ASM_OUTPUT_LABELREF): Same.
1081 2017-11-11 Martin Sebor <msebor@redhat.com>
1084 * doc/extend.texi (attribute nonstring): Remove spurious argument.
1087 * prefic.c (translate_name): Replace strncpy with memcpy to
1088 avoid -Wstringop-truncation.
1090 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1092 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
1094 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1096 * predict.c (maybe_hot_frequency_p): Do not use cfun.
1098 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1100 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
1103 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
1105 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
1106 enable generating XXBRH if the value is in a vector register.
1107 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
1108 value is in a vector register.
1109 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
1110 register to register bswap64's instead of doing the GPR sequence
1111 used on previous machines.
1112 (bswapdi2_xxbrd): New insn.
1113 (bswapdi2_reg): Disallow on ISA 3.0.
1114 (register to register bswap64 splitter): Do not split the insn on
1115 ISA 3.0 systems that use XXBRD.
1117 2017-11-10 Martin Sebor <msebor@redhat.com>
1120 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
1122 (find_subframework_file): Same.
1124 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1126 * auto-profile.c (afdo_indirect_call): Drop frequency.
1127 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
1128 (cgraph_node::create_edge): Drop frequency argument.
1129 (cgraph_node::create_indirect_edge): Drop frequency argument.
1130 (cgraph_edge::make_speculative): Drop frequency arguments.
1131 (cgraph_edge::resolve_speculation): Do not update frequencies
1132 (cgraph_edge::dump_edge_flags): Do not dump frequency.
1133 (cgraph_node::dump): Check consistency in IPA mode.
1134 (cgraph_edge::maybe_hot_p): Use IPA counter.
1135 (cgraph_edge::verify_count_and_frequency): Rename to ...
1136 (cgraph_edge::verify_count): ... this one; drop frequency checking.
1137 (cgraph_node::verify_node): Update.
1138 * cgraph.h (struct cgraph_edge): Drop frequency.
1139 (cgraph_edge::frequency): New function.
1140 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
1142 (cgraph_edge::rebuild_edges): Likewise.
1143 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
1144 (duplicate_thunk_for_node): Do not pass frequency.
1145 (cgraph_node::create_clone): Scale only counts.
1146 (cgraph_node::create_virtual_clone): Do not pass frequency.
1147 (cgraph_node::create_edge_including_clones): Do not pass frequency.
1148 (cgraph_node::create_version_clone): Do not pass frequency.
1149 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
1150 (cgraph_node::expand_thunk): Do not pass frequency.
1151 (cgraph_node::create_wrapper): Do not pass frequency.
1152 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
1154 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
1155 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
1156 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
1157 (gather_caller_stats): Use frequency function.
1158 (ipcp_cloning_candidate_p): Use frequency function.
1159 (ipcp_propagate_stage): Use frequency function.
1160 (get_info_about_necessary_edges): Use frequency function.
1161 (update_profiling_info): Update only IPA profile.
1162 (update_specialized_profile): Use frequency functoin.
1163 (perhaps_add_new_callers): Update only IPA profile.
1164 * ipa-devirt.c (ipa_devirt): Use IPA profile.
1165 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
1166 (dump_ipa_call_summary): Use frequency function.
1167 (estimate_edge_size_and_time): Use frequency function.
1168 (ipa_merge_fn_summary_after_inlining): Use frequency function.
1169 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
1170 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
1171 (update_noncloned_counts): ... ths one; scale counts only.
1172 (clone_inlined_nodes): Do not scale frequency.
1173 (inline_call): Do not pass frequency.
1174 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
1175 (compute_inlined_call_time): Use IPA profile.
1176 (want_inline_small_function_p): Use IPA profile.
1177 (want_inline_self_recursive_call_p): Use IPA profile.
1178 (edge_badness): Use IPA profile.
1179 (lookup_recursive_calls): Use IPA profile.
1180 (recursive_inlining): Do not pass frequency.
1181 (resolve_noninline_speculation): Do not update frequency.
1182 (inline_small_functions): Collect max of IPA profile.
1183 (dump_overall_stats): Dump IPA porfile.
1184 (dump_inline_stats): Dump IPA porfile.
1185 (ipa_inline): Collect IPA stats.
1186 * ipa-inline.h (clone_inlined_nodes): Update prototype.
1187 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
1188 (ipa_propagate_frequency): Use frequency function.
1189 (ipa_profile): Cleanup.
1190 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
1191 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
1192 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
1193 (input_node): Do not stream frequency.
1194 (input_edge): Do not stream frequency.
1195 (merge_profile_summaries): Scale only IPA profiles.
1196 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
1197 * predict.c (drop_profile): Do not recompute frequency.
1198 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
1199 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
1200 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
1201 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
1202 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
1203 * tree-inline.c (copy_bb): Do not scale frequency.
1204 (expand_call_inline): Do not scale frequency.
1205 (tree_function_versioning): Do not scale frequency.
1206 * ubsan.c (ubsan_create_edge): Do not pass frequency.
1208 2017-11-10 Julia Koval <julia.koval@intel.com>
1210 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
1211 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
1212 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
1213 (_mm256_maskz_gf2p8affine_epi64_epi8)
1214 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
1215 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
1216 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
1217 (__builtin_ia32_vgf2p8affineqb_v32qi)
1218 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
1219 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
1221 2017-11-10 Tamar Christina <tamar.christina@arm.com>
1225 (arm_option_override): Refactor.
1226 (arm_option_reconfigure_globals): New.
1227 (arm_options_perform_arch_sanity_checks): New.
1228 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
1230 (arm_options_perform_arch_sanity_checks): Likewise
1232 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
1234 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
1235 (power9-qpmul): New.
1236 * rs6000/rs6000.md ("type" attr): Add qmul.
1237 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
1238 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
1239 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
1241 2017-11-10 Martin Sebor <msebor@redhat.com>
1244 * builtins.c (compute_objsize): Handle arrays that
1245 compute_builtin_object_size likes to fail for. Make extern.
1246 * builtins.h (compute_objsize): Declare.
1247 (check_strncpy_sizes): New function.
1248 (expand_builtin_strncpy): Call check_strncpy_sizes.
1249 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
1250 -Wstringop-truncation.
1251 (gimple_fold_builtin_strncat): Same.
1252 * gimple.c (gimple_build_call_from_tree): Set call location.
1253 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
1254 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
1255 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
1256 (handle_builtin_strlen): Use strlen_to_stridx.
1257 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
1259 Use strlen_to_stridx.
1260 (pass_strlen::execute): Release strlen_to_stridx.
1261 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
1262 (-Wstringop-truncation): Document new option.
1264 2017-11-10 Martin Liska <mliska@suse.cz>
1266 PR gcov-profile/82702
1267 * gcov.c (main): Handle intermediate files in a different
1269 (get_gcov_intermediate_filename): New function.
1270 (output_gcov_file): Remove support of intermediate files.
1271 (generate_results): Allocate intermediate file.
1272 (release_structures): Clean-up properly fn_end.
1273 (output_intermediate_file): Start iterating with line 1.
1275 2017-11-10 Jakub Jelinek <jakub@redhat.com>
1277 PR tree-optimization/82929
1278 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
1279 ops_swapped_p non-static data member.
1280 (store_immediate_info::store_immediate_info): Clear it.
1281 (imm_store_chain_info::coalesce_immediate_stores): If swapping
1282 ops set ops_swapped_p.
1283 (count_multiple_uses): Handle ops_swapped_p.
1285 2017-11-10 Martin Liska <mliska@suse.cz>
1287 * coverage.c (coverage_init): Stream information about
1288 support of has_unexecuted_blocks.
1289 * doc/gcov.texi: Document that.
1290 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
1291 * gcov.c (read_graph_file): Likewise.
1292 (output_line_beginning): Fix a small issue with
1295 2017-11-10 Bin Cheng <bin.cheng@arm.com>
1297 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
1298 reference of trivial component.
1300 2017-11-10 Jakub Jelinek <jakub@redhat.com>
1303 * gimple-ssa-store-merging.c
1304 (pass_store_merging::terminate_all_aliasing_chains): For
1305 gimple_store_p stmts also call refs_output_dependent_p.
1307 PR rtl-optimization/82913
1308 * compare-elim.c (try_merge_compare): Punt if def_insn is not
1311 2017-11-09 Jeff Law <law@redhat.com>
1313 * vr-values.h: New file with vr_values class.
1314 * tree-vrp.c: Include vr-values.h
1315 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
1316 data objects into the vr_values class.
1317 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
1318 (get_value_range): Make it a member function within vr_values class.
1319 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
1320 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
1321 (op_with_constant_singleton_value_range): Likewise.
1322 (extract_range_for_var_from_comparison_expr): Likewise.
1323 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
1324 (extract_range_from_binary_expr): Likewise.
1325 (extract_range_from_unary_expr): Likewise.
1326 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
1327 (check_for_binary_op_overflow, extract_range_basic): Likewise.
1328 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
1329 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
1330 (compare_name_with_value, compare_names): Likewise.
1331 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
1332 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
1333 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
1334 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
1335 (extract_range_from_phi_node): Likewise.
1336 (simplify_truth_ops_using_ranges): Likewise.
1337 (simplify_div_or_mod_using_ranges): Likewise.
1338 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
1339 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
1340 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
1341 (simplify_float_conversion_using_ranges): Likewise.
1342 (simplify_internal_call_using_ranges): Likewise.
1343 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
1344 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
1345 poitner via x_vr_values for calls into gimple folder.
1346 (vrp_initialize_lattice): Make this the vr_values ctor.
1347 (vrp_free_lattice): Make this the vr_values dtor.
1348 (set_vr_value): New function.
1349 (class vrp_prop): Add vr_values data member. Add various member
1350 functions as well as member functions that delegate to vr_values.
1351 (check_array_ref): Make a member function within vrp_prop class.
1352 (search_for_addr_array, vrp_initialize): Likewise.
1353 (vrp_finalize): Likewise. Revamp to avoid direct access to
1354 vr_value, values_propagated, etc.
1355 (check_array_bounds): Extract vrp_prop class instance pointer from
1356 walk info structure. Use it to call member functions.
1357 (check_all_array_refs): Make a member function within vrp_prop class.
1358 Smuggle class instance pointer via walk info structure.
1359 (x_vr_values): New local static.
1360 (vrp_valueize): Use x_vr_values to get class instance.
1361 (vr_valueize_1): Likewise.
1362 (class vrp_folder): Add vr_values data member. Add various member
1363 functions as well as member functions that delegate to vr_values.
1364 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
1365 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
1366 class instance from vr_values. Use it to call member functions.
1367 (vrp_dom_walker): Add vr_values data member.
1368 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
1369 instance via x_vr_values.
1370 (identify_jump_threads): Accept vr_values as argument. Store
1371 it into the walker structure.
1372 (evrp_dom_walker): Add vr_values class data member. Add various
1374 (evrp_dom_walker::try_find_new_range): Use vr_values data
1375 member to access the memory allocator.
1376 (evrp_dom_walker::before_dom_children): Store vr_values class
1377 instance into the vrp_folder class.
1378 (evrp_dom_walker::push_value_range): Rework to avoid direct
1379 access to num_vr_values and vr_value.
1380 (evrp_dom_walker::pop_value_range): Likewise.
1381 (execute_early_vrp): Remove call to vrp_initialize_lattice.
1382 Use vr_values to get to dump_all_value_ranges member function.
1383 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
1384 and simplify_cond_using_ranges_2 via vrp_prop class instance.
1385 Pass vr_values class instance down to identify_jump_threads.
1386 Remove call to vrp_free_lattice.
1387 (debug_all_value_ranges): Remove.
1389 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
1390 (vrp_folder): Likewise.
1392 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
1393 Get it from the existing bitmap instead.
1394 (vrp_intersect_ranges_1): Likewise.
1396 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1398 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
1400 (store_immediate_info::store_immediate_info): Add bitnotp argument,
1401 set bit_not_p to it.
1402 (imm_store_chain_info::coalesce_immediate_stores): Break group
1403 if bit_not_p is different.
1404 (count_multiple_uses, split_group,
1405 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
1406 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
1407 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
1408 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
1409 to store_immediate_info ctor.
1411 2017-11-09 Jim Wilson <jimw@sifive.com>
1413 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
1414 (scan_prog_file): Likewise.
1416 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
1418 * bb-reorder.c (max_entry_frequency): Remove.
1419 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
1420 connect_traces, push_to_next_round_p): Remove prototypes.
1421 (find_traces_1_round): Use counts only.
1422 (push_to_next_round_p): Likewise.
1423 (find_traces): Likewise.
1424 (rotate_loop): Likewise.
1425 (find_traces_1_round): Likewise.
1426 (connect_traces): Likewise.
1427 (edge_order): Likewise.
1429 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
1431 * config/arm/arm.c (output_return_instruction): Add comments to
1432 indicate requirement for cmse_nonsecure_entry return to account
1433 for the size of clearing instruction output here.
1434 (thumb_exit): Likewise.
1435 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
1436 return in hardfloat mode.
1438 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
1440 * config/rs6000/rs6000.c (machine_function): Add a bool,
1441 "toc_is_wrapped_separately".
1442 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
1443 if it wasn't explicitly set or unset, we are optimizing for speed, and
1444 doing separate shrink-wrapping.
1445 (rs6000_get_separate_components): Enable the TOC component if
1446 saving the TOC register in the prologue.
1447 (rs6000_components_for_bb): Handle the TOC component.
1448 (rs6000_emit_prologue_components): Store the TOC register where needed.
1449 (rs6000_set_handled_components): Mark TOC as handled, if handled.
1450 (rs6000_emit_prologue): Don't save the TOC if that is already done.
1452 2017-11-09 Martin Jambor <mjambor@suse.cz>
1454 * ipa-param-manipulation.c: New file.
1455 * ipa-param-manipulation.h: Likewise.
1456 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
1457 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
1458 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
1459 (ipa_parm_adjustment): Likewise.
1460 (ipa_parm_adjustment_vec): Likewise.
1461 (ipa_get_vector_of_formal_parms): Moved declaration to
1462 ipa-param-manipulation.h.
1463 (ipa_get_vector_of_formal_parm_types): Likewise.
1464 (ipa_modify_formal_parameters): Likewise.
1465 (ipa_modify_call_arguments): Likewise.
1466 (ipa_combine_adjustments): Likewise.
1467 (ipa_dump_param_adjustments): Likewise.
1468 (ipa_modify_expr): Likewise.
1469 (ipa_get_adjustment_candidate): Likewise.
1470 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
1471 ipa-param-manipulation.c.
1472 (ipa_get_vector_of_formal_parm_types): Likewise.
1473 (ipa_modify_formal_parameters): Likewise.
1474 (ipa_modify_call_arguments): Likewise.
1475 (ipa_modify_expr): Likewise.
1476 (get_ssa_base_param): Likewise.
1477 (ipa_get_adjustment_candidate): Likewise.
1478 (index_in_adjustments_multiple_times_p): Likewise.
1479 (ipa_combine_adjustments): Likewise.
1480 (ipa_dump_param_adjustments): Likewise.
1481 * tree-sra.c: Also include ipa-param-manipulation.h
1482 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
1485 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1486 Alan Hayward <alan.hayward@arm.com>
1487 David Sherwood <david.sherwood@arm.com>
1489 * doc/sourcebuild.texi (vect_masked_store): Document.
1491 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1492 Alan Hayward <alan.hayward@arm.com>
1493 David Sherwood <david.sherwood@arm.com>
1495 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
1497 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1498 Alan Hayward <alan.hayward@arm.com>
1499 David Sherwood <david.sherwood@arm.com>
1501 * doc/sourcebuild.texi (vect_variable_length): Document.
1503 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1504 Alan Hayward <alan.hayward@arm.com>
1505 David Sherwood <david.sherwood@arm.com>
1507 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
1509 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1510 Alan Hayward <alan.hayward@arm.com>
1511 David Sherwood <david.sherwood@arm.com>
1513 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
1515 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1516 Alan Hayward <alan.hayward@arm.com>
1517 David Sherwood <david.sherwood@arm.com>
1519 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
1520 previously undocumented selectors.
1521 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
1523 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1525 * doc/rtl.texi (const_vector): Say that elements can be
1526 const_wide_ints too.
1527 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
1528 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
1529 (gen_vec_duplicate): Use it instead of CONSTANT_P.
1530 * optabs.c (expand_vector_broadcast): Likewise.
1532 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1533 Alan Hayward <alan.hayward@arm.com>
1534 David Sherwood <david.sherwood@arm.com>
1536 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
1537 scaled index even if the unscaled address was invalid.
1538 Don't increase the complexity of using a scale in that case.
1540 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1541 Alan Hayward <alan.hayward@arm.com>
1542 David Sherwood <david.sherwood@arm.com>
1544 * doc/rtl.texi: Rewrite the subreg rules so that they partition
1545 the inner register into REGMODE_NATURAL_SIZE bytes rather than
1546 UNITS_PER_WORD bytes.
1547 * emit-rtl.c (validate_subreg): Divide subregs into blocks
1548 based on REGMODE_NATURAL_SIZE of the inner mode.
1549 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
1550 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
1551 * expmed.c (lowpart_bit_field_p): Divide the value up into
1552 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
1553 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
1554 whether something is likely to occupy more than one register.
1556 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
1559 * ipa-inline-transform.c (update_noncloned_frequencies): Use
1560 profile_count::adjust_for_ipa_scaling.
1561 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
1562 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
1564 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
1566 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1568 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
1569 (split_group): Add total_orig and total_new arguments, estimate the
1570 number of statements related to the store group without store merging
1571 and with store merging.
1572 (imm_store_chain_info::output_merged_store): Adjust split_group
1573 callers, punt if estimated number of statements with store merging
1574 is not smaller than estimated number of statements without it.
1576 (handled_load): Remove has_single_use checks.
1577 (pass_store_merging::process_store): Likewise.
1579 2017-11-09 Richard Biener <rguenther@suse.de>
1581 PR tree-optimization/82902
1582 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
1584 2017-11-09 Martin Liska <mliska@suse.cz>
1587 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
1590 2017-11-09 Martin Liska <mliska@suse.cz>
1592 PR tree-optimization/82669
1593 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
1595 2017-11-09 Martin Liska <mliska@suse.cz>
1597 PR gcov-profile/48463
1598 * coverage.c (coverage_begin_function): Output also end locus
1599 of a function and information whether the function is
1601 * gcov-dump.c (tag_function): Parse and print the information.
1602 * gcov.c (INCLUDE_MAP): Add include.
1603 (INCLUDE_SET): Likewise.
1604 (struct line_info): Move earlier in the source file because
1605 of vector<line_info> in function_info structure.
1606 (line_info::line_info): Likewise.
1607 (line_info::has_block): Likewise.
1608 (struct source_info): Add new member index.
1609 (source_info::get_functions_at_location): New function.
1610 (function_info::group_line_p): New function.
1611 (output_intermediate_line): New function.
1612 (output_intermediate_file): Use the mentioned function.
1613 (struct function_start): New.
1614 (struct function_start_pair_hash): Likewise.
1615 (process_file): Add code that identifies group functions.
1616 Assign lines either to global or function scope.
1617 (generate_results): Skip artificial functions.
1618 (find_source): Assign index for each source file.
1619 (read_graph_file): Read new flag artificial and end_line.
1620 (add_line_counts): Assign it either to global of function scope.
1621 (accumulate_line_counts): Isolate core of the function to
1622 accumulate_line_info and call it for both function and global
1624 (accumulate_line_info): New function.
1625 (output_line_beginning): Fix GNU coding style.
1626 (print_source_line): New function.
1627 (output_line_details): Likewise.
1628 (output_function_details): Likewise.
1629 (output_lines): Iterate both source (global) scope and function
1631 (struct function_line_start_cmp): New class.
1632 * doc/gcov.texi: Reflect changes in documentation.
1634 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1637 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
1638 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
1639 and similarly for not instead of neg.
1641 2017-11-08 Andi Kleen <ak@linux.intel.com>
1643 * config/i386/i386.opt: Add -mforce-indirect-call.
1644 * config/i386/predicates.md: Check for flag_force_indirect_call.
1645 * doc/invoke.texi: Document -mforce-indirect-call
1647 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
1649 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
1651 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
1652 config/riscv/riscv.c (predict.h): New include.
1653 (riscv_slow_unaligned_access_p): No longer static.
1654 (riscv_block_move_straight): Add require.
1655 config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
1657 2017-11-08 Jakub Jelinek <jakub@redhat.com>
1660 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
1661 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
1662 nonimmediate_operand predicate for operand 1 instead of
1665 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1667 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
1669 * config/aarch64/constraints.md (Uml): New constraint.
1670 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
1673 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1675 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
1676 of two vec_duplicates into a vec_concat.
1678 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1680 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
1681 Simplify vec_merge of vec_duplicate and vec_concat.
1682 * config/aarch64/constraints.md (Utq): New constraint.
1683 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
1686 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1688 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
1689 Simplify vec_merge of vec_duplicate and const_vector.
1690 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
1692 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
1693 mode iterator. Update predicate on operand 1 to
1694 handle non-const_vec constants. Delete constraints.
1695 (*aarch64_combinez_be<mode>): Likewise for operand 2.
1697 2017-11-08 Jakub Jelinek <jakub@redhat.com>
1699 PR tree-optimization/78821
1700 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
1702 (store_operand_info::store_operand_info): Initialize it to false.
1703 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
1704 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
1705 store in the group, and if chain_info is non-NULL, to ignore altogether
1707 (compatible_load_p): Fail if bit_not_p does not match.
1708 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
1709 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
1710 (pass_store_merging::process_store): Adjust
1711 terminate_all_aliasing_chains calls to pass NULL in all current spots,
1712 call terminate_all_aliasing_chains newly when adding a store into
1713 a chain with non-NULL chain_info.
1715 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
1717 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
1719 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
1721 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
1723 (aarch64_layout_frame): Initialise emit_frame_chain.
1724 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
1725 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
1727 2017-11-08 Martin Liska <mliska@suse.cz>
1729 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
1730 of gimple_call_internal_p.
1732 2017-11-07 Tom de Vries <tom@codesourcery.com>
1734 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
1736 2017-11-07 Tom de Vries <tom@codesourcery.com>
1738 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
1741 2017-11-08 Martin Liska <mliska@suse.cz>
1744 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
1746 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
1748 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
1750 2017-11-07 Jakub Jelinek <jakub@redhat.com>
1753 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
1754 (*cmp<mode>_ccz_1): New insn with $k alternative.
1757 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
1758 RTX_COMM_COMPARE as commutative as well.
1759 (ix86_binary_operator_ok): Formatting fix.
1760 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
1761 *<code><mode>3<mask_name><round_saeonly_name>,
1762 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
1763 *<s>mul<mode>3_highpart<mask_name>,
1764 *vec_widen_umult_even_v16si<mask_name>,
1765 *vec_widen_umult_even_v8si<mask_name>,
1766 *vec_widen_umult_even_v4si<mask_name>,
1767 *vec_widen_smult_even_v16si<mask_name>,
1768 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
1769 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
1770 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
1771 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
1772 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
1773 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
1774 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
1775 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
1776 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
1777 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
1778 ix86_binary_operator_ok. Formatting fixes.
1779 (*<plusminus_insn><mode>3<mask_name><round_name>,
1780 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
1783 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
1785 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
1786 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
1787 eq and ne if TARGET_ISEL.
1788 (cmp): New code_iterator.
1789 (UNS, UNSU_, UNSIK): New code_attrs.
1790 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
1791 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
1793 ("eq<mode>3"): ... this.
1794 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
1796 ("ne<mode>3"): ... this.
1798 2017-11-07 Julia Koval <julia.koval@intel.com>
1801 * common/config/i386/i386-common.c
1802 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
1803 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
1804 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
1805 * config/i386/i386.opt: Ditto.
1806 * config/i386/i386.c (ix86_target_string): Ditto.
1807 (ix86_option_override_internal): Ditto.
1808 (ix86_init_mpx_builtins): Move MPX to args2.
1809 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
1810 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
1811 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
1812 __builtin_ia32_vgf2p8affineinvqb_v32qi,
1813 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
1814 __builtin_ia32_vgf2p8affineinvqb_v16qi,
1815 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
1817 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
1820 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
1821 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
1822 (zero-extendsidi peephole2): Remove peephole.
1824 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
1827 * stor-layout.c: Include attribs.h.
1828 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
1829 explicit lookup of "aligned" attribute.
1831 2017-11-07 Andrew Waterman <andrew@sifive.com>
1833 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
1834 (riscv_expand_block_move): Likewise.
1835 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
1837 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
1838 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
1839 * config/riscv/riscv.c (riscv_block_move_straight): New function.
1840 (riscv_adjust_block_mem): Likewise.
1841 (riscv_block_move_loop): Likewise.
1842 (riscv_expand_block_move): Likewise.
1843 gcc/config/riscv/riscv.md (movmemsi): New pattern.
1845 2017-11-07 Michael Clark <michaeljclark@mac.com>
1847 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
1848 (MUSL_DYNAMIC_LINKER): Likewise.
1850 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
1852 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
1853 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
1856 2017-11-07 Richard Biener <rguenther@suse.de>
1858 * match.pd: Fix build.
1860 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
1861 Jackson Woodruff <jackson.woodruff@arm.com>
1863 PR tree-optimization/71026
1864 * match.pd: Canonicalize negate in division.
1866 2017-11-07 Sudakshina Das <sudi.das@arm.com>
1869 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
1871 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
1873 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
1874 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
1876 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
1878 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
1879 non-scalar integral types.
1880 * match.pd (negate_expr_p): Handle MINUS_EXPR.
1881 (-(A-B), -(~A)): New transformations.
1883 2017-11-07 Tom de Vries <tom@codesourcery.com>
1885 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
1886 semicolon after "do {} while (0)".
1887 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1888 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1889 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1890 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1891 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1893 2017-11-07 Tom de Vries <tom@codesourcery.com>
1895 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
1896 after "do {} while (0)".
1897 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1898 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1899 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1900 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1901 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1903 2017-11-07 Tom de Vries <tom@codesourcery.com>
1906 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
1908 (arm_rtx_costs_internal): Add missing semicolon after
1909 HANDLE_NARROW_SHIFT_ARITH call.
1911 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
1913 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
1914 disable isel if it was not set explicitly.
1916 2017-11-06 James Bowman <james.bowman@ftdichip.com>
1918 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE
1920 (add_type_attribute) likewise.
1922 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
1924 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
1925 of struct ix86_frame.
1926 (ix86_initial_elimination_offset): Likewise.
1927 (ix86_expand_split_stack_prologue): Likewise.
1929 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
1931 * tree-vrp.h (enum value_range_type): Update stale comment.
1933 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1934 Alan Hayward <alan.hayward@arm.com>
1935 David Sherwood <david.sherwood@arm.com>
1937 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
1938 (aarch64_expand_vec_perm_const): Take the number of units too.
1939 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
1940 (aarch64_expand_vec_perm_const): Likewise.
1941 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
1942 (vec_perm<mode>): Update accordingly.
1944 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1945 Alan Hayward <alan.hayward@arm.com>
1946 David Sherwood <david.sherwood@arm.com>
1948 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
1949 Take the number of units too.
1950 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
1951 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
1952 but check for a vector mode before rather than after the call.
1953 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
1954 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
1955 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
1956 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
1957 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
1958 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
1959 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
1960 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
1961 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
1962 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
1963 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
1964 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
1965 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
1966 (aarch64_sqdmull2_n<mode>): Update accordingly.
1968 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1969 Alan Hayward <alan.hayward@arm.com>
1970 David Sherwood <david.sherwood@arm.com>
1972 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
1973 the number of units too.
1974 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
1975 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
1976 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
1977 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
1978 (vec_store_lanesxi<mode>): Update accordingly.
1980 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1981 Alan Hayward <alan.hayward@arm.com>
1982 David Sherwood <david.sherwood@arm.com>
1984 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
1985 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
1986 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
1987 of units rather than the mode.
1988 * config/aarch64/iterators.md (nunits): New mode attribute.
1989 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
1990 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
1991 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
1992 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
1993 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
1994 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
1995 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
1996 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
1997 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
1998 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
1999 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
2000 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
2001 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
2002 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
2003 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
2004 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
2005 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
2006 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
2007 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
2008 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
2009 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
2010 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
2011 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
2012 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
2013 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
2014 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
2015 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
2016 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
2017 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
2018 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
2019 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
2020 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
2021 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
2022 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
2023 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
2024 (aarch64_simd_vec_setv2di): Likewise.
2026 2017-11-06 Carl Love <cel@us.ibm.com>
2028 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
2030 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
2031 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
2033 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
2035 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
2036 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
2037 Add power 8 macro expansions.
2038 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
2039 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
2040 power 8 instructions. (VSX_XXBR): Add iterator.
2042 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
2044 * config/arm/arm.md (predicable_short_it): Change default to "no",
2045 improve documentation, remove uses that are identical to the default.
2046 (enabled_for_depr_it): Rename to enabled_for_short_it.
2047 * gcc/config/arm/arm-fixed.md (predicable_short_it):
2048 Remove default uses.
2049 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
2050 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
2051 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
2052 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
2054 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
2057 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
2058 float128 helper macros, which are no longer used after deleting
2059 the old 'q' built-in functions, and moving the round to odd
2060 built-in functions to being special built-in functions.
2061 (BU_FLOAT128_2): Likewise.
2062 (BU_FLOAT128_1_HW): Likewise.
2063 (BU_FLOAT128_2_HW): Likewise.
2064 (BU_FLOAT128_3_HW): Likewise.
2065 (FABSQ): Delete old 'q' built-in functions.
2066 (COPYSIGNQ): Likewise.
2067 (SQRTF128_ODD): Move round to odd built-in functions to be
2068 special built-in functions, so that we can handle
2069 -mabi=ieeelongdouble.
2070 (TRUNCF128_ODD): Likewise.
2071 (ADDF128_ODD): Likewise.
2072 (SUBF128_ODD): Likewise.
2073 (MULF128_ODD): Likewise.
2074 (DIVF128_ODD): Likewise.
2075 (FMAF128_ODD): Likewise.
2076 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
2077 built-in names to 'f128'.
2078 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
2079 old 'q' built-in functions, as the machine independent code for
2080 'f128' built-in functions handles this.
2081 (rs6000_expand_builtin): Add expansion for float128 round to odd
2082 functions, keying off on -mabi=ieeelongdouble of whether to use
2083 the KFmode or TFmode variant.
2084 (rs6000_init_builtins): Initialize the _Float128 round to odd
2086 * doc/extend.texi (PowerPC Built-in Functions): Document the old
2087 _Float128 'q' built-in functions are now mapped into the new
2088 'f128' built-in functions.
2090 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
2092 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
2093 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
2095 2017-11-06 David Malcolm <dmalcolm@redhat.com>
2098 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
2099 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
2100 * toplev.c: Include "ipa-fnsummary.h".
2101 (toplev::finalize): Call ipa_fnsummary_c_finalize.
2103 2017-11-06 Jakub Jelinek <jakub@redhat.com>
2105 PR tree-optimization/82838
2106 * gimple-ssa-store-merging.c
2107 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
2108 on a separate gimple_seq which is then appended to seq.
2110 2017-11-06 Jeff Law <law@redhat.com>
2113 * config/i386/i386.c (PROBE_INTERVAL): Remove.
2114 (get_probe_interval): New functions.
2115 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
2116 (ix86_adjust_stack_and_probe): Likewise.
2117 (output_adjust_stack_and_probe): Likewise.
2118 (ix86_emit_probe_stack_range): Likewise.
2119 (ix86_expand_prologue): Likewise.
2121 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2123 PR tree-optimization/82816
2124 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
2125 if the modes of the two types are the same.
2126 (convert_plusminus_to_widen): Likewise.
2128 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2130 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
2132 (usadv16qi): New define_expand.
2133 (usadv8hi): New define_expand.
2135 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
2138 * ipa-inline-transform.c (update_noncloned_frequencies): Always
2140 (inline_transform): Likewise.
2141 * predict.c (counts_to_freqs): Remove useless conditional.
2142 * profile-count.h (profile_count::apply_scale): Move sanity check.
2143 * tree-inline.c (copy_bb): Always scale.
2144 (copy_cfg_body): Likewise.
2146 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
2149 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
2151 (*cmp_ite0): Add enabled_for_depr_it attribute.
2152 (*cmp_ite1): Likewise.
2154 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
2156 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
2159 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2161 * tree-vrp.c (vrp_int_const_binop): Return true on success and
2162 return the value by pointer.
2163 (extract_range_from_multiplicative_op_1): Update accordingly.
2164 Return as soon as an operation fails.
2166 2017-11-05 Tom de Vries <tom@codesourcery.com>
2169 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
2170 (DEF_SANITIZER_BUILTIN): ... here.
2171 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
2172 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
2174 2017-11-05 Tom de Vries <tom@codesourcery.com>
2177 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
2179 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
2180 ASM_OUTPUT_BEFORE_CASE_LABEL call.
2181 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
2183 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
2184 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
2185 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
2187 2017-11-05 Tom de Vries <tom@codesourcery.com>
2190 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
2193 2017-11-04 Michael Clark <michaeljclark@mac.com>
2195 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
2196 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
2198 (*addsi3_extended): Likewise.
2199 (*addsi3_extended2): Likewise.
2200 (<optab>si3): Likewise.
2201 (<optab>di3): Likewise.
2202 (<optab><mode>3): Likewise.
2203 (<*optabe>si3_internal): Likewise.
2204 (zero_extendqi<SUPERQI:mode>2): Likewise.
2205 (*add<mode>hi3): Likewise.
2206 (*xor<mode>hi3): Likewise.
2207 (<optab>di3): Likewise.
2208 (*<optab>si3_extend): Likewise.
2209 (*sge<u>_<X:mode><GPR:mode>): Likewise.
2210 (*slt<u>_<X:mode><GPR:mode>): Likewise.
2211 (*sle<u>_<X:mode><GPR:mode>): Likewise.
2213 2017-11-04 Andrew Waterman <andrew@sifive.com>
2215 * config/riscv/riscv.c (riscv_option_override): Conditionally set
2216 TARGET_STRICT_ALIGN based upon -mtune argument.
2218 2017-11-04 Andrew Waterman <andrew@sifive.com>
2220 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
2222 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
2224 config/i386/i386.c (choose_basereg): Use optional scratch
2225 register and add assertion.
2226 (x86_emit_outlined_ms2sysv_save): Use scratch register when
2227 needed, and don't allocate stack.
2228 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
2229 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
2230 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
2232 2017-11-03 Jeff Law <law@redhat.com>
2234 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
2235 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
2236 to probe at the start of a noreturn function.
2238 2017-11-03 Jakub Jelinek <jakub@redhat.com>
2240 PR tree-optimization/78821
2241 * gimple-ssa-store-merging.c: Update the file comment.
2242 (MAX_STORE_ALIAS_CHECKS): Define.
2243 (struct store_operand_info): New type.
2244 (store_operand_info::store_operand_info): New constructor.
2245 (struct store_immediate_info): Add rhs_code and ops data members.
2246 (store_immediate_info::store_immediate_info): Add rhscode, op0r
2247 and op1r arguments to the ctor, initialize corresponding data members.
2248 (struct merged_store_group): Add load_align_base and load_align
2250 (merged_store_group::merged_store_group): Initialize them.
2251 (merged_store_group::do_merge): Update them.
2252 (merged_store_group::apply_stores): Pick the constant for
2253 encode_tree_to_bitpos from one of the two operands, or skip
2254 encode_tree_to_bitpos if neither operand is a constant.
2255 (class pass_store_merging): Add process_store method decl. Remove
2256 bool argument from terminate_all_aliasing_chains method decl.
2257 (pass_store_merging::terminate_all_aliasing_chains): Remove
2258 var_offset_p argument and corresponding handling.
2259 (stmts_may_clobber_ref_p): New function.
2260 (compatible_load_p): New function.
2261 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
2262 if there is overlap and rhs_code is not INTEGER_CST. For
2263 non-overlapping stores terminate group if rhs is not mergeable.
2264 (get_alias_type_for_stmts): Change first argument from
2265 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
2266 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
2267 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
2269 (get_location_for_stmts): Change first argument from
2270 auto_vec<gimple *> & to vec<gimple *> &.
2271 (struct split_store): Remove orig_stmts data member, add orig_stores.
2272 (split_store::split_store): Create orig_stores rather than orig_stmts.
2273 (find_constituent_stmts): Renamed to ...
2274 (find_constituent_stores): ... this. Change second argument from
2275 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
2276 to info structures rather than the statements.
2277 (split_group): Rename ALLOW_UNALIGNED argument to
2278 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
2279 it. Adjust find_constituent_stores caller.
2280 (imm_store_chain_info::output_merged_store): Handle rhs_code other
2281 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
2282 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
2283 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
2284 (mem_valid_for_store_merging): New function.
2285 (handled_load): New function.
2286 (pass_store_merging::process_store): New method.
2287 (pass_store_merging::execute): Use process_store method. Adjust
2288 terminate_all_aliasing_chains caller.
2290 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2292 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
2293 Return true for more constants, symbols and label references.
2294 (aarch64_valid_floating_const): Remove unused function.
2296 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2299 * config/aarch64/aarch64.c (aarch64_layout_frame):
2300 Undo forcing of LR at bottom of frame.
2302 2017-11-03 Jeff Law <law@redhat.com>
2305 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
2306 for int_registers_saved.
2308 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
2309 extracted from tree-ssa-dom.c.
2310 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
2311 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
2312 (record_equivalences_from_incoming_edge): Add additional argument
2313 to single_pred_edge_ignoring_loop_edges call.
2314 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
2315 (uncprop_dom_walker::before_dom_children): Add additional argument
2316 to single_pred_edge_ignoring_loop_edges call.
2317 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
2318 single_pred_edge_ignoring_loop_edges rather than open coding.
2319 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
2321 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
2323 * match.pd (-(-A)): Rewrite.
2325 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2327 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
2328 (rs6000_emit_int_cmove): New declaration.
2329 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
2330 (rs6000_emit_sISEL): Delete.
2331 (rs6000_emit_int_cmove): Make non-static.
2332 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
2333 instead of rs6000_emit_sISEL.
2335 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
2337 * asan.c (create_cond_insert_point): Maintain profile.
2338 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
2340 * basic-block.h (struct basic_block_def): Remove frequency.
2341 (EDGE_FREQUENCY): Use to_frequency
2342 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
2344 (find_traces): Update to use to_frequency.
2345 (find_traces_1_round): Likewise; use only IPA counts.
2346 (bb_to_key): Likewise.
2347 (connect_traces): Use IPA counts only.
2348 (copy_bb_p): Update to use to_frequency.
2349 (fix_up_crossing_landing_pad): Likewise.
2350 (sanitize_hot_paths): Likewise.
2351 * bt-load.c (basic_block_freq): Likewise.
2352 * cfg.c (init_flow): Set count_max to uninitialized.
2353 (check_bb_profile): Remove frequencies; check counts.
2354 (dump_bb_info): Do not dump frequencies.
2355 (update_bb_profile_for_threading): Update counts only.
2356 (scale_bbs_frequencies_int): Likewise.
2357 (MAX_SAFE_MULTIPLIER): Remove.
2358 (scale_bbs_frequencies_gcov_type): Update counts only.
2359 (scale_bbs_frequencies_profile_count): Update counts only.
2360 (scale_bbs_frequencies): Update counts only.
2361 * cfg.h (struct control_flow_graph): Add count-max.
2362 (update_bb_profile_for_threading): Update prototype.
2363 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
2364 (find_many_sub_basic_blocks): Likewise.
2365 * cfgcleanup.c (try_forward_edges): Likewise.
2366 (try_crossjump_to_edge): Likewise.
2367 * cfgexpand.c (expand_gimple_cond): Likewise.
2368 (expand_gimple_tailcall): Likewise.
2369 (construct_init_block): Likewise.
2370 (construct_exit_block): Likewise.
2371 * cfghooks.c (verify_flow_info): Check consistency of counts.
2372 (dump_bb_for_graph): Do not dump frequencies.
2373 (split_block_1): Do not update frequencies.
2374 (split_edge): Do not update frequencies.
2375 (make_forwarder_block): Do not update frequencies.
2376 (duplicate_block): Do not update frequencies.
2377 (account_profile_record): Do not update frequencies.
2378 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
2379 for global heuristics.
2380 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
2381 (expected_loop_iterations_unbounded): Use counts only.
2382 * cfgloopmanip.c (scale_loop_profile): Simplify.
2383 (create_empty_loop_on_edge): Simplify
2385 (duplicate_loop_to_header_edge): Simplify
2386 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
2387 (update_br_prob_note): Take care of removing note when profile
2389 (relink_block_chain): Do not dump frequency.
2390 (rtl_account_profile_record): Use to_frequency.
2391 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
2392 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
2393 (cgraph_update_edges_for_call_stmt_node): Likewise.
2394 (cgraph_edge::verify_count_and_frequency): Update.
2395 (cgraph_node::verify_node): Temporarily disable frequency verification.
2396 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
2397 to_cgraph_frequency.
2398 (cgraph_edge::rebuild_edges): Convert to ipa counts.
2399 * cgraphunit.c (init_lowered_empty_function): Do not initialize
2401 (cgraph_node::expand_thunk): Update profile.
2402 * except.c (dw2_build_landing_pads): Do not update frequency.
2403 * final.c (compute_alignments): Use to_frequency.
2404 (dump_basic_block_info): Do not dump frequency.
2405 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
2406 (dump_gimple_bb_header): Do not dump frequency.
2407 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
2409 * gimple-streamer-in.c (input_bb): Do not stream frequency.
2410 * gimple-streamer-out.c (output_bb): Do not stream frequency.
2411 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
2412 (init_before_recovery): Do not update frequency.
2413 (sched_create_recovery_edges): Do not update frequency.
2414 * hsa-gen.c (convert_switch_statements): Do not update frequency.
2415 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
2416 (ipa_cp_c_finalize): Set max_count to uninitialized.
2417 * ipa-fnsummary.c (get_minimal_bb): Use counts.
2418 (param_change_prob): Use counts.
2419 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
2421 * ipa-split.c (consider_split): Use to_frequency.
2422 (split_function): Use to_frequency.
2423 * ira-build.c (loop_compare_func): Likewise.
2424 (mark_loops_for_removal): Likewise.
2425 (mark_all_loops_for_removal): Likewise.
2426 * loop-doloop.c (doloop_modify): Do not update frequency.
2427 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
2429 * lto-streamer-in.c (input_function): Update count_max.
2430 * omp-expand.c (expand_omp_taskreg): Update count_max.
2431 * omp-simd-clone.c (simd_clone_adjust): Update profile.
2432 * predict.c (maybe_hot_frequency_p): Use to_frequency.
2433 (maybe_hot_count_p): Use ipa counts only.
2434 (maybe_hot_bb_p): Simplify.
2435 (maybe_hot_edge_p): Simplify.
2436 (probably_never_executed): Do not take frequency argument.
2437 (probably_never_executed_bb_p): Do not pass frequency.
2438 (probably_never_executed_edge_p): Likewise.
2439 (combine_predictions_for_bb): Check that profile is nonzero.
2440 (propagate_freq): Do not set frequency.
2441 (drop_profile): Simplify.
2442 (counts_to_freqs): Simplify.
2443 (expensive_function_p): Use to_frequency.
2444 (propagate_unlikely_bbs_forward): Simplify.
2445 (determine_unlikely_bbs): Simplify.
2446 (estimate_bb_frequencies): Add hack to silence graphite issues.
2447 (compute_function_frequency): Use ipa counts.
2448 (pass_profile::execute): Update.
2449 (rebuild_frequencies): Use counts only.
2450 (force_edge_cold): Use counts only.
2451 * profile-count.c (profile_count::dump): Dump new count types.
2452 (profile_count::differs_from_p): Check compatiblity.
2453 (profile_count::to_frequency): New function.
2454 (profile_count::to_cgraph_frequency): New function.
2455 * profile-count.h (struct function): Declare.
2456 (enum profile_quality): Add profile_guessed_local and
2457 profile_guessed_global0.
2458 (class profile_proability): Decrease number of bits to 29;
2459 update from_reg_br_prob_note and to_reg_br_prob_note.
2460 (class profile_count: Update comment; decrease number of bits
2461 to 61. Check compatibility.
2462 (profile_count::compatible_p): New private member function.
2463 (profile_count::ipa_p): New member function.
2464 (profile_count::operator<): Handle global zero correctly.
2465 (profile_count::operator>): Handle global zero correctly.
2466 (profile_count::operator<=): Handle global zero correctly.
2467 (profile_count::operator>=): Handle global zero correctly.
2468 (profile_count::nonzero_p): New member function.
2469 (profile_count::force_nonzero): New member function.
2470 (profile_count::max): New member function.
2471 (profile_count::apply_scale): Handle IPA scalling.
2472 (profile_count::guessed_local): New member function.
2473 (profile_count::global0): New member function.
2474 (profile_count::ipa): New member function.
2475 (profile_count::to_frequency): Declare.
2476 (profile_count::to_cgraph_frequency): Declare.
2477 * profile.c (OVERLAP_BASE): Delete.
2478 (compute_frequency_overlap): Delete.
2479 (compute_branch_probabilities): Do not use compute_frequency_overlap.
2480 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
2481 * sched-ebb.c (rank): Use counts only.
2482 * shrink-wrap.c (handle_simple_exit): Use counts only.
2483 (try_shrink_wrapping): Use counts only.
2484 (place_prologue_for_one_component): Use counts only.
2485 * tracer.c (find_best_predecessor): Use to_frequency.
2486 (find_trace): Use to_frequency.
2487 (tail_duplicate): Use to_frequency.
2488 * trans-mem.c (expand_transaction): Do not update frequency.
2489 * tree-call-cdce.c: Do not update frequency.
2490 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
2491 (gimple_merge_blocks): Likewise.
2492 (gimple_split_edge): Likewise.
2493 (gimple_duplicate_sese_region): Likewise.
2494 (gimple_duplicate_sese_tail): Likewise.
2495 (move_sese_region_to_fn): Likewise.
2496 (gimple_account_profile_record): Likewise.
2497 (insert_cond_bb): Likewise.
2498 * tree-complex.c (expand_complex_div_wide): Likewise.
2499 * tree-eh.c (lower_resx): Update profile.
2500 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
2502 (initialize_cfun): Do not initialize frequencies
2503 (freqs_to_counts): Delete.
2504 (copy_cfg_body): Ignore count parameter.
2505 (copy_body): Update.
2506 (expand_call_inline): Update count_max.
2507 (optimize_inline_calls): Update count_max.
2508 (tree_function_versioning): Update count_max.
2509 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
2510 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
2512 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
2513 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
2514 (try_peel_loop): Likewise.
2515 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
2517 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
2518 (tree_transform_and_unroll_loop): Do not use frequencies
2519 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
2520 Use reliable prediction only.
2521 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
2522 * tree-ssa-sink.c (select_best_block): Use to_frequency.
2523 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
2524 probability scaling.
2525 * tree-ssa-threadupdate.c (create_block_for_threading): Do
2526 not update frequency
2527 (any_remaining_duplicated_blocks): Likewise.
2528 (update_profile): Likewise.
2529 (estimated_freqs_path): Delete.
2530 (freqs_to_counts_path): Delete.
2531 (clear_counts_path): Delete.
2532 (ssa_fix_duplicate_block_edges): Likewise.
2533 (duplicate_thread_path): Likewise.
2534 * tree-switch-conversion.c (gen_inbound_check): Use counts.
2535 * tree-tailcall.c (decrease_profile): Do not update frequency.
2536 (eliminate_tail_call): Likewise.
2537 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
2538 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
2539 (optimize_mask_stores): Likewise.
2540 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
2541 * ubsan.c (ubsan_expand_null_ifn): Update profile.
2542 (ubsan_expand_ptr_ifn): Update profile.
2543 * value-prof.c (gimple_ic): Simplify.
2544 * value-prof.h (gimple_ic): Update prototype.
2545 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
2546 * ipa-inline.c (compute_uninlined_call_time): Be sure that
2548 (want_inline_self_recursive_call_p): Likewise.
2549 (resolve_noninline_speculation): Only cummulate defined counts.
2550 (inline_small_functions): Use nonzero_p.
2551 (ipa_inline): Do not access freed node.
2553 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2555 * config/aarch64/aarch64.c (aarch64_override_options_internal):
2556 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
2558 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
2560 * config/riscv/riscv.c (riscv_legitimize_move): Handle
2561 non-legitimate address.
2563 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2565 * config/rs6000/rs6000.md (*lt0_disi): Delete.
2566 (*lt0_<mode>di, *lt0_<mode>si): New.
2568 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2570 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
2571 TARGET_PAIRED_FLOAT.
2573 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
2574 Jim Wilson <jim.wilson@linaro.org>
2576 * config/aarch64/aarch64-cores.def (saphira): New CPU.
2577 * config/aarch64/aarch64-tune.md: Regenerated.
2578 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
2579 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
2581 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
2583 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
2584 (arc_expand_prologue): Restore blink for millicode.
2585 * config/arc/linux.h (LINK_EH_SPEC): Defined.
2587 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
2590 * config/i386/i386.c (ix86_vector_duplicate_value): Use
2591 gen_vec_duplicate after forcing the scalar into a register.
2593 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
2595 * combine (try_combine): Print the insns input to try_combine to the
2598 2017-11-02 Steve Ellcey <sellcey@cavium.com>
2601 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
2602 Remove second argument from aarch64_process_target_attr call.
2603 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
2605 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
2607 (aarch64_handle_attr_arch): Remove second argument.
2608 (aarch64_handle_attr_cpu): Ditto.
2609 (aarch64_handle_attr_tune): Ditto.
2610 (aarch64_handle_attr_isa_flags): Ditto.
2611 (aarch64_process_one_target_attr): Ditto.
2612 (aarch64_process_target_attr): Ditto.
2613 (aarch64_option_valid_attribute_p): Remove second argument.
2614 on aarch64_process_target_attr call.
2616 2017-11-02 David Malcolm <dmalcolm@redhat.com>
2618 * diagnostic.c: Include "selftest-diagnostic.h".
2619 (selftest::assert_location_text): New function.
2620 (selftest::test_diagnostic_get_location_text): New function.
2621 (selftest::diagnostic_c_tests): Call it.
2623 2017-11-02 David Malcolm <dmalcolm@redhat.com>
2625 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
2626 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
2627 (class selftest::test_diagnostic_context): Move to...
2628 * selftest-diagnostic.c: New file.
2629 * selftest-diagnostic.h: New file.
2631 2017-11-02 James Bowman <james.bowman@ftdichip.com>
2633 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
2634 offset range for FT32B.
2635 * config/ft32/ft32.h: option "mcompress" enables relaxation.
2636 * config/ft32/ft32.md: Add TARGET_NOPM.
2637 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
2638 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
2640 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
2642 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
2644 2017-11-02 Jeff Law <law@redhat.com>
2646 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
2647 virtual keyword on FINAL OVERRIDE members.
2649 * tree-ssa-propagate.h (ssa_propagation_engine): Group
2650 virtuals together. Add virtual destructor.
2651 (substitute_and_fold_engine): Similarly.
2653 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
2655 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
2657 2017-11-02 Richard Biener <rguenther@suse.de>
2659 PR tree-optimization/82795
2660 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
2662 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2664 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
2666 (gcc_GAS_CHECK_FEATURE): Remove.
2667 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
2668 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
2669 * configure: Regenerate.
2671 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
2673 * config/arc/arc.c (hwloop_optimize): Account for empty
2676 2017-11-02 Richard Biener <rguenther@suse.de>
2679 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
2680 Truncate ARRAY_REF index and element size.
2682 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
2684 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
2686 2017-11-01 Jeff Law <law@redhat.com>
2688 * tree-ssa-ccp.c (ccp_folder): New class derived from
2689 substitute_and_fold_engine.
2690 (ccp_folder::get_value): New member function.
2691 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
2692 (ccp_fold_stmt): Remove prototype.
2693 (ccp_finalize): Call substitute_and_fold from the ccp_class.
2694 * tree-ssa-copy.c (copy_folder): New class derived from
2695 substitute_and_fold_engine.
2696 (copy_folder::get_value): Renamed from get_value.
2697 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
2698 * tree-vrp.c (vrp_folder): New class derived from
2699 substitute_and_fold_engine.
2700 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
2701 (vrp_folder::get_value): New member function.
2702 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
2703 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
2704 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
2705 provide a class interface to folder/substitute routines.
2706 (ssa_prop_fold_stmt_fn): Remove typedef.
2707 (ssa_prop_get_value_fn): Likewise.
2708 (subsitute_and_fold): Remove prototype.
2709 (replace_uses_in): Likewise.
2710 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
2711 Renamed from replace_uses_in. Call the virtual member function
2712 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
2713 (substitute_and_fold_dom_walker): Remove initialization of
2714 data member entries for calbacks. Add substitute_and_fold_engine
2715 member and initialize it.
2716 (substitute_and_fold_dom_walker::before_dom_children0: Use the
2717 member functions for get_value, replace_phi_args_in c
2718 replace_uses_in, and fold_stmt calls.
2719 (substitute_and_fold_engine::substitute_and_fold): Renamed from
2720 substitute_and_fold. Remove assert. Update ctor call.
2722 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
2723 (ssa_prop_visit_phi_fn): Likewise.
2724 (class ssa_propagation_engine): New class to provide an interface
2726 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
2728 (ssa_prop_visit_phi): Likewise.
2729 (ssa_propagation_engine::simulate_stmt): Moved into class.
2730 Call visit_phi/visit_stmt from the class rather than via
2731 file scoped static variables.
2732 (ssa_propagation_engine::simulate_block): Moved into class.
2733 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
2734 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
2735 set file scoped statics for the visit_stmt/visit_phi callbacks.
2736 * tree-complex.c (complex_propagate): New class derived from
2737 ssa_propagation_engine.
2738 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
2739 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
2740 (tree_lower_complex): Call ssa_propagate via the complex_propagate
2742 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
2743 ssa_propagation_engine.
2744 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
2745 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
2746 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
2747 * tree-ssa-copy.c (copy_prop): New class derived from
2748 ssa_propagation_engine.
2749 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
2750 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
2751 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
2752 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
2753 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
2754 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
2755 (execute_vrp): Call ssa_propagate from the vrp_prop class.
2757 2017-11-01 Jakub Jelinek <jakub@redhat.com>
2759 PR rtl-optimization/82778
2760 PR rtl-optimization/82597
2761 * compare-elim.c (struct comparison): Add in_a_setter field.
2762 (find_comparison_dom_walker::before_dom_children): Remove killed
2763 bitmap and df_simulate_find_defs call, instead walk the defs.
2764 Compute last_setter and initialize in_a_setter. Merge definitions
2765 with first initialization for a few variables.
2766 (try_validate_parallel): Use insn_invalid_p instead of
2767 recog_memoized. Return insn rather than just the pattern.
2768 (try_merge_compare): Fix up comment. Don't uselessly test if
2769 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
2771 (execute_compare_elim_after_reload): Remove df_chain_add_problem
2774 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2775 Alan Hayward <alan.hayward@arm.com>
2776 David Sherwood <david.sherwood@arm.com>
2778 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
2779 aarch64_hard_regno_nregs to get the number of registers
2782 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2783 Alan Hayward <alan.hayward@arm.com>
2784 David Sherwood <david.sherwood@arm.com>
2786 * config/aarch64/constraints.md (Upl): Rename to...
2788 * config/aarch64/aarch64.md
2789 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
2792 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2793 Alan Hayward <alan.hayward@arm.com>
2794 David Sherwood <david.sherwood@arm.com>
2796 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
2797 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
2800 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2801 Alan Hayward <alan.hayward@arm.com>
2802 David Sherwood <david.sherwood@arm.com>
2804 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
2805 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
2806 (aarch64_evpc_dup): Generate rtl direcly, rather than using
2808 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
2809 of a single element.
2810 * config/aarch64/iterators.md: Add a comment above the permute
2811 unspecs to say that they are generated directly by
2812 aarch64_expand_vec_perm_const.
2813 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
2815 2017-11-01 Nathan Sidwell <nathan@acm.org>
2817 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
2819 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
2821 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
2822 and medany code models, and describe what they do.
2824 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2826 Revert accidental duplicate:
2828 * combine.c (can_change_dest_mode): Reject changes in
2829 REGMODE_NATURAL_SIZE.
2831 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
2833 PR rtl-optimization/64682
2834 PR rtl-optimization/69567
2835 PR rtl-optimization/69737
2836 PR rtl-optimization/82683
2837 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
2838 register mentioned in the note, drop the note, unless it came from I3,
2839 in which case it should go to I3 again.
2841 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2843 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
2844 and return false if not.
2845 (clear_bytes_written_by, live_bytes_read): Update accordingly.
2847 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2849 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
2850 range is known to be empty.
2852 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2853 Alan Hayward <alan.hayward@arm.com>
2854 David Sherwood <david.sherwood@arm.com>
2856 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
2857 and CONST_VECTOR_NUNITS instead of computing the number of units from
2858 the byte sizes of the vector and element.
2859 (simplify_binary_operation_1): Likewise.
2860 (simplify_const_binary_operation): Likewise.
2861 (simplify_ternary_operation): Likewise.
2863 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2864 Alan Hayward <alan.hayward@arm.com>
2865 David Sherwood <david.sherwood@arm.com>
2867 * var-tracking.c (INT_MEM_OFFSET): Replace with...
2868 (int_mem_offset): ...this new function.
2869 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
2870 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
2871 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
2874 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2875 Alan Hayward <alan.hayward@arm.com>
2876 David Sherwood <david.sherwood@arm.com>
2878 * lower-subreg.c (interesting_mode_p): New function.
2879 (compute_costs, find_decomposable_subregs, decompose_register)
2880 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
2881 (resolve_clobber, dump_choices): Use it.
2883 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2884 Alan Hayward <alan.hayward@arm.com>
2885 David Sherwood <david.sherwood@arm.com>
2887 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
2889 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2890 Alan Hayward <alan.hayward@arm.com>
2891 David Sherwood <david.sherwood@arm.com>
2893 * alias.c (find_base_value, find_base_term): Only process integer
2894 truncations. Check the precision rather than the size.
2896 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2897 Alan Hayward <alan.hayward@arm.com>
2898 David Sherwood <david.sherwood@arm.com>
2900 * machmode.h (is_narrower_int_mode): New function
2901 * optabs.c (expand_float, expand_fix): Use it.
2902 * dwarf2out.c (rotate_loc_descriptor): Likewise.
2904 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2905 Alan Hayward <alan.hayward@arm.com>
2906 David Sherwood <david.sherwood@arm.com>
2908 * rtl.h (narrower_subreg_mode): New function.
2909 * ira-color.c (update_costs_from_allocno): Use it.
2911 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2912 Alan Hayward <alan.hayward@arm.com>
2913 David Sherwood <david.sherwood@arm.com>
2915 * optabs-query.h (convert_optab_p): New function, split out from...
2916 (convert_optab_handler): ...here.
2917 (widening_optab_handler): Delete.
2918 (find_widening_optab_handler): Remove permit_non_widening parameter.
2919 (find_widening_optab_handler_and_mode): Likewise. Provide an
2920 override that operates on mode class wrappers.
2921 * optabs-query.c (widening_optab_handler): Delete.
2922 (find_widening_optab_handler_and_mode): Remove permit_non_widening
2923 parameter. Assert that the two modes are the same class and that
2924 the "from" mode is narrower than the "to" mode. Use
2925 convert_optab_handler instead of widening_optab_handler.
2926 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
2927 instead of widening_optab_handler.
2928 * expr.c (expand_expr_real_2): Update calls to
2929 find_widening_optab_handler.
2930 * optabs.c (expand_widen_pattern_expr): Likewise.
2931 (expand_binop_directly): Take the insn_code as a parameter.
2932 (expand_binop): Only call find_widening_optab_handler for
2933 conversion optabs; use optab_handler otherwise. Update calls
2934 to find_widening_optab_handler and expand_binop_directly.
2935 Use convert_optab_handler instead of widening_optab_handler.
2936 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
2937 find_widening_optab_handler and use scalar_mode rather than
2939 (convert_plusminus_to_widen): Likewise.
2941 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2942 Alan Hayward <alan.hayward@arm.com>
2943 David Sherwood <david.sherwood@arm.com>
2945 * machmode.h (fixed_size_mode): New class.
2946 * rtl.h (get_pool_mode): Return fixed_size_mode.
2947 * gengtype.c (main): Add fixed_size_mode.
2948 * target.def (get_raw_result_mode): Return a fixed_size_mode.
2949 (get_raw_arg_mode): Likewise.
2950 * doc/tm.texi: Regenerate.
2951 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
2952 * targhooks.c (default_get_reg_raw_mode): Likewise.
2953 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
2954 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
2955 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
2956 (msp430_get_raw_result_mode): Likewise.
2957 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
2958 * dbxout.c (dbxout_parms): Require fixed-size modes.
2959 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
2960 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
2961 * omp-low.c (lower_oacc_reductions): Likewise.
2962 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
2963 (simplify_subreg): Update accordingly.
2964 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
2965 (force_const_mem): Update accordingly. Return NULL_RTX for modes
2966 that aren't fixed-size.
2967 (get_pool_mode): Return a fixed_size_mode.
2968 (output_constant_pool_2): Take a fixed_size_mode.
2970 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2971 Alan Hayward <alan.hayward@arm.com>
2972 David Sherwood <david.sherwood@arm.com>
2974 * doc/rtl.texi (vec_series): Document.
2975 (const): Say that the operand can be a vec_series.
2976 * rtl.def (VEC_SERIES): New rtx code.
2977 * rtl.h (const_vec_series_p_1): Declare.
2978 (const_vec_series_p): New function.
2979 * emit-rtl.h (gen_const_vec_series): Declare.
2980 (gen_vec_series): Likewise.
2981 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
2982 (gen_vec_series): Likewise.
2983 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
2984 * simplify-rtx.c (simplify_unary_operation): Handle negations
2986 (simplify_binary_operation_series): New function.
2987 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
2988 (test_vector_ops_series): New function.
2989 (test_vector_ops): Call it.
2990 * config/powerpcspe/altivec.md (altivec_lvsl): Use
2991 gen_const_vec_series.
2992 (altivec_lvsr): Likewise.
2993 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
2995 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2996 Alan Hayward <alan.hayward@arm.com>
2997 David Sherwood <david.sherwood@arm.com>
2999 * doc/rtl.texi (const): Update description of address constants.
3000 Say that vector constants are allowed too.
3001 * common.md (E, F): Use CONSTANT_P instead of checking for
3003 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
3004 checking for CONST_VECTOR.
3005 * expmed.c (make_tree): Use build_vector_from_val for a CONST
3007 * expr.c (expand_expr_real_2): Check for vector modes instead
3008 of checking for CONST_VECTOR.
3009 * rtl.h (const_vec_p): New function.
3010 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
3011 (unwrap_const_vec_duplicate): Handle them here too.
3013 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3014 David Malcolm <dmalcolm@redhat.com>
3015 Alan Hayward <alan.hayward@arm.com>
3016 David Sherwood <david.sherwood@arm.com>
3018 * rtl.h (vec_duplicate_p): New function.
3019 * selftest-rtl.c (assert_rtx_eq_at): New function.
3020 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
3021 (assert_rtx_eq_at): Declare.
3022 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
3023 * selftest-run-tests.c (selftest::run_tests): Call it.
3024 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
3025 (simplify_unary_operation_1): Recursively handle vector duplicates.
3026 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
3028 (simplify_subreg): Handle subregs of vector duplicates.
3029 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
3030 (selftest::simplify_rtx_c_tests): New functions.
3032 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3033 Alan Hayward <alan.hayward@arm.com>
3034 David Sherwood <david.sherwood@arm.com>
3036 * emit-rtl.h (gen_const_vec_duplicate): Declare.
3037 (gen_vec_duplicate): Likewise.
3038 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
3040 (gen_const_vector): ...here.
3041 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
3042 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
3043 whose elements are all equal.
3044 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
3045 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
3046 (simplify_relational_operation): Likewise.
3047 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
3049 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
3050 (aarch64_expand_vector_init): Likewise.
3051 * config/arm/arm.c (neon_vdup_constant): Likewise.
3052 (neon_expand_vector_init): Likewise.
3053 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
3054 (arm_block_set_unaligned_vect): Likewise.
3055 (arm_block_set_aligned_vect): Likewise.
3056 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
3057 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
3058 (expand_vec_perm_even_odd_pack): Likewise.
3059 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
3060 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
3061 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
3062 gen_const_vec_duplicate.
3063 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
3064 * config/mips/mips.c (mips_gen_const_int_vector): Use
3065 gen_const_vec_duplicate.
3066 (mips_expand_vector_init): Use CONST0_RTX.
3067 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
3068 (define_split): Use gen_const_vec_duplicate.
3069 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
3070 (define_split): Use gen_const_vec_duplicate.
3071 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
3072 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
3073 * config/spu/spu.c (spu_const): Likewise.
3075 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3076 Alan Hayward <alan.hayward@arm.com>
3077 David Sherwood <david.sherwood@arm.com>
3079 * combine.c (can_change_dest_mode): Reject changes in
3080 REGMODE_NATURAL_SIZE.
3082 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
3084 * configure.ac (--enable-libssp): New.
3085 (gcc_cv_libc_provides_ssp): Check for explicit setting before
3086 trying to determine target-specific default. Adjust indentation.
3087 * configure: Regenerated.
3088 * doc/install.texi (Configuration): Expand --disable-libssp
3091 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
3093 config/i386/i386.c (ix86_expand_epilogue): Correct stack
3096 2017-10-31 Martin Jambor <mjambor@suse.cz>
3099 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
3101 2017-10-31 David Malcolm <dmalcolm@redhat.com>
3103 * auto-profile.c (autofdo_source_profile::read): Use
3104 UNKNOWN_LOCATION rather than 0.
3105 * diagnostic-core.h (warning_at_rich_loc): Rename to...
3106 (warning_at): ...this overload.
3107 (warning_at_rich_loc_n): Rename to...
3108 (warning_n): ...this overload.
3109 (error_at_rich_loc): Rename to...
3110 (error_at): ...this overload.
3111 (pedwarn_at_rich_loc): Rename to...
3112 (pedwarn): ...this overload.
3113 (permerror_at_rich_loc): Rename to...
3114 (permerror): ...this overload.
3115 (inform_at_rich_loc): Rename to...
3116 (inform): ...this overload.
3117 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
3118 (diagnostic_n_impl_richloc): Rename to...
3119 (diagnostic_n_impl): ...this rich_location *-based decl.
3120 (inform_at_rich_loc): Rename to...
3121 (inform): ...this, and add an assertion.
3122 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
3123 (warning_at_rich_loc): Rename to...
3124 (warning_at): ...this, and add an assertion.
3125 (warning_at_rich_loc_n): Rename to...
3126 (warning_n): ...this, and add an assertion.
3127 (warning_n): Update location_t-based implementation for removal of
3128 location_t-based diagnostic_n_impl.
3129 (pedwarn_at_rich_loc): Rename to...
3130 (pedwarn): ...this, and add an assertion.
3131 (permerror_at_rich_loc): Rename to...
3132 (permerror): ...this, and add an assertion.
3133 (error_n): Update for removal of location_t-based diagnostic_n_impl.
3134 (error_at_rich_loc): Rename to...
3135 (error_at): ...this, and add an assertion.
3136 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
3137 (driver::do_spec_on_infiles): Likewise.
3138 * substring-locations.c (format_warning_va): Update for renaming
3139 of inform_at_rich_loc.
3141 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
3143 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
3144 _Float<N>X built-in functions so that the variant without the
3145 "__builtin_" prefix is only enabled for the GNU C and Objective C
3146 languages when they are in non-strict ANSI/ISO mode.
3147 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
3148 * target.def (floatn_builtin_p): Add a target hook to control
3149 whether _Float<N> and _Float<N>X built-in functions without the
3150 "__builtin_" prefix are enabled, and return true for C and
3151 Objective C in the default hook. Include langhooks.h in
3153 * targhooks.h (default_floatn_builtin_p): Likewise.
3154 * targhooks.c (default_floatn_builtin_p): Likewise.
3155 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
3156 floatn_builtin_p target hook.
3157 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
3159 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
3160 Eric Botcazou <ebotcazou@adacore.com>
3162 PR rtl-optimization/81803
3163 * lra-constraints.c (curr_insn_transform): Also reload the whole
3164 register for a strict subreg no wider than a word if this is for
3165 a WORD_REGISTER_OPERATIONS target.
3167 2017-10-31 Jason Merrill <jason@redhat.com>
3169 * gdbinit.in: Skip over inlines from timevar.h.
3171 2017-10-31 Martin Liska <mliska@suse.cz>
3173 * doc/gcov.texi: Document new option.
3174 * gcov.c (print_usage): Likewise print it.
3175 (process_args): Support the argument.
3176 (format_count): New function.
3177 (format_gcov): Use the function.
3179 2017-10-31 Martin Liska <mliska@suse.cz>
3181 * gcov.c (struct name_map): do not use typedef.
3182 Define operator== and operator<.
3183 (name_search): Remove.
3184 (name_sort): Remove.
3185 (main): Do not allocate names.
3186 (process_file): Add vertical space.
3187 (generate_results): Use std::find.
3188 (release_structures): Do not release memory.
3189 (find_source): Use std::find.
3191 2017-10-31 Martin Liska <mliska@suse.cz>
3193 * gcov.c (struct line_info): Remove it's typedef.
3194 (line_info::line_info): Add proper ctor.
3195 (line_info::has_block): Do not use a typedef.
3196 (struct source_info): Do not use typedef.
3197 (circuit): Likewise.
3198 (get_cycles_count): Likewise.
3199 (output_intermediate_file): Iterate via vector iterator.
3200 (add_line_counts): Use std::vector methods.
3201 (accumulate_line_counts): Likewise.
3202 (output_lines): Likewise.
3204 2017-10-31 Martin Liska <mliska@suse.cz>
3206 * gcov.c (struct source_info): Remove typedef.
3207 (source_info::source_info): Add proper ctor.
3208 (accumulate_line_counts): Use struct, not it's typedef.
3209 (output_gcov_file): Likewise.
3210 (output_lines): Likewise.
3211 (main): Do not allocate an array.
3212 (output_intermediate_file): Use size of vector container.
3213 (process_file): Resize the vector.
3214 (generate_results): Do not preallocate, use newly added vector
3216 (release_structures): Do not release sources.
3217 (find_source): Use vector methods.
3218 (add_line_counts): Do not use typedef.
3220 2017-10-31 Martin Liska <mliska@suse.cz>
3222 * doc/gcov.texi: Document that.
3223 * gcov.c (add_line_counts): Mark lines with a non-executed
3225 (output_line_beginning): Handle such lines.
3226 (output_lines): Pass new argument.
3227 (output_intermediate_file): Print it in intermediate format.
3229 2017-10-31 Martin Liska <mliska@suse.cz>
3231 * color-macros.h: New file.
3232 * diagnostic-color.c: Factor out color related to macros to
3234 * doc/gcov.texi: Document -k option.
3235 * gcov.c (INCLUDE_STRING): Include string.h.
3236 (print_usage): Add -k option.
3237 (process_args): Parse it.
3238 (pad_count_string): New function.
3239 (output_line_beginning): Likewise.
3240 (DEFAULT_LINE_START): New macro.
3241 (output_lines): Support color output.
3243 2017-10-31 Martin Liska <mliska@suse.cz>
3245 PR gcov-profile/82633
3246 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
3247 their interaction with GCOV infrastructure.
3248 * configure.ac: Add -fkeep-{inline,static}-functions to
3250 * configure: Regenerate.
3252 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
3255 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
3257 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
3260 * config/rs6000/rs6000.md (allocate_stack): Force update interval
3261 into a register if it does not fit into an immediate offset field.
3263 2017-10-31 Olivier Hainque <hainque@adacore.com>
3265 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
3267 2017-10-31 Julia Koval <julia.koval@intel.com>
3269 * config.gcc: Add gfniintrin.h.
3270 * config/i386/gfniintrin.h: New.
3271 * config/i386/i386-builtin-types.def
3272 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
3273 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
3274 __builtin_ia32_vgf2p8affineinvqb_v32qi,
3275 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
3276 __builtin_ia32_vgf2p8affineinvqb_v16qi,
3277 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
3278 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
3279 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
3280 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
3281 V64QI_FTYPE_V64QI_V64QI_INT): New types.
3282 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
3283 * config/i386/immintrin.h: Include gfniintrin.h.
3284 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
3286 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
3288 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
3290 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
3292 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
3293 (arm_ashldi3_1bit): Remove pattern.
3294 (ashrdi3): Remove shift by 1 expansion.
3295 (arm_ashrdi3_1bit): Remove pattern.
3296 (lshrdi3): Remove shift by 1 expansion.
3297 (arm_lshrdi3_1bit): Remove pattern.
3298 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
3299 cost of ashldi3 by 1.
3300 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
3301 (<shift>di3_neon): Likewise.
3303 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
3305 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
3306 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
3307 and (*aarch64_simd_mov<VQ:mode>).
3308 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
3309 pattern alternative.
3310 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
3311 attributes to match pattern alternative.
3313 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
3315 * config.gcc (powerpc*-*-*): Add emmintrin.h.
3316 * config/rs6000/emmintrin.h: New file.
3317 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
3319 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
3321 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
3322 * (movdi_vfp_cortexa8): Remove pattern.
3324 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3326 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
3328 (Specific, alpha*-dec-osf5.1): Remove.
3329 (Specific, mips-sgi-irix5): Remove.
3330 (Specific, mips-sgi-irix6): Remove.
3332 2017-10-30 Jakub Jelinek <jakub@redhat.com>
3335 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
3336 arguments to clear_bit_region_be.
3338 2017-10-30 Jim Wilson <wilson@tuliptree.org>
3340 * gimplify.c: Include memmodel.h.
3342 2017-10-30 Martin Jambor <mjambor@suse.cz>
3344 * omp-grid.c (grid_attempt_target_gridification): Also insert a
3345 condition whether loop should be executed at all.
3347 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
3349 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
3350 gimple folding of vec_madd() intrinsics.
3351 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
3352 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
3353 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
3355 2017-10-30 Richard Biener <rguenther@suse.de>
3357 PR tree-optimization/82762
3359 2017-10-23 Richard Biener <rguenther@suse.de>
3361 PR tree-optimization/82129
3363 2017-08-01 Richard Biener <rguenther@suse.de>
3365 PR tree-optimization/81181
3366 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
3367 (compute_antic): ... end of iteration here.
3369 2017-10-30 Joseph Myers <joseph@codesourcery.com>
3371 * doc/invoke.texi (C Dialect Options): Document -std=c17,
3372 -std=iso9899:2017 and -std=gnu17.
3373 * doc/standards.texi (C Language): Document C17 support.
3374 * doc/cpp.texi (Overview): Mention -std=c17.
3375 (Standard Predefined Macros): Document C11 and C17 values of
3376 __STDC_VERSION__. Do not refer to C99 support as incomplete.
3377 * doc/extend.texi (Inline): Do not list individual options for
3378 standards newer than C99.
3379 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
3381 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
3384 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
3386 * asan.c (asan_finish_file): Align asan globals array by shadow
3389 2017-10-30 Jakub Jelinek <jakub@redhat.com>
3392 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
3393 (struct store_immediate_info): Add bitregion_start and bitregion_end
3395 (store_immediate_info::store_immediate_info): Add brs and bre
3396 arguments and initialize bitregion_{start,end} from those.
3397 (struct merged_store_group): Add bitregion_start, bitregion_end,
3398 align_base and mask fields. Drop unnecessary struct keyword from
3399 struct store_immediate_info. Add do_merge method.
3400 (clear_bit_region_be): Use memset instead of loop storing zeros.
3401 (merged_store_group::do_merge): New method.
3402 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
3403 stores as long as the surrounding bitregions have no gaps.
3404 (merged_store_group::merge_overlapping): Use do_merge.
3405 (merged_store_group::apply_stores): Test that bitregion_{start,end}
3406 is byte aligned, rather than requiring that start and width are
3407 byte aligned. Drop unnecessary struct keyword from
3408 struct store_immediate_info. Allocate and populate also mask array.
3409 Make start of the arrays relative to bitregion_start rather than
3410 start and size them according to bitregion_{end,start} difference.
3411 (struct imm_store_chain_info): Drop unnecessary struct keyword from
3412 struct store_immediate_info.
3413 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
3414 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
3415 struct keyword from struct store_immediate_info.
3416 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
3417 between stores as long as the surrounding bitregions have no gaps.
3419 (struct split_store): Add orig non-static data member.
3420 (split_store::split_store): Initialize orig to false.
3421 (find_constituent_stmts): Return store_immediate_info *, non-NULL
3422 if there is exactly a single original stmt. Change stmts argument
3423 to pointer from reference, if NULL, don't push anything to it. Add
3424 first argument, use it to optimize skipping over orig stmts that
3425 are known to be before bitpos already. Simplify.
3426 (split_group): Return unsigned int count how many stores are or
3427 would be needed rather than a bool. Add allow_unaligned argument.
3428 Change split_stores argument from reference to pointer, if NULL,
3429 only do a dry run computing how many stores would be produced.
3430 Rewritten algorithm to use both alignment and misalign if
3431 !allow_unaligned and handle bitfield stores with gaps.
3432 (imm_store_chain_info::output_merged_store): Set start_byte_pos
3433 from bitregion_start instead of start. Compute allow_unaligned
3434 here, if true, do 2 split_group dry runs to compute which one
3435 produces fewer stores and prefer aligned if equal. Punt if
3436 new count is bigger or equal than original before emitting any
3437 statements, rather than during that. Remove no longer needed
3438 new_ssa_names tracking. Replace num_stmts with
3439 split_stores.length (). Use 32-bit stack allocated entries
3440 in split_stores auto_vec. Try to reuse original store lhs/rhs1
3441 if possible. Handle bitfields with gaps.
3442 (pass_store_merging::execute): Ignore bitsize == 0 stores.
3443 Compute bitregion_{start,end} for the stores and construct
3444 store_immediate_info with that. Formatting fixes.
3446 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
3449 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
3450 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
3452 2017-10-29 Jim Wilson <wilson@tuliptree.org>
3454 * gimplify.c: Include tm_p.h.
3456 * common.opt (gcoff): Re-add as ignored option.
3457 (gcoff1, gcoff2, gcoff3): Likewise.
3459 * Makefile.in (OBJS): Delete sdbout.o.
3460 (GTFILES): Delete $(srcdir)/sdbout.c.
3461 * debug.h: Delete sdb_debug_hooks.
3462 * final.c: Delete sdbout.h include.
3463 (final_scan_insn): Delete SDB_DEBUG check.
3464 (rest_of_clean_state): Likewise.
3465 * output.h: Delete sdb_begin_function_line.
3468 * toplev.c: Delete sdbout.h include.
3469 (process_options): Delete SDB_DEBUG check.
3470 * tree-core.h (tree_type_common): Delete pointer field of
3472 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
3473 TYPE_SYMTAB_POINTER.
3474 * tree.h (TYPE_SYMTAB_POINTER): Delete.
3475 (TYPE_SYMTAB_IS_POINTER): Delete.
3476 (TYPE_SYMTAB_IS_DIE): Renumber.
3477 * xcoffout.c: Refer to former sdbout.c file.
3478 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
3480 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
3481 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
3482 (Debugging Options): Delete -gcoff.
3483 (-gstabs): Delete SDB reference.
3485 (-gcoff@var{level}): Delete.
3486 * doc/passes.texi (Debugging information output): Delete SDB and
3487 sdbout.c references.
3488 * doc/tm.texi: Regenerate.
3489 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
3490 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
3492 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
3493 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
3494 (SDB_DEBUGGING_INFO): Delete.
3495 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
3496 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
3497 * target.def (output_source_filename): Delete COFF reference.
3499 * common.opt (gcoff): Delete.
3500 (gxcoff+): Update Negative chain.
3501 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
3503 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
3504 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
3505 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
3507 (expand_function_start): Change sdb reference to past tense.
3508 (expand_function_end): Change sdb reference to past tense.
3509 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
3510 * opts.c (debug_type_names): Delete coff entry.
3511 (common_handle_option): Delete OPT_gcoff case.
3512 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
3514 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
3515 * config/cris/cris.h: Delete SDB reference in comment.
3516 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
3517 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
3518 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
3519 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
3521 (ix86_expand_prologue): Likewise.
3522 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
3523 * config/ia64/ia64.h: Likewise.
3524 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
3525 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
3527 * config/mmix/mmix.h: Likewise.
3528 * config/nds32/nds32.c: Likewise.
3529 * config/stormy/storym16.h: Likewise.
3530 * config/visium/visium.h: Likewise.
3531 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
3533 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
3535 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
3536 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
3537 FRAME_POINTER_REGNUM point at high end of local var area.
3539 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
3541 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
3542 Move comment around. Do not reset best_edge for a copiable
3543 destination if the copy would cause a partition change.
3544 (better_edge_p): Remove redundant check.
3546 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
3548 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
3550 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
3552 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
3553 for math functions that have _Float<N> and _Float<N>X variants.
3554 (mathfn_built_in_2): Add support for math functions that have
3555 _Float<N> and _Float<N>X variants.
3556 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
3557 (expand_builtin_mathfn_ternary): Add support for fma with
3558 _Float<N> and _Float<N>X variants.
3559 (expand_builtin): Likewise.
3560 (fold_builtin_3): Likewise.
3561 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
3562 create math function _Float<N> and _Float<N>X variants as external
3564 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
3565 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
3566 the __builtin_ prefix and if not strict ansi, without the prefix.
3567 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
3568 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
3569 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
3570 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
3571 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
3572 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
3573 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
3574 function signatures for fma _Float<N> and _Float<N>X variants.
3575 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
3576 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
3577 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
3578 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
3579 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
3580 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
3581 * gencfn-macros.c (print_case_cfn): Add support for math functions
3582 that have _Float<N> and _Float<N>X variants.
3583 (print_define_operator_list): Likewise.
3584 (fltfn_suffixes): Likewise.
3586 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
3587 for math functions that have _Float<N> and _Float<N>X variants.
3588 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
3589 and _Float<N>X variants.
3590 (COPYSIGN): Likewise.
3593 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
3594 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
3596 (integer_valued_read_call_p): Likewise.
3597 * fold-const-call.c (fold_const_call_ss): Likewise.
3598 (fold_const_call_sss): Add support for copysign, fmin, and fmax
3599 _Float<N> and _Float<N>X variants.
3600 (fold_const_call_ssss): Add support for fma _Float<N> and
3601 _Float<N>X variants.
3602 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
3603 support for copysign and fma _Float<N> and _Float<N>X variants.
3604 (backprop::process_builtin_call_use): Likewise.
3605 * tree-call-cdce.c (can_test_argument_range); Add support for
3606 sqrt _Float<N> and _Float<N>X variants.
3607 (edom_only_function): Likewise.
3608 (get_no_error_domain): Likewise.
3609 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
3610 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
3611 copysign _Float<N> and _Float<N>X variants.
3612 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
3613 handled by machine independent code.
3614 (FMAF128): Likewise.
3615 * doc/cpp.texi (Common Predefined Macros): Document defining
3616 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
3617 fma _Float<N> and _Float<N>X variants.
3619 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
3622 * config/i386/i386-modes.def (CCFPU): Remove definition.
3623 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
3624 (ix86_cc_modes_compatible): Ditto.
3625 (ix86_expand_carry_flag_compare): Ditto.
3626 (ix86_expand_int_movcc): Ditto.
3627 (ix86_expand_int_addcc): Ditto.
3628 (ix86_reverse_condition): Ditto.
3629 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
3630 Return true/false for unordered/ordered fp comparisons.
3631 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
3632 (ix86_prepare_fp_compare_args): Update for rename.
3633 (ix86_expand_fp_compare): Update for rename. Generate unordered
3634 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
3635 (ix86_expand_sse_compare_and_jump): Ditto.
3636 * config/i386/predicates.md (fcmov_comparison_operator):
3637 Remove CCFPU mode handling.
3638 (ix86_comparison_operator): Ditto.
3639 (ix86_carry_flag_operator): Ditto.
3640 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
3641 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
3642 (*cmpu<mode>_cc_i387): Ditto.
3643 (FPCMP): Remove mode iterator.
3644 (unord): Remove mode attribute.
3645 (unord_subst): New define_subst transformation
3646 (unord): New define_subst attribute.
3648 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
3649 (*cmpi<unord>xf_i387): Ditto.
3650 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
3651 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
3652 using unord_subst transformation.
3653 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
3654 (round_saeonly): Also handle CCFP mode.
3655 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
3656 Remove UNSPEC_SAHF unspec handling.
3658 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
3660 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
3662 2017-10-27 Jeff Law <law@redhat.com>
3664 * gimple-ssa-sprintf.c: Include domwalk.h.
3665 (class sprintf_dom_walker): New class, derived from dom_walker.
3666 (sprintf_dom_walker::before_dom_children): New function.
3667 (struct call_info): Moved into sprintf_dom_walker class
3668 (compute_formath_length, handle_gimple_call): Likewise.
3669 (sprintf_length::execute): Call the dominator walker rather
3670 than walking the statements.
3672 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
3673 gimple statement locations.
3674 (check_array_bounds): Corresponding changes. Get the statement's
3675 location directly from wi->stmt.
3677 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
3680 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
3682 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
3684 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
3685 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
3687 2017-10-27 Jakub Jelinek <jakub@redhat.com>
3690 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
3691 * config/i386/i386.c (maybe_get_pool_constant): Removed.
3692 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
3693 maybe_get_pool_constant.
3694 * config/i386/predicates.md (zero_extended_scalar_load_operand):
3697 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3699 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
3700 2.26 caveat. Update gas and gld versions.
3701 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
3704 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3706 * cgraph.h (set_malloc_flag): Declare.
3707 * cgraph.c (set_malloc_flag_1): New function.
3708 (set_malloc_flag): Likewise.
3709 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
3710 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
3712 (read_ipa_call_summary): Add support for reading is_return_callee.
3713 (write_ipa_call_summary): Stream is_return_callee.
3714 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
3715 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
3716 ipa-prop.h, ipa-fnsummary.h.
3717 (pure_const_names): Change to static.
3718 (malloc_state_e): Define.
3719 (malloc_state_names): Define.
3720 (funct_state_d): Add field malloc_state.
3721 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
3722 (check_retval_uses): New function.
3723 (malloc_candidate_p): Likewise.
3724 (analyze_function): Add support for malloc attribute.
3725 (pure_const_write_summary): Stream malloc_state.
3726 (pure_const_read_summary): Add support for reading malloc_state.
3727 (dump_malloc_lattice): New function.
3728 (propagate_malloc): New function.
3729 (warn_function_malloc): New function.
3730 (ipa_pure_const::execute): Call propagate_malloc and
3731 ipa_free_fn_summary.
3732 (pass_local_pure_const::execute): Add support for malloc attribute.
3733 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
3734 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
3736 2017-10-27 Martin Liska <mliska@suse.cz>
3738 PR gcov-profile/82457
3739 * doc/invoke.texi: Document that one needs a non-strict ISO mode
3740 for fork-like functions to be properly instrumented.
3742 2017-10-27 Richard Biener <rguenther@suse.de>
3745 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
3746 info when we redirected EH.
3748 2017-10-26 Michael Collison <michael.collison@arm.com>
3750 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
3752 (<optab>_trunchf<GPI:mode>2: New pattern.
3753 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
3754 * config/aarch64/iterators.md (wv): New mode attribute.
3755 (vf, VF): New mode attributes.
3756 (vgp, VGP): New mode attributes.
3757 (s): Update attribute with SImode and DImode prefixes.
3759 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
3761 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
3762 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
3763 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
3764 (nios2_option_overide): Initialize it. Don't allow R0-relative
3765 addressing with PIC.
3766 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
3767 (nios2_symbolic_constant_p): Likewise.
3768 (nios2_legitimate_address_p): Likewise.
3769 (nios2_r0rel_section_name_p): New.
3770 (nios2_symbol_ref_in_r0rel_data_p): New.
3771 (nios2_emit_move_sequence): Handle r0rel_constant_p.
3772 (r0rel_constant_p): New.
3773 (nios2_print_operand_address): Handle r0rel_constant_p.
3774 (nios2_cdx_narrow_form_p): Likewise.
3775 * config/nios2/nios2.opt (mr0rel-sec=): New option.
3776 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
3777 (Nios II Options): Document -mr0rel-sec.
3779 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
3781 * config/nios2/nios2.c: Include xregex.h.
3782 (nios2_gprel_sec_regex): New.
3783 (nios2_option_overide): Initialize it. Don't allow GP-relative
3784 addressing with PIC.
3785 (nios2_small_section_name_p): Check for regex match.
3786 * config/nios2/nios2.opt (mgprel-sec=): New option.
3787 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
3788 (Nios II Options): Document -mgprel-sec.
3790 2017-10-26 Jim Wilson <wilson@tuliptree.org>
3792 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
3794 2017-10-26 Tom de Vries <tom@codesourcery.com>
3796 PR tree-optimization/82707
3797 * gimple.c (gimple_copy): Fix unsharing of
3798 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
3800 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
3802 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
3803 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
3804 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
3805 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
3806 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
3807 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
3808 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
3809 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
3810 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
3811 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
3812 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
3813 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
3814 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
3815 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
3816 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
3817 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
3818 _mm512_mask_cmpunord_ps_mask): New intrinsics.
3820 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
3822 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
3824 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
3825 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
3826 warning to rs6000.c. Remove the Undocumented flag, since it has
3828 (-mabi=ibmlongdouble): Likewise.
3829 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
3830 already set, set the default format for long double.
3831 (rs6000_debug_reg_global): Print whether long double is IBM or
3833 (rs6000_option_override_internal): Rework setting long double
3834 format. Only warn if the user is changing the long double default
3835 and they did not use -Wno-psabi.
3836 * doc/invoke.texi (PowerPC options): Update the documentation for
3837 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
3839 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3840 Alan Hayward <alan.hayward@arm.com>
3841 David Sherwood <david.sherwood@arm.com>
3843 * rtl.h (wider_subreg_mode): New function.
3844 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
3845 rather than an unsigned int *.
3846 * ira-color.c (regno_max_ref_width): Replace with...
3847 (regno_max_ref_mode): ...this new variable.
3848 (coalesced_pseudo_reg_slot_compare): Update accordingly.
3849 Use wider_subreg_mode.
3850 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
3851 rather than an unsigned int *.
3852 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
3853 (process_alt_operands): Likewise.
3854 (invariant_p): Likewise.
3855 * lra-spills.c (assign_mem_slot): Likewise.
3856 (add_pseudo_to_slot): Likewise.
3857 * lra.c (collect_non_operand_hard_regs): Likewise.
3858 (add_regs_to_insn_regno_info): Likewise.
3859 * reload1.c (regno_max_ref_width): Replace with...
3860 (regno_max_ref_mode): ...this new variable.
3861 (reload): Update accordingly. Update call to
3862 ira_sort_regnos_for_alter_reg.
3863 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
3864 (init_eliminable_invariants): Update to use regno_max_ref_mode.
3865 (scan_paradoxical_subregs): Likewise.
3867 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
3869 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
3870 (aarch64_frame): Add emit_frame_chain boolean.
3871 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
3872 Move eh_return case to aarch64_layout_frame.
3873 (aarch64_layout_frame): Initialize emit_frame_chain.
3874 (aarch64_expand_prologue): Use emit_frame_chain.
3876 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
3878 * config/aarch64/aarch64.c (aarch64_layout_frame):
3879 Ensure LR is always stored at the bottom of the callee-saves.
3880 Remove rarely used frame layout which saves callee-saves at top of
3881 frame, so the store of LR can be used as a valid probe in all cases.
3883 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
3885 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
3886 Improve unaligned TImode/TFmode base/offset split.
3888 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3889 Alan Hayward <alan.hayward@arm.com>
3890 David Sherwood <david.sherwood@arm.com>
3892 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
3893 * combine.c (find_single_use_1): Likewise.
3894 (expand_field_assignment): Likewise.
3895 (move_deaths): Likewise.
3896 * lra-constraints.c (simplify_operand_subreg): Likewise.
3897 (curr_insn_transform): Likewise.
3898 * lra.c (collect_non_operand_hard_regs): Likewise.
3899 (add_regs_to_insn_regno_info): Likewise.
3900 * rtlanal.c (reg_referenced_p): Likewise.
3901 (covers_regno_no_parallel_p): Likewise.
3903 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3905 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
3906 Don't print any bits outside the precision of the value.
3907 * wide-int.cc (test_printing): Add some new tests.
3909 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3911 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
3912 supports -xbrace_comment option.
3913 * configure: Regenerate.
3914 * config.in: Regenerate.
3915 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
3916 (ASM_CPU_SPEC): Use it.
3918 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3920 * target.def (static_rtx_alignment): New hook.
3921 * targhooks.h (default_static_rtx_alignment): Declare.
3922 * targhooks.c (default_static_rtx_alignment): New function.
3923 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
3924 * doc/tm.texi: Regenerate.
3925 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
3926 instead of targetm.constant_alignment. Remove call to
3928 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3929 (cris_preferred_mininum_alignment): New function, split out from...
3930 (cris_constant_alignment): ...here.
3931 (cris_static_rtx_alignment): New function.
3932 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
3934 (ix86_constant_alignment): ...here.
3935 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3936 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3937 (mmix_static_rtx_alignment): New function.
3938 * config/spu/spu.c (spu_static_rtx_alignment): New function.
3939 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3941 2017-10-26 Tamar Christina <tamar.christina@arm.com>
3944 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
3945 Add flag_trapping_math and flag_fp_int_builtin_inexact.
3947 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
3949 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
3950 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
3951 mark as a sign-extending load.
3952 (local_pic_load_u): Define.
3954 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
3957 * fold-const.c (operand_equal_for_comparison_p): Also return true
3958 if ARG0 is a simple variant of ARG1 with narrower precision.
3959 (fold_ternary_loc): Always pass unstripped operands to the predicate.
3961 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
3963 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
3965 * i386.h (processor_costs): Add gather_static, gather_per_elt,
3966 scatter_static, scatter_per_elt.
3967 * x86-tune-costs.h: Add new cost entries.
3969 2017-10-25 Richard Biener <rguenther@suse.de>
3971 * tree-ssa-sccvn.h (vn_eliminate): Declare.
3972 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
3973 class pass_fre): Move to ...
3974 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
3975 class pass_fre): ... here and adjust for statistics.
3977 2017-10-25 Jakub Jelinek <jakub@redhat.com>
3980 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
3981 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
3982 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
3983 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
3986 2017-10-25 Richard Biener <rguenther@suse.de>
3988 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
3989 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
3990 eliminate_push_avail, eliminate_insert): Move inside...
3991 (class eliminate_dom_walker): ... this class in preparation
3993 (fini_eliminate): Remove by merging with ...
3994 (eliminate): ... this function. Adjust for class changes.
3995 (pass_pre::execute): Remove fini_eliminate call.
3996 (pass_fre::execute): Likewise.
3998 2017-10-24 Jakub Jelinek <jakub@redhat.com>
4001 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
4002 (VPERMI2, VPERMI2I): New mode iterators.
4003 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
4004 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
4006 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
4007 mode iterator. Remove 3 old define_insn patterns.
4008 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
4009 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
4010 VPERMI2 mode iterator, remove the other two expanders.
4011 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
4012 to use VPERMI2 mode iterator, add another alternative for vpermi2*
4013 instructions, remove the other two patterns.
4014 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
4015 mode iterator, remove the other two patterns.
4016 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
4017 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
4018 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
4019 and adjust argument order accordingly.
4020 (ix86_expand_vec_perm): Adjust caller.
4021 (expand_vec_perm_1): Likewise.
4022 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
4023 (expand_vec_perm_vpermt2_vpshub2): ... this.
4024 (ix86_expand_vec_perm_const_1): Adjust caller.
4025 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
4028 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
4029 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
4030 (vec_shl_<mode>): Remove unused expander.
4031 (avx512bw_<shift_insn><mode>3): New define_insn.
4032 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
4033 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
4035 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
4038 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
4041 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
4043 PR rtl-optimization/82396
4044 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
4045 (autopref_multipass_init): Simplify initialization.
4046 (autopref_rank_data): Simplify sort order.
4047 * gcc/sched-int.h (autopref_multipass_data_): Remove
4048 multi_mem_insn_p, min_offset and max_offset.
4050 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
4053 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
4054 Check special value of flag_omit_frame_pointer.
4055 (aarch64_can_eliminate): Likewise.
4056 (aarch64_override_options_after_change_1): Simplify handling of
4057 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
4059 2017-10-24 Richard Biener <rguenther@suse.de>
4061 PR tree-optimization/82697
4062 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
4063 zero for conditional load and unconditional store.
4065 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
4067 * doc/install.texi: Document bootstrap-cet.
4069 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
4072 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
4073 ENDBR instruction at function entrance if function is only
4076 2017-10-24 Jakub Jelinek <jakub@redhat.com>
4079 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
4080 patterns to better describe from which operation the CF is computed.
4081 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
4082 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
4083 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
4084 is 0, use _0 suffixed expanders instead of emitting a comparison
4087 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
4089 * config/i386/i386.md(*movsf_internal, *movdf_internal):
4090 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
4092 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
4095 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
4096 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
4097 * loop-iv.c (iv_get_reaching_def): Likewise.
4098 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
4099 variable is promoted and the partition contains undefined values.
4101 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4103 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
4105 (nios2_address_cost): Define.
4106 (nios2_legitimize_address): Recognize (exp + constant) directly.
4107 (TARGET_ADDRESS_COST): Define.
4109 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4111 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
4112 (nios2_symbolic_memory_operand_p): Declare.
4113 (nios2_split_large_constant): Declare.
4114 (nios2_split_symbolic_memory_operand): Declare.
4115 * config/nios2/nios2.c: Adjust includes.
4116 (nios2_symbolic_constant_allowed): New.
4117 (nios2_symbolic_constant_p): New.
4118 (nios2_plus_symbolic_constant_p): New.
4119 (nios2_valid_addr_expr_p): Recognize addresses involving
4121 (nios2_legitimate_address_p): Likewise, also LO_SUM.
4122 (nios2_symbolic_memory_operand_p): New.
4123 (nios2_large_constant_p): New.
4124 (nios2_split_large_constant): New.
4125 (nios2_split_plus_large_constant): New.
4126 (nios2_split_symbolic_memory_operand): New.
4127 (nios2_legitimize_address): Code refactoring. Handle addresses
4128 involving symbolic constants.
4129 (nios2_emit_move_sequence): Likewise.
4130 (nios2_print_operand): Improve error output.
4131 (nios2_print_operand_address): Handle LO_SUM.
4132 (nios2_cdx_narrow_form_p): Likewise.
4133 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
4134 operands involving symbolic constants.
4135 (movhi_internal, movsi_internal): Likewise.
4136 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
4137 (extendhisi2, extendqi<mode>2): Likewise.
4139 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4141 * tree-pass.h (PROP_rtl_split_insns): Define.
4142 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
4144 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4146 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
4148 2017-10-23 Jakub Jelinek <jakub@redhat.com>
4151 * target.def (const_not_ok_for_debug_p): Default to
4152 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
4153 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
4154 * targhooks.c (default_const_not_ok_for_debug_p): New function.
4155 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
4156 which targetm.const_not_ok_for_debug_p returned true.
4157 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
4159 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
4161 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
4162 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
4163 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
4165 (ix86_const_not_ok_for_debug_p): New function.
4166 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
4167 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
4169 2017-10-23 David Malcolm <dmalcolm@redhat.com>
4172 * system.h: Conditionally include "unique-ptr.h" if
4173 INCLUDE_UNIQUE_PTR is defined.
4174 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
4175 of defining INCLUDE_UNIQUE_PTR before including "system.h".
4177 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
4179 * config/rl78/rl78.md: New define_expand "subdi3".
4181 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
4184 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
4185 DF_REF_INSN if DF_REF_INSN_INFO is false.
4187 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
4189 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
4190 xmm_move instead of sse_move.
4191 (sse_store_index): New function.
4192 (ix86_register_move_cost): Be more sensible about mismatch stall;
4193 model AVX moves correctly; make difference between sse->integer and
4195 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
4196 moves; make difference between SSE and AVX.
4197 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
4198 and zmm_move. Increase size of sse load and store tables;
4199 add unaligned load and store tables; add ssemmx_to_integer.
4200 * x86-tune-costs.h: Update all entries according to real
4201 move latencies from Agner Fog's manual and chip documentation.
4203 2017-10-23 Jakub Jelinek <jakub@redhat.com>
4206 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
4207 * config/i386/constraints.md (Wf): New constraint.
4208 * config/i386/i386.md (UNSPEC_SBB): New unspec.
4209 (cmp<dwi>_doubleword): Removed.
4210 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
4211 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
4212 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
4213 expand with cmp<dwi>_doubleword. For LTU and GEU use
4214 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
4216 * common.opt (gcolumn-info): Enable by default.
4217 * doc/invoke.texi (gcolumn-info): Document new default.
4219 2017-10-23 Richard Biener <rguenther@suse.de>
4221 PR tree-optimization/82672
4222 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
4223 Fold the stmt if we propagated into it.
4225 2017-10-23 Richard Biener <rguenther@suse.de>
4227 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
4228 (bitmap_remove_expr_from_set): ... this. All callers call this
4229 for non-constant values.
4230 (bitmap_set_subtract): Rename to...
4231 (bitmap_set_subtract_expressions): ... this. Adjust and
4233 (bitmap_set_contains_value): Remove superfluous check.
4234 (bitmap_set_replace_value): Inline into single caller ...
4235 (bitmap_value_replace_in_set): ... here and simplify.
4236 (dependent_clean): Merge into ...
4237 (clean): ... this using an overload. Adjust.
4238 (prune_clobbered_mems): Adjust.
4239 (compute_antic_aux): Likewise.
4240 (compute_partial_antic_aux): Likewise.
4242 2017-10-23 Richard Biener <rguenther@suse.de>
4244 PR tree-optimization/82129
4246 2017-08-01 Richard Biener <rguenther@suse.de>
4248 PR tree-optimization/81181
4249 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
4250 (compute_antic): ... end of iteration here.
4252 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4254 * target.def (starting_frame_offset): New hook.
4255 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
4256 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
4257 * doc/tm.texi.in: Regenerate.
4258 * hooks.h (hook_hwi_void_0): Declare.
4259 * hooks.c (hook_hwi_void_0): New function.
4260 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
4261 STARTING_FRAME_OFFSET.
4262 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
4263 * reload1.c (reload): Likewise.
4264 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
4265 instead of STARTING_FRAME_OFFSET.
4266 * function.c (try_fit_stack_local): Likewise.
4267 (assign_stack_local_1): Likewise
4268 (instantiate_virtual_regs): Likewise.
4269 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
4270 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
4271 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
4272 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
4273 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
4274 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
4275 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
4276 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
4277 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
4278 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
4279 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
4280 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
4281 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
4282 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
4283 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
4284 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
4285 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
4286 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
4287 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
4288 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
4289 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
4290 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
4291 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
4292 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
4293 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
4294 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
4295 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
4296 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
4297 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
4298 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
4299 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
4300 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
4301 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
4302 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
4303 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
4304 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
4305 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
4306 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
4307 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
4308 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
4309 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
4310 * config/avr/avr.c (avr_starting_frame_offset): Make static and
4311 return a HOST_WIDE_INT.
4312 (avr_builtin_setjmp_frame_value): Use it instead of
4313 STARTING_FRAME_OFFSET.
4314 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4315 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
4316 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
4318 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4319 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
4320 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
4321 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4322 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
4323 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
4324 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4325 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
4326 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
4327 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4328 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
4329 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
4331 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4332 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
4333 * config/mips/mips.c (mips_compute_frame_info): Refer to
4334 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
4335 (mips_starting_frame_offset): New function.
4336 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4337 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
4338 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
4339 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
4340 and return a HOST_WIDE_INT.
4341 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4342 (mmix_initial_elimination_offset): Refer to
4343 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
4344 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
4345 * config/pa/pa.c (pa_starting_frame_offset): New function.
4346 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
4347 (pa_expand_prologue): Likewise.
4348 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4349 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
4350 !FRAME_GROWS_DOWNWARD handling to...
4351 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4352 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
4353 !FRAME_GROWS_DOWNWARD handling to...
4354 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4355 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
4356 !FRAME_GROWS_DOWNWARD handling to...
4357 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4358 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
4360 (rs6000_starting_frame_offset): New function.
4361 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
4362 !FRAME_GROWS_DOWNWARD handling to...
4363 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4364 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
4365 !FRAME_GROWS_DOWNWARD handling to...
4366 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4367 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
4368 !FRAME_GROWS_DOWNWARD handling to...
4369 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4370 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
4371 (rs6000_starting_frame_offset): New function.
4372 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
4373 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
4374 * config/vax/vax.c (vax_starting_frame_offset): New function.
4375 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
4376 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4377 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
4378 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
4379 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4380 * system.h (STARTING_FRAME_OFFSET): Poison.
4382 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4384 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
4385 SCALAR_TYPE_MODE instead of TYPE_MODE.
4387 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4388 Alan Hayward <alan.hayward@arm.com>
4389 David Sherwood <david.sherwood@arm.com>
4391 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
4393 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4394 Alan Hayward <alan.hayward@arm.com>
4395 David Sherwood <david.sherwood@arm.com>
4397 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
4399 2017-10-23 Richard Biener <rguenther@suse.de>
4401 PR tree-optimization/82129
4402 * tree-ssa-pre.c (bitmap_set_and): Remove.
4403 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
4404 canonicalizing expressions in the set to those with lowest
4405 ID rather than taking that from the first edge.
4407 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4409 * combine.c (rtx_equal_for_field_assignment_p): Use
4410 byte_lowpart_offset.
4412 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4413 Alan Hayward <alan.hayward@arm.com>
4414 David Sherwood <david.sherwood@arm.com>
4416 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
4417 to a SUBREG_PROMOTED_VAR.
4419 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4420 Alan Hayward <alan.hayward@arm.com>
4421 David Sherwood <david.sherwood@arm.com>
4423 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
4424 (expand_debug_source_expr): Likewise.
4425 * combine.c (combine_simplify_rtx): Likewise.
4426 * cse.c (fold_rtx): Likewise.
4427 * optabs.c (expand_float): Likewise.
4428 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4429 (simplify_binary_operation_1): Likewise.
4431 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4432 Alan Hayward <alan.hayward@arm.com>
4433 David Sherwood <david.sherwood@arm.com>
4435 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
4436 (record_promoted_value): Likewise.
4437 * expr.c (expand_expr_real_2): Likewise.
4438 * ree.c (update_reg_equal_equiv_notes): Likewise.
4439 (combine_set_extension): Likewise.
4440 * rtlanal.c (low_bitmask_len): Likewise.
4441 * simplify-rtx.c (neg_const_int): Likewise.
4442 (simplify_binary_operation_1): Likewise.
4444 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4445 Alan Hayward <alan.hayward@arm.com>
4446 David Sherwood <david.sherwood@arm.com>
4448 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
4449 * regcprop.c (maybe_mode_change): Likewise.
4450 * reload1.c (alter_reg): Likewise.
4452 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4454 * inchash.h (inchash::hash::add_wide_int): New function.
4455 * lto-streamer-out.c (hash_tree): Use it.
4457 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4459 * inchash.h (inchash::hash::add_wide_int): Rename to...
4460 (inchash::hash::add_hwi): ...this.
4461 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
4462 (polymorphic_call_target_hasher::hash): Likewise.
4463 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
4464 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
4465 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
4466 * lto-streamer-out.c (hash_tree): Likewise.
4467 * optc-save-gen.awk: Likewise.
4468 * tree.c (add_expr): Likewise.
4470 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
4473 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
4474 for ordered inequality comparisons even with TARGET_IEEE_FP.
4476 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
4479 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
4480 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
4481 Expand with cmp<dwi>_doubleword.
4483 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4485 * extend.texi: Add x86 specific to 'nocf_check' attribute.
4486 List CET intrinsics.
4487 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
4488 specific to -fcf-protection option.
4490 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4492 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
4493 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
4494 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
4495 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
4496 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
4497 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
4498 (extra_objs): Add cet.o for Linux/x86 targets.
4499 (tmake_file): Add i386/t-cet for Linux/x86 targets.
4500 * config/i386/cet.c: New file.
4501 * config/i386/cetintrin.h: Likewise.
4502 * config/i386/t-cet: Likewise.
4503 * config/i386/cpuid.h (bit_SHSTK): New.
4504 (bit_IBT): Likewise.
4505 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
4506 pass IBT and SHSTK bits.
4507 * config/i386/i386-builtin-types.def
4508 (VOID_FTYPE_UNSIGNED_PVOID): New.
4509 (VOID_FTYPE_UINT64_PVOID): Likewise.
4510 * config/i386/i386-builtin.def: Add CET intrinsics.
4511 * config/i386/i386-c.c (ix86_target_macros_internal): Add
4512 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
4513 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
4514 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
4516 * config/i386/i386.c (rest_of_insert_endbranch): New.
4517 (pass_data_insert_endbranch): Likewise.
4518 (pass_insert_endbranch): Likewise.
4519 (make_pass_insert_endbranch): Likewise.
4520 (ix86_notrack_prefixed_insn_p): Likewise.
4521 (ix86_target_string): Add -mibt, -mshstk flags.
4522 (ix86_option_override_internal): Add flag_cf_protection
4524 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
4525 (ix86_print_operand): Add 'notrack' prefix output.
4526 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
4527 (ix86_expand_builtin): Expand CET intrinsics.
4528 (x86_output_mi_thunk): Add 'endbranch' instruction.
4529 * config/i386/i386.h (TARGET_IBT): New.
4530 (TARGET_IBT_P): Likewise.
4531 (TARGET_SHSTK): Likewise.
4532 (TARGET_SHSTK_P): Likewise.
4533 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
4534 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
4535 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
4536 (builtin_setjmp_setup): New pattern.
4537 (builtin_longjmp): Likewise.
4538 (rdssp<mode>): Likewise.
4539 (incssp<mode>): Likewise.
4540 (saveprevssp): Likewise.
4541 (rstorssp): Likewise.
4542 (wrss<mode>): Likewise.
4543 (wruss<mode>): Likewise.
4544 (setssbsy): Likewise.
4545 (clrssbsy): Likewise.
4546 (nop_endbr): Likewise.
4547 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
4549 * config/i386/immintrin.h: Include <cetintrin.h>.
4550 * config/i386/linux-common.h
4551 (file_end_indicate_exec_stack_and_cet): New prototype.
4552 (TARGET_ASM_FILE_END): New.
4554 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
4556 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
4557 latencies instead of having separate table; make difference between
4558 integer and float costs.
4559 * i386.h (processor_costs): Remove scalar_stmt_cost,
4560 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
4561 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
4563 * x86-tune-costs.h: Remove entries which has been removed in
4564 procesor_costs from all tables; make cond_taken_branch_cost
4565 and cond_not_taken_branch_cost COST_N_INSNS based.
4567 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
4569 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
4571 2017-10-20 Jakub Jelinek <jakub@redhat.com>
4573 * config/i386/i386.md (isa): Remove fma_avx512f.
4574 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
4575 <avx512>_fmadd_<mode>_mask3<round_name>,
4576 <avx512>_fmsub_<mode>_mask<round_name>,
4577 <avx512>_fmsub_<mode>_mask3<round_name>,
4578 <avx512>_fnmadd_<mode>_mask<round_name>,
4579 <avx512>_fnmadd_<mode>_mask3<round_name>,
4580 <avx512>_fnmsub_<mode>_mask<round_name>,
4581 <avx512>_fnmsub_<mode>_mask3<round_name>,
4582 <avx512>_fmaddsub_<mode>_mask<round_name>,
4583 <avx512>_fmaddsub_<mode>_mask3<round_name>,
4584 <avx512>_fmsubadd_<mode>_mask<round_name>,
4585 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
4586 (*vec_widen_umult_even_v16si<mask_name>,
4587 *vec_widen_smult_even_v16si<mask_name>): Likewise.
4588 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
4590 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4592 * extend.texi: Add 'nocf_check' documentation.
4593 * gimple.texi: Add second parameter to
4594 gimple_build_call_from_tree.
4595 * invoke.texi: Add -fcf-protection documentation.
4596 * rtl.texi: Add REG_CALL_NOTRACK documenation.
4598 2017-10-20 Richard Biener <rguenther@suse.de>
4600 PR tree-optimization/82473
4601 * tree-vect-loop.c (vectorizable_reduction): Properly get at
4602 the largest input type.
4604 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4606 * c-attribs.c (handle_nocf_check_attribute): New function.
4607 (c_common_attribute_table): Add 'nocf_check' handling.
4608 * gimple-parser.c: Add second argument NULL to
4609 gimple_build_call_from_tree.
4610 * attrib.c (comp_type_attributes): Check nocf_check attribute.
4611 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
4613 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
4614 * common.opt: Add fcf-protection flag.
4615 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
4616 * flag-types.h: Add enum cf_protection_level.
4617 * gimple.c (gimple_build_call_from_tree): Add second parameter.
4618 Add 'nocf_check' attribute propagation to gimple call.
4619 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
4620 (gimple_build_call_from_tree): Update prototype.
4621 (gimple_call_nocf_check_p): New function.
4622 (gimple_call_set_nocf_check): Likewise.
4623 * gimplify.c: Add second argument to gimple_build_call_from_tree.
4624 * ipa-icf.c: Add nocf_check attribute in statement hash.
4625 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
4626 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
4627 * toplev.c (process_options): Add flag_cf_protection handling.
4629 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4631 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
4633 2017-10-20 Richard Biener <rguenther@suse.de>
4635 PR tree-optimization/82603
4636 * tree-if-conv.c (predicate_mem_writes): Make sure to only
4637 remove false predicated stores.
4639 2017-10-20 Richard Biener <rguenther@suse.de>
4641 * graphite-isl-ast-to-gimple.c
4642 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
4643 Remove return value and simplify, dump copied stmt after lhs
4645 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
4646 Reduce dump verbosity.
4647 (gsi_insert_earliest): Likewise.
4648 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
4649 * graphite.c (print_global_statistics): Adjust dumping.
4650 (print_graphite_scop_statistics): Likewise.
4651 (print_graphite_statistics): Do not dump loops here.
4652 (graphite_transform_loops): But here.
4654 2017-10-20 Nicolas Roche <roche@adacore.com>
4656 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
4657 * configure: Regenerate.
4659 2017-10-20 Jakub Jelinek <jakub@redhat.com>
4662 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
4663 functions when optimizing replace GIMPLE_RETURN stmts with
4664 calls to __builtin_unreachable ().
4667 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
4668 for -fsanitize=thread link of executables.
4669 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
4670 link of executables.
4673 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
4675 (<shift_insn><mode>3<mask_name>): Change the last of the 3
4676 define_insns for logical vector shifts to use VI248_AVX512BW
4677 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
4678 condition, useless isa and prefix attributes. Change the first
4679 2 of these define_insns to ...
4680 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
4681 define_insn for avx512vl.
4682 (<shift_insn><mode>3): ... and this, new define_insn without
4683 masking for non-avx512vl.
4686 * config/i386/sse.md (*andnot<mode>3,
4687 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
4688 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
4689 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
4690 not applied use empty suffix even for TARGET_AVX512VL.
4691 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
4692 is applied, supply evex,evex or evex,evex,evex instead of just
4695 2017-10-20 Julia Koval <julia.koval@intel.com>
4697 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
4698 (OPTION_MASK_ISA_GFNI_UNSET): New.
4699 (ix86_handle_option): Handle OPT_mgfni.
4700 * config/i386/cpuid.h (bit_GFNI): New.
4701 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
4702 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
4703 * config/i386/i386.c (ix86_target_string): Add -mgfni.
4704 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
4705 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
4706 * config/i386/i386.opt: Add mgfni.
4708 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
4710 * config/msp430/msp430.c (msp430_option_override): Disable
4711 -fdelete-null-pointer-checks.
4712 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
4714 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4716 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
4717 of x87 and SSE instructions.
4719 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4721 * asan.c (create_cond_insert_point): Do not update edge count.
4722 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
4723 (afdo_propagate_circuit): Likewise.
4724 (afdo_calculate_branch_prob): Likewise.
4725 (afdo_annotate_cfg): Likewise.
4726 * basic-block.h (struct edge_def): Remove count.
4727 (edge_def::count): New accessor.
4728 * bb-reorder.c (rotate_loop): Update.
4729 (find_traces_1_round): Update.
4730 (connect_traces): Update.
4731 (sanitize_hot_paths): Update.
4732 * cfg.c (unchecked_make_edge): Update.
4733 (make_single_succ_edge): Update.
4734 (check_bb_profile): Update.
4735 (dump_edge_info): Update.
4736 (update_bb_profile_for_threading): Update.
4737 (scale_bbs_frequencies_int): Update.
4738 (scale_bbs_frequencies_gcov_type): Update.
4739 (scale_bbs_frequencies_profile_count): Update.
4740 (scale_bbs_frequencies): Update.
4741 * cfganal.c (connect_infinite_loops_to_exit): Update.
4742 * cfgbuild.c (compute_outgoing_frequencies): Update.
4743 (find_many_sub_basic_blocks): Update.
4744 * cfgcleanup.c (try_forward_edges): Update.
4745 (try_crossjump_to_edge): Update
4746 * cfgexpand.c (expand_gimple_cond): Update
4747 (expand_gimple_tailcall): Update
4748 (construct_exit_block): Update
4749 * cfghooks.c (verify_flow_info): Update
4750 (redirect_edge_succ_nodup): Update
4751 (split_edge): Update
4752 (make_forwarder_block): Update
4753 (duplicate_block): Update
4754 (account_profile_record): Update
4755 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
4756 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
4757 * cfgloopmanip.c (scale_loop_profile): Update.
4759 (lv_adjust_loop_entry_edge): Update.
4760 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
4761 (force_nonfallthru_and_redirect): Update.
4762 (purge_dead_edges): Update.
4763 (rtl_flow_call_edges_add): Update.
4764 * cgraphunit.c (init_lowered_empty_function): Update.
4765 (cgraph_node::expand_thunk): Update.
4766 * gimple-pretty-print.c (dump_probability): Update.
4767 (dump_edge_probability): Update.
4768 * gimple-ssa-isolate-paths.c (isolate_path): Update.
4769 * haifa-sched.c (sched_create_recovery_edges): Update.
4770 * hsa-gen.c (convert_switch_statements): Update.
4771 * ifcvt.c (dead_or_predicable): Update.
4772 * ipa-inline-transform.c (inline_transform): Update.
4773 * ipa-split.c (split_function): Update.
4774 * ipa-utils.c (ipa_merge_profiles): Update.
4775 * loop-doloop.c (add_test): Update.
4776 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
4777 * lto-streamer-in.c (input_cfg): Update.
4778 (input_function): Update.
4779 * lto-streamer-out.c (output_cfg): Update.
4780 * modulo-sched.c (sms_schedule): Update.
4781 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
4782 * predict.c (maybe_hot_edge_p): Update.
4783 (unlikely_executed_edge_p): Update.
4784 (probably_never_executed_edge_p): Update.
4785 (dump_prediction): Update.
4786 (drop_profile): Update.
4787 (propagate_unlikely_bbs_forward): Update.
4788 (determine_unlikely_bbs): Update.
4789 (force_edge_cold): Update.
4790 * profile.c (compute_branch_probabilities): Update.
4791 * reg-stack.c (better_edge): Update.
4792 * shrink-wrap.c (handle_simple_exit): Update.
4793 * tracer.c (better_p): Update.
4794 * trans-mem.c (expand_transaction): Update.
4795 (split_bb_make_tm_edge): Update.
4796 * tree-call-cdce.c: Update.
4797 * tree-cfg.c (gimple_find_sub_bbs): Update.
4798 (gimple_split_edge): Update.
4799 (gimple_duplicate_sese_region): Update.
4800 (gimple_duplicate_sese_tail): Update.
4801 (gimple_flow_call_edges_add): Update.
4802 (insert_cond_bb): Update.
4803 (execute_fixup_cfg): Update.
4804 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
4805 * tree-complex.c (expand_complex_div_wide): Update.
4806 * tree-eh.c (lower_resx): Update.
4807 (unsplit_eh): Update.
4808 (cleanup_empty_eh_move_lp): Update.
4809 * tree-inline.c (copy_edges_for_bb): Update.
4810 (freqs_to_counts): Update.
4811 (copy_cfg_body): Update.
4812 * tree-ssa-dce.c (remove_dead_stmt): Update.
4813 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
4814 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
4815 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
4816 (unloop_loops): Update.
4817 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
4818 * tree-ssa-loop-split.c (connect_loops): Update.
4819 (split_loop): Update.
4820 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
4821 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
4822 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
4823 * tree-ssa-reassoc.c (branch_fixup): Update.
4824 * tree-ssa-tail-merge.c (replace_block_by): Update.
4825 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
4826 (compute_path_counts): Update.
4827 (update_profile): Update.
4828 (recompute_probabilities): Update.
4829 (update_joiner_offpath_counts): Update.
4830 (estimated_freqs_path): Update.
4831 (freqs_to_counts_path): Update.
4832 (clear_counts_path): Update.
4833 (ssa_fix_duplicate_block_edges): Update.
4834 (duplicate_thread_path): Update.
4835 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
4836 (case_bit_test_cmp): Update.
4837 (collect_switch_conv_info): Update.
4838 (gen_inbound_check): Update.
4839 (do_jump_if_equal): Update.
4840 (emit_cmp_and_jump_insns): Update.
4841 * tree-tailcall.c (decrease_profile): Update.
4842 (eliminate_tail_call): Update.
4843 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
4844 (vect_do_peeling): Update.
4845 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
4846 * ubsan.c (ubsan_expand_null_ifn): Update.
4847 (ubsan_expand_ptr_ifn): Update.
4848 * value-prof.c (gimple_divmod_fixed_value): Update.
4849 (gimple_mod_pow2): Update.
4850 (gimple_mod_subtract): Update.
4851 (gimple_ic): Update.
4852 (gimple_stringop_fixed_value): Update.
4854 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
4857 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
4859 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
4861 PR rtl-optimization/82395
4862 * ira-color.c (allocno_priority_compare_func): Fix comparison step
4863 based on non_spilled_static_chain_regno_p.
4865 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
4867 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
4868 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
4869 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
4871 2017-10-19 Martin Sebor <msebor@redhat.com>
4873 PR tree-optimization/82596
4874 * tree.c (array_at_struct_end_p): Handle STRING_CST.
4876 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
4878 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
4879 (get_mem_refs_of_builtin_call): Likewise.
4880 * builtins.c (expand_builtin_apply): Adjust call to
4881 allocate_dynamic_stack_space.
4882 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
4883 the third argument to allocate_dynamic_stack_space, otherwise -1.
4884 (expand_builtin): Deal with all alloca variants.
4885 (is_inexpensive_builtin): Likewise.
4886 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
4887 * calls.c (special_function_p): Deal with all alloca variants.
4888 (initialize_argument_information): Adjust call to
4889 allocate_dynamic_stack_space.
4890 (expand_call): Likewise.
4891 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
4892 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
4893 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
4894 use it for the stack usage computation.
4895 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
4896 * function.c (gimplify_parameters): Call build_alloca_call_expr.
4897 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
4898 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
4899 (in_loop_p): Remove first argument and useless check.
4900 (pass_walloca::execute): Remove useless test and adjust call to above.
4901 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
4902 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
4903 (gimplify_call_expr): Deal with all alloca variants.
4904 * hsa-gen.c (gen_hsa_alloca): Likewise.
4905 (gen_hsa_insns_for_call): Likewise.
4906 * ipa-pure-const.c (special_builtin_state): Likewise.
4907 * tree-chkp.c (chkp_build_returned_bound): Likewise.
4908 * tree-object-size.c (alloc_object_size): Likewise.
4909 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
4910 (call_may_clobber_ref_p_1): Likewise.
4911 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
4912 (ccp_fold_stmt): Likewise.
4913 (optimize_stack_restore): Likewise.
4914 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
4915 (mark_all_reaching_defs_necessary_1): Likewise.
4916 (propagate_necessity): Likewise.
4917 (eliminate_unnecessary_stmts): Likewise.
4918 * tree.c (build_common_builtin_nodes): Build
4919 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
4920 (build_alloca_call_expr): New function.
4921 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
4922 (CASE_BUILT_IN_ALLOCA): Likewise.
4923 (build_alloca_call_expr): Declare.
4924 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
4926 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
4929 * dwarf2out.c (new_die_raw): New static inline function.
4930 (new_die): Use it to create the DIE.
4931 (add_AT_external_die_ref): Likewise.
4932 (clone_die): Likewise.
4933 (clone_as_declaration): Likewise.
4934 (dwarf2out_vms_debug_main_pointer): Likewise.
4935 (base_type_die): Likewise. Remove early return for corner cases.
4936 Do not call add_pubtype on the DIE here.
4937 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
4938 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
4939 typedefs for base types with DW_AT_endianity. Make sure a DIE with
4940 native order exists for base types, attach the DIE manually and call
4941 add_pubtype on it. Do not equate a reverse order DIE to the type.
4943 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
4945 * config/arm/arm.c (align_ok_ldrd_strd): New function.
4946 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
4948 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
4950 2017-10-19 Jakub Jelinek <jakub@redhat.com>
4952 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
4953 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
4954 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
4955 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
4956 * opts.c (sanitizer_opts): Add builtin.
4957 * ubsan.c (instrument_builtin): New function.
4958 (pass_ubsan::execute): Call it.
4959 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
4960 * doc/invoke.texi: Document -fsanitize=builtin.
4962 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
4963 builtins, store max (log2 (align), 0) into uchar field instead of
4964 align into uptr field.
4965 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
4966 store uchar 0 field instead of uptr 0 field.
4967 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
4968 instead of passing one address of struct with 2 locations pass
4969 two addresses of structs with 1 location each.
4970 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
4971 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
4972 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
4973 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
4974 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
4975 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
4976 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
4977 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
4979 2017-10-19 Martin Liska <mliska@suse.cz>
4982 * file-find.c (remove_prefix): Remove.
4983 * file-find.h (remove_prefix): Likewise.
4984 * gcc-ar.c: Remove smartness of lookup.
4986 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
4988 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
4989 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
4990 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
4992 2017-10-19 Jakub Jelinek <jakub@redhat.com>
4995 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
4996 (setcc + and to xor + setcc): New peephole2.
4998 2017-10-19 Tom de Vries <tom@codesourcery.com>
5000 * doc/sourcebuild.texi (Test Directives, Variants of
5001 dg-require-support): Add dg-require-stack-size.
5003 2017-10-19 Martin Liska <mliska@suse.cz>
5006 * gimplify.c (gimplify_decl_expr): Do not instrument variables
5007 that have a large alignment.
5008 (gimplify_target_expr): Likewise.
5010 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
5012 PR rtl-optimization/82602
5013 * ira.c (rtx_moveable_p): Return false for volatile asm.
5015 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
5018 * config/i386/i386-modes.def (CCGZ): New CC mode.
5019 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
5020 * config/i386/predicates.md (ix86_comparison_operator):
5022 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
5023 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
5024 with double-word subtraction.
5025 (put_condition_code): Handle CCGZmode.
5027 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
5029 * wide-int.cc (debug (const wide_int &)): New.
5030 (debug (const wide_int *)): New.
5031 (debug (const widest_int &)): New.
5032 (debug (const widest_int *)): New.
5034 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
5037 * lra-constraints.c (curr_insn_transform): Use non-input operand
5038 instead of output one for matched reload.
5040 2017-10-18 Bin Cheng <bin.cheng@arm.com>
5042 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
5043 (tree-ssa-loop-ivopts.h): New header file.
5044 (struct builtin_info): New fields.
5045 (classify_builtin_1): Compute and record base and offset parts for
5046 memset builtin partition by calling strip_offset.
5047 (offset_cmp, fuse_memset_builtins): New functions.
5048 (finalize_partitions): Fuse adjacent memset partitions by calling
5050 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
5051 Expose the interface.
5052 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
5054 2017-10-18 Bin Cheng <bin.cheng@arm.com>
5056 PR tree-optimization/82574
5057 * tree-loop-distribution.c (find_single_drs): New parameter. Check
5058 that data reference must be executed exactly once per iteration
5059 against the outermost loop in nest.
5060 (classify_partition): Update call to above function.
5062 2017-10-18 Richard Biener <rguenther@suse.de>
5064 PR tree-optimization/82591
5065 * graphite.c (graphite_transform_loops): Move code gen message
5067 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
5068 Here. Handle scop_to_isl_ast failing.
5069 (scop_to_isl_ast): Limit the number of ISL operations.
5071 2017-10-18 Richard Biener <rguenther@suse.de>
5073 * graphite-isl-ast-to-gimple.c
5074 (translate_isl_ast_to_gimple::set_rename): Simplify.
5075 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
5076 (graphite_copy_stmts_from_block): ... here.
5077 (copy_bb_and_scalar_dependences): Simplify.
5078 (add_parameters_to_ivs_params): Canonicalize.
5079 (generate_entry_out_of_ssa_copies): Simplify.
5080 * graphite-sese-to-poly.c (extract_affine_name): Simplify
5081 by passing in ISL dimension.
5082 (parameter_index_in_region_1): Rename to ...
5083 (parameter_index_in_region): ... this.
5084 (extract_affine): Adjust assert, pass down parameter index.
5085 (add_param_constraints): Use range-info when available.
5086 (build_scop_context): Adjust.
5087 * sese.c (new_sese_info): Adjust.
5088 (free_sese_info): Likewise.
5089 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
5090 Remove unused typedefs.
5091 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
5093 2017-10-18 Martin Liska <mliska@suse.cz>
5095 * combine.c (simplify_compare_const): Add gcc_fallthrough.
5097 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
5099 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
5100 (s390_sched_init): Do not reset s390_sched_state if we entered the
5101 current basic block via a fallthru edge and all others are unlikely.
5103 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
5105 * config/s390/s390.c (NUM_SIDES): New variable.
5106 (LONGRUNNING_THRESHOLD): New variable.
5107 (LATENCY_FACTOR): New variable.
5108 (s390_sched_score): Decrease score for long-running instructions on
5110 (s390_sched_variable_issue): Perform bookkeeping for long-running
5113 2017-10-18 Richard Biener <rguenther@suse.de>
5115 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
5116 Simplify with removal of the parameter rename map.
5117 (set_rename): Likewise.
5118 (should_copy_to_new_region): Likewise.
5119 (graphite_copy_stmts_from_block): Likewise.
5120 (copy_bb_and_scalar_dependences): Remove initialization of
5121 unused copied_bb_map.
5123 (copy_internal_parameters): Likewise.
5124 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
5125 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
5126 Use INTEGRAL_TYPE_P.
5127 (parameter_index_in_region_1): Rename to ...
5128 (assign_parameter_index_in_region): ... this. Assert we have
5129 a parameter we handle.
5130 (scan_tree_for_params): Adjust.
5131 * sese.h (parameter_rename_map_t): Remove.
5132 (struct sese_info_t): Remove unused parameter_rename_map and
5133 copied_bb_map members.
5134 * sese.c (new_sese_info): Adjust.
5135 (free_sese_info): Likewise.
5137 2017-10-18 Martin Liska <mliska@suse.cz>
5140 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
5141 on an abnormal edge.
5143 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
5145 * doc/invoke.texi (ffunction-sections and fdata-sections):
5148 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
5150 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
5151 the use statement can throw internally.
5153 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
5155 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
5156 any RTX present on the RHS of a SET.
5157 * compare-elim.c (try_eliminate_compare): Restore comment.
5159 2017-10-17 Jakub Jelinek <jakub@redhat.com>
5161 * langhooks.h (struct lang_hooks): Document that tree_size langhook
5162 may be also called on tcc_type nodes.
5163 * langhooks.c (lhd_tree_size): Likewise.
5165 2017-10-17 David Malcolm <dmalcolm@redhat.com>
5167 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
5168 format_warning_at_substring.
5169 (maybe_warn): Convert source_range * param to a location_t. Pass
5170 UNKNOWN_LOCATION rather than NULL to fmtwarn.
5171 (format_directive): Remove code to extract source_ranges and
5172 source_range * in favor of just a location_t.
5173 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
5175 * substring-locations.c (format_warning_va): Convert
5176 source_range * param to a location_t.
5177 (format_warning_at_substring): Likewise.
5178 * substring-locations.h (format_warning_va): Likewise.
5179 (format_warning_at_substring): Likewise.
5181 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
5183 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
5185 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
5186 and scatter/gather ops.
5188 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
5189 vec_gather_load and vec_scatter_store.
5190 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
5191 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
5192 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
5193 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
5194 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
5195 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
5197 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
5199 * reg-stack.c (compare_for_stack_reg): Add bool argument.
5200 Detect FTST instruction and handle its register pops. Only pop
5201 second operand if can_pop_second_op is true.
5202 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
5203 set can_pop_second_op to false in the compare_for_stack_reg call.
5205 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
5206 output_fp_compare for stack register operands.
5207 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
5208 instructions here. Do not emit stack register pops here. Assert
5209 that FCOMPP pops next to top stack register. Rewrite function.
5211 2017-10-17 Nathan Sidwell <nathan@acm.org>
5214 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
5215 use DECL_ASSEMBLER_NAME_RAW.
5218 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
5221 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
5222 Wilco Dijkstra <wilco.dijkstra@arm.com>
5224 * builtins.c (expand_builtin_update_setjmp_buf): Add a
5225 converstion to Pmode from the buf_addr.
5227 2017-10-17 Richard Biener <rguenther@suse.de>
5229 * graphite-dependences.c (scop_get_reads_and_writes): Change
5230 output parameters to references.
5232 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
5234 PR 71026/tree-optimization
5235 * fold-const.c (distribute_real_division): Removed.
5236 (fold_binary_loc): Remove calls to distribute_real_divison.
5238 2017-10-17 Richard Biener <rguenther@suse.de>
5240 * graphite-scop-detection.c
5241 (scop_detection::stmt_has_simple_data_refs_p): Always use
5242 the full nest as region.
5243 (try_generate_gimple_bb): Likewise.
5244 * sese.c (scalar_evolution_in_region): Simplify now that
5245 SCEV can handle instantiation in regions.
5246 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
5247 in the non-loop part of a function if requested.
5249 2017-10-17 Richard Biener <rguenther@suse.de>
5251 PR tree-optimization/82563
5252 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
5254 (graphite_regenerate_ast_isl): Call it.
5255 * graphite-scop-detection.c (build_scops): Remove entry edge split.
5257 2017-10-17 Jakub Jelinek <jakub@redhat.com>
5259 PR tree-optimization/82549
5260 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
5261 Formatting fixes. Instead of calling make_bit_field_ref with negative
5264 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
5266 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
5267 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
5268 _mm_maskz_reduce_ss): New.
5269 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
5270 __builtin_ia32_reducess_mask): Ditto..
5271 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
5272 * config/i386/sse.md (reduces<mode>): Renamed to ...
5273 (reduces<mode><mask_scalar_name>): ... this.
5274 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
5276 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
5277 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
5279 2017-10-16 David Malcolm <dmalcolm@redhat.com>
5281 * Makefile.in (OBJS): Add unique-ptr-tests.o.
5282 * selftest-run-tests.c (selftest::run_tests): Call
5283 selftest::unique_ptr_tests_cc_tests.
5284 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
5285 * unique-ptr-tests.cc: New file.
5287 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
5290 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5292 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5293 (make_hard_regno_born, make_hard_regno_dead): Update
5294 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
5296 2017-10-16 Jeff Law <law@redhat.com>
5298 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
5300 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
5302 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
5304 2017-10-16 Olivier Hainque <hainque@adacore.com>
5306 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
5307 with_cpu if we were configured for an e500v2 target cpu name.
5309 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
5311 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
5312 * doc/invoke.texi: Document +nodsp as a valid extension for
5315 2017-10-16 Martin Liska <mliska@suse.cz>
5317 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
5318 (test_set_range): Likewise.
5319 (test_range_functions): Rename to ...
5320 (test_bit_in_range): ... this.
5321 (sbitmap_c_tests): Add new test.
5323 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5325 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
5327 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
5328 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
5330 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5332 * config/aarch64/aarch64-builtins.c
5333 (aarch64_types_quadopu_lane_qualifiers): New.
5334 (TYPES_QUADOPU_LANE): New.
5335 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
5336 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
5337 (aarch64_<sur>dot_laneq<vsi2qi>): New.
5338 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
5339 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
5340 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
5341 (Vdottype, DOTPROD): New.
5342 (sur): Add SDOT and UDOT.
5344 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5346 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
5347 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
5348 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
5350 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
5351 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
5352 Enable TARGET_DOTPROD.
5353 (cortex-a75.cortex-a55): Likewise.
5354 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
5356 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5358 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
5359 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
5361 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
5363 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
5364 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
5365 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
5366 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
5367 * config/arm/types.md (neon_dot, neon_dot_q): New.
5368 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
5370 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5372 * config/arm/arm.h (TARGET_DOTPROD): New.
5373 * config/arm/arm.c (arm_arch_dotprod): New.
5374 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
5375 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
5376 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
5377 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
5378 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
5379 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
5380 * doc/invoke.texi (armv8.2-a): Document dotprod
5382 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
5384 * i386.c (ix86_vec_cost): New function.
5385 (ix86_rtx_costs): Handle vector operations better.
5386 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
5387 * x86-tune-costs.h: Add new costs to all tables.
5389 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
5391 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
5393 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
5394 divsd, sqrtss and sqrtsd
5395 * x86-tune-costs.h: Add new entries to all costs.
5396 (znver1_cost): Fix to match real instruction latencies.
5398 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5399 Michael Collison <michael.collison@arm.com>
5401 * compare-elim.c: Include emit-rtl.h.
5402 (can_merge_compare_into_arith): New function.
5403 (try_validate_parallel): Likewise.
5404 (try_merge_compare): Likewise.
5405 (try_eliminate_compare): Call the above when no previous clobber
5407 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
5410 2017-10-14 Jakub Jelinek <jakub@redhat.com>
5414 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
5415 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
5419 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
5420 to be any operand_equal_p operands. For & (B - 1) require
5421 B to be power of 2. Recognize
5422 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
5424 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5427 * optabs.c (expand_memory_blockage): Fix call of
5428 targetm.have_memory_blockage.
5430 2017-10-14 Jakub Jelinek <jakub@redhat.com>
5433 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
5434 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
5435 objects to extra_objs instead of overwriting it.
5437 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5439 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
5440 Use any_fp_register_operand as operand[3] predicate. Simplify
5441 equality test for operands[2] and operands[4] memory location.
5442 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
5443 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
5444 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
5445 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
5446 any_fp_register_operand as operand[1] predicate. Simplify
5447 equality test for operands[0] and operands[3] memory location.
5448 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
5449 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
5450 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
5452 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5454 * target-insns.def: Add memory_blockage.
5455 * optabs.c (expand_memory_blockage): New function.
5456 (expand_asm_memory_barrier): Rename ...
5457 (expand_asm_memory_blockage): ... to this.
5458 (expand_mem_thread_fence): Call expand_memory_blockage
5459 instead of expand_asm_memory_barrier.
5460 (expand_mem_singnal_fence): Ditto.
5461 (expand_atomic_load): Ditto.
5462 (expand_atomic_store): Ditto.
5463 * doc/md.texi (Standard Pattern Names For Generation):
5464 Document memory_blockage instruction pattern.
5466 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
5468 * config/rl78/rl78.c (rl78_emit_libcall): New function.
5469 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
5470 * config/rl78/rl78.md: New define_expand "adddi3".
5472 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
5474 * cfghooks.c (verify_flow_info): Disable check that all probabilities
5477 2017-10-13 Jeff Law <law@redhat.com>
5479 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
5481 2017-10-13 Jakub Jelinek <jakub@redhat.com>
5484 * internal-fn.c (expand_mul_overflow): If both operands have
5485 the same highpart of -1 or 0 and the topmost bit of lowpart
5486 is different, overflow is if res <= 0 rather than res < 0.
5488 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
5490 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
5491 TARGET_P9_VECTOR code for unaligned_load case.
5493 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
5495 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
5497 2017-10-13 Nathan Sidwell <nathan@acm.org>
5499 * tree-core.h (tree_contains_struct): Make bool.
5500 * tree.c (tree_contains_struct): Likewise.
5501 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
5502 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
5503 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
5504 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
5505 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
5507 2017-10-13 Richard Biener <rguenther@suse.de>
5509 * graphite-isl-ast-to-gimple.c
5510 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
5511 parameters and dominance check.
5512 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
5513 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
5514 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
5515 Do not update SSA form here or do intermediate IL verification.
5516 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
5517 (graphite_initialize): Remove check on the number of loops in
5518 the function and inline into graphite_transform_loops.
5519 (graphite_finalize): Inline into graphite_transform_loops.
5520 (graphite_transform_loops): Perform SSA update and IL verification
5522 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
5524 2017-10-13 Richard Biener <rguenther@suse.de>
5526 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
5527 graphite_expression_type_precision): Avoid global constructor
5529 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
5530 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
5531 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
5532 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
5533 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
5534 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
5536 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
5539 * config/i386/i386.h (ix86_red_zone_size): New.
5540 * config/i386/i386.md (push peephole2s): Replace
5541 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
5543 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5544 Alan Hayward <alan.hayward@arm.com>
5545 David Sherwood <david.sherwood@arm.com>
5547 * combine.c (can_change_dest_mode): Reject changes in
5548 REGMODE_NATURAL_SIZE.
5550 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5551 Alan Hayward <alan.hayward@arm.com>
5552 David Sherwood <david.sherwood@arm.com>
5554 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
5555 (expand_debug_source_expr): Likewise.
5556 * combine.c (combine_simplify_rtx): Likewise.
5557 * cse.c (fold_rtx): Likewise.
5558 * fwprop.c (canonicalize_address): Likewise.
5559 * targhooks.c (default_shift_truncation_mask): Likewise.
5561 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5562 Alan Hayward <alan.hayward@arm.com>
5563 David Sherwood <david.sherwood@arm.com>
5565 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
5566 (widened_mode): Likewise.
5567 (expand_unop): Likewise.
5568 * ree.c (transform_ifelse): Likewise.
5569 (merge_def_and_ext): Likewise.
5570 (combine_reaching_defs): Likewise.
5571 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5573 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5574 Alan Hayward <alan.hayward@arm.com>
5575 David Sherwood <david.sherwood@arm.com>
5577 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
5578 * combine.c (gen_lowpart_for_combine): Likewise.
5579 * dwarf2out.c (rtl_for_decl_location): Likewise.
5580 * final.c (alter_subreg): Likewise.
5581 * rtlhooks.c (gen_lowpart_general): Likewise.
5582 (gen_lowpart_if_possible): Likewise.
5584 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5585 Alan Hayward <alan.hayward@arm.com>
5586 David Sherwood <david.sherwood@arm.com>
5588 * calls.c (expand_call): Use subreg_lowpart_offset.
5589 * cse.c (cse_insn): Likewise.
5590 * regcprop.c (copy_value): Likewise.
5591 (copyprop_hardreg_forward_1): Likewise.
5593 2017-10-13 Jakub Jelinek <jakub@redhat.com>
5596 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
5597 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
5598 =Q constraints to +Q and into insn condition add check
5599 that operands[0] and operands[1] are equal.
5600 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
5601 =Q constraints to +Q and into insn condition add check
5602 that operands[0] is equal to either operands[1] or operands[2].
5605 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
5606 instead of handling MINUS_EXPR twice (once for each argument),
5607 canonicalize operand order and handle just once, use rtype where
5608 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
5611 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
5612 any values of __C while still being pattern recognizable as a simple
5615 2017-10-13 Richard Biener <rguenther@suse.de>
5617 PR tree-optimization/82451
5619 2017-10-02 Richard Biener <rguenther@suse.de>
5621 PR tree-optimization/82355
5622 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
5623 a mapping for the enclosing loop but avoid generating one for
5625 (copy_bb_and_scalar_dependences): Remove premature codegen
5626 error on PHIs in blocks duplicated into multiple places.
5627 * graphite-scop-detection.c
5628 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
5629 in the region use it as loop and nest to analyze the DR in.
5630 (try_generate_gimple_bb): Likewise.
5631 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
5632 (add_loop_constraints): For blocks in a loop not in the region
5633 create a dimension with a single iteration.
5634 * sese.h (gbb_loop_at_index): Remove assert.
5636 * cfgloop.c (loop_preheader_edge): For the loop tree root
5637 return the single successor of the entry block.
5638 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
5639 Reset the SCEV hashtable and niters.
5640 * graphite-scop-detection.c
5641 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
5642 assert that we only have POLYNOMIAL_CHREC that vary in loops
5643 contained in the region.
5644 (scop_detection::graphite_can_represent_expr): Adjust.
5645 (scop_detection::stmt_has_simple_data_refs_p): For loops
5646 not in the region set loop to NULL. The nest is now the
5647 entry edge to the region.
5648 (try_generate_gimple_bb): Likewise.
5649 * sese.c (scalar_evolution_in_region): Adjust for
5650 instantiate_scev change.
5651 * tree-data-ref.h (graphite_find_data_references_in_stmt):
5652 Make nest parameter the edge into the region.
5653 (create_data_ref): Likewise.
5654 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
5655 entry edge into a region and adjust instantiate_scev calls.
5656 (create_data_ref): Likewise.
5657 (graphite_find_data_references_in_stmt): Likewise.
5658 (find_data_references_in_stmt): Pass the loop preheader edge
5659 from the nest argument.
5660 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
5661 parameter the edge into the region.
5662 (instantiate_parameters): Use the loop preheader edge as entry.
5663 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
5665 (get_instantiated_value_entry): Make instantiate_below parameter
5666 the edge into the region.
5667 (instantiate_scev_name): Likewise. Adjust dominance checks,
5668 when we cannot use loop-based instantiation instantiate by
5669 walking use-def chains.
5670 (instantiate_scev_poly): Adjust.
5671 (instantiate_scev_binary): Likewise.
5672 (instantiate_scev_convert): Likewise.
5673 (instantiate_scev_not): Likewise.
5674 (instantiate_array_ref): Remove.
5675 (instantiate_scev_3): Likewise.
5676 (instantiate_scev_2): Likewise.
5677 (instantiate_scev_1): Likewise.
5678 (instantiate_scev_r): Do not blindly handle N-operand trees.
5679 Do not instantiate array-refs. Handle all constants and invariants.
5680 (instantiate_scev): Make instantiate_below parameter
5681 the edge into the region.
5682 (resolve_mixers): Use the loop preheader edge for the region
5683 parameter to instantiate_scev_r.
5684 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
5686 2017-10-13 Richard Biener <rguenther@suse.de>
5688 PR tree-optimization/82525
5689 * graphite-isl-ast-to-gimple.c
5690 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
5692 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
5693 Fail code generation when we cannot represent the isl integer.
5694 (binary_op_to_tree): Elide modulo operations that are no-ops
5695 in the type we code generate. Remove now superfluous code
5698 2017-10-13 Richard Biener <rguenther@suse.de>
5700 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
5701 (scop_detection::harmful_loop_in_region): Remove premature
5702 IV type restriction.
5703 (scop_detection::graphite_can_represent_scev): We can handle
5704 pointer IVs just fine.
5706 2017-10-13 Alan Modra <amodra@gmail.com>
5708 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
5709 "Clobbers and Scratch Registers". Add paragraph on
5710 alternative to clobbers for scratch registers and OpenBLAS
5713 2017-10-13 Alan Modra <amodra@gmail.com>
5715 * doc/extend.texi (Clobbers): Correct vax example. Delete old
5716 example of a memory input for a string of known length. Move
5717 commentary out of table. Add a number of new examples
5718 covering array memory inputs.
5720 2017-10-12 Martin Liska <mliska@suse.cz>
5722 PR tree-optimization/82493
5723 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
5724 (test_range_functions): New function.
5725 (sbitmap_c_tests): Likewise.
5726 * selftest-run-tests.c (selftest::run_tests): Run new tests.
5727 * selftest.h (sbitmap_c_tests): New function.
5729 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
5731 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
5733 * config/rs6000/amo.h: Fix spacing issue.
5735 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5738 * config/i386/i386.md (*ashl<mode>3_mask_1,
5739 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
5740 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
5743 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
5745 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
5746 (profile_probability): Set max_probability
5747 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
5749 * profile-count.c (profile_probability::differs_from_p): Do not
5750 rely on max_probaiblity == 10000
5752 2017-10-12 Jeff Law <law@redhat.com>
5754 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
5757 2017-10-12 Martin Sebor <msebor@redhat.com>
5761 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
5762 (handle_alias_pairs): Call it.
5763 * common.opt (-Wattribute-alias): New option.
5764 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
5765 * doc/invoke.texi (-Wattribute-alias): Document.
5767 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
5770 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
5772 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5774 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5775 (make_hard_regno_born, make_hard_regno_dead): Update
5776 bb_killed_pseudos and bb_gen_pseudos.
5778 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
5780 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
5782 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
5784 * config/alpha/alpha.c (alpha_split_conditional_move):
5785 Use std::swap instead of manually swapping.
5786 (alpha_stdarg_optimize_hook): Ditto.
5787 (alpha_canonicalize_comparison): Ditto.
5789 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5791 * tree-loop-distribution.c (struct builtin_info): New struct.
5792 (struct partition): Refactor fields into struct builtin_info.
5793 (partition_free): Free struct builtin_info.
5794 (build_size_arg_loc, build_addr_arg_loc): Delete.
5795 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
5796 information from struct builtin_info.
5797 (find_single_drs): New function refactored from classify_partition.
5798 Also moved builtin validity checks to this function.
5799 (compute_access_range, alloc_builtin): New functions.
5800 (classify_builtin_st, classify_builtin_ldst): New functions.
5801 (classify_partition): Refactor code into functions find_single_drs,
5802 classify_builtin_st and classify_builtin_ldst.
5803 (distribute_loop): Don't do runtime alias check when distributing
5805 (find_seed_stmts_for_distribution): New function.
5806 (pass_loop_distribution::execute): Refactor code finding seed
5807 stmts into above function. Support distribution for the innermost
5808 two-level loop nest. Adjust dump information.
5810 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5812 * tree-loop-distribution.c: Adjust the general comment.
5813 (NUM_PARTITION_THRESHOLD): New macro.
5814 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
5815 (classify_partition): Skip builtin pattern of loop nest's inner loop.
5816 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
5817 in call to build_partition_graph.
5818 (finalize_partitions): New parameter. Make loop distribution more
5819 conservative by fusing more partitions.
5820 (distribute_loop): Don't do runtime alias check in case of loop nest
5822 (find_seed_stmts_for_distribution): New function.
5823 (prepare_perfect_loop_nest): New function.
5824 (pass_loop_distribution::execute): Refactor code finding seed stmts
5825 and loop nest into above functions. Support loop nest distribution.
5826 Adjust dump information accordingly.
5828 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5830 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
5831 and set PTYPE_SEQUENTIAL for merged partition.
5833 2017-10-12 Richard Biener <rguenther@suse.de>
5835 PR tree-optimization/69728
5837 2017-09-19 Richard Biener <rguenther@suse.de>
5839 PR tree-optimization/69728
5840 * graphite-sese-to-poly.c (schedule_error): New global.
5841 (add_loop_schedule): Handle empty domain by failing the
5843 (build_original_schedule): Handle schedule_error.
5845 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
5846 domain by returning an unchanged schedule.
5848 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5850 * genrecog.c (validate_pattern): For VEC_SELECT verify that
5851 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
5853 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
5855 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
5858 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5861 * expr.c (store_field): Don't optimize away bitsize == 0 store
5862 from CALL_EXPR with addressable return type.
5864 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
5866 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
5867 * config/rs6000/rs6000.md (sel): Delete mode attribute.
5868 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
5869 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
5870 TARGET_ISEL instead of TARGET_ISEL<sel>.
5872 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
5874 * config/rs6000/rs6000.c
5875 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
5877 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
5879 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
5881 (reg_or_cint_operand): Fix comment.
5882 (reg_or_zero_operand): New predicate.
5883 * config/rs6000/rs6000-protos.h (output_isel): Delete.
5884 * config/rs6000/rs6000.c (output_isel): Delete.
5885 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
5886 instead of reg_or_cint_operand. Output instruction directly (not via
5888 (isel_unsigned_<mode>): Ditto.
5889 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
5890 gpc_reg_operand. Add an instruction alternative for this. Output
5891 instruction directly.
5892 (*isel_reversed_unsigned_<mode>): Ditto.
5894 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
5896 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
5897 (TARGET_CANONICALIZE_COMPARISON): Define.
5899 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
5902 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
5903 Check whether the dest is REG before adding REG_EQUIV note.
5905 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
5908 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5910 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5911 (make_hard_regno_born, make_hard_regno_dead): Update
5912 bb_killed_pseudos and bb_gen_pseudos.
5914 2017-10-11 Nathan Sidwell <nathan@acm.org>
5916 * incpath.h (enum incpath_kind): Name enum, prefix values.
5917 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
5918 * incpath.c (heads, tails): Use INC_MAX.
5919 (add_env_var_paths, add_standard_paths): Use incpath_kind.
5920 (merge_include_chains, split_quote_chain,
5921 register_include_chains): Update incpath_kind names.
5922 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
5923 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
5925 (add_framework_path, darwin_register_objc_includes): Likewise.
5926 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
5928 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
5930 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
5931 Do not use float_operator operator predicate.
5932 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
5933 * config/i386/predicates.md (float_operator): Remove predicate.
5935 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
5937 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
5938 (*jccxf_i387): Ditto.
5939 (*jcc<mode>_i387): Ditto.
5940 (*jccu<mode>_i387): Ditto.
5941 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
5942 (*jcc_*_i387 splitters): Remove.
5943 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
5944 * config/i386/i386.c (ix86_split_fp_branch): Remove.
5945 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
5948 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
5950 * profile-count.h (slow_safe_scale_64bit): New function.
5951 (safe_scale_64bit): New inline.
5952 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
5953 * profile-count.c: Include wide-int.h
5954 (slow_safe_scale_64bit): New.
5956 2017-10-11 Nathan Sidwell <nathan@acm.org>
5958 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
5959 HAS_DECL_ASSEMBLER_NAME_P.
5960 * gimple-expr.c (gimple_decl_printable_name: Check
5961 HAS_DECL_ASSEMBLER_NAME_P too.
5962 * ipa-utils.h (type_in_anonymous_namespace_p): Check
5963 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
5964 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
5965 * passes.c (rest_of_decl_compilation): Check
5966 HAS_DECL_ASSEMBLER_NAME_P too.
5967 * recog.c (verify_changes): Likewise.
5968 * tree-pretty-print.c (dump_decl_name): Likewise.
5969 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
5971 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
5972 (SET_DECL_ASSEMBLER_NAME): Use it.
5973 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
5974 (COPY_DECL_ASSEMBLER_NAME): Likewise.
5975 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
5977 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
5979 * config.gcc (i386, x86_64): Add extra objects.
5980 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
5981 (ix86_min_insn_size): Declare.
5982 (ix86_issue_rate): Declare.
5983 (ix86_adjust_cost): Declare.
5984 (ia32_multipass_dfa_lookahead): Declare.
5985 (ix86_macro_fusion_p): Declare.
5986 (ix86_macro_fusion_pair_p): Declare.
5987 (ix86_bd_has_dispatch): Declare.
5988 (ix86_bd_do_dispatch): Declare.
5989 (ix86_core2i7_init_hooks): Declare.
5990 (ix86_atom_sched_reorder): Declare.
5991 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
5992 (COSTS_N_BYTES): Move to x86-tune-costs.h.
5993 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
5994 (rip_relative_addr_p): Rename to ...
5995 (ix86_rip_relative_addr_p): ... this one; export.
5996 (memory_address_length): Update.
5997 (ix86_issue_rate): Move to x86-tune-sched.c.
5998 (ix86_flags_dependent): Move to x86-tune-sched.c.
5999 (ix86_agi_dependent): Move to x86-tune-sched.c.
6000 (exact_dependency_1): Move to x86-tune-sched.c.
6001 (exact_store_load_dependency): Move to x86-tune-sched.c.
6002 (ix86_adjust_cost): Move to x86-tune-sched.c.
6003 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
6004 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
6005 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
6006 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
6007 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
6008 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
6009 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
6010 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
6011 (min_insn_size): Rename to ...
6012 (ix86_min_insn_size): ... this one; export.
6013 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
6014 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
6015 (core2i7_first_cycle_multipass_backtrack): Move to
6016 x86-tune-sched-core.c.
6017 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
6018 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
6019 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
6020 (ix86_avoid_jump_mispredicts): Update.
6021 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
6022 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
6023 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
6024 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
6025 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
6026 (MAX_INSN): Move to ix86-tune-sched-bd.c.
6027 (MAX_IMM): Move to ix86-tune-sched-bd.c.
6028 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
6029 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
6030 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
6031 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
6032 (MAX_STORE): Move to ix86-tune-sched-bd.c.
6033 (BIG): Move to ix86-tune-sched-bd.c.
6034 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
6035 (enum insn_path): Move to ix86-tune-sched-bd.c.
6036 (get_mem_group): Move to ix86-tune-sched-bd.c.
6037 (is_cmp): Move to ix86-tune-sched-bd.c.
6038 (dispatch_violation): Move to ix86-tune-sched-bd.c.
6039 (is_branch): Move to ix86-tune-sched-bd.c.
6040 (is_prefetch): Move to ix86-tune-sched-bd.c.
6041 (init_window): Move to ix86-tune-sched-bd.c.
6042 (allocate_window): Move to ix86-tune-sched-bd.c.
6043 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
6044 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
6045 (process_end_window): Move to ix86-tune-sched-bd.c.
6046 (allocate_next_window): Move to ix86-tune-sched-bd.c.
6047 (find_constant): Move to ix86-tune-sched-bd.c.
6048 (get_num_immediates): Move to ix86-tune-sched-bd.c.
6049 (has_immediate): Move to ix86-tune-sched-bd.c.
6050 (get_insn_path): Move to ix86-tune-sched-bd.c.
6051 (get_insn_group): Move to ix86-tune-sched-bd.c.
6052 (count_num_restricted): Move to ix86-tune-sched-bd.c.
6053 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
6054 (add_insn_window): Move to ix86-tune-sched-bd.c.
6055 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
6056 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
6057 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
6058 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
6059 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
6060 (do_dispatch): Move to ix86-tune-sched-bd.c.
6061 (has_dispatch): Move to ix86-tune-sched-bd.c.
6062 * config/i386/t-i386: Add new object files.
6063 * config/i386/x86-tune-costs.h: New file.
6064 * config/i386/x86-tune-sched-atom.c: New file.
6065 * config/i386/x86-tune-sched-bd.c: New file.
6066 * config/i386/x86-tune-sched-core.c: New file.
6067 * config/i386/x86-tune-sched.c: New file.
6069 2017-10-11 Liu Hao <lh_mouse@126.com>
6071 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
6072 the generic version below instead.
6073 (should_colorize): Recognize Windows consoles as terminals
6075 * pretty-print.c [__MINGW32__] (write_all): New function.
6076 [__MINGW32__] (find_esc_head): Likewise.
6077 [__MINGW32__] (find_esc_terminator): Likewise.
6078 [__MINGW32__] (eat_esc_sequence): Likewise.
6079 [__MINGW32__] (mingw_ansi_fputs): New function that handles
6081 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
6084 2017-10-11 Richard Biener <rguenther@suse.de>
6086 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
6087 Properly call analyze_scalar_evolution with the loop of the stmt.
6089 2017-10-11 Richard Biener <rguenther@suse.de>
6091 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
6092 * tree-core.h (tree_base): Add chrec_var union member.
6093 * tree.h (CHREC_VAR): Remove.
6094 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
6095 * tree-chrec.h (build_polynomial_chrec): Adjust.
6096 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
6097 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
6099 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
6101 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
6102 * match.pd: ... here.
6103 ((T) X == (T) Y): Relax condition.
6105 2017-10-11 Bin Cheng <bin.cheng@arm.com>
6107 PR tree-optimization/82472
6108 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
6110 (break_alias_scc_partitions): Update postorder number.
6112 2017-10-11 Martin Liska <mliska@suse.cz>
6115 * opts.c (parse_no_sanitize_attribute): Do not use error_value
6117 * opts.h (parse_no_sanitize_attribute): Remove last argument.
6119 2017-10-11 Martin Liska <mliska@suse.cz>
6121 * print-rtl.c (print_insn): Move declaration of idbuf
6122 to same scope as name.
6124 2017-10-11 Martin Liska <mliska@suse.cz>
6129 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
6132 2017-10-11 Martin Liska <mliska@suse.cz>
6135 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
6138 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
6140 * config.gcc (default_gnu_indirect_function): Default to yes for
6141 arm*-*-linux* with glibc.
6143 2017-10-11 Richard Biener <rguenther@suse.de>
6145 * tree-scalar-evolution.c (get_scalar_evolution): Handle
6146 default-defs and types we do not want to analyze.
6147 (interpret_loop_phi): Replace unreachable code with an assert.
6148 (compute_scalar_evolution_in_loop): Remove and inline ...
6149 (analyze_scalar_evolution_1): ... here, replacing condition with
6150 what makes the intent clearer. Remove handling of cases
6151 get_scalar_evolution now handles.
6153 2017-10-10 Jim Wilson <wilson@tuliptree.org>
6155 PR rtl-optimization/81434
6156 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
6157 comment for main loop. In sched_group_found if, also add checks for
6158 pass and min_cost_group.
6160 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
6162 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
6163 (rs6000_insn_cost): New function.
6164 * config/rs6000/rs6000.md (cost): New attribute.
6166 2017-10-10 Jakub Jelinek <jakub@redhat.com>
6167 H.J. Lu <hongjiu.lu@intel.com>
6171 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
6172 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
6173 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
6174 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
6175 to OPTION_MASK_ISA_AVX512VL - builtins that have both
6176 OPTION_MASK_ISA_MMX and some other bit set require both
6177 mmx and the ISAs without the mmx bit.
6178 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
6179 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
6180 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
6181 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
6182 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
6183 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
6184 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
6185 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
6186 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
6187 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
6188 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
6189 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
6190 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
6191 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
6192 Add OPTION_MASK_ISA_MMX.
6194 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
6196 * config.gcc (armv7*-*-freebsd*): New target.
6197 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
6199 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
6201 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
6202 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
6205 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
6207 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
6208 a template parameter.
6209 (WIDE_INT_REF_FOR): Update accordingly.
6210 * tree.h (wi::int_traits <const_tree>): Delete.
6211 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
6212 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
6213 (wi::tree_to_wide_ref): New typedef.
6214 (wi::to_wide): New function.
6215 * calls.c (get_size_range): Use wi::to_wide when operating on
6217 * cgraph.c (cgraph_node::create_thunk): Likewise.
6218 * config/i386/i386.c (ix86_data_alignment): Likewise.
6219 (ix86_local_alignment): Likewise.
6220 * dbxout.c (stabstr_O): Likewise.
6221 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
6222 * expr.c (const_vector_from_tree): Likewise.
6223 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
6224 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
6225 (fold_negate_expr_1, int_const_binop_1, const_binop)
6226 (fold_convert_const_int_from_real, optimize_bit_field_compare)
6227 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
6228 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
6229 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
6230 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
6231 (fold_not_const, round_up_loc): Likewise.
6232 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
6233 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
6234 (alloca_call_type): Likewise.
6235 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
6236 * godump.c (go_output_typedef): Likewise.
6237 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
6238 * internal-fn.c (get_min_precision): Likewise.
6239 * ipa-cp.c (ipcp_store_vr_results): Likewise.
6240 * ipa-polymorphic-call.c
6241 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
6242 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
6243 (ipa_modify_call_arguments): Likewise.
6244 * match.pd: Likewise.
6245 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
6246 * print-tree.c (print_node_brief, print_node): Likewise.
6247 * stmt.c (expand_case): Likewise.
6248 * stor-layout.c (layout_type): Likewise.
6249 * tree-affine.c (tree_to_aff_combination): Likewise.
6250 * tree-cfg.c (group_case_labels_stmt): Likewise.
6251 * tree-data-ref.c (dr_analyze_indices): Likewise.
6252 (prune_runtime_alias_test_list): Likewise.
6253 * tree-dump.c (dequeue_and_dump): Likewise.
6254 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
6255 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
6256 * tree-pretty-print.c (dump_generic_node): Likewise.
6257 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
6258 (simple_iv_with_niters): Likewise.
6259 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
6260 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
6261 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
6262 * tree-ssa-loop-niter.c (split_to_var_and_offset)
6263 (refine_value_range_using_guard, number_of_iterations_ne_max)
6264 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
6265 (get_cst_init_from_scev, record_nonwrapping_iv)
6266 (scev_var_range_cant_overflow): Likewise.
6267 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
6268 * tree-ssa-pre.c (compute_avail): Likewise.
6269 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
6270 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
6271 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
6272 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
6273 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
6274 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
6275 (dump_case_nodes, try_switch_expansion): Likewise.
6276 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
6277 (vect_do_peeling): Likewise.
6278 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
6279 * tree-vect-stmts.c (vectorizable_load): Likewise.
6280 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
6281 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
6282 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
6283 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
6284 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
6285 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
6286 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
6287 (evrp_dom_walker::before_dom_children): Likewise.
6288 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
6289 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
6290 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
6291 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
6292 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
6293 (get_range_pos_neg): Likewise.
6294 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
6295 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
6296 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
6297 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
6298 * config/avr/avr.c (avr_fold_builtin): Likewise.
6299 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
6300 * config/msp430/msp430.c (msp430_attr): Likewise.
6301 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
6302 * config/powerpcspe/powerpcspe-c.c
6303 (altivec_resolve_overloaded_builtin): Likewise.
6304 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
6305 (rs6000_expand_ternop_builtin): Likewise.
6306 * config/rs6000/rs6000-c.c
6307 (altivec_resolve_overloaded_builtin): Likewise.
6308 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
6309 (rs6000_expand_ternop_builtin): Likewise.
6310 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
6312 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6314 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
6315 when copying loop nest with only one inner loop.
6317 2017-10-10 Richard Biener <rguenther@suse.de>
6319 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
6320 blocks if SCEV is active.
6321 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
6323 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
6324 (scev_initialize): Assert we are not yet initialized.
6326 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6328 * tree-loop-distribution.c (generate_loops_for_partition): Remove
6329 inner loop's exit stmt by making it always exit the loop, otherwise
6330 we would generate an infinite empty loop.
6332 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6334 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
6335 renaming variables in new preheader if it's deleted.
6337 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6339 * tree-loop-distribution.c (struct partition): Remove unused field
6340 loops of the structure.
6341 (partition_alloc, partition_free): Ditto.
6342 (build_rdg_partition_for_vertex): Ditto.
6344 2017-10-09 Jeff Law <law@redhat.com>
6346 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
6347 return type to match prototype and documentation.
6349 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6351 * config/rs6000/rs6000.c (processor_costs): Move to ...
6352 * config/rs6000/rs6000.h: ... here.
6353 (rs6000_cost): Declare.
6355 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
6357 * except.c (setjmp_fn): New global variable.
6358 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
6359 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
6360 if DONT_USE_BUILTIN_SETJMP is defined.
6362 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6364 * target.def (insn_cost): New hook.
6365 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
6366 * doc/tm.texi: Regenerate.
6367 * rtlanal.c (insn_cost): Use the new hook.
6369 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6371 * combine.c (combine_validate_cost): Compute the new insn_cost,
6372 not just pattern_cost.
6373 (try_combine): Adjust comment.
6375 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6377 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
6379 * combine.c (uid_insn_cost): Adjust comment.
6380 (combine_validate_cost): Adjust comment. Use pattern_cost instead
6382 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
6383 * dse.c (find_shift_sequence): Ditto.
6384 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
6385 (bb_valid_for_noce_process_p): Use pattern_cost.
6386 * rtl.h (insn_rtx_cost): Delete.
6387 (pattern_cost): New prototype.
6388 (insn_cost): New prototype.
6389 * rtlanal.c (insn_rtx_cost): Rename to...
6390 (pattern_cost): ... this.
6393 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
6395 * config/i386/i386.md (*jcc_2): Remove insn pattern.
6396 (*jcc<mode>_0_r_i387): Ditto.
6397 (*jccxf_r_i387): Ditto.
6398 (*jcc<mode>_r_i387): Ditto.
6399 (*jccu<mode>_r_i387): Ditto.
6400 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
6401 (*jcc): Rename from *jcc_1.
6403 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6405 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
6406 deferred rescans after the lvx/stvx recombination pre-pass.
6408 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
6410 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
6411 memory operation instruction support.
6412 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
6413 (rs6000-ibm-aix[789]*): Likewise.
6414 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
6415 Document new functions.
6417 2017-10-09 Richard Biener <rguenther@suse.de>
6419 PR tree-optimization/82397
6420 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
6421 equality only for semantically equal trees.
6423 2017-10-09 Richard Biener <rguenther@suse.de>
6425 PR tree-optimization/82449
6426 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
6427 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
6428 allow constant addresses.
6429 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
6432 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6434 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
6437 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6440 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
6443 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6446 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
6448 2017-10-09 Jakub Jelinek <jakub@redhat.com>
6451 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
6452 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
6454 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
6456 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
6457 (WI_BINARY_PREDICATE_RESULT): Likewise.
6458 (wi::binary_traits::operator_result): New type.
6459 (wi::binary_traits::predicate_result): Likewise.
6460 (generic_wide_int::operator~, unary generic_wide_int::operator-)
6461 (generic_wide_int::operator==, generic_wide_int::operator!=)
6462 (generic_wide_int::operator&, generic_wide_int::and_not)
6463 (generic_wide_int::operator|, generic_wide_int::or_not)
6464 (generic_wide_int::operator^, generic_wide_int::operator+
6465 (binary generic_wide_int::operator-, generic_wide_int::operator*):
6467 (operator~, unary operator-, operator==, operator!=, operator&)
6468 (operator|, operator^, operator+, binary operator-, operator*): New
6470 * expr.c (get_inner_reference): Use wi::bit_and_not.
6471 * fold-const.c (fold_binary_loc): Likewise.
6472 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
6473 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
6474 (bit_value_binop): Likewise.
6475 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
6476 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
6477 (extract_range_from_binary_expr_1): Likewise.
6478 (masked_increment): Likewise.
6479 (simplify_bit_ops_using_ranges): Likewise.
6481 2017-10-09 Martin Jambor <mjambor@suse.cz>
6484 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
6485 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
6486 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
6488 (hsa_fixup_mov_insn_type): New function.
6489 (hsa_op_with_type::get_in_type): Use it.
6490 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
6491 immediates in an assert.
6492 (hsa_op_with_type::extend_int_to_32bit): New method.
6493 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
6494 types. Convert to dest type if necessary.
6495 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
6496 (reg_for_gimple_ssa): Pass false as min32int to
6497 hsa_type_for_scalar_tree_type.
6498 (gen_hsa_addr): Fixup type when creating addresable temporary.
6499 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
6500 (gen_hsa_unary_operation): Extend operands and convert to dest type if
6501 necessary. Call hsa_fixup_mov_insn_type.
6502 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
6503 extend operands and convert to dest type if necessary.
6504 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
6505 to dest type if necessary.
6506 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
6507 if conversion nt necessary and size matches.
6508 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
6509 to dest type if necessary.
6510 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
6511 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
6513 (gen_hsa_clrsb): Likewise.
6514 (gen_hsa_ffs): Likewise.
6515 (gen_hsa_divmod): Extend operands and convert to dest type if
6517 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
6519 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
6521 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
6522 Remove empty default arguments. Use a brace block as output
6524 (conditional return): Ditto.
6526 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
6527 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
6529 (group_ending_nop): Ditto.
6530 (doloop_end): Ditto.
6531 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
6532 (splitters for those): Ditto.
6534 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
6536 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
6537 a conditional jump (and the compare for it) so that pc_rtx is the
6539 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
6540 for the deleted and renamed ctr<mode>_internal[234] patterns.
6541 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
6542 Delete second conditional return pattern.
6543 (ctr<mode>_internal2): Delete this second bdnz pattern.
6544 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
6545 (ctr<mode>_internal4): Delete this second bdz pattern.
6547 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
6549 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
6550 (always_initialized_rtx_for_ssa_name_p): New predicate.
6551 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
6552 (finish_out_of_ssa): Free new field of SA.
6553 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
6554 * tree-ssa-coalesce.c: Include tree-ssa.h.
6555 (get_parm_default_def_partitions): Remove extern keyword.
6556 (get_undefined_value_partitions): New function.
6557 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
6558 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
6560 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
6562 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
6564 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
6566 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
6568 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
6569 for targets that preffer 128bit.
6571 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
6573 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
6575 2017-10-08 Olivier Hainque <hainque@adacore.com>
6577 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
6578 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
6580 (thumb_set_return_address): Likewise.
6582 2017-10-08 Olivier Hainque <hainque@adacore.com>
6584 * common/config/arm/arm-common.c (arm_except_unwind_info):
6585 Handle DWARF2_UNWIND_INFO.
6587 2017-10-07 Michael Collison <michael.collison@arm.com>
6589 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
6592 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
6594 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
6595 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
6596 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
6597 defined, force the creation of a new block for a dispatch label.
6599 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
6601 * invoke.texi (Wsuggest-attribute=cold): Document.
6602 * common.opt (Wsuggest-attribute=cold): New
6603 * ipa-pure-const.c (warn_function_cold): New function.
6604 * predict.c (compute_function_frequency): Use it.
6605 * predict.h (warn_function_cold): Declare.
6607 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
6609 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
6612 2017-10-06 Martin Liska <mliska@suse.cz>
6614 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
6615 keyword for member functions.
6616 (struct sanopt_tree_couple): New struct.
6617 (struct sanopt_tree_couple_hash): New function.
6618 (struct sanopt_ctx): Add new hash_map.
6619 (has_dominating_ubsan_ptr_check): New function.
6620 (record_ubsan_ptr_check_stmt): Likewise.
6621 (maybe_optimize_ubsan_ptr_ifn): Likewise.
6622 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
6623 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
6625 2017-10-06 Sudakshina Das <sudi.das@arm.com>
6628 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
6629 aarch64_simd_valid_immediate on CONST_VECTORs.
6630 (aarch64_reg_or_bic_imm): Likewise.
6632 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
6634 PR rtl-optimization/82396
6635 * haifa-sched.c (ready_sort_real): Disable qsort checking.
6637 2017-10-06 Sebastian Pop <sebpop@gmail.com>
6639 * graphite-dependences.c (scop_get_reads): Move code to...
6640 (scop_get_must_writes): Move code to...
6641 (scop_get_may_writes): Move code to...
6642 (scop_get_reads_and_writes): ... here.
6643 (scop_get_dependences): Call scop_get_reads_and_writes.
6645 2017-10-06 Jakub Jelinek <jakub@redhat.com>
6647 PR tree-optimization/82434
6648 * fold-const.h (can_native_encode_type_p,
6649 can_native_encode_string_p): Remove.
6650 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
6651 don't encode anything, just return what would be otherwise returned.
6652 (native_encode_fixed, native_encode_complex, native_encode_vector):
6654 (native_encode_string): Likewise. Inline by hand
6655 can_native_encode_string_p.
6656 (can_native_encode_type_p): Remove.
6657 (can_native_encode_string_p): Remove.
6658 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
6659 STRING_CSTs using can_native_encode_string_p, test all
6660 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
6661 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
6662 argument from native_encode_expr.
6663 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
6664 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
6667 2017-10-06 Richard Biener <rguenther@suse.de>
6669 PR tree-optimization/82397
6670 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
6671 operand_equal_p but rely on data_ref_compare_tree for detecting
6673 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
6674 to match up with dr_group_sort_cmp.
6676 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6679 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
6681 * config/s390/s390-builtin-types.def: Regenerate.
6683 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6686 * config/s390/s390-builtin-types.def: Regenerate.
6687 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
6688 Change flag from B_VXE to B_VX.
6689 (s390_vec_min_dbl): Remove B_VXE flag.
6691 2017-10-06 Richard Biener <rguenther@suse.de>
6693 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
6694 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
6695 translate_isl_ast_to_gimple::is_valid_rename,
6696 translate_isl_ast_to_gimple::get_rename,
6697 translate_isl_ast_to_gimple::get_def_bb_for_const,
6698 translate_isl_ast_to_gimple::get_new_name,
6699 translate_isl_ast_to_gimple::collect_all_ssa_names,
6700 translate_isl_ast_to_gimple::copy_loop_phi_args,
6701 translate_isl_ast_to_gimple::collect_all_ssa_names,
6702 translate_isl_ast_to_gimple::copy_loop_phi_args,
6703 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
6704 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
6705 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
6706 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
6707 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
6708 translate_isl_ast_to_gimple::copy_cond_phi_args,
6709 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
6710 translate_isl_ast_to_gimple::edge_for_new_close_phis,
6711 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
6712 translate_isl_ast_to_gimple::rename_uses,
6713 translate_isl_ast_to_gimple::rename_all_uses): Remove.
6714 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
6715 (set_rename_for_each_def): Likewise.
6716 (graphite_copy_stmts_from_block): Handle debug stmt resetting
6717 here. Handle rewriting SCEV analyzable uses here.
6718 (copy_bb_and_scalar_dependences): Generate code for PHI
6720 (graphite_regenerate_ast_isl): Adjust.
6721 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
6722 (add_write, add_read): New functions.
6723 (build_cross_bb_scalars_def): Use it and simplify.
6724 (build_cross_bb_scalars_use): Likewise.
6725 (graphite_find_cross_bb_scalar_vars): Inline into...
6726 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
6727 simulating out-of-SSA. Compute liveout and add dependencies.
6728 (build_scops): Force an empty entry block.
6729 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
6731 (sese_build_liveouts): Declare.
6732 (sese_trivially_empty_bb_p): Likewise.
6733 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
6734 compute liveout and debug_liveout.
6735 (sese_bad_liveouts_use): Remove.
6736 (sese_reset_debug_liveouts_bb): Likewise.
6737 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
6738 (sese_build_liveouts): Build liveout and debug_liveout and store
6740 (new_sese_info): Adjust.
6741 (free_sese_info): Likewise.
6742 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
6743 do not build liveout here.
6744 (move_sese_in_condition): Adjust region entry.
6745 (scev_analyzable_p): Match up with chrec_apply requirements.
6746 (sese_trivially_empty_bb_p): New.
6747 * tree-into-ssa.c (get_reaching_def): Properly support generating
6748 default-defs for incremental rewrite of anonymous names.
6750 2017-10-06 Richard Biener <rguenther@suse.de>
6752 * graphite-sese-to-poly.c (extract_affine): For casts increasing
6753 precision do not perform modulo reduction.
6755 2017-10-06 Richard Biener <rguenther@suse.de>
6757 PR tree-optimization/82436
6758 * tree-vect-slp.c (vect_supported_load_permutation_p): More
6759 conservatively choose the vectorization factor when checking
6760 whether we can perform the required load permutation.
6761 (vect_transform_slp_perm_load): Assert when we may not fail.
6763 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
6765 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
6766 message for incompatible -msdata=* and -mcall-* options.
6768 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6770 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
6771 rate for post-reload scheduling.
6773 2017-10-05 Tamar Christina <tamar.christina@arm.com>
6775 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
6777 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6779 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
6780 to improve monte carlo in scimark.
6782 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6784 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
6785 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
6786 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
6787 pentium4_cost, nocona_cost): Set reassociation width to 1.
6788 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
6789 width to 2 for fp operations and 1 otherwise.
6790 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
6792 (atom_cost): Set reassociation width to 2.
6793 (slm_cost, generic_cost): Set fp reassociation width
6794 to 2 and 1 otherwise.
6795 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
6796 (core_cost): Set fp reassociation width to 4 and vector to 2.
6797 (ix86_reassociation_width): Rewrite using cost table; special case
6798 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
6799 and TARGET_AVX128_OPTIMAL.
6800 * config/i386/i386.h (processor_costs): Add
6801 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
6802 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
6803 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
6804 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
6805 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
6806 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
6808 2017-10-05 Nathan Sidwell <nathan@acm.org>
6810 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
6812 2017-10-05 Tamar Christina <tamar.christina@arm.com>
6814 * config/arm/arm.c (arm_test_fpu_data): New.
6815 (arm_run_selftests): Call arm_test_fpu_data.
6817 2017-10-04 Nathan Sidwell <nathan@acm.org>
6819 * toplev.c (toplev::main): Remove excess parens on pretty_printer
6821 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
6823 2017-10-04 Sudakshina Das <sudi.das@arm.com>
6825 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
6826 check type for aarch64_simd_valid_immediate.
6827 (aarch64_output_simd_mov_immediate): Update prototype.
6828 (aarch64_simd_valid_immediate): Update prototype.
6829 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
6830 support for ORR-immediate.
6831 (and<mode>3): modified pattern to add support for BIC-immediate.
6832 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
6833 now checks for valid immediate for BIC and ORR based on new enum
6835 (aarch64_output_simd_mov_immediate): Function now used to output
6836 BIC/ORR imm as well based on new enum argument.
6837 * config/aarch64/constraints.md (Do): New vector immediate constraint.
6839 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
6840 (aarch64_reg_or_bic_imm): Likewise.
6842 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6844 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
6845 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
6847 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
6851 PR rtl-optimization/82396
6852 * haifa-sched.c (autopref_multipass_init): Simplify
6854 (autopref_rank_data): Simplify sort order.
6855 * sched-int.h (autopref_multipass_data_): Remove
6856 multi_mem_insn_p, min_offset and max_offset.
6858 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6860 * doc/sourcebuild.texi: Document vect_peeling_profitable.
6862 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6864 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
6867 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6869 * doc/sourcebuild.texi: Document vect_long_mult.
6871 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
6873 PR tree-optimization/82413
6874 * fold-const.c (build_range_check): Use widest_int when comparing
6875 the maximum ETYPE value with HIGH.
6877 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
6879 PR rtl-optimization/82396
6880 * haifa-sched.c (autopref_multipass_init): Simplify
6882 (autopref_rank_data): Simplify sort order.
6883 * sched-int.h (autopref_multipass_data_): Remove
6884 multi_mem_insn_p, min_offset and max_offset.
6886 2017-10-04 Jakub Jelinek <jakub@redhat.com>
6888 PR tree-optimization/82381
6889 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
6890 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
6893 PR tree-optimization/82374
6894 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
6895 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
6896 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
6897 current_function_decl to the new decl.
6899 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
6901 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
6902 helper macro for IEEE float128 hardware built-in functions.
6903 (SQRTF128_ODD): Add built-in functions with the round-to-odd
6905 (TRUNCF128_ODD): Likewise.
6906 (ADDF128_ODD): Likewise.
6907 (SUBF128_ODD): Likewise.
6908 (MULF128_ODD): Likewise.
6909 (DIVF128_ODD): Likewise.
6910 (FMAF128_ODD): Likewise.
6911 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
6912 UNSPEC_TRUNC_ROUND_TO_ODD.
6913 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
6914 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
6915 floating point round to odd instructions.
6916 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
6917 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
6918 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
6919 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
6920 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
6921 (trunc<mode>sf2_hw): Change the truncate with round to odd
6922 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
6923 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
6924 to odd hardware instructions.
6925 (sub<mode>3_odd): Likewise.
6926 (mul<mode>3_odd): Likewise.
6927 (div<mode>3_odd): Likewise.
6928 (sqrt<mode>2_odd): Likewise.
6929 (fma<mode>4_odd): Likewise.
6930 (fms<mode>4_odd): Likewise.
6931 (nfma<mode>4_odd): Likewise.
6932 (nfms<mode>4_odd): Likewise.
6933 (trunc<mode>df2_odd): Change the truncate with round to odd
6934 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
6936 * doc/extend.texi (PowerPC built-in functions): Update documentation
6937 for existing IEEE float128-bit built-in functions. Add built-in
6938 functions that generate the IEEE 128-bit floating point round to
6941 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
6943 PR rtl-optimization/77729
6944 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
6945 to (X&(C1&~C2))|C2 transformations.
6947 2017-10-03 Martin Jambor <mjambor@suse.cz>
6949 PR tree-optimization/82363
6950 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
6951 mismatch, mark lacc written regardless of racc.
6953 2017-10-03 Jakub Jelinek <jakub@redhat.com>
6955 PR tree-optimization/82381
6956 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
6957 stmt_to_insert nor wheather SSA_NAMEs are default defs.
6958 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
6959 fallthrough into reassoc_stmt_dominates_stmt_p.
6962 * combine.c (combine_instructions): Don't combine in unreachable
6965 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
6968 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
6969 function to not use the have_cpu variable. Do not set cpu_index,
6970 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
6972 (rs6000_valid_attribute_p): Remove duplicate initializations of
6973 old_optimize and func_optimize.
6974 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
6975 (rs6000_activate_target_options): Make global.
6976 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
6979 2017-10-02 Jakub Jelinek <jakub@redhat.com>
6981 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
6982 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
6983 Set *poffset to 0 and *psize and *pmax_size to -1 if
6984 *poffset + *psize overflows in HOST_WIDE_INT.
6986 PR tree-optimization/82387
6987 PR tree-optimization/82388
6988 PR tree-optimization/82389
6989 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
6990 instead of live_bytes non-NULL.
6992 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
6995 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
6998 2017-10-02 Richard Biener <rguenther@suse.de>
7000 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
7001 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
7002 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
7004 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
7006 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
7007 requested precision matches the type's.
7008 * calls.c (alloc_max_size): Calculate the new candidate size as
7009 a widest_int and use wi::to_widest when comparing it with the
7010 current candidate size.
7011 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
7012 zero rather than integer_zero_node.
7013 * match.pd: Check for a no-op conversion before using wi::add
7014 rather than after. Use tree_to_uhwi when summing small shift
7015 counts into an unsigned int.
7017 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
7018 Alan Hayward <alan.hayward@arm.com>
7019 David Sherwood <david.sherwood@arm.com>
7022 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
7023 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
7024 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
7025 POINTER_AND_FP_REGS.
7027 2017-10-02 Richard Biener <rguenther@suse.de>
7029 PR tree-optimization/82355
7030 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
7031 a mapping for the enclosing loop but avoid generating one for
7033 (copy_bb_and_scalar_dependences): Remove premature codegen
7034 error on PHIs in blocks duplicated into multiple places.
7035 * graphite-scop-detection.c
7036 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
7037 in the region use it as loop and nest to analyze the DR in.
7038 (try_generate_gimple_bb): Likewise.
7039 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
7040 (add_loop_constraints): For blocks in a loop not in the region
7041 create a dimension with a single iteration.
7042 * sese.h (gbb_loop_at_index): Remove assert.
7044 2017-10-01 Kevin Buettner <kevinb@redhat.com>
7046 * omp-expand.c (adjust_context_scope): New function.
7047 (expand_parallel_call): Call adjust_context_scope.
7049 2017-10-01 Jeff Law <law@redhat.com>
7051 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
7052 dom_opt_dom_walker class with direct access to private members.
7053 Add comments. Call test_for_singularity.
7054 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
7055 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
7056 m_dummy_cond anymore.
7057 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
7059 (pass_dominator:execute): Build the dummy_cond here and pass it
7060 to the dom_opt_dom_walker ctor.
7061 (test_for_singularity): New function.
7063 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
7064 Maya Rashish <coypu@sdf.org>
7066 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
7067 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
7068 (alpha*-*-netbsd*) Use nbsd_tm_file.
7069 (arm*-*-netbsdelf*) Likewise.
7070 (i[34567]86-*-netbsdelf*) Likewise.
7071 (x86_64-*-netbsd*) Likewise.
7072 (mips*-*-netbsd*) Likewise.
7073 (powerpc-*-netbsd*) Likewise.
7074 (sh*-*-netbsd*) Likewise.
7075 (sparc-*-netbsdelf*) Likewise.
7076 (sparc64-*-netbsd*) Likewise.
7077 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
7079 (vax-*-netbsdelf*) Likewise.
7080 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
7081 (UINT_FAST8_TYPE) Likewise.
7082 (INT_FAST16_TYPE) Check CHAR_FAST16.
7083 (UINT_FAST16_TYPE) Likewise.
7085 2017-09-30 Jakub Jelinek <jakub@redhat.com>
7088 * config/i386/i386.md
7089 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
7090 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
7091 *divmodsi4_zext_2): New define_insn_and_split.
7092 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
7093 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
7094 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
7095 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
7096 New define_insn_and_split.
7097 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
7098 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
7099 operands[1] having DImode when mode is SImode.
7101 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
7102 always SImode for DIV and MOD in REG_EQUAL notes.
7104 2017-09-29 Yury Gribov <tetra2005@gmail.com>
7107 * match.pd: Fix handling of NaNs in pattern.
7109 2017-09-29 Jeff Law <law@redhat.com>
7111 * sbitmap.c (bitmap_bit_in_range_p): New function.
7112 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
7113 * tree-ssa-dse.c (live_bytes_read): New function.
7114 (dse_classify_store): Ignore reads of dead bytes.
7116 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
7117 typos and whitespace errors.
7118 * config/i386/predicates.md (address_no_seg_operand): Likewise.
7119 * config/s390/s390.c (s390_emit_prologue): Likewise.
7121 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
7124 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
7125 with a symbol for LRA.
7127 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
7129 PR rtl-optimization/82338
7130 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
7132 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
7134 * genmodes.c (calc_wider_mode): Suppress qsort macro.
7135 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
7136 (qsort_chk): Declare.
7137 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
7138 (qsort_chk): New function.
7140 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7142 PR tree-optimization/82337
7143 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
7144 phi definition if the PHI result appears in an abnormal PHI.
7145 (find_basis_for_base_expr): Don't record a basis if the LHS of the
7146 basis appears in an abnormal PHI.
7148 2017-09-29 Richard Biener <rguenther@suse.de>
7150 * graphite-isl-ast-to-gimple.c
7151 (translate_isl_ast_to_gimple::set_codegen_error): New function.
7152 (binary_op_to_tree): Use it.
7153 (get_rename_from_scev): Likewise.
7154 (copy_loop_phi_nodes): Likewise.
7155 (copy_bb_and_scalar_dependences): Likewise.
7156 (translate_pending_phi_nodes): Likewise.
7158 2017-09-29 Jakub Jelinek <jakub@redhat.com>
7161 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
7162 for movabsq $(i32 << shift), r64.
7164 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
7166 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
7167 index when encoding %esp as %rsp to avoid 0x67 prefix.
7169 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
7171 * config/i386/i386.md (*movsf_internal, *movdf_internal):
7172 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
7174 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
7176 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
7177 Extensions with more than 16 double VFP registers.
7178 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
7179 to_clear_mask and all code related to it. Replace the remaining
7180 entry by a sbitmap and adapt code accordingly.
7182 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
7184 * brig-builtins.def: Change pure attributes to const.
7186 2017-09-28 Joseph Myers <joseph@codesourcery.com>
7188 * config.gcc (default_gnu_indirect_function): Default to yes for
7189 sparc*-*-linux* with glibc.
7191 2017-09-28 Joseph Myers <joseph@codesourcery.com>
7193 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
7194 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
7195 when creating .init_array and .fini_array sections with priority
7198 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
7201 * config/aarch64/aarch64.c
7202 (aarch64_builtin_support_vector_misalignment): Always return false
7203 when misalignment is unknown.
7205 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
7207 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
7208 this function to return false if the definition used by the swap
7209 instruction is artificial, or if the memory address from which the
7210 constant value is loaded is not represented by a base address held
7211 in a register or if the base address register is a frame or stack
7212 pointer. Additionally, return false if the base address of the
7213 loaded constant is a SYMBOL_REF but is not considered to be a
7215 (replace_swapped_load_constant): New function.
7216 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
7217 loaded constant vector with a load of a swapped constant vector.
7219 2017-09-27 Carl Love <cel@us.ibm.com>
7221 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
7222 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
7223 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
7226 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
7228 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
7229 first, always call autopref_rank_data otherwise.
7231 2017-09-27 Richard Biener <rguenther@suse.de>
7233 * graphite-scop-detection.c (find_scop_parameters): Move
7234 loop bound handling ...
7235 (gather_bbs::before_dom_children): ... here, avoiding the need
7236 to build scop_info->loop_nest.
7237 (record_loop_in_sese): Remove.
7238 * sese.h (sese_info_t::loop_nest): Remove.
7239 * sese.c (new_sese_info): Do not allocate loop_nest.
7240 (free_sese_info): Do not free loop_nest.
7242 2017-09-27 Jakub Jelinek <jakub@redhat.com>
7245 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
7246 lhs from calls if the lhs has addressable type.
7248 2017-09-27 Richard Biener <rguenther@suse.de>
7250 * graphite.h (scop::max_alias_set): New member.
7251 * graphite-scop-detection.c: Remove references to non-existing
7252 --param in comments.
7253 (build_alias_sets): Record the maximum alias set used for drs.
7254 (build_scops): Support zero as unlimited for
7255 --param graphite-max-arrays-per-scop.
7256 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
7258 (build_poly_sr_1): ... here. Compute alias set based on the
7259 maximum alias set used for drs rather than
7260 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
7262 2017-09-27 Richard Biener <rguenther@suse.de>
7264 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
7265 --param loop-block-tile-size=0 to disable tiling.
7267 2017-09-27 Richard Biener <rguenther@suse.de>
7269 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
7270 (graphite-max-nb-scop-params): Document special value zero.
7271 * domwalk.h (dom_walker::STOP): New symbolical constant.
7272 (dom_walker::dom_walker): Add optional parameter for bb to
7274 (dom_walker::~dom_walker): Declare.
7275 (dom_walker::before_dom_children): Document STOP return value.
7276 (dom_walker::m_user_bb_to_rpo): New member.
7277 (dom_walker::m_bb_to_rpo): Likewise.
7278 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
7279 mapping here if not provided by the user.
7280 (dom_walker::~dom_walker): Free bb to RPO mapping if not
7281 provided by the user.
7282 (dom_walker::STOP): Define.
7283 (dom_walker::walk): Do not compute bb to RPO mapping here.
7284 Support STOP return value from before_dom_children to stop
7286 * graphite-optimize-isl.c (optimize_isl): If the schedule
7287 is the same still generate code if -fgraphite-identity
7288 or -floop-parallelize-all are given.
7289 * graphite-scop-detection.c: Include cfganal.h.
7290 (gather_bbs::gather_bbs): Get and pass through bb to RPO
7292 (gather_bbs::before_dom_children): Return STOP for BBs
7294 (build_scops): Compute bb to RPO mapping and pass it to
7295 the domwalk. Treat --param graphite-max-nb-scop-params=0
7296 as not limiting the number of params.
7297 * graphite.c (graphite_initialize): Remove limit on the
7298 number of basic-blocks in a function.
7299 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
7300 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
7301 default value of 10.
7303 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7305 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
7306 Adjust code to eliminate needing to do the shift right 32-bits
7307 operation after XSCVDPSPN.
7309 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
7311 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
7312 ((X / Y) != 0 -> X >= Y): Likewise.
7314 2017-09-26 Carl Love <cel@us.ibm.com>
7316 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
7317 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
7318 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
7319 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
7320 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
7321 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
7322 definitions and overloading.
7323 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
7324 statement for P9V_BUILTIN_XST_LEN_R.
7325 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
7326 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
7327 define_expand and define_insn for the instructions and builtins.
7328 * doc/extend.texi: Update the built-in documentation file for the new
7330 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
7331 define_insn for the instructions
7333 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
7336 * gcc/config/netbsd-protos.h: New file.
7337 * gcc/config/netbsd.c: New file.
7338 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
7339 * gcc/config/t-netbsd: New file.
7340 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
7341 (tmake_file) Add t-netbsd.
7342 (extra_objs) Add netbsd.o.
7344 2017-09-26 Janus Weil <janus@gcc.gnu.org>
7348 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
7350 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7352 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
7353 sign extension from a vector register to a GPR by doing a 32-bit
7354 direct move and then an EXTSW.
7355 (extendsi<mode>2 splitter): Likewise.
7356 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
7357 right or vector extract after doing XSCVDPSPN. Use
7358 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
7360 (movdi_from_sf_zero_ext): Likewise.
7361 (reload_gpr_from_vsxsf): Likewise.
7362 (p8_mfvsrd_4_disf): Delete, no longer used.
7363 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
7364 then needing to move the SFmode to a GPR to use the XSCVDPSP
7365 instruction instead of FRSP and XSCVDPSPN.
7366 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
7367 it is adjacent to the other XSCVSPDP insns.
7368 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
7369 SFmode to be in traditional Altivec registers.
7370 (vsx_xscvdpspn): Eliminate useless alternative constraint.
7371 (vsx_xscvspdpn): Likewise.
7372 (vsx_xscvspdpn_scalar): Likewise.
7374 2017-09-26 Martin Jambor <mjambor@suse.cz>
7376 * tree-sra.c (compare_access_positions): Put integral types first,
7377 stabilize sorting of integral types, remove conditions putting
7378 non-full-precision integers last.
7379 (sort_and_splice_var_accesses): Disable scalarization if a
7380 non-integert would be represented by a non-full-precision integer.
7382 2017-09-26 Joseph Myers <joseph@codesourcery.com>
7384 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
7385 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
7386 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
7387 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
7388 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
7389 conditionals inside the function instead of around it. Call
7390 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
7391 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
7393 2017-09-26 Richard Biener <rguenther@suse.de>
7395 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
7397 (scop_detection::build_scop_breadth): ... this. Removed.
7398 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
7399 (scop_detection::harmful_stmt_in_bb): Likewise.
7400 (scop_detection::graphite_can_represent_stmt): Likewise.
7401 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
7402 (scop_detection::can_represent_loop): Remove recursion, fold in ...
7403 (scop_detection::can_represent_loop_1): ... this. Removed.
7404 (scop_detection::harmful_loop_in_region): Simplify after inlining
7405 the above and remove more quadraticness.
7406 (build_scops): Adjust.
7407 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
7410 2017-09-26 Jakub Jelinek <jakub@redhat.com>
7413 * config/i386/i386.c (ix86_print_operand_address_as): Only test
7414 REGNO (base) == SP_REG if base is a REG.
7417 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
7418 if it is different SSA_NAME.
7419 (optimize_range_tests_cmp_bitwise): New function.
7420 (optimize_range_tests): Call it.
7422 2017-09-26 Richard Biener <rguenther@suse.de>
7424 PR tree-optimization/82321
7425 * graphite.c (canonicalize_loop_closed_ssa): Properly check
7426 for the def being inside the loop.
7428 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7430 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
7432 * config/s390/s390-builtins.def: Fix constraint on op4.
7434 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7436 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
7437 independent expanders.
7438 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
7439 ("vec_ordered", "vec_unordered"): New expanders.
7441 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7443 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
7446 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7448 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
7449 vec_unpacks_lo_v16qi.
7450 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
7452 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7454 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
7455 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
7456 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
7458 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7460 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
7462 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
7464 ("*vec_slb<mode>"): New insn pattern.
7465 ("vec_shr_<mode>"): New expander.
7466 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
7467 and force the shift count operand to V16QImode.
7468 ("vec_srb<mode>"): Set shift count mode to V16QI.
7470 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7472 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
7473 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
7474 ("vec_widen_smult_hi_<mode>"): New expander definitions.
7476 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
7479 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
7481 2017-09-26 Richard Biener <rguenther@suse.de>
7483 PR tree-optimization/82320
7484 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
7487 2017-09-25 Jeff Law <law@redhat.com>
7489 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
7490 prototype for new argument.
7491 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
7492 mostly extracted from rs6000_emit_allocate_stack.
7493 (rs6000_emit_probe_stack_range_stack_clash): New function.
7494 (rs6000_emit_allocate_stack): Call
7495 rs6000_emit_probe_stack_range_stack_clash as needed.
7496 (rs6000_emit_probe_stack_range): Add additional argument
7497 to call to gen_probe_stack_range{si,di}.
7498 (output_probe_stack_range): New.
7499 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
7500 (output_probe_stack_range_stack_clash): New.
7501 (rs6000_emit_prologue): Emit notes into dump file as requested.
7502 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
7503 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
7504 Add additional operand and pass it to output_probe_stack_range.
7506 2017-09-25 Bin Cheng <bin.cheng@arm.com>
7508 PR tree-optimization/82163
7509 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
7510 (checking_verify_loop_closed_ssa): New parameter.
7511 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
7512 (check_loop_closed_ssa_stmt): Delete.
7513 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
7514 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
7515 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
7518 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
7520 * brig-builtins.def: Treat HSAIL barrier builtins as
7521 setjmp/longjump style functions.
7523 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
7525 * target.def (constant_alignment): New hook.
7526 * defaults.h (CONSTANT_ALIGNMENT): Delete.
7527 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
7528 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
7529 * doc/tm.texi: Regenerate.
7530 * targhooks.h (default_constant_alignment): Declare.
7531 (constant_alignment_word_strings): Likewise.
7532 * targhooks.c (default_constant_alignment): New function.
7533 (constant_alignment_word_strings): Likewise.
7534 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
7535 instead of CONSTANT_ALIGNMENT.
7536 * varasm.c (align_variable, get_variable_align, build_constant_desc)
7537 (force_const_mem): Likewise.
7538 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
7539 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
7540 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
7541 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7542 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
7544 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
7545 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7546 constant_alignment_word_strings.
7547 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
7548 (CONSTANT_ALIGNMENT): Likewise.
7549 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7550 (arm_constant_alignment): New function.
7551 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
7552 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7553 constant_alignment_word_strings.
7554 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
7555 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7556 constant_alignment_word_strings.
7557 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
7558 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7559 (cris_constant_alignment): New function.
7560 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
7561 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7562 (epiphany_constant_alignment): New function.
7563 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
7564 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7565 constant_alignment_word_strings.
7566 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
7567 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7568 constant_alignment_word_strings.
7569 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
7570 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7571 constant_alignment_word_strings.
7572 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
7573 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
7574 * config/i386/i386.c (ix86_constant_alignment): Make static.
7575 Use the same interface as the target hook.
7576 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7577 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
7578 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7579 constant_alignment_word_strings.
7580 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
7581 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
7582 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7583 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
7584 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7585 constant_alignment_word_strings.
7586 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
7587 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7588 constant_alignment_word_strings.
7589 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
7590 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7591 constant_alignment_word_strings.
7592 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
7593 * config/microblaze/microblaze.c (microblaze_constant_alignment):
7595 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7596 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
7597 * config/mips/mips.c (mips_constant_alignment): New function.
7598 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7599 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
7600 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
7601 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7602 (mmix_constant_alignment): Make static. Use the same interface
7604 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
7605 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7606 constant_alignment_word_strings.
7607 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
7608 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7609 constant_alignment_word_strings.
7610 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
7611 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7612 constant_alignment_word_strings.
7613 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
7614 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7615 (rs6000_constant_alignment): New function.
7616 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
7617 * config/riscv/riscv.c (riscv_constant_alignment): New function.
7618 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7619 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
7620 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7621 (rs6000_constant_alignment): New function.
7622 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
7623 * config/s390/s390.c (s390_constant_alignment): New function.
7624 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7625 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
7626 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7627 constant_alignment_word_strings.
7628 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
7629 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7630 (sparc_constant_alignment): New function.
7631 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
7632 * config/spu/spu.c (spu_constant_alignment): New function.
7633 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7634 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
7635 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7636 constant_alignment_word_strings.
7637 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
7638 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7639 constant_alignment_word_strings.
7640 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
7641 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7642 constant_alignment_word_strings.
7643 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
7644 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7645 (visium_constant_alignment): New function.
7646 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
7647 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7648 (xtensa_constant_alignment): New function.
7649 * system.h (CONSTANT_ALIGNMENT): Poison.
7651 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
7653 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
7654 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
7655 (rs6000_builtin_valid_without_lhs): New helper function.
7656 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7657 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
7659 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
7661 * target.h (vec_perm_indices): Use unsigned short rather than
7663 (auto_vec_perm_indices): Likewise.
7664 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
7665 Use unsigned int rather than unsigned char.
7666 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
7668 2017-09-25 Richard Biener <rguenther@suse.de>
7670 * cfgloop.h (sort_sibling_loops): Declare.
7671 * cfgloop.c (sort_sibling_loops_cmp): New helper.
7672 (sort_sibling_loops): New function sorting the sibling loop list
7674 * graphite.c (graphite_transform_loops): Sort sibling loops.
7676 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
7678 * target.def (vec_perm_const_ok): Change sel parameter to
7680 * optabs-query.c (can_vec_perm_p): Update accordingly.
7681 * doc/tm.texi: Regenerate.
7682 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
7683 auto_vec_perm_indices and remove separate nelt field.
7684 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
7685 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
7686 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
7687 (aarch64_expand_vec_perm_const): Update accordingly.
7688 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
7689 to vec_perm_indices.
7690 * config/arm/arm.c (expand_vec_perm_d): Change perm to
7691 auto_vec_perm_indices and remove separate nelt field.
7692 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
7693 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
7694 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
7696 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
7697 to vec_perm_indices.
7698 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
7699 sel to vec_perm_indices.
7700 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
7701 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
7702 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
7704 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
7707 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
7710 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
7711 on the FUNCTION_DECL function context if it has a DIE that is a
7714 2017-09-25 Richard Biener <rguenther@suse.de>
7716 PR tree-optimization/82285
7717 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
7720 2017-09-25 Tom de Vries <tom@codesourcery.com>
7724 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
7727 2017-09-25 Richard Biener <rguenther@suse.de>
7729 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
7730 ISL errors other than isl_error_quota happen. Dump if the
7731 schedule is the same.
7732 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
7733 errors instead of aborting inside ISL.
7735 2017-09-25 Iain Sandoe <iain@codesourcery.com>
7738 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
7739 of libgcc_eh for m64.
7740 * config/i386/darwin64.h: Likewise.
7742 2017-09-25 Richard Biener <rguenther@suse.de>
7745 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
7746 attribute for incomplete types nor twice for complete ones.
7748 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
7751 * config/i386/i386.c (ix86_print_operand_address_as): Encode
7752 %esp as %rsp to avoid 0x67 prefix if there is no index or base
7755 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
7758 * config/i386/i386.opt (mprefer-avx256): Use
7759 ix86_target_flags variable.
7760 * config/i386/i386.c (ix86_target_string): Move
7761 -mprefer-avx256 to flag2_opts.
7763 2017-09-22 Jakub Jelinek <jakub@redhat.com>
7766 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
7767 and x != -1 | y != -1 into (x & y) != -1.
7769 2017-09-22 Steve Ellcey <sellcey@cavium.com>
7771 * config.gcc: Add new case statement to set
7772 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
7773 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
7774 s390x-*-linux* case statements. Added aarch64 to the list of
7775 supported architectures.
7777 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7779 PR tree-optimization/82289
7780 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
7781 STMT_VINFO_RELEVANT_P.
7783 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7784 Alan Hayward <alan.hayward@arm.com>
7785 David Sherwood <david.sherwood@arm.com>
7787 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
7788 for VR_RANGE only; don't allow VR_ANTI_RANGE.
7789 (extract_range_from_binary_expr_1): Don't call
7790 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
7792 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7793 Alan Hayward <alan.hayward@arm.com>
7794 David Sherwood <david.sherwood@arm.com>
7796 * target.def (preferred_vector_alignment): New hook.
7797 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
7799 * doc/tm.texi: Regenerate.
7800 * targhooks.h (default_preferred_vector_alignment): Declare.
7801 * targhooks.c (default_preferred_vector_alignment): New function.
7802 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
7804 (DR_TARGET_ALIGNMENT): New macro.
7805 (aligned_access_p): Update commentary.
7806 (vect_known_alignment_in_bytes): New function.
7807 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
7809 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
7810 Calculate the misalignment based on the target alignment rather than
7812 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
7813 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
7814 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
7815 the target alignment, rather than masking the element misalignment
7816 with the number of elements in a vector. Also use the target
7817 alignment when calculating the maximum number of peels.
7818 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
7819 instead of TYPE_ALIGN_UNIT.
7820 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
7821 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
7822 (vect_create_addr_base_for_vector_ref): Update call accordingly.
7823 (vect_create_data_ref_ptr): Likewise.
7824 (vect_setup_realignment): Realign by ANDing with
7825 -DR_TARGET_MISALIGNMENT.
7826 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
7827 the number of peels based on DR_TARGET_ALIGNMENT.
7828 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
7829 with the guaranteed alignment boundary when deciding whether
7831 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
7832 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
7833 (ensure_base_align): Remove stmt_info parameter. Get the
7834 target base alignment from DR_TARGET_ALIGNMENT.
7835 (vectorizable_store): Update call accordingly. Interpret
7836 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
7838 (vectorizable_load): Likewise.
7840 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7841 Alan Hayward <alan.hayward@arm.com>
7842 David Sherwood <david.sherwood@arm.com>
7844 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
7845 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
7846 (vect_enhance_data_refs_alignment): Likewise.
7848 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
7850 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
7851 error. Only quit immediately if parsing is complete.
7852 (BEGIN): Initialize fatal_err and parse_done.
7853 (begin fpu, end fpu): Check number of arguments.
7854 (begin arch, end arch): Likewise.
7855 (begin cpu, end cpu): Likewise.
7856 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
7857 (optalias): Likewise.
7859 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
7861 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
7862 * config/arm/arm-isa.h: Delete. Move definitions to ...
7863 * arm-cpus.in: ... here. Use new feature and fgroup values.
7864 * config/arm/arm.c (arm_option_override): Use lower case for feature
7866 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
7867 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
7868 * config/arm/parsecpu.awk (END): Add new command 'isa'.
7870 (print_isa_bits_for): New function.
7871 (gen_isa): New function.
7872 (gen_comm_data): Use print_isa_bits_for.
7873 (define feature): New keyword.
7874 (define fgroup): New keyword.
7875 * config/arm/t-arm (TM_H): Remove.
7876 (GTM_H): Add arm-isa.h.
7877 (arm-isa.h): Add rule to generate file.
7878 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
7879 case for feature bit names.
7881 2017-09-22 Richard Biener <rguenther@suse.de>
7883 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
7885 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
7886 print of no dependency loops ...
7887 * graphite.c (graphite_transform_loops): ... here.
7888 (canonicalize_loop_closed_ssa_form): Work from inner to outer
7890 (same_close_phi_node, remove_duplicate_close_phi,
7891 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
7892 (canonicalize_loop_closed_ssa): ... here and simplify.
7893 * graphite-optimize-isl.c: Include tree-vectorizer.h.
7894 (optimize_isl): Use dump_printf_loc to tell when we stopped
7895 optimizing because of an ISL timeout.
7897 2017-09-22 Richard Biener <rguenther@suse.de>
7899 PR tree-optimization/82291
7900 * tree-if-conv.c (predicate_mem_writes): Make sure to
7901 remove writes in blocks predicated with false.
7903 2017-09-22 Richard Biener <rguenther@suse.de>
7905 * sese.c: Include cfganal.h.
7906 (if_region_set_false_region): Remove.
7907 (create_if_region_on_edge): Likewise.
7908 (move_sese_in_condition): Re-implement without destroying
7911 2017-09-22 Richard Biener <rguenther@suse.de>
7913 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
7914 Verify both BBs contain loop PHI nodes before dispatching to
7916 (graphite_regenerate_ast_isl): Do not recompute dominators,
7917 do not verify three times. Restructure for clarity.
7918 * graphite-scop-detection.c (same_close_phi_node,
7919 remove_duplicate_close_phi, make_close_phi_nodes_unique,
7920 defined_in_loop_p, canonicalize_loop_closed_ssa,
7921 canonicalize_loop_closed_ssa_form): Simplify, remove excess
7922 checking and SSA rewrite, move to ...
7923 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
7924 (graphite_initialize): Do not pass in ctx, do not reset the
7925 SCEV cache, compute only dominators.
7926 (graphite_transform_loops): Allocate ISL ctx after
7927 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
7928 Maintain post-dominators only around build_scops.
7929 * sese.c (if_region_set_false_region): Make static. Free
7930 and recompute dominators.
7931 (move_sese_in_condition): Assert we don't get called with
7932 post-dominators computed.
7933 * sese.h (if_region_set_false_region): Remove.
7935 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
7937 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
7938 mode attribute for TARGET_AVX512VL.
7940 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
7942 * config/i386/i386.opt (mprefer-avx256): New option.
7943 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
7945 (ix86_preferred_simd_mode): Return 256-bit AVX modes
7946 for TARGET_PREFER_AVX256.
7947 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
7949 2017-09-21 Jeff Law <law@redhat.com>
7951 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
7952 Fix dump output if the only stack space is for pushed registers.
7954 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7956 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
7959 2017-09-21 Martin Sebor <msebor@redhat.com>
7962 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
7963 code (in C++) or code that triggers warnings.
7965 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
7967 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
7969 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
7971 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
7972 * sched-rgn.c: Ditto.
7973 * sel-sched-ir.c: Ditto.
7975 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
7977 * toplev.h (set_random_seed): Adjust return type.
7978 * toplev.c (init_local_tick): Move eager initialization of random_seed
7979 to get_random_seed. Adjust comment.
7980 (init_random_seed): Inline to get_random_seed, delete.
7981 (get_random_seed): Initialize random_seed lazily.
7982 (set_random_seed): Do not return previous value.
7983 (print_switch_value): Do not call get_random_seed.
7985 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
7987 * cgraph.c (delete_function_version): New, broken out from...
7988 (cgraph_node::delete_function_version): ...here. Rename to
7989 cgraph_node::delete_function_version_by_decl. Update all uses.
7990 (cgraph_node::remove): Call delete_function_version.
7992 2017-09-21 Jakub Jelinek <jakub@redhat.com>
7995 * tree-inline.c (expand_call_inline): Emit clobber stmts for
7996 VAR_DECLs to which addressable non-volatile parameters are mapped
7997 and for id->retvar after the return value assignment. Clear
7998 id->retval and id->retbnd after inlining.
8000 2017-09-21 Richard Biener <rguenther@suse.de>
8002 PR tree-optimization/82276
8003 PR tree-optimization/82244
8004 * tree-vrp.c (build_assert_expr_for): Set
8005 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
8007 (remove_range_assertions): Revert earlier change.
8009 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
8012 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
8014 2017-09-21 Richard Biener <rguenther@suse.de>
8016 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
8017 Restore valid IL after code generation errors.
8018 * graphite.c (graphite_transform_loops): Diagnose code
8019 generation issues as MSG_MISSED_OPTIMIZATION and continue
8020 with processing SCOPs.
8022 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8023 Alan Hayward <alan.hayward@arm.com>
8024 David Sherwood <david.sherwood@arm.com>
8026 * calls.c (compute_argument_addresses): Use simplify_gen_binary
8027 rather than choosing between plus_constant and gen_rtx_<CODE>.
8028 * expr.c (emit_push_insn): Likewise.
8029 (expand_expr_real_2): Likewise.
8031 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8032 Alan Hayward <alan.hayward@arm.com>
8033 David Sherwood <david.sherwood@arm.com>
8035 * loop-unroll.c (split_iv): Call copy_rtx on the step.
8037 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8038 Alan Hayward <alan.hayward@arm.com>
8039 David Sherwood <david.sherwood@arm.com>
8041 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
8042 calling tree_to_uhwi.
8044 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8045 Alan Hayward <alan.hayward@arm.com>
8046 David Sherwood <david.sherwood@arm.com>
8048 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
8049 INTEGER_CST rather than a negative test for ADDR_EXPR.
8051 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8052 Alan Hayward <alan.hayward@arm.com>
8053 David Sherwood <david.sherwood@arm.com>
8055 * tree-vrp.c (extract_range_from_binary_expr_1): Check
8056 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
8058 2017-09-21 Richard Biener <rguenther@suse.de>
8060 PR tree-optimization/71351
8061 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
8062 graphite_create_new_loop_guard): Remove, fold remaining parts
8064 (translate_isl_ast_node_for): ... here and simplify.
8066 2017-09-21 Jakub Jelinek <jakub@redhat.com>
8069 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
8070 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
8071 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
8072 alternative always use QI mode, for -Os imov (=R,R) alternative
8073 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
8076 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8077 Jeff Law <law@redhat.com>
8079 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
8080 (allocate_stack_space): New function, partially extracted from
8082 (s390_emit_prologue): Track offset to most recent stack probe.
8083 Code to allocate space moved into allocate_stack_space.
8084 Dump actions when no stack is allocated.
8085 (s390_prologue_plus_offset): New function.
8086 (s390_emit_stack_probe): Likewise.
8088 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
8090 * common.opt (Wa, Wl, Wp, g, gz=): Add
8092 (gno-column-info): Remove.
8093 (gcolumn-info): Drop RejectNegative.
8095 (gno-record-gcc-switches): Remove.
8096 (grecord-gcc-switches): Drop RejectNegative.
8097 (gno-split-dwarf): Remove.
8098 (gsplit-dwarf): Drop RejectNegative.
8099 (gno-strict-dwarf): Remove.
8100 (gstrict-dwarf): Drop RejectNegative.
8101 * config/darwin.opt (gfull, gused): Add RejectNegative.
8102 * dwarf2out.c (gen_producer_string): Drop
8103 gno-record-gcc-switches handler.
8104 * optc-gen.awk: Add g to prefixes with negative forms.
8105 * opts-common.c (remapping_prefix_p): New.
8106 (find_opt): Check it.
8107 (generate_canonical_option): Test g prefix.
8108 (option_map): Add -gno- mapping.
8109 (add_misspelling_candidates): Check remapping_prefix_p.
8111 2017-09-20 Jeff Law <law@redhat.com>
8113 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
8114 thinko in stack clash protection support.
8116 * explow.c (compute_stack_clash_protection_loop_data): Use
8117 CONST_INT_P instead of explicit test. Verify object is a
8118 CONST_INT_P before looking at INTVAL.
8119 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
8120 instead of explicit test.
8122 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
8125 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
8126 address instead of to r1 and r11.
8128 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
8130 * config.gcc: Support "knm".
8131 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
8132 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
8134 * config/i386/i386.c (m_KNM): Define.
8135 (processor_target_table): Add "knm".
8137 (ix86_option_override_internal): Add "knm".
8138 (ix86_issue_rate): Add PROCESSOR_KNM.
8139 (ix86_adjust_cost): Ditto.
8140 (ia32_multipass_dfa_lookahead): Ditto.
8141 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
8142 (fold_builtin_cpu): Add M_INTEL_KNM.
8143 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
8144 (processor_type): Add PROCESSOR_KNM.
8145 * config/i386/x86-tune.def: Add m_KNM.
8146 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
8148 2017-09-20 Richard Biener <rguenther@suse.de>
8150 PR tree-optimization/80213
8151 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
8152 are allowed in empty BBs as well.
8153 (canonicalize_loop_closed_ssa): Also look for other complex
8155 (scop_detection::get_sese): Include the loop-closed PHI block
8157 (scop_detection::merge_sese): Remove code adding extra blocks.
8158 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
8159 (build_scops): Assert the final returned scop is invalid.
8161 2017-09-20 Richard Biener <rguenther@suse.de>
8163 PR tree-optimization/82264
8164 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
8166 (vn_phi_lookup): Likewise.
8167 (vn_phi_insert): Likewise.
8169 2017-09-20 Jakub Jelinek <jakub@redhat.com>
8171 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
8172 that fits into uhwi or shwi, add DW_AT_const_value regardless
8173 of early_dwarf without going through RTL, using add_AT_unsigned
8176 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
8177 (DEBUG_LTO_ABBREV_SECTION): Likewise.
8178 (DEBUG_LTO_MACINFO_SECTION): Likewise.
8179 (DEBUG_MACRO_SECTION): Likewise.
8180 (DEBUG_LTO_MACRO_SECTION): Likewise.
8181 (DEBUG_STR_DWO_SECTION): Likewise.
8182 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
8183 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
8184 (DEBUG_LTO_DWO_LINE_SECTION): Define.
8185 (DEBUG_LTO_LINE_STR_SECTION): Define.
8186 (init_sections_and_labels): Initialize debug_line_str_section
8187 variable. Initialize debug_loc_section for -gdwarf-5 to
8188 DEBUG_LOCLISTS_SECTION. Formatting fixes.
8190 2017-09-20 Richard Biener <rguenther@suse.de>
8192 * graphite-sese-to-poly.c (extract_affine): Properly handle
8193 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
8195 2017-09-20 Richard Biener <rguenther@suse.de>
8197 PR tree-optimization/81373
8198 * graphite-scop-detection.c (build_cross_bb_scalars_def):
8199 Force SESE live-out defs to be handled even if they are
8202 2017-09-19 Jeff Law <law@redhat.com>
8204 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
8205 nothing for stack adjustments with REG_STACK_CHECK.
8206 * sched-deps.c (parse_add_or_inc): Reject insns with
8207 REG_STACK_CHECK from dependency breaking.
8208 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
8209 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
8210 * reg-notes.def (STACK_CHECK): New note.
8212 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
8213 (ix86_expand_prologue): Dump stack clash info as needed.
8214 Call ix86_adjust_stack_and_probe_stack_clash as needed.
8216 * function.c (dump_stack_clash_frame_info): New function.
8217 * function.h (dump_stack_clash_frame_info): Prototype.
8218 (enum stack_clash_probes): New enum.
8220 * config/alpha/alpha.c (alpha_expand_prologue): Also check
8221 flag_stack_clash_protection.
8222 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
8223 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
8224 (arm_frame_pointer_required): Likewise.
8225 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
8226 (ia64_expand_prologue): Likewise.
8227 * config/mips/mips.c (mips_expand_prologue): Likewise.
8228 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
8229 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
8230 (sparc_flat_expand_prologue): Likewise.
8231 * config/spu/spu.c (spu_expand_prologue): Likewise.
8233 * explow.c: Include "params.h".
8234 (anti_adjust_stack_and_probe_stack_clash): New function.
8235 (get_stack_check_protect): Likewise.
8236 (compute_stack_clash_protection_loop_data): Likewise.
8237 (emit_stack_clash_protection_loop_start): Likewise.
8238 (emit_stack_clash_protection_loop_end): Likewise.
8239 (allocate_dynamic_stack_space): Use get_stack_check_protect.
8240 Use anti_adjust_stack_and_probe_stack_clash.
8241 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
8242 (emit_stack_clash_protection_loop_start): Likewise.
8243 (emit_stack_clash_protection_loop_end): Likewise.
8244 * rtl.h (get_stack_check_protect): Prototype.
8245 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
8246 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
8247 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
8249 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
8251 * doc/tm.texi: Rebuilt.
8252 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
8253 get_stack_check_protect.
8254 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
8255 * config/arm/arm.c (arm_expand_prologue): Likewise.
8256 (arm_frame_pointer_required): Likewise.
8257 * config/i386/i386.c (ix86_expand_prologue): Likewise.
8258 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
8259 * config/mips/mips.c (mips_expand_prologue): Likewise.
8260 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
8261 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
8262 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
8263 (sparc_flat_expand_prologue): Likewise.
8265 * common.opt (-fstack-clash-protection): New option.
8266 * flag-types.h (enum stack_check_type): Note difference between
8267 -fstack-check= and -fstack-clash-protection.
8268 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
8269 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
8270 * toplev.c (process_options): Issue warnings/errors for cases
8271 not handled with -fstack-clash-protection.
8272 * doc/invoke.texi (-fstack-clash-protection): Document new option.
8273 (-fstack-check): Note additional problem with -fstack-check=generic.
8274 Note that -fstack-check is primarily for Ada and refer users
8275 to -fstack-clash-protection for stack-clash-protection.
8276 Document new params for stack clash protection.
8278 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
8280 * config/i386/i386.c (ix86_split_long_move): Do not handle
8281 address used for LEA in a special way.
8283 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
8285 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
8287 2017-09-19 Martin Sebor <msebor@redhat.com>
8290 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
8291 of incompatible types.
8293 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
8295 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
8296 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
8297 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8298 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
8300 2017-09-19 Richard Biener <rguenther@suse.de>
8302 PR tree-optimization/82244
8303 * tree-vrp.c (remove_range_assertions): Do not propagate
8304 a constant to abnormals but replace the assert with a copy.
8306 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
8308 PR rtl-optimization/57878
8309 PR rtl-optimization/68988
8310 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
8311 avoidance test involving non_reload_pseudos. Move frequency test
8312 below the general fragmentation avoidance test.
8314 2017-09-19 Richard Biener <rguenther@suse.de>
8316 PR tree-optimization/69728
8317 * graphite-sese-to-poly.c (schedule_error): New global.
8318 (add_loop_schedule): Handle empty domain by failing the
8320 (build_original_schedule): Handle schedule_error.
8322 2017-09-19 Richard Biener <rguenther@suse.de>
8324 * graphite-scop-detection.c (scop_detection::can_represent_loop):
8325 Do not iterate to sibling loops but only to siblings of inner
8328 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
8331 * config/m68k/m68k.md (moveq feeding equality comparison): Check
8332 that the registers are different.
8334 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
8336 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
8337 to processor_model and "amdfam17h" to arch_names_table.
8338 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
8340 2017-09-18 Jakub Jelinek <jakub@redhat.com>
8343 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
8345 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8346 Alan Hayward <alan.hayward@arm.com>
8347 David Sherwood <david.sherwood@arm.com>
8349 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
8351 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
8353 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
8354 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
8355 vect_schedule_slp_instance.
8356 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
8357 Update call to vect_slp_analyze_node_operations. Simplify return
8359 (vect_slp_analyze_bb_1): Update call accordingly.
8360 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
8361 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
8362 (vect_schedule_slp): Update call accordingly.
8364 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8365 Alan Hayward <alan.hayward@arm.com>
8366 David Sherwood <david.sherwood@arm.com>
8368 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
8369 with types that aren't in fact scalar.
8371 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8373 * tree-vect-slp.c (vect_record_max_nunits): New function,
8375 (vect_build_slp_tree_1): ...here.
8376 (vect_build_slp_tree_2): Call it for phis too.
8378 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8380 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
8381 to vect_get_vec_def_for_operand when getting the mask operand.
8383 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8384 Alan Hayward <alan.hayward@arm.com>
8385 David Sherwood <david.sherwood@arm.com>
8387 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
8390 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8391 Alan Hayward <alan.hayward@arm.com>
8392 David Sherwood <david.sherwood@arm.com>
8394 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
8395 calculation for vector booleans.
8397 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8398 Alan Hayward <alan.hayward@arm.com>
8399 David Sherwood <david.sherwood@arm.com>
8401 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
8403 (vect_transform_stmt): ...here.
8404 (vect_analyze_stmt): Use it instead of calling
8405 vectorizable_live_operation directly.
8407 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
8409 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
8410 non-SIMT targets in acc vector loops.
8412 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
8414 * configure.ac: Add arc and check if assembler supports gdwarf2.
8415 * configure: Regenerate.
8417 2017-09-18 Richard Biener <rguenther@suse.de>
8419 PR tree-optimization/82220
8420 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
8421 epilogue niters from the min_profitable_iters compute.
8423 2017-09-18 Jakub Jelinek <jakub@redhat.com>
8426 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
8427 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
8428 (ix86_init_pic_reg): Revert 2017-09-01 changes.
8430 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
8433 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
8434 switching to a new text section.
8436 2017-09-18 Richard Biener <rguenther@suse.de>
8438 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
8440 (build_alias_set): Reject aliases with no access function.
8442 2017-09-18 Richard Biener <rguenther@suse.de>
8444 PR tree-optimization/79622
8445 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
8447 (build_cross_bb_scalars_use): Likewise.
8449 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
8451 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
8453 2017-09-18 Alan Modra <amodra@gmail.com>
8456 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
8457 stack_pointer_rtx for count 0. Update comments. Break up
8458 large rtl expression.
8460 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
8462 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
8463 Increase to 20 bytes.
8464 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
8465 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
8466 or avx version of the stub.
8468 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
8471 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
8472 compute the minimum stack alignment. Also update preferred stack
8473 boundary for leaf functions.
8475 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
8477 PR tree-optimization/82228
8478 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
8481 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
8483 * common/config/nds32/nds32-common.c
8484 (nds32_option_optimization_table): Refine formatting.
8485 (nds32_option_optimization_table): Use -fsched-pressure and
8486 -fomit-frame-pointer for specific optimization level.
8488 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
8490 * config/nds32/nds32.c: Refine formatting and comments.
8491 * config/nds32/nds32.h: Likewise.
8492 * config/nds32/nds32.md: Likewise.
8493 * config/nds32/nds32-cost.c: Likewise.
8494 * config/nds32/nds32-isr.c: Likewise.
8495 * config/nds32/nds32-md-auxiliary.c: Likewise.
8496 * config/nds32/nds32-multiple.md: Likewise.
8497 * config/nds32/nds32-predicates.c: Likewise.
8499 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
8500 Jakub Jelinek <jakub@redhat.com>
8502 Add support for -std=c++2a.
8503 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
8505 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
8507 2017-09-15 Steve Ellcey <sellcey@cavium.com>
8510 * doc/extend.texi (Common Function Attributes): Add
8511 references to ARM, AArch64, and S/390 specific attributes.
8512 (Function Specific Option Pragmas): Add AArch64 and S/390
8513 to list of back ends that support the target pragma.
8515 2017-09-15 Nathan Sidwell <nathan@acm.org>
8517 * doc/standards.texi: Fix C++17 description. Update URLs for
8520 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
8522 * common.opt (Wcast-align=strict): New warning option.
8523 * doc/invoke.texi: Document -Wcast-align=strict.
8525 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
8527 * cgraph.h (cgraph_thunk_info): Add comments.
8528 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
8529 assert for VIRTUAL_* arguments stricter.
8531 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
8533 PR tree-optimization/71026
8534 * match.pd: Move RDIV patterns from fold-const.c
8535 * fold-const.c (distribute_real_division): Removed.
8536 (fold_binary_loc): Remove calls to distribute_real_divison.
8538 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8540 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
8541 c++1z and gnu++1z as deprecated. Change other references to
8542 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
8543 Change -Wc++1z-compat to -Wc++17-compat.
8544 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
8545 * dwarf2out.c (highest_c_language): Handle C++17.
8546 (gen_compile_unit_die): Likewise.
8548 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8550 PR rtl-optimization/82192
8551 * combine.c (make_extraction): Don't look through non-paradoxical
8552 SUBREGs or TRUNCATE if pos + len is or might be bigger than
8555 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8556 Alan Hayward <alan.hayward@arm.com>
8557 David Sherwood <david.sherwood@arm.com>
8559 * target.def (function_arg_offset): New hook.
8560 * targhooks.h (default_function_arg_offset): Declare.
8561 * targhooks.c (default_function_arg_offset): New function.
8562 * function.c (locate_and_pad_parm): Use
8563 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
8564 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
8565 (TARGET_FUNCTION_ARG_OFFSET): ...this.
8566 * doc/tm.texi: Regenerate.
8567 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
8568 * config/spu/spu.c (spu_function_arg_offset): New function.
8569 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
8570 * system.h (FUNCTION_ARG_OFFSET): Poison.
8572 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8573 Alan Hayard <alan.hayward@arm.com>
8574 David Sherwood <david.sherwood@arm.com>
8576 * target.def (truly_noop_truncation): New hook.
8577 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
8578 than TRULY_NOOP_TRUNCATION.
8579 * hooks.h (hook_bool_uint_uint_true): Declare.
8580 * hooks.c (hook_bool_uint_uint_true): New function.
8581 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
8582 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
8583 * doc/tm.texi: Regenerate.
8584 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
8585 rather than TRULY_NOOP_TRUNCATION in comments.
8586 (simplify_comparison): Likewise.
8587 (record_truncated_value): Likewise.
8588 * expmed.c (extract_bit_field_1): Likewise.
8589 (extract_split_bit_field): Likewise.
8590 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
8591 instead of TRULY_NOOP_TRUNCATION.
8592 * function.c (assign_parm_setup_block): Likewise.
8593 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
8594 * rtlhooks.c: Include target.h.
8595 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
8596 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
8597 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
8598 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
8599 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
8600 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
8601 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
8602 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
8603 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
8604 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
8605 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
8606 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
8607 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
8608 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
8609 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
8610 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
8611 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
8612 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
8613 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
8614 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
8615 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
8616 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
8617 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
8618 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
8619 * config/mips/mips.c (mips_truly_noop_truncation): New function.
8620 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8621 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8622 rather than TRULY_NOOP_TRUNCATION in comments.
8623 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
8624 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
8625 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
8626 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
8627 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
8628 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
8629 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
8630 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
8631 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
8632 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
8633 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
8634 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8635 rather than TRULY_NOOP_TRUNCATION in comments.
8636 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
8637 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
8638 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
8639 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
8640 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
8641 TRULY_NOOP_TRUNCATION condition.
8642 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
8643 (TRULY_NOOP_TRUNCATION): Delete.
8644 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
8645 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
8646 * config/spu/spu.c (spu_truly_noop_truncation): New function.
8647 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8648 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
8649 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
8650 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
8651 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8652 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8653 rather than TRULY_NOOP_TRUNCATION in comments.
8654 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
8655 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
8656 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
8657 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
8658 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
8659 * system.h (TRULY_NOOP_TRUNCATION): Poison.
8661 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
8664 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
8665 (*cmp_ior): Likewise.
8666 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
8667 (*ior_scc_scc_cmp): Likewise.
8668 (*and_scc_scc): Likewise.
8669 (*and_scc_scc_cmp): Likewise.
8671 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8672 Alan Hayard <alan.hayward@arm.com>
8673 David Sherwood <david.sherwood@arm.com>
8675 * target.def (can_change_mode_class): New hook.
8676 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
8677 (hard_regno_nregs): Likewise.
8678 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
8679 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
8680 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
8681 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
8682 (LOAD_EXTEND_OP): Update accordingly.
8683 * doc/tm.texi: Regenerate.
8684 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
8685 CANNOT_CHANGE_MODE_CLASS.
8686 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
8687 (REG_CAN_CHANGE_MODE_P): ...this new macro.
8688 * combine.c (simplify_set): Update accordingly.
8689 * emit-rtl.c (validate_subreg): Likewise.
8690 * recog.c (general_operand): Likewise.
8691 * regcprop.c (mode_change_ok): Likewise.
8692 * reload1.c (choose_reload_regs): Likewise.
8693 (inherit_piecemeal_p): Likewise.
8694 * rtlanal.c (simplify_subreg_regno): Likewise.
8695 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
8696 instead of CANNOT_CHANGE_MODE_CLASS.
8697 (reload_cse_simplify_operands): Likewise.
8698 * reload.c (push_reload): Use targetm.can_change_mode_class
8699 instead of CANNOT_CHANGE_MODE_CLASS.
8700 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
8701 REG_CANNOT_CHANGE_MODE_P.
8702 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8703 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
8704 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8705 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8706 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8707 (arm_can_change_mode_class): New function.
8708 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
8709 than CANNOT_CHANGE_MODE_CLASS in comments.
8710 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8711 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
8712 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
8713 (ix86_can_change_mode_class): ...this new function, inverting the
8714 sense of the return value.
8715 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8716 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8717 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8718 (ia64_can_change_mode_class): New function.
8719 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8720 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
8721 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
8722 (m32c_can_change_mode_class): ...this new function, inverting the
8723 sense of the return value.
8724 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8725 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8726 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
8727 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
8728 (mips_can_change_mode_class): ...this new function, inverting the
8729 sense of the return value.
8730 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8731 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8732 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8733 (msp430_can_change_mode_class): New function.
8734 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8735 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
8736 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8737 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8738 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8739 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
8740 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8741 (pa_cannot_change_mode_class): Replace with...
8742 (pa_can_change_mode_class): ...this new function, inverting the
8743 sense of the return value.
8744 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
8745 than CANNOT_CHANGE_MODE_CLASS in comments.
8746 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8747 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
8748 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8749 (pdp11_cannot_change_mode_class): Replace with...
8750 (pdp11_can_change_mode_class): ...this new function, inverting the
8751 sense of the return value.
8752 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8753 * config/powerpcspe/powerpcspe-protos.h
8754 (rs6000_cannot_change_mode_class_ptr): Delete.
8755 * config/powerpcspe/powerpcspe.c
8756 (rs6000_cannot_change_mode_class_ptr): Delete.
8757 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8758 (rs6000_option_override_internal): Assign to
8759 targetm.can_change_mode_class instead of
8760 rs6000_cannot_change_mode_class_ptr.
8761 (rs6000_cannot_change_mode_class): Replace with...
8762 (rs6000_can_change_mode_class): ...this new function, inverting the
8763 sense of the return value.
8764 (rs6000_debug_cannot_change_mode_class): Replace with...
8765 (rs6000_debug_can_change_mode_class): ...this new function.
8766 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8767 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
8768 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8769 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8770 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
8772 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
8773 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8774 (rs6000_option_override_internal): Assign to
8775 targetm.can_change_mode_class instead of
8776 rs6000_cannot_change_mode_class_ptr.
8777 (rs6000_cannot_change_mode_class): Replace with...
8778 (rs6000_can_change_mode_class): ...this new function, inverting the
8779 sense of the return value.
8780 (rs6000_debug_cannot_change_mode_class): Replace with...
8781 (rs6000_debug_can_change_mode_class): ...this new function.
8782 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8783 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
8784 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
8785 (s390_can_change_mode_class): ...this new function, inverting the
8786 sense of the return value.
8787 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8788 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8789 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
8790 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8791 (sh_cannot_change_mode_class): Replace with...
8792 (sh_can_change_mode_class): ...this new function, inverting the
8793 sense of the return value.
8794 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8795 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8796 (sparc_can_change_mode_class): New function.
8797 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8798 * config/spu/spu.c (spu_can_change_mode_class): New function.
8799 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8800 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8801 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8802 (visium_can_change_mode_class): New function.
8803 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
8805 2017-09-15 Richard Biener <rguenther@suse.de>
8807 PR tree-optimization/82217
8808 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
8809 but not undefined case.
8811 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8814 * postreload.c (reload_cse_simplify_operands): Skip
8815 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
8817 2017-09-15 Richard Biener <rguenther@suse.de>
8819 PR tree-optimization/68823
8820 * graphite-scop-detection.c (build_alias_set): If we have a
8821 possible dependence check whether we can handle them by just
8822 looking at the DRs DR_ACCESS_FNs.
8823 (build_scops): If build_alias_set fails, fail the SCOP.
8825 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
8827 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
8828 to support float128 built-in functions that require the ISA 3.0
8830 (BU_FLOAT128_3_HW): Likewise.
8831 (SQRTF128): Add support for the IEEE 128-bit square root and fma
8833 (FMAF128): Likewise.
8835 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
8836 support for built-in functions that need the ISA 3.0 IEEE 128-bit
8837 floating point instructions.
8838 (rs6000_invalid_builtin): Likewise.
8839 (rs6000_builtin_mask_names): Likewise.
8840 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
8841 (RS6000_BTM_FLOAT128_HW): Likewise.
8842 (RS6000_BTM_COMMON): Likewise.
8843 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
8845 * doc/extend.texi (RS/6000 built-in functions): Document the
8846 IEEE 128-bit floating point square root and fused multiply-add
8849 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
8851 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
8852 reg (r2) isn't in the set of registers defined in the prologue.
8854 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8855 Alan Hayward <alan.hayward@arm.com>
8856 David Sherwood <david.sherwood@arm.com>
8858 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
8859 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
8860 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
8861 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
8862 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
8864 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
8865 max_vectorization_factor.
8866 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
8868 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8869 Alan Hayward <alan.hayward@arm.com>
8870 David Sherwood <david.sherwood@arm.com>
8872 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
8873 (vect_worthwhile_without_simd_p): Declare.
8874 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
8875 (vectorizable_reduction): Use it.
8876 * tree-vect-stmts.c (vectorizable_shift): Likewise.
8877 (vectorizable_operation): Likewise.
8879 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8880 Alan Hayward <alan.hayward@arm.com>
8881 David Sherwood <david.sherwood@arm.com>
8883 * tree-vectorizer.h (vect_get_num_copies): New function.
8884 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
8885 * tree-vect-loop.c (vectorizable_reduction): Likewise.
8886 (vectorizable_induction): Likewise.
8887 (vectorizable_live_operation): Likewise.
8888 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
8889 (vectorizable_bswap): Likewise.
8890 (vectorizable_call): Likewise.
8891 (vectorizable_conversion): Likewise.
8892 (vectorizable_assignment): Likewise.
8893 (vectorizable_shift): Likewise.
8894 (vectorizable_operation): Likewise.
8895 (vectorizable_store): Likewise.
8896 (vectorizable_load): Likewise.
8897 (vectorizable_condition): Likewise.
8898 (vectorizable_comparison): Likewise.
8899 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
8901 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8902 Alan Hayward <alan.hayward@arm.com>
8903 David Sherwood <david.sherwood@arm.com>
8905 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
8906 of vect_init_vector.
8908 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8909 Alan Hayward <alan.hayward@arm.com>
8910 David Sherwood <david.sherwood@arm.com>
8912 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
8913 an inline wrapper that provides a location.
8914 (gimple_build_vector): Likewise.
8915 * gimple-fold.c (gimple_build_vector_from_val): New function.
8916 (gimple_build_vector): Likewise.
8917 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
8918 functions to build the initial value. Always return a gimple value.
8919 (get_initial_defs_for_reduction): Likewise. Only compute
8921 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
8922 vect_init_vector on the results from get_initial_def(s)_for_reduction.
8923 (vectorizable_induction): Use gimple_build_vector rather than
8926 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8927 Alan Hayward <alan.hayward@arm.com>
8928 David Sherwood <david.sherwood@arm.com>
8930 * target.h (vec_perm_indices): New typedef.
8931 (auto_vec_perm_indices): Likewise.
8932 * optabs-query.h: Include target.h
8933 (can_vec_perm_p): Take a vec_perm_indices *.
8934 * optabs-query.c (can_vec_perm_p): Likewise.
8935 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
8936 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
8937 * tree-vect-generic.c (lower_vec_perm): Likewise.
8938 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
8939 (vect_grouped_load_supported): Likewise.
8940 (vect_shift_permute_load_chain): Likewise.
8941 (vect_permute_store_chain): Use auto_vec_perm_indices.
8942 (vect_permute_load_chain): Likewise.
8943 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
8944 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
8945 Update uses of can_vec_perm_p.
8946 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
8947 mode with a number of elements. Take a vec_perm_indices *.
8948 (vect_create_epilog_for_reduction): Update accordingly.
8949 Use auto_vec_perm_indices.
8950 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
8951 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
8952 (vect_transform_slp_perm_load): Likewise.
8953 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
8954 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
8955 (vect_gen_perm_mask_checked): Likewise.
8956 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
8957 (vect_gen_perm_mask_checked): Likewise.
8958 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
8959 (vectorizable_store): Likewise.
8960 (vectorizable_load): Likewise.
8961 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
8962 (vectorizable_bswap): Likewise.
8964 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8965 Alan Hayward <alan.hayward@arm.com>
8966 David Sherwood <david.sherwood@arm.com>
8968 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
8969 * tree.c (build_vector): Likewise.
8970 (build_vector_from_ctor): Update accordingly.
8971 (build_vector_from_val): Likewise.
8972 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
8973 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
8974 * tree-vect-generic.c (add_rshift): Likewise.
8975 (expand_vector_divmod): Likewise.
8976 (optimize_vector_constructor): Likewise.
8977 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
8978 (vect_transform_slp_perm_load): Likewise.
8979 (vect_schedule_slp_instance): Likewise.
8980 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
8981 (vectorizable_call): Likewise.
8982 (vect_gen_perm_mask_any): Likewise. Add elements in order.
8983 * expmed.c (make_tree): Likewise.
8984 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
8985 a vector passed to build_vector.
8986 (fold_convert_const): Likewise.
8987 (exact_inverse): Likewise.
8988 (fold_ternary_loc): Likewise.
8989 (fold_relational_const): Likewise.
8990 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
8991 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
8992 (const_unop): Likewise. Store the reduction accumulator in a
8993 variable rather than an array.
8994 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
8995 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
8996 the new vector, rather than constructing it after the input arrays.
8997 (native_interpret_vector): Use auto_vec<tree> when building
8998 a vector passed to build_vector. Add elements in order.
8999 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
9000 auto_vec<tree> when building a vector passed to build_vector.
9001 (vect_create_epilog_for_reduction): Likewise.
9002 (vectorizable_induction): Likewise.
9003 (get_initial_def_for_reduction): Likewise. Fix indentation of
9005 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
9007 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
9008 passed to build_vector.
9010 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9011 Alan Hayward <alan.hayward@arm.com>
9012 David Sherwood <david.sherwood@arm.com>
9014 * tree-core.h (tree_base::u): Add an "nelts" field.
9015 (tree_vector): Use VECTOR_CST_NELTS as the length.
9016 * tree.c (tree_size): Likewise.
9017 (make_vector): Initialize VECTOR_CST_NELTS.
9018 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
9019 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
9020 TYPE_VECTOR_SUBPARTS.
9021 * expr.c (const_vector_mask_from_tree): Consistently use "units"
9022 as the number of units, setting it from VECTOR_CST_NELTS.
9023 (const_vector_from_tree): Likewise.
9024 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
9025 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
9026 (fold_negate_expr_1): Likewise.
9027 (fold_convert_const): Likewise.
9028 (const_binop): Likewise. Differentiate the number of output and
9030 (const_unop): Likewise.
9031 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
9032 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
9033 in cases that did the opposite.
9035 2017-09-14 Richard Biener <rguenther@suse.de>
9037 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
9040 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
9042 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
9044 2017-09-14 Jakub Jelinek <jakub@redhat.com>
9047 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
9048 if and where to split a bb, except for splitting before debug insn
9049 sequences followed by non-label real insn. Delete debug insns
9050 in between basic blocks.
9052 * combine.c (make_compound_operation_int): Formatting fixes.
9054 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
9055 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9056 * config/netbsd.h (LINK_EH_SPEC): Likewise.
9057 * config/sol2.h (LINK_EH_SPEC): Likewise.
9058 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9059 * config/s390/linux.h (LINK_SPEC): Likewise.
9060 * config/freebsd.h (LINK_EH_SPEC): Likewise.
9061 * config/openbsd.h (LINK_EH_SPEC): Likewise.
9062 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9063 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
9064 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
9065 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9066 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
9067 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
9069 2017-09-13 Jakub Jelinek <jakub@redhat.com>
9071 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
9073 (ENDFILE_LINUX_SPEC): Likewise.
9074 (LINK_EH_SPEC): Likewise.
9075 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
9076 (LINK_OS_LINUX_SPEC32): Likewise.
9077 (LINK_OS_LINUX_SPEC64): Likewise.
9078 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
9079 (LINK_OS_LINUX_SPEC): Likewise.
9081 2017-09-13 Martin Liska <mliska@suse.cz>
9084 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
9085 CASE_HIGH is NULL_TREE.
9087 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
9088 Alan Hayward <alan.hayward@arm.com>
9089 David Sherwood <david.sherwood@arm.com>
9091 * target.def (secondary_memory_needed): New hook.
9092 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
9093 instead of SECONDARY_MEMORY_NEEDED.
9094 (secondary_memory_needed_mode): Likewise.
9095 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
9096 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
9097 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
9098 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
9099 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
9100 * doc/tm.texi: Regenerate.
9101 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
9102 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
9103 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9104 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
9105 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
9106 * config/i386/i386.c (inline_secondary_memory_needed): Put the
9107 mode argument first and change the reg_class arguments to reg_class_t.
9108 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
9109 Make static. Update the call to inline_secondary_memory_needed.
9110 (ix86_register_move_cost): Update the call to
9111 inline_secondary_memory_needed.
9112 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9113 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
9115 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
9116 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
9118 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
9119 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
9120 * config/mips/mips.c (mips_secondary_memory_needed): Make static
9121 and match hook interface. Add comment from mips.h.
9122 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9123 * config/mmix/mmix.md (truncdfsf2): Refer to
9124 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
9126 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
9127 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
9128 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9129 (pa_secondary_memory_needed): New function.
9130 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
9131 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
9132 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9133 (pdp11_secondary_memory_needed): Make static and match hook interface.
9134 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
9135 * config/powerpcspe/powerpcspe-protos.h
9136 (rs6000_secondary_memory_needed_ptr): Delete.
9137 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
9139 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9140 (rs6000_option_override_internal): Assign to
9141 targetm.secondary_memory_needed rather than
9142 rs6000_secondary_memory_needed_ptr.
9143 (rs6000_secondary_memory_needed): Match hook interface.
9144 (rs6000_debug_secondary_memory_needed): Likewise.
9145 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
9146 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
9147 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
9148 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9149 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
9150 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
9152 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
9153 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9154 (rs6000_option_override_internal): Assign to
9155 targetm.secondary_memory_needed rather than
9156 rs6000_secondary_memory_needed_ptr.
9157 (rs6000_secondary_memory_needed): Match hook interface.
9158 (rs6000_debug_secondary_memory_needed): Likewise.
9159 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
9160 * config/s390/s390.c (s390_secondary_memory_needed): New function.
9161 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9162 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
9163 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9164 (sparc_secondary_memory_needed): New function.
9165 * lra-constraints.c (check_and_process_move): Refer to
9166 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
9168 (curr_insn_transform): Likewise.
9169 (process_alt_operands): Use targetm.secondary_memory_needed
9170 instead of TARGET_SECONDARY_MEMORY_NEEDED.
9171 (check_secondary_memory_needed_p): Likewise.
9172 (choose_split_class): Likewise.
9173 * reload.c: Unconditionally include code that was previously
9174 conditional on SECONDARY_MEMORY_NEEDED.
9175 (push_secondary_reload): Use targetm.secondary_memory_needed
9176 instead of TARGET_SECONDARY_MEMORY_NEEDED.
9177 (push_reload): Likewise.
9178 * reload1.c: Unconditionally include code that was previously
9179 conditional on SECONDARY_MEMORY_NEEDED.
9180 (choose_reload_regs): Use targetm.secondary_memory_needed
9181 instead of TARGET_SECONDARY_MEMORY_NEEDED.
9182 (gen_reload): Likewise.
9183 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
9185 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
9186 Alan Hayward <alan.hayward@arm.com>
9187 David Sherwood <david.sherwood@arm.com>
9189 * target.def (secondary_memory_needed_mode): New hook:
9190 * targhooks.c (default_secondary_memory_needed_mode): Declare.
9191 * targhooks.h (default_secondary_memory_needed_mode): New function.
9192 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
9193 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
9194 * doc/tm.texi: Regenerate.
9195 * lra-constraints.c (check_and_process_move): Use
9196 targetm.secondary_memory_needed_mode instead of
9197 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
9198 (curr_insn_transform): Likewise.
9199 * reload.c (get_secondary_mem): Likewise.
9200 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9201 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
9203 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9204 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9205 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
9206 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9207 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
9209 * config/powerpcspe/powerpcspe-protos.h
9210 (rs6000_secondary_memory_needed_mode): Delete.
9211 * config/powerpcspe/powerpcspe.c
9212 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9213 (rs6000_secondary_memory_needed_mode): Make static.
9214 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9215 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
9217 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
9219 (rs6000_secondary_memory_needed_mode): Make static.
9220 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9221 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
9222 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9223 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9224 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
9226 (sparc_secondary_memory_needed_mode): New function.
9227 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
9229 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
9231 * config/aarch64/constraints.md (Umq): New constraint.
9232 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
9234 (mov<mode>): Update condition.
9236 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9238 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
9239 when bitposition is the same.
9241 2017-09-13 Richard Biener <rguenther@suse.de>
9243 * dwarf2out.c (output_die_symbol): Remove.
9244 (output_die): Do not output a DIEs symbol.
9246 2017-09-13 Richard Biener <rguenther@suse.de>
9249 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
9250 default-def to avoid breaking iterator update with the weird
9251 interaction with cgraph_update_edges_for_call_stmt_node.
9253 2017-09-13 Richard Biener <rguenther@suse.de>
9255 * tree-cfg.c (verify_gimple_assign_binary): Add verification
9256 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
9257 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
9258 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
9260 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
9262 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
9263 Disable pc relative literal load irrespective of
9264 TARGET_FIX_ERR_A53_84341 for default.
9266 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
9268 * config/sparc/sparc.c (output_return): Output the source location of
9269 the insn in the delay slot, if any.
9270 (output_sibcall): Likewise.
9272 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
9275 * common.opt (-static-pie): New alias.
9276 (shared): Negate static-pie.
9277 (-no-pie): Update help text.
9279 (static-pie): New option.
9280 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
9281 -static-pie support.
9282 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
9283 (LINK_EH_SPEC): Likewise.
9284 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9285 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
9286 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
9287 * gcc.c (LINK_COMMAND_SPEC): Likewise.
9288 (init_gcc_specs): Likewise.
9289 (init_spec): Likewise.
9290 (display_help): Update help message for -pie.
9291 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
9294 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
9296 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
9297 (movdi_aarch64): Likewise.
9298 (movti_aarch64): Likewise.
9300 2017-09-12 Simon Wright <simon@pushface.org>
9303 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
9304 calculation of the minor version, always output as 0.
9306 2017-09-12 Jakub Jelinek <jakub@redhat.com>
9309 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
9310 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
9311 on it early, rather than manual conversion late. For
9312 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
9313 instead of performing manual conversion.
9315 2017-09-12 Carl Love <cel@us.ibm.com>
9317 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
9318 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
9320 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
9321 VMULOSW): Add definitions.
9322 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9323 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
9324 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
9325 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
9326 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
9328 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
9330 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
9332 (movti_aarch64): Likewise.
9333 (movdf_aarch64): Likewise.
9334 (movtf_aarch64): Likewise.
9335 (load_pairdi): Likewise.
9336 (store_pairdi): Likewise.
9337 (load_pairdf): Likewise.
9338 (store_pairdf): Likewise.
9339 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
9340 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
9341 (ldr_got_small_<mode>): Likewise.
9342 (ldr_got_small_28k_<mode>): Likewise.
9343 (ldr_got_tiny): Likewise.
9344 * config/aarch64/iterators.md (ldst_sz): New.
9345 (ldpstp_sz): Likewise.
9346 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
9348 (thunderx_load): Split load_8 to load_16.
9349 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
9351 (thunderx2t99_storepair_basic): Split store_8 to store_16.
9352 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
9353 (xgene1_store_pair): Split store_8 to store_16.
9354 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
9355 (falkor_st_0_st_sd): Split store_8 to store_16.
9357 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
9359 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
9360 and store1/2/3/4 to store_4/8/12/16.
9361 * config/aarch64/aarch64.md: Update for rename.
9362 * config/arm/arm.md: Likewise.: Likewise.
9363 * config/arm/arm.c: Likewise.
9364 * config/arm/thumb1.md: Likewise.
9365 * config/arm/thumb2.md: Likewise.
9366 * config/arm/vfp.md: Likewise.
9367 * config/arm/arm-generic.md: Likewise.
9368 * config/arm/arm1020e.md: Likewise.
9369 * config/arm/arm1026ejs.md: Likewise.
9370 * config/arm/arm1136jfs.md: Likewise.
9371 * config/arm/arm926ejs.md: Likewise.
9372 * config/arm/cortex-a15.md: Likewise.
9373 * config/arm/cortex-a17.md: Likewise.
9374 * config/arm/cortex-a5.md: Likewise.
9375 * config/arm/cortex-a53.md: Likewise.
9376 * config/arm/cortex-a57.md: Likewise.
9377 * config/arm/cortex-a7.md: Likewise.
9378 * config/arm/cortex-a8.md: Likewise.
9379 * config/arm/cortex-a9.md: Likewise.
9380 * config/arm/cortex-m4.md: Likewise.
9381 * config/arm/cortex-m7.md: Likewise.
9382 * config/arm/cortex-r4.md: Likewise.
9383 * config/arm/exynos-m1.md: Likewise.
9384 * config/arm/fa526.md: Likewise.
9385 * config/arm/fa606te.md: Likewise.
9386 * config/arm/fa626te.md: Likewise.
9387 * config/arm/fa726te.md: Likewise.
9388 * config/arm/fmp626.md: Likewise.
9389 * config/arm/iwmmxt.md: Likewise.
9390 * config/arm/ldmstm.md: Likewise.
9391 * config/arm/marvell-pj4.md: Likewise.
9392 * config/arm/xgene1.md: Likewise.
9393 * config/aarch64/thunderx.md: Likewise.
9394 * config/aarch64/thunderx2t99.md: Likewise.
9395 * config/aarch64/falkor.md: Likewise.
9397 2017-09-12 Martin Liska <mliska@suse.cz>
9399 * attribs.c (private_lookup_attribute): New function.
9400 * attribs.h (private_lookup_attribute): Declared here.
9401 (lookup_attribute): Called from this place.
9403 2017-09-12 Richard Biener <rguenther@suse.de>
9405 PR tree-optimization/82157
9406 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
9407 stmts with side-effects.
9409 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9410 Alan Hayward <alan.hayward@arm.com>
9411 David Sherwood <david.sherwood@arm.com>
9413 * target.def (hard_regno_nregs): New hook.
9414 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
9415 * targhooks.h (default_hard_regno_nregs): Declare.
9416 * targhooks.c (default_hard_regno_nregs): New function.
9417 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
9418 (TARGET_HARD_REGNO_NREGS): ...this hook.
9419 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
9420 (CLASS_MAX_NREGS): Likewise.
9421 * doc/tm.texi: Regenerate.
9422 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
9423 instead of HARD_REGNO_NREGS.
9424 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
9425 HARD_REGNO_NREGS in the comment.
9426 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
9427 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
9428 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
9429 Return an unsigned int.
9430 (TARGET_HARD_REGNO_NREGS): Redefine.
9431 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
9432 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
9433 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
9434 (arc_hard_regno_nregs): New function.
9435 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
9436 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
9437 (arm_hard_regno_nregs): New function.
9438 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
9439 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
9440 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
9441 (TARGET_HARD_REGNO_NREGS): Redefine.
9442 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
9443 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
9444 (HARD_REGNO_NREGS): Delete.
9445 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
9446 (cr16_hard_regno_nregs): New function.
9447 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
9448 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
9449 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
9450 (cris_hard_regno_nregs): New function.
9451 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
9452 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
9453 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
9454 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
9455 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
9456 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
9457 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
9458 (frv_hard_regno_nregs): Make static. Take and return an
9460 (frv_class_max_nregs): Remove outdated copy of documentation.
9461 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
9462 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
9463 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
9464 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
9465 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
9466 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
9467 (TARGET_HARD_REGNO_NREGS): Redefine.
9468 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
9469 (CLASS_MAX_NREGS): Update comment.
9470 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
9471 (ia64_hard_regno_nregs): New function.
9472 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
9473 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
9474 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
9475 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
9476 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
9478 (m32c_hard_regno_nregs): Likewise. Make static.
9479 (TARGET_HARD_REGNO_NREGS): Redefine.
9480 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
9481 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
9482 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
9483 (m68k_hard_regno_nregs): New function.
9484 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
9485 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
9486 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
9487 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
9488 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
9489 Take and return an unsigned int.
9490 (TARGET_HARD_REGNO_NREGS): Redefine.
9491 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
9492 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
9493 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
9494 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
9495 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
9496 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
9497 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
9498 (msp430_hard_regno_nregs): Make static. Take and return an
9500 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
9501 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
9502 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
9503 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
9504 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
9505 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
9506 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
9507 (TARGET_HARD_REGNO_NREGS): Redefine.
9508 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
9509 (PA_HARD_REGNO_NREGS): ...this.
9510 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
9511 (PA_HARD_REGNO_NREGS): ...this.
9512 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
9513 (pa_hard_regno_nregs): New function.
9514 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
9515 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
9516 (pdp11_hard_regno_nregs): New function.
9517 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
9518 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
9519 (rs6000_hard_regno_nregs_hook): New function.
9520 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
9521 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
9522 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
9523 Take and return an unsigned int. Move earlier in file.
9524 (TARGET_HARD_REGNO_NREGS): Redefine.
9525 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
9526 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
9527 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
9528 (rl78_hard_regno_nregs): Make static. Take and return an
9530 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
9531 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
9532 (rs6000_hard_regno_nregs_hook): New function.
9533 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
9534 * config/rx/rx.c (rx_hard_regno_nregs): New function.
9535 (TARGET_HARD_REGNO_NREGS): Redefine.
9536 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
9537 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
9538 instead of HARD_REGNO_NREGS.
9539 (s390_hard_regno_nregs): New function.
9540 (s390_hard_regno_mode_ok): Add comment from s390.h.
9541 (TARGET_HARD_REGNO_NREGS): Redefine.
9542 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
9543 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
9544 (sh_hard_regno_nregs): New function.
9545 (sh_pass_in_reg_p): Use it.
9546 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
9547 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
9548 (sparc_hard_regno_nregs): New function.
9549 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
9550 * config/spu/spu.c (spu_hard_regno_nregs): New function.
9551 (spu_function_arg_advance): Use it, supplying a valid register number.
9552 (TARGET_HARD_REGNO_NREGS): Redefine.
9553 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
9554 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
9555 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
9556 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
9557 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
9558 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
9559 (CLASS_MAX_NREGS): Remove copy of old documentation.
9560 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
9561 (visium_hard_regno_nregs): New function.
9562 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
9563 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
9564 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
9565 xtensa_hard_regno_nregs): New function.
9566 * system.h (HARD_REGNO_NREGS): Poison.
9568 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9570 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
9571 hard_regno_nregs instead of HARD_REGNO_NREGS.
9572 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
9573 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
9574 (c6x_expand_epilogue): Likewise.
9575 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
9576 (frv_read_iacc_argument): Likewise.
9577 * config/sh/sh.c: Include regs.h.
9578 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
9579 (regs_used): Likewise.
9580 (output_stack_adjust): Likewise.
9581 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
9582 * expmed.c: Include regs.h.
9583 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
9584 * ree.c: Include regs.h.
9585 (combine_reaching_defs): Use hard_regno_nregs instead of
9587 (add_removable_extension): Likewise.
9589 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9591 * regs.h (hard_regno_nregs): Turn into a function.
9592 (end_hard_regno): Update accordingly.
9593 * caller-save.c (setup_save_areas): Likewise.
9594 (save_call_clobbered_regs): Likewise.
9595 (replace_reg_with_saved_mem): Likewise.
9596 (insert_restore): Likewise.
9597 (insert_save): Likewise.
9598 * combine.c (can_change_dest_mode): Likewise.
9599 (move_deaths): Likewise.
9600 (distribute_notes): Likewise.
9601 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
9602 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
9603 (rs6000_split_multireg_move): Likewise.
9604 (rs6000_register_move_cost): Likewise.
9605 (rs6000_memory_move_cost): Likewise.
9606 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
9607 (rs6000_split_multireg_move): Likewise.
9608 (rs6000_register_move_cost): Likewise.
9609 (rs6000_memory_move_cost): Likewise.
9610 * cselib.c (cselib_reset_table): Likewise.
9611 (cselib_lookup_1): Likewise.
9612 * emit-rtl.c (set_mode_and_regno): Likewise.
9613 * function.c (aggregate_value_p): Likewise.
9614 * ira-color.c (setup_profitable_hard_regs): Likewise.
9615 (check_hard_reg_p): Likewise.
9616 (calculate_saved_nregs): Likewise.
9617 (assign_hard_reg): Likewise.
9618 (improve_allocation): Likewise.
9619 (calculate_spill_cost): Likewise.
9620 * ira-emit.c (modify_move_list): Likewise.
9621 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
9622 (ira_hard_reg_in_set_p): Likewise.
9623 * ira.c (setup_reg_mode_hard_regset): Likewise.
9624 (clarify_prohibited_class_mode_regs): Likewise.
9625 (check_allocation): Likewise.
9626 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9627 (lra_setup_reg_renumber): Likewise.
9628 (setup_try_hard_regno_pseudos): Likewise.
9629 (spill_for): Likewise.
9630 (assign_hard_regno): Likewise.
9631 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
9632 * lra-constraints.c (in_class_p): Likewise.
9633 (lra_constraint_offset): Likewise.
9634 (simplify_operand_subreg): Likewise.
9635 (lra_constraints): Likewise.
9636 (split_reg): Likewise.
9637 (split_if_necessary): Likewise.
9638 (invariant_p): Likewise.
9639 (inherit_in_ebb): Likewise.
9640 * lra-lives.c (process_bb_lives): Likewise.
9641 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
9642 (get_hard_regs): Likewise.
9643 (do_remat): Likewise.
9644 * lra-spills.c (assign_spill_hard_regs): Likewise.
9645 * mode-switching.c (create_pre_exit): Likewise.
9646 * postreload.c (reload_combine_recognize_pattern): Likewise.
9647 * recog.c (peep2_find_free_register): Likewise.
9648 * regcprop.c (kill_value_regno): Likewise.
9649 (set_value_regno): Likewise.
9650 (copy_value): Likewise.
9651 (maybe_mode_change): Likewise.
9652 (find_oldest_value_reg): Likewise.
9653 (copyprop_hardreg_forward_1): Likewise.
9654 * regrename.c (check_new_reg_p): Likewise.
9655 (regrename_do_replace): Likewise.
9656 * reload.c (push_reload): Likewise.
9657 (combine_reloads): Likewise.
9658 (find_dummy_reload): Likewise.
9659 (operands_match_p): Likewise.
9660 (find_reloads): Likewise.
9661 (find_equiv_reg): Likewise.
9662 (reload_adjust_reg_for_mode): Likewise.
9663 * reload1.c (count_pseudo): Likewise.
9664 (count_spilled_pseudo): Likewise.
9665 (find_reg): Likewise.
9666 (clear_reload_reg_in_use): Likewise.
9667 (free_for_value_p): Likewise.
9668 (allocate_reload_reg): Likewise.
9669 (choose_reload_regs): Likewise.
9670 (reload_adjust_reg_for_temp): Likewise.
9671 (emit_reload_insns): Likewise.
9672 (delete_output_reload): Likewise.
9673 * rtlanal.c (subreg_get_info): Likewise.
9674 * sched-deps.c (sched_analyze_reg): Likewise.
9675 * sel-sched.c (init_regs_for_mode): Likewise.
9676 (mark_unavailable_hard_regs): Likewise.
9677 (choose_best_reg_1): Likewise.
9678 (verify_target_availability): Likewise.
9679 * valtrack.c (dead_debug_insert_temp): Likewise.
9680 * var-tracking.c (track_loc_p): Likewise.
9681 (emit_note_insn_var_location): Likewise.
9682 * varasm.c (make_decl_rtl): Likewise.
9683 * reginfo.c (choose_hard_reg_mode): Likewise.
9684 (init_reg_modes_target): Refer directly to
9685 this_target_regs->x_hard_regno_nregs.
9687 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9689 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
9690 instead of hard_regno_nregs.
9692 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9694 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
9695 end_hard_regno instead of hard_regno_nregs.
9696 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
9697 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
9698 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
9699 * ira-color.c (improve_allocation): Likewise.
9700 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9701 * lra-lives.c (mark_regno_live): Likewise.
9702 (mark_regno_dead): Likewise.
9703 * lra-remat.c (operand_to_remat): Likewise.
9704 * lra.c (collect_non_operand_hard_regs): Likewise.
9705 * postreload.c (reload_combine_note_store): Likewise.
9706 (move2add_valid_value_p): Likewise.
9707 * reload.c (regno_clobbered_p): Likewise.
9709 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9711 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
9713 * config/v850/v850.c (v850_reorg): Likewise.
9714 * reload.c (refers_to_regno_for_reload_p): Likewise.
9715 (find_equiv_reg): Likewise.
9716 * reload1.c (reload_reg_reaches_end_p): Likewise.
9718 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9720 * caller-save.c (add_used_regs): Use REG_NREGS instead of
9722 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
9723 * config/arm/arm.c (output_move_neon): Likewise.
9724 (arm_attr_length_move_neon): Likewise.
9725 (neon_split_vcombine): Likewise.
9726 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
9727 (c6x_mark_reg_written): Likewise.
9728 (c6x_dwarf_register_span): Likewise.
9729 * config/i386/i386.c (ix86_save_reg): Likewise.
9730 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
9731 (rws_access_reg): Likewise.
9732 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
9733 * mode-switching.c (create_pre_exit): Likewise.
9734 * ree.c (combine_reaching_defs): Likewise.
9735 (add_removable_extension): Likewise.
9736 * regcprop.c (find_oldest_value_reg): Likewise.
9737 (copyprop_hardreg_forward_1): Likewise.
9738 * reload.c (reload_inner_reg_of_subreg): Likewise.
9739 (push_reload): Likewise.
9740 (combine_reloads): Likewise.
9741 (find_dummy_reload): Likewise.
9742 (reload_adjust_reg_for_mode): Likewise.
9743 * reload1.c (find_reload_regs): Likewise.
9744 (forget_old_reloads_1): Likewise.
9745 (reload_reg_free_for_value_p): Likewise.
9746 (reload_adjust_reg_for_temp): Likewise.
9747 (emit_reload_insns): Likewise.
9748 (delete_output_reload): Likewise.
9749 * sel-sched.c (choose_best_reg_1): Likewise.
9750 (choose_best_pseudo_reg): Likewise.
9752 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9753 Alan Hayward <alan.hayward@arm.com>
9754 David Sherwood <david.sherwood@arm.com>
9756 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
9757 * target.def (slow_unaligned_access): New hook.
9758 * targhooks.h (default_slow_unaligned_access): Declare.
9759 * targhooks.c (default_slow_unaligned_access): New function.
9760 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
9761 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
9762 * doc/tm.texi: Regenerate.
9763 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
9764 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
9765 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
9767 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
9768 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
9770 (rs6000_slow_unaligned_access): New function.
9771 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
9772 (expand_block_compare): Likewise.
9773 (expand_strn_compare): Likewise.
9774 (rs6000_rtx_costs): Likewise.
9775 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
9776 (riscv_slow_unaligned_access): Likewise.
9777 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
9778 (riscv_slow_unaligned_access_p): ...this and make static.
9779 (riscv_option_override): Update accordingly.
9780 (riscv_slow_unaligned_access): New function.
9781 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
9782 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
9783 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
9784 (rs6000_slow_unaligned_access): New function.
9785 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
9786 (rs6000_rtx_costs): Likewise.
9787 * config/rs6000/rs6000-string.c (expand_block_compare)
9788 (expand_strn_compare): Use targetm.slow_unaligned_access instead
9789 of SLOW_UNALIGNED_ACCESS.
9790 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
9791 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
9792 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
9793 of SLOW_UNALIGNED_ACCESS.
9794 * expmed.c (simple_mem_bitfield_p): Likewise.
9795 * expr.c (alignment_for_piecewise_move): Likewise.
9796 (emit_group_load_1): Likewise.
9797 (emit_group_store): Likewise.
9798 (copy_blkmode_from_reg): Likewise.
9799 (emit_push_insn): Likewise.
9800 (expand_assignment): Likewise.
9801 (store_field): Likewise.
9802 (expand_expr_real_1): Likewise.
9803 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
9804 * lra-constraints.c (simplify_operand_subreg): Likewise.
9805 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
9806 * gimple-ssa-store-merging.c: Likewise in block comment at start
9808 * tree-ssa-strlen.c: Include target.h.
9809 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
9810 of SLOW_UNALIGNED_ACCESS.
9811 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
9813 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9815 PR rtl-optimization/82185
9816 * expmed.c (emit_store_flag_int): Only test tem if it has been
9819 2017-09-12 Richard Biener <rguenther@suse.de>
9822 * match.pd ((FTYPE) N CMP CST): Fix typo.
9824 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
9826 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
9828 (mips_near_type_p): Add 'short_call' attribute as a synonym
9830 * doc/extend.texi (short_call): Document new function attribute.
9832 2017-09-12 Jakub Jelinek <jakub@redhat.com>
9835 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
9836 assertion check that in the condition.
9837 (get_atomic_generic_size): Likewise. Before testing if parameter
9838 has pointer type, if it has array type, call for C++
9839 default_conversion to perform array-to-pointer conversion.
9841 2017-09-12 Richard Biener <rguenther@suse.de>
9843 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
9844 for operations we cannot scalarize.
9846 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
9848 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
9849 vectors heap vectors. Clean up comments.
9850 Make visited_bbs a reference.
9851 (profitable_jump_thread_path): Make GC
9852 vectors heap vectors. Clean up comments.
9854 (convert_and_register_jump_thread_path): Make GC vectors heap
9856 (check_subpath_and_update_thread_path): Same. Clean up comments.
9857 Make visited_bbs a reference.
9858 (handle_phi): Abstract common code to to
9859 register_jump_thread_path_if_profitable.
9860 Rename VAR_BB to DEF_BB.
9862 Make GC vectors heap vectors.
9863 Make visited_bbs a reference.
9864 (handle_assignment): Same.
9865 (register_jump_thread_path_if_profitable): New.
9866 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
9868 Make GC vectors heap vectors. Clean up comments.
9869 Make visited_bbs a reference.
9870 (find_jump_threads_backwards): Make visited_bbs live in the stack.
9871 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
9874 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
9877 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
9878 words of E_DImode object are reachable by xtensa_uimm8x4 access.
9880 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
9882 Revert r251800 and r251799.
9884 2017-09-11 Martin Jambor <mjambor@suse.cz>
9887 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
9888 arguments in advance.
9889 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
9890 use it to find predecessor edges.
9891 (naive_outof_ssa): Collect vector of predecessors.
9893 2017-09-08 Jason Merrill <jason@redhat.com>
9895 PR c++/70029 - ICE with ref-qualifier and -flto
9896 * langhooks.h (struct lang_hooks_for_types): Add
9897 copy_lang_qualifiers.
9898 * attribs.c (build_type_attribute_qual_variant): Use it.
9899 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
9901 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
9902 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
9904 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
9907 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
9908 (*mulsi3_sp64): New instruction.
9909 (mulsi3): New expander.
9911 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
9913 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
9915 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9917 * sancov.c: Include memmodel.h.
9919 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
9922 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
9925 2017-09-07 Carl Love <cel@us.ibm.com>
9927 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
9928 the sldi instruction.
9930 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
9932 * sancov.c: Include tm_p.h.
9934 2017-09-07 Jakub Jelinek <jakub@redhat.com>
9937 * output.h (switch_to_other_text_partition): New declaration.
9938 * varasm.c (switch_to_other_text_partition): New function.
9939 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
9940 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
9941 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
9942 to the other text partition before emitting LCL label and switch back
9943 after emitting the word after it.
9945 2017-09-07 Richard Biener <rguenther@suse.de>
9947 * passes.def (pass_split_crit_edges): Remove instance before PRE.
9948 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
9949 critical edges here, after loop init.
9950 (pass_data_pre): Remove PROP_no_crit_edges flags.
9951 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
9952 for valueization of call args to avoid leaking VN_TOP.
9953 (visit_use): Assert we do not visit default defs.
9954 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
9955 Use error_mark_node to more easily detect leaking VN_TOP.
9956 All default-defs are varying, not VN_TOP. Mark them visited.
9957 (run_scc_vn): Make code match comment.
9959 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
9961 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
9962 OPTION_MASK_FLOAT128_KEYWORD.
9963 (POWERPC_MASKS): Likewise.
9964 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
9965 support for the -mfloat128-type option, and make -mfloat128
9966 default on PowerPC Linux systems. Define or undefine
9967 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
9968 Define __float128 to be __ieee128 if IEEE 128-bit support is
9969 enabled, or undefine it.
9970 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
9971 Delete defining __FLOAT128_TYPE__.
9972 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
9973 -mfloat128-type option and make -mfloat128 default on PowerPC
9975 (TARGET_FLOAT128_TYPE): Likewise.
9976 (-mfloat128-type): Likewise.
9977 * config/rs6000/rs6000.c (rs6000_option_override_internal):
9978 Delete the -mfloat128-type option and make -mfloat128 default on
9979 PowerPC Linux systems. Always use __ieee128 to be the keyword for
9980 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
9981 128-bit floating point is enabled. Change tests from using
9982 -mfloat128-type to -mfloat128.
9983 (rs6000_mangle_type): Use the correct mangling for the __float128
9984 type even if normal long double is restricted to 64-bits.
9985 (floatn_mode): Enable the _Float128 type by default on VSX Linux
9987 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
9988 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
9989 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
9991 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
9992 documentation for -mfloat128.
9994 2017-09-06 Olivier Hainque <hainque@adacore.com>
9996 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
9998 2017-09-06 Wish Wu <wishwu007@gmail.com>
9999 Jakub Jelinek <jakub@redhat.com>
10001 * asan.c (initialize_sanitizer_builtins): Add
10002 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
10003 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
10004 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
10005 BT_FN_VOID_UINT64_PTR variables.
10006 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
10007 (BT_FN_VOID_UINT16_UINT16): Likewise.
10008 (BT_FN_VOID_UINT32_UINT32): Likewise.
10009 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
10010 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
10011 (BT_FN_VOID_UINT64_PTR): Likewise.
10012 * common.opt (flag_sanitize_coverage): New variable.
10013 (fsanitize-coverage=trace-pc): Remove.
10014 (fsanitize-coverage=): Add.
10015 * flag-types.h (enum sanitize_coverage_code): New enum.
10016 * fold-const.c (fold_range_test): Disable non-short-circuit
10017 optimization if flag_sanitize_coverage.
10018 (fold_truth_andor): Likewise.
10019 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
10020 * opts.c (COVERAGE_SANITIZER_OPT): Define.
10021 (coverage_sanitizer_opts): New array.
10022 (get_closest_sanitizer_option): Add OPTS argument, handle also
10023 OPT_fsanitize_coverage_.
10024 (parse_sanitizer_options): Adjusted to also handle
10025 OPT_fsanitize_coverage_.
10026 (common_handle_option): Add OPT_fsanitize_coverage_.
10027 * sancov.c (instrument_comparison, instrument_switch): New function.
10028 (sancov_pass): Add trace-cmp support.
10029 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
10030 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
10031 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
10032 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
10033 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
10034 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
10035 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
10036 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
10037 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
10038 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
10040 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
10042 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
10043 error. Only quit immediately if parsing is complete.
10044 (BEGIN): Initialize fatal_err and parse_done.
10045 (begin fpu, end fpu): Check number of arguments.
10046 (begin arch, end arch): Likewise.
10047 (begin cpu, end cpu): Likewise.
10048 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
10049 (optalias): Likewise.
10051 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
10053 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
10054 * config/arm/arm-isa.h: Delete. Move definitions to ...
10055 * arm-cpus.in: ... here. Use new feature and fgroup values.
10056 * config/arm/arm.c (arm_option_override): Use lower case for feature
10058 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
10059 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
10060 * config/arm/parsecpu.awk (END): Add new command 'isa'.
10062 (print_isa_bits_for): New function.
10063 (gen_isa): New function.
10064 (gen_comm_data): Use print_isa_bits_for.
10065 (define feature): New keyword.
10066 (define fgroup): New keyword.
10067 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
10068 (arm-isa.h): Add rule to generate file.
10069 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
10070 case for feature bit names.
10072 2017-09-06 Richard Biener <rguenther@suse.de>
10074 * tree-ssa-pre.c (NECESSARY): Remove.
10075 (create_expression_by_pieces): Do not touch pass-local flags.
10076 (insert_into_preds_of_block): Likewise.
10077 (do_pre_regular_insertion): Likewise.
10078 (eliminate_insert): Likewise.
10079 (eliminate_dom_walker::before_dom_children): Likewise.
10080 (fini_eliminate): Do not look at inserted_exprs.
10081 (mark_operand_necessary): Remove.
10082 (remove_dead_inserted_code): Replace with simple work-list
10083 algorithm based on inserted_exprs and SSA uses.
10084 (pass_pre::execute): Re-order fini_eliminate and
10085 remove_dead_inserted_code.
10087 2017-09-06 Olivier Hainque <hainque@adacore.com>
10089 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
10090 for VxWorks 7. Adjust surrounding comments.
10092 2017-09-06 Richard Biener <rguenther@suse.de>
10094 * gimple-ssa-strength-reduction.c
10095 (find_candidates_dom_walker::before_dom_children): Also allow
10098 2017-09-06 Richard Biener <rguenther@suse.de>
10100 PR tree-optimization/82108
10101 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
10102 for gap in the non-permutation SLP case.
10104 2017-09-06 Martin Jambor <mjambor@suse.cz>
10106 PR tree-optimization/82078
10107 * tree-sra.c (sort_and_splice_var_accesses): Move call to
10108 add_access_to_work_queue...
10109 (build_accesses_from_assign): ...here.
10110 (propagate_all_subaccesses): Make sure racc is the group
10111 representative, if there is one.
10113 2017-09-06 Jakub Jelinek <jakub@redhat.com>
10115 PR middle-end/82095
10116 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
10119 2017-09-06 Richard Biener <rguenther@suse.de>
10121 * gimple-ssa-strength-reduction.c
10122 (find_candidates_dom_walker::before_doom_children): Use a
10123 type and not a mode check.
10125 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
10128 * config/arm/predicates.md (arm_general_adddi_operand): Create new
10130 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
10132 2017-09-05 Jeff Law <law@redhat.com>
10134 PR tree-optimization/64910
10135 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
10136 cases where we have 3 or more operands.
10138 2017-09-05 Jakub Jelinek <jakub@redhat.com>
10140 PR middle-end/81768
10141 * omp-low.c (lower_omp_for): Recompute tree invariant if
10142 gimple_omp_for_initial/final is ADDR_EXPR.
10144 PR middle-end/81768
10145 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
10146 into gimple val before gimplification fo the COND_EXPR.
10148 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
10150 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
10151 REGION_COPY argument.
10152 (thread_through_all_blocks): Remove unused argument to
10153 duplicate_thread_path.
10155 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10156 Alan Hayward <alan.hayward@arm.com>
10157 David Sherwood <david.sherwood@arm.com>
10159 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
10160 Take a scalar_mode rather than a machine_mode.
10161 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10162 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
10163 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10164 (aarch64_gen_adjusted_ldpstp): Likewise.
10165 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
10167 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10168 Alan Hayward <alan.hayward@arm.com>
10169 David Sherwood <david.sherwood@arm.com>
10171 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
10172 Take a scalar_int_mode instead of a machine_mode.
10173 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
10174 (aarch64_output_scalar_simd_mov_immediate): Likewise.
10175 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
10176 (aarch64_simd_attr_length_rglist): Delete.
10177 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
10178 a scalar_int_mode instead of a machine_mode.
10179 (aarch64_add_offset): Likewise.
10180 (aarch64_internal_mov_immediate): Likewise
10181 (aarch64_add_constant_internal): Likewise.
10182 (aarch64_add_constant): Likewise.
10183 (aarch64_movw_imm): Likewise.
10184 (aarch64_rtx_arith_op_extract_p): Likewise.
10185 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
10186 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
10187 Remove assert that the mode isn't a vector.
10188 (aarch64_output_scalar_simd_mov_immediate): Likewise.
10189 (aarch64_expand_mov_immediate): Update calls after above changes.
10190 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
10191 (aarch64_and_bitmask_imm): Check for scalar integer modes.
10192 (aarch64_move_imm): Likewise.
10193 (aarch64_can_const_movi_rtx_p): Likewise.
10194 (aarch64_strip_extend): Likewise.
10195 (aarch64_extr_rtx_p): Likewise.
10196 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
10197 a CONST_INT when the mode parameter is VOIDmode.
10198 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
10200 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10202 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
10203 * stor-layout.c (bitwise_mode_for_mode): Likewise.
10204 (bitwise_type_for_mode): Update accordingly.
10206 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10208 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
10209 * stor-layout.c (mode_for_size_tree): Likewise.
10210 (mode_for_array): Update accordingly.
10211 (layout_decl): Likewise.
10212 (compute_record_mode): Likewise. Only set the mode once.
10214 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10216 * target.def (get_mask_mode): Change return type to opt_mode.
10218 * doc/tm.texi: Regenerate.
10219 * targhooks.h (default_get_mask_mode): Return an opt_mode.
10220 * targhooks.c (default_get_mask_mode): Likewise.
10221 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
10222 * optabs-query.c (can_vec_mask_load_store_p): Update use of
10223 targetm.get_mask_mode.
10224 * tree.c (build_truth_vector_type): Likewise.
10226 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10228 * machmode.h (mode_for_vector): Return an opt_mode.
10229 * stor-layout.c (mode_for_vector): Likewise.
10230 (mode_for_int_vector): Update accordingly.
10231 (layout_type): Likewise.
10232 * config/i386/i386.c (emit_memmov): Likewise.
10233 (ix86_expand_set_or_movmem): Likewise.
10234 (ix86_expand_vector_init): Likewise.
10235 (ix86_get_mask_mode): Likewise.
10236 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
10238 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
10239 * expmed.c (extract_bit_field_1): Likewise.
10240 * expr.c (expand_expr_real_2): Likewise.
10241 * optabs-query.c (can_vec_perm_p): Likewise.
10242 (can_vec_mask_load_store_p): Likewise.
10243 * optabs.c (expand_vec_perm): Likewise.
10244 * targhooks.c (default_get_mask_mode): Likewise.
10245 * tree-vect-stmts.c (vectorizable_store): Likewise.
10246 (vectorizable_load): Likewise.
10247 (get_vectype_for_scalar_type_and_size): Likewise.
10249 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10251 * machmode.h (mode_for_int_vector): New function.
10252 * stor-layout.c (mode_for_int_vector): Likewise.
10253 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
10254 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
10255 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
10256 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
10257 (s390_expand_vcond): Likewise.
10259 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10261 * machmode.h (opt_machine_mode): New type.
10262 (opt_mode<T>): Allow construction from anything that can be
10264 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
10265 (mode_for_size): Return an opt_machine_mode.
10266 * stor-layout.c (mode_for_size): Likewise.
10267 (mode_for_size_tree): Update call accordingly.
10268 (bitwise_mode_for_mode): Likewise.
10269 (make_fract_type): Likewise.
10270 (make_accum_type): Likewise.
10271 * caller-save.c (replace_reg_with_saved_mem): Update call
10273 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10274 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10275 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10276 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10277 * expmed.c (extract_bit_field_1): Likewise.
10278 * reload.c (get_secondary_mem): Likewise.
10279 * varasm.c (assemble_integer): Likewise.
10280 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
10283 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10285 * machmode.h (decimal_float_mode_for_size): New function.
10286 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
10287 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
10288 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
10289 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
10290 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
10291 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
10293 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10295 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
10296 (get_builtin_sync_mode): Likewise.
10297 (expand_ifn_atomic_compare_exchange): Likewise.
10298 (expand_builtin_atomic_clear): Likewise.
10299 (expand_builtin_atomic_test_and_set): Likewise.
10300 (fold_builtin_atomic_always_lock_free): Likewise.
10301 * calls.c (compute_argument_addresses): Likewise.
10302 (emit_library_call_value_1): Likewise.
10303 (store_one_arg): Likewise.
10304 * combine.c (combine_instructions): Likewise.
10305 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
10306 * config/arm/arm.c (arm_function_value): Likewise.
10307 (aapcs_allocate_return_reg): Likewise.
10308 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
10309 * config/i386/i386.c (construct_container): Likewise.
10310 (ix86_gimplify_va_arg): Likewise.
10311 (ix86_expand_sse_cmp): Likewise.
10312 (emit_memmov): Likewise.
10313 (emit_memset): Likewise.
10314 (expand_small_movmem_or_setmem): Likewise.
10315 (ix86_expand_pextr): Likewise.
10316 (ix86_expand_pinsr): Likewise.
10317 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
10318 * config/microblaze/microblaze.c (microblaze_block_move_straight):
10320 * config/mips/mips.c (mips_function_value_1) Likewise.
10321 (mips_block_move_straight): Likewise.
10322 (mips_expand_ins_as_unaligned_store): Likewise.
10323 * config/powerpcspe/powerpcspe.c
10324 (rs6000_darwin64_record_arg_advance_flush): Likewise.
10325 (rs6000_darwin64_record_arg_flush): Likewise.
10326 * config/rs6000/rs6000.c
10327 (rs6000_darwin64_record_arg_advance_flush): Likewise.
10328 (rs6000_darwin64_record_arg_flush): Likewise.
10329 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
10330 (sparc_function_value_1): Likewise.
10331 * config/spu/spu.c (adjust_operand): Likewise.
10332 (spu_emit_branch_or_set): Likewise.
10333 (arith_immediate_p): Likewise.
10334 * emit-rtl.c (gen_lowpart_common): Likewise.
10335 * expr.c (expand_expr_real_1): Likewise.
10336 * function.c (assign_parm_setup_block): Likewise.
10337 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
10338 * reload1.c (alter_reg): Likewise.
10339 * stor-layout.c (mode_for_vector): Likewise.
10340 (layout_type): Likewise.
10342 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10344 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
10345 (spu_convert_move): Likewise.
10346 * lower-subreg.c (resolve_simple_move): Likewise.
10348 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10351 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
10352 define_insn to a define_expand.
10353 (altivec_vsum2sws_direct): New define_insn.
10354 (altivec_vsumsws): Convert from a define_insn to a define_expand.
10356 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
10358 * config/arm/arm.c (arm_option_params_internal): Improve setting of
10361 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
10366 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
10367 realign stack if stack alignment needed is less than incoming
10370 2017-09-05 Marek Polacek <polacek@redhat.com>
10373 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
10376 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
10378 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
10380 2017-09-05 Richard Biener <rguenther@suse.de>
10382 PR tree-optimization/82084
10383 * fold-const.c (can_native_encode_string_p): Handle wide characters.
10385 2017-09-05 Richard Biener <rguenther@suse.de>
10387 PR tree-optimization/82102
10388 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
10390 2017-09-05 Martin Liska <mliska@suse.cz>
10392 PR tree-optimization/82032
10393 * tree-cfg.c (generate_range_test): New function.
10394 * tree-cfg.h (generate_range_test): Declared here.
10395 * tree-cfgcleanup.c (convert_single_case_switch): New function.
10396 (cleanup_control_expr_graph): Use it.
10397 * tree-switch-conversion.c (try_switch_expansion): Remove
10399 (emit_case_nodes): Use generate_range_test.
10401 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
10404 * config/i386/i386.md (*<btsc><mode>_mask): Add
10405 TARGET_USE_BT to insn constraint.
10406 (*btr<mode>_mask): Ditto.
10408 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
10410 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
10411 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
10413 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
10416 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
10417 TARGET_NEON and TARGET_IWMMXT.
10418 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
10419 TARGET_NEON and TARGET_IWMMXT.
10420 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
10422 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
10424 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
10425 (ix86_rewrite_tls_address): Ditto.
10426 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
10427 (ix86_rewrite_tls_address_1): Ditto.
10428 (ix86_rewrite_tls_address): Ditto.
10429 * config/i386/predicates.md (tls_address_pattern): New predicate.
10430 * config/i386/i386.md (TLS address splitter): New splitter.
10432 2017-09-04 Richard Biener <rguenther@suse.de>
10434 PR tree-optimization/82084
10435 * fold-const.h (can_native_encode_string_p): Declare.
10436 * fold-const.c (can_native_encode_string_p): Factor out from ...
10437 (native_encode_string): ... here.
10438 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
10439 vectorizing stores from constants we later cannot handle.
10441 2017-09-04 Marek Polacek <polacek@redhat.com>
10444 * doc/invoke.texi: Update -Wtautological-compare documentation.
10446 2017-09-04 Jeff Law <law@redhat.com>
10448 PR tree-optimization/64910
10449 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
10450 swap the first and last operand if the last is a constant.
10452 2017-09-04 Marek Polacek <polacek@redhat.com>
10455 * convert.c (do_narrow): When sanitizing signed integer overflows,
10456 bail out for signed types.
10457 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
10459 2017-09-04 Richard Biener <rguenther@suse.de>
10461 PR tree-optimization/82060
10462 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
10463 Move devirtualization after stmt folding and before EH/AB/noreturn
10464 cleanup to get the stmt refs canonicalized. Use a bool instead
10465 of gimple_modified_p since that doesn't work for NOPs. Schedule
10466 NOPs generated by folding for removal.
10468 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10469 Alan Hayward <alan.hayward@arm.com>
10470 David Sherwood <david.sherwood@arm.com>
10472 * coretypes.h (pad_direction): New enum.
10473 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
10474 (FUNCTION_ARG_PADDING): Likewise.
10475 * target.def (function_arg_padding): New hook.
10476 * targhooks.h (default_function_arg_padding): Declare.
10477 * targhooks.c (default_function_arg_padding): New function.
10478 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
10479 (TARGET_FUNCTION_ARG_PADDING): ...this.
10480 * doc/tm.texi: Regenerate.
10481 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
10482 instead of direction.
10483 (compute_argument_addresses): Likewise.
10484 (load_register_parameters): Likewise.
10485 (emit_library_call_value_1): Likewise.
10486 (store_one_arg): Use targetm.calls.function_arg_padding instead
10487 of FUNCTION_ARG_PADDING.
10488 (must_pass_in_stack_var_size_or_pad): Likewise.
10489 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
10490 (emit_group_store): Likewise.
10491 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
10492 instead of FUNCTION_ARG_PADDING.
10493 (emit_push_insn): Likewise, and propagate enum change throughout
10495 * function.h (direction): Delete.
10496 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
10498 * function.c (assign_parm_find_stack_rtl): Likewise.
10499 (assign_parm_setup_block_p): Likewise.
10500 (assign_parm_setup_block): Likewise.
10501 (gimplify_parameters): Likewise.
10502 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
10503 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
10505 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
10506 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10507 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
10508 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
10509 (aarch64_function_arg_padding): ...this new function.
10510 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
10511 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10512 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
10513 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10514 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
10515 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10516 (arm_pad_arg_upward): Replace with...
10517 (arm_function_arg_padding): ...this new function.
10518 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
10520 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
10521 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
10522 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10523 (ia64_hpux_function_arg_padding): Replace with...
10524 (ia64_function_arg_padding): ...this new function. Use pad_direction
10525 instead of direction. Check for TARGET_HPUX.
10526 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
10527 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10528 (iq2000_function_arg_padding): New function.
10529 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
10530 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
10531 (mips_function_arg_padding): ...this new function.
10532 (mips_pad_reg_upward): Update accordingly.
10533 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10534 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
10535 targetm.calls.function_arg_padding.
10536 (FUNCTION_ARG_PADDING): Delete.
10537 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10538 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
10539 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10540 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
10541 (nios2_block_reg_padding): Return pad_direction instead of direction.
10542 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
10543 instead of direction.
10544 (nios2_function_arg_padding): Likewise. Make static.
10545 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10546 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
10547 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
10548 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
10549 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10550 (pa_function_arg_padding): Make static. Return pad_direction instead
10552 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
10553 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10554 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
10555 instead of direction. Use targetm.calls.function_arg_padding.
10556 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
10557 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
10558 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
10559 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
10560 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
10562 (function_arg_padding): Rename to...
10563 (rs6000_function_arg_padding): ...this. Make static. Return
10564 pad_direction instead of direction.
10565 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
10566 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
10567 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10568 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
10569 instead of direction. Use targetm.calls.function_arg_padding.
10570 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
10571 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
10572 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
10573 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
10574 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10575 (function_arg_padding): Rename to...
10576 (rs6000_function_arg_padding): ...this. Make static. Return
10577 pad_direction instead of direction.
10578 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
10579 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
10580 * config/s390/s390.c (s390_function_arg_padding): New function.
10581 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10582 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
10583 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
10584 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10585 (function_arg_padding): Rename to...
10586 (sparc_function_arg_padding): ...this. Make static. Return
10587 pad_direction instead of direction.
10588 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
10589 * config/spu/spu.c (spu_function_arg_padding): New function.
10590 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10591 * system.h (FUNCTION_ARG_PADDING): Poison.
10593 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10594 Alan Hayward <alan.hayward@arm.com>
10595 David Sherwood <david.sherwood@arm.com>
10597 * target.def (modes_tieable_p): New hook.
10598 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
10599 (TARGET_MODES_TIEABLE_P): ...this.
10600 * doc/tm.texi.in: Regenerate.
10601 * hooks.h (hook_bool_mode_mode_true): Declare.
10602 * hooks.c (hook_bool_mode_mode_true): New function.
10603 * combine.c (subst): Use targetm.modes_tieable_p instead of
10605 * dse.c (find_shift_sequence): Likewise.
10606 * expmed.c (extract_low_bits): Likewise.
10607 * lower-subreg.c: Include target.h.
10608 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
10610 * rtlanal.c (rtx_cost): Likewise.
10611 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
10612 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
10613 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
10614 (TARGET_MODES_TIEABLE_P): Redefine.
10615 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
10616 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
10617 (TARGET_MODES_TIEABLE_P): Redefine.
10618 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
10619 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
10620 (arc_modes_tieable_p): New function.
10621 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
10622 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
10623 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
10624 (arm_modes_tieable_p): Make static.
10625 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
10626 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
10627 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
10628 (TARGET_MODES_TIEABLE_P): Redefine.
10629 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
10630 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
10631 (TARGET_MODES_TIEABLE_P): Redefine.
10632 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
10633 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
10634 (cr16_modes_tieable_p): New function.
10635 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
10636 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
10637 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
10638 (TRULY_NOOP_TRUNCATION): Update comment.
10639 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
10640 (TRULY_NOOP_TRUNCATION): Update comment.
10641 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
10642 (frv_modes_tieable_p): New function.
10643 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
10644 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
10645 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
10646 (TARGET_MODES_TIEABLE_P): Redefine.
10647 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
10648 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
10649 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
10650 (TARGET_MODES_TIEABLE_P): Redefine.
10651 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
10652 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
10653 (ia64_modes_tieable_p): New function.
10654 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
10655 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
10656 (iq2000_modes_tieable_p): New function.
10657 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
10658 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
10659 (lm32_modes_tieable_p): New function.
10660 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
10661 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
10662 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
10663 (TARGET_MODES_TIEABLE_P): Redefine.
10664 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
10665 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
10666 (m32r_modes_tieable_p): New function.
10667 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
10668 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
10669 (m68k_modes_tieable_p): New function.
10670 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
10671 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
10672 (mcore_modes_tieable_p): New function.
10673 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
10674 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
10676 (TARGET_MODES_TIEABLE_P): Redefine.
10677 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
10678 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
10679 * config/mips/mips.c (mips_modes_tieable_p): Make static.
10680 (TARGET_MODES_TIEABLE_P): Redefine.
10681 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
10682 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
10683 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
10684 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
10685 (mn10300_modes_tieable_p): ...this and make static.
10686 (TARGET_MODES_TIEABLE_P): Redefine.
10687 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
10688 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
10689 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
10690 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
10691 (msp430_modes_tieable_p): Make static.
10692 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
10693 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
10694 (TARGET_MODES_TIEABLE_P): Redefine.
10695 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
10696 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
10697 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
10698 (TARGET_MODES_TIEABLE_P): Redefine.
10699 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
10700 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
10701 * config/pa/pa.c (pa_modes_tieable_p): Make static.
10702 (TARGET_MODES_TIEABLE_P): Redefine.
10703 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
10704 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
10705 (pdp11_modes_tieable_p): New function.
10706 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
10707 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
10708 (rs6000_modes_tieable_p): New function.
10709 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
10710 * config/powerpcspe/powerpcspe.md: Update comment.
10711 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
10712 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
10713 (TARGET_MODES_TIEABLE_P): Redefine.
10714 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
10715 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
10716 (rl78_modes_tieable_p): New function.
10717 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
10718 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
10719 (rs6000_modes_tieable_p): New function.
10720 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
10721 * config/rs6000/rs6000.md: Update comment.
10722 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
10723 * config/rx/rx.c (rx_modes_tieable_p): New function.
10724 (TARGET_MODES_TIEABLE_P): Redefine.
10725 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
10726 * config/s390/s390.c (s390_modes_tieable_p): New function.
10727 (TARGET_MODES_TIEABLE_P): Redefine.
10728 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
10729 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
10730 (sh_modes_tieable_p): New function.
10731 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
10732 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
10733 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
10734 (sparc_modes_tieable_p): Make static.
10735 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
10736 * config/spu/spu.c (spu_modes_tieable_p): New function.
10737 (TARGET_MODES_TIEABLE_P): Redefine.
10738 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
10739 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
10740 (TARGET_MODES_TIEABLE_P): Redefine.
10741 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
10742 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
10743 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
10744 * config/v850/v850.c (v850_modes_tieable_p): New function.
10745 (TARGET_MODES_TIEABLE_P): Redefine.
10746 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
10747 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
10748 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
10749 (visium_modes_tieable_p): New function.
10750 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
10751 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
10752 (xtensa_modes_tieable_p): New function.
10753 * system.h (MODES_TIEABLE_P): Poison.
10755 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10756 Alan Hayward <alan.hayward@arm.com>
10757 David Sherwood <david.sherwood@arm.com>
10759 * target.def (hard_regno_mode_ok): New hook.
10760 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
10761 (TARGET_HARD_REGNO_MODE_OK): ...this.
10762 * doc/tm.texi.in: Regenerate.
10763 * hooks.h (hook_bool_uint_mode_true): Declare.
10764 * hooks.c (hook_bool_uint_mode_true): New function.
10765 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
10766 HARD_REGNO_MODE_OK.
10767 * genpreds.c (write_insn_preds_c): Add an include of target.h.
10768 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
10769 instead of HARD_REGNO_MODE_OK.
10770 * caller-save.c: Include target.h.
10771 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
10772 HARD_REGNO_MODE_OK.
10773 * combine.c (can_combine_p): Likewise.
10774 (combinable_i3pat): Likewise.
10775 (can_change_dest_mode): Likewise.
10776 * expr.c (init_expr_target): Likewise.
10777 (convert_move): Likewise.
10778 (convert_modes): Likewise.
10779 * ira.c (setup_prohibited_class_mode_regs): Likewise.
10780 (setup_prohibited_mode_move_regs): Likewise.
10781 * ira.h (target_ira): Likewise.
10782 * lra-assigns.c (find_hard_regno_for_1): Likewise.
10783 * lra-constraints.c (process_alt_operands): Likewise.
10784 (split_reg): Likewise.
10785 * recog.c (peep2_find_free_register): Likewise.
10786 * ree.c (combine_reaching_defs): Likewise.
10787 * regcprop.c (maybe_mode_change): Likewise.
10788 * reginfo.c (init_reg_sets_1): Likewise.
10789 (choose_hard_reg_mode): Likewise.
10790 (simplifiable_subregs): Likewise.
10791 * regrename.c (check_new_reg_p): Likewise.
10792 * reload.c (find_valid_class): Likewise.
10793 (find_valid_class_1): Likewise.
10794 (reload_inner_reg_of_subreg): Likewise.
10795 (push_reload): Likewise.
10796 (combine_reloads): Likewise.
10797 (find_dummy_reload): Likewise.
10798 (find_reloads): Likewise.
10799 * reload1.c (find_reg): Likewise.
10800 (set_reload_reg): Likewise.
10801 (allocate_reload_reg): Likewise.
10802 (choose_reload_regs): Likewise.
10803 (reload_adjust_reg_for_temp): Likewise.
10804 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
10805 (simplify_subreg_regno): Likewise.
10806 * sel-sched.c (init_regs_for_mode): Likewise.
10807 * varasm.c (make_decl_rtl): Likewise.
10808 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
10809 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
10810 HARD_REGNO_MODE_OK.
10811 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
10812 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
10813 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10814 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
10815 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
10816 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10817 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
10818 (arc_mode_class): Delete.
10819 (HARD_REGNO_MODE_OK): Delete.
10820 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10821 (arc_hard_regno_mode_ok): Rename old array to...
10822 (arc_hard_regno_mode_ok_modes): ...this.
10823 (arc_conditional_register_usage): Update accordingly.
10824 (arc_mode_class): Make static.
10825 (arc_hard_regno_mode_ok): New function.
10826 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
10827 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
10828 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10829 (arm_hard_regno_mode_ok): Make static.
10830 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
10831 HARD_REGNO_MODE_OK.
10832 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
10833 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
10834 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
10836 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10837 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
10838 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
10839 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
10840 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
10841 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10842 * config/bfin/predicates.md (valid_reg_operand): Use
10843 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
10844 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
10845 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
10846 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10847 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
10848 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
10849 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10850 (cr16_hard_regno_mode_ok): Make static and return a bool.
10851 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
10852 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10853 (cris_hard_regno_mode_ok): New function.
10854 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
10855 (epiphany_mode_class): Delete.
10856 (HARD_REGNO_MODE_OK): Delete.
10857 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
10858 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10859 (hard_regno_mode_ok): Rename to...
10860 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
10861 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
10862 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
10863 HARD_REGNO_MODE_OK.
10864 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
10865 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
10866 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10867 (frv_hard_regno_mode_ok): Make static and return a bool.
10868 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
10869 HARD_REGNO_MODE_OK.
10870 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
10871 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
10872 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
10873 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
10875 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10876 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
10877 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
10878 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
10880 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10881 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
10882 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10883 (ia64_hard_regno_mode_ok): New function.
10884 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
10885 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10886 (iq2000_hard_regno_mode_ok): New function.
10887 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
10888 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10889 (lm32_hard_regno_mode_ok): New function.
10890 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
10891 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
10892 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
10893 instead of HARD_REGNO_MODE_OK.
10894 (m32c_hard_regno_ok): Rename to...
10895 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
10896 (m32c_cannot_change_mode_class): Update accordingly.
10897 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10898 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
10899 (m32r_mode_class): Delete.
10900 (HARD_REGNO_MODE_OK): Delete.
10901 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10902 (m32r_hard_regno_mode_ok): Rename to...
10903 (m32r_hard_regno_modes): ...this.
10904 (m32r_mode_class): Make static.
10905 (m32r_hard_regno_mode_ok): New function.
10906 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
10907 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
10908 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10909 (m68k_hard_regno_mode_ok): Make static.
10910 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
10911 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10912 (mcore_hard_regno_mode_ok): New function.
10913 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
10914 (HARD_REGNO_MODE_OK): Delete.
10915 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
10917 (microblaze_hard_regno_mode_ok_p): ...this and make static.
10918 (microblaze_hard_regno_mode_ok): New function.
10919 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10920 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
10921 (mips_hard_regno_mode_ok): Delete.
10922 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
10923 (mips_hard_regno_mode_ok_p): ...this and make static.
10924 (mips_hard_regno_mode_ok_p): Rename to...
10925 (mips_hard_regno_mode_ok_uncached): ...this.
10926 (mips_hard_regno_mode_ok): New function.
10927 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
10928 of HARD_REGNO_MODE_OK.
10929 (mips_option_override): Update after above name changes.
10930 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10931 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
10932 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
10933 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
10934 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
10935 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10936 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
10937 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
10938 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
10939 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10940 (msp430_hard_regno_mode_ok): Make static and return a bool.
10941 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
10942 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
10943 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
10945 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10946 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
10947 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
10948 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
10949 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
10950 (PA_HARD_REGNO_MODE_OK): ...this
10951 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
10952 (PA_HARD_REGNO_MODE_OK): ...this.
10953 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10954 (pa_hard_regno_mode_ok): New function.
10955 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
10956 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10957 (pdp11_hard_regno_mode_ok): New function.
10958 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
10959 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
10961 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
10963 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10964 (rs6000_hard_regno_mode_ok): Rename to...
10965 (rs6000_hard_regno_mode_ok_uncached): ...this.
10966 (rs6000_init_hard_regno_mode_ok): Update accordingly.
10967 (rs6000_hard_regno_mode_ok): New function.
10968 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
10969 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
10970 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
10971 (riscv_hard_regno_mode_ok): ...this and make static.
10972 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10973 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
10974 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
10975 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10976 (rl78_hard_regno_mode_ok): Make static and return bool.
10977 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
10978 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
10980 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
10981 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10982 (rs6000_hard_regno_mode_ok): Rename to...
10983 (rs6000_hard_regno_mode_ok_uncached): ...this.
10984 (rs6000_init_hard_regno_mode_ok): Update accordingly.
10985 (rs6000_hard_regno_mode_ok): New function.
10986 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
10987 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
10988 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10989 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
10990 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
10991 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
10992 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10993 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
10994 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
10995 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10996 (sh_hard_regno_mode_ok): Make static.
10997 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
10998 instead of HARD_REGNO_MODE_OK.
10999 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
11000 (sparc_mode_class): Delete.
11001 (HARD_REGNO_MODE_OK): Delete.
11002 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11003 (hard_regno_mode_classes): Make static.
11004 (sparc_mode_class): Likewise.
11005 (sparc_hard_regno_mode_ok): New function.
11006 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
11007 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
11008 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
11010 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11011 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
11012 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
11013 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
11014 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
11015 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11016 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
11017 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
11018 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11019 (visium_hard_regno_mode_ok): New function.
11020 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
11021 instead of HARD_REGNO_MODE_OK.
11022 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
11023 (HARD_REGNO_MODE_OK): Delete.
11024 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
11025 (xtensa_hard_regno_mode_ok_p): ...this and make static.
11026 (xtensa_option_override): Update accordingly.
11027 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11028 (xtensa_hard_regno_mode_ok): New function.
11029 * system.h (HARD_REGNO_MODE_OK): Poison.
11031 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11032 Alan Hayward <alan.hayward@arm.com>
11033 David Sherwood <david.sherwood@arm.com>
11035 * target.def (hard_regno_call_part_clobbered): New hook.
11036 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
11037 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
11038 * doc/tm.texi: Regenerate.
11039 * hooks.h (hook_bool_uint_mode_false): Declare.
11040 * hooks.c (hook_bool_uint_mode_false): New function.
11041 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11042 * cselib.c (cselib_process_insn): Use
11043 targetm.hard_regno_call_part_clobbered instead of
11044 HARD_REGNO_CALL_PART_CLOBBERED.
11045 * ira-conflicts.c (ira_build_conflicts): Likewise.
11046 * ira-costs.c (ira_tune_allocno_costs): Likewise.
11047 * lra-constraints.c (need_for_call_save_p): Likewise.
11048 * lra-lives.c: Include target.h.
11049 (check_pseudos_live_through_calls): Use
11050 targetm.hard_regno_call_part_clobbered instead of
11051 HARD_REGNO_CALL_PART_CLOBBERED.
11052 * regcprop.c: Include target.h.
11053 (copyprop_hardreg_forward_1): Use
11054 targetm.hard_regno_call_part_clobbered instead of
11055 HARD_REGNO_CALL_PART_CLOBBERED.
11056 * reginfo.c (choose_hard_reg_mode): Likewise.
11057 * regrename.c (check_new_reg_p): Likewise.
11058 * reload.c (find_equiv_reg): Likewise.
11059 * reload1.c (emit_reload_insns): Likewise.
11060 * sched-deps.c (deps_analyze_insn): Likewise.
11061 * sel-sched.c (init_regs_for_mode): Likewise.
11062 (mark_unavailable_hard_regs): Likewise.
11063 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
11064 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11065 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
11067 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11068 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11069 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
11071 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
11073 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11074 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11075 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
11077 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11078 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11079 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
11081 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11082 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
11084 * config/powerpcspe/powerpcspe.c
11085 (rs6000_hard_regno_call_part_clobbered): New function.
11086 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11087 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11088 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
11090 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11091 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11092 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
11094 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11095 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11096 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
11098 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11099 Alan Hayward <alan.hayward@arm.com>
11100 David Sherwood <david.sherwood@arm.com>
11102 * rtl.h (subreg_memory_offset): Declare.
11103 * emit-rtl.c (subreg_memory_offset): New function.
11104 * expmed.c (store_bit_field_1): Use it.
11105 * expr.c (undefined_operand_subword_p): Likewise.
11106 * simplify-rtx.c (simplify_subreg): Likewise.
11108 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
11110 PR rtl-optimization/57448
11113 * optabs.c (expand_atomic_load): Place compiler memory barriers if
11114 using atomic_load pattern.
11115 (expand_atomic_store): Likewise.
11117 2017-09-04 Jakub Jelinek <jakub@redhat.com>
11120 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
11121 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
11122 handling. Use replace_call_with_value with NULL instead of
11123 gsi_replace, unlink_stmt_vdef and release_defs.
11125 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
11128 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
11130 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11133 * rtl.h (emit_library_call_value_1): Declare.
11134 (emit_library_call): Replace declaration with a series of overloads.
11135 Remove the parameter count argument.
11136 (emit_library_call_value): Likewise.
11137 * calls.c (emit_library_call_value_1): Make global. Replace varargs
11138 with an "rtx_mode_t *".
11139 (emit_library_call_value): Delete.
11140 (emit_library_call): Likewise.
11141 * asan.c (asan_emit_stack_protection): Update calls accordingly.
11142 (asan_emit_allocas_unpoison): Likewise.
11143 * builtins.c (expand_builtin_powi): Likewise.
11144 (expand_asan_emit_allocas_unpoison): Likewise.
11145 * cfgexpand.c (expand_main_function): Likewise.
11146 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
11147 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
11148 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
11149 * config/arm/arm.c (arm_trampoline_init): Likewise.
11150 (arm_call_tls_get_addr): Likewise.
11151 (arm_expand_divmod_libfunc): Likewise.
11152 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
11153 (smulsi3_highpart): Likewise.
11154 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
11155 (c6x_expand_compare): Likewise.
11156 (c6x_expand_movmem): Likewise.
11157 * config/frv/frv.c (frv_trampoline_init): Likewise.
11158 * config/i386/i386.c (ix86_trampoline_init): Likewise.
11159 (ix86_expand_divmod_libfunc): Likewise.
11160 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
11161 (ia64_expand_compare): Likewise.
11162 (ia64_profile_hook): Likewise.
11163 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
11164 (nonlocal_goto): Likewise.
11165 (restore_stack_nonlocal): Likewise.
11166 * config/m32r/m32r.c (block_move_call): Likewise.
11167 (m32r_trampoline_init): Likewise.
11168 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
11169 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
11170 (m68k_call_m68k_read_tp): Likewise.
11171 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
11172 (microblaze_expand_divide): Likewise.
11173 * config/mips/mips.h (mips_args): Likewise.
11174 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
11175 (MIPS_ICACHE_SYNC): Likewise.
11176 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
11177 (nios2_trampoline_init): Likewise.
11178 * config/pa/pa.c (hppa_tls_call): Likewise.
11179 (pa_trampoline_init): Likewise.
11180 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
11181 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
11182 (expand_strn_compare): Likewise.
11183 (rs6000_generate_compare): Likewise.
11184 (rs6000_expand_float128_convert): Likewise.
11185 (output_profile_hook): Likewise.
11186 (rs6000_trampoline_init): Likewise.
11187 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
11188 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
11189 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
11190 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
11191 (rs6000_generate_compare): Likewise.
11192 (rs6000_expand_float128_convert): Likewise.
11193 (output_profile_hook): Likewise.
11194 (rs6000_trampoline_init): Likewise.
11195 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
11196 * config/sh/sh.c (sh_trampoline_init): Likewise.
11197 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
11198 (sparc_emit_float_lib_cmp): Likewise.
11199 (sparc32_initialize_trampoline): Likewise.
11200 (sparc64_initialize_trampoline): Likewise.
11201 (sparc_profile_hook): Likewise.
11202 * config/spu/spu.c (ea_load_store): Likewise.
11203 * config/spu/spu.md (floatunssidf2): Likewise.
11204 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
11205 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
11206 * config/visium/visium.c (expand_block_move_4): Likewise.
11207 (expand_block_move_2): Likewise.
11208 (expand_block_move_1): Likewise.
11209 (expand_block_set_4): Likewise.
11210 (expand_block_set_2): Likewise.
11211 (expand_block_set_1): Likewise.
11212 (visium_trampoline_init): Likewise.
11213 (visium_profile_hook): Likewise.
11214 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
11215 (xtensa_setup_frame_addresses): Likewise.
11216 (xtensa_trampoline_init): Likewise.
11217 * except.c (sjlj_emit_function_enter): Likewise.
11218 (sjlj_emit_function_exit): Likewise.
11219 * explow.c (allocate_dynamic_stack_space): Likewise.
11220 (probe_stack_range): Likewise.
11221 * expr.c (convert_mode_scalar): Likewise.
11222 * optabs.c (expand_binop): Likewise.
11223 (expand_twoval_binop_libfunc): Likewise.
11224 (expand_unop): Likewise.
11225 (prepare_cmp_insn): Likewise.
11226 (prepare_float_lib_cmp): Likewise.
11227 (expand_float): Likewise.
11228 (expand_fix): Likewise.
11229 (expand_fixed_convert): Likewise.
11230 (maybe_emit_sync_lock_test_and_set): Likewise.
11231 (expand_atomic_compare_and_swap): Likewise.
11232 (expand_mem_thread_fence): Likewise.
11233 (expand_atomic_fetch_op): Likewise.
11235 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
11237 * doc/generic.texi (OpenACC): Adjust URL.
11238 * doc/invoke.texi (C Dialect Options): Ditto.
11240 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
11242 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
11243 predicate for operand 1. Add (m,<S>) constraint.
11244 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
11245 Prevent memory operand 1 with register operand 2.
11247 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
11249 PR rtl-optimization/82024
11250 * combine.c (try_combine): If the combination result is a PARALLEL,
11251 and we only need to retain the SET in there that would be placed
11252 at I2, check that we can place that at I3 instead, before doing so.
11254 2017-09-01 Jakub Jelinek <jakub@redhat.com>
11257 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
11259 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
11260 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
11263 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
11264 Jeff Law <law@redhat.com>
11266 * varasm.c (bss_initializer_p): Do not put constants into .bss
11267 (categorize_decl_for_section): Handle bss_initializer_p returning
11268 false when DECL_INITIAL is NULL.
11270 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11273 * config/s390/s390.c (s390_can_inline_p): New function.
11275 2017-09-01 Jeff Law <law@redhat.com>
11277 PR tree-optimization/82052
11278 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
11279 Always initialize the returned slot after a hash table miss
11280 when INSERT is true.
11282 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
11284 * config/s390/s390.md (mem_signal_fence): Remove.
11285 * doc/md.texi (mem_signal_fence): Remove.
11286 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
11288 * target-insns.def (mem_signal_fence): Remove.
11290 2017-09-01 Jakub Jelinek <jakub@redhat.com>
11293 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
11296 * asan.c (create_odr_indicator): Strip name encoding from assembler
11297 name before appending it after __odr_asan_.
11299 2017-09-01 Martin Liska <mliska@suse.cz>
11301 PR tree-optimization/82059
11302 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
11303 frequency only when an edge is redirected.
11305 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11307 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
11308 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
11309 (arc_conditional_register_usage): Remove ARC600 lp_count
11311 (arc_file_start): Emit Tag_ARC_CPU_variation.
11312 (arc_can_use_doloop_p): New conditions to use ZOLs.
11313 (hwloop_fail): New function.
11314 (hwloop_optimize): Likewise.
11315 (hwloop_pattern_reg): Likewise.
11316 (arc_doloop_hooks): New struct, to be used with reorg_loops.
11317 (arc_reorg_loops): New function, calls reorg_loops.
11318 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
11319 (arc600_corereg_hazard): Remove ZOL checking, case handled by
11321 (arc_loop_hazard): Remove function, functionality moved into
11323 (arc_hazard): Remove arc_loop_hazard call.
11324 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
11325 into hwloop_optimize.
11326 (arc_label_align): Remove ZOL handling.
11327 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
11328 * config/arc/arc.md (doloop_begin): Remove pattern.
11329 (doloop_begin_i): Likewise.
11330 (doloop_end_i): Likewise.
11331 (doloop_fallback): Likewise.
11332 (doloop_fallback_m): Likewise.
11333 (doloop_end): Reimplement expand.
11334 (arc_lp): New pattern for LP instruction.
11335 (loop_end): New pattern.
11336 (loop_fail): Likewise.
11337 (decrement_and_branch_until_zero): Likewise.
11338 * config/arc/arc.opt (mlpc-width): New option.
11339 * doc/invoke.texi (mlpc-width): Document option.
11341 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11343 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
11344 (arc_ccfsm_advance): Fix checking for delay slots.
11345 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
11347 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11349 * config/arc/arc.md (movqi_insn): Add stores to save constant long
11351 (movhi_insn): Update store instruction constraint which are saving
11352 6-bit short immediates.
11353 (movsi_insn): Consider also short scaled load operations.
11354 (zero_extendhisi2_i): Use Usd constraint instead of T.
11355 (extendhisi2_i): Add q constraint.
11356 (arc_clzsi2): Add type and length attributes.
11357 (arc_ctzsi2): Likewise.
11358 * config/arc/constraints.md (Usc): Update constraint, the
11359 assembler can parse two relocations for a single instruction.
11361 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11363 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
11364 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
11366 2017-08-31 Olivier Hainque <hainque@adacore.com>
11368 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
11369 match as powerpc-wrs-vxworks*.
11371 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
11373 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
11374 register constraint for by-element operand.
11375 (aarch64_mls_elt_merge<mode>): Likewise.
11377 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11379 * config/arc/arc.c (arc_can_follow_jump): Check for short
11382 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11384 * config.gcc: Use g.opt for arc.
11385 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
11386 functionality moved to ...
11387 (legitimate_scaled_address_p): New function, ...here.
11388 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
11389 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
11390 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
11392 (arc_override_options): Handle G option.
11393 (arc_output_pic_addr_const): Correct function definition.
11394 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
11395 (arc_decl_anon_ns_mem_p): Delete.
11396 (arc_in_small_data_p): Overhaul this function to take into
11397 consideration the value given via G option.
11398 (arc_rewrite_small_data_1): Renamed and corrected old
11399 arc_rewrite_small_data function.
11400 (arc_rewrite_small_data): New function.
11401 (small_data_pattern): Don't use pic_offset_table_rtx.
11402 * config/arc/arc.h (CC1_SPEC): Recognize G option.
11403 * config/arc/simdext.md (movmisalignv2hi): Use
11404 prepare_move_operands function.
11406 (movmisalign*): Likewise.
11407 * doc/invoke.texi (ARC options): Document -G option.
11409 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11411 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
11413 * config/arc/arc.c (arc_print_operand): Output scalled address for
11414 sdata whenever is possible.
11415 (arc_in_small_data_p): Allow sdata for 64bit datum when double
11416 load/stores are available.
11417 (compact_sda_memory_operand): Check for the alignment required by
11418 code density instructions.
11419 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
11421 * config/arc/constraints.md (Usd): Update constraint.
11422 (Us0): New constraint.
11423 (Usc): Update constraint.
11425 2017-08-31 Richard Biener <rguenther@suse.de>
11427 PR middle-end/82054
11428 * dwarf2out.c (dwarf2out_early_global_decl): Process each
11429 function only once.
11431 2017-08-31 Tamar Christina <tamar.christina@arm.com>
11433 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
11434 Resize type_signature.
11436 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11437 Alan Hayward <alan.hayward@arm.com>
11438 David Sherwood <david.sherwood@arm.com>
11440 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
11441 subregs whose inner modes can be stored in GPRs.
11442 (aarch64_classify_index): Likewise.
11444 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11445 Alan Hayward <alan.hayward@arm.com>
11446 David Sherwood <david.sherwood@arm.com>
11448 * config/aarch64/iterators.md (V_cmp_result): Rename to...
11449 (V_INT_EQUIV): ...this.
11450 (v_cmp_result): Rename to...
11451 (v_int_equiv): ...this.
11452 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
11453 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
11454 (copysign<mode>3): Likewise.
11455 (aarch64_simd_bsl<mode>_internal): Likewise.
11456 (aarch64_simd_bsl<mode>): Likewise.
11457 (vec_cmp<mode><mode>): Likewise.
11458 (vcond<mode><mode>): Likewise.
11459 (vcond<v_cmp_mixed><mode>): Likewise.
11460 (vcondu<mode><v_cmp_mixed>): Likewise.
11461 (aarch64_cm<optab><mode>): Likewise.
11462 (aarch64_cmtst<mode>): Likewise.
11463 (aarch64_fac<optab><mode>): Likewise.
11464 (vec_perm_const<mode>): Likewise.
11465 (vcond_mask_<mode><v_cmp_result>): Rename to...
11466 (vcond_mask_<mode><v_int_equiv>): ...this.
11467 (vec_cmp<mode><v_cmp_result>): Rename to...
11468 (vec_cmp<mode><v_int_equiv>): ...this.
11470 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11471 Alan Hayward <alan.hayward@arm.com>
11472 David Sherwood <david.sherwood@arm.com>
11474 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
11476 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
11477 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
11478 (UNSPEC_LD4_DREG): New unspecs.
11479 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
11480 (aarch64_ld2<mode>_dreg_be): Replace with...
11481 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
11483 (aarch64_ld3<mode>_dreg_le)
11484 (aarch64_ld3<mode>_dreg_be): Replace with...
11485 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
11487 (aarch64_ld4<mode>_dreg_le)
11488 (aarch64_ld4<mode>_dreg_be): Replace with...
11489 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
11492 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11494 PR tree-optimization/81987
11495 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
11496 insert an initializer in a location not dominated by the stride
11499 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
11501 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
11502 on the entire header of the finally block in the fallthru case.
11504 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
11506 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
11508 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
11510 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
11511 rs6000_emit_move_from_cr and call renamed function.
11512 (rs6000_emit_prologue): Call renamed functions.
11513 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
11514 movesi_from_cr, remove volatile CRs.
11516 2017-08-30 Jon Beniston <jon@beniston.com>
11517 Richard Biener <rguenther@suse.de>
11519 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
11520 of VECTOR_MODE_P check.
11521 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
11522 element vector types.
11524 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11526 * df.h (df_read_modify_subreg_p): Remove in favor of...
11527 * rtl.h (read_modify_subreg_p): ...this new function. Take a
11528 const_rtx instead of an rtx.
11529 * cprop.c (local_cprop_find_used_regs): Update accordingly.
11530 * df-problems.c (df_word_lr_mark_ref): Likewise.
11531 * ira-lives.c (mark_pseudo_reg_live): Likewise.
11532 (mark_pseudo_reg_dead): Likewise.
11533 (mark_ref_dead): Likewise.
11534 * reginfo.c (init_subregs_of_mode): Likewise.
11535 * sched-deps.c (sched_analyze_1): Likewise.
11536 * df-scan.c (df_def_record_1): Likewise.
11537 (df_uses_record): Likewise.
11538 (df_read_modify_subreg_p): Remove in favor of...
11539 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
11540 const_rtx instead of an rtx.
11542 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11543 Alan Hayward <alan.hayward@arm.com>
11544 David Sherwood <david.sherwood@arm.com>
11546 * rtl.h (partial_subreg_p): New function.
11547 * caller-save.c (save_call_clobbered_regs): Use it.
11548 * calls.c (expand_call): Likewise.
11549 * combine.c (combinable_i3pat): Likewise.
11550 (simplify_set): Likewise.
11551 (make_extraction): Likewise.
11552 (make_compound_operation_int): Likewise.
11553 (gen_lowpart_or_truncate): Likewise.
11554 (force_to_mode): Likewise.
11555 (make_field_assignment): Likewise.
11556 (reg_truncated_to_mode): Likewise.
11557 (record_truncated_value): Likewise.
11558 (move_deaths): Likewise.
11559 * cse.c (record_jump_cond): Likewise.
11560 (cse_insn): Likewise.
11561 * cselib.c (cselib_lookup_1): Likewise.
11562 * expmed.c (extract_bit_field_using_extv): Likewise.
11563 * function.c (assign_parm_setup_reg): Likewise.
11564 * ifcvt.c (noce_convert_multiple_sets): Likewise.
11565 * ira-build.c (create_insn_allocnos): Likewise.
11566 * lra-coalesce.c (merge_pseudos): Likewise.
11567 * lra-constraints.c (match_reload): Likewise.
11568 (simplify_operand_subreg): Likewise.
11569 (curr_insn_transform): Likewise.
11570 * lra-lives.c (process_bb_lives): Likewise.
11571 * lra.c (new_insn_reg): Likewise.
11572 (lra_substitute_pseudo): Likewise.
11573 * regcprop.c (mode_change_ok): Likewise.
11574 (maybe_mode_change): Likewise.
11575 (copyprop_hardreg_forward_1): Likewise.
11576 * reload.c (push_reload): Likewise.
11577 (find_reloads): Likewise.
11578 (find_reloads_subreg_address): Likewise.
11579 * reload1.c (alter_reg): Likewise.
11580 (eliminate_regs_1): Likewise.
11581 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11583 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
11585 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
11586 back to if statements, including unpack.
11588 2017-08-30 Martin Liska <mliska@suse.cz>
11590 PR inline-asm/82001
11591 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
11593 (func_checker::compare_asm_inputs_outputs): ... this function.
11594 (func_checker::compare_gimple_asm): Use the function to compare
11595 also ASM constrains.
11596 * ipa-icf-gimple.h: Rename the function.
11598 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11599 Alan Hayward <alan.hayward@arm.com>
11600 David Sherwood <david.sherwood@arm.com>
11602 * coretypes.h (complex_mode): New type.
11603 * gdbhooks.py (build_pretty_printer): Handle it.
11604 * machmode.h (complex_mode): New class.
11605 (complex_mode::includes_p): New function.
11606 (is_complex_int_mode): Likewise.
11607 (is_complex_float_mode): Likewise.
11608 * genmodes.c (get_mode_class): Handle complex mode classes.
11609 * function.c (expand_function_end): Use is_complex_int_mode.
11611 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11612 Alan Hayward <alan.hayward@arm.com>
11613 David Sherwood <david.sherwood@arm.com>
11615 * coretypes.h (scalar_mode_pod): New typedef.
11616 * gdbhooks.py (build_pretty_printer): Handle it.
11617 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
11618 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
11619 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
11620 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
11621 as_a <scalar_mode>.
11623 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11624 Alan Hayward <alan.hayward@arm.com>
11625 David Sherwood <david.sherwood@arm.com>
11627 * machmode.h (mode_for_vector): Take a scalar_mode instead
11629 * stor-layout.c (mode_for_vector): Likewise.
11630 * explow.c (promote_mode): Use as_a <scalar_mode>.
11631 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
11633 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11634 Alan Hayward <alan.hayward@arm.com>
11635 David Sherwood <david.sherwood@arm.com>
11637 * target.def (preferred_simd_mode): Take a scalar_mode
11638 instead of a machine_mode.
11639 * targhooks.h (default_preferred_simd_mode): Likewise.
11640 * targhooks.c (default_preferred_simd_mode): Likewise.
11641 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
11642 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
11643 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
11644 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
11645 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
11646 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
11647 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
11648 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
11650 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
11651 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
11652 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
11653 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
11654 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
11655 * doc/tm.texi: Regenerate.
11656 * optabs-query.c (can_vec_mask_load_store_p): Return false for
11659 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11660 Alan Hayward <alan.hayward@arm.com>
11661 David Sherwood <david.sherwood@arm.com>
11663 * target.def (scalar_mode_supported_p): Take a scalar_mode
11664 instead of a machine_mode.
11665 * targhooks.h (default_scalar_mode_supported_p): Likewise.
11666 * targhooks.c (default_scalar_mode_supported_p): Likewise.
11667 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
11668 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
11669 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
11670 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
11671 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
11672 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
11673 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
11674 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
11675 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
11676 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
11677 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
11678 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
11680 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
11681 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
11682 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
11683 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
11684 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
11686 * doc/tm.texi: Regenerate.
11688 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11689 Alan Hayward <alan.hayward@arm.com>
11690 David Sherwood <david.sherwood@arm.com>
11692 * coretypes.h (opt_scalar_mode): New typedef.
11693 * gdbhooks.py (build_pretty_printers): Handle it.
11694 * machmode.h (mode_iterator::get_2xwider): Add overload for
11696 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
11698 * expr.c (convert_mode_scalar): Likewise.
11699 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11700 * optabs.c (expand_float): Likewise.
11701 (expand_fix): Likewise.
11702 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11704 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11705 Alan Hayward <alan.hayward@arm.com>
11706 David Sherwood <david.sherwood@arm.com>
11708 * optabs.c (expand_float): Explicitly check for scalars before
11709 using a branching expansion.
11710 (expand_fix): Likewise.
11712 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11713 Alan Hayward <alan.hayward@arm.com>
11714 David Sherwood <david.sherwood@arm.com>
11716 * expr.c (convert_mode): Split scalar handling out into...
11717 (convert_mode_scalar): ...this new function. Treat the modes
11720 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11721 Alan Hayward <alan.hayward@arm.com>
11722 David Sherwood <david.sherwood@arm.com>
11724 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
11726 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
11728 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11729 Alan Hayward <alan.hayward@arm.com>
11730 David Sherwood <david.sherwood@arm.com>
11732 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
11733 rather than a machine_mode.
11734 (fixed_from_string): Likewise.
11735 (fixed_convert): Likewise.
11736 (fixed_convert_from_int): Likewise.
11737 (fixed_convert_from_real): Likewise.
11738 (real_convert_from_fixed): Likewise.
11739 * fixed-value.c (fixed_from_double_int): Likewise.
11740 (fixed_from_string): Likewise.
11741 (fixed_convert): Likewise.
11742 (fixed_convert_from_int): Likewise.
11743 (fixed_convert_from_real): Likewise.
11744 (real_convert_from_fixed): Likewise.
11745 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
11747 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11748 Alan Hayward <alan.hayward@arm.com>
11749 David Sherwood <david.sherwood@arm.com>
11751 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
11752 of separate mode class checks. Do not allow vector modes here.
11753 (immed_wide_int_const): Use as_a <scalar_mode>.
11754 * explow.c (trunc_int_for_mode): Likewise.
11755 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
11756 (wi::shwi): Likewise.
11757 (wi::min_value): Likewise.
11758 (wi::max_value): Likewise.
11759 * dwarf2out.c (loc_descriptor): Likewise.
11760 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
11761 for CONST_WIDE_INT.
11763 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11764 Alan Hayward <alan.hayward@arm.com>
11765 David Sherwood <david.sherwood@arm.com>
11767 * tree.h (SCALAR_TYPE_MODE): New macro.
11768 * expr.c (expand_expr_addr_expr_1): Use it.
11769 (expand_expr_real_2): Likewise.
11770 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
11771 (fold_convert_const_fixed_from_int): Likewise.
11772 (fold_convert_const_fixed_from_real): Likewise.
11773 (native_encode_fixed): Likewise
11774 (native_encode_complex): Likewise
11775 (native_encode_vector): Likewise.
11776 (native_interpret_fixed): Likewise.
11777 (native_interpret_real): Likewise.
11778 (native_interpret_complex): Likewise.
11779 (native_interpret_vector): Likewise.
11780 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
11781 (simd_clone_adjust_argument_types): Likewise.
11782 (simd_clone_init_simd_arrays): Likewise.
11783 (simd_clone_adjust): Likewise.
11784 * stor-layout.c (layout_type): Likewise.
11785 * tree.c (build_minus_one_cst): Likewise.
11786 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
11787 * tree-inline.c (estimate_move_cost): Likewise.
11788 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
11789 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
11790 (vectorizable_reduction): Likewise.
11791 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
11792 (vect_recog_mixed_size_cond_pattern): Likewise.
11793 (check_bool_pattern): Likewise.
11794 (adjust_bool_pattern): Likewise.
11795 (search_type_for_mask_1): Likewise.
11796 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
11797 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11798 (vectorizable_load): Likewise.
11799 (vectorizable_store): Likewise.
11800 * ubsan.c (ubsan_encode_value): Likewise.
11801 * varasm.c (output_constant): Likewise.
11803 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11804 Alan Hayward <alan.hayward@arm.com>
11805 David Sherwood <david.sherwood@arm.com>
11807 * coretypes.h (scalar_mode): New class.
11808 * machmode.h (scalar_mode): Likewise.
11809 (scalar_mode::includes_p): New function.
11810 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
11811 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
11812 * genmodes.c (get_mode_class): Handle remaining scalar modes.
11813 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
11814 * expmed.c (store_bit_field_1): Likewise.
11815 (extract_bit_field_1): Likewise.
11816 * expr.c (write_complex_part): Likewise.
11817 (read_complex_part): Likewise.
11818 (emit_move_complex_push): Likewise.
11819 (expand_expr_real_2): Likewise.
11820 * function.c (assign_parm_setup_reg): Likewise.
11821 (assign_parms_unsplit_complex): Likewise.
11822 * optabs.c (expand_binop): Likewise.
11823 * rtlanal.c (subreg_get_info): Likewise.
11824 * simplify-rtx.c (simplify_immed_subreg): Likewise.
11825 * varasm.c (output_constant_pool_2): Likewise.
11827 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11828 Alan Hayward <alan.hayward@arm.com>
11829 David Sherwood <david.sherwood@arm.com>
11831 * expmed.c (extract_high_half): Use scalar_int_mode and remove
11833 (expmed_mult_highpart_optab): Likewise.
11834 (expmed_mult_highpart): Likewise.
11836 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11837 Alan Hayward <alan.hayward@arm.com>
11838 David Sherwood <david.sherwood@arm.com>
11840 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
11841 instead of a machine_mode.
11842 (builtin_memset_read_str): Likewise.
11843 * builtins.c (c_readstr): Likewise.
11844 (builtin_memcpy_read_str): Likewise.
11845 (builtin_strncpy_read_str): Likewise.
11846 (builtin_memset_read_str): Likewise.
11847 (builtin_memset_gen_str): Likewise.
11848 (expand_builtin_signbit): Use scalar_int_mode for local variables.
11849 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
11850 instead of a machine_mode.
11851 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
11853 (make_extraction): Likewise.
11854 (try_widen_shift_mode): Take and return scalar_int_modes instead
11856 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
11857 a scalar_int_mode instead of a machine_mode.
11858 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
11859 (avr_addr_space_pointer_mode): Likewise.
11860 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
11861 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
11862 (msp430_unwind_word_mode): Likewise.
11863 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
11864 (spu_addr_space_pointer_mode): Likewise.
11865 (spu_addr_space_address_mode): Likewise.
11866 (spu_libgcc_cmp_return_mode): Likewise.
11867 (spu_libgcc_shift_count_mode): Likewise.
11868 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
11869 (rl78_addr_space_pointer_mode): Likewise.
11870 (fl78_unwind_word_mode): Likewise.
11871 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
11873 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
11874 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
11875 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
11876 (mips_valid_pointer_mode): Likewise.
11877 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
11878 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
11879 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
11881 (ft32_addr_space_address_mode): Likewise.
11882 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
11883 scalar_int_mode instead of a machine_mode.
11884 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
11886 (m32c_addr_space_address_mode): Likewise.
11887 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
11888 (rs6000_eh_return_filter_mode): Likewise.
11889 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
11890 (rs6000_eh_return_filter_mode): Likewise.
11891 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
11892 (s390_libgcc_shift_count_mode): Likewise.
11893 (s390_unwind_word_mode): Likewise.
11894 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
11896 * target.def (mode_rep_extended): Likewise.
11897 (valid_pointer_mode): Likewise.
11898 (addr_space.valid_pointer_mode): Likewise.
11899 (eh_return_filter_mode): Return a scalar_int_mode rather than
11901 (libgcc_cmp_return_mode): Likewise.
11902 (libgcc_shift_count_mode): Likewise.
11903 (unwind_word_mode): Likewise.
11904 (addr_space.pointer_mode): Likewise.
11905 (addr_space.address_mode): Likewise.
11906 * doc/tm.texi: Regenerate.
11907 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
11909 (do_jump): Use scalar_int_mode for local variables.
11910 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
11911 rather than a machine_mode.
11912 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
11913 (scompare_loc_descriptor_wide): Likewise.
11914 (scompare_loc_descriptor_narrow): Likewise.
11915 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
11917 * except.c (sjlj_emit_dispatch_table): Likewise.
11918 (expand_builtin_eh_copy_values): Likewise.
11919 * explow.c (convert_memory_address_addr_space_1): Likewise.
11920 Take a scalar_int_mode rather than a machine_mode.
11921 (convert_memory_address_addr_space): Take a scalar_int_mode rather
11922 than a machine_mode.
11923 (memory_address_addr_space): Use scalar_int_mode for local variables.
11924 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
11925 rather than a machine_mode.
11926 * expmed.c (mask_rtx): Likewise.
11927 (init_expmed_one_conv): Likewise.
11928 (expand_mult_highpart_adjust): Likewise.
11929 (extract_high_half): Likewise.
11930 (expmed_mult_highpart_optab): Likewise.
11931 (expmed_mult_highpart): Likewise.
11932 (expand_smod_pow2): Likewise.
11933 (expand_sdiv_pow2): Likewise.
11934 (emit_store_flag_int): Likewise.
11935 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
11937 (extract_low_bits): Likewise.
11938 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
11940 * expr.c (pieces_addr::adjust): Likewise.
11941 (can_store_by_pieces): Likewise.
11942 (store_by_pieces): Likewise.
11943 (clear_by_pieces_1): Likewise.
11944 (expand_expr_addr_expr_1): Likewise.
11945 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
11946 (expand_expr_real_1): Likewise.
11947 (try_casesi): Likewise.
11948 * final.c (shorten_branches): Likewise.
11949 * fold-const.c (fold_convert_const_int_from_fixed): Change the
11950 type of "mode" to machine_mode.
11951 * internal-fn.c (expand_arith_overflow_result_store): Take a
11952 scalar_int_mode rather than a machine_mode.
11953 (expand_mul_overflow): Use scalar_int_mode for local variables.
11954 * loop-doloop.c (doloop_modify): Likewise.
11955 (doloop_optimize): Likewise.
11956 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
11957 than a machine_mode.
11958 (expand_doubleword_shift_condmove): Likewise.
11959 (expand_doubleword_shift): Likewise.
11960 (expand_doubleword_clz): Likewise.
11961 (expand_doubleword_popcount): Likewise.
11962 (expand_doubleword_parity): Likewise.
11963 (expand_absneg_bit): Use scalar_int_mode for local variables.
11964 (prepare_float_lib_cmp): Likewise.
11965 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
11966 rather than a machine_mode.
11967 (convert_memory_address_addr_space): Likewise.
11968 (get_mode_bounds): Likewise.
11969 (get_address_mode): Return a scalar_int_mode rather than a
11971 * rtlanal.c (get_address_mode): Likewise.
11972 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
11973 than a machine_mode.
11974 * targhooks.c (default_mode_rep_extended): Likewise.
11975 (default_valid_pointer_mode): Likewise.
11976 (default_addr_space_valid_pointer_mode): Likewise.
11977 (default_eh_return_filter_mode): Return a scalar_int_mode rather
11978 than a machine_mode.
11979 (default_libgcc_cmp_return_mode): Likewise.
11980 (default_libgcc_shift_count_mode): Likewise.
11981 (default_unwind_word_mode): Likewise.
11982 (default_addr_space_pointer_mode): Likewise.
11983 (default_addr_space_address_mode): Likewise.
11984 * targhooks.h (default_eh_return_filter_mode): Likewise.
11985 (default_libgcc_cmp_return_mode): Likewise.
11986 (default_libgcc_shift_count_mode): Likewise.
11987 (default_unwind_word_mode): Likewise.
11988 (default_addr_space_pointer_mode): Likewise.
11989 (default_addr_space_address_mode): Likewise.
11990 (default_mode_rep_extended): Take a scalar_int_mode rather than
11992 (default_valid_pointer_mode): Likewise.
11993 (default_addr_space_valid_pointer_mode): Likewise.
11994 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
11996 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
11997 rather than a machine_mode.
11998 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
11999 for local variables.
12000 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12001 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
12002 than a machine_mode.
12004 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12005 Alan Hayward <alan.hayward@arm.com>
12006 David Sherwood <david.sherwood@arm.com>
12008 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
12009 the mode argument to scalar_int_mode.
12010 (do_jump_by_parts_zero_rtx): Likewise.
12011 (do_jump_by_parts_equality_rtx): Likewise.
12012 (do_jump_by_parts_greater): Take a mode argument.
12013 (do_jump_by_parts_equality): Likewise.
12014 (do_jump_1): Update calls accordingly.
12016 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12017 Alan Hayward <alan.hayward@arm.com>
12018 David Sherwood <david.sherwood@arm.com>
12020 * is-a.h (safe_dyn_cast): New function.
12021 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
12022 (jump_table_for_label): Likewise.
12023 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
12024 instead of an rtx_insn *.
12025 (shorten_branches): Use dyn_cast instead of LABEL_P and
12026 JUMP_TABLE_DATA_P. Use jump_table_for_label and
12027 rtx_jump_table_data::get_data_mode.
12028 (final_scan_insn): Likewise.
12030 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12031 Alan Hayward <alan.hayward@arm.com>
12032 David Sherwood <david.sherwood@arm.com>
12034 * combine.c (try_combine): Use is_a <scalar_int_mode> when
12035 trying to combine a full-register integer set with a subreg
12038 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12039 Alan Hayward <alan.hayward@arm.com>
12040 David Sherwood <david.sherwood@arm.com>
12042 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
12043 that is always either address_mode or pointer_mode.
12045 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12046 Alan Hayward <alan.hayward@arm.com>
12047 David Sherwood <david.sherwood@arm.com>
12049 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
12050 when the two are known to be equal.
12052 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12053 Alan Hayward <alan.hayward@arm.com>
12054 David Sherwood <david.sherwood@arm.com>
12056 * simplify-rtx.c (simplify_const_unary_operation): Use
12057 is_a <scalar_int_mode> instead of checking for a nonzero
12058 precision. Forcibly convert op_mode to a scalar_int_mode
12059 in that case. More clearly differentiate the operand and
12060 result modes and use the former when deciding what the value
12061 of a count-bits operation should be. Use is_int_mode instead
12062 of checking for a MODE_INT. Remove redundant check for whether
12063 this mode has a zero precision.
12065 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12066 Alan Hayward <alan.hayward@arm.com>
12067 David Sherwood <david.sherwood@arm.com>
12069 * optabs.c (widen_leading): Change the type of the mode argument
12070 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
12071 (widen_bswap): Likewise.
12072 (expand_parity): Likewise.
12073 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
12074 (expand_ffs): Likewise.
12075 (epand_unop): Check for scalar integer modes before calling the
12078 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12079 Alan Hayward <alan.hayward@arm.com>
12080 David Sherwood <david.sherwood@arm.com>
12082 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
12084 (expand_expr_real_1): Update call accordingly.
12086 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12087 Alan Hayward <alan.hayward@arm.com>
12088 David Sherwood <david.sherwood@arm.com>
12090 * expmed.c (store_bit_field_using_insv): Add op0_mode and
12091 value_mode arguments. Use scalar_int_mode internally.
12092 (store_bit_field_1): Rename the new integer mode from imode
12093 to op0_mode and use it instead of GET_MODE (op0). Update calls
12094 to store_split_bit_field, store_bit_field_using_insv and
12095 store_fixed_bit_field.
12096 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
12097 Use scalar_int_mode internally. Use a bit count rather than a mode
12098 when calculating the largest bit size for get_best_mode.
12099 Update calls to store_split_bit_field and store_fixed_bit_field_1.
12100 (store_fixed_bit_field_1): Add mode and value_mode arguments.
12101 Remove assertion that OP0 has a scalar integer mode.
12102 (store_split_bit_field): Add op0_mode and value_mode arguments.
12103 Update calls to extract_fixed_bit_field.
12104 (extract_bit_field_using_extv): Add an op0_mode argument.
12105 Use scalar_int_mode internally.
12106 (extract_bit_field_1): Rename the new integer mode from imode to
12107 op0_mode and use it instead of GET_MODE (op0). Update calls to
12108 extract_split_bit_field, extract_bit_field_using_extv and
12109 extract_fixed_bit_field.
12110 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
12111 to extract_split_bit_field and extract_fixed_bit_field_1.
12112 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
12113 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
12114 on the target mode.
12115 (extract_split_bit_field): Add an op0_mode argument. Update call
12116 to extract_fixed_bit_field.
12118 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12119 Alan Hayward <alan.hayward@arm.com>
12120 David Sherwood <david.sherwood@arm.com>
12122 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
12123 * explow.c (hard_function_value): Likewise.
12124 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
12125 convert_to_mode call outside the loop.
12126 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
12127 for the mode iterator. Require the mode specified by max_pieces
12129 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
12131 (copy_blkmode_to_reg): Likewise.
12132 (set_storage_via_setmem): Likewise.
12133 * optabs.c (prepare_cmp_insn): Likewise.
12134 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
12135 * stor-layout.c (finish_bitfield_representative): Likewise.
12137 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12138 Alan Hayward <alan.hayward@arm.com>
12139 David Sherwood <david.sherwood@arm.com>
12141 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
12142 * expr.c (convert_move): Use them.
12143 (convert_modes): Likewise.
12144 (store_expr_with_bounds): Likewise.
12146 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12147 Alan Hayward <alan.hayward@arm.com>
12148 David Sherwood <david.sherwood@arm.com>
12150 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
12151 parameter for the mode of "x". Remove the "known_x", "known_mode"
12152 and "known_ret" arguments. Change the type of the mode argument
12153 to scalar_int_mode.
12154 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
12155 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
12156 (reg_num_sign_bit_copies_for_combine): Likewise.
12157 * rtlanal.c (nonzero_bits1): Likewise.
12158 (num_sign_bit_copies1): Likewise.
12159 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
12160 (reg_num_sign_bit_copies_general): Likewise.
12161 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
12162 (reg_nonzero_bits_general): Likewise.
12164 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12165 Alan Hayward <alan.hayward@arm.com>
12166 David Sherwood <david.sherwood@arm.com>
12168 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
12169 than in subroutines. Return 1 for non-integer modes.
12170 (cached_num_sign_bit_copies): Change the type of the mode parameter
12171 to scalar_int_mode.
12172 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
12173 classes. Handle CONST_INT_P first and then check whether X also
12174 has a scalar integer mode. Check the same thing for inner registers
12175 of a SUBREG and for values that are being extended or truncated.
12177 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12178 Alan Hayward <alan.hayward@arm.com>
12179 David Sherwood <david.sherwood@arm.com>
12181 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
12182 in subroutines. Return the mode mask for non-integer modes.
12183 (cached_nonzero_bits): Change the type of the mode parameter
12184 to scalar_int_mode.
12185 (nonzero_bits1): Likewise. Remove early exit for other mode
12186 classes. Handle CONST_INT_P first and then check whether X
12187 also has a scalar integer mode.
12189 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12190 Alan Hayward <alan.hayward@arm.com>
12191 David Sherwood <david.sherwood@arm.com>
12193 * expr.c (widest_int_mode_for_size): Make the comment match the code.
12194 Return a scalar_int_mode and assert that the size is greater than
12196 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
12197 (op_by_pieces_d::op_by_pieces_d): Likewise.
12198 (op_by_pieces_d::run): Likewise.
12199 (can_store_by_pieces): Likewise.
12201 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12202 Alan Hayward <alan.hayward@arm.com>
12203 David Sherwood <david.sherwood@arm.com>
12205 * combine.c (extract_left_shift): Add a mode argument and update
12207 (make_compound_operation_int): Change the type of the mode parameter
12208 to scalar_int_mode and update the call to extract_left_shift.
12210 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12211 Alan Hayward <alan.hayward@arm.com>
12212 David Sherwood <david.sherwood@arm.com>
12214 * combine.c (simplify_and_const_int): Change the type of the mode
12215 parameter to scalar_int_mode.
12216 (simplify_and_const_int_1): Likewise. Update recursive call.
12218 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12219 Alan Hayward <alan.hayward@arm.com>
12220 David Sherwood <david.sherwood@arm.com>
12222 * combine.c (simplify_compare_const): Check that the mode is a
12223 scalar_int_mode (rather than VOIDmode) before testing its
12225 (simplify_comparison): Move COMPARISON_P handling out of the
12226 loop and restrict the latter part of the loop to scalar_int_modes.
12227 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
12228 and when considering SUBREG_REGs. Use is_int_mode instead of
12229 checking GET_MODE_CLASS against MODE_INT.
12231 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12232 Alan Hayward <alan.hayward@arm.com>
12233 David Sherwood <david.sherwood@arm.com>
12235 * combine.c (try_widen_shift_mode): Move check for equal modes to...
12236 (simplify_shift_const_1): ...here. Use scalar_int_mode for
12237 shift_unit_mode and for modes involved in scalar shifts.
12239 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12240 Alan Hayward <alan.hayward@arm.com>
12241 David Sherwood <david.sherwood@arm.com>
12243 * combine.c (force_int_to_mode): New function, split out from...
12244 (force_to_mode): ...here. Keep xmode up-to-date and use it
12245 instead of GET_MODE (x).
12247 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12248 Alan Hayward <alan.hayward@arm.com>
12249 David Sherwood <david.sherwood@arm.com>
12251 * optabs-query.h (extraction_insn::struct_mode): Change type to
12252 opt_scalar_int_mode and update comment.
12253 (extraction_insn::field_mode): Change type to scalar_int_mode.
12254 (extraction_insn::pos_mode): Likewise.
12255 * combine.c (make_extraction): Update accordingly.
12256 * optabs-query.c (get_traditional_extraction_insn): Likewise.
12257 (get_optab_extraction_insn): Likewise.
12258 * recog.c (simplify_while_replacing): Likewise.
12259 * expmed.c (narrow_bit_field_mem): Change the type of the mode
12260 parameter to opt_scalar_int_mode.
12262 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12263 Alan Hayward <alan.hayward@arm.com>
12264 David Sherwood <david.sherwood@arm.com>
12266 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
12267 to a scalar_int_mode instead of a machine_mode.
12268 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
12269 (get_best_mode): Return a boolean and use a pointer argument to store
12270 the selected mode. Replace the limit mode parameter with a bit limit.
12271 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
12272 for the values returned by bit_field_mode_iterator::next_mode.
12273 (store_bit_field): Update call to get_best_mode.
12274 (store_fixed_bit_field): Likewise.
12275 (extract_fixed_bit_field): Likewise.
12276 * expr.c (optimize_bitfield_assignment_op): Likewise.
12277 * fold-const.c (optimize_bit_field_compare): Likewise.
12278 (fold_truth_andor_1): Likewise.
12279 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
12280 Update for new type of m_mode.
12281 (get_best_mode): As above.
12283 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12284 Alan Hayward <alan.hayward@arm.com>
12285 David Sherwood <david.sherwood@arm.com>
12287 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
12288 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
12289 (store_bit_field): Check is_a <scalar_int_mode> before calling
12290 strict_volatile_bitfield_p.
12291 (extract_bit_field): Likewise.
12293 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12294 Alan Hayward <alan.hayward@arm.com>
12295 David Sherwood <david.sherwood@arm.com>
12297 * target.def (cstore_mode): Return a scalar_int_mode.
12298 * doc/tm.texi: Regenerate.
12299 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
12300 * targhooks.h (default_cstore_mode): Likewise.
12301 * targhooks.c (default_cstore_mode): Likewise, using a forced
12303 * expmed.c (emit_cstore): Expect the target of the cstore to be
12306 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12307 Alan Hayward <alan.hayward@arm.com>
12308 David Sherwood <david.sherwood@arm.com>
12310 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
12312 (niter_desc): Likewise mode.
12313 (iv_analyze): Add a mode parameter.
12315 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
12316 and change its type to scalar_int_mode.
12317 * loop-iv.c: Update commentary at head of file.
12318 (iv_constant): Pass the mode paraeter before the rtx it describes
12319 and change its type to scalar_int_mode. Remove VOIDmode handling.
12320 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
12321 (iv_extend): Likewise.
12322 (shorten_into_mode): Likewise.
12323 (iv_add): Use scalar_int_mode.
12324 (iv_mult): Likewise.
12325 (iv_shift): Likewise.
12326 (canonicalize_iv_subregs): Likewise.
12327 (get_biv_step_1): Pass the outer_mode parameter before the rtx
12328 it describes and change its mode to scalar_int_mode. Also change
12329 the type of the returned inner_mode to scalar_int_mode.
12330 (get_biv_step): Likewise, turning outer_mode from a pointer
12331 into a direct parameter. Update call to get_biv_step_1.
12332 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
12333 iv_constant and get_biv_step.
12334 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
12335 and change its type to scalar_int_mode. Don't initialise iv->mode
12336 to VOIDmode and remove later checks for its still being VOIDmode.
12337 Update calls to iv_analyze_op and iv_analyze_expr. Check
12338 is_a <scalar_int_mode> when changing the mode under consideration.
12339 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
12340 Update call to iv_analyze_expr.
12341 (iv_analyze_op): Add a mode parameter. Reject subregs whose
12342 inner register is not also a scalar_int_mode. Update call to
12344 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
12345 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
12346 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
12347 separate mode class checks. Update calls to iv_analyze. Remove
12348 fix-up of VOIDmodes after iv_analyze_biv.
12349 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
12350 don't have a scalar_int_mode. Update call to biv_p.
12352 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12353 Alan Hayward <alan.hayward@arm.com>
12354 David Sherwood <david.sherwood@arm.com>
12356 * cfgexpand.c (convert_debug_memory_address): Use
12357 as_a <scalar_int_mode>.
12358 * combine.c (expand_compound_operation): Likewise.
12359 (make_extraction): Likewise.
12360 (change_zero_ext): Likewise.
12361 (simplify_comparison): Likewise.
12362 * cse.c (cse_insn): Likewise.
12363 * dwarf2out.c (minmax_loc_descriptor): Likewise.
12364 (mem_loc_descriptor): Likewise.
12365 (loc_descriptor): Likewise.
12366 * expmed.c (init_expmed_one_mode): Likewise.
12367 (synth_mult): Likewise.
12368 (emit_store_flag_1): Likewise.
12369 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
12370 of a comparison with size.
12371 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
12372 (reduce_to_bit_field_precision): Likewise.
12373 * function.c (expand_function_end): Likewise.
12374 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
12375 * loop-doloop.c (doloop_modify): Likewise.
12376 * optabs.c (expand_binop): Likewise.
12377 (expand_unop): Likewise.
12378 (expand_copysign_absneg): Likewise.
12379 (prepare_cmp_insn): Likewise.
12380 (maybe_legitimize_operand): Likewise.
12381 * recog.c (const_scalar_int_operand): Likewise.
12382 * rtlanal.c (get_address_mode): Likewise.
12383 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12384 (simplify_cond_clz_ctz): Likewise.
12385 * tree-nested.c (get_nl_goto_field): Likewise.
12386 * tree.c (build_vector_type_for_mode): Likewise.
12387 * var-tracking.c (use_narrower_mode): Likewise.
12389 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12390 Alan Hayward <alan.hayward@arm.com>
12391 David Sherwood <david.sherwood@arm.com>
12393 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
12394 * builtins.c (expand_builtin_signbit): Use it.
12395 * cfgexpand.c (expand_debug_expr): Likewise.
12396 * dojump.c (do_jump): Likewise.
12397 (do_compare_and_jump): Likewise.
12398 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
12399 * expmed.c (make_tree): Likewise.
12400 * expr.c (expand_expr_real_2): Likewise.
12401 (expand_expr_real_1): Likewise.
12402 (try_casesi): Likewise.
12403 * fold-const-call.c (fold_const_call_ss): Likewise.
12404 * fold-const.c (unextend): Likewise.
12405 (extract_muldiv_1): Likewise.
12406 (fold_single_bit_test): Likewise.
12407 (native_encode_int): Likewise.
12408 (native_encode_string): Likewise.
12409 (native_interpret_int): Likewise.
12410 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
12411 * internal-fn.c (expand_addsub_overflow): Likewise.
12412 (expand_neg_overflow): Likewise.
12413 (expand_mul_overflow): Likewise.
12414 (expand_arith_overflow): Likewise.
12415 * match.pd: Likewise.
12416 * stor-layout.c (layout_type): Likewise.
12417 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
12418 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
12419 * tree-ssanames.c (get_range_info): Likewise.
12420 * tree-switch-conversion.c (array_value_type) Likewise.
12421 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
12422 (vect_recog_divmod_pattern): Likewise.
12423 (vect_recog_mixed_size_cond_pattern): Likewise.
12424 * tree-vrp.c (extract_range_basic): Likewise.
12425 (simplify_float_conversion_using_ranges): Likewise.
12426 * tree.c (int_fits_type_p): Likewise.
12427 * ubsan.c (instrument_bool_enum_load): Likewise.
12428 * varasm.c (mergeable_string_section): Likewise.
12429 (narrowing_initializer_constant_valid_p): Likewise.
12430 (output_constant): Likewise.
12432 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12433 Alan Hayward <alan.hayward@arm.com>
12434 David Sherwood <david.sherwood@arm.com>
12436 * machmode.h (NARROWEST_INT_MODE): New macro.
12437 * expr.c (alignment_for_piecewise_move): Use it instead of
12438 GET_CLASS_NARROWEST_MODE (MODE_INT).
12439 (push_block): Likewise.
12440 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
12442 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12444 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12445 Alan Hayward <alan.hayward@arm.com>
12446 David Sherwood <david.sherwood@arm.com>
12448 * postreload.c (move2add_valid_value_p): Change the type of the
12449 mode parameter to scalar_int_mode.
12450 (move2add_use_add2_insn): Add a mode parameter and use it instead
12452 (move2add_use_add3_insn): Likewise.
12453 (reload_cse_move2add): Update accordingly.
12455 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12456 Alan Hayward <alan.hayward@arm.com>
12457 David Sherwood <david.sherwood@arm.com>
12459 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
12461 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
12462 * optabs.c (expand_unop): Likewise.
12464 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12465 Alan Hayward <alan.hayward@arm.com>
12466 David Sherwood <david.sherwood@arm.com>
12468 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
12469 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
12470 (popcount_loc_descriptor): Likewise.
12471 (bswap_loc_descriptor): Likewise.
12472 (rotate_loc_descriptor): Likewise.
12473 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
12474 calling the functions above.
12476 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12477 Alan Hayward <alan.hayward@arm.com>
12478 David Sherwood <david.sherwood@arm.com>
12480 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
12482 (try_combine): Likewise.
12483 (simplify_if_then_else): Likewise.
12484 * cse.c (cse_insn): Likewise.
12485 * dwarf2out.c (mem_loc_descriptor): Likewise.
12486 * emit-rtl.c (gen_lowpart_common): Likewise.
12487 * simplify-rtx.c (simplify_truncation): Likewise.
12488 (simplify_binary_operation_1): Likewise.
12489 (simplify_const_relational_operation): Likewise.
12490 (simplify_ternary_operation): Likewise.
12491 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
12493 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12494 Alan Hayward <alan.hayward@arm.com>
12495 David Sherwood <david.sherwood@arm.com>
12497 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
12498 * reload.c (push_reload): Likewise.
12499 (find_reloads): Likewise.
12501 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12502 Alan Hayward <alan.hayward@arm.com>
12503 David Sherwood <david.sherwood@arm.com>
12505 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
12506 (make_compound_operation_int): Likewise.
12507 (change_zero_ext): Likewise.
12508 * expr.c (convert_move): Likewise.
12509 (convert_modes): Likewise.
12510 * fwprop.c (forward_propagate_subreg): Likewise.
12511 * loop-iv.c (get_biv_step_1): Likewise.
12512 * optabs.c (widen_operand): Likewise.
12513 * postreload.c (move2add_valid_value_p): Likewise.
12514 * recog.c (simplify_while_replacing): Likewise.
12515 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12516 (simplify_binary_operation_1): Likewise. Remove redundant
12517 mode equality check.
12519 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12520 Alan Hayward <alan.hayward@arm.com>
12521 David Sherwood <david.sherwood@arm.com>
12523 * combine.c (combine_simplify_rtx): Add checks for
12524 is_a <scalar_int_mode>.
12525 (simplify_if_then_else): Likewise.
12526 (make_field_assignment): Likewise.
12527 (simplify_comparison): Likewise.
12528 * ifcvt.c (noce_try_bitop): Likewise.
12529 * loop-invariant.c (canonicalize_address_mult): Likewise.
12530 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12532 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12533 Alan Hayward <alan.hayward@arm.com>
12534 David Sherwood <david.sherwood@arm.com>
12536 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
12537 is_a <scalar_int_mode> instead of != BLKmode.
12539 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12540 Alan Hayward <alan.hayward@arm.com>
12541 David Sherwood <david.sherwood@arm.com>
12543 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
12544 instead of != VOIDmode.
12545 * combine.c (if_then_else_cond): Likewise.
12546 (change_zero_ext): Likewise.
12547 * dwarf2out.c (mem_loc_descriptor): Likewise.
12548 (loc_descriptor): Likewise.
12549 * rtlanal.c (canonicalize_condition): Likewise.
12550 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
12552 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12553 Alan Hayward <alan.hayward@arm.com>
12554 David Sherwood <david.sherwood@arm.com>
12556 * simplify-rtx.c (simplify_binary_operation_1): Use
12557 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
12559 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12560 Alan Hayward <alan.hayward@arm.com>
12561 David Sherwood <david.sherwood@arm.com>
12563 * wide-int.h (int_traits<unsigned char>) New class.
12564 (int_traits<unsigned short>) Likewise.
12565 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
12566 Use GET_MODE_UNIT_PRECISION and remove redundant test for
12568 * combine.c (set_nonzero_bits_and_sign_copies): Use
12569 is_a <scalar_int_mode>.
12570 (find_split_point): Likewise.
12571 (combine_simplify_rtx): Likewise.
12572 (simplify_logical): Likewise.
12573 (expand_compound_operation): Likewise.
12574 (expand_field_assignment): Likewise.
12575 (make_compound_operation): Likewise.
12576 (extended_count): Likewise.
12577 (change_zero_ext): Likewise.
12578 (simplify_comparison): Likewise.
12579 * dwarf2out.c (scompare_loc_descriptor): Likewise.
12580 (ucompare_loc_descriptor): Likewise.
12581 (minmax_loc_descriptor): Likewise.
12582 (mem_loc_descriptor): Likewise.
12583 (loc_descriptor): Likewise.
12584 * expmed.c (init_expmed_one_mode): Likewise.
12585 * lra-constraints.c (lra_constraint_offset): Likewise.
12586 * optabs.c (prepare_libcall_arg): Likewise.
12587 * postreload.c (move2add_note_store): Likewise.
12588 * reload.c (operands_match_p): Likewise.
12589 * rtl.h (load_extend_op): Likewise.
12590 * rtlhooks.c (gen_lowpart_general): Likewise.
12591 * simplify-rtx.c (simplify_truncation): Likewise.
12592 (simplify_unary_operation_1): Likewise.
12593 (simplify_binary_operation_1): Likewise.
12594 (simplify_const_binary_operation): Likewise.
12595 (simplify_const_relational_operation): Likewise.
12596 (simplify_subreg): Likewise.
12597 * stor-layout.c (bitwise_mode_for_mode): Likewise.
12598 * var-tracking.c (adjust_mems): Likewise.
12599 (prepare_call_arguments): Likewise.
12601 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12602 Alan Hayward <alan.hayward@arm.com>
12603 David Sherwood <david.sherwood@arm.com>
12605 * machmode.h (is_int_mode): New fuction.
12606 * combine.c (find_split_point): Use it.
12607 (combine_simplify_rtx): Likewise.
12608 (simplify_if_then_else): Likewise.
12609 (simplify_set): Likewise.
12610 (simplify_shift_const_1): Likewise.
12611 (simplify_comparison): Likewise.
12612 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
12613 * cse.c (notreg_cost): Likewise.
12614 (cse_insn): Likewise.
12615 * cselib.c (cselib_lookup_1): Likewise.
12616 * dojump.c (do_jump_1): Likewise.
12617 (do_compare_rtx_and_jump): Likewise.
12618 * dse.c (get_call_args): Likewise.
12619 * dwarf2out.c (rtl_for_decl_init): Likewise.
12620 (native_encode_initializer): Likewise.
12621 * expmed.c (emit_store_flag_1): Likewise.
12622 (emit_store_flag): Likewise.
12623 * expr.c (convert_modes): Likewise.
12624 (store_field): Likewise.
12625 (expand_expr_real_1): Likewise.
12626 * fold-const.c (fold_read_from_constant_string): Likewise.
12627 * gimple-ssa-sprintf.c (get_format_string): Likewise.
12628 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
12629 * optabs.c (expand_binop): Likewise.
12630 (expand_unop): Likewise.
12631 (expand_abs_nojump): Likewise.
12632 (expand_one_cmpl_abs_nojump): Likewise.
12633 * simplify-rtx.c (mode_signbit_p): Likewise.
12634 (val_signbit_p): Likewise.
12635 (val_signbit_known_set_p): Likewise.
12636 (val_signbit_known_clear_p): Likewise.
12637 (simplify_relational_operation_1): Likewise.
12638 * tree.c (vector_type_mode): Likewise.
12640 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12641 Alan Hayward <alan.hayward@arm.com>
12642 David Sherwood <david.sherwood@arm.com>
12644 * machmode.h (smallest_mode_for_size): Fix formatting.
12645 (smallest_int_mode_for_size): New function.
12646 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
12647 instead of smallest_mode_for_size.
12648 * combine.c (make_extraction): Likewise.
12649 * config/arc/arc.c (arc_expand_movmem): Likewise.
12650 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
12651 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
12652 * config/s390/s390.c (s390_expand_insv): Likewise.
12653 * config/sparc/sparc.c (assign_int_registers): Likewise.
12654 * config/spu/spu.c (spu_function_value): Likewise.
12655 (spu_function_arg): Likewise.
12656 * coverage.c (get_gcov_type): Likewise.
12657 (get_gcov_unsigned_t): Likewise.
12658 * dse.c (find_shift_sequence): Likewise.
12659 * expmed.c (store_bit_field_1): Likewise.
12660 * expr.c (convert_move): Likewise.
12661 (store_field): Likewise.
12662 * internal-fn.c (expand_arith_overflow): Likewise.
12663 * optabs-query.c (get_best_extraction_insn): Likewise.
12664 * optabs.c (expand_twoval_binop_libfunc): Likewise.
12665 * stor-layout.c (layout_type): Likewise.
12666 (initialize_sizetypes): Likewise.
12667 * targhooks.c (default_get_mask_mode): Likewise.
12668 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
12670 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12671 Alan Hayward <alan.hayward@arm.com>
12672 David Sherwood <david.sherwood@arm.com>
12674 * machmode.h (opt_mode::else_blk): New function.
12675 (int_mode_for_mode): Declare.
12676 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
12677 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
12679 * cfgexpand.c (expand_debug_expr): Likewise.
12680 * combine.c (gen_lowpart_or_truncate): Likewise.
12681 (gen_lowpart_for_combine): Likewise.
12682 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
12683 * config/avr/avr.c (avr_to_int_mode): Likewise.
12684 (avr_out_plus_1): Likewise.
12685 (avr_out_plus): Likewise.
12686 (avr_out_round): Likewise.
12687 * config/i386/i386.c (ix86_split_to_parts): Likewise.
12688 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
12689 (s390_expand_vcond): Likewise.
12690 * config/spu/spu.c (spu_split_immediate): Likewise.
12691 (spu_expand_mov): Likewise.
12692 * dse.c (get_stored_val): Likewise.
12693 * expmed.c (store_bit_field_1): Likewise.
12694 (convert_extracted_bit_field): Use int_mode_for_mode instead of
12696 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
12697 (extract_low_bits): Likewise.
12698 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
12699 handling rather than repeating the check.
12700 (emit_group_store): Likewise.
12701 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
12702 * optabs.c (expand_absneg_bit): Likewise.
12703 (expand_copysign_absneg): Likewise.
12704 (expand_copysign_bit): Likewise.
12705 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
12706 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
12707 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
12708 * var-tracking.c (prepare_call_arguments): Likewise.
12709 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
12710 int_mode_for_mode instead of mode_for_size.
12711 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
12713 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12714 Alan Hayward <alan.hayward@arm.com>
12715 David Sherwood <david.sherwood@arm.com>
12717 * machmode.h (int_mode_for_size): New function.
12718 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
12719 instead of mode_for_size.
12720 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
12722 * combine.c (expand_field_assignment): Use int_mode_for_size
12723 instead of mode_for_size.
12724 (make_extraction): Likewise.
12725 (simplify_shift_const_1): Likewise.
12726 (simplify_comparison): Likewise.
12727 * dojump.c (do_jump): Likewise.
12728 * dwarf2out.c (mem_loc_descriptor): Likewise.
12729 * emit-rtl.c (init_derived_machine_modes): Likewise.
12730 * expmed.c (flip_storage_order): Likewise.
12731 (convert_extracted_bit_field): Likewise.
12732 * expr.c (copy_blkmode_from_reg): Likewise.
12733 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
12734 * internal-fn.c (expand_mul_overflow): Likewise.
12735 * lower-subreg.c (simple_move): Likewise.
12736 * optabs-libfuncs.c (init_optabs): Likewise.
12737 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12738 * tree.c (vector_type_mode): Likewise.
12739 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
12740 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
12741 * tree-vect-generic.c (expand_vector_parallel): Likewise.
12742 * tree-vect-stmts.c (vectorizable_load): Likewise.
12743 (vectorizable_store): Likewise.
12745 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12746 Alan Hayward <alan.hayward@arm.com>
12747 David Sherwood <david.sherwood@arm.com>
12749 * coretypes.h (pod_mode): New type.
12750 (scalar_int_mode_pod): New typedef.
12751 * machmode.h (pod_mode): New class.
12752 (int_n_data_t::m): Change type to scalar_int_mode_pod.
12753 * genmodes.c (emit_mode_int_n): Update accordingly.
12754 * lower-subreg.h (target_lower_subreg): Change type to
12755 scalar_int_mode_pod.
12756 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
12757 scalar_int_mode_pod.
12759 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12760 Alan Hayward <alan.hayward@arm.com>
12761 David Sherwood <david.sherwood@arm.com>
12763 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
12764 machine_mode to scalar_int_mode.
12765 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
12766 (rs6000_option_override_internal): Remove cast to int.
12767 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
12768 machine_mode to scalar_int_mode.
12769 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
12770 (rs6000_option_override_internal): Remove cast to int.
12771 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
12772 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
12774 * config/s390/s390.c (s390_expand_builtin): Likewise.
12775 * coretypes.h (scalar_int_mode): New type.
12776 (opt_scalar_int_mode): New typedef.
12777 * machmode.h (scalar_int_mode): New class.
12778 (scalar_int_mode::includes_p): New function.
12779 (byte_mode): Change type to scalar_int_mode.
12780 (word_mode): Likewise.
12781 (ptr_mode): Likewise.
12782 * emit-rtl.c (byte_mode): Likewise.
12783 (word_mode): Likewise.
12784 (ptr_mode): Likewise.
12785 (init_derived_machine_modes): Update accordingly.
12786 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
12787 and MODE_PARTIAL_INT.
12788 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
12789 opt_scalar_int_mode.
12791 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12792 Alan Hayward <alan.hayward@arm.com>
12793 David Sherwood <david.sherwood@arm.com>
12795 * target.def (libgcc_floating_mode_supported_p): Take a
12797 * doc/tm.texi: Regenerate.
12798 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
12800 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
12801 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
12804 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12805 Alan Hayward <alan.hayward@arm.com>
12806 David Sherwood <david.sherwood@arm.com>
12808 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
12809 * doc/tm.texi: Regenerate.
12810 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
12811 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
12812 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
12813 * targhooks.h (default_floatn_mode): Likewise.
12814 * targhooks.c (default_floatn_mode): Likewise.
12815 * tree.c (build_common_tree_nodes): Update accordingly.
12817 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12818 Alan Hayward <alan.hayward@arm.com>
12819 David Sherwood <david.sherwood@arm.com>
12821 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
12822 (mode_iterator::iterate_p): Likewise.
12823 (mode_iterator::get_wider): Likewise.
12824 * expr.c (init_expr_target): Use opt_scalar_float_mode.
12826 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12827 Alan Hayward <alan.hayward@arm.com>
12828 David Sherwood <david.sherwood@arm.com>
12830 * coretypes.h (opt_scalar_float_mode): New typedef.
12831 * machmode.h (float_mode_for_size): New function.
12832 * emit-rtl.c (double_mode): Delete.
12833 (init_emit_once): Use float_mode_for_size.
12834 * stor-layout.c (layout_type): Likewise.
12835 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
12837 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12838 Alan Hayward <alan.hayward@arm.com>
12839 David Sherwood <david.sherwood@arm.com>
12841 * output.h (assemble_real): Take a scalar_float_mode.
12842 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
12843 * config/arm/arm.md (consttable_4): Likewise.
12844 (consttable_8): Likewise.
12845 (consttable_16): Likewise.
12846 * config/mips/mips.md (consttable_float): Likewise.
12847 * config/s390/s390.c (s390_output_pool_entry): Likewise.
12848 * varasm.c (assemble_real): Take a scalar_float_mode.
12849 (output_constant_pool_2): Update accordingly.
12850 (output_constant): Likewise.
12852 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12853 Alan Hayward <alan.hayward@arm.com>
12854 David Sherwood <david.sherwood@arm.com>
12856 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
12857 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
12858 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
12859 (native_encode_real): Likewise.
12860 (native_interpret_real): Likewise.
12861 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
12862 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12864 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12865 Alan Hayward <alan.hayward@arm.com>
12866 David Sherwood <david.sherwood@arm.com>
12868 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
12869 <scalar_float_mode>. Simplify.
12870 (gen_extend_conv_libfunc): Likewise.
12872 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12873 Alan Hayward <alan.hayward@arm.com>
12874 David Sherwood <david.sherwood@arm.com>
12876 * coretypes.h (scalar_float_mode): New type.
12877 * machmode.h (mode_traits::from_int): Use machine_mode if
12878 USE_ENUM_MODES is defined.
12879 (is_a): New function.
12881 (dyn_cast): Likewise.
12882 (scalar_float_mode): New class.
12883 (scalar_float_mode::includes_p): New function.
12884 (is_float_mode): Likewise.
12885 * gdbhooks.py (MachineModePrinter): New class.
12886 (build_pretty_printer): Use it for scalar_float_mode.
12887 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
12888 (format_helper::format_helper): Turn into a template.
12889 * genmodes.c (get_mode_class): New function.
12890 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
12891 or machine_mode if none.
12892 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
12893 as_a <scalar_float_mode>.
12894 * dwarf2out.c (mem_loc_descriptor): Likewise.
12895 (insert_float): Likewise.
12896 (add_const_value_attribute): Likewise.
12897 * simplify-rtx.c (simplify_immed_subreg): Likewise.
12898 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
12899 (expand_unop): Update accordingly.
12900 (expand_abs_nojump): Likewise.
12901 (expand_copysign_absneg): Take a scalar_float_mode.
12902 (expand_copysign_bit): Likewise.
12903 (expand_copysign): Update accordingly.
12905 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12906 Alan Hayward <alan.hayward@arm.com>
12907 David Sherwood <david.sherwood@arm.com>
12909 * coretypes.h (opt_mode): New class.
12910 * machmode.h (opt_mode): Likewise.
12911 (opt_mode::else_void): New function.
12912 (opt_mode::require): Likewise.
12913 (opt_mode::exists): Likewise.
12914 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
12915 (GET_MODE_2XWIDER_MODE): Likewise.
12916 (mode_iterator::get_wider): Update accordingly.
12917 (mode_iterator::get_2xwider): Likewise.
12918 (mode_iterator::get_known_wider): Likewise, turning into a template.
12919 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
12920 forcing a wider mode to exist.
12921 * config/cr16/cr16.h (LONG_REG_P): Likewise.
12922 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
12923 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
12924 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
12925 * lower-subreg.c (init_lower_subreg): Likewise.
12926 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
12927 on the final iteration.
12928 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
12929 a wider mode exists before asking for a move pattern.
12930 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
12931 forcing a wider mode to exist.
12932 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
12933 returning false if no such mode exists.
12934 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
12935 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
12936 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
12937 Avoid checking for a MODE_INT if we already know the mode is not a
12939 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
12940 forcing a wider mode to exist.
12941 (expmed_mult_highpart_optab): Likewise.
12942 (expmed_mult_highpart): Likewise.
12943 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
12945 * lto-streamer-in.c (lto_input_mode_table): Likewise.
12946 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
12947 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
12948 * internal-fn.c (expand_mul_overflow): Update use of
12949 GET_MODE_2XWIDER_MODE.
12950 * omp-low.c (omp_clause_aligned_alignment): Likewise.
12951 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
12952 GET_MODE_WIDER_MODE.
12953 (convert_plusminus_to_widen): Likewise.
12954 * tree-switch-conversion.c (array_value_type): Likewise.
12955 * var-tracking.c (emit_note_insn_var_location): Likewise.
12956 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12957 Return false inside rather than outside the loop if no wider mode
12959 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
12960 and GET_MODE_2XWIDER_MODE
12961 (can_compare_p): Use else_void.
12962 * gdbhooks.py (OptMachineModePrinter): New class.
12963 (build_pretty_printer): Use it for opt_mode.
12965 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12966 Alan Hayward <alan.hayward@arm.com>
12967 David Sherwood <david.sherwood@arm.com>
12969 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
12970 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
12972 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12973 Alan Hayward <alan.hayward@arm.com>
12974 David Sherwood <david.sherwood@arm.com>
12976 * machmode.h (mode_traits): New structure.
12977 (get_narrowest_mode): New function.
12978 (mode_iterator::start): Likewise.
12979 (mode_iterator::iterate_p): Likewise.
12980 (mode_iterator::get_wider): Likewise.
12981 (mode_iterator::get_known_wider): Likewise.
12982 (mode_iterator::get_2xwider): Likewise.
12983 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
12984 (FOR_EACH_MODE): Likewise.
12985 (FOR_EACH_MODE_FROM): Likewise.
12986 (FOR_EACH_MODE_UNTIL): Likewise.
12987 (FOR_EACH_WIDER_MODE): Likewise.
12988 (FOR_EACH_2XWIDER_MODE): Likewise.
12989 * builtins.c (expand_builtin_strlen): Use new mode iterators.
12990 * combine.c (simplify_comparison): Likewise
12991 * config/i386/i386.c (type_natural_mode): Likewise.
12992 * cse.c (cse_insn): Likewise.
12993 * dse.c (find_shift_sequence): Likewise.
12994 * emit-rtl.c (init_derived_machine_modes): Likewise.
12995 (init_emit_once): Likewise.
12996 * explow.c (hard_function_value): Likewise.
12997 * expmed.c (extract_fixed_bit_field_1): Likewise.
12998 (extract_bit_field_1): Likewise.
12999 (expand_divmod): Likewise.
13000 (emit_store_flag_1): Likewise.
13001 * expr.c (init_expr_target): Likewise.
13002 (convert_move): Likewise.
13003 (alignment_for_piecewise_move): Likewise.
13004 (widest_int_mode_for_size): Likewise.
13005 (emit_block_move_via_movmem): Likewise.
13006 (copy_blkmode_to_reg): Likewise.
13007 (set_storage_via_setmem): Likewise.
13008 (compress_float_constant): Likewise.
13009 * omp-low.c (omp_clause_aligned_alignment): Likewise.
13010 * optabs-query.c (get_best_extraction_insn): Likewise.
13011 * optabs.c (expand_binop): Likewise.
13012 (expand_twoval_unop): Likewise.
13013 (expand_twoval_binop): Likewise.
13014 (widen_leading): Likewise.
13015 (widen_bswap): Likewise.
13016 (expand_parity): Likewise.
13017 (expand_unop): Likewise.
13018 (prepare_cmp_insn): Likewise.
13019 (prepare_float_lib_cmp): Likewise.
13020 (expand_float): Likewise.
13021 (expand_fix): Likewise.
13022 (expand_sfix_optab): Likewise.
13023 * postreload.c (move2add_use_add2_insn): Likewise.
13024 * reg-stack.c (reg_to_stack): Likewise.
13025 * reginfo.c (choose_hard_reg_mode): Likewise.
13026 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
13027 * stor-layout.c (mode_for_size): Likewise.
13028 (smallest_mode_for_size): Likewise.
13029 (mode_for_vector): Likewise.
13030 (finish_bitfield_representative): Likewise.
13031 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
13032 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
13033 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
13034 * var-tracking.c (prepare_call_arguments): Likewise.
13036 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13037 Alan Hayward <alan.hayward@arm.com>
13038 David Sherwood <david.sherwood@arm.com>
13040 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
13041 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
13042 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
13043 * machmode.h (mode_size): Move earlier in file.
13044 (mode_precision): Likewise.
13045 (mode_inner): Likewise.
13046 (mode_nunits): Likewise.
13047 (mode_unit_size): Likewise.
13048 (unit_unit_precision): Likewise.
13049 (mode_wider): Likewise.
13050 (mode_2xwider): Likewise.
13051 (machine_mode): New class.
13052 (mode_to_bytes): New function.
13053 (mode_to_bits): Likewise.
13054 (mode_to_precision): Likewise.
13055 (mode_to_inner): Likewise.
13056 (mode_to_unit_size): Likewise.
13057 (mode_to_unit_precision): Likewise.
13058 (mode_to_nunits): Likewise.
13059 (GET_MODE_SIZE): Use mode_to_bytes.
13060 (GET_MODE_BITSIZE): Use mode_to_bits.
13061 (GET_MODE_PRECISION): Use mode_to_precision.
13062 (GET_MODE_INNER): Use mode_to_inner.
13063 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
13064 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
13065 (GET_MODE_NUNITS): Use mode_to_nunits.
13066 * system.h (ALWAYS_INLINE): New macro.
13067 * config/powerpcspe/powerpcspe-c.c
13068 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
13069 int for arg1_mode and arg2_mode.
13071 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13072 Alan Hayward <alan.hayward@arm.com>
13073 David Sherwood <david.sherwood@arm.com>
13075 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
13076 Prefix mode names with E_ in case statements.
13077 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13078 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
13079 (aarch64_split_simd_move): Likewise.
13080 (aarch64_gen_storewb_pair): Likewise.
13081 (aarch64_gen_loadwb_pair): Likewise.
13082 (aarch64_gen_store_pair): Likewise.
13083 (aarch64_gen_load_pair): Likewise.
13084 (aarch64_get_condition_code_1): Likewise.
13085 (aarch64_constant_pool_reload_icode): Likewise.
13086 (get_rsqrte_type): Likewise.
13087 (get_rsqrts_type): Likewise.
13088 (get_recpe_type): Likewise.
13089 (get_recps_type): Likewise.
13090 (aarch64_gimplify_va_arg_expr): Likewise.
13091 (aarch64_simd_container_mode): Likewise.
13092 (aarch64_emit_load_exclusive): Likewise.
13093 (aarch64_emit_store_exclusive): Likewise.
13094 (aarch64_expand_compare_and_swap): Likewise.
13095 (aarch64_gen_atomic_cas): Likewise.
13096 (aarch64_emit_bic): Likewise.
13097 (aarch64_emit_atomic_swap): Likewise.
13098 (aarch64_emit_atomic_load_op): Likewise.
13099 (aarch64_evpc_trn): Likewise.
13100 (aarch64_evpc_uzp): Likewise.
13101 (aarch64_evpc_zip): Likewise.
13102 (aarch64_evpc_ext): Likewise.
13103 (aarch64_evpc_rev): Likewise.
13104 (aarch64_evpc_dup): Likewise.
13105 (aarch64_gen_ccmp_first): Likewise.
13106 (aarch64_gen_ccmp_next): Likewise.
13107 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
13108 (alpha_emit_xfloating_libcall): Likewise.
13109 (emit_insxl): Likewise.
13110 (alpha_arg_type): Likewise.
13111 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
13112 (arc_preferred_simd_mode): Likewise.
13113 (arc_secondary_reload): Likewise.
13114 (get_arc_condition_code): Likewise.
13115 (arc_print_operand): Likewise.
13116 (arc_legitimate_constant_p): Likewise.
13117 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13118 * config/arc/arc.md (casesi_load): Likewise.
13119 (casesi_compact_jump): Likewise.
13120 * config/arc/predicates.md (proper_comparison_operator): Likewise.
13121 (cc_use_register): Likewise.
13122 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13123 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
13124 (arm_init_iwmmxt_builtins): Likewise.
13125 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
13126 (neon_expand_vector_init): Likewise.
13127 (arm_attr_length_move_neon): Likewise.
13128 (maybe_get_arm_condition_code): Likewise.
13129 (arm_emit_vector_const): Likewise.
13130 (arm_preferred_simd_mode): Likewise.
13131 (arm_output_iwmmxt_tinsr): Likewise.
13132 (thumb1_output_casesi): Likewise.
13133 (thumb2_output_casesi): Likewise.
13134 (arm_emit_load_exclusive): Likewise.
13135 (arm_emit_store_exclusive): Likewise.
13136 (arm_expand_compare_and_swap): Likewise.
13137 (arm_evpc_neon_vuzp): Likewise.
13138 (arm_evpc_neon_vzip): Likewise.
13139 (arm_evpc_neon_vrev): Likewise.
13140 (arm_evpc_neon_vtrn): Likewise.
13141 (arm_evpc_neon_vext): Likewise.
13142 (arm_validize_comparison): Likewise.
13143 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
13144 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
13145 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
13146 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
13147 (c6x_preferred_simd_mode): Likewise.
13148 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
13149 (epiphany_rtx_costs): Likewise.
13150 * config/epiphany/predicates.md (proper_comparison_operator):
13152 * config/frv/frv.c (condexec_memory_operand): Likewise.
13153 (frv_emit_move): Likewise.
13154 (output_move_single): Likewise.
13155 (output_condmove_single): Likewise.
13156 (frv_hard_regno_mode_ok): Likewise.
13157 (frv_matching_accg_mode): Likewise.
13158 * config/h8300/h8300.c (split_adds_subs): Likewise.
13159 (h8300_rtx_costs): Likewise.
13160 (h8300_print_operand): Likewise.
13161 (compute_mov_length): Likewise.
13162 (output_logical_op): Likewise.
13163 (compute_logical_op_length): Likewise.
13164 (compute_logical_op_cc): Likewise.
13165 (h8300_shift_needs_scratch_p): Likewise.
13166 (output_a_shift): Likewise.
13167 (compute_a_shift_length): Likewise.
13168 (compute_a_shift_cc): Likewise.
13169 (expand_a_rotate): Likewise.
13170 (output_a_rotate): Likewise.
13171 * config/i386/i386.c (classify_argument): Likewise.
13172 (function_arg_advance_32): Likewise.
13173 (function_arg_32): Likewise.
13174 (function_arg_64): Likewise.
13175 (function_value_64): Likewise.
13176 (ix86_gimplify_va_arg): Likewise.
13177 (ix86_legitimate_constant_p): Likewise.
13178 (put_condition_code): Likewise.
13179 (split_double_mode): Likewise.
13180 (ix86_avx256_split_vector_move_misalign): Likewise.
13181 (ix86_expand_vector_logical_operator): Likewise.
13182 (ix86_split_idivmod): Likewise.
13183 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
13184 (ix86_build_const_vector): Likewise.
13185 (ix86_build_signbit_mask): Likewise.
13186 (ix86_match_ccmode): Likewise.
13187 (ix86_cc_modes_compatible): Likewise.
13188 (ix86_expand_branch): Likewise.
13189 (ix86_expand_sse_cmp): Likewise.
13190 (ix86_expand_sse_movcc): Likewise.
13191 (ix86_expand_int_sse_cmp): Likewise.
13192 (ix86_expand_vec_perm_vpermi2): Likewise.
13193 (ix86_expand_vec_perm): Likewise.
13194 (ix86_expand_sse_unpack): Likewise.
13195 (ix86_expand_int_addcc): Likewise.
13196 (ix86_split_to_parts): Likewise.
13197 (ix86_vectorize_builtin_gather): Likewise.
13198 (ix86_vectorize_builtin_scatter): Likewise.
13199 (avx_vpermilp_parallel): Likewise.
13200 (inline_memory_move_cost): Likewise.
13201 (ix86_tieable_integer_mode_p): Likewise.
13202 (x86_maybe_negate_const_int): Likewise.
13203 (ix86_expand_vector_init_duplicate): Likewise.
13204 (ix86_expand_vector_init_one_nonzero): Likewise.
13205 (ix86_expand_vector_init_one_var): Likewise.
13206 (ix86_expand_vector_init_concat): Likewise.
13207 (ix86_expand_vector_init_interleave): Likewise.
13208 (ix86_expand_vector_init_general): Likewise.
13209 (ix86_expand_vector_set): Likewise.
13210 (ix86_expand_vector_extract): Likewise.
13211 (emit_reduc_half): Likewise.
13212 (ix86_emit_i387_round): Likewise.
13213 (ix86_mangle_type): Likewise.
13214 (ix86_expand_round_sse4): Likewise.
13215 (expand_vec_perm_blend): Likewise.
13216 (canonicalize_vector_int_perm): Likewise.
13217 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
13218 (expand_vec_perm_1): Likewise.
13219 (expand_vec_perm_interleave3): Likewise.
13220 (expand_vec_perm_even_odd_pack): Likewise.
13221 (expand_vec_perm_even_odd_1): Likewise.
13222 (expand_vec_perm_broadcast_1): Likewise.
13223 (ix86_vectorize_vec_perm_const_ok): Likewise.
13224 (ix86_expand_vecop_qihi): Likewise.
13225 (ix86_expand_mul_widen_hilo): Likewise.
13226 (ix86_expand_sse2_abs): Likewise.
13227 (ix86_expand_pextr): Likewise.
13228 (ix86_expand_pinsr): Likewise.
13229 (ix86_preferred_simd_mode): Likewise.
13230 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
13231 * config/i386/sse.md (*andnot<mode>3): Likewise.
13232 (<mask_codefor><code><mode>3<mask_name>): Likewise.
13233 (*<code><mode>3): Likewise.
13234 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
13235 (ia64_expand_atomic_op): Likewise.
13236 (ia64_arg_type): Likewise.
13237 (ia64_mode_to_int): Likewise.
13238 (ia64_scalar_mode_supported_p): Likewise.
13239 (ia64_vector_mode_supported_p): Likewise.
13240 (expand_vec_perm_broadcast): Likewise.
13241 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
13242 (iq2000_function_arg_advance): Likewise.
13243 (iq2000_function_arg): Likewise.
13244 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
13245 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
13246 (m68k_libcall_value): Likewise.
13247 (m68k_function_value): Likewise.
13248 (sched_attr_op_type): Likewise.
13249 * config/mcore/mcore.c (mcore_output_move): Likewise.
13250 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
13252 (microblaze_function_arg): Likewise.
13253 * config/mips/mips.c (mips16_build_call_stub): Likewise.
13254 (mips_print_operand): Likewise.
13255 (mips_mode_ok_for_mov_fmt_p): Likewise.
13256 (mips_vector_mode_supported_p): Likewise.
13257 (mips_preferred_simd_mode): Likewise.
13258 (mips_expand_vpc_loongson_even_odd): Likewise.
13259 (mips_expand_vec_unpack): Likewise.
13260 (mips_expand_vi_broadcast): Likewise.
13261 (mips_expand_vector_init): Likewise.
13262 (mips_expand_vec_reduc): Likewise.
13263 (mips_expand_msa_cmp): Likewise.
13264 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
13265 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
13266 (cc_flags_for_mode): Likewise.
13267 * config/msp430/msp430.c (msp430_print_operand): Likewise.
13268 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
13269 (nds32_output_casesi_pc_relative): Likewise.
13270 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13271 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
13272 (nvptx_gen_unpack): Likewise.
13273 (nvptx_gen_pack): Likewise.
13274 (nvptx_gen_shuffle): Likewise.
13275 (nvptx_gen_wcast): Likewise.
13276 (nvptx_preferred_simd_mode): Likewise.
13277 * config/pa/pa.c (pa_secondary_reload): Likewise.
13278 * config/pa/predicates.md (base14_operand): Likewise.
13279 * config/powerpcspe/powerpcspe-c.c
13280 (altivec_resolve_overloaded_builtin): Likewise.
13281 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
13283 (rs6000_preferred_simd_mode): Likewise.
13284 (output_vec_const_move): Likewise.
13285 (rs6000_expand_vector_extract): Likewise.
13286 (rs6000_split_vec_extract_var): Likewise.
13287 (reg_offset_addressing_ok_p): Likewise.
13288 (rs6000_legitimate_offset_address_p): Likewise.
13289 (rs6000_legitimize_address): Likewise.
13290 (rs6000_emit_set_const): Likewise.
13291 (rs6000_const_vec): Likewise.
13292 (rs6000_emit_move): Likewise.
13293 (spe_build_register_parallel): Likewise.
13294 (rs6000_darwin64_record_arg_recurse): Likewise.
13295 (swap_selector_for_mode): Likewise.
13296 (spe_init_builtins): Likewise.
13297 (paired_init_builtins): Likewise.
13298 (altivec_init_builtins): Likewise.
13299 (do_load_for_compare): Likewise.
13300 (rs6000_generate_compare): Likewise.
13301 (rs6000_expand_float128_convert): Likewise.
13302 (emit_load_locked): Likewise.
13303 (emit_store_conditional): Likewise.
13304 (rs6000_output_function_epilogue): Likewise.
13305 (rs6000_handle_altivec_attribute): Likewise.
13306 (rs6000_function_value): Likewise.
13307 (emit_fusion_gpr_load): Likewise.
13308 (emit_fusion_p9_load): Likewise.
13309 (emit_fusion_p9_store): Likewise.
13310 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
13311 (fusion_gpr_mem_load): Likewise.
13312 (fusion_addis_mem_combo_load): Likewise.
13313 (fusion_addis_mem_combo_store): Likewise.
13314 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
13315 (fusion_gpr_mem_load): Likewise.
13316 (fusion_addis_mem_combo_load): Likewise.
13317 (fusion_addis_mem_combo_store): Likewise.
13318 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13320 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
13321 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
13322 (rs6000_preferred_simd_mode): Likewise.
13323 (output_vec_const_move): Likewise.
13324 (rs6000_expand_vector_extract): Likewise.
13325 (rs6000_split_vec_extract_var): Likewise.
13326 (reg_offset_addressing_ok_p): Likewise.
13327 (rs6000_legitimate_offset_address_p): Likewise.
13328 (rs6000_legitimize_address): Likewise.
13329 (rs6000_emit_set_const): Likewise.
13330 (rs6000_const_vec): Likewise.
13331 (rs6000_emit_move): Likewise.
13332 (rs6000_darwin64_record_arg_recurse): Likewise.
13333 (swap_selector_for_mode): Likewise.
13334 (paired_init_builtins): Likewise.
13335 (altivec_init_builtins): Likewise.
13336 (rs6000_expand_float128_convert): Likewise.
13337 (emit_load_locked): Likewise.
13338 (emit_store_conditional): Likewise.
13339 (rs6000_output_function_epilogue): Likewise.
13340 (rs6000_handle_altivec_attribute): Likewise.
13341 (rs6000_function_value): Likewise.
13342 (emit_fusion_gpr_load): Likewise.
13343 (emit_fusion_p9_load): Likewise.
13344 (emit_fusion_p9_store): Likewise.
13345 * config/rx/rx.c (rx_gen_move_template): Likewise.
13346 (flags_from_mode): Likewise.
13347 * config/s390/predicates.md (s390_alc_comparison): Likewise.
13348 (s390_slb_comparison): Likewise.
13349 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
13350 (s390_vector_mode_supported_p): Likewise.
13351 (s390_cc_modes_compatible): Likewise.
13352 (s390_match_ccmode_set): Likewise.
13353 (s390_canonicalize_comparison): Likewise.
13354 (s390_emit_compare_and_swap): Likewise.
13355 (s390_branch_condition_mask): Likewise.
13356 (s390_rtx_costs): Likewise.
13357 (s390_secondary_reload): Likewise.
13358 (__SECONDARY_RELOAD_CASE): Likewise.
13359 (s390_expand_cs): Likewise.
13360 (s390_preferred_simd_mode): Likewise.
13361 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
13362 * config/sh/sh.c (sh_print_operand): Likewise.
13363 (dump_table): Likewise.
13364 (sh_secondary_reload): Likewise.
13365 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13366 * config/sh/sh.md (casesi_worker_1): Likewise.
13367 (casesi_worker_2): Likewise.
13368 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
13369 (fcc_comparison_operator): Likewise.
13370 * config/sparc/sparc.c (sparc_expand_move): Likewise.
13371 (emit_soft_tfmode_cvt): Likewise.
13372 (sparc_preferred_simd_mode): Likewise.
13373 (output_cbranch): Likewise.
13374 (sparc_print_operand): Likewise.
13375 (sparc_expand_vec_perm_bmask): Likewise.
13376 (vector_init_bshuffle): Likewise.
13377 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
13378 (spu_vector_mode_supported_p): Likewise.
13379 (spu_expand_insv): Likewise.
13380 (spu_emit_branch_or_set): Likewise.
13381 (spu_handle_vector_attribute): Likewise.
13382 (spu_builtin_splats): Likewise.
13383 (spu_builtin_extract): Likewise.
13384 (spu_builtin_promote): Likewise.
13385 (spu_expand_sign_extend): Likewise.
13386 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
13387 (tilegx_simd_int): Likewise.
13388 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
13389 (tilepro_simd_int): Likewise.
13390 * config/v850/v850.c (const_double_split): Likewise.
13391 (v850_print_operand): Likewise.
13392 (ep_memory_offset): Likewise.
13393 * config/vax/vax.c (vax_rtx_costs): Likewise.
13394 (vax_output_int_move): Likewise.
13395 (vax_output_int_add): Likewise.
13396 (vax_output_int_subtract): Likewise.
13397 * config/visium/predicates.md (visium_branch_operator): Likewise.
13398 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
13399 (visium_print_operand_address): Likewise.
13400 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13401 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
13402 (xtensa_expand_conditional_branch): Likewise.
13403 (xtensa_copy_incoming_a7): Likewise.
13404 (xtensa_output_literal): Likewise.
13405 * dfp.c (decimal_real_maxval): Likewise.
13406 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
13408 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13409 Alan Hayward <alan.hayward@arm.com>
13410 David Sherwood <david.sherwood@arm.com>
13412 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
13413 (mode_nunits_inline): Likewise.
13414 (mode_inner_inline): Likewise.
13415 (mode_unit_size_inline): Likewise.
13416 (mode_unit_precision_inline): Likewise.
13417 (emit_insn_modes_h): Likewise. Also emit a #define of the
13419 (emit_mode_wider): Add an E_ prefix to mode names.
13420 (emit_mode_complex): Likewise.
13421 (emit_mode_inner): Likewise.
13422 (emit_mode_adjustments): Likewise.
13423 (emit_mode_int_n): Likewise.
13424 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
13425 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
13426 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
13427 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
13428 (CRC32_BUILTIN, ENTRY): Likewise.
13429 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
13430 (aarch64_pop_regs): Likewise.
13431 (aarch64_process_components): Likewise.
13432 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
13433 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
13434 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
13435 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
13436 * config/arm/arm.c (arm_init_libfuncs): Likewise.
13437 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
13439 * config/i386/i386-builtin.def (pcmpestr): Likewise.
13440 (pcmpistr): Likewise.
13441 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
13442 * config/mmix/mmix.c (mmix_output_condition): Likewise.
13443 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
13445 * config/rl78/rl78.c (mduc_regs): Likewise.
13446 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
13447 (htm_expand_builtin): Likewise.
13448 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
13449 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
13450 * config/xtensa/xtensa.c (print_operand): Likewise.
13451 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
13452 (NUM_MODE_VECTOR_INT): Likewise.
13453 * genoutput.c (null_operand): Likewise.
13454 (output_operand_data): Likewise.
13455 * genrecog.c (print_parameter_value): Likewise.
13456 * lra.c (debug_operand_data): Likewise.
13458 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13460 * dwarf2out.c (scompare_loc_descriptor_wide)
13461 (scompare_loc_descriptor_narrow): New functions, split out from...
13462 (scompare_loc_descriptor): ...here.
13463 * expmed.c (emit_store_flag_int): New function, split out from...
13464 (emit_store_flag): ...here.
13466 2017-08-30 Richard Biener <rguenther@suse.de>
13468 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
13469 (dwarf2out_early_finish): Move setting of AT_pubnames from
13470 early debug output to early finish.
13472 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13474 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
13475 and -mdata-region to the assembler.
13477 2017-08-30 Richard Biener <rguenther@suse.de>
13479 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
13481 (gen_subprogram_die): Add DW_AT_object_pointer only early.
13482 (dwarf2out_early_global_decl): Only generate a DIE for the
13483 abstract origin if it doesn't already exist or is a declaration DIE.
13484 (resolve_addr): Do not add the linkage name twice when
13485 generating a stub DIE for the DW_TAG_GNU_call_site target.
13487 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13489 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13490 Use machine_mode rather than int for arg1_mode.
13492 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
13495 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
13496 that the second argument of the built-in functions to unpack
13497 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
13498 switch statement instead a lot of if statements.
13499 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
13500 Allow 64-bit values to be in Altivec registers as well as
13501 traditional floating point registers.
13502 (pack<mode>, FMOVE128_VSX iterator): Likewise.
13504 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
13506 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
13507 MAX_REGS_PER_ADDRESS == 1.
13509 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
13511 * config/i386/i386.opt (flag_fentry): Do not init to -1.
13512 * config/i386/i386.c (ix86_option_override_internal): Simplify
13513 setting of opts->x_flag_entry.
13515 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13516 Jakub Jelinek <jakub@redhat.com>
13517 Richard Biener <rguenther@suse.de>
13519 PR tree-optimization/81503
13520 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
13521 folded constant fits in the target type; reorder tests for clarity.
13523 2017-08-29 Martin Liska <mliska@suse.cz>
13525 * passes.def: Include pass_lower_switch.
13526 * stmt.c (dump_case_nodes): Remove and move to
13527 tree-switch-conversion.
13528 (case_values_threshold): Likewise.
13529 (expand_switch_as_decision_tree_p): Likewise.
13530 (emit_case_decision_tree): Likewise.
13531 (expand_case): Likewise.
13532 (balance_case_nodes): Likewise.
13533 (node_has_low_bound): Likewise.
13534 (node_has_high_bound): Likewise.
13535 (node_is_bounded): Likewise.
13536 (emit_case_nodes): Likewise.
13537 (struct simple_case_node): New struct.
13538 (add_case_node): Remove.
13539 (emit_case_dispatch_table): Use vector instead of case_list.
13540 (reset_out_edges_aux): Remove.
13541 (compute_cases_per_edge): Likewise.
13542 (expand_case): Build list of simple_case_node.
13543 (expand_sjlj_dispatch_table): Use it.
13544 * tree-switch-conversion.c (struct case_node): Moved from
13545 stmt.c and adjusted.
13546 (emit_case_nodes): Likewise.
13547 (node_has_low_bound): Likewise.
13548 (node_has_high_bound): Likewise.
13549 (node_is_bounded): Likewise.
13550 (case_values_threshold): Likewise.
13551 (reset_out_edges_aux): Likewise.
13552 (compute_cases_per_edge): Likewise.
13553 (add_case_node): Likewise.
13554 (dump_case_nodes): Likewise.
13555 (balance_case_nodes): Likewise.
13556 (expand_switch_as_decision_tree_p): Likewise.
13557 (emit_jump): Likewise.
13558 (emit_case_decision_tree): Likewise.
13559 (try_switch_expansion): Likewise.
13560 (do_jump_if_equal): Likewise.
13561 (emit_cmp_and_jump_insns): Likewise.
13562 (fix_phi_operands_for_edge): New function.
13563 (record_phi_operand_mapping): Likewise.
13564 (class pass_lower_switch): New pass.
13565 (pass_lower_switch::execute): New function.
13566 (make_pass_lower_switch): Likewise.
13567 (conditional_probability):
13568 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
13569 * tree-pass.h: Add make_pass_lower_switch.
13571 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13574 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
13577 2017-08-29 Richard Biener <rguenther@suse.de>
13579 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
13580 we do not add a DW_AT_inline attribute twice.
13581 (gen_subprogram_die): Remove code setting DW_AT_inline on
13582 DECL_ABSTRACT_P nodes.
13584 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
13586 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
13587 calls to internal functions.
13588 (gimplify_modify_expr): Likewise.
13589 * tree-call-cdce.c (use_internal_fn): Likewise.
13590 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
13591 (convert_to_divmod): Set the nothrow flag.
13592 * tree-if-conv.c (predicate_mem_writes): Likewise.
13593 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
13594 (vectorizable_call): Likewise.
13595 (vectorizable_store): Likewise.
13596 (vectorizable_load): Likewise.
13597 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
13598 (vect_recog_mask_conversion_pattern): Likewise.
13600 2017-08-29 Martin Liska <mliska@suse.cz>
13603 * gcc.c (driver_handle_option): Add new argument.
13604 * opts-common.c (handle_option): Pass
13605 target_option_override_hook.
13606 * opts-global.c (lang_handle_option): Add new option.
13607 (set_default_handlers): Add new argument.
13608 (decode_options): Likewise.
13609 * opts.c (target_handle_option): Likewise.
13610 (common_handle_option): Call target_option_override_hook.
13611 * opts.h (struct cl_option_handler_func): Add hook for
13612 target option override.
13613 (struct cl_option_handlers): Likewise.
13614 (set_default_handlers): Add new argument.
13615 (decode_options): Likewise.
13616 (common_handle_option): Likewise.
13617 (target_handle_option): Likewise.
13618 * toplev.c (toplev::main): Pass targetm.target_option.override
13621 2017-08-29 Richard Biener <rguenther@suse.de>
13622 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
13624 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
13625 life to the active subtree.
13627 2017-08-28 Jeff Law <law@redhat.com>
13629 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
13630 derive_equivalences.
13631 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
13632 Code moved into....
13633 (edge_info::derive_equivalences): New private member function
13635 * tree-ssa-dom.c (class edge_info): Changed from a struct
13636 to a class. Add ctor/dtor, methods and data members.
13637 (edge_info::edge_info): Renamed from allocate_edge_info.
13638 Initialize additional members.
13639 (edge_info::~edge_info): New.
13640 (free_dom_edge_info): Delete the edge info.
13641 (record_edge_info): Use new class & associated member functions.
13642 Tighten forms for testing for edge equivalences.
13643 (record_temporary_equivalences): Iterate over the simple
13644 equivalences rather than assuming there's only one per edge.
13645 (cprop_into_successor_phis): Iterate over the simple
13646 equivalences rather than assuming there's only one per edge.
13647 (optimize_stmt): Use operand_equal_p rather than pointer
13648 equality for mini-DSE code.
13650 2017-08-28 Nathan Sidwell <nathan@acm.org>
13652 * gcc.c (execute): Fold SIGPIPE handling into switch
13653 statement. Adjust internal error message.
13655 2017-08-28 Richard Biener <rguenther@suse.de>
13658 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
13659 Do nothing for removed DIEs.
13661 2017-08-28 Richard Biener <rguenther@suse.de>
13663 PR tree-optimization/81977
13664 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
13667 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
13670 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
13671 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
13672 using targetm.gen_mem_thread_fence.
13674 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
13677 * config/i386/i386.md (*<btsc><mode>): Change operand 2
13678 predicate to register_operand. Reorder operands.
13679 (*btr<mode>): Ditto.
13680 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
13681 (*btr<mode>_mask): Ditto.
13683 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
13685 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
13686 * config/rs6000/xmmintrin.h: New file.
13687 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
13689 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13692 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
13693 parameter and_insn and return it.
13694 (recombine_lvx_pattern): Insert a copy to ensure availability of
13695 the base register of the copied masking operation at the point of
13696 the instruction replacement.
13697 (recombine_stvx_pattern): Likewise.
13699 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
13701 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
13702 undocumented switches.
13703 (-mpower9-dform-vector): Likewise.
13704 (-mpower9-dform): Likewise.
13705 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
13706 comments to delete references to -mpower9-dform* switches.
13707 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
13708 Delete reference to -mpower9-dform* switches, test for
13709 -mpower9-vector instead.
13710 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
13711 (OTHER_P9_VECTOR_MASKS): Likewise.
13712 (POWERPC_MASKS): Likewise.
13713 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
13714 tests against -mpower9-dform* to -mpower9-vector. Delete code
13715 that checked for -mpower9-dform* consistancy with other options.
13716 Add test for -mpower9-misc to enable other power9 switches.
13717 (rs6000_init_hard_regno_mode_ok): Likewise.
13718 (rs6000_option_override_internal): Likewise.
13719 (rs6000_emit_prologue): Likewise.
13720 (rs6000_emit_epilogue): Likewise.
13721 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
13722 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
13723 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
13725 (emit_fusion_p9_store): Likewise.
13726 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
13727 resetting these macros if the assembler does not support ISA 3.0
13729 (TARGET_P9_DFORM_VECTOR): Likewise.
13730 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
13731 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
13733 2017-08-25 Alan Modra <amodra@gmail.com>
13735 PR rtl-optimization/81747
13736 * cse.c (cse_extended_basic_block): Don't attempt to record
13737 equivalences for degenerate conditional branches that branch
13738 to their fall-through.
13740 2017-08-24 Martin Sebor <msebor@redhat.com>
13742 PR middle-end/81908
13743 * gimple-fold.c (size_must_be_zero_p): New function.
13744 (gimple_fold_builtin_memory_op): Call it.
13746 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
13748 * config/rs6000/mm_malloc.h: New file.
13750 2017-08-24 Bin Cheng <bin.cheng@arm.com>
13752 PR tree-optimization/81913
13753 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
13754 analysis when either IVs in condition can wrap.
13756 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
13758 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
13759 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
13761 2017-08-24 Richard Biener <rguenther@suse.de>
13764 * targhooks.c (default_target_can_inline_p): Properly
13765 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
13766 is present and always compare.
13767 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
13768 infer -mfpmath=sse from TARGET_SSE_P.
13769 (ix86_can_inline_p): Properly use target_option_default_node when
13770 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
13772 2017-08-24 Richard Biener <rguenther@suse.de>
13775 * dwarf2out.c (output_die): Handle flag_generate_offload like
13777 (output_comp_unit): Likewise.
13778 (gen_array_type_die): Likewise.
13779 (dwarf2out_early_finish): Likewise.
13780 (note_variable_value_in_expr): Likewise.
13781 (dwarf2out_finish): Likewise. Adjust assert.
13782 * cgraphunit.c (symbol_table::compile): Move setting of
13783 flag_generate_offload earlier ...
13784 (symbol_table::finalize_compilation_unit): ... here, before
13785 early debug finalization.
13787 2017-08-24 Richard Biener <rguenther@suse.de>
13789 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
13790 and ipa-fnsummary.h.
13791 (ix86_can_inline_p): When ix86_fpmath flags do not match
13792 check whether the callee uses FP math at all.
13794 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
13796 PR middle-end/81931
13797 * tree-ssanames.c (get_nonzero_bits): Use element_precision
13798 instead of TYPE_PRECISION.
13800 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
13801 Alan Hayward <alan.hayward@arm.com>
13802 David Sherwood <david.sherwood@arm.com>
13804 * combine.c (make_extraction): Use subreg_offset_from_lsb.
13806 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
13808 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
13810 (ix86_frame::stack_realign_allocate): New field.
13811 (struct machine_frame_state): Modify comments.
13812 (machine_frame_state::sp_realigned_fp_end): New field.
13813 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
13814 layout calculation.
13815 (sp_valid_at): Add assertion to assure no attempt to access invalid
13816 offset of a realigned stack.
13817 (fp_valid_at): Likewise.
13818 (choose_baseaddr): Modify comments.
13819 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
13820 ix86_expand_prologue.
13821 (ix86_expand_prologue): Modify stack realignment and allocation.
13822 (ix86_expand_epilogue): Modify comments.
13823 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
13824 avx2_runtime, avx512f, and avx512f_runtime.
13826 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
13828 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
13829 (mstackrealign): Do not init to -1.
13830 * config/i386/i386.c (ix86_option_override_internal):
13831 Check opts_set, not opts when setting default value of
13832 opts->x_ix86_force_align_arg_pointer.
13834 2017-08-23 Richard Biener <rguenther@suse.de>
13836 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
13837 lang_hooks.decl_printable_name.
13838 * print-rtl-function.c (print_rtx_function): Likewise.
13839 * tree-pretty-print.c (dump_function_header): Likewise.
13841 2017-08-23 Richard Biener <rguenther@suse.de>
13844 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
13845 -g0 at compile-time.
13847 2017-08-23 Tamar Christina <tamar.christina@arm.com>
13849 PR middle-end/19706
13850 * doc/sourcebuild.texi (Other hardware attributes):
13853 2017-08-23 Tamar Christina <tamar.christina@arm.com>
13855 PR middle-end/19706
13856 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
13857 Fix single-use check.
13859 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
13861 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
13863 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
13865 * doc/install.texi: Modify to add more details on running selected
13868 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
13870 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
13871 is combined with -mabi=ms.
13872 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
13875 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13877 PR tree-optimization/81488
13878 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
13879 and cached_basis fields.
13880 (MAX_SPREAD): New constant.
13881 (alloc_cand_and_find_basis): Initialize new fields.
13882 (clear_visited): New function.
13883 (create_phi_basis_1): Rename from create_phi_basis, set visited
13884 and cached_basis fields.
13885 (create_phi_basis): New wrapper function.
13886 (phi_add_costs_1): Rename from phi_add_costs, add spread
13887 parameter, set visited field, short-circuit when limits reached.
13888 (phi_add_costs): New wrapper function.
13889 (record_phi_increments_1): Rename from record_phi_increments, set
13891 (record_phi_increments): New wrapper function.
13892 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
13893 (phi_incr_cost): New wrapper function.
13894 (all_phi_incrs_profitable_1): Rename from
13895 all_phi_incrs_profitable, set visited field.
13896 (all_phi_incrs_profitable): New wrapper function.
13898 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
13899 Alan Hayward <alan.hayward@arm.com>
13900 David Sherwood <david.sherwood@arm.com>
13902 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
13903 that takes the outer and inner modes.
13904 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
13905 comparison as the canonical test for a paradoxical subreg.
13906 * combine.c (simplify_set): Use paradoxical_subreg_p.
13907 (make_extraction): Likewise.
13908 (force_to_mode): Likewise.
13909 (rtx_equal_for_field_assignment_p): Likewise.
13910 (gen_lowpart_for_combine): Likewise.
13911 (simplify_comparison): Likewise.
13912 * cse.c (equiv_constant): Likewise.
13913 * expmed.c (store_bit_field_1): Likewise.
13914 * final.c (alter_subreg): Likewise.
13915 * fwprop.c (propagate_rtx): Likewise.
13916 (forward_propagate_subreg): Likewise.
13917 * ira-conflicts.c (ira_build_conflicts): Likewise.
13918 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
13919 * lra-constraints.c (curr_insn_transform): Likewise.
13920 (split_reg): Likewise.
13921 * lra-eliminations.c (move_plus_up): Likewise.
13922 (lra_eliminate_regs_1): Likewise.
13923 * recog.c (general_operand): Likewise.
13924 * ree.c (combine_reaching_defs): Likewise.
13925 * reload.c (push_reload): Likewise.
13926 (find_reloads): Likewise.
13927 * reload1.c (elimination_effects): Likewise.
13928 (compute_reload_subreg_offset): Likewise.
13929 (choose_reload_regs): Likewise.
13930 * rtlanal.c (subreg_lsb_1): Likewise.
13931 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13932 (simplify_subreg): Likewise.
13933 * var-tracking.c (track_loc_p): Likewise.
13934 * emit-rtl.c (byte_lowpart_offset): Likewise.
13935 (paradoxical_subreg_p): Delete out-of-line definition.
13937 2017-08-22 Jeff Law <law@redhat.com>
13939 PR tree-optimization/81741
13940 PR tree-optimization/71947
13941 * tree-ssa-dom.c: Include tree-inline.h.
13942 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
13943 equivalences if one is more expensive to compute than the other.
13944 * tree-ssa-scopedtables.h (class const_or_copies): Make
13945 record_const_or_copy_raw method private.
13946 (class avail_exprs_stack): New method simplify_binary_operation.
13947 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
13948 avail_exprs_stack::simplify_binary_operation as needed.
13949 (avail_exprs_stack::simplify_binary_operation): New function.
13951 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
13953 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
13954 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
13955 (DOT_SYMBOLS): Likewise.
13956 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
13957 (RELOCATABLE_NEEDS_FIXUP): Likewise.
13958 (RS6000_ABI_NAME): Likewise.
13959 (TARGET_CMODEL): Likewise.
13960 (TOC_SECTION_ASM_OP): Likewise.
13961 (SET_CMODEL): New macro.
13962 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
13964 2017-08-22 Richard Biener <rguenther@suse.de>
13966 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
13967 to allow for free-lang-data replacements similar to verify_type_variant.
13969 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
13970 Alan Hayward <alan.hayward@arm.com>
13971 David Sherwood <david.sherwood@arm.com>
13973 * config/aarch64/aarch64.md (casesi): Use DImode rather than
13974 VOIDmode for the LABEL_REF.
13976 2017-08-22 Richard Biener <rguenther@suse.de>
13978 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
13980 2017-08-22 Richard Biener <rguenther@suse.de>
13982 * common.opt (feliminate-dwarf2-dups): Ignore.
13983 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
13984 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
13985 same_die_p_wrap, compute_section_prefix,
13986 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
13987 (comdat_symbol_id, comdat_symbol_number): Likewise.
13988 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
13990 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
13991 (output_die): Mark unreachable path unreachable.
13992 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
13993 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
13994 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
13995 (dwarf2out_early_finish): Likewise.
13997 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
13999 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
14001 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
14004 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
14005 not VAR_P. Filter attribute warnings with OPT_Wattributes.
14006 (avr_attribute_table) <io, io_low, address>: Initialize
14007 .decl_required with true.
14009 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
14011 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
14012 undocumented debugging options.
14013 (-mvsx-scalar-double): Likewise.
14014 (-mallow-df-permute): Likewise.
14015 (-mvectorize-builtins): Likewise.
14016 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
14017 (rs6000_builtin_vectorized_function): Likewise.
14018 (rs6000_builtin_md_vectorized_function): Likewise.
14019 (rs6000_opt_vars): Likewise.
14021 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
14024 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
14025 (*btrq_imm): Rename from *btrq.
14026 (*btcq_imm): Rename from *btcq.
14027 (btsc): New code attribute.
14028 (*<btsc><mode>): New insn pattern.
14029 (*btr<mode>): Ditto.
14030 (*<btsc><mode>_mask): New insn_and_split pattern.
14031 (*btr<mode>_mask): Ditto.
14033 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14034 Alan Hayward <alan.hayward@arm.com>
14035 David Sherwood <david.sherwood@arm.com>
14037 * function.c (pad_below): Simplify padding calculation.
14039 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14040 Alan Hayward <alan.hayward@arm.com>
14041 David Sherwood <david.sherwood@arm.com>
14043 * target.def (function_prologue): Remove frame size argument.
14044 (function_epilogue): Likewise.
14045 * doc/tm.texi: Regenerate.
14046 * final.c (final_start_function): Update call to function_prologue.
14047 (final_end_function): Update call to function_epilogue.
14048 (default_function_pro_epilogue): Remove frame size argument.
14049 * output.h (default_function_pro_epilogue): Likewise.
14050 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
14051 (arm_output_function_prologue): Likewise.
14052 * config/frv/frv.c (frv_function_prologue): Likewise.
14053 (frv_function_epilogue): Likewise.
14054 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
14055 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
14056 (ia64_output_function_epilogue): Likewise.
14057 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
14058 (m32r_output_function_epilogue): Likewise.
14059 * config/microblaze/microblaze.c (microblaze_function_prologue)
14060 (microblaze_function_epilogue): Likewise.
14061 * config/mips/mips.c (mips_output_function_prologue): Likewise.
14062 (mips_output_function_epilogue): Likewise.
14063 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
14064 (mmix_target_asm_function_epilogue): Likewise.
14065 * config/msp430/msp430.c (msp430_start_function): Likewise.
14066 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
14067 (nds32_asm_function_epilogue): Likewise.
14068 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
14069 * config/pa/pa.c (pa_output_function_prologue): Likewise.
14070 (pa_output_function_epilogue): Likewise.
14071 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
14072 (rs6000_output_function_epilogue): Likewise.
14073 * config/rl78/rl78.c (rl78_start_function): Likewise.
14074 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
14075 (rs6000_output_function_epilogue): Likewise.
14076 * config/rx/rx.c (rx_output_function_prologue): Likewise.
14077 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
14078 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
14079 (sparc_asm_function_epilogue): Likewise.
14081 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14083 * tree.h (type_has_mode_precision_p): New function.
14084 * convert.c (convert_to_integer_1): Use it.
14085 * expr.c (expand_expr_real_2): Likewise.
14086 (expand_expr_real_1): Likewise.
14087 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
14088 * match.pd: Likewise.
14089 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
14090 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
14091 * tree-tailcall.c (process_assignment): Likewise.
14092 * tree-vect-loop.c (vectorizable_reduction): Likewise.
14093 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
14094 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
14095 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
14096 (vectorizable_assignment): Likewise.
14097 (vectorizable_shift): Likewise.
14098 (vectorizable_operation): Likewise.
14099 * tree-vrp.c (register_edge_assert_for_2): Likewise.
14101 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
14103 * match.pd: Add pow (C, x) simplification.
14105 2017-08-21 Richard Biener <rguenther@suse.de>
14107 PR tree-optimization/81900
14108 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
14109 for blocks with abnormal predecessors.
14110 (compute_antic): Do not set visited flag prematurely.
14112 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
14115 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
14117 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14119 * stor-layout.h (vector_type_mode): Move to...
14120 * tree.h (vector_type_mode): ...here.
14121 * stor-layout.c (vector_type_mode): Move to...
14122 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
14124 2017-08-21 Richard Biener <rguenther@suse.de>
14126 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
14127 register_external_die hooks.
14128 (debug_false_tree_charstarstar_uhwistar): Declare.
14129 (debug_nothing_tree_charstar_uhwi): Likewise.
14130 * debug.c (do_nothing_debug_hooks): Adjust.
14131 (debug_false_tree_charstarstar_uhwistar): New do nothing.
14132 (debug_nothing_tree_charstar_uhwi): Likewise.
14133 * dbxout.c (dbx_debug_hooks): Adjust.
14134 (xcoff_debug_hooks): Likewise.
14135 * sdbout.c (sdb_debug_hooks): Likewise.
14136 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
14137 * dwarf2out.c (macinfo_label_base): New global.
14138 (dwarf2out_register_external_die): New function for the
14139 register_external_die hook.
14140 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
14141 (dwarf2_debug_hooks): Use them.
14142 (dwarf2_lineno_debug_hooks): Adjust.
14143 (struct die_struct): Add with_offset flag.
14144 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
14145 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
14146 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
14147 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
14148 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
14149 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
14150 defining section names for the early LTO debug variants.
14151 (reset_indirect_string): New helper.
14152 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
14153 (print_dw_val): Add support for offsetted symbol references.
14154 (get_ultimate_context): Split out from is_cxx.
14155 (is_cxx): Use get_ultimate_context.
14156 (is_fortran): Add decl overload.
14157 (compute_comp_unit_symbol): Split out worker from
14158 compute_section_prefix.
14159 (compute_section_prefix): Call compute_comp_unit_symbol and
14160 set comdat_type_p here.
14161 (output_die): Skip DIE symbol output for the LTO added one.
14162 Handle DIE symbol references with offset.
14163 (output_comp_unit): Guard section name mangling properly.
14164 For LTO debug sections emit a symbol at the section beginning
14165 which we use to refer to its DIEs.
14166 (add_abstract_origin_attribute): For DIEs registered via
14167 dwarf2out_register_external_die directly refer to the early
14168 DIE rather than indirectly through the shadow one we created.
14169 Remove obsolete call to dwarf2out_abstract_function for
14170 non-function/block origins.
14171 (gen_array_type_die): When generating early LTO debug do
14172 not emit DW_AT_string_length.
14173 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
14174 late when in LTO. As suggested place a gcc_unreachable for
14175 the DECL_ABSTRACT_P case.
14176 (gen_subprogram_die): Avoid another specification DIE
14177 for early built declarations/definitions for the late LTO case.
14178 (gen_variable_die): Add type references for late duplicated VLA dies
14180 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
14181 we have the abstract instance already.
14182 (process_scope_var): Adjust decl DIE contexts in LTO which
14183 first puts them in limbo.
14184 (gen_decl_die): Do not generate type DIEs late apart from
14185 types for VLAs or for decls we do not yet have a DIE. Do not
14186 call dwarf2out_abstract_function late.
14187 (dwarf2out_early_global_decl): Make sure to create DIEs
14188 for abstract instances of a decl first.
14189 (dwarf2out_late_global_decl): Adjust comment.
14190 (output_macinfo_op): With multiple macro sections use
14191 macinfo_label_base to distinguish labels.
14192 (output_macinfo): Likewise. Update macinfo_label_base.
14193 Pass in the line info label.
14194 (note_variable_value_in_expr): When generating LTO resolve
14195 all variable values here by generating DIEs as needed.
14196 (init_sections_and_labels): Add early LTO debug flag parameter
14197 and generate different sections and names if set. Add generation
14198 counter for the labels so we can have multiple of them.
14199 (reset_dies): Helper to allow DIEs to be output multiple times.
14200 (dwarf2out_finish): When outputting DIEs to the fat part of an
14201 LTO object first reset DIEs.
14202 (dwarf2out_early_finish): Output early DIEs when generating LTO.
14203 (modified_type_die): Check for decl_ultimate_origin being self
14205 (gen_type_die_with_usage): Likewise.
14206 (gen_typedef_die): Allow decl_ultimate_origin being self.
14207 (set_decl_abstract_flags): Remove.
14208 (set_block_abstract_flags): Likewise.
14209 (dwarf2out_abstract_function): Treat the early generated DIEs
14210 as the abstract copy and only add DW_AT_inline and
14211 DW_AT_artificial here and call set_decl_origin_self.
14212 If the DIE has an abstract origin don't do anything.
14213 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
14214 if we have none yet (Go fails to build one, PR78628).
14215 (variably_modified_type_p): Prevent endless recursion for Ada
14216 cyclic pointer types.
14217 * lto-streamer-in.c: Include debug.h.
14218 (dref_queue): New global.
14219 (lto_read_tree_1): Stream in DIE references.
14220 (lto_input_tree): Register DIE references.
14221 (input_function): Stream DECL_DEBUG_ARGS.
14222 * lto-streamer-out.c: Include debug.h.
14223 (lto_write_tree_1): Output DIE references.
14224 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
14225 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
14226 (output_function): Stream DECL_DEBUG_ARGS.
14227 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
14228 Stream DECL_ABSTRACT_ORIGIN.
14229 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
14230 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
14231 DECL_CONTEXT for file-scope decls.
14232 * lto-streamer.h (struct dref_entry): Declare.
14233 (dref_queue): Likewise.
14234 * cfgexpand.c (pass_expand::execute): Do not call the
14235 outlining_inline_function hook here.
14236 * lto-wrapper.c (debug_obj): New global.
14237 (tool_cleanup): Unlink it if required.
14238 (debug_objcopy): New function.
14239 (run_gcc): Handle early debug sections in the IL files by
14240 extracting them to separate files, partially linkin them and
14241 feeding the result back as result to the linker.
14242 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
14243 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
14244 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
14245 sections into a separate segment.
14246 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
14248 (darwin_asm_dwarf_section): Likewise.
14249 (darwin_asm_output_dwarf_offset): Likewise.
14250 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
14252 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14253 Alan Hayward <alan.hayward@arm.com>
14254 David Sherwood <david.sherwood@arm.com>
14256 * read-md.h (md_reader::record_potential_iterator_use): Replace
14257 pointer argument with an rtx and an index.
14258 * read-rtl.c (iterator_group::apply_iterator): Likewise.
14259 (apply_mode_iterator): Likewise.
14260 (apply_code_iterator): Likewise.
14261 (apply_int_iterator): Likewise.
14262 (apply_subst_iterator): Likewise.
14263 (record_iterator_use): Likewise.
14264 (record_attribute_use): Likewise.
14265 (md_reader::record_potential_iterator_use): Likewise. Update calls
14266 to record_iterator_use and apply_iterator.
14267 (iterator_use): Replace ptr with x and index.
14268 (attribute_use): Likewise.
14269 (apply_attribute_uses): Update calls to apply_iterator.
14270 (apply_iterators): Likewise. Update initialization of iterator_use.
14271 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
14272 and record_potential_iterator_use.
14273 (rtx_reader::read_rtx_operand): Likewise.
14275 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14276 Alan Hayward <alan.hayward@arm.com>
14277 David Sherwood <david.sherwood@arm.com>
14279 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
14282 2017-08-21 Richard Biener <rguenther@suse.de>
14284 PR middle-end/81884
14285 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
14286 at struct end conservatively when comparing common bases.
14288 2017-08-21 Richard Biener <rguenther@suse.de>
14290 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
14291 (mem_ref_in_stmt): Remove.
14292 (determine_max_movement): Use ref index to get at the reference.
14293 (invariantness_dom_walker::before_dom_children): Deal with
14294 lim data already initialized.
14295 (gather_mem_refs_stmt): Initialize lim data and record ref index.
14297 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
14299 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
14300 (TARGET_ISA_ROUND): Ditto.
14301 (TARGET_ROUND): Ditto.
14302 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
14303 * config/i386/i386.md: Ditto.
14304 * config/i386/sse.md: Ditto.
14305 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
14306 with OPTION_MASK_ISA_SSE4_1.
14308 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
14311 * doc/extend.texi (x86 Built-in Functions): Correct the name of
14312 __builtin_ia32_lzcnt_u16.
14314 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
14317 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
14318 (rs6000_set_current_function): Rewrite function to use it.
14320 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
14323 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
14324 and TYPE_WARN_IF_NOT_ALIGN.
14325 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
14326 (handle_warn_if_not_align): New.
14327 (place_union_field): Call handle_warn_if_not_align.
14328 (place_field): Call handle_warn_if_not_align.
14329 Copy TYPE_WARN_IF_NOT_ALIGN.
14330 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
14331 (layout_type): Likewise.
14332 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
14334 (tree_decl_common): Add warn_if_not_align.
14335 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
14336 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
14337 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
14338 (DECL_WARN_IF_NOT_ALIGN): Likewise.
14339 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
14340 * doc/extend.texi: Document warn_if_not_aligned attribute.
14341 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
14343 2017-08-17 Martin Liska <mliska@suse.cz>
14346 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
14347 (get_data_dependence): Use it as pointer type.
14348 (distribute_loop): Likewise.
14350 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14352 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
14353 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
14354 (p8_vmrgow_<mode>_direct): New define_insn.
14355 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
14356 handle endianness for vmrgew and vmrgow permute patterns.
14358 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
14360 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
14361 * config/rs6000/rs6000-cpus.def: Remove comment.
14362 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
14363 (POWERPC_MASKS): Likewise.
14364 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
14365 use of TARGET_VSX_TIMODE.
14366 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
14367 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
14368 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
14369 (rs6000_option_override_internal): Remove dead code.
14370 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
14371 (rs6000_legitimize_reload_address): Likewise.
14372 (rs6000_legitimate_address_p): Likewise.
14373 (rs6000_opt_masks): Delete "vsx-timode".
14374 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
14375 from function comment.
14376 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
14377 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
14378 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
14380 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
14381 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
14382 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
14383 (VSX_TI): Likewise.
14385 (define_peephole2): Likewise.
14387 2017-08-17 Martin Sebor <msebor@redhat.com>
14390 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
14391 past the end of an array.
14392 (test_pp_format): Add test cases.
14394 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
14396 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
14397 missing ECF_NOTHROW flags.
14399 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
14402 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
14403 operands residing in integer registers.
14404 (*vsx_le_perm_load_<mode>): Likewise.
14405 (*vsx_le_perm_store_<mode>): Likewise.
14406 (define_peephole2): Add peepholes to optimize the above.
14408 2017-08-17 Marek Polacek <polacek@redhat.com>
14410 PR middle-end/81814
14411 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
14412 to mimic what shorten_compare did. Change the return type to bool.
14413 (fold_cond_expr_with_comparison): Update call to
14414 operand_equal_for_comparison_p.
14415 (fold_ternary_loc): Likewise.
14417 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
14419 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
14421 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
14423 2017-08-17 Richard Biener <rguenther@suse.de>
14425 * tree-ssa-structalias.c (solve_graph): When propagating
14426 to successors update the graphs succ edges and avoid duplicate work.
14428 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
14431 * config/i386/i386.c (ix86_option_override_internal): Save target
14432 specific options after ix86_stack_protector_guard_reg was changed.
14434 2017-08-17 Richard Biener <rguenther@suse.de>
14436 PR tree-optimization/81827
14437 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
14438 (new_var_info): Initialize it conservatively.
14439 (get_call_vi): Mark register vars.
14440 (new_scalar_tmp_constraint_exp): Likewise.
14441 (handle_rhs_call): Likewise.
14442 (handle_const_call): Likewise.
14443 (create_function_info_for): Likewise.
14444 (solve_constraints): Sort varinfos to separate register from
14445 non-register vars to pack points-to solution bitmaps during
14448 2017-08-17 Marek Polacek <polacek@redhat.com>
14450 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
14452 2017-08-17 Richard Biener <rguenther@suse.de>
14454 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
14455 to true when overflow is undefined and we saturated the result.
14457 2017-08-17 Alan Modra <amodra@gmail.com>
14460 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
14461 Don't use store multiple if only one reg needs saving.
14462 (interesting_frame_related_regno): New function.
14463 (rs6000_frame_related): Don't emit frame info for regs that
14465 (rs6000_emit_epilogue): Likewise.
14467 2017-08-16 Nathan Sidwell <nathan@acm.org>
14469 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
14470 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
14471 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
14472 * tree.c (free_lang_data_in_type): Use else-if chain. Always
14473 clear TYPE_LANG_1. Remove obsolete member-function stripping.
14474 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
14475 (verify_type): Adjust for TYPE_BINFO move.
14476 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
14477 process TYPE_BINFO directly.
14478 (hash_tree): Likewise.
14479 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
14481 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
14484 2017-08-16 David Malcolm <dmalcolm@redhat.com>
14486 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
14488 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
14491 * config/i386/i386.md (*anddi_1_btr): Change predicates of
14492 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
14493 Add ix86_binary_operator_ok to insn constraint.
14494 (*iordi_1_bts): Ditto.
14495 (*xordi_1_btc): Ditto.
14496 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
14497 Update corresponding peephole2 pattern.
14501 2017-08-16 Bin Cheng <bin.cheng@arm.com>
14503 PR tree-optimization/81832
14504 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
14505 copy loop header which has IFN_LOOP_DIST_ALIAS call.
14507 2017-08-16 Marek Polacek <polacek@redhat.com>
14510 * fold-const.c (fold_indirect_ref_1): Restore original behavior
14511 regarding size_zero_node.
14513 2017-08-16 Martin Liska <mliska@suse.cz>
14516 * config.gcc: Respect previously set extra_objs in case
14519 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
14521 PR tree-optimization/81835
14522 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
14523 the phi SSA_NAME. Check that the condition in a COND_EXPR does
14524 not depend on the phi.
14526 2017-08-16 Alan Modra <amodra@gmail.com>
14528 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
14531 2017-08-16 Alan Modra <amodra@gmail.com>
14533 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
14534 (save_reg_p): ..into this. Update all callers.
14535 (first_reg_to_save): Simplify.
14537 2017-08-16 Alan Modra <amodra@gmail.com>
14539 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
14542 2017-08-15 Joseph Myers <joseph@codesourcery.com>
14546 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
14547 constant count if that count is less than 32.
14549 2017-08-15 Nathan Sidwell <nathan@acm.org>
14551 * gcc.c (execute): Emit friendlier message if inferior is killed
14552 by an external cause.
14554 2017-08-15 Richard Biener <rguenther@suse.de>
14556 PR tree-optimization/81790
14557 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
14558 CONSTRUCTORs from simplifying and VN.
14560 2017-08-14 Martin Sebor <msebor@redhat.com>
14562 * builtin-attrs.def: Add comments.
14564 2017-08-14 Martin Sebor <msebor@redhat.com>
14567 * doc/extend.texi (attribute nonstring): Document new attribute.
14569 2017-08-14 Martin Sebor <msebor@redhat.com>
14572 * tree-diagnostic.c (default_tree_printer): Handle %G.
14573 * gimple-pretty-print.h (percent_G_format): Declare new function.
14574 * gimple-pretty-print.c (percent_G_format): Define.
14575 * tree-pretty-print.c (percent_K_format): Add argument.
14577 2017-08-14 Martin Sebor <msebor@redhat.com>
14579 PR translation/79998
14580 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
14581 Remove a stray space.
14583 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
14586 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
14587 (*iordi_1_bts): Ditto.
14588 (*xordi_1_btc): Ditto.
14590 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14593 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
14594 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14596 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
14597 quoted strings, and make more translator-friendly.
14598 (darwin_rs6000_override_options): Likewise.
14599 (rs6000_option_override_internal): Likewise.
14600 (rs6000_return_in_memory): Fix overlong line.
14601 (init_cmulative_args): Use quoted strings, and make more
14602 translator-friendly.
14603 (rs6000_pass_by_reference): Fix overlong line.
14604 (def_builtin): Use quoted strings.
14605 (altivec_expand_predicate_builtin): Use quoted strings, and make
14606 more translator-friendly.
14607 (htm_expand_builtin): Use quoted strings.
14608 (cpu_expand_builtin): Use quoted strings, and make more
14609 translator-friendly.
14610 (altivec_expand_builtin): Likewise.
14611 (paired_expand_predicate_builtin): Likewise.
14612 (rs6000_invalid_builtin): Likewise.
14613 (builtin_function_type): Use quoted strings.
14614 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
14615 more translator-friendly.
14616 (rs6000_trampoline_init): Likewise.
14617 (rs6000_handle_altivec_attribute): Likewise.
14618 (rs6000_inner_target_options): Use quoted strings.
14619 (rs6000_disable_incompatible_switches): Likewise.
14620 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
14621 strings, and make more translator-friendly.
14622 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
14624 2017-08-14 Bin Cheng <bin.cheng@arm.com>
14626 PR tree-optimization/81799
14627 * tree-loop-distribution.c (version_loop_by_alias_check): Force
14628 cond_expr to simple gimple operand.
14630 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
14632 PR middle-end/46932
14633 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
14635 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
14639 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
14640 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
14641 TARGET_GASISR_PROLOGUES.
14642 * config/avr/avr.c (avr_option_override): Same.
14643 (avr_pass_pre_proep::execute): Same.
14645 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
14648 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
14649 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
14651 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
14653 * config/i386/i386.md (*load_tp_<mode>): Redefine as
14654 define_insn_and_split. Split to a memory load from 0 in
14655 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
14656 using PTR mode iterator.
14657 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
14658 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
14659 (*add_tp_<mode>): Redefine as define_insn_and_split.
14660 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
14661 address space. Merge with *add_tp_x32 using PTR mode iterator.
14662 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
14663 Split to an add with a memory load from 0 in
14664 DEFAULT_TLS_SEG_REG address space.
14666 2017-08-12 Andrew Pinski <apinski@cavium.com>
14668 * config/aarch64/aarch64-option-extensions.def (rdma):
14669 Fix feature string to what Linux prints out in /proc/cpuinfo.
14671 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
14674 * dwarf2out.c (modified_type_die): For C typedef types that have
14675 an ultimate origin, process the ultimate origin instead of the
14677 (gen_typedef_die): Assert that input DECLs have no ultimate
14679 (gen_type_die_with_usage): For typedef variants that have an
14680 ultimate origin, just call gen_decl_die on the original DECL.
14681 (process_scope_var): Avoid creating DIEs for local typedefs and
14682 concrete static variables.
14684 2017-08-12 Alan Modra <amodra@gmail.com>
14688 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
14689 match gnu-user.h startfile.
14690 (ENDFILE_LINUX_SPEC): Similarly.
14692 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
14695 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
14697 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
14699 2017-08-11 Tamar Christina <tamar.christina@arm.com>
14700 * config/aarch64/aarch64.md (mov<mode>): Change.
14701 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
14702 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
14703 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
14705 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
14707 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
14708 for storage order barriers.
14710 2017-08-11 Martin Liska <mliska@suse.cz>
14713 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
14714 variables of void type.
14716 2017-08-11 Martin Liska <mliska@suse.cz>
14718 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
14719 TARGET_SUPPORTS_ALIASES.
14720 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
14721 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
14722 (optimize_weakref): Likewise.
14723 * symtab.c (symtab_node::noninterposable_alias): Likewise.
14724 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
14725 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
14727 2017-08-11 Martin Liska <mliska@suse.cz>
14730 * config/i386/i386.c (make_resolver_func): Do complete
14731 refactoring of the function.
14733 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
14736 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
14737 * config/i386/i386.c (ix86_stack_protect_guard): Use
14738 ix86_stack_protect_guard_symbol_str to generate varible declaration.
14739 * doc/invoke.texi (x86 Options): Document
14740 -mstack-protector-guard-symbol= option.
14742 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
14744 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
14745 * config/i386/i386.c (ix86_split_stack_guard): New function.
14746 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
14747 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
14748 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
14749 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
14750 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
14751 (split_stack_space_check): Call ix86_split_stack_guard.
14753 2017-08-10 Martin Sebor <msebor@redhat.com>
14755 * print-tree.c (print_node): Print location using the established
14757 Replace spaces with colons.
14758 (debug_raw, debug): Ditto.
14760 2017-08-10 Martin Sebor <msebor@redhat.com>
14763 * pretty-print.c (pp_format): Correct the handling of %s precision.
14765 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
14768 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
14770 (ix86_finalize_stack_frame_flags): This. Also clear
14771 frame_pointer_needed if -fno-omit-frame-pointer is used without
14773 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
14774 with ix86_finalize_stack_frame_flags.
14775 (ix86_expand_epilogue): Likewise.
14776 (ix86_expand_split_stack_prologue): Likewise.
14777 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
14779 2017-08-10 Martin Liska <mliska@suse.cz>
14782 * c-attribs.c (handle_target_attribute):
14783 Report warning for an empty string argument of target attribute.
14785 2017-08-09 Jakub Jelinek <jakub@redhat.com>
14788 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
14790 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
14791 or DECL_NONLOCAL labels.
14792 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
14793 or DECL_NONLOCAL labels here.
14795 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
14797 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
14798 to indicate when early gimple folding has been disabled.
14799 (rs6000_gimple_fold_builtin): Add debug content.
14800 (rs6000_invalid_builtin): Fix whitespace.
14801 (rs6000_expand_builtin): Fix whitespace.
14802 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
14804 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
14807 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
14808 SAVE_MULTIPLE if not all the registers that saves, should be saved.
14810 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
14812 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
14813 (qdf24xx): Likewise.
14814 * config/aarch64/aarch64.md: Include falkor.md.
14815 * config/aarch64/falkor.md: New.
14817 2017-08-09 Marek Polacek <polacek@redhat.com>
14820 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
14821 * diagnostic.c (emit_diagnostic): Add a comment.
14822 (emit_diagnostic_valist): New function.
14824 2017-08-09 Marek Polacek <polacek@redhat.com>
14827 * input.c (make_location): New overload.
14828 * input.h (make_location): Declare.
14830 2017-08-08 Alan Modra <amodra@gmail.com>
14831 H.J. Lu <hongjiu.lu@intel.com>
14834 * gcc.c (NO_PIE_SPEC): Delete.
14835 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
14837 (LINK_PIE_SPEC): ..to here.
14838 (LINK_COMMAND_SPEC): Support -no-pie.
14839 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
14840 chain of crtbegin*.o selection, update for PIE_SPEC changes and
14842 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
14843 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
14844 (ENDFILE_CRTEND_SPEC): Similarly.
14846 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
14849 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
14850 (mstack-protector-guard-offset=): Ditto.
14851 * config/i386/i386.c (ix86_option_override): Handle
14852 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
14854 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
14855 ix86_stack_protect_guard_offset variables.
14856 (TARGET_STACK_PROTECT_GUARD): Always define.
14857 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
14858 and -mstack-protector-guard-offset= options.
14860 2017-08-08 Bin Cheng <bin.cheng@arm.com>
14862 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
14863 boundary case for the last candidate.
14865 2017-08-08 Bin Cheng <bin.cheng@arm.com>
14867 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
14868 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
14870 2017-08-08 Tamar Christina <tamar.christina@arm.com>
14872 PR middle-end/19706
14873 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
14874 * config/aarch64/aarch64-builtins.c
14875 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
14876 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
14877 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
14879 2017-08-08 Tamar Christina <tamar.christina@arm.com>
14880 Andrew Pinski <pinskia@gmail.com>
14882 PR middle-end/19706
14883 * internal-fn.def (XORSIGN): New.
14884 * optabs.def (xorsign_optab): New.
14885 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
14886 (convert_expand_mult_copysign): New.
14887 (pass_optimize_widening_mul::execute): Call
14888 convert_expand_mult_copysign.
14890 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14892 PR tree-optimization/81354
14893 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
14894 Insert on edges rather than explicitly creating landing pads.
14895 (analyze_candidates_and_replace): Commit edge inserts.
14897 2017-08-08 Richard Biener <rguenther@suse.de>
14899 PR middle-end/81719
14900 * tree-ssa-loop-niter.c: Include tree-dfa.h.
14901 (expand_simple_operations): Also look through ADDR_EXPRs with
14902 MEM_REF bases treating them as POINTER_PLUS_EXPR.
14904 2017-08-08 Richard Biener <rguenther@suse.de>
14906 PR tree-optimization/81723
14907 * tree-vect-slp.c (struct bst_traits): New hash traits.
14908 (bst_fail): New global.
14909 (vect_build_slp_tree_2): New worker, split out from ...
14910 (vect_build_slp_tree): ... this now wrapping it with using
14911 bst_fail set to cache SLP tree build fails. Properly handle
14913 (vect_analyze_slp_instance): Allocate and free bst_fail.
14915 2017-08-08 Martin Liska <mliska@suse.cz>
14918 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
14919 LABEL_DECLs that can be from a different function.
14921 2017-08-08 Bin Cheng <bin.cheng@arm.com>
14923 PR tree-optimization/81744
14924 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
14925 loop's number of iterations.
14927 2017-08-08 Martin Liska <mliska@suse.cz>
14929 * asan.c: Include header files.
14930 * attribs.c (build_decl_attribute_variant): New function moved
14932 (build_type_attribute_qual_variant): Likewise.
14933 (cmp_attrib_identifiers): Likewise.
14934 (simple_cst_list_equal): Likewise.
14935 (omp_declare_simd_clauses_equal): Likewise.
14936 (attribute_value_equal): Likewise.
14937 (comp_type_attributes): Likewise.
14938 (build_type_attribute_variant): Likewise.
14939 (lookup_ident_attribute): Likewise.
14940 (remove_attribute): Likewise.
14941 (merge_attributes): Likewise.
14942 (merge_type_attributes): Likewise.
14943 (merge_decl_attributes): Likewise.
14944 (merge_dllimport_decl_attributes): Likewise.
14945 (handle_dll_attribute): Likewise.
14946 (attribute_list_equal): Likewise.
14947 (attribute_list_contained): Likewise.
14948 * attribs.h (lookup_attribute): New function moved from tree.[ch].
14949 (lookup_attribute_by_prefix): Likewise.
14950 * bb-reorder.c: Include header files.
14951 * builtins.c: Likewise.
14952 * calls.c: Likewise.
14953 * cfgexpand.c: Likewise.
14954 * cgraph.c: Likewise.
14955 * cgraphunit.c: Likewise.
14956 * convert.c: Likewise.
14957 * dwarf2out.c: Likewise.
14958 * final.c: Likewise.
14959 * fold-const.c: Likewise.
14960 * function.c: Likewise.
14961 * gimple-expr.c: Likewise.
14962 * gimple-fold.c: Likewise.
14963 * gimple-pretty-print.c: Likewise.
14964 * gimple.c: Likewise.
14965 * gimplify.c: Likewise.
14966 * hsa-common.c: Likewise.
14967 * hsa-gen.c: Likewise.
14968 * internal-fn.c: Likewise.
14969 * ipa-chkp.c: Likewise.
14970 * ipa-cp.c: Likewise.
14971 * ipa-devirt.c: Likewise.
14972 * ipa-fnsummary.c: Likewise.
14973 * ipa-inline.c: Likewise.
14974 * ipa-visibility.c: Likewise.
14976 * lto-cgraph.c: Likewise.
14977 * omp-expand.c: Likewise.
14978 * omp-general.c: Likewise.
14979 * omp-low.c: Likewise.
14980 * omp-offload.c: Likewise.
14981 * omp-simd-clone.c: Likewise.
14982 * opts-global.c: Likewise.
14983 * passes.c: Likewise.
14984 * predict.c: Likewise.
14985 * sancov.c: Likewise.
14986 * sanopt.c: Likewise.
14987 * symtab.c: Likewise.
14988 * toplev.c: Likewise.
14989 * trans-mem.c: Likewise.
14990 * tree-chkp.c: Likewise.
14991 * tree-eh.c: Likewise.
14992 * tree-into-ssa.c: Likewise.
14993 * tree-object-size.c: Likewise.
14994 * tree-parloops.c: Likewise.
14995 * tree-profile.c: Likewise.
14996 * tree-ssa-ccp.c: Likewise.
14997 * tree-ssa-live.c: Likewise.
14998 * tree-ssa-loop.c: Likewise.
14999 * tree-ssa-sccvn.c: Likewise.
15000 * tree-ssa-structalias.c: Likewise.
15001 * tree-ssa.c: Likewise.
15002 * tree-streamer-in.c: Likewise.
15003 * tree-vectorizer.c: Likewise.
15004 * tree-vrp.c: Likewise.
15005 * tsan.c: Likewise.
15006 * ubsan.c: Likewise.
15007 * varasm.c: Likewise.
15008 * varpool.c: Likewise.
15009 * tree.c: Remove functions moved to attribs.[ch].
15010 * tree.h: Likewise.
15011 * config/aarch64/aarch64.c: Add attrs.h header file.
15012 * config/alpha/alpha.c: Likewise.
15013 * config/arc/arc.c: Likewise.
15014 * config/arm/arm.c: Likewise.
15015 * config/avr/avr.c: Likewise.
15016 * config/bfin/bfin.c: Likewise.
15017 * config/c6x/c6x.c: Likewise.
15018 * config/cr16/cr16.c: Likewise.
15019 * config/cris/cris.c: Likewise.
15020 * config/darwin.c: Likewise.
15021 * config/epiphany/epiphany.c: Likewise.
15022 * config/fr30/fr30.c: Likewise.
15023 * config/frv/frv.c: Likewise.
15024 * config/ft32/ft32.c: Likewise.
15025 * config/h8300/h8300.c: Likewise.
15026 * config/i386/winnt.c: Likewise.
15027 * config/ia64/ia64.c: Likewise.
15028 * config/iq2000/iq2000.c: Likewise.
15029 * config/lm32/lm32.c: Likewise.
15030 * config/m32c/m32c.c: Likewise.
15031 * config/m32r/m32r.c: Likewise.
15032 * config/m68k/m68k.c: Likewise.
15033 * config/mcore/mcore.c: Likewise.
15034 * config/microblaze/microblaze.c: Likewise.
15035 * config/mips/mips.c: Likewise.
15036 * config/mmix/mmix.c: Likewise.
15037 * config/mn10300/mn10300.c: Likewise.
15038 * config/moxie/moxie.c: Likewise.
15039 * config/msp430/msp430.c: Likewise.
15040 * config/nds32/nds32-isr.c: Likewise.
15041 * config/nds32/nds32.c: Likewise.
15042 * config/nios2/nios2.c: Likewise.
15043 * config/nvptx/nvptx.c: Likewise.
15044 * config/pa/pa.c: Likewise.
15045 * config/pdp11/pdp11.c: Likewise.
15046 * config/powerpcspe/powerpcspe.c: Likewise.
15047 * config/riscv/riscv.c: Likewise.
15048 * config/rl78/rl78.c: Likewise.
15049 * config/rx/rx.c: Likewise.
15050 * config/s390/s390.c: Likewise.
15051 * config/sh/sh.c: Likewise.
15052 * config/sol2.c: Likewise.
15053 * config/sparc/sparc.c: Likewise.
15054 * config/spu/spu.c: Likewise.
15055 * config/stormy16/stormy16.c: Likewise.
15056 * config/tilegx/tilegx.c: Likewise.
15057 * config/tilepro/tilepro.c: Likewise.
15058 * config/v850/v850.c: Likewise.
15059 * config/vax/vax.c: Likewise.
15060 * config/visium/visium.c: Likewise.
15061 * config/xtensa/xtensa.c: Likewise.
15063 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
15066 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
15067 constraints since the -mupper-regs-* switches have been
15069 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
15070 into a vector from a double word element that was extracted from
15071 another vector, and eliminate extra XXPERMDI instructions.
15072 (vsx_concat_<mode>_2): Likewise.
15073 (vsx_concat_<mode>_3): Likewise.
15074 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
15075 concat to allow optimizing inserts from previous extracts.
15077 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
15079 * config/i386/i386.c (ix86_stack_protect_guard): Generate
15080 memory reference to a SSP offset in TLS address space.
15081 (ix86_print_operand) <case '@'>: Remove.
15082 (ix86_print_operand_punct_valid_p): Remove '@' code.
15083 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
15084 UNSPEC_SP_TLS_TEST.
15085 (stack_tls_protect_set_<mode>): Remove.
15086 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
15087 (stack_tls_protect_test_<mode>): Remove.
15088 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
15090 2017-08-07 Olivier Hainque <hainque@adacore.com>
15093 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
15095 2017-08-07 Douglas Rupp <rupp@adacore.com>
15097 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
15098 variable was referenced as multidir in command.
15100 2017-08-07 Jakub Jelinek <jakub@redhat.com>
15103 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
15106 2017-08-07 Martin Liska <mliska@suse.cz>
15108 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
15109 * config/rl78/rl78.c: Add include of attribs.h.
15110 * config/sh/sh.c: Likewise.
15111 * config/v850/v850.c: Likewise.
15113 2017-08-07 Tom de Vries <tom@codesourcery.com>
15115 PR middle-end/78266
15116 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
15118 2017-08-07 Martin Liska <mliska@suse.cz>
15120 * config/mips/mips.c: Include attribs.h.
15122 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
15125 * doc/invoke.texi: Document change in behvaior for -Ofast for
15128 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
15130 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
15132 (aarch64_pop_regs): Likewise.
15133 (aarch64_gen_load_pair): Likewise.
15134 (aarch64_save_callee_saves): Likewise.
15135 (aarch64_restore_callee_saves): Likewise.
15137 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
15139 * config/i386/i386.c: Revert the last change.
15141 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
15144 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
15146 (ix86_finalize_stack_frame_flags): This. Also clear
15147 frame_pointer_needed if -fno-omit-frame-pointer is used without
15149 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
15150 with ix86_finalize_stack_frame_flags.
15151 (ix86_expand_epilogue): Likewise.
15152 (ix86_expand_split_stack_prologue): Likewise.
15154 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
15157 * config/i386/i386.c (get_builtin_code_for_version): Set priority
15158 to P_AES for Westmere.
15160 2017-08-07 Jonathan Yong <10walls@gmail.com>
15162 * config/i386/mingw.opt (fset-stack-executable): Removed.
15163 * config/i386/cygming.opt (fset-stack-executable): Moved
15165 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
15167 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
15169 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
15171 2017-08-07 Marek Polacek <polacek@redhat.com>
15173 PR middle-end/81737
15174 * fold-const.c (fold_indirect_ref_1): Check type_domain.
15176 2017-08-07 Martin Liska <mliska@suse.cz>
15178 * attribs.h (canonicalize_attr_name): New function.
15179 (cmp_attribs): Move from c-format.c and adjusted.
15180 (is_attribute_p): Moved from tree.h.
15181 * tree-inline.c: Add new includes.
15182 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
15183 (private_is_attribute_p): Remove.
15184 (private_lookup_attribute): Likewise.
15185 (private_lookup_attribute_by_prefix): Simplify.
15186 (remove_attribute): Use is_attribute_p.
15187 * tree.h: Remove removed declarations.
15189 2017-08-07 Jakub Jelinek <jakub@redhat.com>
15191 PR middle-end/81698
15192 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
15193 instead of computing it in the function. Formatting fix.
15194 (expand_case): Don't rely on default_edge being the first edge,
15195 clear it if removing it, pass default_edge to
15196 emit_case_dispatch_table.
15197 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
15200 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
15202 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
15203 insn in the function, emit NOP after the insn.
15205 2017-08-06 Tom de Vries <tom@codesourcery.com>
15207 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
15210 2017-08-06 Tom de Vries <tom@codesourcery.com>
15212 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
15215 2017-08-04 Yury Gribov <tetra2005@gmail.com>
15217 PR tree-optimization/57371
15218 * match.pd: New pattern.
15220 2017-08-04 Marek Polacek <polacek@redhat.com>
15222 PR middle-end/81695
15223 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
15224 perform the computation in offset_int.
15226 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15228 PR tree-optimization/81136
15229 * tree-vectorizer.h: Include tree-hash-traits.h.
15230 (vec_base_alignments): New typedef.
15231 (vec_info): Add a base_alignments field.
15232 (vect_record_base_alignments): Declare.
15233 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
15235 (DR_IS_CONDITIONAL_IN_STMT): New macro.
15236 (create_data_ref): Add an is_conditional_in_stmt argument.
15237 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
15238 the is_conditional_in_stmt field.
15239 (data_ref_loc): Add an is_conditional_in_stmt field.
15240 (get_references_in_stmt): Set the is_conditional_in_stmt field.
15241 (find_data_references_in_stmt): Update call to create_data_ref.
15242 (graphite_find_data_references_in_stmt): Likewise.
15243 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
15244 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
15245 (vect_record_base_alignment): New function.
15246 (vect_record_base_alignments): Likewise.
15247 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
15248 for nested statements even if we fail to compute a misalignment.
15249 Use pooled base alignments for unconditional references.
15250 (vect_find_same_alignment_drs): Compare base addresses instead
15252 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
15253 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
15255 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15257 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
15258 Add an explicit name for the enum. Use auto_vec for slp_instances
15259 and grouped_stores.
15260 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
15262 (_bb_vec_info): Add a constructor and destructor.
15263 (vinfo_for_stmt): Return NULL for uids of -1 as well.
15264 (destroy_loop_vec_info): Delete.
15265 (vect_destroy_datarefs): Likewise.
15266 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
15267 (vec_info::vec_info): New function.
15268 (vec_info::~vec_info): Likewise.
15269 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
15270 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
15271 destroy_loop_vec_info.
15272 * tree-vect-loop.c (new_loop_vec_info): Replace with...
15273 (_loop_vec_info::_loop_vec_info): ...this.
15274 (destroy_loop_vec_info): Replace with...
15275 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
15276 the stmt_vec_infos. Leave handling of vec_info information to its
15277 destructor. Remove explicit vector releases.
15278 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
15279 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
15280 * tree-vect-slp.c (new_bb_vec_info): Replace with...
15281 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
15282 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
15283 (destroy_bb_vec_info): Replace with...
15284 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
15285 information to its destructor.
15286 (vect_slp_analyze_bb_1): Use new and delete instead of
15287 new_bb_vec_info and destroy_bb_vec_info.
15288 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
15291 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15293 * tree-data-ref.h (subscript): Add access_fn field.
15294 (data_dependence_relation): Add could_be_independent_p.
15295 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
15296 (same_access_functions): Move to tree-data-ref.c.
15297 * tree-data-ref.c (ref_contains_union_access_p): New function.
15298 (access_fn_component_p): Likewise.
15299 (access_fn_components_comparable_p): Likewise.
15300 (dr_analyze_indices): Add a reference to access_fn_component_p.
15301 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
15303 (constant_access_functions): Likewise.
15304 (add_other_self_distances): Likewise.
15305 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
15306 (initialize_data_dependence_relation): Use XCNEW and remove
15307 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
15308 of access functions that have the same type. Allow the
15309 subsequence to end with different bases in some circumstances.
15310 Record the chosen access functions in SUB_ACCESS_FN.
15311 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
15312 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
15313 (subscript_dependence_tester_1): Likewise dra and drb.
15314 (build_classic_dist_vector): Update calls accordingly.
15315 (subscript_dependence_tester): Likewise.
15316 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
15317 DDR_COULD_BE_INDEPENDENT_P.
15318 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
15319 comp_alias_ddrs instead of may_alias_ddrs.
15320 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
15322 (vect_analyze_data_ref_dependence): Use it if
15323 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
15324 distance vectors if that fails.
15325 (dependence_distance_ge_vf): New function.
15326 (vect_prune_runtime_alias_test_list): Use it. Don't clear
15327 LOOP_VINFO_MAY_ALIAS_DDRS.
15329 2017-08-04 Richard Biener <rguenther@suse.de>
15331 PR middle-end/81705
15332 * fold-const.c (fold_binary_loc): Properly restrict
15333 minus_var0 && minus_var1 case when associating undefined overflow
15336 2017-08-04 Tom de Vries <tom@codesourcery.com>
15338 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
15340 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15342 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15343 Don't start diagnostic messages with a capital letter.
15344 * config/rs6000/rs6000.c (rs6000_option_override_internal):
15346 (rs6000_invalid_builtin): Likewise.
15347 (rs6000_trampoline_init): Likewise.
15349 2017-08-03 Jakub Jelinek <jakub@redhat.com>
15352 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
15353 after setting changeable df flags.
15355 2017-08-03 Richard Biener <rguenther@suse.de>
15357 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
15358 up if the use is in USE - X.
15360 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
15362 * toplev.c (dumpfile.h): New include.
15363 (internal_error_reentered): New static function. Use it...
15364 (internal_error_function): ...here to handle reentered internal_error.
15366 2017-08-03 Richard Biener <rguenther@suse.de>
15368 PR middle-end/81148
15369 * fold-const.c (split_tree): Add minus_var and minus_con
15370 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
15371 here but always use minus_*.
15372 (associate_trees): Assert we never associate with MINUS_EXPR
15373 and NULL first operand. Do not recurse for PLUS_EXPR operands
15374 when associating as MINUS_EXPR either.
15375 (fold_binary_loc): Track minus_var and minus_con.
15377 2017-08-03 Tom de Vries <tom@codesourcery.com>
15380 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
15381 ACCEL_COMPILER, apply finish_options on
15382 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
15384 2017-08-03 Tom de Vries <tom@codesourcery.com>
15387 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
15388 function_entry_patch_area_size > 0.
15390 2017-08-03 Jakub Jelinek <jakub@redhat.com>
15393 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
15394 instead of 10??LU, perform unit multiplication in wide_int,
15395 don't change alloc_object_size_limit if the limit is larger
15398 PR tree-optimization/81655
15399 PR tree-optimization/81588
15400 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
15401 the case when ranges[i].low and high are 1 for unsigned type with
15404 PR middle-end/81052
15405 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
15406 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
15408 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15410 * tree-vrp.h: Add include guard.
15412 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
15415 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
15416 (ud2): New insn pattern.
15417 * config/i386/i386.c (ix86_expand_epilogue):
15418 For naked functions, generate ud2 instead of trap insn.
15420 2017-08-02 Marek Polacek <polacek@redhat.com>
15423 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
15425 2017-08-02 Tom de Vries <tom@codesourcery.com>
15426 Cesar Philippidis <cesar@codesourcery.com>
15428 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
15429 Add missing edge probabilities.
15431 2017-08-02 Tamar Christina <tamar.christina@arm.com>
15433 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
15434 Correct endianness.
15436 2017-08-02 Jakub Jelinek <jakub@redhat.com>
15438 PR middle-end/79499
15439 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
15440 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
15441 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
15442 prologue_seq sequences - if any.
15444 2017-08-02 Richard Biener <rguenther@suse.de>
15446 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
15447 via vectors if supported, integer extracts via punning if supported
15448 or otherwise vector extracts.
15450 2017-08-02 Richard Biener <rguenther@suse.de>
15452 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
15454 (bitmap_insert_into_set): ... this.
15456 2017-08-02 Richard Biener <rguenther@suse.de>
15458 PR tree-optimization/81633
15460 2015-08-17 Alan Hayward <alan.hayward@arm.com>
15462 PR tree-optimization/71752
15463 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
15465 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
15467 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
15468 (machine_function::call_ms2sysv_pad_out): Remove field.
15469 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
15470 (ix86_compute_frame_layout): Likewise.
15472 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
15475 * config/i386/i386.c (ix86_set_func_type): Disallow naked
15476 attribute with interrupt attribute.
15478 2017-08-01 Andrew Pinski <apinski@cavium.com>
15480 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
15481 BIT_INSERT_EXPR's operand 1
15482 to see if the types precision matches.
15484 2017-08-01 Martin Liska <mliska@suse.cz>
15486 PR middle-end/70140
15487 * builtins.c (expand_builtin_memcpy_args): Remove.
15488 (expand_builtin_memcpy): Call newly added function
15489 expand_builtin_memory_copy_args.
15490 (expand_builtin_memcpy_with_bounds): Likewise.
15491 (expand_builtin_mempcpy): Remove last argument.
15492 (expand_builtin_mempcpy_with_bounds): Likewise.
15493 (expand_builtin_memory_copy_args): New function created from
15494 expand_builtin_mempcpy_args with small modifications.
15495 (expand_builtin_mempcpy_args): Remove.
15496 (expand_builtin_stpcpy): Remove unused argument.
15497 (expand_builtin): Likewise.
15498 (expand_builtin_with_bounds): Likewise.
15500 2017-08-01 Martin Liska <mliska@suse.cz>
15503 Make mempcpy more optimal (PR middle-end/70140).
15505 2017-08-01 Jakub Jelinek <jakub@redhat.com>
15508 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
15509 __builtin_vec_cmpne verify both arguments are compatible vectors
15510 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
15511 verify arg1_type is a pointer or array type. For __builtin_vec_st,
15512 move computation of aligned to after checking the argument types.
15516 * config/rs6000/vsx.md (vextract_fp_from_shorth,
15517 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
15520 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
15521 Doug Rupp <rupp@adacore.com>
15522 Olivier Hainque <hainque@adacore.com>
15524 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
15525 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
15527 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
15528 for TARGET_OS_CPP_BUILTIN.
15529 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
15530 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
15532 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
15533 passing required abi options to the assembler for EABI configurations.
15534 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
15535 of .text.hot and .text.unlikely sections for kernel modules when
15536 using ARM style exceptions.
15537 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
15538 options. Add EXTRA_CC1_SPEC.
15539 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
15541 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
15543 (ARM_TARGET2_DWARF_FORMAT): Define.
15544 * config/arm/t-vxworks: Adjust multilib control to removal of the
15545 Diab command line options.
15547 2017-08-01 Martin Liska <mliska@suse.cz>
15549 PR gcov-profile/81561
15550 * gcov.c (unblock): Make unblocking safe as we need to preserve
15551 index correspondence of blocks and block_lists.
15553 2017-08-01 Richard Biener <rguenther@suse.de>
15555 PR tree-optimization/81181
15556 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
15557 (compute_antic): ... end of iteration here.
15559 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
15561 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
15562 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
15563 (ftree-slp-vectorize): Likewise.
15564 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
15565 can no longer be set independent of flag_tree_loop_vectorize.
15566 * omp-general.c (emp_max_vf): Likewise.
15567 * opts.c (enable_fdo_optimizations): Remove references to
15568 flag_tree_vectorize, these are now implicit.
15569 (common_handle_option): Remove handling for OPT_ftree_vectorize,
15570 and leave it for the options machinery.
15572 2017-08-01 Martin Liska <mliska@suse.cz>
15574 PR middle-end/70140
15575 * builtins.c (expand_builtin_memcpy_args): Remove.
15576 (expand_builtin_memcpy): Call newly added function
15577 expand_builtin_memory_copy_args.
15578 (expand_builtin_memcpy_with_bounds): Likewise.
15579 (expand_builtin_mempcpy): Remove last argument.
15580 (expand_builtin_mempcpy_with_bounds): Likewise.
15581 (expand_builtin_memory_copy_args): New function created from
15582 expand_builtin_mempcpy_args with small modifications.
15583 (expand_builtin_mempcpy_args): Remove.
15584 (expand_builtin_stpcpy): Remove unused argument.
15585 (expand_builtin): Likewise.
15586 (expand_builtin_with_bounds): Likewise.
15588 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
15591 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
15592 print "ds:" only for immediates in generic address space.
15594 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
15597 * config/i386/i386.c (ix86_funciton_naked): New prototype.
15598 (ix86_function_ok_for_sibcall): Return false for naked functions.
15600 2017-08-01 Richard Biener <rguenther@suse.de>
15602 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
15603 (compute_antic): Seed worklist with exit block predecessors.
15604 * cfganal.c (dfs_find_deadend): For a cycle return the source
15605 of the edge closing it.
15607 2017-08-01 Tamar Christina <tamar.christina@arm.com>
15609 * config/aarch64/aarch64.c
15610 (aarch64_can_const_movi_rtx_p): Move 0 check.
15612 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15614 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
15615 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
15617 * match.pd: Ditto in address comparison pattern.
15619 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15621 PR tree-optimization/81627
15622 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
15623 closed ssa form for store-store chain.
15625 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15627 PR tree-optimization/81620
15628 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
15629 for store-store chain.
15631 2017-08-01 Jakub Jelinek <jakub@redhat.com>
15633 PR tree-optimization/81588
15634 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
15635 ranges[i].in_p, invert comparison code ccode. For >/>=,
15636 swap rhs1 and rhs2 and comparison code unconditionally,
15637 for </<= don't do that. Don't swap rhs1/rhs2 again if
15638 ranges[i].in_p, instead invert comparison code ccode if
15639 opcode or oe->rank is BIT_IOR_EXPR.
15642 * optabs.def (vec_extract_optab, vec_init_optab): Change from
15643 a direct optab to conversion optab.
15644 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
15645 with GET_MODE_INNER as last argument instead of optab_handler.
15646 * expmed.c (extract_bit_field_1): Likewise. Use vector from
15647 vector extraction if possible and optab is available.
15648 * expr.c (store_constructor): Use convert_optab_handler instead
15649 of optab_handler. Use vector initialization from smaller
15650 vectors if possible and optab is available.
15651 * tree-vect-stmts.c (vectorizable_load): Likewise.
15652 * doc/md.texi (vec_extract, vec_init): Document that the optabs
15653 now have two modes.
15654 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
15655 of vec_init from half-sized vectors with the same element mode.
15656 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
15657 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
15658 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
15659 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
15660 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
15661 after mode in gen_vec_extract* calls.
15662 (vec_extract<mode>): Renamed to ...
15663 (vec_extract<mode><ssescalarmodelower>): ... this.
15664 (vec_extract<mode><ssehalfvecmodelower>): New expander.
15665 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
15666 element mode after mode in gen_vec_init* calls.
15667 (VEC_INIT_HALF_MODE): New mode iterator.
15668 (vec_init<mode>): Renamed to ...
15669 (vec_init<mode><ssescalarmodelower>): ... this.
15670 (vec_init<mode><ssehalfvecmodelower>): New expander.
15671 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
15672 (vec_extractv2sfsf): ... this.
15673 (vec_initv2sf): Renamed to ...
15674 (vec_initv2sfsf): ... this.
15675 (vec_extractv2si): Renamed to ...
15676 (vec_extractv2sisi): ... this.
15677 (vec_initv2si): Renamed to ...
15678 (vec_initv2sisi): ... this.
15679 (vec_extractv4hi): Renamed to ...
15680 (vec_extractv4hihi): ... this.
15681 (vec_initv4hi): Renamed to ...
15682 (vec_initv4hihi): ... this.
15683 (vec_extractv8qi): Renamed to ...
15684 (vec_extractv8qiqi): ... this.
15685 (vec_initv8qi): Renamed to ...
15686 (vec_initv8qiqi): ... this.
15687 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
15688 (vec_init<mode>): Renamed to ...
15689 (vec_init<mode><VEC_base_l>): ... this.
15690 (vec_extract<mode>): Renamed to ...
15691 (vec_extract<mode><VEC_base_l>): ... this.
15692 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
15693 (vec_initv2sfsf): ... this.
15694 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
15695 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
15696 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
15697 element mode after mode in gen_vec_init* calls.
15698 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
15699 (vec_init<mode><Vel>): ... this.
15700 (vec_extract<mode>): Renamed to ...
15701 (vec_extract<mode><Vel>): ... this.
15702 * config/aarch64/iterators.md (Vel): New mode attribute.
15703 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
15704 Add element mode after mode in gen_vec_extract* calls.
15705 * config/s390/vector.md (non_vec_l): New mode attribute.
15706 (vec_extract<mode>): Renamed to ...
15707 (vec_extract<mode><non_vec_l>): ... this.
15708 (vec_init<mode>): Renamed to ...
15709 (vec_init<mode><non_vec_l>): ... this.
15710 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
15711 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
15713 * config/arm/iterators.md (V_elem_l): New mode attribute.
15714 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
15715 (vec_extract<mode><V_elem_l>): ... this.
15716 (vec_extractv2di): Renamed to ...
15717 (vec_extractv2didi): ... this.
15718 (vec_init<mode>): Renamed to ...
15719 (vec_init<mode><V_elem_l>): ... this.
15720 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
15721 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
15722 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
15723 Add element mode after gen_vec_extract* calls.
15724 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
15725 (vec_init<mode><unitmode>): ... this.
15726 (vec_extract<mode>): Renamed to ...
15727 (vec_extract<mode><unitmode>): ... this.
15728 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
15729 (vec_init<mode><unitmode>): ... this.
15730 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
15731 (vec_initv2sfsf): ... this.
15732 (vec_extractv2sf): Renamed to ...
15733 (vec_extractv2sfsf): ... this.
15734 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
15735 Add element mode after gen_vec_extract* calls.
15736 * config/mips/mips.md (unitmode): New mode iterator.
15737 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
15738 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
15739 * config/spu/spu.md (inner_l): New mode attribute.
15740 (vec_init<mode>): Renamed to ...
15741 (vec_init<mode><inner_l>): ... this.
15742 (vec_extract<mode>): Renamed to ...
15743 (vec_extract<mode><inner_l>): ... this.
15744 * config/sparc/sparc.md (veltmode): New mode iterator.
15745 (vec_init<VMALL:mode>): Renamed to ...
15746 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
15747 * config/ia64/vect.md (vec_initv2si): Renamed to ...
15748 (vec_initv2sisi): ... this.
15749 (vec_initv2sf): Renamed to ...
15750 (vec_initv2sfsf): ... this.
15751 (vec_extractv2sf): Renamed to ...
15752 (vec_extractv2sfsf): ... this.
15753 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
15754 (vec_init<mode>): Renamed to ...
15755 (vec_init<mode><VEC_base_l>): ... this.
15756 (vec_extract<mode>): Renamed to ...
15757 (vec_extract<mode><VEC_base_l>): ... this.
15758 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
15759 (vec_initv2sfsf): ... this.
15760 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
15761 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
15762 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
15763 gen_vec_init* calls.
15765 2017-08-01 Richard Biener <rguenther@suse.de>
15767 PR tree-optimization/81297
15768 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
15769 TREE_OVERFLOW from INTEGER_CSTs.
15771 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
15773 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
15775 2017-07-31 Carl Love <cel@us.ibm.com>
15777 * config/rs6000/rs6000-c: Add support for built-in functions
15778 vector signed char vec_xl_be (signed long long, signed char *);
15779 vector unsigned char vec_xl_be (signed long long, unsigned char *);
15780 vector signed int vec_xl_be (signed long long, signed int *);
15781 vector unsigned int vec_xl_be (signed long long, unsigned int *);
15782 vector signed long long vec_xl_be (signed long long, signed long long *);
15783 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
15784 vector signed short vec_xl_be (signed long long, signed short *);
15785 vector unsigned short vec_xl_be (signed long long, unsigned short *);
15786 vector double vec_xl_be (signed long long, double *);
15787 vector float vec_xl_be (signed long long, float *);
15788 * config/rs6000/altivec.h (vec_xl_be): Add #define.
15789 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
15790 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
15792 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
15793 (altivec_expand_builtin): Add switch statement to call
15794 altivec_expand_xl_be for each builtin.
15795 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
15796 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
15797 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
15798 __builtin_vsx_le_be_v16qi.
15799 * doc/extend.texi: Update the built-in documentation file for the
15800 new built-in functions.
15802 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
15805 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
15807 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
15809 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15811 * config.gcc: Add z14.
15812 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
15813 CPU model numbers for z13s and z14.
15814 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
15816 * config/s390/s390-opts.h (enum processor_type): Rename
15817 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
15818 * config/s390/s390.c (processor_table): Add field for CPU name to
15819 be passed to Binutils.
15820 (s390_asm_output_machine_for_arch): Use the new field in
15821 processor_table for Binutils.
15822 (s390_expand_builtin): Replace arch12 with z14.
15823 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
15824 (s390_get_sched_attrmask): Likewise.
15825 (s390_get_unit_mask): Likewise.
15826 * config/s390/s390.opt: Add z14 to processor_type enum.
15828 2017-07-31 Martin Jambor <mjambor@suse.cz>
15831 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
15832 regardless of optimization level.
15834 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
15835 Martin Liska <mliska@suse.cz>
15837 * predict.def: Remove old comment and adjust probability.
15838 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
15839 PREDICT statements.
15841 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
15844 * config/i386/i386.c (ix86_function_naked): New function.
15845 (ix86_can_use_return_insn_p): Return false for naked functions.
15846 (ix86_expand_prologue): Skip prologue for naked functions.
15847 (ix86_expand_epilogue): Skip epilogue for naked functions
15848 and emit trap instruction.
15849 (ix86_warn_func_return): New function.
15850 (ix86_attribute_table): Add "naked" attribute specification.
15851 (TARGET_WARN_FUNC_RETURN): Define.
15852 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
15854 2017-07-31 Martin Liska <mliska@suse.cz>
15856 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
15857 (dump_gimple_bb_header): Always dump BB info.
15858 (pp_cfg_jump): Do not append info about BB when dumping a jump.
15860 2017-07-31 Martin Liska <mliska@suse.cz>
15863 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
15864 also with current_function_decl non-null equality.
15866 2017-07-31 Jakub Jelinek <jakub@redhat.com>
15869 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
15870 change type to the element type, instead add eltype variable and
15871 use it where we are interested in the element type.
15873 PR tree-optimization/81603
15874 * ipa-polymorphic-call.c
15875 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
15876 offset arithmetic in offset_int, bail out if the resulting bit offset
15877 doesn't fit into shwi.
15879 2017-07-31 Martin Liska <mliska@suse.cz>
15881 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
15882 (gimplify_save_expr): Fix comment.
15884 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
15887 * config/i386/i386.c (ix86_function_arg): Update arguments for
15889 (ix86_compute_frame_layout): Set the initial stack offset to
15890 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
15891 INCOMING_FRAME_SP_OFFSET.
15892 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
15893 stack before exception handler returns.
15894 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
15895 the 'ERROR_CODE' for exception handler.
15897 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
15899 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
15900 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
15901 (ASM_OUTPUT_REG_POP): Ditto.
15902 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
15903 instead of asm_fprintf to output pure string.
15905 2017-07-29 Jakub Jelinek <jakub@redhat.com>
15907 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
15908 to imported_module_or_decl hook.
15909 (debug_nothing_tree_tree_tree_bool): Remove.
15910 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
15911 * debug.c (do_nothing_debug_hooks): Use
15912 debug_nothing_tree_tree_tree_bool_bool instead of
15913 debug_nothing_tree_tree_tree_bool.
15914 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
15915 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
15916 * sdbout.c (sdb_debug_hooks): Likewise.
15917 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
15918 (gen_namespace_die): Add DW_AT_export_symbols attribute if
15920 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
15921 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
15922 attribute, don't add anything.
15924 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15926 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
15927 (fold_build2_stat_loc): Likewise.
15928 (fold_build3_stat_loc): Likewise.
15929 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
15930 (fold_build1_loc): Remove macro.
15931 (fold_build2_loc): Likewise.
15932 (fold_build3_loc): Likewise.
15934 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15936 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
15937 (gimple_build_debug_bind_source_stat): Likewise.
15938 * gimple.h (gimple_build_debug_bind): Remove macro.
15939 (gimple_build_debug_bind_source): Likewise.
15941 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15943 * bitmap.c (bitmap_alloc): Adjust.
15944 (bitmap_gc_alloc): Likewise.
15945 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
15947 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15949 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
15950 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
15951 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
15952 (bitmap_gc_alloc_stat): Likewise.
15953 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
15955 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15957 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
15958 * rtl.h (shallow_copy_rtx): Remove macro.
15960 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15962 * emit-rtl.c (gen_raw_REG): Adjust.
15963 * gengenrtl.c (gendef): Likewise.
15964 * rtl.c (rtx_alloc_stat): Remove _stat from name.
15965 * rtl.h (rtx_alloc): Remove macro.
15967 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15969 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
15970 (build_tree_list_stat): Likewise.
15971 * tree.h (build_tree_list): Remove macro.
15972 (build_tree_list_vec): Likewise.
15974 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15976 * tree.c (make_vector_stat): Remove _stat from name.
15977 (build_vector_stat): Likewise.
15978 * tree.h (make_vector_stat): Remove macro.
15979 (build_vector_stat): Likewise.
15981 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15983 * tree.h (build_var_debug_value): Remove prototype.
15985 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15987 * tree.c (tree_cons_stat): Remove _stat from name.
15988 * tree.h (tree_cons): Remove macro.
15990 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15992 * tree.c (build_vl_exp_stat): Remove _stat from name.
15993 * tree.h (build_vl_exp): Remove macro.
15995 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15997 * tree.c (build_decl_stat): Remove _stat from name.
15998 * tree.h (build_decl): Remove macro.
16000 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16002 * gimple.c (gimple_build_with_ops_stat): Adjust.
16003 (gimple_alloc_stat): Remove _stat from name.
16004 * gimple.h (gimple_alloc): Remove macro.
16006 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16008 * tree.c (make_tree_vec_stat): Remove _stat from name.
16009 (grow_tree_vec_stat): Likewise.
16010 * tree.h (make_tree_vec_stat): Adjust prototype.
16011 (grow_tree_vec_stat): Likewise.
16012 (make_tree_vec): Remove macro.
16013 (grow_tree_vec): Likewise.
16015 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16017 * fold-const.c (fold_build1_stat_loc): Adjust.
16018 (fold_build2_stat_loc): Likewise.
16019 (fold_build3_stat_loc): Likewise.
16020 * tree.c (build0_stat): Remove _stat from name.
16021 (build1_stat): Likewise.
16022 (build2_stat): Likewise.
16023 (build3_stat): Likewise.
16024 (build4_stat): Likewise.
16025 (build5_stat): Likewise.
16026 * tree.h (build1_loc): Remove macro, and rename _stat function
16028 (build2_loc): Likewise.
16029 (build3_loc): Likewise.
16030 (build4_loc): Likewise.
16031 (build5_loc): Likewise.
16033 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16035 * tree.c (make_int_cst_stat): Remove _stat from name.
16036 * tree.h (make_int_cst_stat): Adjust prototype.
16037 (make_int_cst): Remove macro.
16039 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16041 * tree.c (make_tre_binfo_stat): Remove _stat from name.
16042 * tree.h (make_tree_binfo_stat): Adjust prototype.
16043 (make_tree_binfo): Remove.
16045 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16047 * tree.c (copy_node_stat): Rename to copy_node.
16048 (build_distinct_type_copy): Adjust.
16049 * tree.h (copy_node_stat): Adjust prototype.
16050 (copy_node): Remove macro.
16052 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16054 * tree.c (make_node_stat): rename to make_node.
16055 (build_tree_list_stat): Adjust.
16056 (build0_stat): Likewise.
16057 (build2_stat): Likewise.
16058 (build3_stat): Likewise.
16059 (build4_stat): Likewise.
16060 (build5_stat): Likewise.
16061 (build_decl_stat): Likewise.
16062 * tree.h (make_node_stat): Adjust prototype.
16063 (make_node): remove macro.
16065 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
16067 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
16068 (PPC_FEATURE2_SCV): Likewise.
16069 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
16071 2017-07-28 Tamar Christina <tamar.christina@arm.com>
16073 * config/aarch64/aarch64.c
16074 (aarch64_internal_mov_immediate): Add new special pattern.
16075 * config/aarch64/aarch64.md (*movdi_aarch64):
16076 Add reg/32bit const mov case.
16078 2017-07-28 Tamar Christina <tamar.christina@arm.com>
16079 Richard Sandiford <richard.sandiford@linaro.org>
16081 * config/aarch64/aarch64.md (mov<mode>): Generalize.
16082 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
16083 Add integer and movi cases.
16084 (movi-split-hf-df-sf split, fp16): New.
16085 (enabled): Added TARGET_FP_F16INST.
16086 * config/aarch64/iterators.md (GPF_HF): New.
16087 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
16089 2017-07-28 Tamar Christina <tamar.christina@arm.com>
16091 * config/aarch64/aarch64.c
16092 (aarch64_simd_container_mode): Add prototype.
16093 (aarch64_expand_mov_immediate): Add HI support.
16094 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
16095 (aarch64_can_const_movi_rtx_p): New.
16096 (aarch64_preferred_reload_class):
16097 Remove restrictions of using FP registers for certain SIMD operations.
16098 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
16099 (aarch64_valid_floating_const): Add integer move validation.
16100 (aarch64_simd_imm_scalar_p): Remove.
16101 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
16102 (aarch64_legitimate_constant_p): Expand list of supported cases.
16103 * config/aarch64/aarch64-protos.h
16104 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
16105 (aarch64_reinterpret_float_as_int): New.
16106 (aarch64_simd_imm_scalar_p): Remove.
16107 * config/aarch64/constraints.md (Uvi): New.
16108 (Dd): Split into Ds and new Dd.
16109 * config/aarch64/aarch64.md (*movsi_aarch64):
16111 (*movdi_aarch64): Add SIMD mov case.
16113 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16115 * tree-predcom.c: (struct chain): Handle store-store chain in which
16116 stores for elimination only store loop invariant values.
16117 (execute_pred_commoning_chain): Ditto.
16118 (prepare_initializers_chain_store_elim): Ditto.
16119 (prepare_finalizers): Ditto.
16120 (is_inv_store_elimination_chain): New function.
16121 (initialize_root_vars_store_elim_1): New function.
16123 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16125 * tree-predcom.c: Revise general description of the pass.
16126 (enum chain_type): New enum type for store elimination.
16127 (struct chain): New field supporting store elimination.
16128 (struct component): Ditto.
16129 (dump_chain): Dump store-stores chain.
16130 (release_chain): Release resources.
16131 (split_data_refs_to_components): Compute and create component
16132 contains only stores for elimination.
16133 (get_chain_last_ref_at): New function.
16134 (make_invariant_chain): Initialization.
16135 (make_rooted_chain): Specify chain type in parameter and record it.
16136 (add_looparound_copies): Skip for store-stores chain.
16137 (determine_roots_comp): Compute type of chain and pass it to
16139 (initialize_root_vars_store_elim_2): New function.
16140 (finalize_eliminated_stores): New function.
16141 (remove_stmt): Handle store for elimination.
16142 (execute_pred_commoning_chain): Execute predictive commoning on
16143 store-store chains.
16144 (determine_unroll_factor): Skip unroll for store-stores chain.
16145 (prepare_initializers_chain_store_elim): New function.
16146 (prepare_initializers_chain): Hanlde store-store chain.
16147 (prepare_finalizers_chain, prepare_finalizers): New function.
16148 (tree_predictive_commoning_loop): Return integer value indicating
16149 if loop is unrolled or lcssa form is corrupted.
16150 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
16152 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16154 * tree-predcom.c (initialize_root): Delete.
16155 (execute_pred_commoning_chain): Initialize root vars and replace
16156 reference of non-combined chain directly, rather than call above
16159 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16161 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
16162 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
16164 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16166 * tree-predcom.c (struct chain): New field init_seq.
16167 (release_chain): Release init_seq.
16168 (prepare_initializers_chain): Record intialization stmts in above
16170 (insert_init_seqs): New function.
16171 (tree_predictive_commoning_loop): Call insert_init_seqs.
16173 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16175 * tree-predcom.c (determine_roots_comp): Skip trivial components.
16177 2017-07-28 Richard Biener <rguenther@suse.de>
16179 * match.pd: Remove superfluous :c.
16180 * genmatch.c (simplify::id): Add member.
16181 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
16183 (current_id): New global.
16184 (dt_node::parent): Move from ...
16185 (dt_operand::parent): ... here. Add for_id member.
16186 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
16187 (decision_tree::find_node): Relax order requirement when
16188 merging DT_TRUE nodes to ones inbetween the current simplify
16189 and the one we try to merge with. Add diagnostic whenever
16190 we need to enforce pattern order by not merging.
16191 (decision_tree::insert): Set current_id.
16192 (decision_tree::print_node): Dump parent node and for_id.
16193 (parser::last_id): Add member.
16194 (parser::push_simplify): Assign unique id.
16195 (parser::parser): Initialize last_id.
16197 2017-07-28 Martin Liska <mliska@suse.cz>
16200 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
16201 gimple_build_debug_bind.
16203 2017-07-28 Richard Biener <rguenther@suse.de>
16205 PR tree-optimization/81502
16206 * match.pd: Add pattern combining BIT_INSERT_EXPR with
16208 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
16210 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
16211 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
16212 for BIT_FIELD_REF args.
16213 * fold-const.c (make_bit_field_ref): Likewise.
16214 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
16216 2017-07-28 Jakub Jelinek <jakub@redhat.com>
16219 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
16220 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
16221 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
16222 Or it into SANITIZER_UNDEFINED.
16223 * ubsan.c: Include gimple-fold.h and varasm.h.
16224 (ubsan_expand_ptr_ifn): New function.
16225 (instrument_pointer_overflow): New function.
16226 (maybe_instrument_pointer_overflow): New function.
16227 (instrument_object_size): Formatting fix.
16228 (pass_ubsan::execute): Call instrument_pointer_overflow
16229 and maybe_instrument_pointer_overflow.
16230 * internal-fn.c (expand_UBSAN_PTR): New function.
16231 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
16232 * sanitizer.def (__ubsan_handle_pointer_overflow,
16233 __ubsan_handle_pointer_overflow_abort): New builtins.
16234 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
16235 * internal-fn.def (UBSAN_PTR): New internal function.
16236 * opts.c (sanitizer_opts): Add pointer-overflow.
16237 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
16238 * fold-const.c (build_range_check): Compute pointer range check in
16239 integral type if pointer arithmetics would be needed. Formatting
16242 2017-07-28 Martin Liska <mliska@suse.cz>
16245 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
16246 parameters that are of a variable-length.
16248 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
16250 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
16252 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
16253 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
16254 (CRT_CALL_STATIC_FUNCTION): Likewise.
16255 (ASM_DEFAULT_SPEC): New define.
16256 (ASM_SPEC32): Likewise.
16257 (ASM_SPEC64): Likewise.
16258 (ASM_SPEC_COMMON): Likewise.
16259 (ASM_SPEC): Likewise.
16260 (INVALID_64BIT): Likewise.
16261 (LINK_OS_DEFAULT_SPEC): Likewise.
16262 (LINK_OS_SPEC32): Likewise.
16263 (LINK_OS_SPEC64): Likewise.
16264 (POWERPC_LINUX): Likewise.
16265 (PTRDIFF_TYPE): Likewise.
16266 (RESTORE_FP_PREFIX): Likewise.
16267 (RESTORE_FP_SUFFIX): Likewise.
16268 (SAVE_FP_PREFIX): Likewise.
16269 (SAVE_FP_SUFFIX): Likewise.
16270 (SIZE_TYPE): Likewise.
16271 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
16272 (TARGET_64BIT): Likewise.
16273 (TARGET_64BIT): Likewise.
16274 (TARGET_AIX): Likewise.
16275 (WCHAR_TYPE_SIZE): Likewise.
16276 (WCHAR_TYPE): Undefine.
16277 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
16278 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
16279 (CPP_OS_RTEMS_SPEC): Delete.
16280 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
16281 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
16283 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
16285 2017-07-28 Jakub Jelinek <jakub@redhat.com>
16287 PR tree-optimization/81578
16288 * tree-parloops.c (build_new_reduction): Bail out if
16289 reduction_code isn't one of the standard OpenMP reductions.
16290 Move the details printing after that decision.
16292 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16294 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
16295 related to reload_in_progress.
16296 (splat_input_operand): Likewise.
16297 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
16299 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
16301 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
16302 (TARGET_INSTANTIATE_DECLS): Likewise.
16303 (legitimate_indexed_address_p): Delete reload_in_progress code.
16304 (rs6000_debug_legitimate_address_p): Likewise.
16305 (rs6000_eliminate_indexed_memrefs): Likewise.
16306 (rs6000_emit_le_vsx_store): Likewise.
16307 (rs6000_emit_move_si_sf_subreg): Likewise.
16308 (rs6000_emit_move): Likewise.
16309 (register_to_reg_type): Likewise.
16310 (rs6000_pre_atomic_barrier): Likewise.
16311 (rs6000_machopic_legitimize_pic_address): Likewise.
16312 (rs6000_allocate_stack_temp): Likewise.
16313 (rs6000_address_for_fpconvert): Likewise.
16314 (rs6000_address_for_altivec): Likewise.
16315 (rs6000_secondary_memory_needed_rtx): Delete function.
16316 (rs6000_check_sdmode): Likewise.
16317 (rs6000_alloc_sdmode_stack_slot): Likewise.
16318 (rs6000_instantiate_decls): Likewise.
16319 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
16320 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
16321 Delete reload_in_progress.
16322 (*vec_reload_and_plus_<mptrsize>): Likewise.
16323 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
16324 (vsx_div_v2di): Likewise.
16325 (vsx_udiv_v2di): Likewise.
16327 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16329 * config/rs6000/rs6000.opt (mlra): Replace with stub.
16330 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
16331 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
16332 (rs6000_debug_reg_global): Delete print of LRA status.
16333 (rs6000_option_override_internal): Delete dead LRA related code.
16334 (rs6000_lra_p): Delete function.
16335 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
16337 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
16339 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
16340 * config/riscv/rtems.h: New file.
16342 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16343 Sudakshina Das <sudi.das@arm.com>
16345 * config/aarch64/aarch64.md
16346 (define_split for and<mode>3nr_compare): Move
16347 non aarch64_logical_operand to a register.
16348 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
16349 register immediate operand to a register.
16350 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
16352 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16354 PR middle-end/81564
16355 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
16357 2017-07-27 Richard Biener <rguenther@suse.de>
16359 PR tree-optimization/81573
16360 PR tree-optimization/81494
16361 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
16362 multi defuse cycle case.
16364 2017-07-27 Richard Biener <rguenther@suse.de>
16366 PR tree-optimization/81571
16367 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
16370 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
16372 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
16373 earlier and only if MASK_FPU is set. Adjust formatting.
16375 2017-07-27 Martin Liska <mliska@suse.cz>
16377 * opt-functions.awk: Add validation of value of Init.
16378 * optc-gen.awk: Pass new argument.
16380 2017-07-27 Martin Liska <mliska@suse.cz>
16382 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
16383 Fix wrong condition.
16385 2017-07-27 Martin Liska <mliska@suse.cz>
16387 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
16388 BBs and edges seen by autoFDO.
16390 2017-07-27 Richard Biener <rguenther@suse.de>
16392 PR tree-optimization/81502
16393 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
16394 with incompatible but same sized type.
16395 (execute_update_addresses_taken): Likewise.
16397 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
16399 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
16400 flag_tree_loop_vectorize rather than flag_tree_vectorize.
16402 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16405 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
16406 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
16407 Change s_operand to memory_operand.
16409 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
16411 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
16412 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
16413 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
16414 Emit instructions rather than returning an expression. Handle TFmode
16415 and KFmode by casting to TImode.
16416 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
16417 (rs6000_emit_le_vsx_store): Likewise.
16418 * config/rs6000/vsx.md (VSX_TI): New iterator.
16419 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
16420 (*vsx_le_undo_permute_<mode>): Likewise.
16421 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
16422 emit the split sequence.
16423 (*vsx_le_perm_store_<mode>): Likewise.
16425 2017-07-27 Jakub Jelinek <jakub@redhat.com>
16427 PR tree-optimization/81555
16428 PR tree-optimization/81556
16429 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
16430 if true, force CHANGED for the recursive invocation.
16431 (reassociate_bb): Remember original length of ops array, pass
16432 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
16434 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
16435 attributes for noipa attribute. For naked attribute use
16436 lookup_attribute first before lookup_attribute_spec.
16437 * final.c (rest_of_handle_final): Disable IPA RA for functions with
16439 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
16440 for functions with noipa attribute.
16441 (cgraph_externally_visible_p): Return true for functions with noipa
16443 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
16444 for functions with noipa attribute.
16445 * doc/extend.texi: Document noipa function attribute.
16446 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
16447 also for functions with noipa attribute.
16448 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
16450 2017-07-26 Andrew Pinski <apinski@cavium.com>
16452 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
16453 vec_unalign_load_cost and vec_unalign_store_cost.
16455 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
16457 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
16458 -mvsx-small-integer option.
16459 (ISA_3_0_MASKS_IEEE): Likewise.
16460 (OTHER_VSX_VECTOR_MASKS): Likewise.
16461 (POWERPC_MASKS): Likewise.
16462 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
16463 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
16464 code, only testing for DImode being allowed in non-VSX floating
16466 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
16467 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
16469 (rs6000_option_override_internal): Delete -mvsx-small-integer.
16470 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
16471 TARGET_P8_VECTOR test.
16472 (rs6000_secondary_reload_simple_move): Likewise.
16473 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
16474 since TARGET_P9_VECTOR was already tested.
16475 (rs6000_opt_masks): Remove -mvsx-small-integer.
16476 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
16477 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16479 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
16480 test for TARGET_VEXTRACTUB was used, and that uses
16482 (p9 extract splitter): Likewise.
16483 (vsx_extract_<mode>_di_p9): Likewise.
16484 (vsx_extract_<mode>_store_p9): Likewise.
16485 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
16486 for TARGET_P9_VECTOR was used. Delete code that is now dead with
16487 the elimination of TARGET_VSX_SMALL_INTEGER.
16488 (vsx_extract_<mode>_p8): Likewise.
16489 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
16490 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
16491 (vsx_set_<mode>_p9): Likewise.
16492 (vsx_set_v4sf_p9): Likewise.
16493 (vsx_set_v4sf_p9_zero): Likewise.
16494 (vsx_insert_extract_v4sf_p9): Likewise.
16495 (vsx_insert_extract_v4sf_p9_2): Likewise.
16496 * config/rs6000/rs6000.md (sign extend splitter): Change
16497 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
16498 (floatsi<mode>2_lfiwax_mem): Likewise.
16499 (floatunssi<mode>2_lfiwzx_mem): Likewise.
16500 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
16501 since a test for TARGET_P9_VECTOR was used.
16502 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16503 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
16504 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16505 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
16506 TARGET_P8_VECTOR test.
16507 (fix_trunc<mode>si2_stfiwx): Likewise.
16508 (fix_trunc<mode>si2_internal): Likewise.
16509 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
16510 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16512 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
16513 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
16514 TARGET_P8_VECTOR test.
16515 (fixuns_trunc<mode>si2_stfiwx): Likewise.
16516 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
16517 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16519 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
16520 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
16521 since a test for TARGET_P9_VECTOR was used.
16522 (splitter for loading small constants): Likewise.
16524 2017-07-26 Andrew Pinski <apinski@cavium.com>
16526 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
16529 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
16532 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
16533 (fp_valid_at): Likewise.
16535 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
16537 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
16538 (qdf24xx_addrcost_table): Likewise.
16539 (cortexa57_tunings): Update to use generic_branch_cost.
16540 (cortexa72_tunings): Likewise.
16541 (cortexa73_tunings): Likewise.
16542 (qdf24xx_tunings): Likewise.
16544 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
16546 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
16547 (thunderx2t99_branch_cost): Likewise.
16548 (cortexa35_tunings): Update to use generic_branch_cost.
16549 (cortexa53_tunings): Likewise.
16550 (cortexa57_tunings): Likewise.
16551 (cortexa72_tunings): Likewise.
16552 (cortexa73_tunings): Likewise.
16553 (thunderx2t99_tunings): Likewise.
16555 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
16557 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
16558 (sparc_option_override): Honour MASK_FSMULD.
16559 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
16560 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
16561 * config/sparc/sparc.opt (mfsmuld): New option.
16562 * doc/invoke.texi (mfsmuld): Document option.
16564 2017-07-26 Marek Polacek <polacek@redhat.com>
16566 PR middle-end/70992
16567 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
16569 2017-07-26 Richard Biener <rguenther@suse.de>
16571 * gimple-match-head.c (do_valueize): Return OP if valueize
16573 (get_def): New helper to get at the def stmt of a SSA name
16574 if valueize allows.
16575 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
16576 do_valueize to get at the def stmt.
16577 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
16579 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
16581 PR middle-end/46932
16582 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
16584 2017-07-26 Martin Liska <mliska@suse.cz>
16587 * function.c (expand_function_start): Make expansion of
16588 nonlocal_goto_save_area after parm_birth_insn.
16590 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
16592 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
16593 from all CPU target flags enable members.
16595 2017-07-26 Richard Biener <rguenther@suse.de>
16597 * genmatch.c (dt_simplify::gen): Make iterator vars const.
16598 (decision_tree::gen): Make 'type' const.
16599 (write_predicate): Likewise.
16601 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
16603 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
16604 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
16605 (rs6000_option_override_internal): Likewise.
16606 (rs6000_expand_vector_set): Likewise.
16607 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
16608 (TARGET_UPPER_REGS_SF): Likewise.
16609 (TARGET_UPPER_REGS_DI): Likewise.
16610 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
16611 (TARGET_DIRECT_MOVE_64BIT): Likewise.
16612 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
16613 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16614 (Splitters for DI constants in Altivec registers): Likewise.
16615 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
16616 (vsx_set_v4sf_p9): Likewise.
16617 (vsx_set_v4sf_p9_zero): Likewise.
16618 (vsx_insert_extract_v4sf_p9): Likewise.
16619 (vsx_insert_extract_v4sf_p9_2): Likewise.
16621 2017-07-25 Carl Love <cel@us.ibm.com>
16623 * doc/extend.texi: Update the built-in documentation file for the
16624 existing built-in functions
16625 vector signed char vec_cnttz (vector signed char);
16626 vector unsigned char vec_cnttz (vector unsigned char);
16627 vector signed short vec_cnttz (vector signed short);
16628 vector unsigned short vec_cnttz (vector unsigned short);
16629 vector signed int vec_cnttz (vector signed int);
16630 vector unsigned int vec_cnttz (vector unsigned int);
16631 vector signed long long vec_cnttz (vector signed long long);
16632 vector unsigned long long vec_cnttz (vector unsigned long long);
16634 2017-07-25 Andrew Pinski <apinski@cavium.com>
16636 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
16637 accesses where the use is for the first operand of a BIT_INSERT.
16639 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
16642 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
16643 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
16645 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
16647 * config/i386/gstabs.h: Delete.
16648 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
16650 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
16652 * config/i386/i386.c (ix86_decompose_address): Do not check for
16653 register RTX when looking at index_reg or base_reg.
16654 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
16656 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
16658 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
16659 to update EH info here.
16661 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16663 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
16665 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16667 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
16669 2017-07-25 Torsten Duwe <duwe@suse.de>
16671 * common.opt: Introduce -fpatchable-function-entry
16672 command line option, and its variables function_entry_patch_area_size
16673 and function_entry_patch_area_start.
16674 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
16675 including a two-value parser.
16676 * target.def (print_patchable_function_entry): New target hook.
16677 * targhooks.h (default_print_patchable_function_entry): New function.
16678 * targhooks.c (default_print_patchable_function_entry): Likewise.
16679 * toplev.c (process_options): Switch off IPA-RA if
16680 patchable function entries are being generated.
16681 * varasm.c (assemble_start_function): Look at the
16682 patchable-function-entry command line switch and current
16683 function attributes and maybe generate NOP instructions by
16684 calling the print_patchable_function_entry hook.
16685 * doc/extend.texi: Document patchable_function_entry attribute.
16686 * doc/invoke.texi: Document -fpatchable_function_entry
16687 command line option.
16688 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
16690 * doc/tm.texi: Re-generate.
16692 2017-07-25 Jakub Jelinek <jakub@redhat.com>
16695 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
16696 TARGET_AVX512DQ rather than TARGET_AVX512BW.
16698 2017-07-25 Tamar Christina <tamar.christina@arm.com>
16700 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
16702 2017-07-25 Richard Biener <rguenther@suse.de>
16704 PR tree-optimization/81455
16705 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
16706 not walk in cycles when looking for guards.
16708 2017-07-25 Richard Biener <rguenther@suse.de>
16710 PR tree-optimization/81529
16711 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
16712 when optimizing backedge uses.
16714 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
16716 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
16718 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
16719 to dl_section_ref. On AIX, append an expression to subtract
16720 the size of the section length to dl_section_ref.
16722 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
16724 * configure.ac: If any of the config.* scripts fail, exit 1.
16725 * configure: Regenerate.
16727 2017-07-25 Richard Biener <rguenther@suse.de>
16729 PR middle-end/81546
16730 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
16731 of immediate uses, be more verbose on errors.
16733 2017-07-25 Richard Biener <rguenther@suse.de>
16735 PR tree-optimization/81510
16736 * tree-vect-loop.c (vect_is_simple_reduction): When the
16737 reduction stmt is not inside the loop bail out.
16739 2017-07-25 Richard Biener <rguenther@suse.de>
16741 PR tree-optimization/81303
16742 * tree-vect-loop-manip.c (vect_loop_versioning): Build
16743 profitability check against LOOP_VINFO_NITERSM1.
16745 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16747 * domwalk.c (cmp_bb_postorder): Simplify.
16748 (sort_bbs_postorder): New function. Use it...
16749 (dom_walker::walk): ...here to optimize common cases.
16751 2017-07-25 Martin Liska <mliska@suse.cz>
16754 * ipa-visibility.c (function_and_variable_visibility): Make the
16755 redirection just on target that supports aliasing.
16756 Fix GNU coding style.
16758 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
16761 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
16763 * config/arm/rtems.h: Likewise.
16764 * config/bfin/rtems.h: Likewise.
16765 * config/i386/rtemself.h: Likewise.
16766 * config/lm32/rtems.h: Likewise.
16767 * config/m32c/rtems.h: Likewise.
16768 * config/m68k/rtemself.h: Likewise.
16769 * config/microblaze/rtems.h: Likewise.
16770 * config/mips/rtems.h: Likewise.
16771 * config/moxie/rtems.h: Likewise.
16772 * config/nios2/rtems.h: Likewise.
16773 * config/powerpcspe/rtems.h: Likewise.
16774 * config/rs6000/rtems.h: Likewise.
16775 * config/rtems.h: Likewise.
16776 * config/sh/rtems.h: Likewise.
16777 * config/sh/rtemself.h: Likewise.
16778 * config/sparc/rtemself.h: Likewise.
16780 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
16783 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
16784 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
16785 * tree-ssa-structalias.c (alias_get_name): Same.
16787 2017-07-25 Bin Cheng <bin.cheng@arm.com>
16790 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
16791 instructions if no du chain is found.
16793 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
16795 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
16797 2017-07-25 Richard Biener <rguenther@suse.de>
16799 PR middle-end/81505
16800 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
16803 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
16805 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
16806 upper-regs options.
16807 (ISA_2_7_MASKS_SERVER): Likewise.
16808 (ISA_3_0_MASKS_IEEE): Likewise.
16809 (OTHER_P8_VECTOR_MASKS): Likewise.
16810 (OTHER_VSX_VECTOR_MASKS): Likewise.
16811 (POWERPC_MASKS): Likewise.
16812 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
16813 duplicate list of options.
16814 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
16815 explicit -mupper-regs options.
16816 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
16817 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
16818 alias for -mupper-regs-df.
16819 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
16820 (rs6000_init_hard_regno_mode_ok): Likewise.
16821 (rs6000_option_override_internal): Likewise.
16822 (rs6000_opt_masks): Likewise.
16823 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
16824 options in terms of whether -mvsx or -mpower8-vector was used.
16825 (TARGET_UPPER_REGS_DI): Likewise.
16826 (TARGET_UPPER_REGS_SF): Likewise.
16827 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
16828 -mupper-regs-* options.
16830 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
16832 * passes.c (emergency_dump_function): Print some empty lines and a
16833 header before the RTL dump.
16835 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
16837 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
16839 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
16842 * config/aarch64/aarch64.c (aarch64_classify_symbol):
16843 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
16845 2017-07-24 Carl Love <cel@us.ibm.com>
16847 * config/rs6000/rs6000-c.c: Add support for built-in functions
16848 vector float vec_extract_fp32_from_shorth (vector unsigned short);
16849 vector float vec_extract_fp32_from_shortl (vector unsigned short);
16850 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
16851 vec_extract_fp_from_shortl): Add defines for the two builtins.
16852 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
16853 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
16855 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
16856 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
16857 * doc/extend.texi: Update the built-in documentation file for the
16858 new built-in function.
16860 2017-07-24 Jakub Jelinek <jakub@redhat.com>
16863 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
16865 * doc/generic.texi: Likewise.
16866 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
16867 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
16869 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
16871 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
16872 (aarch64_mls_elt_merge<mode>): Likewise.
16874 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
16876 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
16877 having __cxa_atexit.
16879 2017-07-23 Michael Collison <michael.collison@arm.com>
16881 * config/arm/arm.c (arm_option_override): Deprecate
16882 use of -mstructure-size-boundary.
16883 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
16884 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
16886 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16889 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
16890 Reduce cost estimate for direct moves.
16892 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
16895 * config/i386/i386.c (ix86_option_override_internal): Disable
16896 BMI, BMI2 and TBM instructions for -m16.
16898 2017-07-21 Carl Love <cel@us.ibm.com>
16900 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16901 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16902 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16903 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16904 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16905 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
16906 VMULOSW): New enum "unspec" values.
16907 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
16908 altivec_vmulosw): New patterns.
16909 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16910 VMULOSW): Add definitions.
16912 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
16914 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
16915 (qdf24xx): Likewise.
16916 * config/aarch64/aarch64-options-extensions.def (rdma); New.
16917 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
16918 (AARCH64_FL_V8_1): Renumber.
16919 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
16920 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
16921 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
16922 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
16923 rdma to feature modifiers list.
16925 2017-07-21 Yury Gribov <tetra2005@gmail.com>
16927 PR middle-end/56727
16928 * ipa-visibility (function_and_variable_visibility): Convert
16929 recursive PLT call to direct call if appropriate.
16931 2017-07-21 Andrew Pinski <apinski@cavium.com>
16933 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
16934 operand 1 to see if the types precision matches.
16935 * fold-const.c (operand_equal_p): Likewise.
16937 2017-07-21 Richard Biener <rguenther@suse.de>
16939 PR tree-optimization/81303
16940 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
16941 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
16942 (vect_peeling_hash_get_lowest_cost): Adjust.
16943 (vect_enhance_data_refs_alignment): Likewise. Use
16944 vect_get_peeling_costs_all_drs to compute the penalty for no
16945 peeling to match up costs.
16947 2017-07-21 Richard Biener <rguenther@suse.de>
16949 PR tree-optimization/81500
16950 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
16951 we didn't identify a reduction path.
16953 2017-07-21 Tom de Vries <tom@codesourcery.com>
16954 Cesar Philippidis <cesar@codesourcery.com>
16956 PR gcov-profile/81442
16957 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
16960 2017-07-21 Tom de Vries <tom@codesourcery.com>
16963 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
16965 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
16966 nvptx_override_options_after_change.
16968 2017-07-21 Ulrich Drepper <drepper@redhat.com>
16970 * dwarf2out.c (output_file_names): Avoid double testing for
16971 dwarf_version >= 5.
16973 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
16975 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
16977 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
16979 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
16981 (try_crossjump_to_edge): Do not punt on partitioned functions.
16983 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
16985 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
16986 Put all BBs reachable only via paths crossing cold region to cold
16988 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
16990 2016-07-21 Richard Biener <rguenther@suse.de>
16992 PR tree-optimization/81303
16993 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
16994 into account prologue and epilogue iterations when raising
16995 min_profitable_iters to sth at least covering one vector iteration.
16997 2017-07-21 Tamar Christina <tamar.christina@arm.com>
16999 * config/arm/arm.c (arm_test_cpu_arch_dat):
17002 2017-07-20 Nathan Sidwell <nathan@acm.org>
17004 Remove TYPE_METHODS.
17005 * tree.h (TYPE_METHODS): Delete.
17006 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
17007 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
17008 (dbxout_type_methods): Scan TYPE_FIELDS.
17009 (dbxout_type): Don't check TYPE_METHODS here.
17010 * function.c (use_register_for_decl): Always ignore register for
17011 class types when not optimizing.
17012 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
17013 * tree.c (free_lang_data_in_type): Stitch out member functions and
17014 templates from TYPE_FIELDS.
17015 (build_distinct_type_copy, verify_type_variant,
17016 verify_type): Member fns are on TYPE_FIELDS.
17017 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
17018 * tree-pretty-print.c (dump_generic_node): Likewise.
17020 2017-07-20 Jakub Jelinek <jakub@redhat.com>
17023 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
17024 V2TImode and V4TImode.
17025 (ix86_expand_vector_extract): Likewise.
17026 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
17027 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
17028 (ssescalarmode): Handle V4TImode and V2TImode.
17029 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
17030 (*vec_extractv2ti, *vec_extractv4ti): New insns.
17031 (VEXTRACTI128_MODE): New mode iterator.
17032 (splitter for *vec_extractv?ti first element): New.
17033 (VEC_INIT_MODE): New mode iterator.
17034 (vec_init<mode>): Consolidate 3 expanders into one using
17035 VEC_INIT_MODE mode iterator.
17037 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
17039 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
17040 non_spilled_static_chain_regno_p.
17042 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
17044 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
17046 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
17048 * bb-reorder.c (connect_traces): Allow copying of blocks within
17051 2017-07-20 Richard Biener <rguenther@suse.de>
17053 * gimple.h (gimple_phi_result): Add gphi * overload.
17054 (gimple_phi_result_ptr): Likewise.
17055 (gimple_phi_arg): Likewise. Adjust index assert to only
17056 allow actual argument accesses rather than all slots available
17058 (gimple_phi_arg_def): Add gphi * overload.
17059 * tree-phinodes.c (make_phi_node): Initialize only actual
17061 (resize_phi_node): Clear memory not covered by old node,
17062 do not initialize excess argument slots.
17063 (reserve_phi_args_for_new_edge): Initialize new argument slot
17066 2017-07-20 Bin Cheng <bin.cheng@arm.com>
17068 PR tree-optimization/81388
17070 2016-07-21 Bin Cheng <bin.cheng@arm.com>
17072 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
17073 by removing computation of may_be_zero.
17075 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
17076 Tom de Vries <tom@codesourcery.com>
17078 PR middle-end/81030
17079 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
17080 when gimple level profile disagrees with what RTL expander did.
17082 2017-07-20 Richard Biener <rguenther@suse.de>
17084 PR tree-optimization/61171
17085 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
17086 (vect_analyze_stmt): Add slp instance parameter.
17087 (vectorizable_reduction): Likewise.
17088 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
17089 (vect_is_simple_reduction): Deal with chains not detected
17090 as SLP reduction chain, specifically not properly associated
17091 chains containing a mix of plus/minus.
17092 (get_reduction_op): Remove.
17093 (get_initial_defs_for_reduction): Simplify, pass in whether
17094 this is a reduction chain, pass in the SLP node for the PHIs.
17095 (vect_create_epilog_for_reduction): Get the SLP instance as
17097 (vectorizable_reduction): Get the SLP instance as arg.
17098 During analysis remember the SLP node with the PHIs in the
17099 instance. Simplify getting at the vectorized reduction PHIs.
17100 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
17101 through SLP instance.
17102 (vect_slp_analyze_operations): Likewise.
17103 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
17104 (vect_transform_stmt): Likewise.
17106 2017-07-20 Tom de Vries <tom@codesourcery.com>
17108 PR tree-optimization/81489
17109 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
17110 read of phi arg location to before loop that modifies phi.
17112 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17114 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
17117 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
17119 PR middle-end/81331
17120 * except.c (execute): Fix ordering issue.
17122 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
17124 PR rtl-optimization/81423
17125 * combine.c (make_compound_operation_int): Don't try to optimize
17126 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
17128 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
17130 PR rtl-optimization/81423
17131 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
17132 with a constant that is -1 in the truncated to mode.
17134 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
17136 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
17137 (determine_unlikely_bbs): ... here.
17138 * predict.h (propagate_unlikely_bbs_forward): Declare.
17139 * cfgexpand.c (pass_expand::execute): Use it.
17140 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
17142 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
17143 propagate_unlikely_bbs_forward.
17145 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
17147 PR middle-end/81331
17148 * except.c (maybe_add_nop_after_section_switch): New function.
17151 2017-07-19 Tom de Vries <tom@codesourcery.com>
17153 * gimple.h (gimple_phi_set_arg): Make assert more strict.
17155 2017-07-19 Tom de Vries <tom@codesourcery.com>
17157 * gimple.h (gimple_phi_arg): Make assert more strict.
17159 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
17161 * config.gcc (powerpc*-*-*): Add mmintrin.h.
17162 * config/rs6000/mmintrin.h: New file.
17163 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
17165 2017-07-19 Jakub Jelinek <jakub@redhat.com>
17167 PR tree-optimization/81346
17168 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
17170 2017-07-19 Tom de Vries <tom@codesourcery.com>
17172 * config/nvptx/nvptx.md (VECIM): Add V2DI.
17174 2017-07-19 Tom de Vries <tom@codesourcery.com>
17176 * config/nvptx/nvptx-modes.def: Add V2DImode.
17177 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
17178 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
17179 (nvptx_output_mov_insn): Handle lack of mov.b128.
17180 (nvptx_print_operand): Handle 'H' and 'L' codes.
17181 (nvptx_vector_mode_supported): Allow V2DImode.
17182 (nvptx_preferred_simd_mode): New function.
17183 (nvptx_data_alignment): New function.
17184 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
17185 nvptx_preferred_simd_mode.
17186 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
17188 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
17190 2017-07-19 Tom de Vries <tom@codesourcery.com>
17192 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
17193 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
17194 (nvptx_vector_mode_supported): New function. Allow V2SImode.
17195 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
17196 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
17197 (mov<VECIM>_insn): New define_insn.
17198 (define_expand "mov<VECIM>): New define_expand.
17200 2017-07-19 Tom de Vries <tom@codesourcery.com>
17202 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
17204 2017-07-19 Jakub Jelinek <jakub@redhat.com>
17206 PR tree-optimization/81346
17207 * fold-const.h (fold_div_compare, range_check_type): Declare.
17208 * fold-const.c (range_check_type): New function.
17209 (build_range_check): Use range_check_type.
17210 (fold_div_compare): No longer static, rewritten into
17211 a match.pd helper function.
17212 (fold_comparison): Don't call fold_div_compare here.
17213 * match.pd (X / C1 op C2): New optimization using fold_div_compare
17214 as helper function.
17216 2017-07-19 Nathan Sidwell <nathan@acm.org>
17218 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
17219 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
17220 * tree.c (find_decls_types_r, verify_type): Use
17221 TYPE_{MIN,MAX}_VALUE_RAW.
17222 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
17223 (hash_tree): Likewise.
17224 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
17226 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
17229 2017-07-18 Tom de Vries <tom@codesourcery.com>
17231 PR middle-end/81464
17232 * omp-expand.c (expand_omp_for_static_chunk): Handle
17233 equal-argument loop exit phi.
17235 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
17238 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
17239 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
17240 operand 2 predicate.
17241 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
17242 operand 2 predicate.
17243 (ror,rol -> rorx splitters): Use const_int_operand as
17244 operand 2 predicate.
17246 2017-06-18 Richard Biener <rguenther@suse.de>
17248 PR tree-optimization/81410
17249 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
17250 the gap in the ! slp_perm SLP case after each group.
17252 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
17254 PR middle-end/81463
17255 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
17258 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
17260 PR middle-end/81462
17261 * predict.c (set_even_probabilities): Cleanup; do not affect
17262 probabilities that are already known.
17263 (combine_predictions_for_bb): Call even when count is set.
17265 2017-07-18 Nathan Sidwell <nathan@acm.org>
17267 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
17270 2017-07-18 Bin Cheng <bin.cheng@arm.com>
17273 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
17274 optimization for loop niter analysis.
17276 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
17279 * config/avr/avr.c (avr_optimize_casesi): Don't use
17280 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
17282 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
17284 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
17285 body_cost_vec from _vect_peel_extended_info.
17286 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
17287 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
17290 2017-07-18 Bin Cheng <bin.cheng@arm.com>
17292 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
17294 2017-07-18 Richard Biener <rguenther@suse.de>
17296 PR tree-optimization/80620
17297 PR tree-optimization/81403
17298 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
17299 info when re-using a VN table entry.
17301 2017-07-18 Richard Biener <rguenther@suse.de>
17303 PR tree-optimization/81418
17304 * tree-vect-loop.c (vectorizable_reduction): Properly compute
17305 vectype_in. Verify that with lane-reducing reduction operations
17306 we have a single def-use cycle.
17308 2017-07-17 Carl Love <cel@us.ibm.com>
17310 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
17312 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17313 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
17314 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
17315 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
17316 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
17317 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
17318 VMULOSW): New enum "unspec" values.
17319 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
17320 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
17321 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
17322 altivec_vmulosw): New patterns.
17323 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
17324 VMULOSW): Add definitions.
17325 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
17327 * config/alpha/alpha.c: Include predict.h.
17329 2017-07-17 Yury Gribov <tetra2005@gmail.com>
17331 * tree-vrp.c (compare_assert_loc): Fix comparison function
17332 to return predictable results.
17334 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17336 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
17338 (subdi3): Likewise.
17339 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
17340 * doc/invoke.texi (mexpand-adddi): Update text.
17342 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17344 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
17345 that also clobbers the CC register. The old expand code is moved
17347 (*arc_clzsi2): ... here.
17348 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
17349 the CC register. The old expand code is moved to ...
17350 (arc_ctzsi2): ... here.
17352 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17354 * config/arc/arc.opt (mindexed-loads): Use initial value
17355 TARGET_INDEXED_LOADS_DEFAULT.
17356 (mauto-modify-reg): Use initial value
17357 TARGET_AUTO_MODIFY_REG_DEFAULT.
17358 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
17359 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
17360 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
17361 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
17363 2017-07-17 Martin Liska <mliska@suse.cz>
17366 * opts.c (finish_options): Do not allow -fgnu-tm
17367 w/ -fsanitize={kernel-,}address. Say sorry.
17369 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17372 * tree-loop-distribution.c (classify_partition): Only assert on
17373 numer of iterations.
17374 (merge_dep_scc_partitions): Delete prameter. Update function call.
17375 (distribute_loop): Remove code handling loop with unknown niters.
17376 (pass_loop_distribution::execute): Skip loop with unknown niters.
17378 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17381 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
17382 function sort_partitions_by_post_order.
17384 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17386 PR tree-optimization/81374
17387 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
17388 the max index of basic blocks, rather than number of basic blocks.
17390 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17392 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
17394 (arc_legitimate_pic_operand_p): Likewise.
17395 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
17397 (arc_needs_pcl_p): Likewise.
17398 (arc_legitimate_pc_offset_p): Likewise.
17399 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
17400 function is also used in constrains.md.
17401 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
17402 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
17403 PLUS. Only return true/false in known cases, otherwise assert.
17404 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
17405 is already called in arc_legitimate_constant_p.
17406 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
17408 (LEGITIMATE_PIC_OPERAND_P): Use
17409 arc_raw_symbolic_reference_mentioned_p function.
17410 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
17415 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17416 Andrew Burgess <andrew.burgess@embecosm.com>
17418 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
17419 (arc_return_address_register): New function.
17420 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
17421 (arc_handle_fndecl_attribute): Add naked attribute.
17422 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
17423 (TARGET_WARN_FUNC_RETURN): Likewise.
17424 (arc_allocate_stack_slots_for_args): New function.
17425 (arc_warn_func_return): Likewise.
17426 (machine_function): Change type fn_type.
17427 (arc_compute_function_type): Consider new naked function type,
17428 change function return type.
17429 (arc_must_save_register): Adapt to handle new
17430 arc_compute_function_type's return type.
17431 (arc_expand_prologue): Likewise.
17432 (arc_expand_epilogue): Likewise.
17433 (arc_return_address_regs): Delete.
17434 (arc_return_address_register): New function.
17435 (arc_epilogue_uses): Use above function.
17436 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
17437 (arc_function_type): Change encoding, add naked type.
17438 (ARC_INTERRUPT_P): Change to handle the new encoding.
17439 (ARC_FAST_INTERRUPT_P): Likewise.
17440 (ARC_NORMAL_P): Define.
17441 (ARC_NAKED_P): Likewise.
17442 (arc_compute_function_type): Delete prototype.
17443 * config/arc/arc.md (in_ret_delay_slot): Use
17444 arc_return_address_register function.
17445 (simple_return): Likewise.
17446 (p_return_i): Likewise.
17448 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17450 PR tree-optimization/81428
17451 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
17452 can't be built for those types.
17454 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
17456 Remove stuff dead since r239246.
17458 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
17459 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
17460 (avr_inform_devices): Remove dead stuff.
17462 2017-07-17 Tamar Christina <tamar.christina@arm.com>
17464 * config/arm/arm_neon.h: Fix softp typo.
17466 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17468 PR tree-optimization/81365
17469 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
17470 aggregate moves onto bb predecessor edges, make sure there are no
17471 loads that could alias the lhs in between the start of bb and the
17474 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
17477 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
17478 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
17479 [LSHIFTRT, outer_code = TRUNCATE]: Same.
17481 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17483 PR tree-optimization/81396
17484 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
17485 (init_symbolic_number): Initialize it to 1.
17486 (perform_symbolic_merge): Add n_ops from both operands into the new
17488 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
17489 without base_addr as useless if they need more than one operation.
17490 (bswap_replace): Handle !bswap case for NULL base_addr.
17492 2017-07-17 Tom de Vries <tom@codesourcery.com>
17495 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
17498 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
17500 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
17501 conditional builtin define __FIX_LEON3FT_B2BST.
17503 2017-07-17 Daniel Cederman <cederman@gaisler.com>
17505 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
17506 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
17509 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
17511 PR rtl-optimization/81424
17512 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
17513 to remove potential trapping from operands if -fnon-call-exceptions.
17515 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17517 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
17518 profile_proability for scalling.
17519 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
17521 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17523 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
17525 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17527 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
17528 fixpoint arithmetics.
17530 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17532 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
17533 fixpoint arithmetics.
17535 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17537 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
17538 fixpoint arithmetics.
17540 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17542 * profile-count.h (profile_probability::from_reg_br_prob_note,
17543 profile_probability::to_reg_br_prob_note): New functions.
17544 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
17545 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
17546 * predict.c (probability_reliable_p): Update.
17547 (edge_probability_reliable_p): Update.
17548 (br_prob_note_reliable_p): Update.
17549 (invert_br_probabilities): Update.
17550 (add_reg_br_prob_note): New function.
17551 (combine_predictions_for_insn): Update.
17552 * asan.c (asan_clear_shadow): Update.
17553 * cfgbuild.c (compute_outgoing_frequencies): Update.
17554 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
17555 (update_br_prob_note): Update.
17556 (rtl_verify_edges): Update.
17557 (purge_dead_edges): Update.
17558 (fixup_reorder_chain): Update.
17559 * emit-rtl.c (try_split): Update.
17560 * ifcvt.c (cond_exec_process_insns): Update.
17561 (cond_exec_process_if_block): Update.
17562 (dead_or_predicable): Update.
17563 * internal-fn.c (expand_addsub_overflow): Update.
17564 (expand_neg_overflow): Update.
17565 (expand_mul_overflow): Update.
17566 * loop-doloop.c (doloop_modify): Update.
17567 * loop-unroll.c (compare_and_jump_seq): Update.
17568 * optabs.c (emit_cmp_and_jump_insn_1): Update.
17569 * predict.h: Update.
17570 * reorg.c (mostly_true_jump): Update.
17572 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
17573 * config/alpha/alpha.c (emit_unlikely_jump): Update.
17574 * config/arc/arc.c: (emit_unlikely_jump): Update.
17575 * config/arm/arm.c: (emit_unlikely_jump): Update.
17576 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
17577 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
17578 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
17579 (ix86_print_operand): Update.
17580 (ix86_split_fp_branch): Update.
17581 (predict_jump): Update.
17582 * config/ia64/ia64.c (ia64_print_operand): Update.
17583 * config/mmix/mmix.c (mmix_print_operand): Update.
17584 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
17585 (rs6000_expand_split_stack_prologue): Update.
17586 * config/rs6000/rs6000.c: Update.
17587 * config/s390/s390.c (s390_expand_vec_strlen): Update.
17588 (s390_expand_vec_movstr): Update.
17589 (s390_expand_cs_tdsi): Update.
17590 (s390_expand_split_stack_prologue): Update.
17591 * config/sh/sh.c (sh_print_operand): Update.
17592 (expand_cbranchsi4): Update.
17593 (expand_cbranchdi4): Update.
17594 * config/sparc/sparc.c (output_v9branch): Update.
17595 * config/spu/spu.c (get_branch_target): Update.
17596 (ea_load_store_inline): Update.
17597 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
17598 * config/tilepro/tilepro.c: Update.
17600 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
17602 * gimplify.c (mostly_copy_tree_r): Revert latest change.
17603 (gimplify_save_expr): Likewise.
17605 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17607 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
17609 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17611 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
17613 * timevar.def (TV_IPA_FNSUMMARY): Define.
17615 2017-07-16 Daniel Cederman <cederman@gaisler.com>
17617 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
17618 to back store errata sensitive sequence from being generated.
17619 (sqrtdf2_fix): Likewise.
17621 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17623 * tree-ssa-threadupdate.c (compute_path_counts,
17624 update_joiner_offpath_counts): Use profile_probability.
17626 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
17629 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17631 * config/arm/arm-c.c (arm_cpu_builtins): Define
17632 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
17634 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
17636 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17637 array entries to represent __ieee128 versions of the
17638 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
17639 scalar_extract_sig, and scalar_insert_exp built-in functions.
17640 (altivec_resolve_overloaded_builtin): Add special case handling
17641 for the __builtin_scalar_insert_exp function, as represented by
17642 the P9V_BUILTIN_VEC_VSIEDP constant.
17643 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
17644 exponent support for __ieee128 argument.
17645 (VSESQP): Add scalar extract signature support for __ieee128
17647 (VSTDCNQP): Add scalar test negative support for __ieee128
17649 (VSIEQP): Add scalar insert exponent support for __int128 argument
17650 with __ieee128 result.
17651 (VSIEQPF): Add scalar insert exponent support for __ieee128
17652 argument with __ieee128 result.
17653 (VSTDCQP): Add scalar test data class support for __ieee128
17655 (VSTDCNQP): Add overload support for scalar test negative with
17656 __ieee128 argument.
17657 (VSTDCQP): Add overload support for scalar test data class
17658 __ieee128 argument.
17659 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
17660 UNSPEC_VSX_SXSIGDP.
17661 (UNSPEC_VSX_SIEXPQP): New constant.
17662 (xsxexpqp): New insn for VSX scalar extract exponent quad
17664 (xsxsigqp): New insn for VSX scalar extract significand quad
17666 (xsiexpqpf): New insn for VSX scalar insert exponent quad
17667 precision with floating point argument.
17668 (xststdcqp): New expand for VSX scalar test data class quad
17670 (xststdcnegqp): New expand for VSX scalar test negative quad
17672 (xststdcqp): New insn to match expansions for VSX scalar test data
17673 class quad precision and VSX scalar test negative quad precision.
17674 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
17675 special case operand checking to enforce that second operand of
17676 VSX scalar test data class with quad precision argument is a 7-bit
17678 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
17679 prototypes and descriptions of __ieee128 versions of
17680 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
17681 scalar_test_data_class, and scalar_test_neg built-in functions.
17683 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17685 PR tree-optimization/81162
17686 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
17687 replace a negate with an add.
17689 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
17691 * doc/invoke.texi (arm/-mcpu): Document +crypto.
17693 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17695 * config/arm/arm-c.c (arm_cpu_builtins): Define
17696 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
17698 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17700 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
17701 (armv8-r): Set ARM Cortex-R52 as default CPU.
17702 * config/arm/arm-tables.opt: Regenerate.
17703 * config/arm/arm-tune.md: Regenerate.
17704 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
17706 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
17707 extension for -mcpu=cortex-r52.
17709 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17711 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
17712 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
17713 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
17714 (fp-armv8): Define it as FP_ARMv8 only.
17715 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
17716 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
17718 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
17719 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
17720 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
17721 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
17722 than TARGET_FPU_ARMV8.
17723 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
17724 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
17725 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
17727 * config/arm/neon.md (neon_vrint): Likewise.
17728 (neon_vcvt): Likewise.
17729 (neon_<fmaxmin_op><mode>): Likewise.
17730 (<fmaxmin><mode>3): Likewise.
17731 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
17732 * config/arm/predicates.md (arm_cond_move_operator): Check against
17733 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
17735 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
17737 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
17738 to top of function.
17740 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17742 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
17743 loop in comment with memset.
17745 2017-07-14 Martin Liska <mliska@suse.cz>
17747 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
17748 * dwarf2out.c (is_java): Remove the function.
17749 (output_pubname): Remove usage of the function.
17750 (lower_bound_default): Remove usage of DW_LANG_Java.
17751 (gen_compile_unit_die): Likewise.
17752 * gcc.c: Remove compiler defaults for .java and .zip files.
17753 * gimple-expr.c (remove_suffix): Change as there's no longer
17754 extension than 4-letter one.
17755 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
17756 (gimplify_save_expr): Likewise.
17757 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
17758 as it's possible even for other languages than Java.
17759 * langhooks.h (struct lang_hooks): Remove Java from a comment.
17760 * lto-opts.c (lto_write_options): Remove reference to Java.
17761 * opts.c (strip_off_ending): Update file extension handling.
17762 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
17763 * tree-eh.c (lower_resx): Likewise.
17764 * tree.c (free_lang_data_in_type): Remove dead code.
17765 (find_decls_types_r): Likewise.
17766 (build_common_builtin_nodes): Remove Java from a comment.
17767 (verify_type): Remove dead code.
17768 * varasm.c (assemble_external): Remove Java from a comment.
17770 2017-07-14 Martin Liska <mliska@suse.cz>
17772 * opts.c (finish_options): Add quotes.
17773 (common_handle_option): Likewise.
17775 2017-07-14 Martin Liska <mliska@suse.cz>
17777 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
17778 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
17779 Remove N_SO_PASCAL.
17780 * dwarf2out.c (lower_bound_default): Do not handle
17782 (gen_compile_unit_die): Likewise.
17783 * gcc.c: Remove default extension binding for GNU Pascal.
17784 * stmt.c: Remove Pascal language from a comment.
17785 * xcoffout.c: Likewise.
17787 2017-07-13 David Malcolm <dmalcolm@redhat.com>
17790 * diagnostic-show-locus.c (fixit_cmp): New function.
17791 (layout::layout): Sort m_fixit_hints.
17792 (column_range::column_range): Assert that the values are valid.
17793 (struct char_span): New struct.
17794 (correction::overwrite): New method.
17795 (struct source_line): New struct.
17796 (line_corrections::add_hint): Add assertions. Reimplement memcpy
17797 calls in terms of classes source_line and char_span, and
17798 correction::overwrite.
17799 (selftest::test_overlapped_fixit_printing_2): New function.
17800 (selftest::diagnostic_show_locus_c_tests): Call it.
17802 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
17804 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
17805 early if there is no lhs.
17807 2017-07-13 Martin Liska <mliska@suse.cz>
17809 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
17810 (gen_reference_type_die): Likewise.
17811 * stor-layout.c: Remove Pascal-related comment.
17813 2017-07-13 Martin Liska <mliska@suse.cz>
17815 * opts.c (finish_options): Add quotes to error messages.
17816 (parse_sanitizer_options): Likewise.
17818 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
17820 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
17822 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
17824 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
17826 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
17828 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
17830 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
17832 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
17835 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
17836 provides the hardware capability bits, define the macro
17837 __BUILTIN_CPU_SUPPORTS__.
17838 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
17839 if GLIBC does not provide the hardware capability bits. Add a
17840 gcc_unreachable call if the built-in cpu function is neither
17841 __builtin_cpu_is nor __builtin_cpu_supports.
17842 (rs6000_get_function_versions_dispatcher): Change the warning
17843 that an old GLIBC is used which does not export the capability
17844 bits to be an error.
17845 * doc/extend.texi (target_clones attribute): Document the
17846 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
17847 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
17848 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
17849 the macros defined by GCC if the newer GLIBC is available.
17851 2017-07-12 Jeff Law <law@redhat.com>
17853 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
17854 remaining includes slightly.
17855 * config/riscv/riscv-builtins.c: Include profile-count.h.
17857 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
17860 * config/avr/avr.c (avr_set_current_function): In diagnostic
17861 messages: Quote keywords and (parts of) identifiers.
17862 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
17865 2017-07-12 Carl Love <cel@us.ibm.com>
17867 * config/rs6000/rs6000-c.c: Add support for built-in functions
17868 vector bool char vec_revb (vector bool char);
17869 vector bool short vec_revb (vector short char);
17870 vector bool int vec_revb (vector bool int);
17871 vector bool long long vec_revb (vector bool long long);
17872 * doc/extend.texi: Update the built-in documentation file for the
17873 new built-in functions.
17875 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17877 * config/s390/s390.md: Remove movcc splitter.
17879 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17881 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
17882 load/store on condition.
17884 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
17887 * config/avr/avr.c (avr_encode_section_info)
17888 [progmem && !TREE_READONLY]: Error if progmem object needs
17891 2017-07-11 Michael Collison <michael.collison@arm.com>
17893 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
17896 2017-07-11 Carl Love <cel@us.ibm.com>
17898 * config/rs6000/rs6000-c.c: Add support for builtins
17899 vector unsigned int vec_parity_lsbb (vector signed int);
17900 vector unsigned int vec_parity_lsbb (vector unsigned int);
17901 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
17902 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
17903 vector unsigned long long vec_parity_lsbb (vector signed long long);
17904 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
17905 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
17906 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
17907 * doc/extend.texi: Update the built-in documentation file for the
17908 new built-in functions.
17910 2017-07-11 David Malcolm <dmalcolm@redhat.com>
17912 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
17913 (layout::m_primary_loc): New field.
17914 (layout::layout): Initialize new field. Move location filtering
17915 logic from here to...
17916 (layout::maybe_add_location_range): ...this new method. Add
17917 support for filtering to just the lines already specified by other
17919 (layout::will_show_line_p): New method.
17920 (gcc_rich_location::add_location_if_nearby): New method.
17921 (selftest::test_add_location_if_nearby): New test function.
17922 (selftest::diagnostic_show_locus_c_tests): Call it.
17923 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
17926 2017-07-11 Tom de Vries <tom@codesourcery.com>
17928 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
17929 (bb_first_real_insn): New function.
17930 (nvptx_single): Add extra initialization of broadcasted condition
17933 2017-07-11 Nathan Sidwell <nathan@acm.org>
17935 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
17937 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
17939 * doc/extend.texi (AVR Function Attributes): Remove weblink to
17940 Binutils doc as TEXI will mess them up.
17941 * doc/invoke.texi (AVR Options): Same here.
17943 2017-07-11 Daniel Cederman <cederman@gaisler.com>
17945 * config/sparc/sparc.opt (mfix-ut700): New option.
17946 (mfix-gr712rc): Likewise.
17947 (sparc_fix_b2bst): New variable.
17948 * doc/invoke.texi (SPARC options): Document them.
17949 (ARM options): Fix warnings.
17950 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
17951 instructions to prevent sequences that can trigger the store-store
17952 errata for certain LEON3FT processors.
17953 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
17954 (sparc_option_override): Set sparc_fix_b2bst appropriately.
17955 * config/sparc/sparc.md (fix_b2bst): New attribute.
17956 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
17958 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
17961 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
17963 (*rsqrtsf2_sse): Ditto.
17965 (div<mode>3): Macroize insn from divdf3 and divsf3
17966 using MODEF mode iterator.
17968 2017-07-10 Martin Sebor <msebor@redhat.com>
17970 PR tree-optimization/80397
17971 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
17972 instead of testing for equality to INTEGER_TYPE.
17974 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
17976 * config.gcc: Remove uclibc from arc target spec.
17978 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
17980 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
17982 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17985 * lto-wrapper.c (remove_option): New function.
17986 (merge_and_complain): Merge PIC/PIE options more realistically.
17988 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
17990 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
17994 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
17995 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
17996 * config.in: Regenerate.
17997 * configure: Regenerate.
17998 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
17999 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
18000 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
18001 (TARGET_GASISR_PROLOGUES): ...target mask.
18002 * common/config/avr/avr-common.c
18003 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
18004 Set -mgas-isr-prologues.
18005 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
18006 INSERT_PASS_BEFORE for it.
18007 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
18008 * config/avr/avr.c (avr_option_override)
18009 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
18010 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
18011 (avr_attribute_table) <no_gccisr>: Add new function attribute.
18012 (avr_set_current_function) <is_no_gccisr>: Init machine field.
18013 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
18015 (make_avr_pass_pre_proep): New function.
18016 (emit_push_sfr) <treg>: Add argument to function and use it
18017 instead of TMP_REG.
18018 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
18019 and set machine->gasisr.yes.
18020 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
18021 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
18022 __gcc_isr.n_pushed to .L__stack_usage.
18023 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
18024 (avr_asm_final_postscan_insn): ...this new static function.
18025 * config/avr/avr.h (machine_function)
18026 <is_no_gccisr, use_L__stack_usage>: New fields.
18027 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
18028 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
18029 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
18030 (gasisr, *gasisr): New expander and insn.
18031 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
18032 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
18033 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
18035 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
18037 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
18040 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
18042 Move jump-tables out of .text again.
18045 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
18046 (ASM_OUTPUT_ADDR_VEC): New function.
18047 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
18048 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
18049 INSN_ADDRESSes as asm comment.
18050 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
18051 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
18052 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
18053 * config/avr/avr.md (*tablejump): Adjust comment.
18054 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
18055 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
18057 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
18058 (avr_output_addr_vec): New proto.
18059 (avr_log_t) <insn_addresses>: New field.
18061 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
18064 * config/i386/i386.c (ix86_function_arg_advance): Set
18065 outgoing_args_on_stack to true if there are outgoing arguments
18067 (ix86_function_arg): Likewise.
18068 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
18069 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
18070 * config/i386/i386.h (machine_function): Add
18071 outgoing_args_on_stack.
18073 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
18075 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
18076 supporting pthreds.
18077 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
18079 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
18081 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
18082 (REAL_H): Remove $(MACHMODE_H).
18083 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
18085 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
18086 $(MACHMODE_H) and double-int.h.
18087 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
18089 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
18092 2017-07-07 Andrew Pinski <apinski@cavium.com>
18094 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
18095 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
18097 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
18099 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
18100 Add warning if GCC was not configured to link against a GLIBC that
18101 exports the hardware capability bits.
18102 (make_resolver_func): Make resolver function private and not a
18103 COMDAT function. Create the name with clone_function_name instead
18104 of make_unique_name.
18107 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
18108 correct operand in doing the split.
18110 2017-07-07 Carl Love <cel@us.ibm.com>
18112 * config/rs6000/rs6000-c: Add support for built-in function
18113 vector unsigned short vec_pack_to_short_fp32 (vector float,
18115 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
18116 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
18117 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
18118 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
18119 (convert_4f32_8i16): Add define_expand.
18120 * doc/extend.texi: Update the built-in documentation file for the
18121 new built-in function.
18123 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18125 * config/sparc/m8.md: New file.
18126 * config/sparc/sparc.md: Include m8.md.
18128 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18130 * config/sparc/sparc.opt: New option -mvis4b.
18131 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
18132 (sparc_option_override): Handle VIS4B.
18133 (enum sparc_builtins): Define
18134 SPARC_BUILTIN_DICTUNPACK{8,16,32},
18135 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
18136 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
18137 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
18138 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
18139 (check_constant_argument): New function.
18140 (sparc_vis_init_builtins): Define builtins
18141 __builtin_vis_dictunpack{8,16,32},
18142 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
18143 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
18144 __builtin_vis_fpcmpde{8,16,32}shl and
18145 __builtin_vis_fpcmpur{8,16,32}shl.
18146 (sparc_expand_builtin): Check that the constant operands to
18147 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
18148 constant and in range.
18149 * config/sparc/sparc-c.c (sparc_target_macros): Handle
18151 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
18152 (SPARC_IMM5_P): Likewise.
18153 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
18154 (enabled): Handle vis4b.
18155 (UNSPEC_DICTUNPACK): New unspec.
18156 (UNSPEC_FPCMPSHL): Likewise.
18157 (UNSPEC_FPUCMPSHL): Likewise.
18158 (UNSPEC_FPCMPDESHL): Likewise.
18159 (UNSPEC_FPCMPURSHL): Likewise.
18160 (cpu_feature): New CPU feature `vis4b'.
18161 (dictunpack{8,16,32}): New insns.
18162 (FPCSMODE): New mode iterator.
18163 (fpcscond): New code iterator.
18164 (fpcsucond): Likewise.
18165 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
18166 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
18167 (fpcmpde{8,16,32}{si,di}shl): Likewise.
18168 (fpcmpur{8,16,32}{si,di}shl): Likewise.
18169 * config/sparc/constraints.md: Define constraints `q' for unsigned
18170 2-bit integer constants and `t' for unsigned 5-bit integer
18172 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
18174 (imm5_operand_dictunpack16): Likewise.
18175 (imm5_operand_dictunpack32): Likewise.
18176 (imm2_operand): Likewise.
18177 * doc/invoke.texi (SPARC Options): Document -mvis4b.
18178 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
18179 ditunpack* and fpcmp*shl builtins.
18181 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18183 * config.gcc: Handle m8 in --with-{cpu,tune} options.
18184 * config.in: Add HAVE_AS_SPARC6 define.
18185 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
18187 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
18189 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
18190 (CPP_CPU_SPEC): Handle m8.
18191 (ASM_CPU_SPEC): Likewise.
18192 * config/sparc/sparc-opts.h (enum processor_type): Add
18194 * config/sparc/sparc.c (m8_costs): New struct.
18195 (sparc_option_override): Handle TARGET_CPU_m8.
18196 (sparc32_initialize_trampoline): Likewise.
18197 (sparc64_initialize_trampoline): Likewise.
18198 (sparc_issue_rate): Likewise.
18199 (sparc_register_move_cost): Likewise.
18200 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
18201 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
18202 (ASM_CPU64_DEFAULT_SPEC): Likewise.
18203 (CPP_CPU_SPEC): Handle M8.
18204 (ASM_CPU_SPEC): Likewise.
18205 (AS_M8_FLAG): Define.
18206 * config/sparc/sparc.md: Add m8 to the cpu attribute.
18207 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
18208 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
18210 * configure: Regenerate.
18211 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
18214 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18216 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
18218 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
18219 ("*movdi_insn_sp32"): Do not set v3pipe.
18220 ("*movsi_insn"): Likewise.
18221 ("*movdi_insn_sp64"): Likewise.
18222 ("*movsf_insn"): Likewise.
18223 ("*movdf_insn_sp32"): Likewise.
18224 ("*movdf_insn_sp64"): Likewise.
18225 ("*zero_extendsidi2_insn_sp64"): Likewise.
18226 ("*sign_extendsidi2_insn"): Likewise.
18227 ("*mov<VM32:mode>_insn"): Likewise.
18228 ("*mov<VM64:mode>_insn_sp64"): Likewise.
18229 ("*mov<VM64:mode>_insn_sp32"): Likewise.
18230 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
18231 ("<vlop:code><VL:mode>3"): Likewise.
18232 ("*not_<vlop:code><VL:mode>3"): Likewise.
18233 ("*nand<VL:mode>_vis"): Likewise.
18234 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
18235 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
18236 ("one_cmpl<VL:mode>2"): Likewise.
18237 ("faligndata<VM64:mode>_vis"): Likewise.
18238 ("alignaddrsi_vis"): Likewise.
18239 ("alignaddrdi_vis"): Likweise.
18240 ("alignaddrlsi_vis"): Likewise.
18241 ("alignaddrldi_vis"): Likewise.
18242 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
18243 ("bmaskdi_vis"): Likewise.
18244 ("bmasksi_vis"): Likewise.
18245 ("bshuffle<VM64:mode>_vis"): Likewise.
18246 ("cmask8<P:mode>_vis"): Likewise.
18247 ("cmask16<P:mode>_vis"): Likewise.
18248 ("cmask32<P:mode>_vis"): Likewise.
18249 ("pdistn<P:mode>_vis"): Likewise.
18250 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
18252 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18254 * config/sparc/sparc.md ("subtype"): New insn attribute.
18255 ("*wrgsr_sp64"): Set insn subtype.
18256 ("*rdgsr_sp64"): Likewise.
18257 ("alignaddrsi_vis"): Likewise.
18258 ("alignaddrdi_vis"): Likewise.
18259 ("alignaddrlsi_vis"): Likewise.
18260 ("alignaddrldi_vis"): Likewise.
18261 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
18262 ("fexpand_vis"): Likewise.
18263 ("fpmerge_vis"): Likewise.
18264 ("faligndata<VM64:mode>_vis"): Likewise.
18265 ("bshuffle<VM64:mode>_vis"): Likewise.
18266 ("cmask8<P:mode>_vis"): Likewise.
18267 ("cmask16<P:mode>_vis"): Likewise.
18268 ("cmask32<P:mode>_vis"): Likewise.
18269 ("fchksm16_vis"): Likewise.
18270 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
18271 ("fmean16_vis"): Likewise.
18272 ("fp<plusminus_insn>64_vis"): Likewise.
18273 ("<plusminus_insn>v8qi3"): Likewise.
18274 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
18275 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
18276 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
18277 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
18278 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
18279 ("*movqi_insn"): Likewise.
18280 ("*movhi_insn"): Likewise.
18281 ("*movsi_insn"): Likewise.
18282 ("movsi_pic_gotdata_op"): Likewise.
18283 ("*movdi_insn_sp32"): Likewise.
18284 ("*movdi_insn_sp64"): Likewise.
18285 ("movdi_pic_gotdata_op"): Likewise.
18286 ("*movsf_insn"): Likewise.
18287 ("*movdf_insn_sp32"): Likewise.
18288 ("*movdf_insn_sp64"): Likewise.
18289 ("*zero_extendhisi2_insn"): Likewise.
18290 ("*zero_extendqihi2_insn"): Likewise.
18291 ("*zero_extendqisi2_insn"): Likewise.
18292 ("*zero_extendqidi2_insn"): Likewise.
18293 ("*zero_extendhidi2_insn"): Likewise.
18294 ("*zero_extendsidi2_insn_sp64"): Likewise.
18295 ("ldfsr"): Likewise.
18296 ("prefetch_64"): Likewise.
18297 ("prefetch_32"): Likewise.
18298 ("tie_ld32"): Likewise.
18299 ("tie_ld64"): Likewise.
18300 ("*tldo_ldub_sp32"): Likewise.
18301 ("*tldo_ldub1_sp32"): Likewise.
18302 ("*tldo_ldub2_sp32"): Likewise.
18303 ("*tldo_ldub_sp64"): Likewise.
18304 ("*tldo_ldub1_sp64"): Likewise.
18305 ("*tldo_ldub2_sp64"): Likewise.
18306 ("*tldo_ldub3_sp64"): Likewise.
18307 ("*tldo_lduh_sp32"): Likewise.
18308 ("*tldo_lduh1_sp32"): Likewise.
18309 ("*tldo_lduh_sp64"): Likewise.
18310 ("*tldo_lduh1_sp64"): Likewise.
18311 ("*tldo_lduh2_sp64"): Likewise.
18312 ("*tldo_lduw_sp32"): Likewise.
18313 ("*tldo_lduw_sp64"): Likewise.
18314 ("*tldo_lduw1_sp64"): Likewise.
18315 ("*tldo_ldx_sp64"): Likewise.
18316 ("*mov<VM32:mode>_insn"): Likewise.
18317 ("*mov<VM64:mode>_insn_sp64"): Likewise.
18318 ("*mov<VM64:mode>_insn_sp32"): Likewise.
18320 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18322 * config/sparc/sparc.md ("type"): New insn type viscmp.
18323 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
18325 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
18326 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
18327 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
18328 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
18330 ("n7_vis_logical_11cycle"): Likewise.
18331 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
18332 * config/sparc/niagara2.md ("niag3_vis": Likewise.
18333 * config/sparc/niagara.md ("niag_vis"): Likewise.
18334 * config/sparc/ultra3.md ("us3_fga"): Likewise.
18335 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
18337 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18339 * config/sparc/sparc.md: New instruction type `bmask'.
18340 (bmaskdi_vis): Use the `bmask' type.
18341 (bmasksi_vis): Likewise.
18342 * config/sparc/ultra3.md (us3_array): Likewise.
18343 * config/sparc/niagara7.md (n7_array): Likewise.
18344 * config/sparc/niagara4.md (n4_array): Likewise.
18345 * config/sparc/niagara2.md (niag2_vis): Likewise.
18346 (niag3_vis): Likewise.
18347 * config/sparc/niagara.md (niag_vis): Likewise.
18349 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18351 * ipa-comdats.c: Remove optimize check from gate.
18352 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
18353 for functions not optimized.
18354 (ipa_fn_summary_read): Skip optimize check.
18355 (ipa_fn_summary_write): Likewise.
18356 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
18358 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
18360 (can_inline_edge_p): Check flag_pcc_struct_return for match.
18361 (check_callers): Give up on caller which is not optimized.
18362 (inline_small_functions): Likewise.
18363 (ipa_inline): Do not give up when not optimizing.
18364 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
18365 away unoptimizes cdtors.
18366 (whole_program_function_and_variable_visibility): Do
18367 ipa_discover_readonly_nonaddressable_vars in LTO mode.
18368 * ipa.c (process_references): Do not check optimize.
18369 (symbol_table::remove_unreachable_nodes): Update optimize check.
18370 (set_writeonly_bit): Update optimize check.
18371 (pass_ipa_cdtor_merge::gate): Do not check optimize.
18372 (pass_ipa_single_use::gate): Remove.
18374 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18376 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
18377 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
18378 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
18379 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
18380 permute_load, permute_store, adjust_extract, adjust_splat,
18381 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
18382 replace_swap_with_copy, dump_swap_insn_table,
18383 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
18384 recombine_lvx_pattern, recombine_stvx_pattern,
18385 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
18386 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
18387 to file rs6000-p8swap.c.
18388 * config/rs6000/rs6000-p8swap.c: New file.
18389 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
18390 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
18391 and rs6000*-*-* targets.
18393 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18395 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
18397 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18399 * lto-wrapper.c (merge_and_complain): Do not merge
18400 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
18401 fsigned_zeros, ftrapping_math, fwrapv.
18402 (append_compiler_options): Do not track these options.
18403 (append_linker_options): Likewie
18405 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18407 * cgraphunit.c (cgraph_node::finalize_function): When
18408 !flag_toplevel_reorde set no_reorder flag.
18409 (varpool_node::finalize_decl): Likewise.
18410 (symbol_table::compile): Drop no toplevel reorder path.
18412 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18414 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
18415 edges; zero probability is not better than uninitialized.
18417 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
18419 * asan.h (asan_sanitize_allocas_p): Declare.
18420 * asan.c (asan_sanitize_allocas_p): New function.
18421 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
18422 (handle_builtin_alloca): Likewise.
18423 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
18424 if !asan_sanitize_allocas_p.
18425 * params.def (asan-instrument-allocas): Add new option.
18426 * params.h (ASAN_PROTECT_ALLOCAS): Define.
18427 * opts.c (common_handle_option): Disable allocas sanitization for
18430 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
18432 * asan.c: Include gimple-fold.h.
18433 (get_last_alloca_addr): New function.
18434 (handle_builtin_stackrestore): Likewise.
18435 (handle_builtin_alloca): Likewise.
18436 (asan_emit_allocas_unpoison): Likewise.
18437 (get_mem_refs_of_builtin_call): Add new parameter, remove const
18438 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
18439 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
18440 (instrument_builtin_call): Pass gimple iterator to
18441 get_mem_refs_of_builtin_call.
18442 (last_alloca_addr): New global.
18443 * asan.h (asan_emit_allocas_unpoison): Declare.
18444 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
18445 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
18446 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
18447 if function calls alloca.
18448 * gimple-fold.c (replace_call_with_value): Remove static keyword.
18449 * gimple-fold.h (replace_call_with_value): Declare.
18450 * internal-fn.c: Include asan.h.
18451 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
18452 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
18454 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18456 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
18457 (C_SELFTEST_FLAGS): New.
18458 (CPP_SELFTEST_FLAGS): New.
18459 (SELFTEST_DEPS): New, from deps of s-selftest.
18460 (C_SELFTEST_DEPS): New, from deps of s-selftest.
18461 (CPP_SELFTEST_DEPS): New.
18462 (selftest): Add dependency on s-selftest-c++.
18463 (s-selftest): Rename to...
18464 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
18465 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
18466 than SELFTEST_FLAGS.
18467 (selftest-gdb): Rename to...
18468 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
18470 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
18471 (selftest-valgrind): Rename to...
18472 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
18474 (selftest-valgrind): Reintroduce as an alias for
18475 selftest-c-valgrind.
18476 (s-selftest-c++): New.
18477 (selftest-c++-gdb): New.
18478 (selftest-c++-valgrind): New.
18480 2017-07-06 Olivier Hainque <hainque@adacore.com>
18482 * gcc.c (process_command): When deciding if undefined variables
18483 should be ignored when processing specs, accept "gcc -v" as well.
18485 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18487 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
18488 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
18490 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
18492 * config/arm/arm-cpus.in (armv8-r): Add new entry.
18493 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
18494 * config/arm/arm-tables.opt: Regenerate.
18495 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
18497 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
18499 2017-07-06 Carl Love <cel@us.ibm.com>
18501 * ChangeLog: Clean up from mid air collision
18503 2017-07-06 Carl Love <cel@us.ibm.com>
18505 * config/rs6000/rs6000-c.c: Add support for built-in functions
18506 vector signed int vec_subc (vector signed int, vector signed int);
18507 vector signed __int128 vec_subc (vector signed __int128,
18508 vector signed __int128);
18509 vector unsigned __int128 vec_subc (vector unsigned __int128,
18510 vector unsigned __int128);
18511 vector signed int vec_sube (vector signed int, vector signed int,
18512 vector signed int);
18513 vector unsigned int vec_sube (vector unsigned int,
18514 vector unsigned int,
18515 vector unsigned int);
18516 vector signed __int128 vec_sube (vector signed __int128,
18517 vector signed __int128,
18518 vector signed__int128);
18519 vector unsigned __int128 vec_sube (vector unsigned __int128,
18520 vector unsigned __int128,
18521 vector unsigned __int128);
18522 vector signed int vec_subec (vector signed int, vector signed int,
18523 vector signed int);
18524 vector unsigned int vec_subec (vector unsigned int,
18525 vector unsigned int,
18526 vector unsigned int);
18527 vector signed __int128 vec_subec (vector signed __int128,
18528 vector signed __int128,
18529 vector signed__int128);
18530 vector unsigned __int128 vec_subec (vector unsigned __int128,
18531 vector unsigned __int128,
18532 vector unsigned __int128);
18533 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
18534 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
18535 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
18536 BU_ALTIVEC_OVERLOAD_X definitions.
18537 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
18538 * doc/extend.texi: Update the built-in documentation file for the new
18539 built-in functions.
18541 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18544 * diagnostic-show-locus.c (layout::layout): Use start and finish
18545 spelling location for the start and finish of each range.
18546 * genmatch.c (linemap_client_expand_location_to_spelling_point):
18547 Add unused aspect param.
18548 * input.c (expand_location_1): Add "aspect" param, and use it
18549 to access the correct part of the location.
18550 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
18552 (expand_location_to_spelling_point): Likewise.
18553 (linemap_client_expand_location_to_spelling_point): Add "aspect"
18554 param, and pass it to expand_location_1.
18556 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
18558 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
18559 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
18560 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
18561 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
18562 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
18563 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
18564 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
18565 _mm_maskz_getmant_ss): New intrinsics.
18566 (__builtin_ia32_getexpss128_mask): Changed to ...
18567 __builtin_ia32_getexpss128_round ... this.
18568 (__builtin_ia32_getexpsd128_mask): Changed to ...
18569 __builtin_ia32_getexpsd128_round ... this.
18570 * config/i386/i386-builtin-types.def
18571 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
18572 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
18573 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
18574 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
18575 __builtin_ia32_getmantss_mask_round): New builtins.
18576 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
18577 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
18578 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
18579 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
18580 * config/i386/sse.md
18581 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
18582 avx512f_sgetexp<mode><mask_scalar_name>
18583 <round_saeonly_scalar_name> ... this.
18584 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
18585 %0, %1, %2<round_saeonly_op3>}): Changed to ...
18586 vgetexp<ssescalarmodesuffix>
18587 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18588 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
18589 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
18590 avx512f_vgetmant<mode><mask_scalar_name>
18591 <round_saeonly_scalar_name> ... this.
18592 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
18593 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
18594 vgetmant<ssescalarmodesuffix>
18595 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
18596 %0<mask_scalar_operand4>, %1, %2
18597 <round_saeonly_scalar_mask_op4>, %3} ... this.
18598 * config/i386/subst.md (mask_scalar_operand4,
18599 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
18600 round_saeonly_scalar_nimm_predicate): New subst attributes.
18602 2017-07-06 Julia Koval <julia.koval@intel.com>
18604 * config/i386/i386.c (ix86_erase_embedded_rounding):
18605 Remove code for old rounding pattern.
18607 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
18609 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
18611 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
18613 * doc/sourcebuild.texi (Test Directives, Variants of
18614 dg-require-support): Add documentation for dg-require-stack-check.
18616 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
18618 * config/i386/subst.md (mask_scalar, round_scalar,
18619 round_saeonly_scalar): New meta-templates.
18620 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
18621 round_scalar_mask_operand3, round_scalar_mask_op3,
18622 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
18623 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
18624 round_saeonly_scalar_constraint,
18625 round_saeonly_scalar_prefix): New subst attribute.
18626 * config/i386/sse.md
18627 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
18628 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
18629 <round_scalar_name> ... this.
18630 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
18631 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
18632 <round_scalar_name> ... this.
18633 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
18634 <sse>_vm<code><mode>3<mask_scalar_name>
18635 <round_saeonly_scalar_name> ... this.
18636 (v<plusminus_mnemonic><ssescalarmodesuffix>
18637 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
18638 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
18639 v<plusminus_mnemonic><ssescalarmodesuffix>
18640 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18641 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
18642 (v<multdiv_mnemonic><ssescalarmodesuffix>
18643 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
18644 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
18645 v<multdiv_mnemonic><ssescalarmodesuffix>
18646 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18647 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
18648 (v<maxmin_float><ssescalarmodesuffix>
18649 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
18650 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
18651 v<maxmin_float><ssescalarmodesuffix>
18652 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18653 %0<mask_scalar_operand3>, %1, %<iptr>2
18654 <round_saeonly_scalar_mask_op3>} ... this.
18656 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
18658 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
18659 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
18661 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
18662 Alan Hayward <alan.hayward@arm.com>
18663 David Sherwood <david.sherwood@arm.com>
18665 * combine.c (simplify_if_then_else): Remove "enum" before
18667 * compare-elim.c (can_eliminate_compare): Likewise.
18668 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
18670 (aarch64_lookup_simd_builtin_type): Likewise.
18671 (aarch64_simd_builtin_type): Likewise.
18672 (aarch64_init_simd_builtin_types): Likewise.
18673 (aarch64_simd_expand_args): Likewise.
18674 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
18676 (aarch64_reverse_mask): Likewise.
18677 (aarch64_simd_emit_reg_reg_move): Likewise.
18678 (aarch64_gen_adjusted_ldpstp): Likewise.
18679 (aarch64_ccmp_mode_to_code): Likewise.
18680 (aarch64_operands_ok_for_ldpstp): Likewise.
18681 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
18682 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
18684 (aarch64_min_divisions_for_recip_mul): Likewise.
18685 (aarch64_reassociation_width): Likewise.
18686 (aarch64_get_condition_code_1): Likewise.
18687 (aarch64_simd_emit_reg_reg_move): Likewise.
18688 (aarch64_simd_attr_length_rglist): Likewise.
18689 (aarch64_reverse_mask): Likewise.
18690 (aarch64_operands_ok_for_ldpstp): Likewise.
18691 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
18692 (aarch64_gen_adjusted_ldpstp): Likewise.
18693 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
18695 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
18696 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
18697 (arm_lookup_simd_builtin_type): Likewise.
18698 (arm_simd_builtin_type): Likewise.
18699 (arm_init_simd_builtin_types): Likewise.
18700 (arm_expand_builtin_args): Likewise.
18701 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
18702 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
18703 (ft32_setup_incoming_varargs): Likewise.
18704 (ft32_function_arg): Likewise.
18705 (ft32_function_arg_advance): Likewise.
18706 (ft32_pass_by_reference): Likewise.
18707 (ft32_arg_partial_bytes): Likewise.
18708 (ft32_valid_pointer_mode): Likewise.
18709 (ft32_addr_space_pointer_mode): Likewise.
18710 (ft32_addr_space_legitimate_address_p): Likewise.
18711 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
18713 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
18714 (ix86_emit_outlined_ms2sysv_restore): Likewise.
18715 (iamcu_alignment): Likewise.
18716 (canonicalize_vector_int_perm): Likewise.
18717 (ix86_noce_conversion_profitable_p): Likewise.
18718 (ix86_mpx_bound_mode): Likewise.
18719 (ix86_operands_ok_for_move_multiple): Likewise.
18720 * config/microblaze/microblaze-protos.h
18721 (microblaze_expand_conditional_branch_reg): Likewise.
18722 * config/microblaze/microblaze.c
18723 (microblaze_expand_conditional_branch_reg): Likewise.
18724 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
18726 (rs6000_reassociation_width): Likewise.
18727 (rs6000_invalid_binary_op): Likewise.
18728 (fusion_p9_p): Likewise.
18729 (emit_fusion_p9_load): Likewise.
18730 (emit_fusion_p9_store): Likewise.
18731 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
18733 (riscv_hard_regno_mode_ok_p): Likewise.
18734 (riscv_address_insns): Likewise.
18735 (riscv_split_symbol): Likewise.
18736 (riscv_legitimize_move): Likewise.
18737 (riscv_function_value): Likewise.
18738 (riscv_hard_regno_nregs): Likewise.
18739 (riscv_expand_builtin): Likewise.
18740 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
18741 (riscv_build_integer): Likewise.
18742 (riscv_split_integer): Likewise.
18743 (riscv_legitimate_constant_p): Likewise.
18744 (riscv_cannot_force_const_mem): Likewise.
18745 (riscv_regno_mode_ok_for_base_p): Likewise.
18746 (riscv_valid_base_register_p): Likewise.
18747 (riscv_valid_offset_p): Likewise.
18748 (riscv_valid_lo_sum_p): Likewise.
18749 (riscv_classify_address): Likewise.
18750 (riscv_legitimate_address_p): Likewise.
18751 (riscv_address_insns): Likewise.
18752 (riscv_load_store_insns): Likewise.
18753 (riscv_force_binary): Likewise.
18754 (riscv_split_symbol): Likewise.
18755 (riscv_force_address): Likewise.
18756 (riscv_legitimize_address): Likewise.
18757 (riscv_move_integer): Likewise.
18758 (riscv_legitimize_const_move): Likewise.
18759 (riscv_legitimize_move): Likewise.
18760 (riscv_address_cost): Likewise.
18761 (riscv_subword): Likewise.
18762 (riscv_output_move): Likewise.
18763 (riscv_canonicalize_int_order_test): Likewise.
18764 (riscv_emit_int_order_test): Likewise.
18765 (riscv_function_arg_boundary): Likewise.
18766 (riscv_pass_mode_in_fpr_p): Likewise.
18767 (riscv_pass_fpr_single): Likewise.
18768 (riscv_pass_fpr_pair): Likewise.
18769 (riscv_get_arg_info): Likewise.
18770 (riscv_function_arg): Likewise.
18771 (riscv_function_arg_advance): Likewise.
18772 (riscv_arg_partial_bytes): Likewise.
18773 (riscv_function_value): Likewise.
18774 (riscv_pass_by_reference): Likewise.
18775 (riscv_setup_incoming_varargs): Likewise.
18776 (riscv_print_operand): Likewise.
18777 (riscv_elf_select_rtx_section): Likewise.
18778 (riscv_save_restore_reg): Likewise.
18779 (riscv_for_each_saved_reg): Likewise.
18780 (riscv_register_move_cost): Likewise.
18781 (riscv_hard_regno_mode_ok_p): Likewise.
18782 (riscv_hard_regno_nregs): Likewise.
18783 (riscv_class_max_nregs): Likewise.
18784 (riscv_memory_move_cost): Likewise.
18785 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
18786 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
18787 (rl78_addr_space_address_mode): Likewise.
18788 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18790 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
18791 (rs6000_reassociation_width): Likewise.
18792 (rs6000_invalid_binary_op): Likewise.
18793 (fusion_p9_p): Likewise.
18794 (emit_fusion_p9_load): Likewise.
18795 (emit_fusion_p9_store): Likewise.
18796 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
18797 (ok_for_simple_move_operands): Likewise.
18798 (ok_for_simple_move_strict_operands): Likewise.
18799 (ok_for_simple_arith_logic_operands): Likewise.
18800 (visium_legitimize_reload_address): Likewise.
18801 (visium_select_cc_mode): Likewise.
18802 (output_cbranch): Likewise.
18803 (visium_split_double_move): Likewise.
18804 (visium_expand_copysign): Likewise.
18805 (visium_expand_int_cstore): Likewise.
18806 (visium_expand_fp_cstore): Likewise.
18807 * config/visium/visium.c (visium_pass_by_reference): Likewise.
18808 (visium_function_arg): Likewise.
18809 (visium_function_arg_advance): Likewise.
18810 (visium_libcall_value): Likewise.
18811 (visium_setup_incoming_varargs): Likewise.
18812 (visium_legitimate_constant_p): Likewise.
18813 (visium_legitimate_address_p): Likewise.
18814 (visium_legitimize_address): Likewise.
18815 (visium_secondary_reload): Likewise.
18816 (visium_register_move_cost): Likewise.
18817 (visium_memory_move_cost): Likewise.
18818 (prepare_move_operands): Likewise.
18819 (ok_for_simple_move_operands): Likewise.
18820 (ok_for_simple_move_strict_operands): Likewise.
18821 (ok_for_simple_arith_logic_operands): Likewise.
18822 (visium_function_value_1): Likewise.
18823 (rtx_ok_for_offset_p): Likewise.
18824 (visium_legitimize_reload_address): Likewise.
18825 (visium_split_double_move): Likewise.
18826 (visium_expand_copysign): Likewise.
18827 (visium_expand_int_cstore): Likewise.
18828 (visium_expand_fp_cstore): Likewise.
18829 (visium_split_cstore): Likewise.
18830 (visium_select_cc_mode): Likewise.
18831 (visium_split_cbranch): Likewise.
18832 (output_cbranch): Likewise.
18833 (visium_print_operand_address): Likewise.
18834 * expmed.c (flip_storage_order): Likewise.
18835 * expmed.h (emit_cstore): Likewise.
18836 (flip_storage_order): Likewise.
18837 * genrecog.c (validate_pattern): Likewise.
18838 * hsa-gen.c (gen_hsa_addr): Likewise.
18839 * internal-fn.c (expand_arith_overflow): Likewise.
18840 * ira-color.c (allocno_copy_cost_saving): Likewise.
18841 * lra-assigns.c (find_hard_regno_for_1): Likewise.
18842 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
18843 (process_invariant_for_inheritance): Likewise.
18844 * lra-eliminations.c (move_plus_up): Likewise.
18845 * omp-low.c (lower_oacc_reductions): Likewise.
18846 * simplify-rtx.c (simplify_subreg): Likewise.
18847 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
18848 (TARGET_CHKP_BOUND_MODE): Likewise..
18849 * targhooks.c (default_chkp_bound_mode): Likewise.
18850 (default_setup_incoming_vararg_bounds): Likewise.
18851 * targhooks.h (default_chkp_bound_mode): Likewise.
18852 (default_setup_incoming_vararg_bounds): Likewise.
18853 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
18854 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
18855 (have_whole_vector_shift): Likewise.
18856 * tree-vect-stmts.c (vectorizable_load): Likewise.
18857 * doc/tm.texi: Regenerate.
18859 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
18861 Graceful degrade if Binutils PR21472 is not available.
18864 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
18865 .rodata in flash test fails.
18866 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
18867 * confgure: Regenerate.
18868 * config.in: Regenerate.
18869 * config/avr/avr.c (avr_asm_named_section)
18870 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
18871 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
18872 (avr_asm_init_sections): Same.
18874 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18876 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
18877 (fma<VH:mode>4_intrinsic): Likewise.
18878 (*fmsub<VCVTF:mode>4): Likewise.
18879 (*fmsub<VH:mode>4_intrinsic): Likewise.
18881 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
18884 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
18885 Don't depend on "optimize > 0".
18886 (out_movhi_r_mr, out_movqi_mr_r): Same.
18887 (out_movhi_mr_r, out_movqi_r_mr): Same.
18888 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
18889 io_address_operand on "optimize > 0".
18891 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18893 * tree-loop-distribution.c: Add general explanantion on the pass.
18894 (generate_loops_for_partition): Mark distributed loop.
18895 (pg_add_dependence_edges): New parameter. Handle alias data
18896 dependence specially and record it in the parameter if asked.
18897 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
18898 (init_partition_graph_vertices, add_partition_graph_edge): New.
18899 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
18900 (free_partition_graph_vdata, build_partition_graph): New.
18901 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
18902 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
18903 (data_ref_segment_size, latch_dominated_by_data_ref): New.
18904 (compute_alias_check_pairs, version_loop_by_alias_check): New.
18905 (version_for_distribution_p, finalize_partitions): New.
18906 (distribute_loop): Handle alias data dependence specially. Factor
18907 out loop fusion code as functions and call these functions.
18909 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18911 * tree-loop-distribution.c (classify_partition): New parameter and
18912 better handle reduction statement.
18913 (rdg_build_partitions): Revise comment.
18914 (distribute_loop): Compute statements in all partitions and pass it
18915 to classify_partition.
18917 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18919 * tree-loop-distribution.c (enum partition_type): New.
18920 (struct partition): New field type.
18921 (partition_merge_into): Add parameter. Update partition type.
18922 (data_dep_in_cycle_p, update_type_for_merge): New functions.
18923 (build_rdg_partition_for_vertex): Compute partition type.
18924 (rdg_build_partitions): Dump partition type.
18925 (distribute_loop): Update calls to partition_merge_into.
18927 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18929 * tree-loop-distribution.c (struct ddr_hasher): New.
18930 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
18932 (classify_partition): Call get_data_dependence.
18933 (pg_add_dependence_edges): Ditto.
18934 (distribute_loop): Release data dependence hash table.
18936 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18938 * tree-loop-distribution.c (ref_base_address): Delete.
18939 (similar_memory_accesses): Rename ...
18940 (share_memory_accesses): ... to this. Check if partitions access
18941 the same memory reference.
18942 (distribute_loop): Call share_memory_accesses.
18944 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18946 * tree-loop-distribution.c (struct partition): New field recording
18947 its data reference.
18948 (partition_alloc, partition_free): Init and release data refs.
18949 (partition_merge_into): Merge data refs.
18950 (build_rdg_partition_for_vertex): Collect data refs for partition.
18951 (pg_add_dependence_edges): Change parameters from vector to bitmap.
18953 (distribute_loop): Remve data refs from vertice data of partition
18956 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18958 * tree-loop-distribution.c (params.h): Include header file.
18959 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
18960 (datarefs_vec): New global var.
18961 (create_rdg_vertices): Use datarefs_vec directly.
18962 (free_rdg): Don't free data references.
18963 (build_rdg): Update use. Don't free data references.
18964 (distribute_loop): Compute global variable for data references.
18965 Bail out if there are too many data references.
18967 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18969 * tree-loop-distribution.c (loop_nest): New global var.
18970 (build_rdg): Use loop directly, rather than loop nest.
18971 (pg_add_dependence_edges): Remove loop nest parameter. Use global
18973 (distribute_loop): Compute global variable loop nest. Update use.
18975 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18977 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
18978 (partition_merge_into): New parameter. Dump reason for fusion.
18979 (distribute_loop): Update use of partition_merge_into.
18981 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18983 * tree-loop-distribution.c (bb_top_order_index): New.
18984 (bb_top_order_index_size, bb_top_order_cmp): New.
18985 (stmts_from_loop): Use topological order.
18986 (pass_loop_distribution::execute): Compute and release topological
18987 order for basic blocks.
18989 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18991 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
18994 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18996 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
18997 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
18998 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
18999 * internal-fn.def (LOOP_DIST_ALIAS): New.
19000 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
19001 (fold_loop_internal_call): ... this.
19002 (vect_loop_dist_alias_call): New function.
19003 (set_uid_loop_bbs): Call fold_loop_internal_call.
19004 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
19007 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
19010 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
19011 Require dead FLAGS_REG at the beginning of a peephole.
19013 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
19016 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
19017 arguments in the call to __builtin_ia32_sbb_u32.
19018 (_subborrow_u64): Swap _X and _Y arguments in the call to
19019 __builtin_ia32_sbb_u64.
19021 2017-07-04 Jakub Jelinek <jakub@redhat.com>
19024 * tree-vrp.c (compare_assert_loc): Turn into a function template
19025 with stable template parameter. Only test if a->e is NULL,
19026 !a->e == !b->e has been verified already. Use e == NULL or
19027 e != NULL instead of e or ! e tests. If stable is true, don't use
19028 iterative_hash_expr, on the other side allow a or b or both NULL
19029 and sort the NULLs last.
19030 (process_assert_insertions): Sort using compare_assert_loc<false>
19031 instead of compare_assert_loc, later sort using
19032 compare_assert_loc<true> before calling process_assert_insertions_for
19033 in a loop. Use break instead of continue once seen NULL pointer.
19035 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
19037 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
19038 Cortex-R7 and Cortex-R8 processors.
19040 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
19042 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
19043 uninitialized while src is not.
19045 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
19047 * common/config/arm/arm-common.c: Adjust include path for
19049 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
19050 (arm-cpu.h): Create in build directory. Adjust dependency rules.
19051 (arm-cpu-data.h): Likewise.
19052 (arm-cpu-cdata.h): Likewise.
19053 * config/arm/arm-cpu.h: Delete.
19054 * config/arm/arm-cpu-cdata.h: Delete.
19055 * config/arm/arm-cpu-data.h: Delete.
19057 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
19059 * config/arm/arm-cpus.in (cortex-a55): New.
19060 (cortex-a75): Likewise.
19061 (cortex-a75.cortex-a55): Likewise.
19062 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
19064 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
19065 * config/arm/arm-cpu-cdata.h: Regenerate.
19066 * config/arm/arm-cpu-data.h: Regenerate.
19067 * config/arm/arm-cpu.h: Regenerate.
19068 * config/arm/arm-tables.opt: Regenerate.
19069 * config/arm/arm-tune.md: Regenerate.
19071 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
19073 * haifa-sched.c (sched_create_recovery_edges): Update profile.
19075 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
19077 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
19080 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
19082 PR tree-optimization/81292
19083 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
19084 full_string_p, also call adjust_related_strinfos if the adjustment
19085 is simple, otherwise invalidate related strinfos.
19087 2017-07-04 Martin Liska <mliska@suse.cz>
19090 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
19091 newly created variable as DECL_IGNORED_P.
19093 2017-07-04 Martin Liska <mliska@suse.cz>
19096 * ipa-inline.c (inline_small_functions):
19097 Use xstrdup_for_dump.
19099 2017-07-04 Tom de Vries <tom@codesourcery.com>
19101 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
19103 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
19106 * config/darwin.c (darwin_function_switched_text_sections):
19109 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
19111 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
19113 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
19115 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
19117 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19119 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
19120 min_profitable_iters, and th as inclusive lower bounds.
19121 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
19122 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
19123 for min_profitable_iters and min_profitable_estimate.
19124 (vect_transform_loop): Treat th as an inclusive lower bound.
19125 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
19127 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
19130 * config/darwin.c (darwin_function_switched_text_sections):
19131 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
19132 in two pieces, and suppress the use of buf.
19134 2017-07-03 Nathan Sidwell <nathan@acm.org>
19136 * hash-table.h (hash_table_mod1): Fix indentation.
19138 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19140 PR middle-end/81290
19141 * predict.c (force_edge_cold): Be more careful about propagation
19143 * profile-count.h (profile_probability::guessed,
19144 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
19146 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
19148 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
19150 * doc/invoke.texi (rcpc architecture extension): Document it.
19152 2017-07-03 Richard Biener <rguenther@suse.de>
19154 PR tree-optimization/60510
19155 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
19156 the scalar reduction PHI and use it.
19157 (vectorizable_reduction): Properly guard the single_defuse_cycle
19158 path for non-SLP reduction chains where we cannot use it.
19159 Rework reduc_def/index and vector type deduction. Rework
19160 vector operand gathering during reduction op code-gen.
19161 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
19162 chains dissolve the chain and leave it to non-SLP reduction
19165 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19167 * tree-data-ref.h (dr_alignment): Declare.
19168 * tree-data-ref.c (dr_alignment): New function.
19169 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
19170 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
19172 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
19174 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19176 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
19177 and base_misalignment fields.
19178 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
19179 * tree-data-ref.c: Include builtins.h.
19180 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
19181 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
19182 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
19183 * tree-vect-data-refs.c: Include tree-cfg.h.
19184 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
19185 fields instead of calculating an alignment here.
19186 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
19187 innermost_loop_behavior fields.
19189 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19191 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
19193 (DR_STEP_ALIGNMENT): New macro.
19194 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
19195 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
19196 (create_data_ref): Print it.
19197 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
19198 to tell whether the step preserves vector (mis)alignment.
19199 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
19200 Move the check for an integer step and generalise to all INTEGER_CST.
19201 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
19202 Print the outer step alignment.
19204 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19206 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
19207 with offset_alignment.
19208 (DR_ALIGNED_TO): Delete.
19209 (DR_OFFSET_ALIGNMENT): New macro.
19210 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
19211 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
19212 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
19213 (create_data_ref): Likewise.
19214 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
19215 (vect_analyze_data_refs): Likewise.
19216 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
19217 creating dummy innermost behavior.
19219 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19221 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
19222 with a "innermost_loop_behavior *" and refeence tree.
19223 * tree-data-ref.c (dr_analyze_innermost): Likewise.
19224 (create_data_ref): Update call accordingly.
19225 * tree-predcom.c (find_looparound_phi): Likewise.
19227 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19229 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
19230 fields with dr_wrt_vec_loop.
19231 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
19232 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
19233 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
19234 (vect_dr_behavior): New function.
19235 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
19236 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
19237 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
19238 track whether the step preserves the misalignment.
19239 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
19240 Use vect_dr_behavior.
19241 (vect_setup_realignment): Update call accordingly.
19242 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
19243 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
19244 call to vect_create_addr_base_for_vector_ref.
19245 (vect_create_cond_for_align_checks): Likewise.
19246 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
19247 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
19248 (vect_recog_mask_conversion_pattern): Likewise.
19249 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
19250 (new_stmt_vec_info): Remove redundant zeroing.
19252 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
19254 * common/config/arm/arm-common.c (arm_be8_option): New function.
19255 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
19256 (ISA_ARMv6): Add isa_bit_be8.
19257 * config/arm/arm.h (arm_be8_option): Add prototype.
19258 (BE8_SPEC_FUNCTION): New define.
19259 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
19260 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
19261 (mlittle-endian): Similarly.
19262 (mbe8, mbe32): New options.
19263 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
19264 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
19266 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19268 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
19270 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19272 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
19273 (cleanup_tree_cfg_bb): Use it.
19274 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
19276 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
19278 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19281 * loop-doloop.c (add_test): Update profile.
19283 2017-07-03 Martin Liska <mliska@suse.cz>
19286 * sanopt.c (rewrite_usage_of_param): New function.
19287 (sanitize_rewrite_addressable_params): Likewise.
19288 (pass_sanopt::execute): Call rewrite_usage_of_param.
19290 2017-07-03 Richard Biener <rguenther@suse.de>
19292 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
19293 back to using VIEW_CONVERT_EXPR.
19295 2017-07-03 Martin Liska <mliska@suse.cz>
19298 * doc/extend.texi: Document when a resolver function is
19299 generated for target_clones.
19301 2017-07-03 Martin Liska <mliska@suse.cz>
19303 * asan.c (asan_emit_stack_protection): Unpoison just red zones
19304 and shadow memory of auto variables which are subject of
19305 use-after-scope sanitization.
19306 (asan_expand_mark_ifn): Add do set only when is_poison.
19308 2016-07-03 Richard Biener <rguenther@suse.de>
19310 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
19312 (vect_force_simple_reduction): Record reduction def -> phi mapping.
19313 (vectorizable_reduction): Perform reduction PHI creation when
19314 visiting a reduction PHI and adjust and simplify code generation
19315 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
19316 (vect_transform_loop): Visit reduction PHIs.
19317 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
19318 defs into the SLP tree.
19319 (vect_build_slp_tree): Reduction defs terminate the recursion.
19320 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
19322 (vect_get_vec_defs_for_stmt_copy): Export.
19323 (vect_get_vec_defs): Likewise.
19324 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
19326 (vect_get_vec_defs_for_stmt_copy): Declare.
19327 (vect_get_vec_defs): Likewise.
19329 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19331 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
19332 parameter with a "loop" parameter and use it instead of the
19333 loop containing DR_STMT. Don't check simple_iv when doing
19334 BB analysis. Describe the two analysis modes in the comment.
19336 2017-07-03 Tom de Vries <tom@codesourcery.com>
19338 PR tree-optimization/69468
19339 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
19340 (find_same_succ_bb): Handle ignore_edge_flags.
19342 2017-07-03 Tom de Vries <tom@codesourcery.com>
19344 PR tree-optimization/81192
19345 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
19347 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
19349 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
19351 2017-07-03 Tom de Vries <tom@codesourcery.com>
19353 PR tree-optimization/81192
19354 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
19355 BB_SAME_SUCC (bb) == NULL.
19357 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19359 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
19362 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19364 * dumpfile.c: Include profile-count.h
19365 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
19367 (insert_cond_bb): Update profile.
19368 * tree-cfg.h (insert_cond_bb): Update prototype.
19369 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
19370 * tree-dump.c: Do not include tree-cfg.
19372 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19374 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
19376 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19378 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
19381 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19383 * tree-complex.c (expand_complex_div_wide): update profile.
19385 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19386 Alan Hayward <alan.hayward@arm.com>
19387 David Sherwood <david.sherwood@arm.com>
19389 * Makefile.in (MACHMODE_H): Remove insn-modes.h
19390 (CORETYPES_H): New define.
19391 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
19392 (insn-modes-inline.h, s-modes-inline-h): New rules.
19393 (generated_files): Add insn-modes-inline.h.
19394 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
19395 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
19396 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
19397 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
19398 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
19399 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
19400 (build/gencodes.o, build/genconditions.o): Likewise.
19401 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
19402 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
19403 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
19404 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
19405 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
19406 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
19407 * coretypes.h: Include everything up to real.h for generators.
19408 Include insn-modes.h first. Include wide-int-print.h after
19409 wide-int.h. Include insn-modes-inline.h and then machmode.h.
19410 * machmode.h: Don't include insn-modes.h here.
19411 * function-tests.c: Remove includes of signop.h, machmode.h,
19412 double-int.h and wide-int.h.
19414 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
19416 * optc-save-gen.awk: Likewise.
19417 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
19418 * godump.c: Remove include of wide-int-print.h.
19419 * pretty-print.h: Likewise.
19420 * wide-int-print.cc: Likewise.
19421 * wide-int.cc: Likewise.
19422 * hash-map-tests.c: Remove include of signop.h.
19423 * hash-set-tests.c: Likewise.
19424 * rtl-tests.c: Likewise.
19425 * mkconfig.sh: Remove include of machmode.h.
19426 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
19428 (emit_insn_modes_inline_h): ...this new function. Emit the code
19429 into an insn-modes-inline.h header file, adding appropriate
19430 include guards and end comments.
19431 (emit_insn_modes_c_header): Remove include of machmode.h.
19432 (emit_min_insn_modes_c_header): Include coretypes.h rather than
19434 (main): Handle -i flag and call emit_insn_modes_inline_h when
19437 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19439 * tree-ssa-strlen.c (strinfo): Rename the length field to
19440 nonzero_chars. Add a full_string_p field.
19441 (compare_nonzero_chars, zero_length_string_p): New functions.
19442 (get_addr_stridx): Add an offset_out parameter.
19443 Use compare_nonzero_chars.
19444 (get_stridx): Update accordingly. Use compare_nonzero_chars.
19445 (new_strinfo): Update after above changes to strinfo.
19446 (set_endptr_and_length): Set full_string_p.
19447 (get_string_length): Update after above changes to strinfo.
19448 (unshare_strinfo): Update call to new_strinfo.
19449 (maybe_invalidate): Likewise.
19450 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
19451 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
19452 as a uhwi instead of an shwi. Update after above changes to
19453 strinfo and new_strinfo.
19454 (zero_length_string): Assert that chainsi contains full strings.
19455 Use zero_length_string_p. Update call to new_strinfo.
19456 (adjust_related_strinfos): Update after above changes to strinfo.
19457 Copy full_string_p from origsi.
19458 (adjust_last_stmt): Use zero_length_string_p.
19459 (handle_builtin_strlen): Update after above changes to strinfo and
19460 new_strinfo. Install the lhs as the string length if the previous
19461 entry didn't describe a full string.
19462 (handle_builtin_strchr): Update after above changes to strinfo
19464 (handle_builtin_strcpy): Likewise.
19465 (handle_builtin_strcat): Likewise.
19466 (handle_builtin_malloc): Likewise.
19467 (handle_pointer_plus): Likewise.
19468 (handle_builtin_memcpy): Likewise. Track nonzero characters
19469 that aren't necessarily followed by a nul terminator.
19470 (handle_char_store): Likewise.
19472 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19474 PR tree-optimization/80769
19475 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
19476 for malloc and calloc. Document the new invariant that all related
19477 strinfos have delayed lengths or none do.
19478 (verify_related_strinfos): Move earlier in file.
19479 (set_endptr_and_length): New function, split out from...
19480 (get_string_length): ...here. Also set the lengths of related
19482 (zero_length_string): Assert that chainsi has known (rather than
19484 (adjust_related_strinfos): Likewise.
19486 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19488 PR tree-optimization/81136
19489 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
19490 assert that two references with the same misalignment have the same
19491 compile-time misalignment if those compile-time misalignments
19494 2017-07-01 Andi Kleen <ak@linux.intel.com>
19496 * print-tree.c (print_node): Print all attributes.
19498 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19500 * cfg.c (scale_bbs_frequencies): New function.
19501 * cfg.h (scale_bbs_frequencies): Declare it.
19502 * cfgloopanal.c (single_likely_exit): Cleanup.
19503 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
19505 (scale_loop_profile): Likewise.
19506 (loop_version): Likewise.
19507 (create_empty_loop_on_edge): Update.
19508 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
19509 scale_loop_frequencies, scale_loop_profile, loopify,
19510 loop_version): Update prototypes.
19511 * modulo-sched.c (sms_schedule): Update.
19512 * predict.c (unlikely_executed_edge_p): Also check probability.
19513 (probably_never_executed_edge_p): Fix typo.
19514 * tree-if-conv.c (version_loop_for_if_conversion): Update.
19515 * tree-parloops.c (gen_parallel_loop): Update.
19516 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
19517 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
19518 * tree-ssa-loop-split.c (split_loop): Update.
19519 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
19520 * tree-vect-loop-manip.c (vect_do_peeling): Update.
19521 (vect_loop_versioning): Update.
19522 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
19524 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19526 * trans-mem.c (split_bb_make_tm_edge): Update profile.
19528 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19530 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
19531 to keep profile consistent.
19533 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19535 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
19536 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
19537 * profile-count.h (max_safe_multiplier): Make unsigned.
19538 (profile_count::guessed_zero): New.
19540 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19542 * bb-reorder.c (fix_up_crossing_landing_pad,
19543 fix_crossing_conditional_branches): Use make_single_succ_edge
19544 to keep profile consistent.
19546 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19548 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
19551 2017-07-01 Jakub Jelinek <jakub@redhat.com>
19554 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
19555 the right scopes, make sure cond_jump isn't preserved between multiple
19556 iterations. Search for fallthru edge whenever there are 3+ edges and
19557 use find_fallthru_edge for it.
19559 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19561 Patch by Alexander Monakov <amonakov@ispras.ru>
19562 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
19563 probabilities consistently.
19565 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19567 * pa.c (pa_expand_compare_and_swap_loop): Update call of
19568 emit_cmp_and_jump_insns.
19570 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19573 * tree-inline.c (expand_call_inline): Combine profile statuses.
19575 2017-06-30 Andrew Pinski <apinski@cavium.com>
19577 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
19578 fold_stmt returned true.
19580 2017-06-30 Nathan Sidwell <nathan@acm.org>
19582 * ggc.h (empty_string): Delete.
19583 * cfgexpand.c (expand_asm_stmt): Use plain "".
19584 * optabs.c (expand_asm_memory_barrier): Likewise.
19585 * stringpool.c (empty_string): Delete.
19586 (digit_vector, digit_string): Delete.
19587 (ggc_alloc_string): Use plain "", don't optimize single digit
19588 strings. Use ggc_alloc_atomic.
19590 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
19592 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
19593 comparison set and one other set, use the cost of the non-comparison
19596 2017-06-30 Nathan Sidwell <nathan@acm.org>
19598 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
19601 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
19603 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
19604 loops. Remove now unneeded calls to gimple_switch_set_label() that
19605 just set removed labels to NULL_TREE.
19607 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
19609 * tree-ssanames.c (set_range_info_raw): Abstract from ...
19610 (set_range_info): ...here. Only call set_range_info_raw if domain
19612 (set_nonzero_bits): Call set_range_info_raw.
19613 * tree-ssanames.h (set_range_info_raw): New.
19615 2017-06-30 Jakub Jelinek <jakub@redhat.com>
19618 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
19619 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
19620 of nonimmediate_operand and <store_mask_constraint> instead of m
19621 for the input operand. For V8FI iterator, always split if input
19622 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
19623 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
19624 <store_mask_predicate> instead of register_operand and
19625 <store_mask_constraint> instead of v for the input operand. Make
19626 sure both operands aren't MEMs for if not <mask_applied>.
19628 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
19630 * lto-wrapper.c (copy_file) Close both file descriptors before
19633 2017-06-30 Martin Liska <mliska@suse.cz>
19636 * multiple_target.c (create_dispatcher_calls): Make ifunc
19637 also for function that don't have calls or are not referenced.
19639 2017-06-30 Richard Biener <rguenther@suse.de>
19641 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
19642 analyze the first scalar stmt. Move vector type computation
19643 for the BB case here from ...
19644 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
19645 live operation processing in the SLP case properly.
19647 2017-06-30 Richard Biener <rguenther@suse.de>
19649 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
19651 2017-06-30 Martin Liska <mliska@suse.cz>
19654 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
19655 before BUILT_IN_UNWIND_RESUME when ASAN is used.
19657 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
19659 * doc/invoke.texi (AArch64): Add missing options and remove redundant
19662 2017-06-30 Richard Biener <rguenther@suse.de>
19664 PR tree-optimization/81249
19665 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
19666 condition reduction result to original scalar type.
19668 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19670 * profile-count.h (enum profile_quality): Fix typos and whitespace
19673 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19675 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
19676 type for branch probabilities.
19678 2017-06-29 Julian Brown <julian@codesourcery.com>
19679 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19681 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
19682 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
19683 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
19684 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
19686 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19688 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
19689 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
19690 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
19691 CC usage from generic code to here.
19692 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
19693 CC usage into the target macros.
19695 2017-06-29 Maya Rashish <coypu@sdf.org>
19697 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
19700 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19702 * arm/arm-builtins.c: Include profile-count.h
19703 * except.c (sjlj_emit_function_enter): Use
19704 profile_probability::unlikely.
19706 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19708 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
19709 and tocrel_offset be pointer args rather than implicitly using
19711 (legitimate_constant_pool_address_p, rs6000_emit_move,
19712 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
19713 tocrel_offset and use in toc_relative_expr_p call.
19714 (print_operand, print_operand_address): Use static tocrel_base_oac
19715 and tocrel_offset_oac.
19716 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
19719 2017-06-29 Maya Rashish <coypu@sdf.org>
19721 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
19723 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
19725 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
19726 objects, take into account only the alignment of 'op0' and 'mode1' if
19729 2017-06-29 Steve Ellcey <sellcey@cavium.com>
19731 * ccmp.c (ccmp_tree_comparison_p): New function.
19732 (ccmp_candidate_p): Update to use above function.
19733 (get_compare_parts): New function.
19734 (expand_ccmp_next): Update to use new functions.
19735 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
19737 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
19738 take mode as argument.
19739 * ccmp.h (expand_ccmp_expr): Add mode as argument.
19740 * expr.c (expand_expr_real_1): Pass mode as argument.
19742 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
19744 * combine.c (combine_instructions): Print insns to dump_file, together
19747 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19749 * asan.c (asan_emit_stack_protection): Update.
19750 (create_cond_insert_point): Update.
19751 * auto-profile.c (afdo_propagate_circuit): Update.
19752 * basic-block.h (struct edge_def): Turn probability to
19753 profile_probability.
19754 (EDGE_FREQUENCY): Update.
19755 * bb-reorder.c (find_traces_1_round): Update.
19756 (better_edge_p): Update.
19757 (sanitize_hot_paths): Update.
19758 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
19759 (make_single_succ_edge): Update.
19760 (check_bb_profile): Update.
19761 (dump_edge_info): Update.
19762 (update_bb_profile_for_threading): Update.
19763 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
19764 probabilitycount to 0.
19765 * cfgbuild.c (compute_outgoing_frequencies): Update.
19766 * cfgcleanup.c (try_forward_edges): Update.
19767 (outgoing_edges_match): Update.
19768 (try_crossjump_to_edge): Update.
19769 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
19770 (expand_gimple_tailcall): Update.
19771 (construct_init_block): Use make_single_succ_edge.
19772 (construct_exit_block): Use make_single_succ_edge.
19773 * cfghooks.c (verify_flow_info): Update.
19774 (redirect_edge_succ_nodup): Update.
19775 (split_edge): Update.
19776 (account_profile_record): Update.
19777 * cfgloopanal.c (single_likely_exit): Update.
19778 * cfgloopmanip.c (scale_loop_profile): Update.
19779 (set_zero_probability): Remove.
19780 (duplicate_loop_to_header_edge): Update.
19781 * cfgloopmanip.h (loop_version): Update prototype.
19782 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
19783 (force_nonfallthru_and_redirect): Update.
19784 (update_br_prob_note): Update.
19785 (rtl_verify_edges): Update.
19786 (purge_dead_edges): Update.
19787 (rtl_lv_add_condition_to_bb): Update.
19788 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
19789 * cgraphunit.c (init_lowered_empty_function): Update.
19790 (cgraph_node::expand_thunk): Update.
19791 * cilk-common.c: Include profile-count.h
19792 * dojump.c (inv): Remove.
19793 (jumpifnot): Update.
19794 (jumpifnot_1): Update.
19795 (do_jump_1): Update.
19797 (do_jump_by_parts_greater_rtx): Update.
19798 (do_compare_rtx_and_jump): Update.
19799 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
19800 do_jump_1. do_compare_rtx_and_jump): Update prototype.
19801 * dwarf2cfi.c: Include profile-count.h
19802 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
19803 (sjlj_emit_dispatch_table): Likewise.
19804 * explow.c: Include profile-count.h
19805 * expmed.c (emit_store_flag_force): Update.
19806 (do_cmp_and_jump): Update.
19807 * expr.c (compare_by_pieces_d::generate): Update.
19808 (compare_by_pieces_d::finish_mode): Update.
19809 (emit_block_move_via_loop): Update.
19810 (store_expr_with_bounds): Update.
19811 (store_constructor): Update.
19812 (expand_expr_real_2): Update.
19813 (expand_expr_real_1): Update.
19814 * expr.h (try_casesi, try_tablejump): Update prototypes.
19815 * gimple-pretty-print.c (dump_probability): Update.
19816 (dump_profile): New.
19817 (dump_gimple_label): Update.
19818 (dump_gimple_bb_header): Update.
19819 * graph.c (draw_cfg_node_succ_edges): Update.
19820 * hsa-gen.c (convert_switch_statements): Update.
19821 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
19822 (find_if_case_1): Update.
19823 (find_if_case_2): Update.
19824 * internal-fn.c (expand_arith_overflow_result_store): Update.
19825 (expand_addsub_overflow): Update.
19826 (expand_neg_overflow): Update.
19827 (expand_mul_overflow): Update.
19828 (expand_vector_ubsan_overflow): Update.
19829 * ipa-cp.c (good_cloning_opportunity_p): Update.
19830 * ipa-split.c (split_function): Use make_single_succ_edge.
19831 * ipa-utils.c (ipa_merge_profiles): Update.
19832 * loop-doloop.c (add_test): Update.
19833 (doloop_modify): Update.
19834 * loop-unroll.c (compare_and_jump_seq): Update.
19835 (unroll_loop_runtime_iterations): Update.
19836 * lra-constraints.c (lra_inheritance): Update.
19837 * lto-streamer-in.c (input_cfg): Update.
19838 * lto-streamer-out.c (output_cfg): Update.
19839 * mcf.c (adjust_cfg_counts): Update.
19840 * modulo-sched.c (sms_schedule): Update.
19841 * omp-expand.c (expand_omp_for_init_counts): Update.
19842 (extract_omp_for_update_vars): Update.
19843 (expand_omp_ordered_sink): Update.
19844 (expand_omp_for_ordered_loops): Update.
19845 (expand_omp_for_generic): Update.
19846 (expand_omp_for_static_nochunk): Update.
19847 (expand_omp_for_static_chunk): Update.
19848 (expand_cilk_for): Update.
19849 (expand_omp_simd): Update.
19850 (expand_omp_taskloop_for_outer): Update.
19851 (expand_omp_taskloop_for_inner): Update.
19852 * omp-simd-clone.c (simd_clone_adjust): Update.
19853 * optabs.c (expand_doubleword_shift): Update.
19854 (expand_abs): Update.
19855 (emit_cmp_and_jump_insn_1): Update.
19856 (expand_compare_and_swap_loop): Update.
19857 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
19858 * predict.c (predictable_edge_p): Update.
19859 (edge_probability_reliable_p): Update.
19860 (set_even_probabilities): Update.
19861 (combine_predictions_for_insn): Update.
19862 (combine_predictions_for_bb): Update.
19863 (propagate_freq): Update.
19864 (estimate_bb_frequencies): Update.
19865 (force_edge_cold): Update.
19866 * profile-count.c (profile_count::dump): Add missing space into dump.
19867 (profile_count::debug): Add newline.
19868 (profile_count::differs_from_p): Explicitly convert to unsigned.
19869 (profile_count::stream_in): Update.
19870 (profile_probability::dump): New member function.
19871 (profile_probability::debug): New member function.
19872 (profile_probability::differs_from_p): New member function.
19873 (profile_probability::differs_lot_from_p): New member function.
19874 (profile_probability::stream_in): New member function.
19875 (profile_probability::stream_out): New member function.
19876 * profile-count.h (profile_count_quality): Rename to ...
19877 (profile_quality): ... this one.
19878 (profile_probability): New.
19879 (profile_count): Update.
19880 * profile.c (compute_branch_probabilities): Update.
19881 * recog.c (peep2_attempt): Update.
19882 * sched-ebb.c (schedule_ebbs): Update.
19883 * sched-rgn.c (find_single_block_region): Update.
19884 (compute_dom_prob_ps): Update.
19885 (schedule_region): Update.
19886 * sel-sched-ir.c (compute_succs_info): Update.
19887 * stmt.c (struct case_node): Update.
19888 (do_jump_if_equal): Update.
19889 (get_outgoing_edge_probs): Update.
19890 (conditional_probability): Update.
19891 (emit_case_dispatch_table): Update.
19892 (expand_case): Update.
19893 (expand_sjlj_dispatch_table): Update.
19894 (emit_case_nodes): Update.
19895 * targhooks.c: Update.
19896 * tracer.c (better_p): Update.
19897 (find_best_successor): Update.
19898 * trans-mem.c (expand_transaction): Update.
19899 * tree-call-cdce.c: Update.
19900 * tree-cfg.c (gimple_split_edge): Upate.
19901 (move_sese_region_to_fn): Upate.
19902 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
19903 * tree-eh.c (lower_resx): Upate.
19904 (cleanup_empty_eh_move_lp): Upate.
19905 * tree-if-conv.c (version_loop_for_if_conversion): Update.
19906 * tree-inline.c (copy_edges_for_bb): Update.
19907 (copy_cfg_body): Update.
19908 * tree-parloops.c (gen_parallel_loop): Update.
19909 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
19910 (gimple_gen_time_profiler): Update.
19911 * tree-ssa-dce.c (remove_dead_stmt): Update.
19912 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
19913 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
19914 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
19915 (unloop_loops): Update.
19916 (try_peel_loop): Update.
19917 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
19918 * tree-ssa-loop-split.c (connect_loops): Update.
19919 (split_loop): Update.
19920 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
19921 (hoist_guard): Update.
19922 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
19923 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
19924 (value_replacement): Update.
19925 * tree-ssa-reassoc.c (branch_fixup): Update.
19926 * tree-ssa-tail-merge.c (replace_block_by): Update.
19927 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
19928 (create_edge_and_update_destination_phis): Update.
19929 (compute_path_counts): Update.
19930 (recompute_probabilities): Update.
19931 (update_joiner_offpath_counts): Update.
19932 (freqs_to_counts_path): Update.
19933 (duplicate_thread_path): Update.
19934 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
19935 (struct switch_conv_info): Update.
19936 (gen_inbound_check): Update.
19937 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
19938 (vect_do_peeling): Update.
19939 (vect_loop_versioning): Update.
19940 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
19941 (optimize_mask_stores): Update.
19942 * ubsan.c (ubsan_expand_null_ifn): Update.
19943 * value-prof.c (gimple_divmod_fixed_value): Update.
19944 (gimple_divmod_fixed_value_transform): Update.
19945 (gimple_mod_pow2): Update.
19946 (gimple_mod_pow2_value_transform): Update.
19947 (gimple_mod_subtract): Update.
19948 (gimple_mod_subtract_transform): Update.
19949 (gimple_ic): Update.
19950 (gimple_stringop_fixed_value): Update.
19951 (gimple_stringops_transform): Update.
19952 * value-prof.h: Update.
19954 2017-06-29 Carl Love <cel@us.ibm.com>
19956 * config/rs6000/rs6000-c.c: Add support for built-in functions
19957 vector signed int vec_signed (vector float);
19958 vector signed long long vec_signed (vector double);
19959 vector signed int vec_signed2 (vector double, vector double);
19960 vector signed int vec_signede (vector double);
19961 vector signed int vec_signedo (vector double);
19962 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
19963 instruction generator.
19964 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
19965 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
19966 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
19968 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
19969 vunsignede_v2df): Add define_expands.
19970 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
19971 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
19972 VEC_UNSIGNEDO): Add definitions.
19973 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
19974 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
19975 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
19976 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
19977 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
19978 * config/rs6000/altivec.h (vec_signed, vec_signed2,
19979 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
19980 vec_unsignede, vec_unsignedo): Add builtin defines.
19981 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
19983 * doc/extend.texi: Update the built-in documentation file for the
19984 new built-in functions.
19986 2017-06-29 Richard Biener <rguenther@suse.de>
19988 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
19989 reduction chains to LOOP_VINFO_REDUCTIONS.
19990 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
19991 SLP reductions after processing reduction chains.
19993 2017-06-29 Nathan Sidwell <nathan@acm.org>
19995 * builtins.c (fold_builtin_FUNCTION): Use
19996 lang_hooks.decl_printable_name.
19998 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
20000 PR middle-end/81194
20001 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
20002 with only one label.
20003 * stmt.c (expand_case): Assert NCASES is greater than one.
20005 2017-06-29 Richard Biener <rguenther@suse.de>
20007 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
20009 (group_case_labels): Likewise.
20010 (find_taken_edge): Push sanity checking on val to workers...
20011 (find_taken_edge_cond_expr): ... here
20012 (find_taken_edge_switch_expr): ... and here, handle cases
20013 with just a default label.
20014 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
20015 (group_case_labels): Likewise.
20016 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
20017 group_case_labels does anything cleanup the CFG again.
20019 2017-06-29 Bin Cheng <bin.cheng@arm.com>
20021 PR tree-optimization/81196
20022 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
20023 exit condition comparing two IVs.
20025 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
20027 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
20028 profile to the dummy entry at the end of the list of architectures.
20029 * config/arm/arm-cpu-cdata.h: Regenerated.
20031 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20032 Michael Collison <michael.collison@arm.com>
20035 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
20037 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
20038 (*aarch64_reg_<mode>3_minus_mask): New pattern.
20039 (*aarch64_<optab>_reg_di3_mask2): New pattern.
20040 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
20041 of shift when the shift amount is masked with constant equal to
20042 the size of the mode.
20043 * config/aarch64/predicates.md (subreg_lowpart_operator): New
20046 2017-06-29 Martin Liska <mliska@suse.cz>
20048 * config/i386/i386.opt: Change range from [1,5] to [0,5].
20050 2017-06-29 Yury Gribov <tetra2005@gmail.com>
20053 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
20056 (edge_growth_cache_entry::edge_growth_cache_entry): New
20058 (reset_edge_growth_cache): Update to use constructor.
20060 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20062 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
20063 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
20064 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
20066 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
20068 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
20069 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
20071 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
20073 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
20074 (*-linux-uclibc*): Add t-uclibc tmake_file.
20075 * config/t-musl: New.
20076 * config/t-uclibc: New.
20078 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
20080 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
20082 (gen_comm_data): Emit architectural setting of arch_prof.
20083 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
20085 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
20086 (armv8-m.base, armv8-m.main): Likewise.
20087 * arm-protos.h (arm_build_target): Add profile field.
20088 (arch_option): Likewise.
20089 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
20091 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
20092 arm_active_target.profile.
20094 2017-06-28 Richard Biener <rguenther@suse.de>
20096 PR middle-end/81227
20097 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
20098 TYPE_OVERFLOW_WRAPS.
20099 * match.pd (negate_expr_p): Likewise.
20100 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
20101 fold_build2, not fold_binary.
20103 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
20105 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
20106 Convert memory address to Pmode.
20107 (aarch64_print_operand): Assert MEM operands are always Pmode.
20109 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
20112 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
20113 Remove redundant if.
20114 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
20115 * config/arm/aarch-common-protos.h
20116 (aarch_forward_to_shift_is_not_shifted_re): Remove.
20117 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
20119 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
20122 * multiple_target.c (create_dispatcher_calls): Set the default
20123 clone to be static, not public.
20125 2017-06-28 Richard Biener <rguenther@suse.de>
20127 * tree-vect-loop.c (vectorizable_reduction): Move special
20128 cond reduction IV var creation ...
20129 (vect_create_epilog_for_reduction): ... here. Remove induction_index
20130 parameter. Use STMT_VINFO_VECTYPE.
20131 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
20134 2017-06-28 Martin Liska <mliska@suse.cz>
20137 * ipa-visibility.c (non_local_p): Handle visibility.
20139 2017-06-28 Martin Liska <mliska@suse.cz>
20142 * common.opt: Add IntegerRange to various options.
20143 * opt-functions.awk (integer_range_info): New function.
20144 * optc-gen.awk: Add integer_range_info to cl_options struct.
20145 * opts-common.c (decode_cmdline_option): Handle
20146 CL_ERR_INT_RANGE_ARG.
20147 (cmdline_handle_error): Likewise.
20148 * opts.c (print_filtered_help): Show valid interval in
20149 when --help is provided.
20150 * opts.h (struct cl_option): Add range_min and range_max fields.
20151 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
20153 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
20155 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
20156 (x * C EQ/NE y * C): New transformation.
20158 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
20160 * genmultilib (combination_space): Accept '+' in option names.
20162 2017-06-28 Martin Liska <mliska@suse.cz>
20165 * asan.c (instrument_derefs): Bail out inner references
20166 that are hard register variables.
20168 2017-06-28 Jakub Jelinek <jakub@redhat.com>
20171 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
20172 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
20174 2017-06-28 Richard Biener <rguenther@suse.de>
20176 * tree-vectorizer.h (vect_get_vec_defs): Remove.
20177 (vect_get_slp_defs): Adjust.
20178 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
20180 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
20182 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
20183 get_initial_defs_for_reduction instead of vect_get_vec_defs.
20184 (vectorizable_reduction): Adjust.
20185 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
20187 (vect_get_slp_defs): Likewise.
20188 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
20189 (vectorizable_bswap): Adjust.
20190 (vectorizable_call): Likewise.
20191 (vectorizable_conversion): Likewise.
20192 (vectorizable_assignment): Likewise.
20193 (vectorizable_shift): Likewise.
20194 (vectorizable_operation): Likewise.
20195 (vectorizable_store): Likewise.
20196 (vectorizable_condition): Likewise.
20197 (vectorizable_comparison): Likewise.
20199 2017-06-28 Michael Collison <michael.collison@arm.com>
20202 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
20204 (arm_gen_movmemqi): Removed unused variable 'i'.
20205 Convert 'for' loop into 'while' loop.
20206 (arm_expand_prologue): Remove last unnecessary set of insn.
20207 (thumb_pop): Remove unused variable 'pushed_words'.
20208 (thumb_exit): Remove last unnecessary set of regs_to_pop.
20210 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20212 * config/s390/predicates.md: Use s390_rel_address_ok_p.
20213 * config/s390/s390-protos.h: Add prototype of
20214 s390_rel_address_ok_p.
20215 * config/s390/s390.c (s390_got_symbol): New function.
20216 (s390_rel_address_ok_p): New function.
20217 (legitimize_pic_address): Use s390_rel_address_ok_p.
20218 (s390_load_got): Use s390_got_symbol.
20219 (s390_option_override): Issue error if
20220 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
20221 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
20223 * config/s390/s390.opt: New option mpic-data-is-text-relative.
20225 2017-06-27 Andrew Pinski <apinski@cavium.com>
20227 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
20228 (X * copysign (1.0, X)): New pattern.
20229 (X * copysign (1.0, -X)): New pattern.
20230 (copysign (-1.0, CST)): New pattern.
20232 2017-06-27 Joseph Myers <joseph@codesourcery.com>
20234 * genmultilib (combination_space): Remove variable.
20235 Validate reuse rules against regular expression for any sequence
20236 of multilib options in any order.
20238 2017-06-27 Michael Collison <michael.collison@arm.com>
20240 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
20241 call aarch64_split_simd_combine.
20242 * (aarch64_combine_internal<mode>): Delete pattern.
20243 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
20244 Allow register and subreg operands.
20246 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20248 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
20249 specific need, just fallback on defaults.
20250 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
20252 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20253 Olivier Hainque <hainque@adacore.com>
20255 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
20257 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
20258 targets. Pick a default if no particular attempt applied.
20259 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
20262 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20264 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
20265 (x86_64-wrs-vxworks7): Likewise.
20267 2017-06-27 Marek Polacek <polacek@redhat.com>
20270 * ubsan.c (instrument_null): Check get_base_address's result for null.
20272 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
20274 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
20276 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
20278 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
20279 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
20280 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
20281 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
20282 New function types.
20283 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
20284 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
20285 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
20286 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
20287 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
20288 BUILT_IN_FEUPDATEENV): New builtins.
20289 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
20290 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
20291 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
20292 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
20294 (builtin_structptr_types): Adjust size.
20295 * tree.c (builtin_structptr_types): Add four entries.
20297 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20298 Olivier Hainque <hainque@adacore.com>
20300 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
20301 (TLS_SYM): New local macro, forcing reference to __tls__ on
20302 link command lines for VxWorks 7 RTPs, triggering initialization
20304 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
20305 OS features TLS support, true for RTPs on VxWorks 7.
20306 * config/vxworks.c (vxworks_override_options): Setup emutls
20309 2017-06-27 Jakub Jelinek <jakub@redhat.com>
20311 * predict.c (test_prediction_value_range): Use -1U instead of -1
20312 to avoid narrowing conversion warning.
20313 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
20314 to avoid narrowing conversion warning.
20315 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
20317 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
20319 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20321 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
20322 64bit configurations.
20323 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
20324 (SIZE_TYPE): Likewise.
20325 * config/vxworks.c (vxworks_emutls_var_fields): Use
20326 long_unsigned_type_node instead of unsigned_type_node as the offset
20327 field type, which is "pointer" mode in emutls.c.
20329 2017-06-27 Jakub Jelinek <jakub@redhat.com>
20332 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
20334 PR middle-end/81207
20335 * gimple-fold.c (replace_call_with_call_and_fold): Handle
20336 gimple_vuse copying separately from gimple_vdef copying.
20338 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20340 * value-prof.c (free_hist): Remove call to memset and the enclosing if
20343 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
20344 Olivier Hainque <hainque@adacore.com>
20346 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
20347 for all vxworks7 targets.
20348 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
20349 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
20350 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
20351 variations for VX6/VX7 and 32/64bits later on in ...
20352 (VXWORKS_LIB_SPEC): Leverage new macros.
20353 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
20354 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
20356 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
20358 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
20359 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
20361 2017-06-26 Carl Love <cel@us.ibm.com>
20363 * config/rs6000/rs6000-c.c: Add support for built-in functions
20364 vector bool char vec_reve (vector bool char);
20365 vector signed char vec_reve (vector signed char);
20366 vector unsigned char vec_reve (vector unsigned char);
20367 vector bool int vec_reve (vector bool int);
20368 vector signed int vec_reve (vector signed int);
20369 vector unsigned int vec_reve (vector unsigned int);
20370 vector bool long long vec_reve (vector bool long long);
20371 vector signed long long vec_reve (vector signed long long);
20372 vector unsigned long long vec_reve (vector unsigned long long);
20373 vector bool short vec_reve (vector bool short);
20374 vector signed short vec_reve (vector signed short);
20375 vector double vec_reve (vector double);
20376 vector float vec_reve (vector float);
20377 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
20378 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
20379 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
20380 (altivec_vreve): New pattern.
20381 * config/rs6000/altivec.h (vec_reve): New define.
20382 * doc/extend.texi (vec_rev): Update the built-in documentation file
20383 for the new built-in functions.
20385 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20387 PR tree-optimization/71815
20388 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
20390 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
20392 (slsr_process_phi): Likewise.
20393 (replace_uncond_cands_and_profitable_phis): Don't replace a
20394 multiply candidate with a stride of 1 (copy or cast).
20395 (phi_incr_cost): Call uses_consumed_by_stmt rather than
20397 (lowest_cost_path): Likewise.
20398 (total_savings): Likewise.
20400 2017-06-26 Richard Biener <rguenther@suse.de>
20403 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
20404 Use def_builtin_pure for all gather builtins.
20406 2017-06-26 Richard Biener <rguenther@suse.de>
20408 PR tree-optimization/81203
20409 * tree-tailcall.c (find_tail_calls): Do not move stmts into
20410 non-dominating BBs.
20412 2017-06-26 Marek Polacek <polacek@redhat.com>
20415 * doc/invoke.texi: Document -Wmultistatement-macros.
20417 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
20419 * doc/sourcebuild.texi (ARM-specific attributes): Document new
20420 arm_neon_ok_no_float_abi effective target.
20422 2017-06-26 Richard Biener <rguenther@suse.de>
20424 PR tree-optimization/80928
20425 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
20426 (copy_bbs): Set BB_DUPLICATED flag early.
20427 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
20429 (execute_on_shrinking_pred): Likewise.
20430 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
20431 BB_DUPLICATED blocks.
20432 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
20433 iterate over all PHIs considering removal of *gsi.
20435 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
20437 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
20440 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20442 * config/rs6000/rs6000-string.c: (expand_block_clear,
20443 do_load_for_compare, select_block_compare_mode,
20444 compute_current_alignment, expand_block_compare,
20445 expand_strncmp_align_check, expand_strn_compare,
20446 expand_block_move, rs6000_output_load_multiple)
20447 Move functions related to string/block move/compare
20448 to a separate file.
20449 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
20450 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
20451 for this function which is now used in two files.
20452 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
20453 * config.gcc: Add rs6000-string.o to extra_objs for
20454 targets powerpc*-*-* and rs6000*-*-*.
20456 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
20459 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
20460 32-bit, since indexed is not valid for DImode.
20461 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
20462 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
20463 (define_peephole2 for Altivec d-form load): Add 32-bit support.
20464 (define_peephole2 for Altivec d-form store): Likewise.
20467 * multiple_target.c (create_dispatcher_calls): Only create the
20468 dispatcher call if the function is the default clone of a
20469 versioned function.
20471 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
20473 PR middle-end/80902
20474 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
20475 a call, force the call to not be a tail call.
20477 2017-06-23 Jeff Law <law@redhat.com>
20479 * doc/contrib.texi: Add entry for Steven Pemberton's work on
20482 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
20484 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
20485 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
20486 handling for early expansion of vector shifts (sl,sr,sra,rl).
20487 (builtin_function_type): Add vector shift right instructions
20488 to the unsigned argument list.
20490 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
20492 rtl-optimizatoin/79286
20493 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
20494 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
20495 trap. PIC register plus a const unspec without offset can never trap.
20497 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
20499 * tree.h (builtin_structptr_type): New type.
20500 (builtin_structptr_types): Declare new array.
20501 * tree.c (builtin_structptr_types): New array.
20502 (free_lang_data, build_common_tree_nodes): Use it.
20504 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
20507 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
20510 2017-06-22 Matt Turner <mattst88@gmail.com>
20512 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
20513 Lake models to skylake case. Assume skylake for unknown
20514 models with clflushopt.
20516 2017-06-22 Jeff Law <law@redhat.com>
20518 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
20519 frame sizes that do not satisfy aarch64_uimm12_shift.
20521 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
20523 * profile-count.h (apply_probability,
20524 apply_scale, probability_in): Fix checks for zero.
20526 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20528 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
20529 * doc/cppdiropts.texi (-I @var{dir}): Document it.
20531 2016-06-22 Richard Biener <rguenther@suse.de>
20533 * tree-vect-loop.c (vect_model_reduction_cost): Handle
20534 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
20535 REDUC_MAX_EXPR support.
20536 (vectorizable_reduction): Likewise.
20537 (vect_create_epilog_for_reduction): Likewise.
20539 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
20541 * match.pd (A / (1 << B) -> A >> B): New.
20542 * generic-match-head.c: Include optabs-tree.h.
20543 * gimple-match-head.c: Likewise.
20544 * optabs-tree.h (target_supports_op_p): New.
20545 * optabs-tree.c (target_supports_op_p): New.
20547 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20549 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
20550 $gcc_cv_ld --help output.
20551 (gcc_cv_ld_demangle): Likewise.
20552 (gcc_cv_ld_eh_frame_hdr): Likewise.
20553 (gcc_cv_ld_pie): Likewise.
20554 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
20555 (gcc_cv_ld_buildid): Likewise.
20556 (gcc_cv_ld_sysroot): Likewise.
20557 (ld_bndplt_support): Likewise.
20558 (ld_pushpopstate_support): Likewise.
20559 * configure: Regenerate.
20560 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
20562 2017-06-21 Jakub Jelinek <jakub@redhat.com>
20565 * config/i386/sse.md (round<mode>2): Renumber match_dup and
20566 operands indexes to avoid gap between operands and match_dups.
20568 2017-06-21 Andrew Pinski <apinski@cavium.com>
20570 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
20571 Increment Arith_shift and Arith_shift_reg by 1.
20572 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
20574 * config/aarch64/aarch64.c (thunderx_tunings): Enable
20575 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
20576 (aarch64_strip_extend): Add new argument and test for it.
20577 (aarch64_cheap_mult_shift_p): New function.
20578 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
20579 add a cost if it is true.
20580 Update calls to aarch64_strip_extend.
20581 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
20583 2017-06-21 Andrew Pinski <apinski@cavium.com>
20585 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
20587 (thunderxt88): Likewise.
20588 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
20589 (thunderx_prefetch_tune): New variable.
20590 (thunderx2t99_prefetch_tune): Update for the correct values.
20591 (thunderxt88_tunings): New variable.
20592 (thunderx_tunings): Use thunderx_prefetch_tune instead of
20593 generic_prefetch_tune.
20594 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
20596 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20598 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
20599 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
20600 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
20601 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
20602 (aarch64_atomic_cas<mode>, GPI): Likewise.
20604 2017-06-21 Martin Liska <mliska@suse.cz>
20606 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
20607 statements on cold and hot labels.
20608 * predict.c (tree_estimate_probability_bb): Remove the
20609 prediction from this place.
20611 2017-06-21 Martin Liska <mliska@suse.cz>
20613 PR tree-optimization/79489
20614 * gimplify.c (maybe_add_early_return_predict_stmt): New
20616 (gimplify_return_expr): Call the function.
20617 * predict.c (tree_estimate_probability_bb): Remove handling
20619 * predict.def: Update comment about early return predictor.
20620 * gimple-predict.h (is_gimple_predict): New function.
20621 * predict.def: Change default value of early return to 66.
20622 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
20624 * passes.def: Put pass_strip_predict_hints to the beginning of
20627 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
20629 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
20630 FUNCTION_DECL declarations.
20631 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
20633 (dwaf2out_decl): Likewise.
20634 * godump.c (go_early_global_decl): Skip call to the real debug hook
20635 for FUNCTION_DECL declarations.
20636 * passes.c (rest_of_decl_compilation): Skip call to the
20637 early_global_decl debug hook for FUNCTION_DECL declarations, unless
20638 -fdump-go-spec is passed.
20640 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
20642 * config/i386/i386.c (struct builtin_isa): New field pure_p.
20643 Reorder for compactness.
20644 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
20645 (def_builtin_pure, def_builtin_pure2): New functions.
20646 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
20648 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
20650 * match.pd (nop_convert): New predicate.
20651 ((A +- CST1) +- CST2): Allow some NOP conversions.
20653 2017-06-21 Jakub Jelinek <jakub@redhat.com>
20656 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
20657 with ctors/dtors if GOVD_SHARED is set.
20659 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20661 * config/aarch64/aarch64.md (movti_aarch64):
20662 Emit mov rather than orr.
20663 (movtf_aarch64): Likewise.
20664 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
20665 Emit mov rather than orr.
20667 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20669 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
20670 Swap alternatives, make integer dup more expensive.
20672 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20674 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
20675 Return true for non-tls symbols.
20677 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
20679 * config/aarch64/aarch64-cores.def (cortex-a55): New.
20680 (cortex-a75): Likewise.
20681 (cortex-a75.cortex-a55): Likewise.
20682 * config/aarch64/aarch64-tune.md: Regenerate.
20683 * doc/invoke.texi (-mtune): Document new values for -mtune.
20685 2017-06-21 Tom de Vries <tom@codesourcery.com>
20687 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
20688 stack_size feature.
20689 (Effective-Target Keywords, Other attributes): Suggest using
20690 dg-add-options stack_size feature to get stack limit in stack_size
20691 effective target documentation.
20693 2017-06-21 Julian Brown <julian@codesourcery.com>
20694 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20696 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
20697 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
20698 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
20700 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
20701 attribute type list for neon_multiply.
20702 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
20703 attribute type list for neon_multiply.
20704 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
20705 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
20706 attribute type list for neon_multiply.
20707 * config/arm/types.md (crypto_pmull): Add.
20708 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
20709 attribute type list.
20711 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
20713 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
20716 2017-06-20 Jakub Jelinek <jakub@redhat.com>
20718 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
20719 to make sure not to dereference a NULL cost_classes_ptr pointer.
20721 2017-06-20 Carl Love <cel@us.ibm.com>
20723 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20724 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
20725 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
20726 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
20727 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
20728 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
20729 VMULOSW): New enum "unspec" values.
20730 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
20731 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
20732 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
20733 altivec_vmulosw): New patterns.
20734 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
20735 VMULOSW): Add definitions.
20737 2017-06-20 Julia Koval <julia.koval@intel.com>
20739 * config/i386/i386.c: Fix rounding expand for new pattern.
20740 * config/i386/subst.md: Fix pattern (parallel -> unspec).
20742 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20744 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
20745 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
20747 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20749 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
20752 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20754 * config/aarch64/aarch64-cores.def: Rearrange to sort by
20755 architecture, then by implementer ID.
20756 * config/aarch64/aarch64-tune.md: Regenerate.
20758 2017-06-20 Richard Biener <rguenther@suse.de>
20760 PR middle-end/81097
20761 * fold-const.c (split_tree): Fold to type before negating.
20763 2017-06-20 David Malcolm <dmalcolm@redhat.com>
20765 * diagnostic-show-locus.c
20766 (selftest::test_fixit_deletion_affecting_newline): New function.
20767 (selftest::diagnostic_show_locus_c_tests): Call it.
20769 2017-06-20 Andreas Schwab <schwab@suse.de>
20772 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
20775 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
20777 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
20778 __ARM_FEATURE_COPROC according to support.
20780 2017-06-20 Jakub Jelinek <jakub@redhat.com>
20782 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
20783 Rewritten to avoid overflow for > 32-bit pointers.
20786 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
20787 by removing enum keyword.
20788 (ubsan_type_descriptor): Likewise. Formatting fix.
20791 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
20792 splitter): Require TARGET_SSE2 in the condition.
20794 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
20797 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
20798 for doing vector set of SFmode on ISA 3.0.
20799 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
20800 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
20802 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
20803 SFmode value into a V4SF variable that was extracted from another
20804 V4SF variable without converting the element to double precision
20805 and back to single precision vector format.
20806 (vsx_insert_extract_v4sf_p9_2): Likewise.
20808 2017-06-19 Jakub Jelinek <jakub@redhat.com>
20810 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
20811 in UWHI to avoid undefined overflow.
20814 * ubsan.h (enum ubsan_encode_value_phase): New.
20815 (ubsan_encode_value): Change second argument to
20816 enum ubsan_encode_value_phase with default value of
20817 UBSAN_ENCODE_VALUE_GENERIC.
20818 * ubsan.c (ubsan_encode_value): Change second argument to
20819 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
20820 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
20821 create_tmp_var_raw instead of create_tmp_var and use a
20823 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
20824 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
20825 ubsan_encode_value callers.
20828 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
20829 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
20830 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
20832 2017-06-19 Richard Biener <rguenther@suse.de>
20834 PR middle-end/81118
20835 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
20836 estimates if we changed anything.
20838 2017-06-19 Richard Biener <rguenther@suse.de>
20840 PR tree-optimization/80887
20841 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
20842 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
20843 simplified lookups, then reset mprts_hook.
20844 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
20846 (try_to_simplify): Likewise.
20848 2017-06-19 Martin Liska <mliska@suse.cz>
20851 * gimplify.c (gimplify_switch_expr):
20852 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
20854 2017-06-19 Martin Liska <mliska@suse.cz>
20856 * doc/install.texi: Document that PGO runs in 4 stages.
20858 2017-06-19 Martin Liska <mliska@suse.cz>
20861 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
20862 to dispatcher function name.
20863 * multiple_target.c (replace_function_decl): New function.
20864 (create_dispatcher_calls): Redirect both edges and references.
20866 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
20868 * profile-count.c (profile_count::dump): Dump quality.
20869 (profile_count::differs_from_p): Update for unsigned val.
20870 * profile-count.h (profile_count_quality): New enum.
20871 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
20873 2017-06-19 Richard Biener <rguenther@suse.de>
20875 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
20876 struct function as arg.
20877 (estimate_numbers_of_iterations): Export overload with loop arg.
20878 (free_numbers_of_iterations_estimates_loop): Use an overload of
20879 free_numbers_of_iterations_estimates instead.
20880 * tree-cfg.c (remove_bb): Adjust.
20881 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
20882 * tree-parloops.c (gen_parallel_loop): Likewise.
20883 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
20885 (tree_unroll_loops_completely): Likewise.
20886 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
20887 Use an overload instead and export.
20888 (estimated_loop_iterations): Adjust.
20889 (max_loop_iterations): Likewise.
20890 (likely_max_loop_iterations): Likewise.
20891 (estimate_numbers_of_iterations): Take struct function as arg
20893 (loop_exits_before_overflow): Adjust.
20894 (free_numbers_of_iterations_estimates_loop): Use an overload.
20895 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
20896 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
20898 2017-06-19 Richard Biener <rguenther@suse.de>
20901 * ipa-prop.c (find_constructor_constant_at_offset): Handle
20902 RANGE_EXPR conservatively.
20904 2017-06-16 Carl Love <cel@us.ibm.com>
20906 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20907 definitions for vec_float, vec_float2, vec_floato,
20908 vec_floate built-ins.
20909 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
20910 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
20912 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
20913 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
20914 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
20915 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
20916 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
20917 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
20918 vec_floato): Add builtin defines.
20919 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
20920 Update the built-in documentation file for the new built-in
20923 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20925 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
20926 (mthumb): Mark as the negative of -marm.
20928 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20930 * doc/invoke.texi (ARM Options, -mcpu): Document supported
20932 (ARM Options, -mtune): Document that this accepts the same
20933 extension options as -mcpu.
20934 (ARM Options, -mfpu): Document addition of -mfpu=auto.
20936 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20938 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
20939 permitted extensions.
20941 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20943 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
20944 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
20945 (armv8-m.main): Add option +nodsp.
20946 * config/arm/arm-cpu-cdata.h: Regenerated.
20948 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20950 * config/arm/t-fuchsia: New file.
20951 * config.gcc (arm*-*-fuchsia*): Use it.
20953 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20955 * config/arm/t-symbian: Rewrite for new option infrastructure.
20957 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20959 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
20960 (MULTILIB_REQUIRED): Likewise.
20962 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20964 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
20965 (MULTILIB_RESUE): Likewise.
20966 (MULTILIB_MATCHES): Likewise.
20967 (MULTLIB_REQUIRED): Likewise.
20969 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20971 * config/arm/t-rtems: Rewrite for new option framework.
20973 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20975 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
20976 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
20977 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
20978 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
20979 * config/arm/t-multilib: ... here.
20980 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
20981 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
20982 armv7-a and armv8*-a when A-profile libraries have not been built.
20983 * config/arm/t-rmprofile: Rewrite.
20985 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20987 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
20988 with a backslash. Remove the backslash after substituting unescaped
20990 * doc/fragments.texi (MULTILIB_REUSE): Document it.
20992 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20994 * config.gcc: (arm*-*-*): When building a-profile libraries, force
20995 the driver to pass through the default setting of -mfloat-abi.
20996 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
20998 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
20999 (all_feat_combs): New rule.
21000 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
21002 * config/arm/t-aprofile: Rewrite.
21004 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21006 * config/arm/arm.h (FPUTYPE_AUTO): Define.
21007 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
21008 fpu is not specified by the user/command-line.
21009 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
21010 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
21011 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
21012 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
21013 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
21014 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
21016 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21018 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
21019 * config/arm/t-arm-elf: Rewritten.
21021 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21023 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
21024 have some floating-point instructions.
21025 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
21026 (TARGET_MAYBE_HARD_FLOAT): New macro.
21027 * config/arm/arm-builtins.c (arm_init_builtins): Use
21028 TARGET_MAYBE_HARD_FLOAT.
21029 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
21031 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21033 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
21034 (configargs.h): Include it.
21035 (arm_print_hint_for_fpu_option): New function.
21036 (arm_parse_fpu_option): New function.
21037 (candidate_extension): New class.
21038 (arm_canon_for_multilib): New function.
21039 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
21040 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
21041 (ARCH_CANONICAL_SPECS): New macro.
21042 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
21044 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21046 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
21047 are set after handling multilib fragments. Set target_cpu_default2
21050 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21052 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
21054 (arm*-*-*): Set target_cpu_default2 to a quoted string.
21055 * config/arm/parsecpu.awk (check_cpu): Validate any extension
21057 (check_arch): Likewise.
21058 * config/arm/arm.c (arm_configure_build_target): Handle
21059 TARGET_CPU_DEFAULT being a string constant. Scan any feature
21060 options in the default.
21062 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21064 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
21065 when an option is an alias of another.
21066 * config/arm/parsecpu.awk (optalias): New parser token.
21067 (gen_comm_data): Mark non-alias options as such. Emit entries
21068 for extension aliases.
21069 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
21070 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
21071 (armv6kz, armv6zk, armv6t2): Likewise.
21072 (armv7): Make vfpv3-d16 an alias.
21073 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
21075 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
21076 Sort in canonical order.
21077 (armv8-a): Sort in canonical order.
21078 (armv8.1-a, armv8.2-a): Likewise.
21079 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
21081 (cortex-a9): Sort in canonical order.
21082 * config/arm/arm.c (selftests.h): Include it.
21083 (arm_test_cpu_arch_data): New function.
21084 (arm_run_self_tests): New function.
21085 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
21086 (targetm): Move declaration to the end of the file.
21087 * arm-cpu-cdata.h: Regenerated.
21089 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21091 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
21092 call to target_mode_check describing the type of option passed.
21093 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
21094 (arm_target_thumb_only): Use arm_parse_arch_option_name or
21095 arm_parse_cpu_option_name to match parameters against list of
21097 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
21098 arm_arch_core_flags data structure.
21099 * config/arm/arm-cpu_cdata.h: Regenerated.
21101 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21103 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
21105 (arm_print_hint_for_cpu_option): Likewise.
21106 (arm_print_hint_for_arch_option): Likewise.
21107 (arm_parse_cpu_option_name): Likewise.
21108 (arm_parse_arch_option_name): Likewise.
21109 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
21110 of entries in the all_fpus list.
21111 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
21112 (arm_parse_cpu_option_name): Declare.
21113 (arm_parse_arch_option_name): Declare.
21114 (arm_parse_option_features): Declare.
21115 (arm_intialize_isa): Declare.
21116 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
21118 (gen_comm_data): ... here. Make definitions non-static.
21119 * config/arm/arm-cpu-data.h: Regenerated.
21120 * config/arm/arm-cpu-cdata.h: Regenerated.
21122 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21124 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
21125 (cpu_arch_extension): New structure.
21126 (cpu_arch_option, arch_option, cpu_option): New structures.
21127 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
21128 architecture types.
21129 (gen_data): Generate new format data tables.
21130 * config/arm/arm.c (cpu_tune): New structure.
21131 (cpu_option, processors): Delete.
21132 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
21133 (arm_print_hint_for_cpu_option): ... this and ...
21134 (arm_print_hint_for_arch_option): ... this.
21135 (arm_parse_arch_cpu_name): Delete. Replace with ...
21136 (arm_parse_cpu_option_name): ... this and ...
21137 (arm_parse_arch_option_name): ... this.
21138 (arm_unrecognized_feature): Change type of target parameter to
21140 (arm_parse_arch_cpu_features): Delete. Replace with ...
21141 (arm_parse_option_features): ... this.
21142 (arm_configure_build_target): Rework to use new configuration data
21144 (arm_print_tune_info): Rework for new configuration data tables.
21145 * config/arm/arm-cpu-data.h: Regenerated.
21146 * config/arm/arm-cpu.h: Regenerated.
21148 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21150 * Makefile.in (OBJS): Move sbitmap.o from here ...
21151 (OBJS-libcommon): ... to here.
21153 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21155 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
21156 (ISA_ALL_CRYPTO): New macro.
21157 (ISA_ALL_SIMD): New macro
21158 (ISA_ALL_FP): New macro.
21159 * config/arm/arm.c (fpu_bitlist): Update initializer.
21160 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
21162 (arm9e): Add fpu. Add option for nofp
21163 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
21164 (arm926ej-s, arm1026ej-s): Likewise.
21165 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
21166 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
21167 neon-fp16, neon-vfpv4, nofp and nosimd.
21168 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
21169 (cortex-a8): Add fpu. Add option for nofp.
21170 (cortex-a9): Add fpu. Add options for nosimd and nofp.
21171 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
21172 (cortex-r4f): Add fpu.
21173 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
21174 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
21176 (cortex-r8): Likewise.
21177 (cortex-m4): Add fpu. Add option for nofp.
21178 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
21179 (cortex-a17.cortex-a7): Likewise.
21180 (cortex-a32): Add fpu. Add options for crypto and nofp.
21181 (cortex-a35, cortex-a53): Likewise.
21182 (cortex-a57): Add fpu. Add option for crypto.
21183 (cortex-a72, cortex-a73): Likewise.
21184 (exynos-m1): Likewise.
21185 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
21186 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
21187 (cortex-m33): Add fpu. Add option for nofp.
21188 * config/arm/arm-cpu-cdata.h: Regenerated
21189 * config/arm/arm-cpu-data.h: Regenerated.
21191 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21193 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
21194 (armv5te, armv5tej): Likewise.
21195 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
21196 (armv7): Add options fp and vfpv3-d16.
21197 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
21198 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
21200 (armv7ve): Likewise.
21201 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
21202 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
21203 (armv8-a): Add nocrypto option.
21204 (armv8.1-a, armv8.2-a): Likewise.
21205 (armv8-m.main): add options fp, fp.dp and nofp.
21207 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21209 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
21211 (armv8-a+crc): Delete.
21212 (armv8.1-a): Add options simd, crypto and nofp.
21213 (armv8.2-a): Add options fp16, simd, crypto and nofp.
21214 (armv8.2-a+fp16): Delete.
21215 (armv8-m.main): Add option dsp.
21216 (armv8-m.main+dsp): Delete.
21217 (cortex-a8): Add fpu. Add nofp option.
21218 (cortex-a9): Add fpu. Add nofp and nosimd options.
21219 * config/arm/parsecpu.awk (gen_data): Generate option tables and
21220 link to main cpu and architecture data structures.
21221 (gen_comm_data): Only put isa attributes from the main architecture
21223 (option): New statement for architecture and CPU entries.
21224 * arm.c (struct cpu_option): New structure.
21225 (struct processors): Add entry for options.
21226 (arm_unrecognized_feature): New function.
21227 (arm_parse_arch_cpu_name): Ignore any characters after the first
21229 (arm_parse_arch_cpu_feature): New function.
21230 (arm_configure_build_target): Separate out any CPU and architecture
21231 features and parse separately. Don't error out if -mfpu=auto is
21232 used with only an architecture string.
21233 (arm_print_asm_arch_directives): New function.
21234 (arm_file_start): Call it.
21235 * config/arm/arm-cpu-cdata.h: Regenerated.
21236 * config/arm/arm-cpu-data.h: Likewise.
21237 * config/arm/arm-tables.opt: Likewise.
21239 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21241 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
21242 assembler when it is not -mfpu=auto.
21244 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21246 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
21247 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
21248 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
21249 (ASM_CPU_SPEC): Rewrite.
21250 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
21251 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
21252 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
21253 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
21254 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
21255 copied string is NUL-terminated. Also strip any characters prefixed
21257 (arm_rewrite_selected_arch): New function.
21258 (arm_rewrite_march): New function.
21260 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21262 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
21263 (x_arm_cpu_string, x_arm_tune_string): Likewise.
21264 (march, mcpu, mtune): Convert to string-based options.
21265 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
21266 (arm_parse_arch_cpu_name): New function.
21267 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
21268 identify selected architecture or CPU.
21269 (arm_option_save): New function.
21270 (TARGET_OPTION_SAVE): Redefine.
21271 (arm_option_restore): Restore string options.
21272 (arm_option_print): Print string options.
21274 2017-06-16 Martin Sebor <msebor@redhat.com>
21276 PR tree-optimization/80933
21277 PR tree-optimization/80934
21278 * builtins.c (fold_builtin_3): Do not handle bcmp here.
21279 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
21280 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
21281 (gimple_fold_builtin): Call them.
21283 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21285 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
21286 as unlikely; update profile.
21288 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21290 * predict.c (force_edge_cold): Handle declaring edges impossible
21293 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21295 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
21297 (try_unroll_loop_completely): Fix reporting.
21299 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21301 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
21303 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
21306 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
21307 if given a non-constant argument for an intrinsic which requires a
21310 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21312 * profile.c (compare_freqs): New function.
21313 (branch_prob): Sort edge list.
21314 (find_spanning_tree): Assume that the list is priority sorted.
21316 2017-06-16 Richard Biener <rguenther@suse.de>
21318 PR tree-optimization/81090
21319 * passes.def (pass_record_bounds): Remove.
21320 * tree-pass.h (make_pass_record_bounds): Likewise.
21321 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
21322 make_pass_record_bounds): Likewise.
21323 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
21324 not free niter estimates at the beginning but at the end.
21325 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
21327 2017-06-16 Richard Biener <rguenther@suse.de>
21329 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
21330 initializer to workaround ICE in host GCC 4.8.
21332 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21334 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
21336 (clone_inlined_nodes): Update.
21338 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21340 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
21341 prefetch settings, and enable prefetching by default at -O3.
21343 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21345 * config/aarch64/aarch64.c (aarch64_override_options_internal):
21346 Set flag_prefetch_loop_arrays according to tuning data.
21348 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21350 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
21351 New tune structure.
21352 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
21353 [Unrelated to main purpose of the patch] Place the pointer field last
21354 to enable type checking errors when tune structure are wrongly merged.
21355 * config/aarch64/aarch64.c (generic_prefetch_tune,)
21356 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
21357 (thunderx2t99_prefetch_tune): New tune constants.
21358 (tune_params *_tunings): Update all tunings (no functional change).
21359 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
21360 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
21361 from tunings structures.
21363 2017-06-16 Jakub Jelinek <jakub@redhat.com>
21366 * ubsan.c (instrument_null): Add T argument, use it instead
21367 of computing it based on IS_LHS.
21368 (instrument_object_size): Likewise.
21369 (pass_ubsan::execute): Adjust instrument_null and
21370 instrument_object_size callers to pass gimple_get_lhs or
21371 gimple_assign_rhs1 result to it. Use instrument_null instead of
21372 calling get_base_address and instrument_mem_ref. Handle
21373 aggregate call arguments for object-size sanitization.
21375 2017-06-16 Yury Gribov <tetra2005@gmail.com>
21377 PR tree-optimization/81089
21378 * tree-vrp.c (is_masked_range_test): Validate operands of
21381 2017-06-15 Martin Sebor <msebor@redhat.com>
21384 * dumpfile.c (dump_register): Avoid calling memset to initialize
21385 a class with a default ctor.
21386 * gcc.c (struct compiler): Remove const qualification.
21387 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
21388 * hash-table.h: Ditto.
21389 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
21391 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
21392 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
21394 * params.h (struct param_info): Make struct members non-const.
21395 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
21396 with default initialization.
21397 * vec.h (vec_copy_construct, vec_default_construct): New helper
21399 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
21400 with vec_copy_construct.
21401 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
21402 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
21403 * doc/invoke.texi (-Wclass-memaccess): Document.
21405 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
21407 * emit-rtl.h (is_leaf): Update comment about local
21408 register allocator.
21410 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
21413 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
21414 for a variable to have a section before checking if the section has a
21416 Set section to.persistent if persistent attribute is set.
21417 Warn if .persistent attribute is used on an automatic variable.
21419 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
21421 PR rtl-optimization/80474
21422 * reorg.c (update_block): Do not ignore instructions in a delay slot.
21424 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
21426 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
21429 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
21431 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
21432 (casesi): Emit bounds checking as RTL.
21433 (casesi_internal_mips16_<mode>): Remove bounds checking.
21435 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
21437 * config/xtensa/xtensa.c (xtensa_option_override): Append
21438 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
21439 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
21440 xtensa_doloop_hooks): Define unconditionally.
21441 (xtensa_reorg_loops): Only call reorg_loops in the presence of
21443 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
21444 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
21445 for it in xtensa_option_override.
21446 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
21447 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
21449 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
21451 * doc/cppopts.texi: Document '-' special value to -MF.
21453 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
21455 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
21456 (cortex_a53_fconst): Likewise.
21457 (cortex_a53_fpmul): Likewise.
21458 (cortex_a53_f_load_64): Likewise.
21459 (cortex_a53_f_load_many): Likewise.
21460 (cortex_a53_advsimd_alu): Likewise.
21461 (cortex_a53_advsimd_alu_q): Likewise.
21462 (cortex_a53_advsimd_mul): Likewise.
21463 (cortex_a53_advsimd_mul_q): Likewise.
21464 (fpmac bypass): Add new bypass for fpmac-fpmac case.
21465 Add missing fmul, r2f_cvt and fconst cases.
21467 2017-06-14 Richard Biener <rguenther@suse.de>
21469 PR middle-end/81088
21470 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
21472 (fold_binary_loc): When associating do not treat pre-existing
21473 TREE_OVERFLOW on literal constants as a reason to allow
21474 TREE_OVERFLOW on associated literal constants.
21476 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21478 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
21479 (MASK_FEATURES): New macro.
21480 * config/sparc/sparc.c (sparc_option_override): Remove the special
21481 handling of -mfpu and generalize it to all MASK_FEATURES switches.
21483 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21485 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
21486 a division of 0 if non-call exceptions are enabled.
21488 2017-06-14 Andrew Pinski <apinski@cavium.com>
21489 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21492 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
21493 Improve vector initialization code gen for only variable case.
21495 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21497 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
21499 2017-06-14 Richard Biener <rguenther@suse.de>
21501 PR tree-optimization/81083
21502 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
21505 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21507 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
21508 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
21509 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
21510 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
21511 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
21512 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
21514 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21516 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
21517 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
21519 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21521 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
21523 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21525 * config/rs6000/t-rtems: Don't handle SPE.
21527 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21529 * config/rs6000/t-linux: Don't handle SPE.
21531 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21533 * config/rs6000/eabispe.h: Delete file.
21535 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21537 * config/rs6000/t-spe: Delete file.
21539 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21541 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
21542 (rs6000_legitimate_offset_address_p): Return false for anything in
21543 V2SImode or V2SFmode.
21545 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21547 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
21548 except V2SF and V2SI. Rearrange the vector modes, and add comments.
21549 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
21551 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
21552 (rs6000_legitimate_offset_address_p): Ditto.
21553 (rs6000_emit_move): Ditto.
21554 (rs6000_init_builtins): Remove V4HI_type_node.
21556 2017-06-13 Martin Liska <mliska@suse.cz>
21559 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
21560 (gate_asan): Likewise.
21561 * asan.h (asan_no_sanitize_address_p): Remove the function.
21562 (sanitize_flags_p): New function.
21563 * builtins.def: Fix coding style.
21564 * common.opt: Use renamed enum value.
21565 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
21566 * doc/extend.texi: Document no_sanitize attribute.
21567 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
21568 to SANITIZE_UNDEFINED_NONDEFAULT.
21569 * gcc.c (sanitize_spec_function): Use the renamed enum value.
21570 * gimple-fold.c (optimize_atomic_compare_exchange_p):
21571 Use sanitize_flags_p.
21572 * gimplify.c (gimplify_function_tree): Likewise.
21573 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
21574 * opts.c (parse_no_sanitize_attribute): New function.
21575 (common_handle_option): Use renamed enum value.
21576 * opts.h (parse_no_sanitize_attribute): Declare.
21577 * tree.c (sanitize_flags_p): New function.
21578 * tree.h: Declared here.
21579 * tsan.c: Use sanitize_flags_p.
21580 * ubsan.c (ubsan_expand_null_ifn): Likewise.
21581 (instrument_mem_ref): Likewise.
21582 (instrument_bool_enum_load): Likewise.
21583 (do_ubsan_in_current_function): Remove the function.
21584 (pass_ubsan::execute): Use sanitize_flags_p.
21585 * ubsan.h: Remove do_ubsan_in_current_function
21586 * tree-cfg.c (print_no_sanitize_attr_value): New function.
21587 (dump_function_to_file): Use it here.
21589 2017-06-13 Martin Jambor <mjambor@suse.cz>
21591 PR tree-optimization/80803
21592 PR tree-optimization/81063
21593 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
21594 (propagate_subaccesses_across_link): Enqueue subtree whenever
21595 necessary instead of relying on the caller.
21597 2017-06-13 Martin Jambor <mjambor@suse.cz>
21599 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
21600 that have a first_link.
21601 (sort_and_splice_var_accesses): Do not check first_link before
21603 (subtree_mark_written_and_enqueue): Likewise.
21604 (propagate_all_subaccesses): Likewise and do not stop at first
21605 parent with a first_link.
21607 2017-06-13 Martin Jambor <mjambor@suse.cz>
21609 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
21612 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21614 * match.pd: New pattern.
21616 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21618 * tree-vrp.c (is_masked_range_test): New function.
21619 (register_edge_assert_for): Determine ranges for
21622 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21624 PR tree-optimization/67328
21625 * fold-const.c (maskable_range_p): New function.
21626 (build_range_check): Generate bittests if possible.
21628 2017-06-13 Martin Liska <mliska@suse.cz>
21630 * gimple-pretty-print.c (dump_probability): Add new argument.
21631 (dump_edge_probability): Dump both probability and count.
21632 (dump_gimple_label): Likewise.
21633 (dump_gimple_bb_header): Likewise.
21635 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
21638 * config/avr/avr-devices.c: Fix indentation.
21639 * config/avr/gen-avr-mmcu-specs.c: Dito.
21641 2017-06-13 Richard Biener <rguenther@suse.de>
21643 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
21644 instead get vector type from stmt_info.
21645 (vectorizable_reduction): Adjust. Remove dead code.
21647 2017-06-13 Richard Biener <rguenther@suse.de>
21649 PR middle-end/81065
21650 * fold-const.c (extract_muldiv_1): Remove bogus distribution
21651 case of C * (x * C2 + C3).
21652 (fold_addr_of_array_ref_difference): Properly fold index difference.
21654 2017-06-12 David S. Miller <davem@davemloft.net>
21657 * config/sparc/sparc.md (return expander): Emit frame blockage if
21658 function uses alloca.
21660 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
21662 * combine.c (make_field_assignment): Check len rather than the mode
21663 precision when calling force_to_mode.
21665 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
21667 Support multilibs and devices that see flash in RAM address range.
21670 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
21671 (avr_mcu_t) <flash_pm_offset>: New field.
21672 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
21673 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
21674 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
21675 (AVR_TINY_PM_OFFSET): Remove macro.
21676 * config/avr/avr.opt (-mshort-calls): New option.
21677 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
21678 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
21679 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
21680 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
21681 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
21682 instead of avr_arch->have_jmp_call.
21683 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
21684 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
21685 avr_arch->flash_pm_offset to define.
21686 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
21687 new field flash_pm_offset. Add entry for avrxmega3.
21688 (avr_texinfo): Add entry for avrxmega3.
21689 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
21690 attiny212, attiny214,
21691 attiny412, attiny414, attiny416, attiny417,
21692 attiny814, attiny816, attiny817,
21693 attiny1614, attiny1616, attiny1617,
21694 attiny3214, attiny3216, attiny3217.
21695 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
21696 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
21697 (avr_print_operand_address) [AVR_TINY]: Same.
21698 (avr_asm_init_sections) <readonly_data_section>: Only patch
21699 callback if avr_arch->flash_pm_offset = 0.
21700 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
21701 for rodata if avr_arch->flash_pm_offset != 0.
21702 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
21703 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
21704 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
21705 (m_options): Append opt_rcall.
21706 (m_dirnames): Append dir_rcall.
21707 * config/avr/t-multilib: Regenerate.
21709 * configure.ac [target=avr]: Check whether avrxmega3 default
21710 linker description file works as needed.
21711 * configure: Regenerate.
21712 * doc/avr-mmcu.texi: Regenerate.
21713 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
21714 <__AVR_ARCH__>: Document avrxmega3 and 103.
21715 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
21716 <__AVR_SHORT_CALLS__>: Document it.
21717 <__AVR_PM_BASE_ADDRESS__>: Document it.
21718 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
21719 (AVR Variable Attributes) <progmem>: Document this is
21720 not needed for avrxmega3.
21721 (AVR Named Address Spaces) <__flash>: Dito.
21723 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
21725 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
21727 2017-06-12 Doug Rupp <rupp@adacore.com>
21729 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
21730 Append vxworks-stdint.h to the tm_file list.
21731 * config/vxworks-stdint.h: New file.
21733 2017-06-12 Martin Liska <mliska@suse.cz>
21735 PR tree-optimization/81041
21736 * tree-profile.c (gimple_gen_ic_func_profiler):
21737 Create an extra BB in profile-generate
21738 (gimple_gen_time_profiler): Likewise.
21740 2017-06-12 Jakub Jelinek <jakub@redhat.com>
21742 PR tree-optimization/81003
21743 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
21744 (update_range_test): Use it instead of force_gimple_operand_gsi.
21746 2017-06-12 Richard Biener <rguenther@suse.de>
21748 PR tree-optimization/81053
21749 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
21750 with backedge value not defined in loop. Simplify def stmt
21753 2017-06-11 Tom de Vries <tom@codesourcery.com>
21756 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
21758 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
21759 nvptx_cannot_force_const_mem.
21761 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21763 * opts.c (finish_options): Move test for flag_split_stack after
21764 it has been initialized.
21766 2017-06-11 Jason Merrill <jason@redhat.com>
21768 * tree.h (id_equal): New.
21769 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
21770 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
21771 instead of strcmp of IDENTIFIER_POINTER.
21773 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21775 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
21776 (mark_all_inlined_calls_cdtor): Fix formating.
21777 (inline_transform): Rescale profile before inlining.
21779 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21781 * cgraph.h (cgraph_edge::clone): Update prototype.
21782 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
21783 (cgraph_node::create_clone): Update.
21784 (cgraph_node::create_version_clone): Update.
21785 * tree-inline.c (copy_bb): Update.
21786 (expand_call_inline): Update.
21788 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
21790 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
21791 factored out from ...
21792 (rs6000_emit_prologue): ... here.
21794 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
21796 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
21797 factored out from ...
21798 (rs6000_emit_prologue): ... here.
21800 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21802 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
21804 (handle_missing_profiles): Fix computation of tp_first_run.
21805 (counts_to_freqs): Do not touch freqs when count is 0.
21807 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21809 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
21812 2017-06-10 Tom de Vries <tom@codesourcery.com>
21814 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
21815 attributes): Document signal effective target.
21817 2017-06-10 Tom de Vries <tom@codesourcery.com>
21819 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
21820 Document effective target stack_size.
21822 2017-06-09 David Malcolm <dmalcolm@redhat.com>
21824 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
21825 to the edit_context if they can be auto-applied.
21827 2017-06-9 Ian Lance Taylor <iant@golang.org>
21829 * opts.c (finish_options): If -fsplit-stack, disable implicit
21830 -forder-blocks-and-partition.
21831 * doc/invoke.texi (Optimize Options): Document that when using
21832 -fsplit-stack -forder-blocks-and-partition is not implicitly
21835 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21837 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
21838 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
21839 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
21840 * builtins.def (abort, trap, unreachable): Declare cold.
21841 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
21842 * tree-core.h (ECF_COLD): New.
21843 * tree.c (set_call_expr_flags): Handle ECF_COLD.
21844 (build_common_builtin_nodes): Mark unreachable and abort as cold.
21846 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21848 * predict.c (unlikely_executed_stmt_p): Cleanup.
21850 2017-06-09 Richard Biener <rguenther@suse.de>
21852 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
21853 model if the ref is always written to.
21855 2017-06-09 Tamar Christina <tamar.christina@arm.com>
21857 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
21859 2017-06-09 Tamar Christina <tamar.christina@arm.com>
21861 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
21864 2017-06-09 Tom de Vries <tom@codesourcery.com>
21867 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
21868 "target cannot support label values" when encountering LABEL_REF.
21870 2017-06-09 Martin Liska <mliska@suse.cz>
21872 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
21873 (gimple_gen_ic_func_profiler): Emit direct comparison
21874 of __gcov_indirect_call_callee with NULL.
21875 (gimple_gen_time_profiler): Change probability from
21876 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
21878 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21880 * profile.c (edge_gcov_counts): Turn to pointer.
21881 (compute_branch_probabilities, compute_branch_probabilities): Update.
21882 (branch_prob): Do not clear edge_gcov_count.
21883 * profile.h (edge_gcov_counts): Turn to pointer.
21884 (edge_gcov_count): Update.
21886 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21888 * gimple.h (gimple_check_failed): Mark cold.
21890 2017-06-09 Richard Biener <rguenther@suse.de>
21892 PR tree-optimization/66623
21893 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
21894 refactor check_reduction into two parts, properly computing
21895 whether we have to check reduction validity for outer loop
21898 2017-06-09 Richard Biener <rguenther@suse.de>
21900 PR tree-optimization/79483
21901 * graphite-scop-detection.c (order): New global.
21902 (get_order): Compute bb to order mapping that satisfies code
21903 generation constraints.
21904 (cmp_pbbs): New helper.
21905 (build_scops): Start domwalk at entry block, sort generated
21908 2017-06-09 Richard Biener <rguenther@suse.de>
21910 PR middle-end/81007
21911 * ipa-polymorphic-call.c
21912 (ipa_polymorphic_call_context::restrict_to_inner_class):
21913 Skip FIELD_DECLs with error_mark_node type.
21914 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
21917 2017-06-09 Martin Liska <mliska@suse.cz>
21919 * predict.c (struct branch_predictor): New struct.
21920 (test_prediction_value_range): New test.
21921 (predict_c_tests): New function.
21922 * selftest-run-tests.c (selftest::run_tests): Run the function.
21923 * selftest.h: Declare new tests.
21925 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
21928 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
21929 gen_add3_insn did not fail.
21930 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
21931 r0, construct that number in a temporary reg and add that reg to r0.
21932 If asked to put the result in r0 as well, fail.
21934 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
21936 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
21937 for early expansion of vec_eqv.
21939 2017-06-08 Jakub Jelinek <jakub@redhat.com>
21941 PR middle-end/81005
21942 * ubsan.c (instrument_null): Avoid pointless code temporary.
21943 (pass_ubsan::execute): Instrument aggregate arguments of calls.
21945 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
21949 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
21952 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
21955 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21957 * predict.c (unlikely_executed_edge_p): Move ahead.
21958 (probably_never_executed_edge_p): Use it.
21960 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
21962 PR middle-end/79988
21963 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
21964 gimple_call_builtin_p call.
21966 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21968 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
21969 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
21970 rtl_check_failed_type2, rtl_check_failed_code1,
21971 rtl_check_failed_code2, rtl_check_failed_code_mode,
21972 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
21973 rtvec_check_failed_bounds, rtl_check_failed_flag,
21974 _fatal_insn_not_found, _fatal_insn): Likewise.
21975 * tree.h (tree_contains_struct_check_failed,
21976 tree_check_failed, tree_not_check_failed,
21977 tree_class_check_failed, tree_range_check_failed,
21978 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
21979 tree_vec_elt_check_failed, phi_node_elt_check_failed,
21980 tree_operand_check_failed, omp_clause_check_failed,
21981 omp_clause_operand_check_failed, omp_clause_range_check_failed):
21984 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21986 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
21987 flag_branch_probabilities.
21988 * ipa-inline.c (edge_badness): Likewise.
21989 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
21990 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
21991 * predict.c (maybe_hot_frequency_p): Likewise.
21992 (probably_never_executed): Likewise.
21993 * sched-ebb.c (schedule_ebbs): Likewise.
21994 * sched-rgn.c (find_single_block_region): Likewise.
21995 * tracer.c (tail_duplicate): Likewise.
21997 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21999 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
22000 longer requires x_flag_profile_use.
22002 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22004 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
22005 instead of flag_reorder_blocks_and_partition.
22006 * dbxout.c (dbxout_function_end): Likewise.
22007 * dwarf2out.c (gen_subprogram_die): Likewise.
22008 * haifa-sched.c (sched_create_recovery_edges): Likewise.
22009 * hw-doloop.c (reorg_loops): Likewise.
22010 * varasm.c (assemble_start_function,
22011 assemble_end_function): Likewise.
22012 (decide_function_section): Do not check for
22013 flag_reorder_blocks_and_partition.
22015 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
22017 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
22019 (chkp_get_hard_register_fake_addr_expr): Ditto.
22020 (chkp_build_addr_expr): Add check for hard reg case.
22021 (chkp_parse_array_and_component_ref): Ditto.
22022 (chkp_find_bounds_1): Ditto.
22023 (chkp_process_stmt): Don't generate bounds store for
22026 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22028 * predict.c (maybe_hot_bb_p): Do not check profile status.
22029 (maybe_hot_edge_p): Likewise.
22030 (probably_never_executed): Check for zero counts even if profile
22032 (unlikely_executed_edge_p): New function.
22033 (unlikely_executed_stmt_p): New function.
22034 (unlikely_executed_bb_p): New function.
22035 (set_even_probabilities): Use unlikely predicates.
22036 (combine_predictions_for_bb): Likewise.
22037 (predict_paths_for_bb): Likewise.
22038 (predict_paths_leading_to_edge): Likewise.
22039 (determine_unlikely_bbs): New function.
22040 (estimate_bb_frequencies): Use it.
22041 (compute_function_frequency): Use zero counts even if profile is
22043 * profile-count.h: Fix typo.
22045 2017-08-08 Julia Koval <julia.koval@intel.com>
22047 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
22048 _mm512_mask_cvtsepi16_storeu_epi8,
22049 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
22050 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
22051 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
22052 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
22053 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
22054 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
22055 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
22056 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
22057 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
22058 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
22059 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
22060 __builtin_ia32_pmovuswb256mem_mask,
22061 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
22062 __builtin_ia32_pmovwb512mem_mask): New builtins.
22064 2017-08-08 Julia Koval <julia.koval@intel.com>
22066 PR target/73350,80862
22067 * config/i386/subst.md (round): Fix round pattern.
22068 * config/i386/i386.c (ix86_erase_embedded_rounding):
22069 Fix erasing rounding for the fixed pattern.
22071 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22073 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
22075 2017-06-08 Martin Liska <mliska@suse.cz>
22077 PR gcov-profile/80911
22078 * gcov.c (block_info::block_info): New constructor.
22080 2017-06-07 Carl Love <cel@us.ibm.com>
22082 * config/rs6000/rs6000-c: The return type of the following
22083 built-in functions was implemented as int not long long. Fix sign
22084 of return value for the unsigned version of vec_mulo and vec_mule.
22085 vector unsigned long long vec_bperm (vector unsigned long long,
22086 vector unsigned char)
22087 vector signed long long vec_mule (vector signed int,
22089 vector unsigned long long vec_mule (vector unsigned int,
22090 vector unsigned int)
22091 vector signed long long vec_mulo (vector signed int,
22093 vector unsigned long long vec_mulo (vector unsigned int,
22094 vector unsigned int)
22095 * doc/extend.texi: Fix the documentation for the built-in
22098 2017-06-07 Carl Love <cel@us.ibm.com>
22101 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
22104 2017-06-07 Carl Love <cel@us.ibm.com>
22106 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
22107 support, Generate doublehv for signed int/float for BE case only.
22109 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
22111 * doc/invoke.texi (mcx16): Rewrite.
22113 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22115 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
22116 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
22117 *mov<mode>_softfloat, and an anonymous splitter): Use
22118 nonimmediate_operand instead of rs6000_nonimmediate_operand.
22120 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22122 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
22124 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
22125 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
22126 (rs6000_debug_reg_global): Adjust.
22127 (rs6000_init_hard_regno_mode_ok): Adjust.
22128 (rs6000_dbx_register_number): Adjust.
22129 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
22130 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
22131 Remove SPE_ACC and SPEFSCR.
22132 (REG_ALLOC_ORDER): Ditto.
22133 (FRAME_POINTER_REGNUM): Change to 111.
22134 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
22135 (REG_CLASS_NAMES): Ditto.
22136 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
22137 (REGISTER_NAMES): Ditto.
22138 (ADDITIONAL_REG_NAMES): Ditto.
22139 (rs6000_reg_names): Ditto.
22140 * config/rs6000/rs6000.md: Renumber some register number
22143 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22145 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
22147 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
22148 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
22150 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
22151 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
22152 Delete the SPE high registers.
22153 (REG_ALLOC_ORDER): Ditto.
22154 (enum reg_class): Remove SPE_HIGH_REGS.
22155 (REG_CLASS_NAMES): Ditto.
22156 (REG_CLASS_CONTENTS): Delete the SPE high registers.
22157 (REGISTER_NAMES): Ditto.
22158 (rs6000_reg_names): Ditto.
22159 * doc/tm.texi.in: Remove SPE as example.
22160 * doc/tm.texi: Regenerate.
22162 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22164 * config/rs6000/8540.md (ppc8540_brinc): Delete.
22165 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
22166 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
22167 * config/rs6000/rs6000.md (type): Remove "brinc".
22169 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22171 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
22172 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
22173 * config/rs6000/linuxspe.h: Delete file.
22174 * config/rs6000/rs6000.md: Don't include spe.md.
22175 * config/rs6000/spe.h: Delete file.
22176 * config/rs6000/spe.md: Delete file.
22177 * config/rs6000/t-rs6000: Remove spe.md.
22179 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22181 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
22182 (reg_or_none500mem_operand): Delete.
22183 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
22184 instead of reg_or_none500mem_operand.
22186 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22188 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
22189 handling of SPE flags.
22190 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
22192 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22194 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
22196 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
22197 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
22198 paired_divv2sf3): Similar.
22199 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
22200 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
22201 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
22203 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
22204 Rename the paired_* instruction patterns.
22205 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
22207 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
22208 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
22209 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
22210 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
22211 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
22212 PAIRED_VECTOR_MODE.
22213 (struct machine_function): Delete field spe_insn_chain_scanned_p.
22214 (spe_func_has_64bit_regs_p): Delete.
22215 (spe_expand_predicate_builtin): Delete.
22216 (spe_expand_evsel_builtin): Delete.
22217 (TARGET_DWARF_REGISTER_SPAN): Do not define.
22218 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
22219 (invalid_e500_subreg): Delete.
22220 (rs6000_legitimize_address): Always force_reg op2 as well, for
22221 paired single memory accesses.
22222 (rs6000_member_type_forces_blk): Delete.
22223 (rs6000_spe_function_arg): Delete.
22224 (rs6000_expand_unop_builtin): Delete SPE handling.
22225 (rs6000_expand_binop_builtin): Ditto.
22226 (spe_expand_stv_builtin): Delete.
22227 (bdesc_2arg_spe): Delete.
22228 (spe_expand_builtin): Delete.
22229 (spe_expand_predicate_builtin): Delete.
22230 (spe_expand_evsel_builtin): Delete.
22231 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
22232 (spe_init_builtins): Delete.
22233 (spe_func_has_64bit_regs_p): Delete.
22234 (savres_routine_name): Delete "info" parameter. Adjust callers.
22235 (rs6000_emit_stack_reset): Ditto.
22236 (rs6000_dwarf_register_span): Delete.
22237 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
22238 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
22239 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
22241 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
22243 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
22244 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
22245 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
22246 mulv2sf3, divv2sf3): Delete expanders.
22248 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22250 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
22252 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22254 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
22255 * config/rs6000/rs6000.c: Ditto.
22257 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22259 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
22260 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
22261 comparison_operator.
22263 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22265 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
22266 * config/rs6000/rs6000.opt: Ditto.
22267 * config/rs6000/t-rtems: Ditto.
22269 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22271 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
22272 TARGET_E500_SINGLE by 0, simplify.
22273 * config/rs6000/rs6000.c: Ditto.
22274 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
22275 (spe_build_register_parallel): Delete.
22276 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
22277 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
22278 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
22279 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
22280 (E500_CONVERT): Delete.
22281 * config/rs6000/spe.md: Remove many patterns and all define_constants.
22283 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22285 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
22286 * config/rs6000/dfp.md: Ditto.
22287 (negdd2, *negdd2_fpr): Merge.
22288 (absdd2, *absdd2_fpr): Merge.
22289 (negtd2, *negtd2_fpr): Merge.
22290 (abstd2, *abstd2_fpr): Merge.
22291 * config/rs6000/e500.h: Delete file.
22292 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
22293 TARGET_FPRS by 1 and simplify.
22294 * config/rs6000/rs6000-c.c: Ditto.
22295 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
22296 TARGET_DF_SPE by 0.
22297 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
22299 * config/rs6000/rs6000.md: Ditto.
22300 (floatdidf2, *floatdidf2_fpr): Merge.
22301 (move_from_CR_gt_bit): Delete.
22302 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
22303 (E500_CR_IOR_COMPARE): Delete.
22304 (All patterns that require !TARGET_FPRS): Delete.
22305 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
22307 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22309 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
22311 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22313 * graphds.c (add_edge): Intitialize edge's attached data.
22314 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
22315 pointer parameter. Call pointed function on each edge during
22316 graph traversing. Skip traversing the edge when the function
22318 (graphds_dfs, graphds_scc): Ditto.
22319 (for_each_edge): New parameter. Pass the new parameter to callback
22321 * graphds.h (skip_edge_callback): New function pointer type.
22322 (graphds_dfs, graphds_scc): New function pointer parameter.
22323 (graphds_edge_callback, for_each_edge): New parameter.
22325 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22327 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
22328 out code checking if runtime alias check is possible to below ...
22329 Call the new function.
22330 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
22331 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
22333 2017-06-07 Marek Polacek <polacek@redhat.com>
22336 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
22337 TYPE_OVERFLOW_WRAPS check.
22339 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22341 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
22342 if versioning is required.
22343 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
22344 peeling with the check for versioning.
22346 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22348 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
22349 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
22350 Set true to new parameter if new ssa variable is defined.
22351 (vect_gen_vector_loop_niters): Refactor. Set range information
22352 for the new vector loop bound variable.
22353 (vect_do_peeling): Ditto.
22355 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22357 * tree-affine.c (ssa.h): Include header file.
22358 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
22359 has wrapping overflow behavior.
22361 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22363 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
22365 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22367 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
22368 (tree_to_aff_combination): ... here.
22370 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22372 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
22373 reg_pressure model function.
22374 (ivopts_global_cost_for_size): Delete.
22375 (determine_set_costs, iv_ca_recount_cost): Call new model function
22376 ivopts_estimate_reg_pressure.
22378 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22380 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
22381 expensive than udiv. Remove floating point cases from mod.
22383 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22385 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
22386 Increase idiv cost.
22388 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22390 * config/aarch64/aarch64.md
22391 (copysignsf3): Fix mask generation.
22393 2017-06-07 Jakub Jelinek <jakub@redhat.com>
22395 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
22397 (class dump_manager): Add register_dumps method.
22398 * dumpfile.c: Include langhooks.h.
22399 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
22400 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
22401 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
22402 (dump_manager::dump_register): Start with 512 entries instead of 32.
22403 (dump_manager::register_dumps): New method.
22404 * toplev.c (general_init): Instead of invoking register_dumps
22405 langhook, invoke register_dumps method on the dump manager.
22406 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
22409 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
22411 * doc/md.texi: Clarify the restrictions on a define_insn condition.
22412 Say that # requires an associated define_split to exist, and that
22413 the define_split must be suitable for use after register allocation.
22415 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22417 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
22418 (compute_outgoing_frequencies): Also initialize zero counts.
22419 (find_many_sub_basic_blocks): Do not produce uninitialized profile
22420 around loops; preserve more of profile when nothing changes.
22422 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
22424 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
22426 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
22427 * config/arm/arm-cpu-cdata.h: Regenerate.
22428 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
22429 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
22430 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
22431 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
22432 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
22434 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
22436 * config/arm/t-rmprofile: Likewise.
22437 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
22439 2017-06-06 David S. Miller <davem@davemloft.net>
22442 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
22443 blockage if function uses alloca.
22445 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22447 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
22448 New "uid" fields to hold pretty-print IDs of group and ref.
22449 Memory references are now identified as <group_id>:<ref_id>
22450 instead of using [random] addresses.
22451 (dump_mem_details): Simplify, no functional change.
22452 (dump_mem_ref): Simplify and make output more concise.
22453 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
22454 (find_or_create_group): Initialize group uid.
22455 (record_ref): Initialize ref uid. Improve debug output.
22456 (prune_group_by_reuse, should_issue_prefetch_p,)
22457 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
22458 (mark_nontemporal_store, determine_loop_nest_reuse):
22459 Improve debug output.
22461 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22463 * dbgcnt.def (prefetch): New debug counter.
22464 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
22465 (schedule_prefetches): Stop issueing prefetches if debug counter
22468 2017-06-06 Tom de Vries <tom@codesourcery.com>
22470 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
22471 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
22473 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22475 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
22476 Use aarch64_reg_or_zero predicate for operand 4.
22477 (aarch64_compare_and_swap<mode> define_insn_and_split):
22478 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
22479 (aarch64_store_exclusive<mode>): Likewise for operand 2.
22481 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
22483 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
22484 (arm_compute_save_core_reg_mask): This.
22485 (thumb1_compute_save_reg_mask): Rename into ...
22486 (thumb1_compute_save_core_reg_mask): This.
22487 (arm_compute_save_reg0_reg12_mask): Adapt comment.
22488 (arm_compute_frame_layout): Likewise.
22490 2017-06-06 Richard Biener <rguenther@suse.de>
22492 PR tree-optimization/80974
22493 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
22494 keep or clear leaders SSA info.
22496 2017-06-06 Tom de Vries <tom@codesourcery.com>
22498 * config/nvptx/nvptx.c (split_mode_p): New function.
22499 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
22501 2017-06-06 Tom de Vries <tom@codesourcery.com>
22503 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
22505 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22508 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
22511 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22513 * shrink-wrap.c (handle_simple_exit): Update profile.
22514 (try_shrink_wrapping): Upate profile.
22516 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22518 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
22519 (tree_guess_outgoing_edge_probabilities): New.
22520 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
22521 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
22523 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22525 * ipa-split.c (split_function): Initialize return bb profile.
22527 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22529 * profile.c (compute_branch_probabilities): Also initialize
22530 EXIT_BLOCK profile.
22532 2017-06-06 Richard Biener <rguenther@suse.de>
22534 PR tree-optimization/80928
22535 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
22536 (vect_analyze_loop_operations): Properly guard analysis for
22538 (vect_transform_loop): Likewise.
22539 (vect_analyze_loop_2): Also reset SLP type on PHIs.
22540 (vect_model_induction_cost): Do not cost for pure SLP.
22541 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
22542 of induction in inner loop vectorization.
22543 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
22544 (vect_get_and_check_slp_defs): Handle vect_induction_def.
22545 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
22547 (vect_analyze_slp_cost_1): Cost induction.
22548 (vect_detect_hybrid_slp_stmts): Handle PHIs.
22549 (vect_get_slp_vect_defs): Likewise.
22550 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
22551 (vect_transform_stmt): Handle SLP reductions.
22552 * tree-vectorizer.h (vectorizable_induction): Adjust.
22554 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
22556 * config/rs6000/rs6000.c (make_resolver_func): Update
22557 init_lowered_empty_function call.
22559 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
22561 * doc/invoke.texi: Document the -fprofile-abs-path option.
22562 * common.opt (fprofile-abs-path): New option.
22563 * gcov-io.h (gcov_write_filename): Declare.
22564 * gcov-io.c (gcov_write_filename): New function.
22565 * coverage.c (coverage_begin_function): Use gcov_write_filename.
22566 * profile.c (output_location): Likewise.
22568 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22570 * shring-wrap.c: Revert accidental commit.
22572 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
22574 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
22576 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22578 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
22580 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
22581 profile in callgraph edge.
22582 * profile-count.h (apply_probability): If THIS is 0, then result is 0
22583 (apply_scale): Likewise.
22584 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
22585 Also scale profile when inlining function with zero profile.
22586 (initialize_cfun): Update exit block profile even when it is zero.
22587 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
22588 when profile is read.
22590 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
22592 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
22593 (CLONE_*): New constants to define the processors we can generate
22594 code for with the target_clone attribute.
22595 (rs6000_clone_map): New array to identify which clone processors
22596 the current program is running on.
22597 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
22598 target_clone attribute.
22599 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
22600 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
22601 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
22602 (cpu_expand_builtin): Add support for target_clone attribute.
22603 (rs6000_valid_attribute_p): Allow "default" attribute.
22604 (get_decl_name): New debug function to simplify printing the
22605 current function name in debugging statements.
22606 (rs6000_clone_priority): New functions to support the target_clone
22607 attribute, and be able to generate code to switch between ISA 2.05
22608 through ISA 3.0 (power6 through power9).
22609 (rs6000_compare_version_priority): Likewise.
22610 (rs6000_get_function_versions_dispatcher): Likewise.
22611 (make_resolver_func): Likewise.
22612 (add_condition_to_bb): Likewise.
22613 (dispatch_function_versions): Likewise.
22614 (rs6000_generate_version_dispatcher_body): Likewise.
22615 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
22616 (fusion_gpr_load_p): Fix a spacing issue.
22617 * doc/extend.texi (Common Function Attributes): Document that the
22618 PowerPC supports the target_clone attribute.
22620 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
22622 * config/arm/arm.h: explain F symbol found in description of ARM
22623 register allocation in its legend.
22625 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22627 * config/mips/frame-header-opt.c: Include profile-count.h.
22628 * config/riscv/riscv.c: Include profile-count.h
22630 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22632 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
22634 (sm_set_flag_if_changed): Add bbs field.
22635 (execute_sm_if_changed_flag_set): Pass BBS.
22636 (execute_sm): Update.
22638 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22640 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
22643 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22645 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
22646 (peephole2): New peephole2 to emit the above.
22647 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
22649 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22651 * config/aarch64/aarch64.c (define_peephole2 above
22652 *sub_<shift>_<mode>): New peephole.
22654 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22656 * config/i386/i386.c (make_resolver_func): Update.
22657 * Makefile.in: Add profile-count.h and profile-count.o
22658 * auto-profile.c (afdo_indirect_call): Update to new API.
22659 (afdo_set_bb_count): Update.
22660 (afdo_propagate_edge): Update.
22661 (afdo_propagate_circuit): Update.
22662 (afdo_calculate_branch_prob): Update.
22663 (afdo_annotate_cfg): Update.
22664 * basic-block.h: Include profile-count.h
22665 (struct edge_def): Turn count to profile_count.
22666 (struct basic_block_def): Likewie.
22667 (REG_BR_PROB_BASE): Move to profile-count.h
22668 (RDIV): Move to profile-count.h
22669 * bb-reorder.c (max_entry_count): Turn to profile_count.
22670 (find_traces): Update.
22671 (rotate_loop):Update.
22672 (connect_traces):Update.
22673 (sanitize_hot_paths):Update.
22674 * bt-load.c (migrate_btr_defs): Update.
22675 * cfg.c (RDIV): Remove.
22676 (init_flow): Use alloc_block.
22677 (alloc_block): Uninitialize count.
22678 (unchecked_make_edge): Uninitialize count.
22679 (check_bb_profile): Update.
22680 (dump_edge_info): Update.
22681 (dump_bb_info): Update.
22682 (update_bb_profile_for_threading): Update.
22683 (scale_bbs_frequencies_int): Update.
22684 (scale_bbs_frequencies_gcov_type): Update.
22685 (scale_bbs_frequencies_profile_count): New.
22686 * cfg.h (update_bb_profile_for_threading): Update.
22687 (scale_bbs_frequencies_profile_count): Declare.
22688 * cfgbuild.c (compute_outgoing_frequencies): Update.
22689 (find_many_sub_basic_blocks): Update.
22690 * cfgcleanup.c (try_forward_edges): Update.
22691 (try_crossjump_to_edge): Update.
22692 * cfgexpand.c (expand_gimple_tailcall): Update.
22693 (construct_exit_block): Update.
22694 * cfghooks.c (verify_flow_info): Update.
22695 (dump_bb_for_graph): Update.
22696 (split_edge): Update.
22697 (make_forwarder_block): Update.
22698 (duplicate_block): Update.
22699 (account_profile_record): Update.
22700 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
22701 (get_estimated_loop_iterations): Update.
22702 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
22703 (single_likely_exit): Update.
22704 * cfgloopmanip.c (scale_loop_profile): Update.
22706 (set_zero_probability): Update.
22707 (lv_adjust_loop_entry_edge): Update.
22708 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
22709 (purge_dead_edges): Update.
22710 (rtl_account_profile_record): Update.
22711 * cgraph.c (cgraph_node::create): Uninitialize count.
22712 (symbol_table::create_edge): Uninitialize count.
22713 (cgraph_update_edges_for_call_stmt_node): Update.
22714 (cgraph_edge::dump_edge_flags): Update.
22715 (cgraph_node::dump): Update.
22716 (cgraph_edge::maybe_hot_p): Update.
22717 * cgraph.h: Include profile-count.h
22718 (create_clone), create_edge, create_indirect_edge): Update.
22719 (cgraph_node): Turn count to profile_count.
22720 (cgraph_edge0: Likewise.
22721 (make_speculative, clone): Update.
22722 (create_edge): Update.
22723 (init_lowered_empty_function): Update.
22724 * cgraphclones.c (cgraph_edge::clone): Update.
22725 (duplicate_thunk_for_node): Update.
22726 (cgraph_node::create_clone): Update.
22727 * cgraphunit.c (cgraph_node::analyze): Update.
22728 (cgraph_node::expand_thunk): Update.
22729 * final.c (dump_basic_block_info): Update.
22730 * gimple-streamer-in.c (input_bb): Update.
22731 * gimple-streamer-out.c (output_bb): Update.
22732 * graphite.c (print_global_statistics): Update.
22733 (print_graphite_scop_statistics): Update.
22734 * hsa-brig.c: Include basic-block.h.
22735 * hsa-dump.c: Include basic-block.h.
22736 * hsa-gen.c (T sum_slice): Update.
22737 (convert_switch_statements):Update.
22738 * hsa-regalloc.c: Include basic-block.h.
22739 * ipa-chkp.c (chkp_produce_thunks): Update.
22740 * ipa-cp.c (struct caller_statistics): Update.
22741 (init_caller_stats): Update.
22742 (gather_caller_stats): Update.
22743 (ipcp_cloning_candidate_p): Update.
22744 (good_cloning_opportunity_p): Update.
22745 (get_info_about_necessary_edges): Update.
22746 (dump_profile_updates): Update.
22747 (update_profiling_info): Update.
22748 (update_specialized_profile): Update.
22749 (perhaps_add_new_callers): Update.
22750 (decide_about_value): Update.
22751 (ipa_cp_c_finalize): Update.
22752 * ipa-devirt.c (struct odr_type_warn_count): Update.
22753 (struct decl_warn_count): Update.
22754 (struct final_warning_record): Update.
22755 (possible_polymorphic_call_targets): Update.
22756 (ipa_devirt): Update.
22757 * ipa-fnsummary.c (redirect_to_unreachable): Update.
22758 * ipa-icf.c (sem_function::merge): Update.
22759 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
22760 * ipa-inline.c (compute_uninlined_call_time): Update.
22761 (compute_inlined_call_time): Update.
22762 (want_inline_small_function_p): Update.
22763 (want_inline_self_recursive_call_p): Update.
22764 (edge_badness): Update.
22765 (lookup_recursive_calls): Update.
22766 (recursive_inlining): Update.
22767 (inline_small_functions): Update.
22768 (dump_overall_stats): Update.
22769 (dump_inline_stats): Update.
22770 * ipa-profile.c (ipa_profile_generate_summary): Update.
22771 (ipa_propagate_frequency): Update.
22772 (ipa_profile): Update.
22773 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22774 * ipa-utils.c (ipa_merge_profiles): Update.
22775 * loop-doloop.c (doloop_modify): Update.
22776 * loop-unroll.c (report_unroll): Update.
22777 (unroll_loop_runtime_iterations): Update.
22778 * lto-cgraph.c (lto_output_edge): Update.
22779 (lto_output_node): Update.
22780 (input_node): Update.
22781 (input_edge): Update.
22782 (merge_profile_summaries): Update.
22783 * lto-streamer-in.c (input_cfg): Update.
22784 * lto-streamer-out.c (output_cfg): Update.
22785 * mcf.c (create_fixup_graph): Update.
22786 (adjust_cfg_counts): Update.
22787 (sum_edge_counts): Update.
22788 * modulo-sched.c (sms_schedule): Update.
22789 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
22790 * predict.c (maybe_hot_count_p): Update.
22791 (probably_never_executed): Update.
22792 (dump_prediction): Update.
22793 (combine_predictions_for_bb): Update.
22794 (propagate_freq): Update.
22795 (handle_missing_profiles): Update.
22796 (counts_to_freqs): Update.
22797 (rebuild_frequencies): Update.
22798 (force_edge_cold): Update.
22799 * predict.h: Include profile-count.h
22800 (maybe_hot_count_p, counts_to_freqs): UPdate.
22801 * print-rtl-function.c: Do not include cfg.h
22802 * print-rtl.c: Include basic-block.h
22803 * profile-count.c: New file.
22804 * profile-count.h: New file.
22805 * profile.c (is_edge_inconsistent): Update.
22806 (correct_negative_edge_counts): Update.
22807 (is_inconsistent): Update.
22808 (set_bb_counts): Update.
22809 (read_profile_edge_counts): Update.
22810 (compute_frequency_overlap): Update.
22811 (compute_branch_probabilities): Update; Initialize and deinitialize
22813 (branch_prob): Update.
22814 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
22815 (edge_gcov_count): New.
22816 (bb_gcov_count): New.
22817 * shrink-wrap.c (try_shrink_wrapping): Update.
22818 * tracer.c (better_p): Update.
22819 * trans-mem.c (expand_transaction): Update.
22820 (ipa_tm_insert_irr_call): Update.
22821 (ipa_tm_insert_gettmclone_call): Update.
22822 * tree-call-cdce.c: Update.
22823 * tree-cfg.c (gimple_duplicate_sese_region): Update.
22824 (gimple_duplicate_sese_tail): Update.
22825 (gimple_account_profile_record): Update.
22826 (execute_fixup_cfg): Update.
22827 * tree-inline.c (copy_bb): Update.
22828 (copy_edges_for_bb): Update.
22829 (initialize_cfun): Update.
22830 (freqs_to_counts): Update.
22831 (copy_cfg_body): Update.
22832 (expand_call_inline): Update.
22833 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
22834 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
22835 (try_unroll_loop_completely): Update.
22836 (try_peel_loop): Update.
22837 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
22838 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
22839 * tree-ssa-loop-split.c (connect_loops): Update.
22840 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
22841 * tree-ssa-reassoc.c (branch_fixup): Update.
22842 * tree-ssa-tail-merge.c (replace_block_by): Update.
22843 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
22844 (compute_path_counts): Update.
22845 (update_profile): Update.
22846 (recompute_probabilities): Update.
22847 (update_joiner_offpath_counts): Update.
22848 (estimated_freqs_path): Update.
22849 (freqs_to_counts_path): Update.
22850 (clear_counts_path): Update.
22851 (ssa_fix_duplicate_block_edges): Update.
22852 (duplicate_thread_path): Update.
22853 * tree-switch-conversion.c (case_bit_test_cmp): Update.
22854 (struct switch_conv_info): Update.
22855 * tree-tailcall.c (decrease_profile): Update.
22856 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
22857 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
22858 * value-prof.c (check_counter): Update.
22859 (gimple_divmod_fixed_value): Update.
22860 (gimple_mod_pow2): Update.
22861 (gimple_mod_subtract): Update.
22862 (gimple_ic_transform): Update.
22863 (gimple_stringop_fixed_value): Update.
22864 * value-prof.h (gimple_ic): Update.
22866 2017-06-02 Carl Love <cel@us.ibm.com>
22868 * config/rs6000/rs6000-c: Add support for built-in functions
22869 vector double vec_doublee (vector signed int);
22870 vector double vec_doublee (vector unsigned int);
22871 vector double vec_doublee (vector float);
22872 vector double vec_doubleh (vector signed int);
22873 vector double vec_doubleh (vector unsigned int);
22874 vector double vec_doubleh (vector float);
22875 vector double vec_doublel (vector signed int);
22876 vector double vec_doublel (vector unsigned int);
22877 vector double vec_doublel (vector float);
22878 vector double vec_doubleo (vector signed int);
22879 vector double vec_doubleo (vector unsigned int);
22880 vector double vec_doubleo (vector float);.
22881 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
22882 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
22884 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
22885 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
22886 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
22888 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
22889 vec_doublel, vec_doubleh.
22890 * doc/extend.texi: Update the built-in documentation file for the
22891 new built-in functions.
22893 2017-06-02 David Malcolm <dmalcolm@redhat.com>
22896 * ipa-inline-analysis.c (free_growth_caches): Set
22897 edge_removal_hook_holder to NULL after removing it.
22899 2017-06-02 Sudakshina Das <sudi.das@arm.com>
22901 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
22902 comparision with zero.
22904 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
22905 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
22906 for early expansion of vec_min and vec_max builtins.
22907 (builtin_function_type): Add min/max unsigned variants to those
22908 identified as having unsigned arguments.
22910 2017-06-02 Olivier Hainque <hainque@adacore.com>
22912 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
22914 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22916 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
22917 Use VALL_F16 iterator rather than VALL.
22919 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22921 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
22922 Emit CBNZ inside loop when doing a strong exchange and comparing
22923 against zero. Generate the CC flags after the loop.
22925 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
22927 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
22928 (dl_section_ref): New.
22929 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
22930 On AIX, append an expression to subtract the size of the
22931 section length to dl_section_ref.
22933 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
22935 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
22936 for early expansion of vector absolute builtins.
22938 2017-06-02 Richard Biener <rguenther@suse.de>
22940 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
22941 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
22943 2017-06-02 Richard Biener <rguenther@suse.de>
22945 PR tree-optimization/80948
22946 * tree-tailcall.c (find_tail_calls): Track stmts to move in
22947 stmt order as well.
22949 2017-06-02 Richard Biener <rguenther@suse.de>
22951 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
22953 * tree-vect-stmts.c (process_use): Do not mark backedge defs
22954 for inductions as relevant.
22956 2017-06-02 Richard Biener <rguenther@suse.de>
22958 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
22959 (vectorizable_induction): ... this. Remove dead code.
22961 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
22963 * builtins. (expand_builtin_alloca): Remove second parameter and
22964 infer its value from the first parameter instead.
22965 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
22967 2017-06-02 Jakub Jelinek <jakub@redhat.com>
22969 PR rtl-optimization/80903
22970 * loop-doloop.c (add_test): Unshare sequence.
22972 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
22974 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
22976 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
22978 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
22980 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
22981 xlogue_layout::get_instance, logue_layout::xlogue_layout,
22982 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
22983 (xlogue_layout::get_stub_rtx): Make static.
22984 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
22985 (xlogue_layout::compute_stub_managed_regs): Rename to...
22986 (xlogue_layout::count_stub_managed_regs): ...this.
22987 (xlogue_layout::is_stub_managed_reg): New function.
22988 (xlogue_layout::m_stub_names): Rename to...
22989 (xlogue_layout::s_stub_names): ...this, make static.
22990 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
22991 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
22992 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
22993 xlogue_layout::s_stub_names): Instantiate statics.
22994 (stub_managed_regs): Remove.
22995 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
22996 (disable_call_ms2sysv_xlogues): Rename to...
22997 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
22998 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
23000 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
23001 change after reload_completed.
23002 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
23004 (ix86_expand_prologue): Likewise.
23005 (ix86_expand_epilogue): Likewise.
23006 (ix86_expand_split_stack_prologue): Likewise.
23007 (ix86_compute_frame_layout): Remove frame parameter ...
23008 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
23009 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
23011 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
23012 (ix86_frame): Move from here ...
23013 * config/i386/i386.h (ix86_frame): ... to here.
23014 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
23015 complete ix86_frame data structure instead. Remove some_ld_name.
23017 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
23019 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
23020 symbols that hold a DECL_VALUE_EXPR.
23022 2017-06-01 Martin Jambor <mjambor@suse.cz>
23024 PR tree-optimization/80898
23025 * tree-sra.c (process_subtree_disqualification): Removed.
23026 (disqualify_candidate): Do not acll
23027 process_subtree_disqualification.
23028 (subtree_mark_written_and_enqueue): New function.
23029 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
23030 RHS has been disqualified and re-queue LHS if necessary. Apart
23031 from that, ignore disqualified RHS.
23033 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23035 * config/s390/s390.c (s390_emit_epilogue): Disable early return
23036 address fetch for z10 or later.
23038 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23040 * config/arc/arc.md (tst_movb): Add guard when splitting.
23042 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23044 * config/arc/arc.c (arc_can_eliminate): Test against
23045 arc_frame_pointer_needed.
23047 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23049 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
23050 to prevent store reordering.
23051 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
23052 (type): Add block type.
23053 (stack_tie): Define special instruction to be used in
23056 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23058 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
23059 constraint. It is not valid for the pattern.
23060 (noncommutative_binary_comparison): Likewise.
23062 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23064 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
23067 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23069 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
23070 be used by the reg-alloc.
23072 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23074 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
23075 reg-alloc when having mul64 or mul32x16 instructions.
23076 (mulsidi3): Likewise.
23077 (umulsidi3): Likewise.
23078 (mulsi32x16): New pattern.
23079 (mulsi64): Likewise.
23080 (mulsidi64): Likewise.
23081 (umulsidi64): Likewise.
23082 (MUL32x16_REG): Define.
23083 (mul64_600): Use MUL32x16_REG.
23084 (mac64_600): Likewise.
23085 (umul64_600): Likewise.
23086 (umac64_600): Likewise.
23088 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23090 * config/arc/arc.md (mulsi3_700): Make it commutative.
23092 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
23094 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
23096 (*sign_extendsidi2_insn): Likewise for movstosw.
23098 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
23100 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
23101 the type of the input discriminant value. Convert the
23102 discriminant value of signedness vary.
23104 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
23106 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
23107 Add to -Wall section.
23109 2017-06-01 Richard Biener <rguenther@suse.de>
23111 PR middle-end/66313
23112 * fold-const.c (fold_plusminus_mult_expr): If the factored
23113 factor may be zero use a wrapping type for the inner operation.
23114 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
23115 and handle moved defs.
23116 (process_assignment): Properly guard the unary op case. Return a
23117 tri-state indicating that moving the stmt before the call may allow
23118 to continue. Pass through to_move.
23119 (find_tail_calls): Handle moving unrelated defs before
23122 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
23125 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
23126 splitter result in the canonical way.
23128 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
23130 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
23131 also for 32bit target. Update insn attributes.
23132 (zero-extendsidi2 splitter): Allow all registers for operand 1.
23134 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
23136 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
23137 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
23138 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
23139 (_mm_maskz_min_ss): New intrinsics.
23141 2017-05-31 Martin Liska <mliska@suse.cz>
23143 * tree-vect-loop.c (vect_create_epilog_for_reduction):
23144 Change comment style to one we normally use.
23145 (vectorizable_reduction): Likewise.
23146 (vectorizable_induction): Likewise.
23147 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
23148 (vectorizable_call): Likewise.
23149 (vectorizable_simd_clone_call): Likewise.
23150 (vectorizable_conversion): Likewise.
23151 (vectorizable_assignment): Likewise.
23152 (vectorizable_shift): Likewise.
23153 (vectorizable_operation): Likewise.
23154 (vectorizable_store): Likewise.
23155 (vectorizable_load): Likewise.
23156 * tree-vectorizer.h: Likewise.
23158 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
23160 * passes.c (emergency_dump_function): New.
23161 * tree-pass.h (emergency_dump_function): Declare.
23162 * plugin.c (plugins_internal_error_function): Remove.
23163 * plugin.h (plugins_internal_error_function): Remove declaration.
23164 * toplev.c (internal_error_function): New static function. Use it...
23165 (general_init): ...here.
23167 2017-05-31 Graham Markall <graham.markall@embecosm.com>
23169 * config/arc/arc.c (arc_print_operand): Handle constant operands.
23170 (arc_rtx_costs): Add costs for new patterns.
23171 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
23172 * config/arc/predicates.md: Add _1_2_3_operand predicate.
23174 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
23176 * tree-ssa-strlen.c (get_next_strinfo): New function.
23177 (get_stridx_plus_constant): Use it.
23178 (zero_length_string): Likewise.
23179 (adjust_related_strinfos): Likewise.
23180 (adjust_last_stmt): Likewise.
23182 2017-05-31 Richard Biener <rguenther@suse.de>
23185 * config/i386/i386.c (ix86_expand_builtin): Remove assert
23186 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
23188 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
23190 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
23191 loop_vinfo argument and use of dependence distance vectors.
23192 Check instead whether the two references differ only in their
23193 initial value and assume that they have the same alignment if the
23194 difference is a multiple of the vector alignment.
23195 (vect_analyze_data_refs_alignment): Update call accordingly.
23197 2017-05-31 Martin Liska <mliska@suse.cz>
23200 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
23202 2017-05-31 Bin Cheng <bin.cheng@arm.com>
23204 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
23205 (create_intersect_range_checks): Move from ...
23206 * tree-data-ref.c (create_intersect_range_checks_index)
23207 (create_intersect_range_checks): ... to here.
23208 (create_runtime_alias_checks): New function factored from ...
23209 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
23210 here. Call above function.
23211 * tree-data-ref.h (create_runtime_alias_checks): New function.
23213 2017-05-31 Bin Cheng <bin.cheng@arm.com>
23215 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
23216 segment length for dr_b and compute it in wide_int.
23218 2017-05-31 Richard Biener <rguenther@suse.de>
23220 PR tree-optimization/80906
23221 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
23222 and pass through iv_map.
23223 (copy_bb_and_scalar_dependences): Adjust.
23224 (translate_pending_phi_nodes): Likewise.
23225 (copy_loop_close_phi_args): Handle code-generating IVs instead
23228 2017-05-30 David Malcolm <dmalcolm@redhat.com>
23230 * diagnostic-color.c (color_dict): Add "type-diff".
23231 (parse_gcc_colors): Update comment.
23232 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
23233 -fdiagnostics-show-template-tree and -fno-elide-type.
23234 (GCC_COLORS): Add type-diff to example.
23236 (-fdiagnostics-show-template-tree): New.
23237 (-fno-elide-type): New.
23238 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
23239 the pp_format_decoder callback. Call any m_format_postprocessor's
23241 (pretty_printer::pretty_printer): Initialize
23242 m_format_postprocessor.
23243 (pretty_printer::~pretty_printer): Delete any
23244 m_format_postprocessor.
23245 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
23246 (class format_postprocessor): New class.
23247 (struct pretty_printer::format_decoder): Document the new parameters.
23248 (struct pretty_printer::m_format_postprocessor): New field.
23249 * tree-diagnostic.c (default_tree_printer): Update for new
23250 bool and const char ** params.
23251 * tree-diagnostic.h (default_tree_printer): Likewise.
23253 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
23255 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
23256 (lwa_operand): Delete rs6000_gen_cell_microcode test.
23257 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
23258 rs6000_gen_cell_microcode code.
23259 (rs6000_final_prescan_insn): Delete.
23260 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
23261 "warn-cell-microcode" entries.
23262 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
23263 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
23264 throughout. Change cc_reg_not_micro_cr0_operand to
23265 cc_reg_not_cr0_operand throughout.
23266 (*extendhi<mode>2_noload): Delete.
23267 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
23268 (mwarn-cell-microcode): Delete.
23269 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
23270 -mgen-cell-microcode and -mwarn-cell-microcode.
23272 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
23275 * config/i386/constraints.md (Yd): New constraint.
23277 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
23278 and (?Yd, r) alternatives. Update insn attributes.
23279 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
23280 and (?*Yd, r) alternatives. Update insn attributes.
23281 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
23283 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
23285 * gimplify.c (gimplify_modify_expr): Don't create a
23286 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
23289 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
23291 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
23293 2017-05-30 Richard Biener <rguenther@suse.de>
23295 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
23296 and reduc_def fields.
23297 (STMT_VINFO_REDUC_TYPE): New define.
23298 (STMT_VINFO_REDUC_DEF): Likewise.
23299 (vect_force_simple_reduction): Adjust prototype.
23300 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
23301 (vect_is_simple_reduction): Remove check_reduction argument.
23302 (vect_force_simple_reduction): Adjust and set
23303 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
23304 (vectorizable_reduction): Do not re-do reduction analysis
23305 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
23306 * tree-parloops.c (gather_scalar_reductions): Adjust.
23308 2017-05-30 Richard Biener <rguenther@suse.de>
23310 PR middle-end/80901
23311 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
23314 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
23316 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
23317 Introduce unknown_misalignment parameter and remove vf.
23318 (vect_peeling_hash_get_lowest_cost):
23319 Pass unknown_misalignment parameter.
23320 (vect_enhance_data_refs_alignment):
23321 Fix unsupportable data ref treatment.
23323 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23325 * tree-vect-data-refs.c (vect_get_data_access_cost):
23326 Workaround for SLP handling.
23327 (vect_enhance_data_refs_alignment):
23328 Compute costs for doing no peeling at all, compare to the best
23329 peeling costs so far and avoid peeling if cheaper.
23331 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23333 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
23334 Return peeling info and set costs to zero for unlimited cost
23336 (vect_enhance_data_refs_alignment): Also inspect all datarefs
23337 with unknown misalignment. Compute and costs for unknown
23338 misalignment, compare them to the costs for known misalignment
23339 and choose the cheapest for peeling.
23341 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23343 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
23344 (vect_get_peeling_costs_all_drs): Create function.
23345 (vect_peeling_hash_get_lowest_cost):
23346 Use vect_get_peeling_costs_all_drs.
23347 (vect_peeling_supportable): Create function.
23348 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
23350 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23352 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
23353 DR_HAS_NEGATIVE_STEP.
23354 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
23355 (vect_enhance_data_refs_alignment): Use.
23356 (vect_duplicate_ssa_name_ptr_info): Use.
23357 * tree-vectorizer.h (dr_misalignment): Use.
23358 (known_alignment_for_access_p): Use.
23360 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
23363 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
23365 (has_section_name): New function.
23367 2017-05-30 Martin Liska <mliska@suse.cz>
23370 * auto-profile.c (get_function_decl_from_block): Fix
23373 2017-05-30 Richard Biener <rguenther@suse.de>
23375 PR middle-end/80876
23376 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
23378 2017-05-30 Martin Liska <mliska@suse.cz>
23380 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
23381 * dumpfile.h (struct dump_file_info): Remove ctors.
23383 2017-05-30 Martin Liska <mliska@suse.cz>
23385 * predict.def: Fix GNU coding style.
23387 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
23389 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
23390 Mark 'to' argument with ATTRIBUTE_UNUSED.
23392 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
23394 * config/xtensa/xtensa.c (xtensa_emit_call): Use
23395 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
23396 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
23399 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
23401 * doc/install.texi (Options specification): Restore entry of
23402 --enable-sjlj-exceptions.
23404 2017-05-27 Michael Eager <eager@eagercon.com>
23407 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
23409 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
23411 * config/microblaze/microblaze.h
23412 (FIXED_REGISTERS): Update in macro.
23413 (CALL_USED_REGISTERS): Update in macro.
23415 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
23417 * doc/install.texi: Add links to macOS binary distributions.
23419 2017-05-27 Jakub Jelinek <jakub@redhat.com>
23423 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23425 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
23427 2017-05-26 Martin Liska <mliska@suse.cz>
23429 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
23431 2017-05-26 Martin Liska <mliska@suse.cz>
23433 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
23434 always leading ';; '.
23435 (dump_bb_info): Likewise.
23436 (brief_dump_cfg): Likewise.
23437 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
23438 * dumpfile.c: Remove usage of TDF_VERBOSE.
23439 * dumpfile.h (enum dump_kind): Likewise.
23440 (dump_gimple_bb_header): Do not use TDF_COMMENT.
23441 * print-tree.c (debug_verbose): Remove.
23442 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
23443 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
23444 * tree-diagnostic.c (default_tree_printer): Replace
23445 TDF_DIAGNOSTIC with TDF_SLIM.
23447 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23449 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
23450 in parameter loop, rather than loop_vinfo.
23451 (create_intersect_range_checks): Ditto.
23452 (vect_create_cond_for_alias_checks): Update call to above functions.
23454 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23456 PR tree-optimization/80815
23457 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
23458 for merging runtime alias checks. Handle negative DR_STEPs.
23460 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23462 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
23464 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
23465 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
23466 out code pruning runtime alias checks.
23467 * tree-data-ref.c (prune_runtime_alias_test_list): New function
23468 factored out from above.
23469 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
23471 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
23473 (prune_runtime_alias_test_list): New decalaration.
23475 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23477 * tree-vect-data-refs.c (compare_tree): Rename and move ...
23478 * tree-data-ref.c (data_ref_compare_tree): ... to here.
23479 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
23480 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
23481 (operator==, comp_dr_with_seg_len_pair): Ditto.
23482 (vect_prune_runtime_alias_test_list): Ditto.
23484 2017-05-26 Martin Liska <mliska@suse.cz>
23487 * params.def: Bound partial-inlining-entry-probability param.
23489 2017-05-26 Marek Polacek <polacek@redhat.com>
23492 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
23495 2017-05-26 Richard Biener <rguenther@suse.de>
23497 PR tree-optimization/80842
23498 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
23501 2017-05-26 Richard Biener <rguenther@suse.de>
23503 PR tree-optimization/80844
23504 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
23506 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
23508 * doc/md.texi (Machine Constraints): Update x86 family
23509 machine constraints section to match 'config/i386/constraints.md'.
23511 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
23513 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
23515 2017-05-25 Nathan Sidwell <nathan@acm.org>
23517 * doc/invoke.texi (--enable-languages): Update documentation.
23519 2017-05-25 Martin Liska <mliska@suse.cz>
23521 * dumpfile.c: Add TDF_FOLDING.
23522 * dumpfile.h (enum dump_kind): Likewise.
23523 * genmatch.c (dt_simplify::gen_1): Use it.
23525 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23527 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
23529 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23531 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
23532 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
23534 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23536 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
23537 * match.pd (X == C): Rewrite it here.
23538 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
23539 with_certain_nonzero_bits2): New predicates.
23540 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
23542 2017-05-24 Nathan Sidwell <nathan@acm.org>
23544 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
23547 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
23550 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
23552 * config/powerpcspe: New port. Files are copied from the rs6000
23553 port, with "rs6000" in filenames replaced by "powerpcspe".
23555 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
23557 PR rtl-optimization/80754
23558 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
23560 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
23562 * config/sparc/sparc.md (length): Return the correct value for -mflat
23563 sibcalls to match output_sibcall.
23565 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
23569 * config/rs6000/rs6000.c (struct machine_function): Add new field
23571 (rs6000_get_separate_components): Init that field, use it.
23572 (rs6000_components_for_bb): Use the field.
23574 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
23576 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
23578 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
23580 PR middle-end/80823
23581 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
23583 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23586 * config/s390/s390.c (s390_check_qrst_address): Check incoming
23587 address against address_operand predicate.
23588 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
23590 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
23592 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
23593 parameters passed indirectly.
23595 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
23597 * config/i386/i386.md (*movdi_internal): Remove SSE4
23598 alternative 18 (?r, *v). Update insn attributes.
23599 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
23600 Update insn attributes.
23601 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
23602 Update insn attributes.
23603 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
23604 alternative 1 (r, v). Remove isa attribute.
23605 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
23606 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
23607 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
23609 2017-05-23 Tom de Vries <tom@codesourcery.com>
23611 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
23614 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23616 * cgraphunit.c (symbol_table::process_new_functions): Update.
23617 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
23618 (inline_generate_summary): Rename to ...
23619 (ipa_fn_summary_generate): ... this one.
23620 (inline_read_summary): Rename to ...
23621 (ipa_fn_summary_read): ... this one.
23622 (inline_write_summary): Rename to ...
23623 (ipa_fn_summary_write): ... this one.
23624 (inline_free_summary): Rename to ...
23625 (ipa_free_fn_summary): ... this one.
23626 (pass_data_local_fn_summary, pass_local_fn_summary,
23627 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
23628 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
23629 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
23630 make_pass_ipa_fn_summary): New.
23631 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
23632 inline_write_summary, inline_free_summary): Remove.
23633 (ipa_free_fn_summary) : New.
23634 * ipa-inline.c (ipa_inline): Update.
23635 (pass_ipa_inline): Do not generate summaries.
23636 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
23638 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
23639 and add pass_ipa_fn_summary.
23640 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
23642 (make_pass_inline_parameters): Remove.
23644 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
23646 * omp-low.c (struct omp_context): Remove "default_kind" member.
23649 * omp-offload.c (execute_oacc_device_lower): Remove the
23650 parallelism dimensions function attributes for unparallelized
23651 OpenACC kernels constructs.
23653 2017-05-23 Martin Liska <mliska@suse.cz>
23655 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
23657 (cgraph_edge::make_speculative): Likewise.
23658 (cgraph_edge::resolve_speculation): Likewise.
23659 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
23660 (cgraph_node::dump): Likewise.
23661 * cgraph.h: Likewise.
23662 * cgraphunit.c (analyze_functions): Likewise.
23663 (symbol_table::compile): Likewise.
23664 * ipa-cp.c (print_all_lattices): Likewise.
23665 (determine_versionability): Likewise.
23666 (initialize_node_lattices): Likewise.
23667 (ipcp_verify_propagated_values): Likewise.
23668 (estimate_local_effects): Likewise.
23669 (update_profiling_info): Likewise.
23670 (create_specialized_node): Likewise.
23671 (perhaps_add_new_callers): Likewise.
23672 (decide_about_value): Likewise.
23673 (decide_whether_version_node): Likewise.
23674 (identify_dead_nodes): Likewise.
23675 (ipcp_store_bits_results): Likewise.
23676 * ipa-devirt.c (dump_targets): Likewise.
23677 (ipa_devirt): Likewise.
23678 * ipa-icf.c (sem_item::dump): Likewise.
23679 (sem_function::equals): Likewise.
23680 (sem_variable::equals): Likewise.
23681 (sem_item_optimizer::read_section): Likewise.
23682 (sem_item_optimizer::execute): Likewise.
23683 (congruence_class::dump): Likewise.
23684 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
23685 (dump_inline_summary): Likewise.
23686 (estimate_node_size_and_time): Likewise.
23687 (inline_analyze_function): Likewise.
23688 * ipa-inline-transform.c (inline_call): Likewise.
23689 * ipa-inline.c (report_inline_failed_reason): Likewise.
23690 (want_early_inline_function_p): Likewise.
23691 (edge_badness): Likewise.
23692 (update_edge_key): Likewise.
23693 (inline_small_functions): Likewise.
23694 * ipa-profile.c (ipa_profile): Likewise.
23695 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
23696 (ipa_make_edge_direct_to_target): Likewise.
23697 (remove_described_reference): Likewise.
23698 (ipa_impossible_devirt_target): Likewise.
23699 (propagate_controlled_uses): Likewise.
23700 (ipa_print_node_params): Likewise.
23701 (ipcp_transform_function): Likewise.
23702 * ipa-pure-const.c (pure_const_read_summary): Likewise.
23703 (propagate_pure_const): Likewise.
23704 * ipa-reference.c (generate_summary): Likewise.
23705 (read_write_all_from_decl): Likewise.
23706 (propagate): Likewise.
23707 (ipa_reference_read_optimization_summary): Likewise.
23708 * ipa-utils.c (ipa_merge_profiles): Likewise.
23709 * ipa.c (walk_polymorphic_call_targets): Likewise.
23710 (symbol_table::remove_unreachable_nodes): Likewise.
23711 (ipa_single_use): Likewise.
23712 * passes.c (execute_todo): Likewise.
23713 * predict.c (drop_profile): Likewise.
23714 * symtab.c (symtab_node::get_dump_name): New function.
23715 (symtab_node::dump_name): Likewise.
23716 (symtab_node::dump_asm_name): Likewise.
23717 (symtab_node::dump_references): Likewise.
23718 (symtab_node::dump_referring): Likewise.
23719 (symtab_node::dump_base): Likewise.
23720 (symtab_node::debug_symtab): Likewise.
23721 * tree-sra.c (convert_callers_for_node): Likewise.
23722 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
23723 * value-prof.c (init_node_map): Likewise.
23725 2017-05-23 Martin Liska <mliska@suse.cz>
23727 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
23728 and symtab_node::debug_symtab to symbol_table::debug.
23729 * cgraphunit.c (analyze_functions): Use the renamed function.
23730 (symbol_table::compile): Likewise.
23731 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
23732 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
23733 * passes.c (execute_todo): Likewise.
23734 * symtab.c (symbol_table::dump): New function.
23735 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
23737 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23739 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
23740 that nonconst implies exec.
23742 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23744 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
23745 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
23746 (inline_edge_summary_vec): Turn into ...
23747 (ipa_call_summaries): ... this one.
23748 (redirect_to_unreachable, edge_set_predicate,
23749 evaluate_properties_for_edge, inline_summary_alloc,
23750 reset_ipa_call_summary, reset_inline_summary,
23751 inline_summary_t::duplicate): Update.
23752 (inline_edge_duplication_hook): Turn to ...
23753 (ipa_call_summary_t::duplicate): ... this one.
23754 (inline_edge_removal_hook): Turn to ...
23755 (ipa_call_summary_t::remove): ... this one.
23756 (dump_inline_edge_summary): Turn to ...
23757 (dump_ipa_call_summary): ... this one.
23758 (estimate_function_body_sizes): Update.
23759 (inline_update_callee_summaries): Update.
23760 (remap_edge_change_prob): Update.
23761 (remap_edge_summaries): Update.
23762 (inline_merge_summary): Update.
23763 (do_estimate_edge_time): Update.
23764 (inline_generate_summary): Update.
23765 (inline_read_section): Update.
23766 (inline_read_summary): Update.
23767 (inline_free_summary): Update.
23768 * ipa-inline.c (can_inline_edge_p): Update.
23769 (compute_inlined_call_time): Update.
23770 (want_inline_small_function_p): Update.
23771 (edge_badness): Update.
23772 (early_inliner): Update.
23773 * ipa-inline.h (inline_edge_summary): Turn to ...
23774 (ipa_call_summary): ... this one.
23775 (ipa_call_summary_t): New class.
23776 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
23777 (ipa_call_summaries): New.
23778 (inline_edge_summary): Remove.
23779 (estimate_edge_growth): Update.
23780 * ipa-profile.c (ipa_propagate_frequency_1): Update.
23781 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
23782 * ipa-split.c (execute_split_functions): Update.
23783 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
23785 2017-05-23 Tom de Vries <tom@codesourcery.com>
23787 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
23788 attributes): Document rdrand effective target.
23790 2017-05-23 Tom de Vries <tom@codesourcery.com>
23792 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
23793 attributes): Sort alphabetically.
23795 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
23797 * config/avr/genmultilib.awk: Use gsub instead of gensub.
23799 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
23802 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
23803 V2DF/V2DI splat into two separate patterns, one that handles
23804 registers, and the other that only handles memory. Drop support
23805 for splatting from a GPR on ISA 2.07 and then splitting the
23806 splat into direct move and splat.
23807 (vsx_splat_<mode>_reg): Likewise.
23808 (vsx_splat_<mode>_mem): Likewise.
23810 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
23812 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
23814 2017-05-22 Jakub Jelinek <jakub@redhat.com>
23816 PR middle-end/80809
23817 * omp-low.c (finish_taskreg_remap): New function.
23818 (finish_taskreg_scan): If unit size of ctx->record_type
23819 is non-constant, unshare the size expression and replace
23820 decls in it with possible outer var refs.
23822 PR middle-end/80809
23823 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
23824 GOVD_SHARED rather than GOVD_PRIVATE with it.
23825 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
23826 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
23828 PR middle-end/80853
23829 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
23830 as last argument to build_outer_var_ref for pointer bases of array
23831 section reductions.
23833 2017-05-19 Martin Sebor <msebor@redhat.com>
23835 * print-tree.c (print_node): Print DECL_READ_P flag.
23837 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23839 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
23840 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
23841 * cgraph.c: Likewise.
23842 * cgraphunit.c: Likewise.
23843 * gengtype.c: Likewise.
23844 * ipa-cp.c: Likewise.
23845 * ipa-devirt.c: Likewise.
23846 * ipa-icf.c: Likewise.
23847 * ipa-predicate.c: Likewise.
23848 * ipa-profile.c: Likewise.
23849 * ipa-prop.c: Likewise.
23850 * ipa-split.c: Likewise.
23852 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
23853 edge_predicate_pool, dump_inline_hints,
23854 inline_summary::account_size_time, redirect_to_unreachable,
23855 edge_set_predicate, set_hint_predicate,
23856 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
23857 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
23858 inline_summary_t::remove, remap_hint_predicate_after_duplication,
23859 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
23860 ipa_call_summary_t::remove, initialize_growth_caches,
23861 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
23862 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
23863 mark_modified, unmodified_parm_1, unmodified_parm,
23864 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
23865 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
23866 compute_bb_predicates, will_be_nonconstant_expr_predicate,
23867 will_be_nonconstant_predicate, record_modified_bb_info,
23868 get_minimal_bb, record_modified, param_change_prob,
23869 phi_result_unknown_predicate, predicate_for_phi_result,
23870 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
23871 estimate_function_body_sizes, compute_inline_parameters,
23872 compute_inline_parameters_for_curren, pass_data_inline_parameters,
23873 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
23874 inline_update_callee_summaries, remap_edge_change_prob,
23875 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
23876 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
23877 inline_analyze_function, inline_summary_t::insert,
23878 inline_generate_summary, read_ipa_call_summary, inline_read_section,
23879 inline_read_summary, write_ipa_call_summary, inline_write_summary,
23880 inline_free_summary): Move to ipa-fnsummary.h
23881 (predicate_t): Remove.
23882 * ipa-fnsummary.c: New file.
23883 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
23884 (enum inline_hints_vals, inline_hints, agg_position_info,
23885 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
23886 inline_summaries, ipa_call_summary, ipa_call_summary_t,
23887 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
23888 dump_inline_summary, dump_inline_hints, inline_generate_summary,
23889 inline_read_summary, inline_write_summary, inline_free_summary,
23890 inline_analyze_function, initialize_inline_failed,
23891 inline_merge_summary, inline_update_overall_summary,
23892 compute_inline_parameters): Move to ipa-fnsummary.h
23893 * ipa-fnsummary.h: New file.
23894 * ipa-inline-transform.h: Include ipa-inline.h.
23895 * ipa-inline.c: LIkewise.
23897 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23899 * ipa-inline.c (edge_badness): Use inlined_time instead of
23900 inline_summaries->get.
23902 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23904 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
23906 2017-05-22 Nathan Sidwell <nathan@acm.org>
23908 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
23909 (fdump-lang): Document 'raw' option.
23910 * dumpfile.h (TDI_tu): Delete.
23911 * dumpfile.c (dump_files): Remove translation-unit.
23912 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
23914 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
23916 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
23917 command option from $(AWK) call.
23918 * config/avr/genmultilib.awk: Simplify and rewrite so that it
23919 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
23920 [FORMAT]: Remove handling of variable.
23921 * config/avr/t-multilib: Regenerate.
23923 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23925 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
23927 (dump_inline_summary): Do not print self_time.
23928 (estimate_function_body_sizes): Do not set self_time.
23929 (compute_inline_parameters): Likewise.
23930 (inline_read_section, inline_write_summary): Do not stream self_time.
23931 * ipa-inline.h (inline_summary): Drop self_time.
23933 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23935 * ipa-inline-analysis.c (account_size_time): Rename to ...
23936 (inline_summary::account_size_time): ... this one.
23937 (reset_ipa_call_summary): Turn to ...
23938 (ipa_call_summary::reset): ... this one.
23939 (reset_inline_summary): Turn to ...
23940 (inline_summary::reset): ... this one.
23941 (inline_summary_t::remove): Update.
23942 (inline_summary_t::duplicate): Update.
23943 (ipa_call_summary_t::remove): Update.
23944 (dump_inline_summary): Update.
23945 (estimate_function_body_sizes): Update.
23946 (compute_inline_parameters): Update.
23947 (estimate_node_size_and_time): Update.
23948 (inline_merge_summary): Update.
23949 (inline_update_overall_summary): Update.
23950 (inline_read_section): Update.
23951 (inline_write_summary): Update.
23952 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
23953 add account_size_time and reset member functions.
23954 (ipa_call_summary): Add reset function.
23955 * ipa-predicate.h (predicate::operator &): Constify.
23957 2017-05-22 Richard Biener <rguenther@suse.de>
23959 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
23961 2017-05-19 Jason Merrill <jason@redhat.com>
23963 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
23965 2017-05-19 Marek Polacek <polacek@redhat.com>
23968 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
23969 TYPE_OVERFLOW_WRAPS checks.
23971 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
23973 * tree-core.h (enum omp_clause_default_kind): Add
23974 "OMP_CLAUSE_DEFAULT_PRESENT".
23975 * tree-pretty-print.c (dump_omp_clause): Handle it.
23976 * gimplify.c (enum gimplify_omp_var_data): Add
23977 "GOVD_MAP_FORCE_PRESENT".
23978 (gimplify_adjust_omp_clauses_1): Map it to
23979 "GOMP_MAP_FORCE_PRESENT".
23980 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
23982 * gimplify.c (oacc_default_clause): Clarify.
23984 2017-05-19 Nathan Sidwell <nathan@acm.org>
23986 LANG_HOOK_REGISTER_DUMPS
23987 * toplev.c (general_init): Call register dump lang hook.
23988 * doc/invoke.texi: Document -fdump-lang option family.
23989 * dumpfile.c (dump_files): Remove class dump here.
23990 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
23991 * dumpfile.h (tree_dump_index): Remove TDI_class.
23992 * langhooks-def.h (lhd_register_dumps): Declare.
23993 (LANG_HOOKS_REGISTER_DUMPS): Define.
23994 (LANG_HOOKS_INITIALIZER): Add it.
23995 * langhooks.c (lhd_register_dumps): Define.
23996 * langhooks.h (struct lang_hooks): Add register_dumps.
23998 2017-05-19 Nathan Sidwell <nathan@acm.org>
24000 * context.h (context::set_passes): New.
24001 * context.c (context::context): Do not create pass manager.
24002 * toplev.c (general_init): Create pass manager here.
24004 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
24006 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
24007 use this splitter if two add or or instructions would also work for
24008 the constant we want to generate.
24010 2017-05-19 Richard Biener <rguenther@suse.de>
24013 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
24014 predicate evaluation.
24016 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
24018 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
24020 * ipa-inline.c (want_inline_small_function_p): Do not cast to
24023 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
24025 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
24026 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
24027 (inline_edge_summary_vec): Turn into ...
24028 (ipa_call_summaries): ... this one.
24029 (redirect_to_unreachable, edge_set_predicate,
24030 evaluate_properties_for_edge, inline_summary_alloc,
24031 reset_ipa_call_summary, reset_inline_summary,
24032 inline_summary_t::duplicate): Update.
24033 (inline_edge_duplication_hook): Turn to ...
24034 (ipa_call_summary_t::duplicate): ... this one.
24035 (inline_edge_removal_hook): Turn to ...
24036 (ipa_call_summary_t::remove): ... this one.
24037 (dump_inline_edge_summary): Turn to ...
24038 (dump_ipa_call_summary): ... this one.
24039 (estimate_function_body_sizes): Update.
24040 (inline_update_callee_summaries): Update.
24041 (remap_edge_change_prob): Update.
24042 (remap_edge_summaries): Update.
24043 (inline_merge_summary): Update.
24044 (do_estimate_edge_time): Update.
24045 (inline_generate_summary): Update.
24046 (inline_read_section): Update.
24047 (inline_read_summary): Update.
24048 (inline_free_summary): Update.
24049 * ipa-inline.c (can_inline_edge_p): Update.
24050 (compute_inlined_call_time): Update.
24051 (want_inline_small_function_p): Update.
24052 (edge_badness): Update.
24053 (early_inliner): Update.
24054 * ipa-inline.h (inline_edge_summary): Turn to ...
24055 (ipa_call_summary): ... this one.
24056 (ipa_call_summary_t): New class.
24057 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
24058 (ipa_call_summaries): New.
24059 (inline_edge_summary): Remove.
24060 (estimate_edge_growth): Update.
24061 * ipa-profile.c (ipa_propagate_frequency_1): Update.
24062 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
24063 * ipa-split.c (execute_split_functions): Update.
24064 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
24066 2017-05-19 Richard Biener <rguenther@suse.de>
24068 PR middle-end/80764
24069 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
24071 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
24073 * config/rs6000/rs6000.c (struct machine_function): Add field
24074 fpr_is_wrapped_separately.
24075 (rs6000_get_separate_components): Use 64 components. Handle the
24076 new FPR components.
24077 (rs6000_components_for_bb): Handle the FPR components.
24078 (rs6000_emit_prologue_components): Handle the FPR components.
24079 (rs6000_emit_epilogue_components): Handle the FPR components.
24080 (rs6000_set_handled_components): Handle the FPR components.
24081 (rs6000_emit_prologue): Don't output prologue code for those FPRs
24082 that are already separately shrink-wrapped.
24083 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
24084 that are already separately shrink-wrapped.
24086 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
24089 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
24092 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
24093 (define_peephole2 for Altivec d-form load): Add peepholes to catch
24094 cases where the register allocator uses a move and an offsettable
24095 memory operation to/from a FPR register on ISA 2.06/2.07.
24096 (define_peephole2 for Altivec d-form store): Likewise.
24098 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
24101 * config/i386/mmx.md (*mov<mode>_internal): Enable
24102 alternatives 11, 12, 13 and 14 also for 32bit targets.
24103 Remove alternatives 15, 16, 17 and 18.
24104 * config/i386/sse.md (vec_concatv2di): Change
24105 alternative (!x, *y) to (x, ?!*Yn).
24107 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
24109 * dumpfile.h (enum dump_kind): Remove stray comma.
24111 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
24113 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
24114 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
24115 predicate::num_conditions
24116 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
24117 (CHANGED): turn into predicate::changed.
24118 (agg_position_info): Move to ipa-predicate.h
24119 (add_condition, predicate::add_clause, predicate::operator &=,
24120 predicate::or_with, predicate::evaluate, predicate::probability,
24121 dump_condition, dump_clause, predicate::dump,
24122 predicate::remap_after_duplication, predicate::remap_after_inlining,
24123 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
24124 (evaluate_conditions_for_known_args): Update.
24125 (set_cond_stmt_execution_predicate): Update.
24126 * ipa-inline.h: Include ipa-predicate.h
24127 (condition, inline_param_summary, conditions, agg_position_info,
24128 predicate): Move to ipa-predicate.h
24129 * ipa-predicate.c: New file.
24130 * ipa-predicate.h: New file.
24132 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
24134 * final.c (leaf_function_p): Check we are not in a sequence.
24136 2017-05-18 Martin Liska <mliska@suse.cz>
24138 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
24139 * dumpfile.c (dump_register): Use new enum dump_kind.
24140 (get_dump_file_name): Likewise.
24141 (dump_enable_all): Likewise.
24142 (dump_switch_p_1): Likewise.
24143 (enable_rtl_dump_file): Remove usage of TDF_RTL.
24144 * dumpfile.h (enum dump_kind): New enum type.
24145 (struct dump_file_info): Create constructor and
24146 format fields and comments.
24147 * passes.c (pass_manager::register_one_dump_file):
24149 * statistics.c (statistics_early_init): Likewise.
24150 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
24151 TDF_TREE with TDF_SLIM.
24152 (gather_memory_references_ref): Likewise.
24154 2017-05-18 Martin Liska <mliska@suse.cz>
24156 * vec.h (struct vnull): Use it.
24158 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
24160 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
24161 (true_predicate, false_predicate, true_predicate_p,
24162 false_predicate_p): Remove.
24163 (single_cond_predicate, not_inlined_predicate): Turn to member function
24165 (add_condition): Update.
24166 (add_clause): Turn to...
24167 (predicate::add_clause): ... this one; update; allow passing NULL
24169 (and_predicates): Turn to ...
24170 (predicate::operator &=): ... this one.
24171 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
24172 (or_predicates): Turn to ...
24173 (predicate::or_with): ... this one.
24174 (evaluate_predicate): Turn to ...
24175 (predicate::evaluate): ... this one.
24176 (predicate_probability): Turn to ...
24177 (predicate::probability): ... this one.
24178 (dump_condition): Update.
24179 (dump_predicate): Turn to ...
24180 (predicate::dump): ... this one.
24181 (account_size_time): Update.
24182 (edge_set_predicate): Update.
24183 (set_hint_predicate): UPdate.
24184 (evaluate_conditions_for_known_args): Update.
24185 (evaluate_properties_for_edge): Update.
24186 (remap_predicate_after_duplication): Turn to...
24187 (predicate::remap_after_duplication): ... this one.
24188 (remap_hint_predicate_after_duplication): Update.
24189 (inline_summary_t::duplicate): UPdate.
24190 (dump_inline_edge_summary): Update.
24191 (dump_inline_summary): Update.
24192 (set_cond_stmt_execution_predicate): Update.
24193 (set_switch_stmt_execution_predicate): Update.
24194 (compute_bb_predicates): Update.
24195 (will_be_nonconstant_expr_predicate): Update.
24196 (will_be_nonconstant_predicate): Update.
24197 (phi_result_unknown_predicate): Update.
24198 (predicate_for_phi_result): Update.
24199 (array_index_predicate): Update.
24200 (estimate_function_body_sizes): Update.
24201 (estimate_node_size_and_time): Update.
24202 (estimate_ipcp_clone_size_and_time): Update.
24203 (remap_predicate): Rename to ...
24204 (predicate::remap_after_inlining): ... this one.
24205 (remap_hint_predicate): Update.
24206 (inline_merge_summary): Update.
24207 (inline_update_overall_summary): Update.
24208 (estimate_size_after_inlining): Update.
24209 (read_predicate): Rename to ...
24210 (predicate::stream_in): ... this one.
24211 (read_inline_edge_summary): Update.
24212 (write_predicate): Rename to ...
24213 (predicate::stream_out): ... this one.
24214 (write_inline_edge_summary): Update.
24215 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
24216 (clause_t): Turn to uint32_t
24217 (predicate): Turn to class; implement constructor and operators
24219 (size_time_entry): Update.
24220 (inline_summary): Update.
24221 (inline_edge_summary): Update.
24223 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
24225 * fold-const.c (fold_binary_loc): Move transformation...
24226 * match.pd (C - X CMP X): ... here.
24228 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
24230 * config/sparc/sparc.c (sparc_option_override): Set function
24231 alignment for -mcpu=niagara7 to 64 to match the I$ line.
24232 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
24234 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
24236 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
24238 2017-05-18 Marek Polacek <polacek@redhat.com>
24241 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
24242 (pass_ubsan::execute): Call gimple_assign_single_p instead of
24243 gimple_assign_load_p.
24245 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
24247 PR middle-end/80692
24248 * real.c (do_compare): Give decimal_do_compare preference over
24249 comparing just the signs.
24251 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
24253 * doc/md.texi (Canonicalization of Instructions): Describe the
24254 canonical form of instructions that inherently set a condition
24257 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
24259 PR middle-end/80775
24260 * tree-cfg.c: Move deletion of unreachable case statements to after
24261 the merging of consecutive case labels.
24263 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
24265 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
24266 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
24267 restoring of callee-saved registers.
24269 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
24271 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
24272 * config/visium/visium.c (single_set_and_flags): Likewise.
24273 * config/visium/visium.md (Substitutions): Likewise.
24275 2017-05-17 Martin Liska <mliska@suse.cz>
24277 * cfg.c: Introduce dump_flags_t type and
24278 use it instead of int type.
24280 * cfghooks.c: Likewise.
24281 * cfghooks.h (struct cfg_hooks): Likewise.
24282 * cfgrtl.c: Likewise.
24283 * cfgrtl.h: Likewise.
24284 * cgraph.c (cgraph_node::get_body): Likewise.
24285 * coretypes.h: Likewise.
24286 * domwalk.c: Likewise.
24287 * domwalk.h: Likewise.
24288 * dumpfile.c (struct dump_option_value_info): Likewise.
24289 (dump_enable_all): Likewise.
24290 (dump_switch_p_1): Likewise.
24291 (opt_info_switch_p): Likewise.
24292 * dumpfile.h (enum tree_dump_index): Likewise.
24293 (struct dump_file_info): Likewise.
24294 * genemit.c: Likewise.
24295 * generic-match-head.c: Likewise.
24296 * gengtype.c (open_base_files): Likewise.
24297 * gimple-pretty-print.c: Likewise.
24298 * gimple-pretty-print.h: Likewise.
24299 * graph.c (print_graph_cfg): Likewise.
24300 * graphite-scop-detection.c (dot_all_sese): Likewise.
24301 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
24302 * loop-unroll.c (report_unroll): Likewise.
24303 * passes.c (pass_manager::register_one_dump_file): Likewise.
24304 * print-tree.c: Likewise.
24305 * statistics.c: Likewise.
24306 * tree-cfg.c: Likewise.
24307 * tree-cfg.h: Likewise.
24308 * tree-dfa.c: Likewise.
24309 * tree-dfa.h: Likewise.
24310 * tree-dump.c (dump_function): Likewise.
24311 * tree-dump.h (struct dump_info): Likewise.
24312 * tree-pretty-print.c: Likewise.
24313 * tree-pretty-print.h: Likewise.
24314 * tree-ssa-live.c: Likewise.
24315 * tree-ssa-live.h: Likewise.
24316 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
24317 * tree-vect-loop.c: Likewise.
24318 * tree-vect-slp.c: Likewise.
24320 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
24321 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24323 PR tree-optimization/80457
24324 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
24325 of all arguments to a statement as scalar_to_vec operations.
24326 (vectorizable_call): Adjust call to vect_model_simple_cost for
24328 (vectorizable_conversion): Likewise.
24329 (vectorizable_assignment): Likewise.
24330 (vectorizable_shift): Likewise.
24331 (vectorizable_operation): Likewise.
24332 (vectorizable_comparison): Likewise.
24333 (vect_is_simple_cond): Record the def types for operands.
24334 (vectorizable_condition): Likewise, call vect_model_simple_cost.
24335 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
24336 for statement argument count.
24338 2017-05-16 Carl Love <cel@us.ibm.com>
24340 * config/rs6000/rs6000-c: Add support for built-in functions
24341 vector unsigned long long vec_bperm (vector unsigned long long,
24342 vector unsigned char)
24343 vector signed long long vec_mule (vector signed int,
24345 vector unsigned long long vec_mule (vector unsigned int,
24346 vector unsigned int)
24347 vector signed long long vec_mulo (vector signed int,
24349 vector unsigned long long vec_mulo (vector unsigned int,
24350 vector unsigned int)
24351 vector signed char vec_sldw (vector signed char,
24352 vector signed char,
24354 vector unsigned char vec_sldw (vector unsigned char,
24355 vector unsigned char,
24357 vector signed short vec_sldw (vector signed short,
24358 vector signed short,
24360 vector unsigned short vec_sldw (vector unsigned short,
24361 vector unsigned short,
24363 vector signed int vec_sldw (vector signed int,
24366 vector unsigned int vec_sldw (vector unsigned int,
24367 vector unsigned int,
24369 vector signed long long vec_sldw (vector signed long long,
24370 vector signed long long,
24372 vector unsigned long long vec_sldw (vector unsigned long long,
24373 vector unsigned long long,
24375 * config/rs6000/rs6000-c: Add support for built-in functions
24376 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
24377 * config/rs6000/altivec.h: Add defintion for vec_sldw.
24378 * doc/extend.texi: Update the built-in documentation for the
24379 new built-in functions.
24381 2017-05-16 Marek Polacek <polacek@redhat.com>
24385 * tree.c (save_expr): Don't fold the expression.
24387 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
24389 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
24390 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
24391 and (?*y,m). Update insn attributes.
24393 2017-05-16 Martin Liska <mliska@suse.cz>
24395 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
24396 flags argument of print_gimple_stmt, print_gimple_expr,
24397 print_generic_stmt and print_generic_expr.
24398 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
24399 * coretypes.h: Likewise.
24400 * except.c (dump_eh_tree): Likewise.
24401 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
24402 * gimple-pretty-print.h: Likewise.
24403 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
24404 (backprop::push_to_worklist): Likewise.
24405 (backprop::pop_from_worklist): Likewise.
24406 (backprop::process_use): Likewise.
24407 (backprop::intersect_uses): Likewise.
24408 (note_replacement): Likewise.
24409 * gimple-ssa-store-merging.c
24410 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
24411 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
24412 (pass_store_merging::execute): Likewise.
24413 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
24414 (ssa_base_cand_dump_callback): Likewise.
24415 (dump_incr_vec): Likewise.
24416 (replace_refs): Likewise.
24417 (replace_mult_candidate): Likewise.
24418 (create_add_on_incoming_edge): Likewise.
24419 (create_phi_basis): Likewise.
24420 (insert_initializers): Likewise.
24421 (all_phi_incrs_profitable): Likewise.
24422 (introduce_cast_before_cand): Likewise.
24423 (replace_one_candidate): Likewise.
24424 * gimplify.c (gimplify_expr): Likewise.
24425 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
24426 (set_rename): Likewise.
24427 (rename_uses): Likewise.
24428 (copy_loop_phi_nodes): Likewise.
24429 (add_close_phis_to_merge_points): Likewise.
24430 (copy_loop_close_phi_args): Likewise.
24431 (copy_cond_phi_args): Likewise.
24432 (graphite_copy_stmts_from_block): Likewise.
24433 (translate_pending_phi_nodes): Likewise.
24434 * graphite-poly.c (print_pdr): Likewise.
24435 (dump_gbb_cases): Likewise.
24436 (dump_gbb_conditions): Likewise.
24437 (print_scop_params): Likewise.
24438 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
24439 (build_cross_bb_scalars_use): Likewise.
24440 (gather_bbs::before_dom_children): Likewise.
24441 * hsa-dump.c (dump_hsa_immed): Likewise.
24442 * ipa-cp.c (print_ipcp_constant_value): Likewise.
24443 (get_replacement_map): Likewise.
24444 * ipa-inline-analysis.c (dump_condition): Likewise.
24445 (estimate_function_body_sizes): Likewise.
24446 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
24447 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
24448 * ipa-prop.c (ipa_dump_param): Likewise.
24449 (ipa_print_node_jump_functions_for_edge): Likewise.
24450 (ipa_modify_call_arguments): Likewise.
24451 (ipa_modify_expr): Likewise.
24452 (ipa_dump_param_adjustments): Likewise.
24453 (ipa_dump_agg_replacement_values): Likewise.
24454 (ipcp_modif_dom_walker::before_dom_children): Likewise.
24455 * ipa-pure-const.c (check_stmt): Likewise.
24456 (pass_nothrow::execute): Likewise.
24457 * ipa-split.c (execute_split_functions): Likewise.
24458 * omp-offload.c (dump_oacc_loop_part): Likewise.
24459 (dump_oacc_loop): Likewise.
24460 * trans-mem.c (tm_log_emit): Likewise.
24461 (tm_memopt_accumulate_memops): Likewise.
24462 (dump_tm_memopt_set): Likewise.
24463 (dump_tm_memopt_transform): Likewise.
24464 * tree-cfg.c (gimple_verify_flow_info): Likewise.
24465 (print_loop): Likewise.
24466 * tree-chkp-opt.c (chkp_print_addr): Likewise.
24467 (chkp_gather_checks_info): Likewise.
24468 (chkp_get_check_result): Likewise.
24469 (chkp_remove_check_if_pass): Likewise.
24470 (chkp_use_outer_bounds_if_possible): Likewise.
24471 (chkp_reduce_bounds_lifetime): Likewise.
24472 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
24473 (chkp_mark_completed_bounds): Likewise.
24474 (chkp_register_incomplete_bounds): Likewise.
24475 (chkp_mark_invalid_bounds): Likewise.
24476 (chkp_maybe_copy_and_register_bounds): Likewise.
24477 (chkp_build_returned_bound): Likewise.
24478 (chkp_get_bound_for_parm): Likewise.
24479 (chkp_build_bndldx): Likewise.
24480 (chkp_get_bounds_by_definition): Likewise.
24481 (chkp_generate_extern_var_bounds): Likewise.
24482 (chkp_get_bounds_for_decl_addr): Likewise.
24483 * tree-chrec.c (chrec_apply): Likewise.
24484 * tree-data-ref.c (dump_data_reference): Likewise.
24485 (dump_subscript): Likewise.
24486 (dump_data_dependence_relation): Likewise.
24487 (analyze_overlapping_iterations): Likewise.
24488 * tree-inline.c (expand_call_inline): Likewise.
24489 (tree_function_versioning): Likewise.
24490 * tree-into-ssa.c (dump_defs_stack): Likewise.
24491 (dump_currdefs): Likewise.
24492 (dump_names_replaced_by): Likewise.
24493 (dump_update_ssa): Likewise.
24494 (update_ssa): Likewise.
24495 * tree-object-size.c (pass_object_sizes::execute): Likewise.
24496 * tree-parloops.c (build_new_reduction): Likewise.
24497 (try_create_reduction_list): Likewise.
24498 (ref_conflicts_with_region): Likewise.
24499 (oacc_entry_exit_ok_1): Likewise.
24500 (oacc_entry_exit_single_gang): Likewise.
24501 * tree-pretty-print.h: Likewise.
24502 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
24503 (get_scalar_evolution): Likewise.
24504 (add_to_evolution): Likewise.
24505 (get_loop_exit_condition): Likewise.
24506 (analyze_evolution_in_loop): Likewise.
24507 (analyze_initial_condition): Likewise.
24508 (analyze_scalar_evolution): Likewise.
24509 (instantiate_scev): Likewise.
24510 (number_of_latch_executions): Likewise.
24511 (gather_chrec_stats): Likewise.
24512 (final_value_replacement_loop): Likewise.
24513 (scev_const_prop): Likewise.
24514 * tree-sra.c (dump_access): Likewise.
24515 (disqualify_candidate): Likewise.
24516 (create_access): Likewise.
24517 (reject): Likewise.
24518 (maybe_add_sra_candidate): Likewise.
24519 (create_access_replacement): Likewise.
24520 (analyze_access_subtree): Likewise.
24521 (analyze_all_variable_accesses): Likewise.
24522 (sra_modify_assign): Likewise.
24523 (initialize_constant_pool_replacements): Likewise.
24524 (find_param_candidates): Likewise.
24525 (decide_one_param_reduction): Likewise.
24526 (replace_removed_params_ssa_names): Likewise.
24527 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
24528 * tree-ssa-copy.c (dump_copy_of): Likewise.
24529 (copy_prop_visit_cond_stmt): Likewise.
24530 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
24531 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
24532 (record_equivalences_from_stmt): Likewise.
24533 * tree-ssa-dse.c (compute_trims): Likewise.
24534 (delete_dead_call): Likewise.
24535 (delete_dead_assignment): Likewise.
24536 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
24537 (forward_propagate_into_cond): Likewise.
24538 (pass_forwprop::execute): Likewise.
24539 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
24540 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
24542 (move_computations_worker): Likewise.
24543 (execute_sm): Likewise.
24544 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
24545 (remove_exits_and_undefined_stmts): Likewise.
24546 (remove_redundant_iv_tests): Likewise.
24547 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
24548 (adjust_iv_update_pos): Likewise.
24549 * tree-ssa-math-opts.c (bswap_replace): Likewise.
24550 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
24551 (value_replacement): Likewise.
24552 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
24553 * tree-ssa-pre.c (print_pre_expr): Likewise.
24554 (get_representative_for): Likewise.
24555 (create_expression_by_pieces): Likewise.
24556 (insert_into_preds_of_block): Likewise.
24557 (eliminate_insert): Likewise.
24558 (eliminate_dom_walker::before_dom_children): Likewise.
24559 (eliminate): Likewise.
24560 (remove_dead_inserted_code): Likewise.
24561 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
24562 * tree-ssa-reassoc.c (get_rank): Likewise.
24563 (eliminate_duplicate_pair): Likewise.
24564 (eliminate_plus_minus_pair): Likewise.
24565 (eliminate_not_pairs): Likewise.
24566 (undistribute_ops_list): Likewise.
24567 (eliminate_redundant_comparison): Likewise.
24568 (update_range_test): Likewise.
24569 (optimize_range_tests_var_bound): Likewise.
24570 (optimize_vec_cond_expr): Likewise.
24571 (rewrite_expr_tree): Likewise.
24572 (rewrite_expr_tree_parallel): Likewise.
24573 (linearize_expr): Likewise.
24574 (break_up_subtract): Likewise.
24575 (linearize_expr_tree): Likewise.
24576 (attempt_builtin_powi): Likewise.
24577 (attempt_builtin_copysign): Likewise.
24578 (transform_stmt_to_copy): Likewise.
24579 (transform_stmt_to_multiply): Likewise.
24580 (dump_ops_vector): Likewise.
24581 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
24582 (print_scc): Likewise.
24583 (set_ssa_val_to): Likewise.
24584 (visit_reference_op_store): Likewise.
24585 (visit_use): Likewise.
24586 (sccvn_dom_walker::before_dom_children): Likewise.
24587 (run_scc_vn): Likewise.
24588 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
24590 (expr_hash_elt::print): Likewise.
24591 (const_and_copies::pop_to_marker): Likewise.
24592 (const_and_copies::record_const_or_copy_raw): Likewise.
24593 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
24594 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
24595 (dump_predicates): Likewise.
24596 (find_uninit_use): Likewise.
24597 (warn_uninitialized_phi): Likewise.
24598 (pass_late_warn_uninitialized::execute): Likewise.
24599 * tree-ssa.c (verify_vssa): Likewise.
24600 (verify_ssa): Likewise.
24601 (maybe_optimize_var): Likewise.
24602 * tree-vrp.c (dump_value_range): Likewise.
24603 (dump_all_value_ranges): Likewise.
24604 (dump_asserts_for): Likewise.
24605 (register_edge_assert_for_2): Likewise.
24606 (vrp_visit_cond_stmt): Likewise.
24607 (vrp_visit_switch_stmt): Likewise.
24608 (vrp_visit_stmt): Likewise.
24609 (vrp_visit_phi_node): Likewise.
24610 (simplify_cond_using_ranges_1): Likewise.
24611 (fold_predicate_in): Likewise.
24612 (evrp_dom_walker::before_dom_children): Likewise.
24613 (evrp_dom_walker::push_value_range): Likewise.
24614 (evrp_dom_walker::pop_value_range): Likewise.
24615 (execute_early_vrp): Likewise.
24617 2017-05-16 Richard Biener <rguenther@suse.de>
24619 * dwarf2out.c (loc_list_from_tree_1): Do not create
24620 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
24622 2017-05-16 Richard Biener <rguenther@suse.de>
24624 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
24626 (note_variable_value_in_expr): If we resolved the decl ref
24627 do not push to the stack.
24629 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
24631 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
24632 operations in fast-math mode.
24633 (vaddq_f16): Likewise.
24634 (vmul_f16): Likewise.
24635 (vmulq_f16): Likewise.
24636 (vsub_f16): Likewise.
24637 (vsubq_f16): Likewise.
24638 * config/arm/neon.md (add<mode>3): New.
24640 (fma:<VH:mode>3): New. Also remove outdated comment.
24643 2017-05-16 Martin Liska <mliska@suse.cz>
24647 * ipa-devirt.c (warn_types_mismatch): Fix typo.
24648 (odr_types_equivalent_p): Likewise.
24650 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
24652 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
24654 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
24657 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
24658 non-interunit SSE move alternatives with '?'.
24659 (zero-extendsidi peephole2): New peephole to skip intermediate
24660 general register in SSE zero-extend sequence.
24662 2017-05-15 Jeff Law <law@redhat.com>
24664 * reorg.c (relax_delay_slots): Create a new variable to hold
24665 the temporary target rather than clobbering TARGET_LABEL.
24667 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
24668 missing argument to extract_bit_field call.
24669 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
24671 2017-05-15 Martin Liska <mliska@suse.cz>
24674 * gcc.c (process_command): Do not allow empty argument of -o option.
24676 2017-05-15 Renlin Li <renlin.li@arm.com>
24678 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
24679 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
24680 * config/aarch64/constraints.md (Usf): Add long call check.
24681 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
24682 (call_value): Likewise.
24683 (sibcall): Likewise.
24684 (sibcall_value): Likewise.
24686 (call_value_insn): New.
24687 (sibcall_insn): Update rtx pattern.
24688 (sibcall_value_insn): Likewise.
24689 (call_internal): Remove.
24690 (call_value_internal): Likewise.
24691 (sibcall_internal): Likewise.
24692 (sibcall_value_internal): Likewise.
24693 (call_reg): Likewise.
24694 (call_symbol): Likewise.
24695 (call_value_reg): Likewise.
24696 (call_value_symbol): Likewise.
24698 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
24701 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
24703 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
24705 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
24706 compatible with CCGOCmode and with CCZmode.
24708 2017-05-14 Martin Sebor <msebor@redhat.com>
24710 PR middle-end/77671
24711 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
24712 (gimple_fold_builtin_snprintf): Same.
24713 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
24714 (gimple_fold_builtin_snprintf): Same.
24715 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
24716 of character types.
24717 (is_call_safe): New function.
24718 (try_substitute_return_value): Call it.
24719 (try_simplify_call): New function.
24720 (pass_sprintf_length::handle_gimple_call): Call it.
24722 2017-05-14 Martin Sebor <msebor@redhat.com>
24724 PR middle-end/80669
24725 * builtins.c (expand_builtin_stpncpy): Simplify.
24727 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
24729 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
24730 * config/i386/i386.h
24731 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
24732 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
24733 (struct machine_function): Add new members call_ms2sysv,
24734 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
24735 (struct machine_frame_state): New fields sp_realigned and
24736 sp_realigned_offset.
24737 * config/i386/i386.c
24738 (enum xlogue_stub): New enum.
24739 (enum xlogue_stub_sets): New enum.
24740 (class xlogue_layout): New class.
24741 (struct ix86_frame): New fields stack_realign_allocate_offset,
24742 stack_realign_offset and outlined_save_offset. Modify comments to
24743 detail stack layout when using out-of-line stubs.
24744 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
24745 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
24746 -mcall-ms2sysv-xlogues.
24747 (stub_managed_regs): New static variable.
24748 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
24749 registers managed by out-of-line stub.
24750 (disable_call_ms2sysv_xlogues): New function.
24751 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
24752 m->call_ms2sysv when appropriate and compute frame layout for
24754 (sp_valid_at, fp_valid_at): New inline functions.
24755 (choose_basereg): New function.
24756 (choose_baseaddr): Add align parameter, use choose_basereg and modify
24758 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
24759 Use align parameter of choose_baseaddr to generated aligned SSE movs
24761 (pro_epilogue_adjust_stack): Modify to track
24762 machine_frame_state::sp_realigned.
24763 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
24764 (ix86_nsaved_sseregs): Likewise.
24765 (ix86_emit_save_regs): Likewise.
24766 (ix86_emit_save_regs_using_mov): Likewise.
24767 (ix86_emit_save_sse_regs_using_mov): Likewise.
24768 (get_scratch_register_on_entry): Likewise.
24769 (gen_frame_set): New function.
24770 (gen_frame_load): Likewise.
24771 (gen_frame_store): Likewise.
24772 (emit_outlined_ms2sysv_save): Likewise.
24773 (emit_outlined_ms2sysv_restore): Likewise.
24774 (ix86_expand_prologue): Modify stack re-alignment code and call
24775 emit_outlined_ms2sysv_save when appropriate.
24776 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
24777 parameter rtx_insn *insn, which allows the function to be used to only
24778 generate the notes.
24779 (ix86_expand_epilogue): Modify validity checks of frame and stack
24780 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
24781 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
24782 * config/i386/predicates.md
24783 (save_multiple): New predicate.
24784 (restore_multiple): Likewise.
24785 * config/i386/sse.md
24786 (save_multiple<mode>): New pattern.
24787 (save_multiple_realign<mode>): Likewise.
24788 (restore_multiple<mode>): Likewise.
24789 (restore_multiple_and_return<mode>): Likewise.
24790 (restore_multiple_leave_return<mode>): Likewise.
24791 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
24793 2017-05-14 Julia Koval <julia.koval@intel.com>
24795 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
24796 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
24797 (__builtin_ia32_xsetbv): New builtins.
24798 * config/i386/i386.c (ix86_expand_special_args_builtin):
24800 (ix86_expand_builtin): Special expand for new intrinsics.
24801 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
24802 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
24803 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
24805 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24807 * cfganal.c (inverted_post_order_compute): Change argument type
24809 * cfganal.h (inverted_post_order_compute): Adjust prototype.
24810 * df-core.c (rest_of_handle_df_initialize): Adjust.
24811 (rest_of_handle_df_finish): Likewise.
24812 (df_analyze_1): Likewise.
24813 (df_analyze): Likewise.
24814 (loop_inverted_post_order_compute): Change argument to be a vec *.
24815 (df_analyze_loop): Adjust.
24816 (df_get_n_blocks): Likewise.
24817 (df_get_postorder): Likewise.
24818 * df.h (struct df_d): Change field to be a vec.
24819 * lcm.c (compute_laterin): Adjust.
24820 (compute_available): Likewise.
24821 * lra-lives.c (lra_create_live_ranges_1): Likewise.
24822 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
24823 * tree-ssa-pre.c (compute_antic): Likewise.
24825 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24827 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
24828 (depth_first_search::depth_first_search): Change structure init
24829 function to this constructor.
24830 (depth_first_search::add_bb): Rename function to this member.
24831 (depth_first_search::execute): Likewise.
24832 (flow_dfs_compute_reverse_finish): Adjust.
24834 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24836 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
24837 (longest_simple_path): Likewise.
24838 * shrink-wrap.c (spread_components): Likewise.
24839 (disqualify_problematic_components): Likewise.
24840 (emit_common_heads_for_components): Likewise.
24841 (emit_common_tails_for_components): Likewise.
24842 (insert_prologue_epilogue_for_components): Likewise.
24844 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24846 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
24849 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24851 * df-core.c (df_set_blocks): Start using auto_bitmap.
24852 (df_compact_blocks): Likewise.
24853 * df-problems.c (df_rd_confluence_n): Likewise.
24854 * df-scan.c (df_insn_rescan_all): Likewise.
24855 (df_process_deferred_rescans): Likewise.
24856 (df_update_entry_block_defs): Likewise.
24857 (df_update_exit_block_uses): Likewise.
24858 (df_entry_block_bitmap_verify): Likewise.
24859 (df_exit_block_bitmap_verify): Likewise.
24860 (df_scan_verify): Likewise.
24861 * lra-constraints.c (lra_constraints): Likewise.
24862 (undo_optional_reloads): Likewise.
24863 (lra_undo_inheritance): Likewise.
24864 * lra-remat.c (calculate_gen_cands): Likewise.
24865 (do_remat): Likewise.
24866 * lra-spills.c (assign_spill_hard_regs): Likewise.
24867 (spill_pseudos): Likewise.
24868 * tree-ssa-pre.c (bitmap_set_and): Likewise.
24869 (bitmap_set_subtract_values): Likewise.
24871 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24873 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
24874 management with auto_bitmap.
24875 (fix_inter_tick): Likewise.
24876 (fix_recovery_deps): Likewise.
24877 * ira.c (add_store_equivs): Likewise.
24878 (find_moveable_pseudos): Likewise.
24879 (split_live_ranges_for_shrink_wrap): Likewise.
24880 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
24881 (rtx_reuse_manager::seen_def_p): Likewise.
24882 (rtx_reuse_manager::set_seen_def): Likewise.
24883 * print-rtl.h (class rtx_reuse_manager): Likewise.
24885 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24887 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
24889 (migrate_btr_def): Likewise.
24890 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
24891 * df-core.c (loop_post_order_compute): Likewise.
24892 (loop_inverted_post_order_compute): Likewise.
24893 * hsa-common.h: Likewise.
24894 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
24895 * init-regs.c (initialize_uninitialized_regs): Likewise.
24896 * ipa-inline.c (resolve_noninline_speculation): Likewise.
24897 (inline_small_functions): Likewise.
24898 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
24899 * ira.c (combine_and_move_insns): Likewise.
24900 (build_insn_chain): Likewise.
24901 * loop-invariant.c (find_invariants): Likewise.
24902 * lower-subreg.c (propagate_pseudo_copies): Likewise.
24903 * predict.c (tree_predict_by_opcode): Likewise.
24904 (predict_paths_leading_to): Likewise.
24905 (predict_paths_leading_to_edge): Likewise.
24906 (estimate_loops_at_level): Likewise.
24907 (estimate_loops): Likewise.
24908 * shrink-wrap.c (try_shrink_wrapping): Likewise.
24909 (spread_components): Likewise.
24910 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
24911 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
24912 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
24913 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
24914 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
24915 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
24916 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
24917 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
24918 (mark_threaded_blocks): Likewise.
24919 (thread_through_all_blocks): Likewise.
24920 * tree-ssa.c (verify_ssa): Likewise.
24921 (execute_update_addresses_taken): Likewise.
24922 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
24924 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24926 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
24928 (post_order_compute): Likewise.
24929 (inverted_post_order_compute): Likewise.
24930 (pre_and_rev_post_order_compute_fn): Likewise.
24932 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24934 * genrecog.c (int_set::int_set): Explicitly construct our
24935 auto_vec base class.
24936 * vec.h (auto_vec::auto_vec): New constructor.
24938 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24940 * bitmap.h (class auto_bitmap): New constructor taking
24941 bitmap_obstack * argument.
24943 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24945 * bitmap.h (class auto_bitmap): Change type of m_bits to
24946 bitmap_head, and adjust ctor / dtor and member operators.
24948 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
24950 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
24951 when returned register mode doesn't match original mode.
24953 2017-05-12 Jeff Law <law@redhat.com>
24954 Jakub Jelinek <jakub@redhat.com>
24956 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
24957 we look for cc setter after the compare-elim changes.
24958 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
24959 within the vector to match what compare-elim now expects.
24960 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
24961 (xorsi3_flags, one_cmplsi2_flags): Likewise.
24963 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
24964 after the compare-elim changes.
24965 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
24966 the vector to match what compare-elim now expects.
24967 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
24968 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
24969 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
24970 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
24971 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
24973 * config/visium/visium.c (single_set_and_flags): Fix where
24974 we look for cc setter after the compare-elim changes.
24975 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
24976 with the vector to match what compare-elim now expects.
24977 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
24978 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
24979 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
24980 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
24981 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
24982 (neg<mode>2_insn_set_overflow): Likewise.
24984 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
24986 PR middle-end/79794
24987 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
24988 maybe_expand_insn call, set ops[0].target. If still set after call,
24989 set alt_rtl. Add extra arg to recursive calls.
24990 (extract_bit_field): Add alt_rtl argument. Pass to
24992 * expmed.h (extract_bit_field): Fix prototype.
24993 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
24994 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
24995 to extract_bit_field_calls.
24996 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
24997 Pass alt_rtl to extract_bit_field calls.
24998 * calls.c (store_unaligned_arguments_into_psuedos)
24999 load_register_parameters): Pass extra NULL to extract_bit_field calls.
25000 * optabs.c (maybe_legitimize_operand): Clear op->target when call
25002 * optabs.h (struct expand_operand): Add target bitfield.
25004 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
25006 * compare-elim.c (try_eliminate_compare): Canonicalize
25007 operation with embedded compare to
25008 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
25009 (set (reg) (operation)].
25011 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
25013 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
25016 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
25017 cost of adding a carry flag for ADC instruction.
25018 [case MINUS]: Ignore the cost of subtracting a carry flag
25019 for SBB instruction.
25021 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
25023 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
25025 * config/rs6000/bmiintrin.h: New file.
25026 * config/rs6000/bmi2intrin.h: New file.
25027 * config/rs6000/x86intrin.h: New file.
25029 2017-05-12 Jeff Law <law@redhat.com>
25031 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
25034 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
25036 PR middle-end/80707
25037 * tree-cfg.c: Remove cfg edges of unreachable case statements.
25039 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25041 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
25042 early expansion of vector divide builtins.
25043 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
25044 builtins identified as having unsigned arguments.
25046 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25048 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
25049 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
25050 expansion of vector logical operations (and, andc, or, xor,
25053 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25055 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
25056 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
25058 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25060 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
25061 early GIMPLE expansion of vector multiplies.
25063 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25065 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
25066 TARGET_HAVE_MOVT conditional.
25067 (movt splitter): Likewise.
25069 2017-05-12 Richard Biener <rguenther@suse.de>
25071 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
25072 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
25073 Fold all stmts not inplace.
25075 2017-05-12 Richard Biener <rguenther@suse.de>
25077 PR tree-optimization/80713
25078 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
25079 inserted_exprs bit for not removed stmts.
25081 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
25083 PR middle-end/69921
25084 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
25085 parallelized" attribute for parallelized OpenACC kernels.
25086 * omp-offload.c (execute_oacc_device_lower): Use it.
25088 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
25089 Set "oacc kernels" attribute.
25090 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
25091 parameter. Adjust all users.
25092 (oacc_fn_attrib_kernels_p): Remove function.
25093 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
25094 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
25095 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
25096 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
25097 assert "oacc kernels" attribute is set.
25099 2017-05-11 Carl Love <cel@us.ibm.com>
25101 * config/rs6000/rs6000-c: Add support for built-in functions
25102 vector unsigned char vec_popcnt (vector signed char)
25103 vector unsigned char vec_popcnt (vector unsigned char)
25104 vector unsigned short vec_popcnt (vector signed short)
25105 vector unsigned short vec_popcnt (vector unsigned short)
25106 vector unsigned int vec_popcnt (vector signed int)
25107 vector unsigned int vec_popcnt (vector unsigned int)
25108 vector unsigned long long vec_popcnt (vector signed long long)
25109 vector unsigned long long vec_popcnt (vector unsigned long long)
25110 vector signed long long vec_slo (vector signed long long,
25111 vector signed char)
25112 vector signed long long vec_slo (vector signed long long,
25113 vector unsigned char)
25114 vector unsigned long long vec_slo (vector unsigned long long,
25115 vector signed char)
25116 vector unsigned long long vec_slo (vector unsigned long long,
25117 vector unsigned char)
25118 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
25119 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
25120 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
25121 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
25122 * doc/extend.texi: Update the built-in documentation file for the
25123 new built-in functions.
25125 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
25127 * attribs.h (sorted_attr_string): Move machine independent
25128 functions for target clone support from the i386 port to common
25129 code. Rename ix86_function_versions to common_function_versions.
25130 Rename make_name to make_unique_name.
25131 (common_function_versions): Likewise.
25132 (make_unique_name): Likewise.
25133 (make_dispatcher_decl): Likewise.
25134 (is_function_default_version): Likewise.
25135 * attribs.c (attr_strcmp): Likewise.
25136 (sorted_attr_string): Likewise.
25137 (common_function_versions): Likewise.
25138 (make_unique_name): Likewise.
25139 (make_dispatcher_decl): Likewise.
25140 (is_function_default_version): Likewise.
25141 * config/i386/i386.c (attr_strcmp): Likewise.
25142 (sorted_attr_string): Likewise.
25143 (ix86_function_versions): Likewise.
25144 (make_name): Likewise.
25145 (make_dispatcher_decl): Likewise.
25146 (is_function_default_version): Likewise.
25147 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
25149 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25152 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
25153 Account for direct move costs for vec_construct of integer
25156 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
25159 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
25160 (UNSPEC_STX_ATOMIC): Ditto.
25161 (loaddi_via_sse): New insn.
25162 (storedi_via_sse): Ditto.
25163 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
25164 Update corresponding peephole2 patterns.
25165 (atomic_storedi_fpu): Ditto.
25167 2017-05-11 Julia Koval <julia.koval@intel.com>
25169 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
25170 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
25172 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
25173 (__builtin_ia32_rsqrt14ss_mask): New builtins.
25174 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
25176 2017-05-11 Nathan Sidwell <nathan@acm.org>
25178 * graphite-poly.c: Include dumpfile.h.
25180 * dumpfle.h (dump_function): Declare here ...
25181 * tree-dump.h (dump_function): ... not here.
25182 * dumpfile.c: #include tree-cfg.h.
25183 (dump_function): Move here from ...
25184 * tree-dump.c (dump_function): ... here.
25185 * gimplify.c: #include splay-tree.h, not tree-dump.h.
25186 * graphite-poly.c: Don't include tree-dump.h.
25187 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
25188 * print-tree.c: Likewise.
25189 * stor-layout.c: Likewise.
25190 * tree-nested.c: Likewise.
25192 * dumpfile.c (dump_start): Use TDF_FLAGS.
25193 (dump_enable_all): Fix TDF_KIND check thinko.
25195 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
25197 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
25198 array entries to represent two legal parameterizations of the
25199 overloaded __builtin_cmpb function, as represented by the
25200 P6_OV_BUILTIN_CMPB constant.
25201 (altivec_resolve_overloaded_builtin): Add special case handling
25202 for the __builtin_cmpb function, as represented by the
25203 P6_OV_BUILTIN_CMPB constant.
25204 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
25205 (BU_P6_64BIT_2): New macro.
25206 (BU_P6_OVERLOAD_2): New macro
25207 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
25208 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
25209 (CMPB): Add overload support to represent both 32-bit and 64-bit
25210 compare-bytes function.
25211 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
25212 support for TARGET_CMPB.
25213 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
25214 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
25215 documentation of the __builtin_cmpb overloaded built-in function.
25217 2017-05-11 Richard Biener <rguenther@suse.de>
25219 PR tree-optimization/80705
25220 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
25221 bases are not vectorizable.
25223 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25225 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
25226 when counting register pressure.
25228 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25230 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
25231 (iv_ca_more_deps): Renamed to ...
25232 (iv_ca_compare_deps): ... this.
25233 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
25235 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25237 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
25239 (determine_group_iv_costs): ... here.
25240 (find_inv_vars_cb): Record inv var if it's not recorded before.
25242 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25244 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
25245 (get_shiftadd_cost): Ditto.
25247 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25249 * tree-ssa-address.c: Include header file.
25250 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
25252 (add_to_parts): Refactor.
25253 (addr_to_parts): New parameter. Update use of move_hint_to_base.
25254 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
25257 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25259 PR tree-optimization/53090
25260 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
25262 (extract_cond_operands): Detect condition with IV on both sides
25263 and return COMP_IV_EXPR_2.
25264 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
25265 (rewrite_use_compare): Simplify by removing call to function
25266 extract_cond_operands.
25268 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25270 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
25271 (extract_cond_operands): Detect condition comparing against non-
25272 invariant bound and return appropriate enum value.
25273 (find_interesting_uses_cond): Update use of extract_cond_operands.
25274 Handle its return value accordingly.
25275 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
25277 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25279 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
25280 nonlinear iv_use computation in loop invariant sensitive way.
25282 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25284 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
25285 (find_iv_candidates): Call relate_compare_use_with_all_cands.
25287 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25289 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
25290 (dump_cand): Support iv_cand.inv_exprs.
25291 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
25293 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
25296 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25298 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
25300 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
25301 as local function. Include necessary header files.
25302 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
25304 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25306 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
25308 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25310 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
25311 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
25312 RSHIFT_EXPR and BIT_NOT_EXPR.
25314 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25316 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
25317 (adjust_setup_cost): New parameter supporting round up adjustment.
25318 (struct address_cost_data): Delete.
25319 (force_expr_to_var_cost): Don't bound cost with spill_cost.
25320 (split_address_cost, ptr_difference_cost): Delete.
25321 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
25322 (struct ainc_cost_data): New struct.
25323 (get_address_cost_ainc): New function.
25324 (get_address_cost, get_computation_cost): Reimplement.
25325 (determine_group_iv_cost_address): Record inv_expr for all uses of
25327 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
25328 (iv_ca_has_deps): Reimplemented to ...
25329 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
25331 (iv_ca_extend): Call iv_ca_more_deps.
25333 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25335 * tree-ssa-address.c (struct mem_address): Move to header file.
25336 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
25337 * tree-ssa-address.h (struct mem_address): Move from C file.
25338 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
25340 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25342 * tree-affine.h (aff_combination_type): New interface.
25343 (aff_combination_zero_p): Remove static.
25344 (aff_combination_const_p): New interface.
25345 (aff_combination_singleton_var_p): New interfaces.
25347 2017-05-11 Richard Biener <rguenther@suse.de>
25349 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
25350 Skip unreachable blocks and destinations.
25351 (eliminate): Move stmt removal and fixup ...
25352 (fini_eliminate): ... here. Skip inserted exprs.
25353 (pass_pre::execute): Move fini_pre after fini_eliminate.
25354 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
25355 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
25356 PRE to get rid of dead code that has invalid SSA form and
25357 split critical edges again.
25359 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25361 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
25363 2017-05-11 Richard Biener <rguenther@suse.de>
25365 * passes.c (execute_function_todo): Verify loops if they are
25366 said to be up-to-date.
25367 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
25368 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
25370 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
25373 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
25374 handle calling assemble_external ourself.
25377 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
25378 modes with zero size. Enhance comment.
25380 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25382 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
25383 built-ins for vec_xl and vec_xst with short and char pointer
25386 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
25388 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
25389 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
25390 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
25391 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
25392 (_mm_maskz_min_round_ss): New intrinsics.
25393 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
25394 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
25395 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
25396 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
25397 (__builtin_ia32_minss_mask_round): New builtins.
25398 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25399 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25400 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
25402 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
25403 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
25405 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
25408 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
25410 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
25411 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
25412 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
25413 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
25414 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
25415 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
25416 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
25417 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25418 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
25419 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
25420 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
25421 (__builtin_ia32_mulss_mask_round): New builtins.
25422 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25423 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25424 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
25426 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
25427 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
25429 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
25432 2017-05-10 Julia Koval <julia.koval@intel.com>
25434 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
25435 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
25436 (_mm256_setr_m128i): New intrinsics.
25438 2017-05-10 Julia Koval <julia.koval@intel.com>
25440 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
25441 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
25442 (_mm_maskz_rcp14_ss): New intrinsics.
25443 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
25444 (__builtin_ia32_rcp14ss_mask): New builtins.
25445 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
25447 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
25449 PR tree-optimization/51513
25450 * tree-cfg.c (gimple_seq_unreachable_p): New function.
25451 (assert_unreachable_fallthru_edge_p): Use it.
25452 (group_case_labels_stmt): Likewise.
25453 * tree-cfg.h: Prototype it.
25454 * stmt.c: Include cfghooks.h and tree-cfg.h.
25455 (emit_case_dispatch_table) <gap_label>: New local variable.
25456 Use it to fill dispatch table gaps.
25457 Test for default_label before updating probabilities.
25458 (expand_case) <default_label>: Remove unneeded initialization.
25459 Test for unreachable default case statement and remove its edge.
25460 Set default_label accordingly.
25461 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
25463 2017-05-10 Carl Love <cel@us.ibm.com>
25465 * config/rs6000/rs6000-c: Add support for built-in functions
25466 vector signed char vec_neg (vector signed char)
25467 vector signed short int vec_neg (vector short int)
25468 vector signed int vec_neg (vector signed int)
25469 vector signed long long vec_neg (vector signed long long)
25470 vector float vec_neg (vector float)
25471 vector double vec_neg (vector double)
25472 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
25474 * config/rs6000/altivec.h: Add define for vec_neg
25475 * doc/extend.texi: Update the built-in documentation for the
25476 new built-in functions.
25478 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25480 PR tree-optimization/77644
25481 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
25483 2017-05-10 Nathan Sidwell <nathan@acm.org>
25485 * dumpfile.h (TDI_lang_all): New.
25486 (TDF_KIND): New. Renumber others
25487 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
25489 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
25491 (get_dump_file_name): Adjust suffix generation.
25492 (dump_enable_all): Use TDF_KIND.
25493 * doc/invoke.texi (-fdump-lang-all): Document.
25495 * dumpfile.h: Tabify.
25497 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
25500 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
25501 Move member access before delete.
25503 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
25505 * tree-inline.c (expand_call_inline): Split block at stmt
25508 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
25511 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
25512 are now unused after splitting mov{sf,sd}_hardfloat.
25513 (f32_lr2): Likewise.
25514 (f32_lm): Likewise.
25515 (f32_lm2): Likewise.
25516 (f32_li): Likewise.
25517 (f32_li2): Likewise.
25518 (f32_lv): Likewise.
25519 (f32_sr): Likewise.
25520 (f32_sr2): Likewise.
25521 (f32_sm): Likewise.
25522 (f32_sm2): Likewise.
25523 (f32_si): Likewise.
25524 (f32_si2): Likewise.
25525 (f32_sv): Likewise.
25526 (f32_dm): Likewise.
25527 (f32_vsx): Likewise.
25528 (f32_av): Likewise.
25529 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
25530 For movsf, order stores so the VSX stores occur before the GPR
25531 store which encourages the register allocator to use a traditional
25532 FPR instead of a GPR. For movsd, order the stores so that the GPR
25533 store comes before the VSX stores to allow the power6 to work.
25534 This is due to the power6 not having a 32-bit integer store
25535 instruction from a FPR.
25536 (movsf_hardfloat): Likewise.
25537 (movsd_hardfloat): Likewise.
25539 2017-05-09 Martin Sebor <msebor@redhat.com>
25541 PR translation/80280
25542 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
25545 PR translation/80280
25546 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
25547 data member added in r247778.
25548 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
25550 2017-05-09 Nathan Sidwell <nathan@acm.org>
25552 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
25554 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
25557 2017-05-09 Marek Polacek <polacek@redhat.com>
25559 * doc/invoke.texi: Fix typo.
25561 2017-05-09 Richard Biener <rguenther@suse.de>
25563 * tree-vrp.c (vrp_val_is_max): Adjust comment.
25564 (vrp_val_is_min): Likewise.
25565 (set_value_range_to_value): Likewise.
25566 (set_value_range_to_nonnegative): Likewise.
25567 (gimple_assign_nonzero_p): Likewise.
25568 (gimple_stmt_nonzero_p): Likewise.
25569 (vrp_int_const_binop): Likewise. Remove unreachable case.
25570 (adjust_range_with_scev): Adjust comments.
25571 (compare_range_with_value): Likewise.
25572 (extract_range_from_phi_node): Likewise.
25573 (test_for_singularity): Likewise.
25575 2017-05-09 Richard Biener <rguenther@suse.de>
25577 * tree-vrp.c (get_single_symbol): Add assert that we don't
25578 get overflowed constants as invariant part.
25579 (compare_values_warnv): Add comment before the TREE_NO_WARNING
25580 checks. Use wi::cmp instead of recursing for integer constants.
25581 (compare_values): Just ignore whether we assumed undefined
25582 overflow instead of failing the compare.
25583 (extract_range_for_var_from_comparison_expr): Add comment before the
25584 TREE_NO_WARNING sets.
25585 (test_for_singularity): Likewise.
25586 (extract_range_from_comparison): Do not disable optimization
25587 when we assumed undefined overflow.
25588 (extract_range_basic): Remove init of unused var.
25590 2017-05-09 Richard Biener <rguenther@suse.de>
25592 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
25593 (extract_range_from_multiplicative_op_1): Adjust.
25594 (extract_range_from_binary_expr_1): Use int_const_binop.
25596 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
25599 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
25600 rs6000_store_data_bypass_p in seven define_bypass directives and
25601 in several comments.
25602 * config/rs6000/rs6000-protos.h: Add prototype for
25603 rs6000_store_data_bypass_p function.
25604 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
25605 function implements slightly different (rs6000-specific) semantics
25606 than store_data_bypass_p, returning false rather than aborting
25607 with assertion error when arguments do not satisfy the
25608 requirements of store data bypass.
25609 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
25610 rs6000_store_data_bypass_p.
25612 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
25614 * config/xtensa/xtensa-protos.h
25615 (xtensa_initial_elimination_offset): New declaration.
25616 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
25617 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
25618 macro definition, add case for FRAME_POINTER_REGNUM when
25619 FRAME_GROWS_DOWNWARD.
25620 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
25621 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
25622 xtensa_initial_elimination_offset.
25624 2017-05-08 Nathan Sidwell <nathan@acm.org>
25626 * doc/invoke.texi: Alphabetize -fdump options.
25628 2017-05-08 Martin Sebor <msebor@redhat.com>
25630 PR translation/80280
25631 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
25633 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
25635 * target.def (compute_frame_layout): New optional target hook.
25636 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
25637 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
25638 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
25640 * reload1.c (verify_initial_elim_offsets): Likewise.
25641 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
25642 (use_simple_return_p): Call arm_compute_frame_layout if needed.
25643 (arm_get_frame_offsets): Split up into this ...
25644 (arm_compute_frame_layout): ... and this function.
25646 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
25648 * config/aarch64/constraints.md (Usa): New constraint.
25649 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
25651 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
25653 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
25654 with_multilib_list after it has been checked.
25656 2017-05-08 Richard Biener <rguenther@suse.de>
25658 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
25659 (bitmap_set_subtract_values): Likewise.
25661 2017-05-08 Richard Biener <rguenther@suse.de>
25663 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
25664 (gimple_assign_nonzero): ... this and remove strict_overflow_p
25666 (gimple_stmt_nonzero_warnv_p): Rename to ...
25667 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
25669 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
25670 (extract_range_basic): Adjust, do not disable propagation on
25671 strict overflow sensitive simplification.
25672 (vrp_visit_cond_stmt): Likewise.
25674 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
25676 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
25677 body size unconditionally.
25679 2017-05-07 Jeff Law <law@redhat.com>
25682 2017-05-06 Jeff Law <law@redhat.com>
25683 PR tree-optimization/78496
25684 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
25687 PR tree-optimization/78496
25688 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
25689 (simplify_stmt_using_ranges): Call it.
25690 (vrp_dom_walker::before_dom_children): Extract equivalences
25691 from an ASSERT_EXPR with an equality comparison against a
25694 2017-05-06 Jeff Law <law@redhat.com>
25696 PR tree-optimization/78496
25697 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
25700 PR tree-optimization/78496
25701 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
25702 (simplify_stmt_using_ranges): Call it.
25703 (vrp_dom_walker::before_dom_children): Extract equivalences
25704 from an ASSERT_EXPR with an equality comparison against a
25707 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
25709 * lra-constraints.c (lra_copy_reg_equiv): New function.
25710 (split_reg): Use it to copy equivalence information from the
25711 original register to the spill register.
25713 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
25715 PR rtl-optimization/75964
25716 * simplify-rtx.c (simplify_const_relational_operation): Remove
25717 invalid handling of comparisons of integer ABS.
25719 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
25721 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
25722 initialize to zero.
25723 (init_regs): Remove declaration.
25724 (function_arg_advance_32): Initialize error_p as boolean variable.
25726 2017-05-05 Nathan Sidwell <nathan@acm.org>
25728 * store-motion.c (remove_reachable_equiv_notes): Reformat long
25729 lines. Use for (;;).
25731 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25733 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
25734 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
25735 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
25736 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
25737 VF=2 that require versioning.
25739 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25741 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
25744 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25746 * diagnostic.h (diagnostic_override_option_index): Convert from
25747 macro to inline function.
25749 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25751 * diagnostic.c (last_module_changed_p): New function.
25752 (set_last_module): New function.
25753 (diagnostic_report_current_module): Convert macro usage to
25754 the above functions.
25755 * diagnostic.h (diagnostic_context::last_module): Strengthen
25756 from const line_map * to const line_map_ordinary *.
25757 (diagnostic_last_module_changed): Delete macro.
25758 (diagnostic_set_last_module): Delete macro.
25760 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25762 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
25763 with diagnostic_report_diagnostic.
25764 (diagnostic_n_impl_richloc): Likewise.
25765 * diagnostic.h (report_diagnostic): Delete macro.
25766 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
25767 with diagnostic_report_diagnostic.
25768 * substring-locations.c (format_warning_va): Likewise.
25770 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25772 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
25773 save/restor of format_spec. Move option-printing code to...
25774 (print_option_information): ...this new function, and
25775 reimplement by simply printing to the pretty_printer,
25776 rather than appending to the format string.
25778 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25780 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
25781 handling logic into...
25782 (update_effective_level_from_pragmas): ...this new function.
25784 2017-05-04 Andrew Waterman <andrew@sifive.com>
25786 * config/riscv/riscv.opt (mstrict-align): New option.
25787 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
25788 (SLOW_UNALIGNED_ACCESS): Define.
25789 (riscv_slow_unaligned_access): Declare.
25790 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
25792 (riscv_slow_unaligned_access): New variable.
25793 (rocket_tune_info): Set slow_unaligned_access to true.
25794 (optimize_size_tune_info): Set slow_unaligned_access to false.
25795 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
25796 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
25797 (riscv_option_override): Set riscv_slow_unaligned_access.
25798 * doc/invoke.texi: Add -mstrict-align to RISC-V.
25800 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
25802 * config/riscv/riscv.md: Unify indentation.
25804 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
25809 * config/rs6000/rs6000.md (u code attribute): Add FIX and
25811 (extendsi<mode>2): Add support for doing sign extension via
25812 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
25813 don't have ISA 3.0 instructions.
25814 (extendsi<mode>2 splitter): Likewise.
25815 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
25816 generate the normal insns since SImode can now go in vector
25817 registers. Disallow the special UNSPECs needed for previous
25818 machines to hide SImode being used. Add new insns
25819 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
25820 (fix_trunc<mode>si2_stfiwx): Likewise.
25821 (fix_trunc<mode>si2_internal): Likewise.
25822 (fixuns_trunc<mode>si2): Likewise.
25823 (fixuns_trunc<mode>si2_stfiwx): Likewise.
25824 (fctiw<u>z_<mode>_smallint): Likewise.
25825 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
25826 of floating point to 32-bit integer from doing a direct move to
25827 the GPR registers to do a store.
25828 (fctiwz_<mode>): Break long line.
25830 2017-05-05 Bin Cheng <bin.cheng@arm.com>
25832 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
25833 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
25834 (addr_list, addr_offset_valid_p): New.
25835 (split_address_groups): Check offset validity with above function.
25836 (gt-tree-ssa-loop-ivopts.h): Include header file.
25838 2017-05-05 Nathan Sidwell <nathan@acm.org>
25840 * config.gcc (arm*-*-*): Add missing 'fi'.
25842 2017-05-05 Steve Ellcey <sellcey@cavium.com>
25844 * doc/invoke.texi (-fopt-info): Explicitly say order of options
25845 included in -fopt-info does not matter.
25846 * doc/optinfo.texi (-fopt-info): Fix description of default
25847 behavour. Explicitly say order of options included in -fopt-info
25850 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
25852 * config.gcc: Allow combinations of aprofile and rmprofile values for
25853 --with-multilib-list.
25854 * config/arm/t-multilib: New file.
25855 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
25856 variables. Remove setting of ISA and floating-point ABI in
25857 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
25858 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
25859 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
25860 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
25862 * config/arm/t-rmprofile: Likewise except for the matches changes.
25863 * doc/install.texi (--with-multilib-list): Document the combination of
25864 aprofile and rmprofile values and warn about pitfalls in doing that.
25866 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
25868 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
25869 (movdi_aarch64): Likewise.
25871 2017-05-05 Jakub Jelinek <jakub@redhat.com>
25873 PR tree-optimization/80632
25874 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
25876 (build_arrays): Initialize it for virtual phis.
25877 (fix_phi_nodes): Use it for virtual phis.
25879 PR tree-optimization/80558
25880 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
25881 [x, y] op z into [x op, y op z] for op & or | if conditions
25884 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
25885 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25888 * config/arm/arm.md (use_literal_pool): Remove.
25889 (64-bit immediate split): No longer takes cost into consideration
25890 if arm_disable_literal_pool is enabled.
25891 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
25892 used when arm_disable_literal_pool is enabled.
25893 (arm_max_const_double_inline_cost): Remove use of
25894 arm_disable_literal_pool.
25895 (push_minipool_fix): Add assert.
25896 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
25897 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
25898 (no_literal_pool_sf_immediate): New.
25900 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25902 PR tree-optimization/80613
25903 * tree-ssa-dce.c (propagate_necessity): Remove cases for
25904 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
25906 2017-05-05 Richard Biener <rguenther@suse.de>
25908 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
25910 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
25912 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
25913 of this flag from insn conditions due to removal from r247495.
25915 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
25917 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
25919 (arm_early_store_addr_dep_ptr): Likewise.
25920 * config/arm/aarch-common-protos.h
25921 (arm_early_load_addr_dep_ptr): Add prototype.
25922 (arm_early_store_addr_dep_ptr): Likewise.
25923 * config/arm/cortex-a53.md: Add new bypasses.
25925 2017-05-05 Jakub Jelinek <jakub@redhat.com>
25927 * tree.c (next_type_uid): Change type to unsigned.
25928 (type_hash_canon): Decrement back next_type_uid if
25929 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
25930 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
25933 2017-05-04 Martin Sebor <msebor@redhat.com>
25935 * builtins.c: Fix a trivial typo in a comment.
25937 PR middle-end/79234
25938 * builtins.c (check_sizes): Adjust to handle reading past the end.
25939 Avoid printing excessive upper bound of ranges. Use %E to print
25940 tree nodes instead of converting them to %wu.
25941 (expand_builtin_memchr): New function.
25942 (compute_dest_size): Rename...
25943 (compute_objsize): ...to this.
25944 (expand_builtin_memcpy): Adjust.
25945 (expand_builtin_mempcpy): Adjust.
25946 (expand_builtin_strcat): Adjust.
25947 (expand_builtin_strcpy): Adjust.
25948 (check_strncat_sizes): Adjust.
25949 (expand_builtin_strncat): Adjust.
25950 (expand_builtin_strncpy): Adjust and simplify.
25951 (expand_builtin_memset): Adjust.
25952 (expand_builtin_bzero): Adjust.
25953 (expand_builtin_memcmp): Adjust.
25954 (expand_builtin): Handle memcmp.
25955 (maybe_emit_chk_warning): Check strncat just once.
25957 2017-05-04 Martin Sebor <msebor@redhat.com>
25959 PR preprocessor/79214
25960 PR middle-end/79222
25961 PR middle-end/79223
25962 * builtins.c (check_sizes): Add inlining context and issue
25963 warnings even when -Wno-system-headers is set.
25964 (check_strncat_sizes): Same.
25965 (expand_builtin_strncat): Same.
25966 (expand_builtin_memmove): New function.
25967 (expand_builtin_stpncpy): Same.
25968 (expand_builtin): Handle memmove and stpncpy.
25970 2017-05-04 Bin Cheng <bin.cheng@arm.com>
25972 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
25973 which is not used any more.
25975 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25977 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
25979 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25981 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
25982 (cortexa53_tunings): Likewise.
25983 (cortexa57_tunings): Likewise.
25984 (cortexa72_tunings): Likewise.
25985 (cortexa73_tunings): Likewise.
25987 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25989 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
25990 Set loop alignment to 8.
25992 2017-05-04 Martin Sebor <msebor@redhat.com>
25994 PR translation/80280
25995 * builtins.c (expand_builtin_object_size): Add missing quoting to
25996 %D and like directives.
25997 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
25998 (hsa_type_for_tree_type): Same.
25999 (verify_function_arguments): Same.
26000 * symtab.c (symbol_table::change_decl_assembler_name): Same.
26001 * varasm.c (get_section): Same.
26004 2017-05-04 Martin Sebor <msebor@redhat.com>
26006 PR translation/80280
26007 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
26009 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26011 * config/aarch64/aarch64.c (generic_addrcost_table):
26012 Change HI/TI mode setting.
26014 2017-05-04 Martin Jambor <mjambor@suse.cz>
26016 PR tree-optimization/80622
26017 * tree-sra.c (comes_initialized_p): New function.
26018 (build_accesses_from_assign): Only set write lazily when
26019 comes_initialized_p is false.
26020 (analyze_access_subtree): Use comes_initialized_p.
26021 (propagate_subaccesses_across_link): Assert !comes_initialized_p
26022 instead of testing for PARM_DECL.
26024 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26026 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
26027 constraint on operand 0 to allow more general addressing modes.
26028 Adjust output template.
26029 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
26031 * config/aarch64/aarch64-protos.h
26032 (aarch64_address_valid_for_prefetch_p): Declare prototype.
26033 * config/aarch64/constraints.md (Dp): New address constraint.
26034 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
26037 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
26039 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
26040 update use of estimate_ipcp_clone_size_and_time.
26041 (estimate_local_effects): Update use of
26042 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
26043 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
26044 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
26045 Return nonspecialized time.
26047 2017-05-04 Richard Biener <rguenther@suse.de>
26049 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
26050 for the last VUSE which def dominates the PHI. Directly call
26052 (get_continuation_for_phi_1): Remove.
26054 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
26056 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
26057 to explain the use of truncating division. Cap the number of
26058 iterations to the maximum given by nb_iterations_upper_bound,
26061 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
26063 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
26064 * configure: Regenerate.
26065 * config.in: Regenerate.
26066 * config/i386/driver-mingw32.c: new file.
26067 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
26068 * config.host: Link driver-mingw32.o on MinGW host.
26069 * doc/install.texi: Document new --enable-mingw-wildcard configure
26072 2017-05-04 Marek Polacek <polacek@redhat.com>
26074 PR tree-optimization/80612
26075 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
26077 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26078 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
26080 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
26081 (movt splitter): Likewise.
26082 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
26083 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
26084 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
26085 block for Thumb-1 with MOVT.
26086 (thumb2_legitimate_address_p): Move code block ...
26087 (can_avoid_literal_pool_for_label_p): ... into this new function.
26088 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
26090 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
26091 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
26092 "M-profile targets with the MOVT instruction".
26094 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26096 * config/arm/arm-builtins.c (arm_init_builtins): Rename
26097 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
26098 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
26100 2017-05-04 Martin Liska <mliska@suse.cz>
26102 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
26103 variable cond_code.
26105 2017-05-04 Richard Biener <rguenther@suse.de>
26107 * tree.c (array_at_struct_end_p): Handle arrays at struct
26108 end with flexarrays more conservatively. Refactor and treat
26109 arrays of arrays or aggregates more strict. Fix
26110 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
26111 * tree.c (array_at_struct_end_p): Adjust prototype.
26112 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
26113 * gimple-fold.c (get_range_strlen): Likewise.
26114 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
26116 2017-05-04 Richard Biener <rguenther@suse.de>
26118 PR tree-optimization/31130
26119 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
26121 (supports_overflow_infinity): Likewise.
26122 (is_negative_overflow_infinity): Likewise.
26123 (is_positive_overflow_infinity): Likewise.
26124 (is_overflow_infinity): Likewise.
26125 (stmt_overflow_infinity): Likewise.
26126 (overflow_infinity_range_p): Likewise.
26127 (usable_range_p): Remove as always returning true.
26128 (make_overflow_infinity): Remove.
26129 (negative_overflow_infinity): Likewise.
26130 (positive_overflow_infinity): Likewise.
26131 (avoid_overflow_infinity): Likewise.
26132 (set_value_range): Adjust accordingly.
26133 (set_value_range_to_nonnegative): Likewise, remove now unused
26134 overflow_infinity arg.
26135 (vrp_operand_equal_p): Adjust.
26136 (update_value_range): Likewise.
26137 (range_int_cst_singleton_p): Likewise.
26138 (operand_less_p): Likewise.
26139 (compare_values_warnv): Likewise.
26140 (extract_range_for_var_from_comparison_expr): Likewise.
26141 (vrp_int_const_binop): Likewise.
26142 (zero_nonzero_bits_from_vr): Likewise.
26143 (extract_range_from_multiplicative_op_1): Likewise.
26144 (extract_range_from_binary_expr_1): Likewise.
26145 (extract_range_from_unary_expr): Likewise.
26146 (extract_range_from_comparison): Likewise.
26147 (extract_range_basic): Likewise.
26148 (adjust_range_with_scev): Likewise.
26149 (compare_ranges): Likewise.
26150 (compare_range_with_value): Likewise.
26151 (dump_value_range): Likewise.
26152 (test_for_singularity): Likewise, remove strict_overflow_p parameter
26154 (simplify_cond_using_ranges): Adjust.
26156 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
26158 * brig-builtins.def: Added a builtin for class_f64.
26159 * builtin-types.def: Added a builtin type needed by class_f64.
26161 2017-05-03 Jason Merrill <jason@redhat.com>
26163 * timevar.def: Add TV_CONSTEXPR.
26165 2017-05-03 David Malcolm <dmalcolm@redhat.com>
26167 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
26169 2017-05-03 Martin Jambor <mjambor@suse.cz>
26171 * ipa-prop.c (ipa_update_after_lto_read): Removed.
26172 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
26173 * ipa-cp.c (ipcp_propagate_stage): Do not call
26174 ipa_update_after_lto_read.
26175 * ipa-inline.c (ipa_inline): Likewise.
26177 2017-05-03 Martin Jambor <mjambor@suse.cz>
26179 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
26180 tag. Added a default constructor and a destructor.
26181 (ipa_edge_args_sum_t): New class;
26182 (ipa_edge_args_sum): Declare.
26183 (ipa_edge_args_vector): Remove declaration.
26184 (IPA_EDGE_REF): Use ipa_edge_args_sum.
26185 (ipa_free_edge_args_substructures): Remove declaration.
26186 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
26187 (ipa_edge_args_info_available_for_edge_p): Likewise.
26188 * ipa-prop.c (ipa_edge_args_vector): Removed.
26189 (edge_removal_hook_holder): Likewise.
26190 (edge_duplication_hook_holder): Likewise.
26191 (ipa_edge_args_sum): New variable.
26192 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
26193 ipa_edge_args_vector.
26194 (ipa_free_edge_args_substructures): Likewise.
26195 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
26196 ipa_edge_args_vector.
26197 (ipa_edge_removal_hook): Turned into method
26198 ipa_edge_args_sum_t::remove.
26199 (ipa_edge_duplication_hook): Turned into method
26200 ipa_edge_args_sum_t::duplicate.
26201 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
26202 registering edge hooks.
26203 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
26204 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
26205 ipa_edge_args_sum instead of ipa_edge_args_vector.
26206 * ipa-profile.c (ipa_profile): Likewise.
26208 2017-05-03 Martin Jambor <mjambor@suse.cz>
26210 * symbol-summary.h (function_summary): New method exists.
26211 (function_summary::symtab_removal): Deallocate through release.
26212 (call_summary): New class.
26213 (gt_ggc_mx): New overload.
26214 (gt_pch_nx): Likewise.
26215 (gt_pch_nx): Likewise.
26217 2017-05-03 Jeff Law <law@redhat.com>
26219 PR tree-optimization/78496
26220 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
26221 from simplify_cond_using_ranges. Split off code to walk
26222 backwards through casts into ...
26223 (simplify_cond_using_ranges_2): New function.
26224 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
26225 (execute_vrp): After identifying jump threads, call
26226 simplify_cond_using_ranges_2.
26228 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
26231 * ipa-inline.h (inline_summary): Add ctor.
26232 (create_ggc): Do not use ggc_cleared_alloc.
26234 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
26235 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26237 * gcc.c (handle_braces): Support escaping in switch matching
26239 * doc/invoke.texi (Spec Files): Document it.
26240 Remove superfluous @code markup in items.
26242 2017-05-03 David Malcolm <dmalcolm@redhat.com>
26244 * diagnostic-show-locus.c (struct column_range): New struct.
26245 (get_affected_columns): New function.
26246 (get_printed_columns): New function.
26247 (struct correction): New struct.
26248 (correction::ensure_capacity): New function.
26249 (correction::ensure_terminated): New function.
26250 (struct line_corrections): New struct.
26251 (line_corrections::~line_corrections): New dtor.
26252 (line_corrections::add_hint): New function.
26253 (layout::print_trailing_fixits): Reimplement in terms of the new
26255 (selftest::test_overlapped_fixit_printing): New function.
26256 (selftest::diagnostic_show_locus_c_tests): Call it.
26258 2017-05-03 Nathan Sidwell <nathan@acm.org>
26260 Canonicalize canonical type hashing
26261 * tree.h (type_hash_canon_hash): Declare.
26262 * tree.c (type_hash_list, attribute_hash_list): Move into
26263 type_hash_canon_hash.
26264 (build_type_attribute_qual_variant): Break out hash code calc into
26265 type_hash_canon_hash.
26266 (type_hash_canon_hash): New. Generic type hash computation.
26267 (build_range_type_1, build_array_type_1, build_function_type,
26268 build_method_type_directly, build_offset_type, build_complex_type,
26269 make_vector_type): Call it.
26271 2017-05-03 Richard Biener <rguenther@suse.de>
26273 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
26274 When all DRs have unknown misaligned do not always peel
26275 when there is a store but apply the same costing model as if
26276 there were only loads.
26278 2017-05-03 Richard Biener <rguenther@suse.de>
26281 PR tree-optimization/80492
26282 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
26283 compare_base_decls returning dont-know properly.
26285 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
26287 * config/arm/iterators.md (CCSI): New mode iterator.
26288 (arch): New mode attribute.
26289 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
26290 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
26291 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
26292 code iterator for success result mode.
26293 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
26294 the corresponding new insn generators.
26296 2017-05-03 Bin Cheng <bin.cheng@arm.com>
26299 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26300 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
26302 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
26304 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
26305 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
26306 (DDR_A): Wrap DDR argument in brackets.
26307 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
26308 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
26309 (DDR_REVERSED_P): Likewise.
26311 2017-05-03 Jakub Jelinek <jakub@redhat.com>
26313 PR tree-optimization/79472
26314 * tree-switch-conversion.c (struct switch_conv_info): Add
26315 contiguous_range and default_case_nonstandard fields.
26316 (collect_switch_conv_info): Compute contiguous_range and
26317 default_case_nonstandard fields, don't clear final_bb if
26318 contiguous_range and only the default case doesn't have the required
26320 (check_all_empty_except_final): Set default_case_nonstandard instead
26321 of failing if contiguous_range and the default case doesn't have empty
26323 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
26324 and only the default case doesn't have the required constants. Skip
26326 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
26327 if default_case_nonstandard.
26328 (build_constructors): Build constant 1 just once. Assert that default
26329 values aren't inserted in between cases if contiguous_range. Skip
26331 (build_arrays): Skip virtual phis.
26332 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
26333 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
26334 Handle virtual phis.
26335 (gen_inbound_check): Handle default_case_nonstandard case.
26336 (process_switch): Adjust check_final_bb caller. Call
26337 gather_default_values with the first non-default case instead of
26338 default case if default_case_nonstandard.
26340 2017-05-02 Nathan Sidwell <nathan@acm.org>
26342 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
26343 check. Fix formatting.
26345 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
26347 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
26348 errors when comparing specialized and unspecialized times.
26350 2017-05-02 David Malcolm <dmalcolm@redhat.com>
26352 * diagnostic-show-locus.c
26353 (layout::should_print_annotation_line_p): Make private.
26354 (layout::print_annotation_line): Make private.
26355 (layout::annotation_line_showed_range_p): Make private.
26356 (layout::show_ruler): Make private.
26357 (layout::print_source_line): Make private. Pass in line and
26358 line_width, rather than calling location_get_source_line. Drop
26360 (layout::print_leading_fixits): New method.
26361 (layout::print_any_fixits): Rename to...
26362 (layout::print_trailing_fixits): ...this, and make private.
26363 Don't print newline fixits.
26364 (diagnostic_show_locus): Move logic for printing one row into...
26365 (layout::print_line): ...this new function. Move the
26366 location_get_source_line call and error-handling from
26367 print_source_line to here. Call print_leading_fixits, and rename
26368 print_any_fixits to print_trailing_fixits.
26369 (selftest::test_fixit_insert_containing_newline): Update now that
26370 newlines are partially supported.
26371 (selftest::test_fixit_insert_containing_newline_2): New test.
26372 (selftest::test_fixit_replace_containing_newline): Update comments.
26373 (selftest::diagnostic_show_locus_c_tests): Call the new test.
26374 * edit-context.c (class added_line): New class.
26375 (class edited_line): Describe newline handling in comment.
26376 (edited_line::actually_edited_p): New method.
26377 (edited_line::print_content): Delete redundant decl.
26378 (edited_line::m_predecessors): New field.
26379 (edited_file::print_content): Call edited_line::print_content.
26380 (edited_file::print_diff): Update to support newlines.
26381 (edited_file::print_diff_hunk): Likewise.
26382 (edited_file::print_run_of_changed_lines): New function.
26383 (edited_file::print_diff_line): Convert to...
26384 (print_diff_line): ...this.
26385 (edited_file::get_effective_line_count): New function.
26386 (edited_line::edited_line): Initialize new field m_predecessors.
26387 (edited_line::~edited_line): Clean up m_predecessors.
26388 (edited_line::apply_fixit): Handle newlines.
26389 (edited_line::get_effective_line_count): New function.
26390 (edited_line::print_content): New function.
26391 (edited_line::print_diff_lines): New function.
26392 (selftest::test_applying_fixits_insert_containing_newline): New
26394 (selftest::test_applying_fixits_replace_containing_newline): New
26396 (selftest::insert_line): New function.
26397 (selftest::test_applying_fixits_multiple_lines): Add example of
26399 (selftest::edit_context_c_tests): Call the new tests.
26401 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26403 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
26404 parameter cand. Update dump information.
26405 (get_computation_cost): Update uses.
26407 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26409 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
26410 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
26411 (get_computation_at, rewrite_use_address): Update use of
26412 get_computation_aff.
26414 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26416 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
26417 (get_computation): Delete.
26418 (get_computation_cost): Implement like get_computation_cost_at.
26419 Use get_computation_at.
26420 (get_computation_cost_at): Delete.
26421 (rewrite_use_nonlinear_expr): Use get_computation_at.
26422 (rewrite_use_compare, remove_unused_ivs): Ditto.
26424 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26426 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
26428 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26430 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
26431 (ivopts_global_cost_for_size): Rename parameter and update uses.
26432 (iv_ca_recount_cost): Update uses.
26433 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
26434 candidates seperately in n_invs and n_cands.
26435 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
26437 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26439 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
26440 (find_inv_vars_cb): New.
26441 (find_depends): Renamed to ...
26442 (find_inv_vars): ... this.
26443 (add_candidate_1, force_var_cost): Call find_inv_vars.
26444 (split_address_cost, determine_group_iv_cost_cond): Ditto.
26446 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26448 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
26449 inv_vars. Add inv_exprs.
26450 (struct iv_cand): Rename depends_on to inv_vars.
26451 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
26452 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
26453 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
26454 (dump_cand): Dump inv_vars.
26455 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
26456 (record_invariant, find_depends, add_candidate_1): Ditto.
26457 (set_group_iv_cost, force_var_cost): Ditto.
26458 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
26459 (get_computation_cost_at, get_computation_cost): Ditto.
26460 (determine_group_iv_cost_generic): Ditto.
26461 (determine_group_iv_cost_address): Ditto.
26462 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
26463 (determine_group_iv_costs): Ditto.
26464 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
26465 (iv_ca_set_remove_invariants): Renamed to ...
26466 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
26467 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
26468 (iv_ca_set_add_invariants): Renamed to ...
26469 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
26470 (iv_ca_set_cp): Use iv_ca_set_add_invs.
26471 (iv_ca_has_deps): Support inv_vars and inv_exprs.
26472 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
26473 (create_new_ivs): Remove useless dump.
26475 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26477 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
26479 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
26480 (iv_ca_set_no_cp, create_new_iv): Ditto.
26482 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26484 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
26486 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
26488 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
26489 function tree_check2.
26491 2017-05-02 Martin Liska <mliska@suse.cz>
26493 * doc/gcov.texi: Add missing preposition.
26494 * gcov.c (function_info::function_info): Properly fill up
26495 all member variables.
26497 2017-05-02 Tamar Christina <tamar.christina@arm.com>
26499 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
26501 2017-05-02 Tamar Christina <tamar.christina@arm.com>
26503 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
26505 2017-05-02 Martin Liska <mliska@suse.cz>
26508 * lto-streamer-in.c (lto_read_tree_1): Remove
26509 LTO_STREAMER_DEBUG.
26510 * lto-streamer.c (struct tree_hash_entry): Likewise.
26511 (struct tree_entry_hasher): Likewise.
26512 (tree_entry_hasher::hash): Likewise.
26513 (tree_entry_hasher::equal): Likewise.
26514 (lto_streamer_init): Likewise.
26515 (lto_orig_address_map): Likewise.
26516 (lto_orig_address_get): Likewise.
26517 (lto_orig_address_remove): Likewise.
26518 * lto-streamer.h: Likewise.
26519 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
26520 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
26522 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
26524 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
26525 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
26526 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
26527 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
26528 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
26529 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
26530 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
26531 (mm_maskz_sub_ss): New intrinsics.
26532 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26533 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
26534 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
26535 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
26536 (__builtin_ia32_subss_mask_round): New builtins.
26537 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26538 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
26539 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
26541 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
26542 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
26544 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
26547 2017-05-02 Martin Jambor <mjambor@suse.cz>
26549 PR tree-optimization/78687
26550 * tree-sra.c (access): New field parent.
26551 (process_subtree_disqualification): New function.
26552 (disqualify_candidate): Call it.
26553 (build_accesses_from_assign): Reset write flag if creating an
26555 (build_access_subtree): Fill in parent field and also prpagate
26556 down grp_write flag.
26557 (create_artificial_child_access): New parameter set_grp_write, set
26558 grp_write to its value.
26559 (propagate_subaccesses_across_link): Also propagate grp_write flag
26561 (propagate_all_subaccesses): Push the closest parent back to work
26562 queue if add_access_to_work_queue returned true.
26564 2017-05-02 Richard Biener <rguenther@suse.de>
26566 * common.opt (fstrict-overflow): Alias negative to fwrapv.
26567 * doc/invoke.texi (fstrict-overflow): Remove all traces of
26568 -fstrict-overflow documentation.
26569 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
26570 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
26571 flag_strict_overflow.
26572 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
26573 * lto-opts.c (lto_write_options): Do not stream it.
26574 * lto-wrapper.c (merge_and_complain): Do not handle it.
26575 * opts.c (default_options_table): Do not set -fstrict-overflow.
26576 (finish_options): Likewise do not clear it when sanitizing.
26577 * simplify-rtx.c (simplify_const_relational_operation): Do not
26578 test flag_strict_overflow.
26580 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
26582 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
26583 using enabled attribute.
26584 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
26585 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
26586 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
26587 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
26588 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
26589 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
26590 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
26591 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
26592 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
26593 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
26595 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
26597 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
26599 2017-05-02 Richard Biener <rguenther@suse.de>
26601 PR tree-optimization/80591
26603 2017-04-10 Richard Biener <rguenther@suse.de>
26605 * tree-ssa-structalias.c (find_func_aliases): Properly handle
26608 2017-05-02 Richard Biener <rguenther@suse.de>
26610 PR tree-optimization/80549
26611 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
26612 (cleanup_tree_cfg_noloop): Create forwarders to known loop
26613 headers if they do not have a preheader.
26615 2017-05-02 Martin Liska <mliska@suse.cz>
26618 * common.opt: Fix typo.
26619 * doc/invoke.texi: Likewise.
26621 2017-05-01 Jan Beulich <jbeulich@suse.com>
26623 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
26624 swapping, add (x,x,m,x,n) alternative.
26626 2017-05-01 Nathan Sidwell <nathan@acm.org>
26628 * calls.c (combine_pending_stack_adjustment_and_call): Remove
26629 unnecessary unadjusted_alignment check.
26631 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
26634 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
26636 * gimplify.c (gimplify_cilk_detach): New function.
26637 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
26638 * tree-core.h: Document EXPR_CILK_SPAWN.
26639 * tree.h (EXPR_CILK_SPAWN): Define.
26641 2017-05-01 David Malcolm <dmalcolm@redhat.com>
26643 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
26644 to use new fixit_hint representation, using the "replace" logic.
26645 (get_line_span_for_fixit_hint): Likewise.
26646 (layout::print_any_fixits): Likewise.
26647 (selftest::test_one_liner_many_fixits): Rename to...
26648 (selftest::test_one_liner_many_fixits_1): ...this, and update
26649 comment and expected output to reflect that the multiple fix-it
26650 hints are now consolidated into one insertion.
26651 (selftest::test_one_liner_many_fixits_2): New test.
26652 (selftest::test_diagnostic_show_locus_one_liner): Update for
26654 (selftest::test_fixit_consolidation): Update for fix-it API
26656 * diagnostic.c (print_parseable_fixits): Likewise.
26657 * edit-context.c (edited_line::m_line_events): Convert from
26658 auto_vec <line_event *> to auto_vec <line_event>.
26659 (class line_event): Convert from abstract base class to a concrete
26660 class, taking over the role of replace_event.
26661 (class insert_event): Delete.
26662 (class replace_event): Rename to class line_event. Convert to
26664 (edit_context::add_fixits): Reimplement.
26665 (edit_context::apply_insert): Delete.
26666 (edit_context::apply_replace): Rename to...
26667 (edit_context::apply_fixit): ...this. Convert to half-open range.
26668 (edited_file::apply_insert): Delete.
26669 (edited_file::apply_replace): Rename to...
26670 (edited_file::apply_fixit): ...this.
26671 (edited_line::~edited_line): Drop deletion of events.
26672 (edited_line::apply_insert): Delete.
26673 (edited_line::apply_replace): Rename to...
26674 (edited_line::apply_fixit): ...this. Convert to half-open range.
26675 Update for change to type of m_line_events.
26676 * edit-context.h (edit_context::apply_insert): Delete.
26677 (edit_context::apply_replace): Rename to...
26678 (edit_context::apply_fixit): ...this.
26680 2017-05-01 Martin Sebor <msebor@redhat.com>
26682 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
26685 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
26688 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
26689 __get_cpuid_max returns 0.
26690 (__get_cpuid_count): Ditto.
26692 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
26694 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
26695 replacement expression is another instance of one of its arguments.
26697 2017-05-01 Jakub Jelinek <jakub@redhat.com>
26700 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
26701 check for stack push/pop autoinc.
26702 * config/i386/i386.c (ix86_agi_dependent): Return false
26703 if the only reason why modified_in_p returned true is that
26704 addr is SP based and set_insn is a push or pop.
26706 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
26708 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
26711 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
26714 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
26715 (account_size_time): Use two predicates - exec_pred and
26717 (evaluate_conditions_for_known_args): Compute both clause and
26719 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
26720 (inline_summary_t::duplicate): Update.
26721 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
26723 (compute_inline_parameters): Likewise.
26724 (estimate_edge_size_and_time): Update caluclation of time.
26725 (estimate_node_size_and_time): Compute both time and nonspecialized
26727 (estimate_ipcp_clone_size_and_time): Update.
26728 (inline_merge_summary): Update.
26729 (do_estimate_edge_time): Update.
26730 (do_estimate_edge_size): Update.
26731 (do_estimate_edge_hints): Update.
26732 (inline_read_section, inline_write_summary): Stream both new predicates.
26733 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
26735 (compute_inlined_call_time): Cleanup.
26736 (big_speedup_p): Update.
26737 (edge_badness): Update.
26738 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
26739 (size_time_entry): Replace predicate by exec_predicate and
26740 nonconst_predicate.
26741 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
26742 (estimate_edge_time): Return also nonspec_time.
26743 (reset_edge_growth_cache): Update.
26745 2017-04-29 Jakub Jelinek <jakub@redhat.com>
26747 PR rtl-optimization/80491
26748 * ifcvt.c (noce_process_if_block): When looking for x setter
26749 with missing else_bb, don't check only the insn right before
26750 cond_earliest, but look for the last insn that x is modified in
26751 within the same bb.
26753 PR rtl-optimization/80491
26754 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
26756 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
26758 PR tree-optimization/80487
26759 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
26761 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26763 PR tree-optimization/79697
26764 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
26765 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
26766 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
26768 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
26769 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
26771 2017-04-28 Martin Sebor <msebor@redhat.com>
26773 PR tree-optimization/80523
26774 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
26775 (init_target_to_host_charmap, target_to_host, target_strtol10): New
26777 (maybe_warn, format_directive, parse_directive): Use new functions.
26778 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
26780 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
26782 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
26784 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
26786 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
26787 target_header_dir): Set correctly.
26788 * configure: Regenerated.
26789 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
26790 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
26791 instead of SYSTEM_HEADER_DIR.
26793 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
26795 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
26796 (estimate_local_effects): Likewise.
26797 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
26798 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
26799 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
26800 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
26801 do_estimate_edge_time, estimate_edge_time): Likewise.
26802 * ipa-inline-analysis.c (estimate_node_size_and_time,
26803 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
26804 (estimate_time_after_inlining): Remove.
26806 2017-04-28 Martin Liska <mliska@suse.cz>
26808 * doc/gcov.texi: Enhance documentation of gcov.
26810 2017-04-28 Martin Liska <mliska@suse.cz>
26812 * doc/gcov.texi: Sort options in alphabetic order.
26813 * doc/gcov-dump.texi: Likewise.
26814 * doc/gcov-tool.texi: Likewise.
26815 * gcov.c (print_usage): Likewise.
26816 * gcov-dump.c (print_usage): Likewise.
26817 * gcov-tool.c (print_merge_usage_message): Likewise.
26818 (print_rewrite_usage_message): Likewise.
26819 (print_overlap_usage_message): Likewise.
26821 2017-04-28 Martin Liska <mliska@suse.cz>
26823 PR gcov-profile/53915
26824 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
26826 2017-04-28 Martin Liska <mliska@suse.cz>
26828 PR gcov-profile/79891
26829 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
26830 is marked by compiler as living on a line.
26831 (get_cycles_count): Remove usage of the union.
26832 (output_intermediate_file): Likewise.
26833 (find_source): Fix GNU coding style.
26834 (accumulate_line_counts): Remove old non-all block mode.
26835 (output_lines): Remove usage of the union.
26836 * profile.c (output_location): Include all BBs, even if
26837 belonging to a same line (and file) as a previous BB.
26839 2017-04-28 Martin Liska <mliska@suse.cz>
26841 * gcov.c (process_args): Handle new argument 'w'.
26842 (read_graph_file): Assign ID to BBs.
26843 (output_branch_count): Display BB # if verbose flag is set.
26844 (output_lines): Likewise for arcs.
26845 (print_usage): Add '--verbose' option help.
26846 * doc/gcov.texi: Document --verbose (-w) option.
26848 2017-04-28 Martin Liska <mliska@suse.cz>
26850 * gcov.c (struct block_location_info): New struct.
26851 (process_file): Fill up the new structure.
26852 (read_graph_file): Replace usage of encoding by the newly added
26854 (add_line_counts): Likewise.
26855 (accumulate_line_counts): Remove usage of the union.
26856 (function_info::function_info): New function.
26857 (function_info::~function_info): Likewise.
26858 (process_file): Call delete instead of release_function.
26859 (release_function): Release the function.
26860 (release_structures): Call delete instead of release_function.
26861 (solve_flow_graph): Replace usage of num_blocks.
26862 (find_exception_blocks): Likewise.
26863 (output_lines): Fix GNU coding style.
26865 2017-04-28 Martin Liska <mliska@suse.cz>
26868 * coverage.c (coverage_remove_note_file): New function.
26869 * coverage.h: Declare the function.
26870 * toplev.c (finalize): Clean if an error has been seen.
26872 2017-04-28 Martin Liska <mliska@suse.cz>
26874 PR gcov-profile/80031
26875 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
26876 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
26877 * gcov.c (read_graph_file): Read just number of blocks.
26878 * profile.c (branch_prob): Do not stream 0 flags per a basic
26881 2017-04-28 Martin Liska <mliska@suse.cz>
26883 * gcov-dump.c (tag_*): Add new argument to declarations.
26884 (dump_gcov_file): Likewise.
26885 (tag_blocks): Add and use new argument depth.
26886 (tag_arcs): Likewise.
26887 (tag_lines): Likewise.
26888 (tag_counters): Likewise.
26889 (tag_summary): Likewise.
26890 (dump_working_sets): Use depth to do a proper indentation.
26892 2017-04-28 Jakub Jelinek <jakub@redhat.com>
26895 * cgraph.h (symtab_node::debug_symtab): No longer inline.
26896 * symtab.c (symtab_node::debug_symtab): Move definition here.
26898 2017-04-28 Richard Biener <rguenther@suse.de>
26900 * lto-streamer.h (LTO_major_version): Bump to 7.
26902 2017-04-28 Richard Biener <rguenther@suse.de>
26904 * tree-vrp.c (assert_info): New struct.
26905 (add_assert_info): New helper.
26906 (register_edge_assert_for_2): Refactor to add asserts to a vector
26908 (register_edge_assert_for_1): Likewise.
26909 (register_edge_assert_for): Likewise.
26910 (finish_register_edge_assert_for): New helper actually registering
26911 asserts where live on edge.
26912 (find_conditional_asserts): Adjust.
26913 (find_switch_asserts): Likewise.
26914 (evrp_dom_walker::try_find_new_range): Generalize.
26915 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
26917 2017-04-27 Marek Polacek <polacek@redhat.com>
26920 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
26921 arg10 and arg11 to itype.
26923 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
26925 * doc/extend.texi (Object Size Checking): Improve grammar.
26927 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
26930 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
26931 that the logic for permitting reciprocal estimates matches that
26934 2017-04-27 Jakub Jelinek <jakub@redhat.com>
26937 * tree.c (type_cache_hasher::equal): Only compare
26938 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
26939 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
26940 non-aggregate element types.
26941 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
26942 about the flag on ARRAY_TYPEs in the comment, formatting fix.
26944 2017-04-27 Richard Biener <rguenther@suse.de>
26946 PR middle-end/80533
26947 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
26948 stripping ARRAY_REFs from MEM_EXPR make sure we're not
26949 keeping a reference to a trailing array.
26951 2017-04-27 Richard Biener <rguenther@suse.de>
26953 PR middle-end/80539
26954 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
26955 being in loop-closed SSA form conservatively.
26956 (chrec_fold_multiply_poly_poly): Likewise.
26958 2017-04-27 Tamar Christina <tamar.christina@arm.com>
26960 PR middle-end/79665
26961 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
26962 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
26964 2017-04-27 Jakub Jelinek <jakub@redhat.com>
26967 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
26968 (aarch64_function_arg_alignment): Return unsigned int again, but still
26969 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
26970 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
26971 Don't emit -Wpsabi note.
26972 (aarch64_function_arg_boundary): Likewise.
26973 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
26976 2017-04-26 Nathan Sidwell <nathan@acm.org>
26978 * tree.h (crc32_unsigned_n): Declare.
26979 (crc32_unsigned, crc32_unsigned): Make inline.
26980 * tree.c (crc32_unsigned_bits): Replace with ...
26981 (crc32_unsigned_n): ... this.
26982 (crc32_unsigned, crc32_byte): Remove.
26983 (crc32_string): Remove unnecessary braces.
26985 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
26987 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
26988 * ipa-inline-analysis.c (MAX_TIME): Remove.
26989 (account_size_time): Use sreal for time.
26990 (dump_inline_summary): Update.
26991 (estimate_function_body_sizes): Update.
26992 (estimate_edge_size_and_time): Update.
26993 (estimate_calls_size_and_time): Update.
26994 (estimate_node_size_and_time): Update.
26995 (inline_merge_summary): Update.
26996 (inline_update_overall_summary): Update.
26997 (estimate_time_after_inlining): Update.
26998 (inline_read_section): Update.
26999 (inline_write_summary): Update.
27000 * ipa-inline.c (compute_uninlined_call_time): Update.
27001 (compute_inlined_call_time): Update.
27002 (recursive_inlining): Update.
27003 (inline_small_functions): Update.
27004 (dump_overall_stats): Update.
27005 * ipa-inline.h: Include sreal.h.
27006 (size_time_entry): Turn time to sreal.
27007 (inline_summary): Turn self_time nad time to sreal.
27009 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
27011 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
27013 (sreal::stream_out, sreal::stream_in): New.
27014 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
27016 2017-04-25 Jakub Jelinek <jakub@redhat.com>
27018 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
27021 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
27024 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
27026 (dimode_scalar_chain::compute_convert_gain): Ditto.
27027 (dimode_scalar_chain::make_vector_copies): Ditto.
27028 (dimode_scalar_chain::convert_reg): Ditto.
27029 (dimode_scalar_chain::convert_insn): Ditto.
27030 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
27031 (VI248_AVX512BW_1): New mode iterator.
27032 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
27033 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
27036 2017-04-25 Martin Sebor <msebor@redhat.com>
27038 PR tree-optimization/80497
27039 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
27040 constants are representable in HOST_WIDE_INT.
27041 (parse_directive): Ditto.
27043 2017-04-25 Martin Sebor <msebor@redhat.com>
27046 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
27047 (new_zero_array): Adjust signature.
27048 (dom_info::dom_init): Used unsigned rather that size_t.
27049 (dom_info::dom_info): Same.
27051 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27052 Jakub Jelinek <jakub@redhat.com>
27055 * config/arm/arm.c: Include gimple.h.
27056 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
27057 returns negative, increment ncrn only if it returned positive.
27058 (arm_needs_doubleword_align): Return int instead of bool,
27059 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
27060 members, but if there is any such non-FIELD_DECL
27061 > PARM_BOUNDARY aligned decl, return -1 instead of false.
27062 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
27063 returns negative, increment nregs only if it returned positive.
27064 (arm_setup_incoming_varargs): Likewise.
27065 (arm_function_arg_boundary): Emit -Wpsabi note if
27066 arm_needs_doubleword_align returns negative, return
27067 DOUBLEWORD_ALIGNMENT only if it returned positive.
27069 2017-04-25 Marek Polacek <polacek@redhat.com>
27072 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
27073 first argument to type.
27075 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
27078 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
27079 type checks to test for compatibility instead of equality.
27081 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27082 Jakub Jelinek <jakub@redhat.com>
27085 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
27087 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
27088 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
27089 the alignment computation, but return their maximum in warn_alignment.
27090 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
27091 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
27093 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
27094 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
27097 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27099 * config/arc/simdext.md (dmpyh): Fix typo.
27101 2017-04-25 Richard Biener <rguenther@suse.de>
27103 PR tree-optimization/80492
27104 * alias.c (compare_base_decls): Handle registers with asm
27105 specification conservatively.
27106 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
27107 compare_base_decls returning dont-know properly.
27109 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27111 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
27112 (legitimate_offset_address_p): New function.
27113 (arc_legitimate_address_p): Use above function.
27115 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27117 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
27119 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27121 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
27122 ACCH registers whenever they are available.
27124 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27126 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
27127 double regs fix when not used.
27129 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27131 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
27133 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
27134 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
27136 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27138 * config/arc/arc.c (arc_output_addsi): Check for h-register class
27139 when emitting short ADD instructions.
27141 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27143 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
27145 (cmpsi_cc_c_insn): Likewise.
27146 (cbranchsi4_scratch): Compute proper instruction length using
27147 compact_hreg_operand.
27148 * config/arc/predicates.md (compact_hreg_operand): New predicate.
27150 2017-04-25 Richard Biener <rguenther@suse.de>
27152 PR middle-end/80509
27153 * passes.c (pass_manager::pass_manager): Initialize
27154 m_name_to_pass_map.
27156 2017-04-25 Richard Biener <rguenther@suse.de>
27158 PR tree-optimization/79201
27159 * tree-ssa-sink.c (statement_sink_location): Handle calls.
27161 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27164 * config/s390/vector.md: Split MEM->GPR vector moves for
27165 non-s_operand addresses.
27167 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27170 * config/s390/predicates.md (reload_const_wide_int_operand): New
27172 * config/s390/s390.md ("movti"): Remove d/P alternative.
27173 ("movti_bigconst"): New pattern definition.
27175 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
27178 * s390-protos.h (s390_expand_cs_hqi): Removed.
27179 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
27180 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
27181 modes as well as CCZ1mode and CCZmode.
27182 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
27183 signature of s390_emit_compare_and_swap.
27184 (s390_expand_cs_hqi): Likewise, make static.
27185 (s390_expand_cs_tdsi): Generate an explicit compare before trying
27186 compare-and-swap, in some cases.
27187 (s390_expand_cs): Wrapper function.
27188 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
27190 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
27191 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
27192 patterns for small and large integers. Forbid symref memory operands.
27193 Move expander to s390.c. Require cc register.
27194 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
27195 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
27196 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
27197 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
27198 symref memory operands. Remove CC mode and call s390_match_ccmode
27200 ("atomic_exchange<mode>"): Allow and implement all integer modes.
27202 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
27204 * config/s390/s390.md (define_peephole2): New peephole to help
27205 combining the load-and-test pattern with volatile memory.
27207 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
27209 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
27210 with CCZmode for TARGET_Z196.
27212 2017-04-25 Jakub Jelinek <jakub@redhat.com>
27214 PR rtl-optimization/80501
27215 * combine.c (make_compound_operation_int): Set subreg_code to SET
27216 even for AND with mask of the sign bit of mode.
27218 PR rtl-optimization/80500
27219 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
27220 sum's initial value.
27222 2017-04-25 Julian Brown <julian@codesourcery.com>
27223 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27225 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
27227 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
27229 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
27231 2017-04-25 Julian Brown <julian@codesourcery.com>
27232 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27234 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
27235 (thunderx2t99_sha): New Reservation.
27237 2017-04-25 Julian Brown <julian@codesourcery.com>
27238 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27240 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
27241 type for 1-element load.
27243 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
27245 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
27247 2017-04-24 Martin Jambor <mjambor@suse.cz>
27249 PR tree-optimization/80293
27250 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
27251 char arrays not totally scalarizable if it is false.
27252 (analyze_all_variable_accesses): Pass correct value in the new
27253 parameter. Add a statistics counter.
27255 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
27257 PR middle-end/79931
27258 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
27260 2017-04-24 Richard Biener <rguenther@suse.de>
27262 PR tree-optimization/80494
27263 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
27264 out for complex types.
27266 2017-04-24 Richard Biener <rguenther@suse.de>
27268 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
27269 * tree-ssa-sccvn.c (print_scc): Print SCC size.
27270 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
27271 (DFS): Adjust and never fail.
27272 (sccvn_dom_walker::fail): Remove.
27273 (sccvn_dom_walker::before_dom_children): Adjust.
27274 (run_scc_vn): Likewise and never fail.
27275 * tree-ssa-pre.c (pass_pre::execute): Adjust.
27276 (pass_fre::execute): Likewise.
27278 2017-04-24 Richard Biener <rguenther@suse.de>
27280 PR tree-optimization/79725
27281 * tree-ssa-sink.c (statement_sink_location): Return whether
27282 failure reason was zero uses. Move that check later.
27283 (sink_code_in_bb): Deal with zero uses by removing the stmt
27286 2017-04-24 Richard Biener <rguenther@suse.de>
27289 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
27290 pointer-based references.
27292 2017-04-24 Richard Biener <rguenther@suse.de>
27295 * pass_manager.h (pass_manager::operator new): Remove.
27296 (pass_manager::operator delete): Likewise.
27297 * passes.c (pass_manager::operator new): Remove.
27298 (pass_manager::operator delete): Likewise.
27299 (pass_manager::pass_manager): Zero individual pass members.
27301 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
27304 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
27305 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
27306 Check "XEXP (src, 1)" operand here.
27307 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
27308 Check "XEXP (src, 1)" operand here.
27309 (dimode_scalar_chain::make_vector_copies): Detect count register
27310 of a shift instruction. Zero extend count register from QImode
27311 to DImode to satisfy vector shift pattern count operand predicate.
27312 Substitute vector shift count operand with a DImode copy.
27313 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
27316 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
27318 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
27319 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
27320 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
27321 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
27322 (UNSPEC_NOREX_MEM): Remove definition.
27324 2017-04-21 Richard Biener <rguenther@suse.de>
27326 PR tree-optimization/79547
27327 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
27328 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
27329 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
27330 without any constraints.
27332 2017-04-21 Richard Biener <rguenther@suse.de>
27334 PR tree-optimization/78847
27335 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
27337 2017-04-21 Richard Biener <rguenther@suse.de>
27339 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
27340 (build_distinct_type_copy): Likewise.
27341 (build_variant_type_copy): Likewise.
27342 * tree.c (build_qualified_type): Pass down mem-stat info.
27343 (build_distinct_type_copy): Likewise.
27344 (build_variant_type_copy): Likewise.
27346 2017-04-21 Richard Biener <rguenther@suse.de>
27348 PR tree-optimization/80237
27349 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
27351 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
27352 for a simplified result.
27354 2016-04-21 Richard Biener <rguenther@suse.de>
27356 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
27357 sth as strict as a simple_iv but a chrec without symbols and an
27358 operand defined in the loop we are peeling (and not some subloop).
27359 (propagate_constants_for_unrolling): Propagate all constants.
27361 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
27364 * config/i386/i386.c (print_reg): Remove assert for disalowed
27365 regno values, call output_operand_lossage instead.
27367 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
27370 * config/i386/constraints.md (Yc): New register constraint.
27371 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
27372 Use Yc constraint for alternative 2 of operand 0. Remove
27373 preferred_for_speed attribute.
27375 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
27377 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
27378 lastprivate clauses in SIMT case.
27380 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
27382 * doc/invoke.texi (-Wextra-semi): Document new warning option.
27384 2017-04-20 Richard Biener <rguenther@suse.de>
27386 PR tree-optimization/57796
27387 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
27388 as N scalar stores.
27389 (vect_model_load_cost): Cost gathers as N scalar loads.
27391 2017-04-20 Richard Biener <rguenther@suse.de>
27393 * ggc-page.c (ggc_allocated_p): Rename to ...
27394 (safe_lookup_page_table_entry): ... this and return the lookup
27396 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
27398 2017-04-20 Richard Biener <rguenther@suse.de>
27400 PR tree-optimization/80453
27401 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
27402 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
27403 from the conditions.
27404 (vn_phi_eq): Pass them down.
27405 (vn_phi_lookup): Record them.
27406 (vn_phi_insert): Likewise.
27408 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
27410 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
27411 uninitialized variable warning to avoid buffer overrun.
27413 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
27416 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
27417 is suppressed for '{ 0 }' in C.
27419 2017-04-20 Jakub Jelinek <jakub@redhat.com>
27421 * BASE-VER: Set to 8.0.0.
27423 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
27425 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
27426 priority .init_array and .fini_array section with SECTION_NOTYPE
27429 2017-04-20 Jakub Jelinek <jakub@redhat.com>
27431 PR middle-end/80423
27432 * tree.h (build_array_type): Add typeless_storage default argument.
27433 * tree.c (type_cache_hasher::equal): Also compare
27434 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
27435 (build_array_type): Add typeless_storage argument, set
27436 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
27438 (build_nonshared_array_type): Adjust build_array_type_1 caller.
27439 (build_array_type): Likewise. Add typeless_storage argument.
27441 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
27442 Jakub Jelinek <jakub@redhat.com>
27444 PR tree-optimization/80426
27445 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
27446 operation on symbolic operands, also compute the overflow for the
27447 invariant part when the operation degenerates into a negation.
27449 2017-04-19 Jakub Jelinek <jakub@redhat.com>
27452 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
27453 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
27456 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
27458 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
27461 * config/avr/avr.c (tree.h): Include it.
27462 (cgraph.h): Include it.
27463 (avr_encode_section_info): Don't warn for uninitialized progmem
27464 variable if it's just an alias.
27466 2017-04-19 Richard Biener <rguenther@suse.de>
27469 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
27470 when needed by AutoPGO.
27472 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
27475 * doc/lto.texi: Remove an extra 'that'.
27477 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
27479 PR rtl-optimization/80429
27480 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
27481 are only used in debug insns.
27483 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
27484 Vladimir Makarov <vmakarov@redhat.com>
27486 * config/sparc/predicates.md (input_operand): Add comment. Return
27487 true for any memory operand when LRA is in progress.
27488 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
27490 2017-04-18 Jeff Law <law@redhat.com>
27493 * mips.md ({return,simple_return}_internal): Do not overwrite
27496 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27498 PR tree-optimization/80443
27499 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
27500 instead of adding 1, subtract -1 and similarly instead of subtracting
27503 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
27505 PR rtl-optimization/80357
27506 * haifa-sched.c (tmp_bitmap): New variable.
27507 (model_recompute): Handle duplicate use records.
27508 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
27509 (free_global_sched_pressure_data): Free it.
27511 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
27514 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
27515 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
27516 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
27517 instead of SYSTEM_HEADER_DIR.
27519 2017-04-18 Jeff Law <law@redhat.com>
27521 PR middle-end/80422
27522 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
27523 predecessors after walking up the insn chain.
27525 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27528 * dwarf2out.c (modified_type_die): Try harder not to emit internal
27529 sizetype type into debug info.
27531 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
27534 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
27535 unneeded test for TARGET_UPPER_REGS_SF.
27536 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
27538 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27541 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
27542 instead of gsi_after_labels.
27544 2017-04-18 Jeff Law <law@redhat.com>
27546 * regcprop.c (maybe_mode_change): Avoid creating copies of the
27550 2017-04-13 Jeff Law <law@redhat.com>
27551 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
27552 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
27554 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
27557 * config/avr/avr.c (intl.h): Include it.
27558 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
27560 2017-04-18 Martin Liska <mliska@suse.cz>
27562 PR gcov-profile/78783
27563 * gcov-tool.c (gcov_output_files): Validate that destination
27564 file is either removed by the tool or by a user.
27566 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
27567 Guy Benyei <guybe@mellanox.com>
27569 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
27570 block, and do not negate it, the stored id is already negative.
27572 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
27574 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
27576 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
27579 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
27580 masks of options that should be turned off if the VSX vector
27581 options are turned off.
27582 (OTHER_P8_VECTOR_MASKS): Likewise.
27583 (OTHER_VSX_VECTOR_MASKS): Likewise.
27584 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
27585 rs6000_disable_incompatible_switches to validate no type switches
27587 (rs6000_incompatible_switch): New function to disallow turning on
27588 other vector options if -mno-vsx, -mno-power8-vector, or
27589 -mno-power9-vector are specified.
27591 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27593 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
27595 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27597 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
27598 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
27599 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
27600 (ARG_POINTER_CFA_OFFSET): Likewise.
27602 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27604 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
27605 conditions to take advantage of various optimizations.
27607 2017-04-13 Jeff Law <law@redhat.com>
27609 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
27610 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
27611 (zero_extendsidi2_dext): Likewise.
27613 2017-04-13 Jakub Jelinek <jakub@redhat.com>
27616 * fold-const.c (fold_ternary_loc): Revert
27617 use op0 instead of fold_convert_loc (loc, type, arg0) part of
27620 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
27622 PR rtl-optimization/80343
27623 * lra-remat.c (update_scratch_ops): Assign original hard reg to
27624 new scratch pseudo.
27626 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
27629 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
27630 to ubsan_encode_value.
27632 2017-04-13 Jeff Law <law@redhat.com>
27634 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
27635 appearing in DEBUG_INSNs.
27637 2017-04-13 Martin Liska <mliska@suse.cz>
27639 PR gcov-profile/80413
27640 * gcov-io.c (gcov_write_string): Copy to buffer just when
27641 allocated size is greater than zero.
27643 2017-04-13 Jakub Jelinek <jakub@redhat.com>
27646 * dwarf2out.c (decls_for_scope): Ignore declarations of
27647 current_function_decl in BLOCK_NONLOCALIZED_VARS.
27649 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
27652 * ipa-visibility.c (non_local_p): Fix typos.
27653 (localize_node): When localizing symbol in same comdat group,
27654 dissolve the group only when we know external symbols are going
27656 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
27658 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27660 PR tree-optimization/79390
27661 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
27662 order does not result in usable sequence, retry with reversed operand
27668 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
27669 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
27670 op0 instead of fold_convert_loc (loc, type, arg0).
27672 2017-04-12 Jeff Law <law@redhat.com>
27674 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
27675 has a delay slot in the generated code.
27677 * config/cris/cris.md (cris_preferred_reload_class): Return
27678 GENNONACR_REGS rather than GENERAL_REGS.
27680 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27683 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
27684 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
27685 signedness of the result type.
27687 2017-04-12 Richard Biener <rguenther@suse.de>
27688 Jeff Law <law@redhat.com>
27690 PR tree-optimization/80359
27691 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
27692 trim stores to TARGET_MEM_REFs.
27694 2017-04-12 Richard Biener <rguenther@suse.de>
27696 PR tree-optimization/79390
27697 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
27698 threading case even more.
27700 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
27703 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
27704 for quad_address_p for TImode, instead of just not indexed_address.
27706 2017-04-12 Richard Biener <rguenther@suse.de>
27707 Bernd Edlinger <bernd.edlinger@hotmail.de>
27709 PR middle-end/79671
27710 * alias.c (component_uses_parent_alias_set_from): Handle
27711 TYPE_TYPELESS_STORAGE.
27712 (get_alias_set): Likewise.
27713 * tree-core.h (tree_type_common): Add typeless_storage flag.
27714 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
27715 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
27716 for types containing members with TYPE_TYPELESS_STORAGE.
27717 (place_field): Likewise.
27718 (layout_type): Likewise for ARRAY_TYPE.
27719 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
27720 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
27721 TYPE_TYPELESS_STORAGE.
27722 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
27724 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27727 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
27728 first argument to type.
27730 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27734 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
27735 CONST0_RTX (mode) rather than const0_rtx where appropriate.
27736 (rs6000_expand_binop_builtin): Likewise.
27737 (rs6000_expand_ternop_builtin): Likewise; also add missing
27738 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
27739 vshasigma built-ins.
27740 * doc/extend.texi: Document that vec_xxpermdi's third argument
27741 must be a constant.
27743 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
27745 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
27746 Use shift_const cost parameter when calculating gain of STV shifts.
27748 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
27750 PR rtl-optimization/70478
27751 * lra-constraints.c (process_alt_operands): Check memory for
27752 disfavoring memory insn operand.
27754 2017-04-11 Jakub Jelinek <jakub@redhat.com>
27756 PR middle-end/80100
27757 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
27758 left shift in unsigned HOST_WIDE_INT type.
27760 PR rtl-optimization/80385
27761 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
27762 (not (neg X)) into (plus X -1) for complex or non-integral modes.
27765 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
27766 if they have any depend clauses.
27768 2017-04-11 Martin Liska <mliska@suse.cz>
27771 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
27772 * ipa-split.c (split_function): Create a local comdat symbol
27773 if caller is in a comdat group.
27775 2017-04-11 Martin Liska <mliska@suse.cz>
27778 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
27781 2017-04-11 Martin Sebor <msebor@redhat.com>
27783 PR middle-end/80364
27784 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
27785 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
27787 (directive::set_width, directive::set_precision, format_character):
27789 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
27792 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
27795 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
27796 conflict, set target->arch_name instead of target->cpu_name.
27798 2017-04-11 Richard Biener <rguenther@suse.de>
27800 PR tree-optimization/80374
27801 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
27802 build_zero_cst, remove fold_convertible_p check again.
27804 2017-04-11 Martin Liska <mliska@suse.cz>
27807 * ubsan.c (instrument_object_size): Do not instrument register
27810 2017-04-11 Jakub Jelinek <jakub@redhat.com>
27813 * config/i386/i386-builtin-types.def
27814 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
27815 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
27816 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
27817 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
27818 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
27819 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
27820 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
27821 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
27822 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
27823 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
27824 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
27825 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
27826 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
27827 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
27828 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
27829 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
27830 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
27831 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
27832 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
27833 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
27834 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
27835 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
27836 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
27837 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
27838 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
27839 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
27840 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
27841 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
27842 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
27843 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
27844 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
27845 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
27846 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
27847 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
27848 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
27849 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
27850 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
27851 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
27852 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
27853 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
27854 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
27855 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
27856 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
27857 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
27858 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
27859 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
27861 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
27862 flag to second_arg_count, handle 4 argument function type _COUNT
27863 aliases, handle second_arg_count on second argument rather than last.
27865 2017-04-10 Jeff Law <law@redhat.com>
27867 PR tree-optimization/80374
27868 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
27869 record anything if we can not convert integer_zero_node to the
27872 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
27875 * config/rs6000/rs6000.c (rs6000_option_override_internal):
27876 Enhance special handling given to the TARGET_P9_MINMAX option in
27877 relation to certain other options.
27879 2017-04-10 Bin Cheng <bin.cheng@arm.com>
27881 PR tree-optimization/80153
27882 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
27883 remove POINTER_PLUS_EXPR's base part directly, rather than through
27886 2017-04-10 Richard Biener <rguenther@suse.de>
27887 Bin Cheng <bin.cheng@arm.com>
27889 PR tree-optimization/80153
27890 * tree-affine.c (aff_combination_to_tree): Get base pointer from
27891 the first element of pointer type aff_tree. Build result expr in
27893 (add_elt_to_tree): Convert to type unconditionally. Remove other
27894 fold_convert calls.
27895 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
27896 (rewrite_use_nonlinear_expr): Check invariant using iv information.
27898 2017-04-10 Richard Biener <rguenther@suse.de>
27900 * tree-ssa-structalias.c (find_func_aliases): Properly handle
27903 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
27905 PR rtl-optimization/70478
27906 * lra-constraints.c (curr_small_class_check): New.
27907 (update_and_check_small_class_inputs): New.
27908 (process_alt_operands): Update curr_small_class_check. Disfavor
27909 alternative insn memory operands. Check available regs for small
27912 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
27915 * config/mips/mips.opt (-mvirt): Update description.
27916 * doc/invoke.texi (-mvirt): Likewise.
27918 2017-04-10 Richard Biener <rguenther@suse.de>
27920 PR middle-end/80362
27921 * fold-const.c (fold_binary_loc): Look at unstripped ops when
27922 looking for NEGATE_EXPR in -A / -B to A / B folding.
27924 2017-04-10 Martin Liska <mliska@suse.cz>
27926 PR gcov-profile/80224
27927 * gcov.c (print_usage): Fix usage string.
27928 (get_gcov_intermediate_filename): Remove.
27929 (output_gcov_file): Use both for normal and intermediate format.
27930 (generate_results): Do not initialize special file for
27931 intermediate format.
27933 2017-04-10 Richard Biener <rguenther@suse.de>
27935 PR tree-optimization/80304
27936 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
27939 2017-04-10 Nathan Sidwell <nathan@acm.org>
27942 * config/rs6000/rs6000.c (rs6000_vector_type): New.
27943 (rs6000_init_builtins): Use it.
27945 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27947 * config/arm/arm.md (<mrc>): Add mode to SET source.
27948 (<mrrc>): Likewise.
27950 2017-04-10 Richard Biener <rguenther@suse.de>
27952 PR middle-end/80344
27953 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
27955 2017-04-10 Jakub Jelinek <jakub@redhat.com>
27958 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
27959 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
27960 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
27961 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
27962 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
27963 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
27964 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
27965 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
27966 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
27967 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
27968 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
27969 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
27970 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
27971 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
27972 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
27973 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
27974 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
27975 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
27976 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
27977 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
27978 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
27979 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
27980 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
27982 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
27984 PR rtl-optimization/70478
27985 * lra-constraints.c: Reverse the last patch.
27987 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
27989 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
27990 Add comment for WCHAR_T.
27992 2017-04-08 Martin Liska <mliska@suse.cz>
27995 2017-04-07 Martin Liska <mliska@suse.cz>
27998 * ipa-split.c (split_function): Add function part to a same comdat
28001 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28004 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
28006 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
28008 * rs6000/rs6000.c (vec_load_pendulum): Rename...
28009 (vec_pairing): ...to this.
28010 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
28011 (rs6000_sched_init): Adjust for name change.
28012 (struct rs6000_sched_context): Likewise.
28013 (rs6000_init_sched_context): Likewise.
28014 (rs6000_set_sched_context): Likewise.
28016 2017-04-07 Jakub Jelinek <jakub@redhat.com>
28022 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
28024 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
28025 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
28026 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
28028 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
28030 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
28032 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
28034 PR rtl-optimization/70703
28035 * ira-color.c (update_conflict_hard_regno_costs): Use
28036 int64_t instead of HOST_WIDE_INT.
28038 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
28040 PR rtl-optimization/70478
28041 * lra-constraints.c (process_alt_operands): Disfavor alternative
28042 insn memory operands.
28044 2017-04-07 Jeff Law <law@redhat.com>
28046 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
28047 CALL and NOTE_INSN_CALL_ARG_LOCATION.
28049 2017-04-07 Martin Liska <mliska@suse.cz>
28052 * config/aarch64/aarch64.c (aarch64_process_target_attr):
28053 Show error message instead of an ICE.
28055 2017-04-07 Martin Liska <mliska@suse.cz>
28058 * ipa-split.c (split_function): Add function part to a same comdat
28061 2017-04-07 Richard Biener <rguenther@suse.de>
28063 PR middle-end/80341
28064 * tree.c (get_unwidened): Also handle ! for_type case for
28066 * convert.c (do_narrow): Split out from ...
28067 (convert_to_integer_1): ... here. Do not pass final truncation
28068 type to get_unwidened for TRUNC_DIV_EXPR.
28070 2017-04-07 Richard Biener <rguenther@suse.de>
28072 * tree-affine.c (wide_int_ext_for_comb): Take type rather
28074 (aff_combination_const): Adjust.
28075 (aff_combination_scale): Likewise.
28076 (aff_combination_add_elt): Likewise.
28077 (aff_combination_add_cst): Likewise.
28078 (aff_combination_convert): Likewise.
28079 (add_elt_to_tree): Likewise. Remove unused argument.
28080 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
28082 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
28084 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
28086 * config/arm/arm.c (arm_default_short_enums): Use
28087 ARM_DEFAULT_SHORT_ENUMS.
28088 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
28090 2017-04-06 Jakub Jelinek <jakub@redhat.com>
28093 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
28094 members with redundant out-of-class redeclaration.
28096 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
28099 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
28100 * config/i386/i386.md (*zero_extendsidi2):
28101 Add (?*x,*x) and (?*v,*v) alternatives.
28103 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
28106 * config/i386/i386.c (ix86_expand_builtin)
28107 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
28108 mode from insn data. Convert operands to insn operand mode.
28109 Copy operands that don't satisfy insn predicate to a register.
28111 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
28113 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
28116 2017-04-06 Richard Biener <rguenther@suse.de>
28118 PR tree-optimization/80334
28119 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
28120 preserve alignment of accesses.
28122 2017-04-06 Richard Biener <rguenther@suse.de>
28124 PR tree-optimization/80262
28125 * tree-sra.c (build_ref_for_offset): Preserve address-space
28127 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
28128 Drop useless address-space information on MEM_REF offsets.
28130 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
28132 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
28134 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
28136 PR rtl-optimization/70703
28137 * ira-color.c (update_conflict_hard_regno_costs): Use
28138 HOST_WIDE_INT instead of long.
28140 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
28143 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
28144 not defined for x86_64 target. Add -mmmx target option when __SSE2__
28146 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
28147 for x86_64 target. Handle -m3dnowa option.
28149 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
28151 PR rtl-optimization/70703
28152 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
28153 (update_conflict_hard_regno_costs): Use long instead of unsigned
28154 arithmetic for cost calculation.
28156 2017-04-05 Jakub Jelinek <jakub@redhat.com>
28157 Bernd Edlinger <bernd.edlinger@hotmail.de>
28160 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
28163 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
28166 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
28167 ptr_mode with Pmode throughout.
28168 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
28169 into probe_stack_range and use DImode.
28171 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
28174 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
28175 call_eh_return is true.
28177 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28179 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
28180 Initialize last_match_fntype_index.
28182 2017-04-05 Jakub Jelinek <jakub@redhat.com>
28185 * tree-nvr.c: Include internal-fn.h.
28186 (pass_return_slot::execute): Ignore internal calls without
28189 2017-04-04 Jakub Jelinek <jakub@redhat.com>
28190 Richard Biener <rguenther@suse.de>
28193 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
28194 captures used multiple times, except for the last use.
28195 * generic-match-head.c: Include gimplify.h.
28197 2017-04-04 Jakub Jelinek <jakub@redhat.com>
28199 PR tree-optimization/79390
28200 * target.h (struct noce_if_info): Declare.
28201 * targhooks.h (default_noce_conversion_profitable_p): Declare.
28202 * target.def (noce_conversion_profitable_p): New target hook.
28203 * ifcvt.h (struct noce_if_info): New type, moved from ...
28204 * ifcvt.c (struct noce_if_info): ... here.
28205 (noce_conversion_profitable_p): Renamed to ...
28206 (default_noce_conversion_profitable_p): ... this. No longer
28208 (noce_try_store_flag_constants, noce_try_addcc,
28209 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
28210 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
28211 instead of noce_conversion_profitable_p.
28212 * config/i386/i386.c: Include ifcvt.h.
28213 (ix86_option_override_internal): Don't override
28214 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
28215 (ix86_noce_conversion_profitable_p): New function.
28216 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
28217 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
28218 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
28219 * doc/tm.texi: Regenerated.
28221 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28223 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
28226 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
28229 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
28230 instructions for small multiply cores.
28232 2017-04-04 Jeff Law <law@redhat.com>
28234 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
28236 (mips_expand_vec_perm_const): Initialize elements in orig_perm
28237 that are not set by the loop over the elements.
28239 2017-04-04 Jakub Jelinek <jakub@redhat.com>
28242 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
28243 int mode, convert_modes it to mode as unsigned, otherwise use
28244 lowpart_subreg to mode rather than SImode.
28245 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
28246 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
28247 Use DImode instead of SImode for the shift count operand.
28248 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
28251 2017-04-04 Richard Biener <rguenther@suse.de>
28253 PR middle-end/80281
28254 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
28255 arithmetic done for the negate or the plus. Simplify.
28256 (A - (-B) -> A + B): Likewise.
28257 * fold-const.c (split_tree): Make sure to not negate pointers.
28259 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
28261 PR rtl-optimization/60818
28262 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
28263 a compare of comparisons with the thing compared if this results
28264 in a different machine mode.
28266 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
28268 * alias.c (base_alias_check): Fix typo in comment.
28269 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
28270 * cgraphunit.c (symbol_table::compile): Likewise.
28271 * collect2.c (maybe_run_lto_and_relink): Likewise.
28272 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
28273 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
28274 * config/avr/avr.c (avr_map_op_t): Likewise.
28275 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
28276 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
28277 * config/epiphany/epiphany.md (movcc): Likewise.
28278 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
28279 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
28281 * config/mips/mips.c (mips_save_restore_reg): Likewise.
28282 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
28283 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
28284 * config/sh/sh.c (sh_rtx_costs): Likewise.
28285 * fold-const.c (fold_truth_andor): Likewise.
28286 * genautomata.c (collapse_flag): Likewise.
28287 * gengtype.h (struct type::u::s): Likewise.
28288 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
28289 * input.c (FORMAT_AMOUNT): Likewise.
28290 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
28291 (known_aggs_to_agg_replacement_list): Likewise.
28292 * ipa-inline-analysis.c: Likewise.
28293 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
28294 * ipa-polymorphic-call.c
28295 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
28296 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
28297 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
28299 * modulo-sched.c (apply_reg_moves): Likewise.
28300 * omp-expand.c (build_omp_regions_1): Likewise.
28301 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
28302 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
28303 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
28304 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
28305 * value-prof.c: Likewise.
28306 * var-tracking.c (val_reset): Likewise.
28308 2017-04-03 Richard Biener <rguenther@suse.de>
28310 PR tree-optimization/80275
28311 * fold-const.c (split_address_to_core_and_offset): Handle
28314 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
28316 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
28317 descriptors is at least equal to that of functions.
28319 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
28321 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
28323 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
28326 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
28327 (mov<IMOD4:mode>): New expander.
28328 (*mov<IMOD4:mode>_internal): New insn and split pattern.
28330 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
28332 PR rtl-optimization/79405
28333 * fwprop.c (propagations_left): New variable.
28334 (forward_propagate_into): Decrement it.
28335 (fwprop_init): Initialize it.
28336 (fw_prop): If the variable has reached zero, stop propagating.
28337 (fwprop_addr): Ditto.
28339 2017-03-31 Jakub Jelinek <jakub@redhat.com>
28342 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
28343 a FUNCTION_DECL, pass it as decl instead of origin to
28346 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
28348 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
28351 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
28354 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
28355 TARGET_VSX_SMALL_INTEGER.
28357 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28359 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
28360 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
28362 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
28364 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
28365 extraction from odd-numbered MSA register.
28367 2017-03-31 Jakub Jelinek <jakub@redhat.com>
28369 PR middle-end/80173
28370 * expmed.c (store_bit_field_1): Don't attempt to create
28371 a word subreg out of hard registers wider than word if they
28372 have HARD_REGNO_NREGS of 1 for their mode.
28374 PR middle-end/80163
28375 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
28376 conversions to integer types wider than word and pointer.
28379 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
28380 (rtx_equal_for_cselib_p): Pass 0 to it.
28381 * cselib.c (cselib_hasher::equal): Likewise.
28382 (rtx_equal_for_cselib_1): Add depth argument. If depth
28383 is 128, don't look up VALUE locs and punt. Increment
28384 depth in recursive calls when walking VALUE locs.
28386 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
28388 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
28389 (make_gcov_file_name): Use the canonical path name for generating
28391 (read_line): Fix handling of files with ascii null bytes.
28393 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
28395 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
28396 to initialise a vector register instead
28397 of using a const_int.
28399 2017-03-30 Jakub Jelinek <jakub@redhat.com>
28401 PR translation/80189
28402 * gimplify.c (omp_default_clause): Use %qs instead of %s in
28403 diagnostic messages.
28405 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
28408 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
28409 (dfp_diex_<mode>): Update mode of operand 1.
28410 * doc/extend.texi (dxex, dxexq): Document change to return type.
28411 (diex, diexq): Document change to argument type.
28413 2017-03-30 Martin Jambor <mjambor@suse.cz>
28416 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
28417 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
28418 it reflects the signature changes performed at the callee side.
28419 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
28420 to cgraph_build_function_type_skip_args.
28421 (build_function_decl_skip_args): Adjust call to the above function.
28423 2017-03-30 Jakub Jelinek <jakub@redhat.com>
28426 * config/i386/sse.md
28427 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
28428 register as dest whenever it is a MEM not rtx_equal_p to the
28429 corresponding dup operand, and when forcing into reg move the
28430 reg into the memory afterwards.
28431 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
28432 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
28433 for the force_reg mode.
28434 (avx512vl_vextractf128<mode>): Use register as dest either
28435 always when a MEM, or when it is a MEM not rtx_equal_p to the
28436 corresponding dup operand, or even not when it is a CONST_VECTOR
28437 depending on the mode and lo vs. hi.
28438 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
28440 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
28441 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
28442 Likewise. Require that operands[2] is even.
28443 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
28444 Remove extraneous parens. Require that operands[2] is a multiple
28446 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
28447 operands[0] is a MEM if <mask_applied>, the predicates/constraints
28448 disallow memory then.
28450 2017-03-30 Richard Biener <rguenther@suse.de>
28452 PR tree-optimization/77498
28453 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
28454 to non-constants over backedges.
28456 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
28458 PR rtl-optimization/80233
28459 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
28460 as last_combined_insn. Do not test for BARRIER_P separately.
28462 2017-03-29 Andreas Schwab <schwab@suse.de>
28465 * calls.c (prepare_call_address): Convert funexp to Pmode before
28466 copying to temp reg.
28468 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28470 PR tree-optimization/80158
28471 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
28472 Handle possible future case of more than one alternate
28474 (replace_rhs_if_not_dup): Likewise.
28475 (replace_one_candidate): Likewise.
28477 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
28479 PR rtl-optimization/80193
28480 * ira.c (ira): Do not check allocation for LRA.
28482 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
28484 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
28485 (nvptx_output_simt_exit): Declare.
28486 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
28487 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
28488 (init_softstack_frame): Move initialization of crtl->is_leaf to...
28489 (nvptx_declare_function_name): ...here. Emit declaration of local
28490 memory space buffer for omp_simt_enter insn.
28491 (nvptx_output_unisimt_switch): New.
28492 (nvptx_output_softstack_switch): New.
28493 (nvptx_output_simt_enter): New.
28494 (nvptx_output_simt_exit): New.
28495 * config/nvptx/nvptx.h (struct machine_function): New fields
28496 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
28497 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
28498 (UNSPECV_SIMT_EXIT): Ditto.
28499 (omp_simt_enter_insn): New insn.
28500 (omp_simt_enter): New expansion.
28501 (omp_simt_exit): New insn.
28502 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
28504 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
28505 (expand_GOMP_SIMT_ENTER_ALLOC): New.
28506 (expand_GOMP_SIMT_EXIT): New.
28507 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
28508 (GOMP_SIMT_ENTER_ALLOC): Ditto.
28509 (GOMP_SIMT_EXIT): Ditto.
28510 * target-insns.def (omp_simt_enter): New insn.
28511 (omp_simt_exit): Ditto.
28512 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
28514 (lower_rec_simd_input_clauses): Implement SIMT privatization.
28515 (lower_rec_input_clauses): Likewise.
28516 (lower_lastprivate_clauses): Handle SIMT privatization.
28518 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
28519 (ompdevlow_adjust_simt_enter): New.
28520 (find_simtpriv_var_op): New.
28521 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
28522 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
28524 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
28525 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
28526 (copy_decl_for_dup_finish): Ditto.
28528 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
28530 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
28533 * config/i386/i386.c (ix86_option_override_internal): Always
28534 allow -mpreferred-stack-boundary=3 for 64-bit targets.
28536 2017-03-28 Bin Cheng <bin.cheng@arm.com>
28538 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
28540 2017-03-28 Bin Cheng <bin.cheng@arm.com>
28542 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
28543 mark new edge's irreducible flag accordign to it.
28544 (vect_do_peeling): Check loop preheader edge's irreducible flag
28545 and pass it to function slpeel_add_loop_guard.
28547 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
28549 PR tree-optimization/80218
28550 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
28551 Update block frequencies and counts.
28553 2017-03-28 Richard Biener <rguenther@suse.de>
28555 PR tree-optimization/78644
28556 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
28557 of a simplification result we may not use it at all.
28559 2017-03-28 Richard Biener <rguenther@suse.de>
28562 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
28563 without arguments, generate default definition of a SSA name.
28565 2017-03-28 Richard Biener <rguenther@suse.de>
28567 PR middle-end/80222
28568 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
28569 TYPE_REF_CAN_ALIAS_ALL references.
28570 * fold-const.c (fold_indirect_ref_1): Likewise.
28572 2017-03-28 Martin Liska <mliska@suse.cz>
28575 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
28576 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
28578 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
28579 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
28581 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
28582 (EXTRA_SPECS): Define.
28583 (SUBTARGET_EXTRA_SPECS): Likewise.
28584 (SUBTARGET_CPP_SPEC): Likewise.
28585 * config/arc/elf.h (EXTRA_SPECS): Renamed to
28586 SUBTARGET_EXTRA_SPECS.
28587 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
28589 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
28591 * config/arc/simdext.md (vst64_insn): Update pattern.
28592 (vld32wh_insn): Likewise.
28593 (vld32wl_insn): Likewise.
28594 (vld64_insn): Likewise.
28595 (vld32_insn): Likewise.
28597 2017-03-28 Marek Polacek <polacek@redhat.com>
28600 * fold-const.c (fold_comparison): Use protected_set_expr_location
28601 instead of SET_EXPR_LOCATION.
28603 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
28605 * tree.c (add_expr): Avoid name lookup warning.
28607 2017-03-27 Jeff Law <law@redhat.com>
28609 PR tree-optimization/80216
28610 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
28611 function name. Limit recursion depth.
28612 (record_temporary_equivalences): Corresponding changes.
28614 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
28616 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
28619 2017-03-27 Jakub Jelinek <jakub@redhat.com>
28622 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
28624 * cfgcleanup.c (reg_note_cfa_p): New array.
28625 (insns_have_identical_cfa_notes): New function.
28626 (old_insns_match_p): Don't cross-jump in between /f
28627 and non-/f instructions. If both i1 and i2 are frame related,
28628 verify all CFA notes, their order and content.
28630 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
28633 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
28634 HImode and SImode with zero extend to DImode to one insn.
28635 (bswap<mode>2_extenddi): Likewise.
28636 (bswapsi2_extenddi): Likewise.
28637 (bswaphi2_extendsi): Likewise.
28638 (bswaphi2): Combine bswap HImode and SImode into one insn.
28639 Separate memory insns from swapping register.
28640 (bswapsi2): Likewise.
28641 (bswap<mode>2): Likewise.
28642 (bswaphi2_internal): Delete, no longer used.
28643 (bswapsi2_internal): Likewise.
28644 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
28645 store, and gpr<-gpr swap insns.
28646 (bswap<mode>2_store): Likewise.
28647 (bswaphi2_reg): Register only splitter, combine with the splitter.
28648 (bswaphi2 splitter): Likewise.
28649 (bswapsi2_reg): Likewise.
28650 (bswapsi2 splitter): Likewise.
28651 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
28652 the insns into load, store, and register/register insns.
28653 (bswapdi2_ldbrx): Likewise.
28654 (bswapdi2_load): Likewise.
28655 (bswapdi2_store): Likewise.
28656 (bswapdi2_reg): Likewise.
28658 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
28660 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
28661 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
28663 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
28666 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
28668 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
28669 special handling for target option conflicts between dform
28670 options (-mpower9-dform, -mpower9-dform-vector,
28671 -mpower9-dform-scalar) and -mno-direct-move.
28673 2017-03-27 Richard Biener <rguenther@suse.de>
28675 PR tree-optimization/80181
28676 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
28678 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28680 * config/arc/predicates.md (move_double_src_operand): Replace the
28681 call to move_double_src_operand with a call to address_operand.
28683 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28685 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
28686 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
28687 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
28689 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28691 * config/arc/predicates.md (long_immediate_loadstore_operand):
28692 Consider scaled addresses cases.
28694 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28696 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
28697 restored when in interrupt.
28698 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
28699 doesn't have delay slot.
28701 2017-03-27 Richard Biener <rguenther@suse.de>
28704 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
28705 inlined thunk clones.
28707 2017-03-27 Jakub Jelinek <jakub@redhat.com>
28710 * asan.c (instrument_derefs): Copy over last operand from
28711 original COMPONENT_REF to the new COMPONENT_REF with
28712 DECL_BIT_FIELD_REPRESENTATIVE.
28713 * ubsan.c (instrument_object_size): Likewise.
28715 2017-03-27 Richard Biener <rguenther@suse.de>
28717 PR tree-optimization/80170
28718 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
28719 sure DR/SCEV didnt fold in constants we do not see when looking
28720 at the reference base alignment.
28722 2017-03-27 Richard Biener <rguenther@suse.de>
28724 PR middle-end/80171
28725 * gimple-fold.c (fold_ctor_reference): Properly guard against
28726 NULL return value from canonicalize_constructor_val.
28728 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
28731 * config/i386/i386.c (ix86_expand_builtin)
28732 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
28733 flags reg setting and flags reg using instructions.
28734 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
28735 clobbering instructions to zero extend op2.
28737 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
28739 * doc/install.texi (Configuration) <--with-aix-soname>:
28740 Update link to AIX ld.
28742 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
28744 PR rtl-optimization/80160
28745 PR rtl-optimization/80159
28746 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
28747 reg_alternate_class into account.
28749 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
28752 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
28753 to consider in curr_insn_transform.
28755 2017-03-24 Jakub Jelinek <jakub@redhat.com>
28757 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
28758 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
28759 and emit_mode_inner.
28761 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28763 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
28764 argument to the overloaded builtin variants. Use the new flag to
28765 deprecate certain builtin variants.
28766 * config/s390/s390-builtin-types.def: Add new builtin types.
28767 * config/s390/s390-builtins.h: Support new flags field for
28768 overloaded builtins.
28769 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
28770 (s390_macro_to_expand): Enable vector float data type.
28771 (s390_cpu_cpp_builtins_internal): Indicate support of the new
28772 builtins by incrementing the __VEC__ version number.
28773 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
28775 (s390_resolve_overloaded_builtin): Emit error messages depending
28776 on the builtin flags.
28777 * config/s390/s390.c (s390_expand_builtin): Support additional
28778 flags argument. Change error message to match the messages
28779 emitted in s390-c.c.
28780 * config/s390/s390.md: New UNSPEC_* constants.
28781 (op_type): Add new instruction types.
28782 * config/s390/vecintrin.h: Add new builtins and test data class
28784 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
28785 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
28786 (VEC_INEXACT, VEC_NOINEXACT): New constants.
28787 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
28788 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
28789 ("vec_mergel<mode>"): V_HW -> VEC_HW.
28791 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
28792 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
28793 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
28794 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
28796 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
28797 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
28798 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
28799 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
28801 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
28802 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
28803 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
28804 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
28805 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
28806 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
28807 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
28809 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
28810 ("vec_scatter_element<V_HW_4:mode>_DI")
28811 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
28812 ("vec_fpint<mode>", "vflls")
28813 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
28814 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
28815 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
28816 ("*vec_cmphe<mode>_cc"): ... these.
28818 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
28819 mode constant instead of magic value.
28821 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28823 * config/s390/s390.c (s390_expand_vec_compare): Support other
28824 vector floating point modes than just V2DF.
28825 (s390_expand_vcond): Likewise.
28826 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
28827 (s390_cannot_change_mode_class): Prevent mode changes between TF
28828 and V1TF in vector registers.
28829 * config/s390/s390.md (DF, SF): New mode attributes.
28830 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
28831 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
28832 SFmode support for VRs.
28833 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
28835 (VFT, VF_HW): New mode iterators.
28836 (vw, sdx): New mode attributes.
28837 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
28838 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
28839 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
28840 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
28841 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
28842 also the new vector floating point modes. Renaming to ...
28844 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
28845 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
28846 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
28847 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
28848 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
28849 ("vec_unordered<mode>"): ... these.
28851 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
28852 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
28853 ("*vec_extendv2df"): New insn definitions.
28855 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28857 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
28858 ("mulditi3_2", "*muldi3_sign"): New patterns.
28859 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
28860 rename the pattern definition.
28862 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28864 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
28866 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
28868 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28870 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
28871 instruction if possible.
28872 * config/s390/vector.md (vec_halfnumelts): New mode
28874 ("*vec_vllezlf<mode>"): New pattern.
28876 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28878 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
28879 ("popcountv4si2", "popcountv2di2"): Rename to ...
28880 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
28881 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
28883 ("popcount<mode>2_vxe"): New pattern.
28885 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28887 * common/config/s390/s390-common.c (processor_flags_table): Add
28889 * config.gcc: Add arch12.
28890 * config/s390/driver-native.c (s390_host_detect_local_cpu):
28891 Default to arch12 for unknown CPU model numbers.
28892 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
28893 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
28894 PROCESSOR_max sanity check.
28895 * config/s390/s390-opts.h (enum processor_type): Add
28897 * config/s390/s390.c (processor_table): Add arch12.
28898 (s390_expand_builtin): Add check for B_VXE flag.
28899 (s390_issue_rate): Add PROCESSOR_ARCH12.
28900 (s390_get_sched_attrmask): Likewise.
28901 (s390_get_unit_mask): Likewise.
28902 (s390_sched_score): Enable z13 scheduling for arch12.
28903 (s390_sched_reorder): Likewise.
28904 (s390_sched_variable_issue): Likewise.
28905 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
28907 (s390_tune_attr): Use z13 scheduling also for arch12.
28908 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
28909 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
28910 (TARGET_VXE_P): New macros.
28911 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
28912 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
28913 * config/s390/s390.opt: Add arch12 as processor_type.
28915 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28917 * config/s390/s390.md
28918 ("fixuns_truncdddi2", "fixuns_trunctddi2")
28919 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
28920 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
28922 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
28923 Rename expanders to ...
28925 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
28926 ("fixuns_truncdddi2_emu"): ... these.
28928 ("fixuns_trunc<mode>si2_emu"): New expander.
28930 ("*fixuns_truncdfdi2_z13"): Rename to ...
28931 ("*fixuns_truncdfdi2_vx"): ... this.
28933 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28935 * config/s390/2964.md: Remove the single element vector compare
28936 instructions which are no longer used.
28937 * config/s390/s390.c (s390_select_ccmode): Remove handling of
28939 (s390_canonicalize_comparison): Remove handling of DFmode
28941 (s390_expand_vec_compare_scalar): Remove function.
28942 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
28943 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
28945 ("*cmp<mode>_ccs"): Add wfcdb instruction.
28947 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28949 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
28951 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
28952 will anyway by matched by mov<mode>_64dfp.
28954 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28956 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
28957 vlef/vstef. Add missing operand to vleif.
28959 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28961 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
28962 pair for all vector types with 64 bit elements.
28963 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
28964 * config/s390/vector.md (V_HW_64): ... here.
28965 (V_128_NOSINGLE): New mode iterator.
28966 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
28967 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
28968 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
28969 ("*vec_load_pairv2di"): Change to ...
28970 ("*vec_load_pair<mode>"): ... this one.
28972 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28974 * config/s390/constraints.md: Add comments.
28975 (jKK): Reject element sizes > 8 bytes.
28976 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
28978 * config/s390/s390.md: Add the s_operand checks formerly in
28979 s390_split_ok_p to various splitters where they are still
28981 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
28982 for 128 bit vectors. Plus two splitters.
28984 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28986 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
28989 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28992 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
28993 error if the boundary argument is not constant.
28995 2017-03-24 Jakub Jelinek <jakub@redhat.com>
28997 PR rtl-optimization/80112
28998 * loop-doloop.c (doloop_condition_get): Don't check condition
28999 if cmp isn't SET with IF_THEN_ELSE src.
29001 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29003 PR tree-optimization/80158
29004 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
29005 replacing a candidate statement, also replace it for the
29006 candidate's alternate interpretation.
29007 (replace_rhs_if_not_dup): Likewise.
29008 (replace_one_candidate): Likewise.
29010 2017-03-24 Richard Biener <rguenther@suse.de>
29012 PR tree-optimization/80167
29013 * graphite-isl-ast-to-gimple.c
29014 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
29016 (translate_isl_ast_to_gimple::get_rename): Likewise.
29018 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
29020 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
29021 handling of certain combinations of target options, including the
29022 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
29023 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
29025 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29028 * config/arm/arm.md (*load_multiple): Add reload_completed to
29029 matching condition.
29031 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29032 Richard Biener <rguenth@suse.de>
29034 PR tree-optimization/79908
29035 PR tree-optimization/80136
29036 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
29037 been cast away, gimplify_and_add suffices.
29039 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
29041 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
29043 2017-03-23 Richard Biener <rguenther@suse.de>
29045 PR tree-optimization/80032
29046 * gimplify.c (gimple_push_cleanup): Forced unconditional
29047 cleanups still have to go to the conditional_cleanups
29050 2017-03-22 Jakub Jelinek <jakub@redhat.com>
29052 PR tree-optimization/80072
29053 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
29055 (next_operand_entry_id): Change type to unsigned int.
29056 (sort_by_operand_rank): Make sure to return the right return value
29057 even if unsigned fields are bigger than INT_MAX.
29058 (struct oecount): Change cnt and id type to unsigned int.
29059 (oecount_hasher::equal): Formatting fix.
29060 (oecount_cmp): Make sure to return the right return value
29061 even if unsigned fields are bigger than INT_MAX.
29062 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
29065 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
29066 TREE_READONLY on result if writing it more than once.
29069 * doc/invoke.texi (-fsanitize=thread): Document that with
29070 -fnon-call-exceptions atomics are not able to throw
29074 * tsan.c: Include tree-eh.h.
29075 (instrument_builtin_call): Call maybe_clean_eh_stmt or
29076 maybe_clean_or_replace_eh_stmt where needed.
29077 (instrument_memory_accesses): Add cfg_changed argument.
29078 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
29079 if it returned true.
29080 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
29082 PR rtl-optimization/63191
29083 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
29084 wrapper function, moved the whole old content into ...
29085 (ix86_delegitimize_address_1): ... this. New inline function.
29086 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
29087 true as last argument instead of ix86_delegitimize_address.
29089 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
29091 * config/aarch64/aarch64.c (generic_branch_cost): Copy
29092 cortexa57_branch_cost.
29094 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
29096 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
29098 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29101 * doc/md.texi (Constraints): Document wA constraint.
29102 * config/rs6000/constraints.md (wA): New.
29103 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
29104 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
29105 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
29106 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
29108 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
29111 * gimplify.c (is_oacc_declared): New function.
29112 (oacc_default_clause): Use it to set default flags for acc declared
29113 variables inside parallel regions.
29114 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
29115 declared variables.
29116 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
29117 declare attribute to any decl as necessary.
29119 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
29122 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
29123 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
29124 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
29125 (arm_arch_lpae): This.
29126 * config/arm/arm.c (arm_arch7ve): Rename into ...
29127 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
29128 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
29131 2017-03-22 Martin Liska <mliska@suse.cz>
29134 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
29135 error message instead of an ICE.
29137 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29139 * doc/extend.texi (6.11 Additional Floating Types): Revise.
29141 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
29143 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
29145 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
29148 2017-03-21 Martin Sebor <msebor@redhat.com>
29150 * doc/extend.texi: Use "cannot" instead of "can't."
29151 * doc/hostconfig.texi: Same.
29152 * doc/install.texi: Same.
29153 * doc/invoke.texi: Same.
29154 * doc/loop.texi: Same.
29155 * doc/md.texi: Same.
29156 * doc/objc.texi: Same.
29157 * doc/rtl.texi: Same.
29158 * doc/tm.texi: Same.
29159 * doc/tm.texi.in: Same.
29160 * doc/trouble.texi: Same.
29162 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
29165 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
29166 (collect_checksum_attributes): Set it.
29167 (die_checksum_ordered): Use it.
29169 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29171 PR tree-optimization/79908
29172 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
29173 change: For a VA_ARG whose LHS has been cast away, use
29174 force_gimple_operand to construct the side effects.
29176 2017-03-21 David Malcolm <dmalcolm@redhat.com>
29178 PR translation/80001
29179 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
29180 more amenable to translation.
29181 (oacc_loop_auto_partitions): Likewise.
29183 2017-03-21 Marek Polacek <polacek@redhat.com>
29184 Martin Sebor <msebor@redhat.com>
29186 PR tree-optimization/80109
29187 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
29188 on INTEGRAL_TYPE_P.
29190 2017-03-21 Jakub Jelinek <jakub@redhat.com>
29191 Segher Boessenkool <segher@kernel.crashing.org>
29194 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
29195 check reg_used_between_p between insn and one of succ or succ2
29196 depending on if succ is artificial insn not inserted into insn
29199 2017-03-21 Martin Liska <mliska@suse.cz>
29201 PR gcov-profile/80081
29202 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
29203 * doc/gcc.texi: Include gcov-dump stuff.
29204 * doc/gcov-dump.texi: New file.
29206 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
29208 PR rtl-optimization/79150
29209 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
29210 conditional jump, if the jump is the last insn of the loop.
29212 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29213 Richard Biener <rguenth@suse.de>
29215 PR tree-optimization/79908
29216 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
29217 been cast away, use force_gimple_operand to construct the side
29220 2017-03-21 Martin Liska <mliska@suse.cz>
29222 PR libfortran/79956
29223 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
29226 2017-03-21 Brad Spengler <spender@grsecurity.net>
29229 * plugin.c (htab_hash_plugin): New function.
29230 (add_new_plugin): Use it and adjust.
29231 (parse_plugin_arg_opt): Adjust.
29232 (init_one_plugin): Likewise.
29234 2017-03-21 Richard Biener <rguenther@suse.de>
29236 PR tree-optimization/80032
29237 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
29238 if set force the cleanup to happen unconditionally.
29239 (gimplify_target_expr): Push inserted clobbers with force_uncond
29240 to avoid them being removed by control-dependent DCE.
29242 2017-03-21 Richard Biener <rguenther@suse.de>
29244 PR tree-optimization/80122
29245 * tree-inline.c (copy_bb): Do not expans va-arg packs or
29246 va_arg_pack_len when the inlined call stmt requires pack
29248 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
29250 2017-03-21 Jakub Jelinek <jakub@redhat.com>
29253 * tsan.c (instrument_builtin_call): If the memory model argument
29254 is not a constant, assume it is valid.
29257 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
29260 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
29262 PR rtl-optimization/79910
29263 * combine.c (can_combine_p): Do not allow combining an I0 or I1
29264 if its dest is used by an insn before I2 (other than the combined
29265 insns themselves, which are properly handled already).
29267 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
29270 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
29272 * combine.c (record_used_regs): New static function.
29273 (try_combine): Handle situations where there is an additional
29274 instruction between I2 and I3 which needs to have a LOG_LINK
29278 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
29280 * combine.c (try_combine): Delete redundant i1 test. Call
29281 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
29283 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29286 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
29287 alternatives 13/14.
29289 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29291 PR tree-optimization/80054
29292 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
29293 the optimization if a PHI or any of its arguments is not dominated
29294 by the candidate's basis. Use gphi* rather than gimple* as
29296 (replace_profitable_candidates): Clean up a gimple* variable that
29297 should be a gphi* variable.
29299 2017-03-20 Martin Sebor <msebor@redhat.com>
29302 * doc/extend.texi (attribute constructor): Document present limitation.
29304 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
29307 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
29308 __POWER9_VECTOR__ #ifdef control, change template definition to
29309 use Power9-specific built-in function.
29310 (vec_any_eq): Likewise.
29311 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
29312 to control outcomes from this test.
29313 (vector_ae_<mode>p): For VEC_F modes, likewise.
29315 2017-03-20 Ian Lance Taylor <iant@google.com>
29317 * config/i386/i386.c (ix86_function_regparm): Save an extra
29318 register for -fsplit-stack with DECL_STATIC_CHAIN.
29320 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
29323 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
29324 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
29326 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
29328 * config/riscv/riscv.c (riscv_print_operand): Use "fence
29330 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
29333 2017-03-20 Marek Polacek <polacek@redhat.com>
29336 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
29338 2017-03-20 Richard Biener <rguenther@suse.de>
29340 PR tree-optimization/80113
29341 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
29342 allocate extra SSA name for PHI def.
29343 (add_close_phis_to_outer_loops): Likewise.
29344 (add_close_phis_to_merge_points): Likewise.
29345 (copy_loop_close_phi_args): Likewise.
29346 (copy_cond_phi_nodes): Likewise.
29348 2017-03-20 Martin Liska <mliska@suse.cz>
29350 PR middle-end/79753
29351 * tree-chkp.c (chkp_build_returned_bound): Do not build
29352 returned bounds for a LHS that's not a BOUNDED_P type.
29354 2017-03-20 Martin Liska <mliska@suse.cz>
29358 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
29359 COMPLEX_CST and VECTOR_CST.
29361 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29364 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
29365 target operand. A new splitter adds the clobber statement in case
29366 the target operand is dead anyway.
29368 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
29370 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
29371 to age-old versions of binutils and glibc.
29373 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
29375 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
29377 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
29379 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
29381 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
29383 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
29384 requirement for binutils 2.13.
29386 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
29388 * combine.c (try_combine): Delete redundant i1 test. Call
29389 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
29391 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
29393 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
29394 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
29396 <riscv64-*-elf>: Re-arrange section
29397 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
29398 <riscv32-*-linux>: Likewise.
29399 <riscv64-*-elf>: Likewise
29400 <riscv64-*-linux>: Likewise.
29402 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
29405 * aarch64.opt(verbose-cost-dump): Fix typo.
29407 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
29410 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
29411 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
29413 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
29415 * reload.c (find_reloads): When reloading a nonoffsettable address,
29416 use RELOAD_OTHER for it and its address reloads.
29418 PR rtl-optimization/79910
29419 * combine.c (record_used_regs): New static function.
29420 (try_combine): Handle situations where there is an additional
29421 instruction between I2 and I3 which needs to have a LOG_LINK
29424 2017-03-17 Jeff Law <law@redhat.com>
29426 PR tree-optimization/71437
29427 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
29428 conditional in the hash table first.
29429 (vrp_dom_walker::before_dom_children): Extract condition from
29430 ASSERT_EXPR. Record condition, its inverion and any implied
29431 conditions as well.
29433 2017-03-17 Marek Polacek <polacek@redhat.com>
29434 Markus Trippelsdorf <markus@trippelsdorf.de>
29436 PR tree-optimization/80079
29437 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
29440 2017-03-17 Richard Biener <rguenther@suse.de>
29442 PR middle-end/80075
29443 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
29444 Properly verify the LHS before the RHS possibly claims to be
29446 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
29449 2017-03-17 Martin Jambor <mjambor@suse.cz>
29451 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
29452 (List of -O2 options): Likewise.
29453 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
29456 2017-03-17 Tom de Vries <tom@codesourcery.com>
29458 * gcov-dump.c (print_usage): Print bug_report_url.
29460 2017-03-17 Richard Biener <rguenther@suse.de>
29462 PR middle-end/80050
29463 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
29464 (parser::peek): Likewise.
29466 2017-03-17 Richard Biener <rguenther@suse.de>
29468 PR tree-optimization/80048
29469 * sese.c (free_sese_info): Properly release rename_map and
29470 copied_bb_map elements.
29472 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
29474 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
29475 Add linked-list forward and backlinks. Insert on
29476 construction, remove on destruction.
29477 (class pass_store_merging): Add m_stores_head field.
29478 (pass_store_merging::terminate_and_process_all_chains):
29479 Iterate over m_stores_head list.
29480 (pass_store_merging::terminate_all_aliasing_chains):
29482 (pass_store_merging::execute): Check for debug stmts first.
29483 Push new chains onto the m_stores_head stack.
29485 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
29488 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
29489 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
29490 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
29492 2017-03-16 Jeff Law <law@redhat.com>
29494 PR tree-optimization/71437
29495 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
29496 member function. Implementation moved into after_dom_children
29497 member function and into the threader's thread_outgoing_edges
29499 (dom_opt_dom_walker::after_dom_children): Simplify by moving
29500 some code into new thread_outgoing_edges.
29501 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
29502 definition. Simplify marker handling (do it here). Assume we always
29503 have the available expression and the const/copies tables.
29504 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
29506 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
29507 * tree-vrp.c (equiv_stack): No longer file scoped.
29508 (vrp_dom_walker): New class.
29509 (vrp_dom_walker::before_dom_children): New member function.
29510 (vrp_dom_walker::after_dom_children): Likewise.
29511 (identify_jump_threads): Setup domwalker. Use it rather than
29512 walking edges in a random order by hand. Simplify setup/finalization.
29513 (finalize_jump_threads): Remove.
29514 (vrp_finalize): Do not call identify_jump_threads here.
29515 (execute_vrp): Do it here instead and call thread_through_all_blocks
29518 PR tree-optimization/71437
29519 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
29521 (simplify_stmt_for_jump_threading): Add basic_block argument. All
29523 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
29524 (dom_opt_dom_walker::thread_across_edge): Remove
29525 handle_dominating_asserts argument. All callers changed.
29526 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
29527 changes. Remove calls to lhs_of_dominating_assert. Other
29528 uses of handle_dominating_asserts turn into unconditional code
29529 (simplify_control_stmt_condition_1): Likewise.
29530 (simplify_control_stmt_condition): Likewise.
29531 (thread_through_normal_block, thread_across_edge): Likewise.
29532 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
29533 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
29534 object if it is not an SSA_NAME.
29535 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
29536 before calling into the VRP specific simplifiers.
29537 (identify_jump_threads): Remove handle_dominating_asserts
29540 2017-03-16 Jakub Jelinek <jakub@redhat.com>
29543 * tree-diagnostic.c (default_tree_printer): No longer static.
29544 * tree-diagnostic.h (default_tree_printer): New prototype.
29546 2017-03-16 Tamar Christina <tamar.christina@arm.com>
29548 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
29549 Change ins into fmov.
29551 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29553 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
29554 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
29555 Use h_con constraint for operand 1.
29556 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
29557 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
29559 2017-03-15 Jeff Law <law@redhat.com>
29561 PR tree-optimization/71437
29562 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
29563 (record_temporary_equivalences): Use it.
29565 PR tree-optimization/71437
29566 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
29567 tree-ssa-scopedtables.
29568 (lookup_avail_expr, build_and_record_new_cond): Likewise.
29569 (record_conditions, record_cond, vuse_eq): Likewise.
29570 (record_edge_info): Adjust to API tweak of record_conditions.
29571 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
29572 (record_temporary_equivalences, optimize_stmt): Likewise.
29573 (eliminate_redundant_computations): Likewise.
29574 (record_equivalences_from_stmt): Likewise.
29575 * tree-ssa-scopedtables.c: Include options.h and params.h.
29576 (vuse_eq): New function, moved from tree-ssa-dom.c
29577 (build_and_record_new_cond): Likewise.
29578 (record_conditions): Likewise. Accept vector of conditions rather
29579 than edge_equivalence structure for first argument.
29580 for the first argument.
29581 (avail_exprs_stack::lookup_avail_expr): New member function, moved
29582 from tree-ssa-dom.c.
29583 (avail_exprs_stack::record_cond): Likewise.
29584 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
29585 from tree-ssa-dom.c.
29586 (avail_exprs_stack): Add new member functions lookup_avail_expr
29588 (record_conditions): Declare.
29590 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
29593 * lra-constraints.c (process_alt_operands): Increase reject for
29594 reloading an input/output operand.
29596 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
29599 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
29600 insns to convert from signed/unsigned char/short to IEEE 128-bit
29602 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
29604 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
29607 * config/i386/i386.c (ix86_vector_duplicate_value): Create
29608 subreg of inner mode for values already in registers.
29610 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
29612 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
29613 iteration reg is used after the loop.
29615 2017-03-14 Martin Sebor <msebor@redhat.com>
29617 PR tree-optimization/79800
29618 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
29619 precision in negative-positive range.
29620 (format_floating): Call non-const overload with adjusted precision.
29622 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
29625 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
29628 2017-03-14 Martin Sebor <msebor@redhat.com>
29630 PR middle-end/80020
29631 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
29632 * builtins.def (aligned_alloc): Use it.
29635 * Makefile.in (GTFILES): Add calls.c.
29636 * calls.c: Include "gt-calls.h".
29638 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
29640 PR rtl-optimization/79728
29641 * regs.h (struct target_regs): New field
29642 x_contains_allocatable_regs_of_mode.
29643 (contains_allocatable_regs_of_mode): New macro.
29644 * reginfo.c (init_reg_sets_1): Initialize it, and change
29645 contains_reg_of_mode so it includes global regs as well.
29646 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
29647 rather than contains_regs_of_mode.
29649 2017-03-14 Martin Liska <mliska@suse.cz>
29651 * doc/invoke.texi: Document options that can't be combined with
29652 -fcheck-pointer-bounds.
29654 2017-03-14 Martin Liska <mliska@suse.cz>
29656 PR middle-end/79831
29657 * doc/invoke.texi (-Wchkp): Document the option.
29659 2017-03-14 Martin Liska <mliska@suse.cz>
29661 * Makefile.in: Install gcov-dump.
29663 2017-03-14 Martin Liska <mliska@suse.cz>
29665 * multiple_target.c (expand_target_clones): Bail out for
29666 an invalid attribute.
29668 2017-03-14 Richard Biener <rguenther@suse.de>
29670 * alias.c (struct alias_set_entry): Pack properly.
29671 * cfgloop.h (struct loop): Likewise.
29672 * cse.c (struct set): Likewise.
29673 * ipa-utils.c (struct searchc_env): Likewise.
29674 * loop-invariant.c (struct invariant): Likewise.
29675 * lra-remat.c (struct cand): Likewise.
29676 * recog.c (struct change_t): Likewise.
29677 * rtl.h (struct address_info): Likewise.
29678 * symbol-summary.h (function_summary): Likewise.
29679 * tree-loop-distribution.c (struct partition): Likewise.
29680 * tree-object-size.c (struct object_size_info): Likewise.
29681 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
29682 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
29683 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
29684 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
29685 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
29686 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
29687 (struct _stmt_vec_info): Likewise.
29689 2017-03-14 Martin Liska <mliska@suse.cz>
29692 * multiple_target.c (create_dispatcher_calls): Check that
29693 a target can create a function dispatcher.
29695 2017-03-14 Martin Liska <mliska@suse.cz>
29698 * multiple_target.c (expand_target_clones): Drop local.local
29699 flag for default implementation.
29701 2017-03-14 Richard Biener <rguenther@suse.de>
29703 PR tree-optimization/80030
29704 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
29706 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
29708 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
29709 gcc_fallthrough() instead of __attribute__((fallthrough));
29711 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
29713 * doc/gcc.texi: Remove "up" link to (DIR).
29714 * doc/gccint.texi: Ditto.
29716 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
29718 * doc/install.texi (Specific) <avr>: Remove reference to
29721 2017-03-13 Jeff Law <law@redhat.com>
29723 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
29724 attribute rather than comments.
29726 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
29727 match_scratch operand is highest.
29729 2017-03-13 Martin Liska <mliska@suse.cz>
29731 PR middle-end/78339
29732 * ipa-pure-const.c (warn_function_noreturn): If the declarations
29733 is a CHKP clone, use original declaration.
29735 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29737 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
29738 (arc_conditional_register_usage): Use a different allocation order
29739 when optimizing for size.
29740 * common/config/arc/arc-common.c (arc_option_optimization_table):
29741 Section anchors default on when optimizing for size.
29743 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29745 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
29747 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29749 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
29750 * config/arc/arc.md (cpu_facility): Add cd variant.
29751 (*movqi_insn): Add code density variant.
29752 (*movhi_insn): Likewise.
29753 (*movqi_insn): Likewise.
29754 (*addsi3_mixed): Likewise.
29755 (subsi3_insn): Likewise.
29757 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29759 * config/arc/arc.md (movsi_cond_exec): Update constraint.
29761 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29763 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
29764 expressions with MINUS and UNARY ops.
29766 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29769 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
29771 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
29772 between vec_select and vector argument.
29773 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
29774 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
29775 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
29776 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
29777 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
29778 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
29780 2017-03-13 Richard Biener <rguenther@suse.de>
29783 * params.def (vect-max-peeling-for-alignment): Fix typo.
29785 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29787 * doc/install.texi (Specific) <mips-*-*>: Remove description of
29788 issue that only occurred with binutils below 2.18.
29790 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29792 * doc/install.texi (Specific) <cris-axis-elf>: No longer
29793 refer to binutils 2.11/2.12 minimum.
29795 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29797 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
29798 ftp.kernel.org and simplify binutils requirement.
29800 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
29802 * doc/invoke.texi (Warning Options): Fix spelling of link-time
29804 (Optimize Options): Ditto. Also remove redundancy.
29806 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29808 PR translation/79848
29809 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
29811 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
29812 to G_ to avoid double translation.
29814 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29816 PR translation/79923
29817 * auto-profile.c (get_combined_location): Convert leading
29818 character of diagnostics to lower case and remove trailing period.
29819 (read_profile): Likewise for various diagnostics.
29820 * config/arm/arm.c (arm_option_override): Remove trailing period
29821 from various diagnostics.
29822 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
29823 (msp430_expand_delay_cycles): Likewise.
29825 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29828 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
29829 full command-line argument, rather than just "str".
29830 (aarch64_validate_march): Likewise.
29831 (aarch64_validate_mtune): Likewise.
29833 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
29835 PR rtl-optimization/78911
29836 * lra-assigns.c (must_not_spill_p): New function.
29837 (spill_for): Use it.
29839 2017-03-10 Jakub Jelinek <jakub@redhat.com>
29841 PR tree-optimization/79981
29842 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
29843 ATOMIC_COMPARE_EXCHANGE ifn result.
29844 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
29845 IFN_ATOMIC_COMPARE_EXCHANGE.
29847 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29850 * opts.c (parse_sanitizer_options): Add missing question mark to
29851 "did you mean" message.
29853 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29855 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
29857 (VMULEUH_UNS): Likewise.
29858 (VMULOUB_UNS): Likewise.
29859 (VMULOUH_UNS): Likewise.
29860 * config/rs6000/rs6000.c (builtin_function_type): Remove
29861 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
29863 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29866 * read-rtl-function.c (function_reader::read_rtx_operand): Update
29867 x with result of extra_parsing_for_operand_code_0.
29868 (function_reader::extra_parsing_for_operand_code_0): Convert
29869 return type from void to rtx, returning x. When reading
29870 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
29871 larger size containing struct block_symbol.
29873 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
29875 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
29876 -mfloat128-hardware without -m64.
29878 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
29881 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
29882 entries to the case statement that marks unsigned arguments to
29883 overloaded functions.
29885 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
29887 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
29888 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
29890 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
29893 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
29894 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
29896 2017-03-10 Martin Liska <mliska@suse.cz>
29900 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
29901 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
29904 2017-03-10 Olivier Hainque <hainque@adacore.com>
29906 * tree-switch-conversion (array_value_type): Start by resetting
29907 candidate type to it's main variant.
29909 2017-03-10 Jakub Jelinek <jakub@redhat.com>
29911 PR rtl-optimization/79909
29912 * combine.c (try_combine): Use simplify_replace_rtx on individual
29913 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
29914 of the whole CALL_INSN_FUNCTION_USAGE.
29916 PR tree-optimization/79972
29917 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
29918 get_range_info on SSA_NAMEs. Formatting fixes.
29920 2017-03-10 Richard Biener <rguenther@suse.de>
29921 Jakub Jelinek <jakub@redhat.com>
29923 PR tree-optimization/77975
29924 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
29925 edge to be constant.
29926 (get_val_for): For constant x return it. Formatting fix.
29927 (loop_niter_by_eval): Avoid pointless looping if the next iteration
29928 would use the same bases as the current one.
29930 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29932 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
29933 instead of vec_select for V1TImode.
29934 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
29936 (VSX_LE_128): Add V1TI to this mode iterator.
29937 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
29938 (*vsx_le_perm_store_<mode>): Likewise.
29939 (pre-reload splitter for VSX stores): Likewise.
29940 (post-reload splitter for VSX stores): Likewise.
29941 (*vsx_xxpermdi2_le_<mode>): Likewise.
29942 (*vsx_lxvd2x2_le_<mode>): Likewise.
29943 (*vsx_stxvd2x2_le_<mode>): Likewise.
29945 2017-03-09 Michael Eager <eager@eagercon.com>
29947 Correct failures with --enable-checking=yes,rtl.
29949 * config/microblaze/microblaze.c (microblaze_expand_shift):
29950 Replace GET_CODE test with CONST_INT_P and INTVAL test with
29951 test for const0_rtx.
29952 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
29953 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
29955 2017-03-09 Richard Biener <rguenther@suse.de>
29957 PR tree-optimization/79977
29958 * graphite-scop-detection.c (scop_detection::merge_sese):
29959 Handle the case of extra exits to blocks dominating the entry.
29961 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
29963 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
29966 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
29968 PR rtl-optimization/79949
29969 * lra-constraints.c (process_alt_operands): Check memory when
29970 trying to predict a cycle. Print about the overall increase.
29972 2017-03-09 Richard Biener <rguenther@suse.de>
29974 PR middle-end/79971
29975 * gimple-expr.c (useless_type_conversion_p): Preserve
29976 TYPE_SATURATING for fixed-point types.
29978 2017-03-09 Richard Biener <rguenther@suse.de>
29981 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
29982 alignment of BLKmode params.
29984 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29987 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
29988 (VALL_NO_V2Q): Likewise.
29990 * config/aarch64/aarch64-simd.md
29991 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
29993 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
29994 VALL_NO_V2Q mode iterator.
29995 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
29997 2017-03-09 Martin Liska <mliska@suse.cz>
29999 PR tree-optimization/79631
30000 * tree-chkp-opt.c (chkp_is_constant_addr): Call
30001 tree_int_cst_sign_bit just for INTEGER constants.
30003 2017-03-09 Martin Liska <mliska@suse.cz>
30007 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
30010 2017-03-09 Marek Polacek <polacek@redhat.com>
30013 * tree.c (inchash::add_expr): Handle TREE_VEC.
30015 2017-03-09 Martin Liska <mliska@suse.cz>
30018 (chkp_narrow_size_and_offset): New function.
30019 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
30020 (void chkp_parse_bit_field_ref): New function.
30021 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
30022 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
30024 2017-03-09 Martin Liska <mliska@suse.cz>
30027 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
30028 (chkp_find_bounds_1): Remove gcc_unreachable.
30030 2017-03-09 Jakub Jelinek <jakub@redhat.com>
30033 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
30034 BUILT_IN_SYNC*, determine the access type from the size suffix and
30035 always build a MEM_REF with that type. Handle forgotten
30036 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
30039 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
30040 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
30041 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
30042 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
30043 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
30044 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
30045 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
30046 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
30047 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
30048 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
30049 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
30050 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
30051 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
30052 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
30053 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
30054 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
30055 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
30056 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
30057 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
30058 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
30059 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
30060 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
30061 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
30062 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
30063 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
30064 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
30065 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
30066 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
30067 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
30068 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
30069 definitions outside of __OPTIMIZE__ guarded section.
30072 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
30073 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
30074 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
30075 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
30078 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30080 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
30081 ("vfenez<mode>"): Add missing constraints.
30083 2017-03-08 Martin Sebor <msebor@redhat.com>
30086 * config/nds32/nds32.c (nds32_option_override):
30087 Fix misspelled diagnostic.
30089 2017-03-08 Jakub Jelinek <jakub@redhat.com>
30092 * gimplify.c (gimplify_omp_for): Replace index var in outer
30093 taskloop statement with an artificial variable and add
30094 OMP_CLAUSE_PRIVATE clause for it.
30096 2017-03-08 Richard Biener <rguenther@suse.de>
30098 PR tree-optimization/79955
30099 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
30100 for accesses that are completely outside of the variable.
30102 2017-03-08 Andrew Haley <aph@redhat.com>
30104 PR tree-optimization/79943
30105 * tree-ssa-loop-split.c (compute_new_first_bound): When
30106 calculating the new upper bound, (END-BEG) should be added, not
30109 2017-03-08 Jakub Jelinek <jakub@redhat.com>
30111 * config/avr/avr.md (setmemhi): Make sure match_dup
30112 operand number comes before match_scratch.
30114 2017-03-08 Richard Biener <rguenther@suse.de>
30116 PR tree-optimization/79920
30117 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
30118 with ncopies == 1 to ...
30119 (vect_transform_slp_perm_load): ... here. Properly compute
30120 all element loads by iterating VF times over the group. Do
30121 not handle ncopies (computed in a broken way) in
30122 vect_create_mask_and_perm.
30124 2017-03-08 Jakub Jelinek <jakub@redhat.com>
30127 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
30128 is a uniform vector, use uniform_vector_p return value instead of
30129 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
30131 2017-03-07 Marek Polacek <polacek@redhat.com>
30133 PR middle-end/79809
30134 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
30135 (alloca_call_type): Likewise.
30137 2017-03-07 Martin Liska <mliska@suse.cz>
30139 * gcov.c (process_args): Put comment to correct location.
30141 2017-03-07 Martin Liska <mliska@suse.cz>
30143 PR middle-end/68270
30144 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
30145 Use array_at_struct_end_p instead of DECL_CHAIN (field).
30146 (chkp_narrow_bounds_for_field): Likewise.
30147 (chkp_parse_array_and_component_ref): Pass one more argument to
30150 2017-03-07 Richard Biener <rguenther@suse.de>
30152 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
30155 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
30157 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
30158 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
30160 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30163 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
30164 to end of description.
30165 (PARAM_MAX_STORES_TO_MERGE): Likewise.
30167 2017-03-07 Jakub Jelinek <jakub@redhat.com>
30169 PR rtl-optimization/79901
30170 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
30172 (*avx512f_<code><mode>3<mask_name>): ... this.
30173 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
30174 iterator instead of VI8_AVX2_AVX512BW.
30176 PR rtl-optimization/79901
30177 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
30178 min/max expander, expand it using expand_vec_cond_expr.
30181 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
30184 2017-03-06 Jakub Jelinek <jakub@redhat.com>
30187 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
30188 to void * for PCH reasons.
30189 * dwarf2out.c (output_loc_operands, output_die): Cast
30190 v.val_vec.array to unsigned char *.
30192 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
30195 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
30198 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
30200 PR rtl-optimization/79571
30201 * lra-constraints.c (process_alt_operands): Calculate static
30202 reject and subtract it from overall when only addresses will be
30205 2017-03-06 Julia Koval <julia.koval@intel.com>
30208 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
30209 incoming stack boundary to 128 for 64-bit targets.
30211 2017-03-06 Richard Biener <rguenther@suse.de>
30213 PR tree-optimization/79894
30214 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
30215 to NULL after folding it.
30217 2017-03-06 Richard Biener <rguenther@suse.de>
30219 PR tree-optimization/79824
30220 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
30221 check disabling peeling for gaps.
30223 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
30225 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
30226 attributes): Document gettimeofday.
30228 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
30230 * config/s390/s390.c (s390_option_override_internal): Set
30231 PARAM_MIN_VECT_LOOP_BOUND
30233 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
30235 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
30236 * config/s390/s390.md: Likewise.
30238 2017-03-06 Jakub Jelinek <jakub@redhat.com>
30241 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
30242 (<avx2_avx512>_perm<mode>): Rename to ...
30243 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
30245 (<avx512>_perm<mode>_mask): Rename to ...
30246 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
30248 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
30249 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
30250 instead of VI8F_256_512.
30251 (avx512f_perm<mode>): New define_expand.
30252 (avx512f_perm<mode>_mask): Likewise.
30253 (avx512f_perm<mode>_1<mask_name>): New define_insn.
30254 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
30256 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30258 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
30259 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
30261 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
30263 2017-03-06 Martin Liska <mliska@suse.cz>
30266 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
30267 when having a SSA NAME w/o VAR_DECL assigned to it.
30269 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30271 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
30272 msa_dpsub_<su>_d): Fix MODE for vec_select.
30274 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30276 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
30278 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
30280 2017-03-06 Richard Biener <rguenther@suse.de>
30282 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
30283 * plugin.c (register_plugin_info): Likewise.
30284 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
30286 2017-03-05 Jakub Jelinek <jakub@redhat.com>
30288 * config/i386/sse.md (sse_storehps, sse_storelps,
30289 avx_<castmode><avxsizesuffix>_<castmode>,
30290 avx512f_<castmode><avxsizesuffix>_<castmode>,
30291 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
30292 in condition that at least one operand is not a MEM.
30294 2017-03-03 Jakub Jelinek <jakub@redhat.com>
30296 PR middle-end/79805
30297 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
30298 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
30300 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
30301 gimple_call_nothrow_p flag based on whether original builtin can throw.
30302 If it can, emit following stmts on the fallthrough edge.
30303 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
30304 don't create new bb if inserting just debug stmts on the edge, try to
30305 insert them on the fallthru bb or just reset debug stmts.
30307 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
30310 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
30311 restore recog_data (including the operand rtxes inside it) around
30312 the call to get_insn_template.
30314 2017-03-03 Martin Sebor <msebor@redhat.com>
30316 PR tree-optimization/79699
30317 * context.c (context::~context): Free MPFR caches to avoid
30318 a memory leak on program exit.
30320 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30322 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
30323 Use wide_int::ulow () instead of .elt (0).
30325 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
30327 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
30328 (*pushxf): Limit oF constraint to 32bit targets and add oC
30329 constraint for 64bit targets.
30330 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
30331 (*pushdf): Change rmF constraint to rmC.
30333 2017-03-03 Martin Liska <mliska@suse.cz>
30335 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
30336 Remove unused variable.
30338 2017-03-03 Jakub Jelinek <jakub@redhat.com>
30341 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
30342 is a memory operand, increase num_memory.
30343 (ix86_expand_args_builtin): Likewise.
30345 2017-03-03 Jan Hubicka <jh@suse.cz>
30348 * ipa-devirt.c (maybe_record_node): Properly handle
30349 __cxa_pure_virtual visibility.
30351 2017-03-03 Martin Liska <mliska@suse.cz>
30353 PR tree-optimization/79803
30354 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
30356 (pass_loop_prefetch::execute): Disabled optimization if an
30357 assumption about L1 cache size is not met.
30359 2017-03-03 Martin Liska <mliska@suse.cz>
30361 PR rtl-optimization/79574
30362 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
30363 (hash_scan_set): Likewise.
30364 (dump_hash_table): Likewise.
30365 (hoist_code): Likewise.
30367 2017-03-03 Richard Biener <rguenther@suse.de>
30369 * fixed-value.c (fixed_from_string): Restore use of elt (1)
30370 in place of uhigh ().
30371 (fixed_convert_from_real): Likewise.
30373 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
30376 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
30378 2017-03-03 Richard Biener <rguenther@suse.de>
30380 PR middle-end/79818
30381 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
30382 TYPE_OVERFLOW_UNDEFINED check.
30384 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30386 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
30388 (vector_ae_<mode>_p): Likewise.
30389 (vector_nez_<mode>_p): Likewise.
30390 (vector_ne_v2di_p): Likewise.
30391 (vector_ae_v2di_p): Likewise.
30392 (vector_ne_<mode>_p): Likewise.
30393 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
30395 (vsx_tsqrt<mode>2_fe): Likewise.
30397 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
30400 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
30402 2017-03-02 Jakub Jelinek <jakub@redhat.com>
30404 PR rtl-optimization/79780
30405 * cprop.c (one_cprop_pass): When second and further conditional trap
30406 in a single basic block is turned into an unconditional trap, turn it
30407 into a deleted note to avoid RTL verification failures.
30409 2017-03-02 Richard Biener <rguenther@suse.de>
30411 * fold-const.c (const_binop): Use ulow () instead of elt (0).
30413 2017-03-02 Richard Biener <rguenther@suse.de>
30415 PR tree-optimization/79345
30417 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
30418 param and abort the walk, returning -1 if it is hit.
30419 (walk_aliased_vdefs): Take a limit param and pass it on.
30420 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
30421 defaulting to 0 and return a signed int.
30422 * tree-ssa-uninit.c (struct check_defs_data): New struct.
30423 (check_defs): New helper.
30424 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
30425 about uninitialized memory.
30426 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
30427 bogus uninitialized warning.
30428 (fixed_convert_from_real): Likewise.
30430 2017-03-02 Bin Cheng <bin.cheng@arm.com>
30432 PR tree-optimization/66768
30433 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
30434 iv_use if base object can't be determined.
30436 2017-03-02 Jakub Jelinek <jakub@redhat.com>
30438 PR tree-optimization/79345
30439 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
30440 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
30441 (get_pattern_stats): Initialize it.
30442 * genemit.c (gen_expand): Verify match_scratch numbers come after
30443 match_operand/match_dup numbers.
30444 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
30445 match_scratch numbers.
30446 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
30448 * config/s390/s390.md (trunctdsd2): Likewise.
30450 2017-03-02 Richard Biener <rguenther@suse.de>
30452 * wide-int.h (wide_int_storage::operator=): Implement in terms
30455 2017-03-02 Richard Biener <rguenther@suse.de>
30457 PR tree-optimization/79777
30458 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
30459 the to insert expression to sth existing.
30461 2017-03-01 Martin Sebor <msebor@redhat.com>
30463 PR middle-end/79692
30464 * gimple-ssa-sprintf.c
30465 (directive::known_width_and_precision): New function.
30466 (format_integer): Use it.
30467 (get_mpfr_format_length): Consider the full range of precision
30468 when computing %g output with the # flag. Set the likely byte
30469 count to 3 rather than 1 when precision is indeterminate.
30470 (format_floating): Correct the lower bound of precision.
30472 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30474 * doc/invoke.texi: Document default code model for 64-bit Linux.
30476 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30479 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
30480 udiv rather than div since input pattern is unsigned.
30482 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
30484 * config/i386/i386.c (print_reg): Warn for values of
30485 unsupported size in integer register.
30487 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
30490 * config/rs6000/predicates.md (current_file_function_operand): Do
30491 not allow self calls to be local if the function is replaceable.
30493 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
30496 * config/rs6000/altivec.h (vec_ctz and others): Change the
30497 preprocessor macro that controls conditional compilation from
30498 _ARCH_PWR9 to __POWER9_VECTOR__.
30499 (vec_all_ne): Change parameterization of __altivec_scalar_pred
30500 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
30501 control (instead of _ARCH_PWR9 control) so that template
30502 definition uses power9-specific function.
30503 (vec_any_eq): Likewise.
30504 (vec_all_ne): Change macro definition to use a power9-specific
30505 expansion under #ifdef __POWER9_VECTOR__ control (instead of
30506 _ARCH_PWR9 control).
30507 (vec_any_eq) Likewise.
30508 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
30509 expansion for CMPNEF to remove support for xvcmpnesp instruction.
30510 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
30511 support for xvcmpnedp instruction.
30512 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
30513 macro expansion so that Power9 implementation of vec_all_ne does
30514 not use the AltiVec predicate framework.
30515 (VCMPNEH_P): Likewise.
30516 (VCMPNEW_P): Likewise.
30517 (VCMPNED_P): Likewise.
30518 (VCMPNEFP_P): Likewise.
30519 (VCMPNEDP_P): Likewise.
30520 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
30521 implementation of vec_any_eq to not use AltiVec predicate
30523 (VCMPAEH_P): Likewise.
30524 (VCMPAEW_P): Likewise.
30525 (VCMPAED_P): Likewise.
30526 (VCMPAEFP_P): Likewise.
30527 (VCMPAEDP_P): Likewise.
30528 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
30529 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
30530 not use the AltiVec predicate framework.
30531 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
30532 of vec_any_eq to not use AltiVec predicate framework.
30533 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
30534 support for predefined __POWER9_VECTOR__ macro to indicate that
30535 Power9 instruction selection is enabled.
30536 (altivec_overloaded_builtins): Remove extraneous
30537 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
30538 function argument types RS6000_BTI_bool_V16QI and
30539 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
30540 entry for overloaded function argument types RS6000_BTI_bool_V4SI
30541 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
30542 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
30543 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
30544 Power9 for implementations of vec_cmpne. Change the signature for
30545 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
30546 (representing vec_all_ne) to remove the previously described first
30547 argument of type RS6000_BTI_INTSI, as this was an artifact of
30548 reliance on the AltiVec predicate framework, which is no longer
30549 used in the implementation of these functions. Add
30550 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
30551 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
30552 since, unlike the AltiVec predicate framework implementation, we
30553 do not share function descriptors between vec_alle and vec_anyeq.
30554 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
30555 set of modes that receive special treatment even when
30556 TARGET_P9_VECTOR is true. The special treatment emits code that
30557 does not depend on Power9 instructions.
30558 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
30559 define_expand to not rely on AltiVec predicate framework.
30560 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
30562 (vector_ne_v2di_p): Change this define_expand to not rely on
30563 AltiVec predicate framework.
30564 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
30566 (vector_ne_<mode>_p): Change this define_expand to not rely on
30567 AltiVec predicate framework.
30568 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
30570 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
30571 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
30572 define_insn pattern.
30573 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
30574 define_insn pattern because the xvcmpne<VSs>. instruction is not
30576 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
30577 instruction is not supported.
30579 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30581 * config/nvptx/nvptx.c: Include intl.h.
30583 2017-03-01 Martin Jambor <mjambor@suse.cz>
30586 * ipa-prop.h (ipa_bits): Removed field known.
30587 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
30588 to pointers. Adjusted their comments to warn about their sharing.
30589 (ipcp_transformation_summary): Change bits to a vector of pointers.
30590 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
30591 (ipa_get_ipa_bits_for_value): Declare.
30592 * tree-vrp.h (value_range): Mark as GTY((for_user)).
30593 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
30594 (ipa_bits_hash_table): Likewise.
30595 (ipa_vr_ggc_hash_traits): Likewise.
30596 (ipa_vr_hash_table): Likewise.
30597 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
30598 being pointers and vr_known being removed.
30599 (ipa_set_jf_unknown): Likewise.
30600 (ipa_get_ipa_bits_for_value): New function.
30601 (ipa_set_jfunc_bits): Likewise.
30602 (ipa_get_value_range): New overloaded functions.
30603 (ipa_set_jfunc_vr): Likewise.
30604 (ipa_compute_jump_functions_for_edge): Use the above functions to
30605 construct bits and vr parts of jump functions.
30606 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
30607 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
30609 (ipcp_grow_transformations_if_necessary): Also allocate
30610 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
30612 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
30613 them. Fix too long lines.
30614 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
30615 vr_known being removed.
30616 (ipa_read_jump_function): Use new setter functions to construct bits
30617 and vr parts of jump functions or set them to NULL.
30618 (write_ipcp_transformation_info): Adjust for bits being pointers.
30619 (read_ipcp_transformation_info): Likewise.
30620 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
30622 Include gt-ipa-prop.h.
30623 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
30625 (ipcp_store_bits_results): Likewise.
30626 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
30627 Do not write to existing jump functions but use a temporary instead.
30629 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30632 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
30633 attempt to use its first operand as BIT_FIELD_REF base.
30635 2017-03-01 Richard Biener <rguenther@suse.de>
30637 PR middle-end/79721
30638 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
30639 interpolating formula in wrapping arithmetic.
30640 (chrec_apply): Convert chrec_evaluate return value to wanted type.
30642 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30644 PR tree-optimization/79734
30645 * tree-vect-generic.c (expand_vector_condition): Optimize
30646 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
30647 Handle VEC_COND_EXPR where comparison has different inner width from
30648 type's inner width.
30650 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
30652 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
30653 markup, and similar issues. Remove @opindex entries for things
30654 that aren't options. Add missing -mmpy-option entries.
30656 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30658 PR tree-optimization/79737
30659 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
30660 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
30661 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
30662 instead of byte_size. Formatting fix.
30663 (shift_bytes_in_array_right): Formatting fix.
30665 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
30668 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
30669 condition on optimize for the leaf function test.
30671 2017-02-28 Martin Liska <mliska@suse.cz>
30674 * read-rtl-function.c (function_reader::handle_unknown_directive):
30675 Bail out when one uses -flto.
30677 2017-02-28 Martin Liska <mliska@suse.cz>
30679 * common.opt: Replace space with tabular for options of <number>
30681 * config/i386/i386.opt: Show <number> value for
30682 -mlarge-data-threshold.
30683 * opts.c (print_filtered_help): Do not display number in hexadecimal
30686 2017-02-28 Martin Liska <mliska@suse.cz>
30688 * common.opt: Fix --help=option -Q for options which are of
30691 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
30693 * config/i386/i386.c (print_reg): Error out for values
30694 of 8-bit size in invalid integer register.
30696 2017-02-28 Martin Sebor <msebor@redhat.com>
30698 PR tree-optimization/79691
30699 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
30701 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30704 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
30705 gcc_unreachable with output_operand_lossage.
30707 2017-02-28 Richard Biener <rguenther@suse.de>
30709 PR tree-optimization/79740
30710 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
30712 (visit_nary_op): Insert the nary into the hashtable if we
30713 pattern-matched sth.
30714 * tree-ssa-pre.c (eliminate_insert): Robustify.
30716 2017-02-28 Richard Biener <rguenther@suse.de>
30718 PR middle-end/79731
30719 * fold-const.c (decode_field_reference): Reject out-of-bound
30722 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30724 * config/i386/i386.c: Include intl.h.
30725 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
30726 instead of just cond ? "..." : "...".
30727 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
30728 * coverage.c (read_counts_file): Likewise.
30729 * omp-offload.c: Include intl.h.
30730 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
30731 of just cond ? "..." : "...".
30732 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
30733 of just cond ? "..." : "...".
30735 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
30738 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
30740 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
30741 'tune for' CPU name.
30742 * config/arm/arm-cpu-data.h: Regenerated.
30744 2017-02-28 Richard Biener <rguenther@suse.de>
30746 PR tree-optimization/79732
30747 * tree-inline.c (expand_call_inline): Do not shadow var.
30749 2017-02-28 Richard Biener <rguenther@suse.de>
30751 PR tree-optimization/79723
30752 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
30753 address-space properly.
30755 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
30757 * doc/optinfo.texi (Optimization groups): Fix option used for
30759 * doc/invoke.texi (-fopt-info): Document "omp".
30760 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
30761 (OPTGROUP_ALL): Add OPTGROUP_OMP.
30762 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
30763 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
30764 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
30766 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
30768 * dumpfile.c (optgroup_options): Instead of "openmp", associate
30769 OPTGROUP_OMP with "omp".
30771 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
30774 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
30775 for arithmetic shift of unsigned V2DI.
30777 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
30779 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
30780 arc/linux.h headers.
30781 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
30782 (LINK_SPEC): Likewise.
30783 (ARC_TLS_EXTRA_START_SPEC): Likewise.
30784 (EXTRA_SPECS): Likewise.
30785 (STARTFILE_SPEC): Likewise.
30786 (ENDFILE_SPEC): Likewise.
30787 (LIB_SPEC): Likewise.
30788 (TARGET_SDATA_DEFAULT): Likewise.
30789 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
30790 (MULTILIB_DEFAULTS): Likewise.
30791 (DWARF2_UNWIND_INFO): Likewise.
30792 * config/arc/big.h: New file.
30793 * config/arc/elf.h: Likewise.
30794 * config/arc/linux.h: Likewise.
30795 * config/arc/t-uClibc: Remove.
30797 2017-02-27 Bin Cheng <bin.cheng@arm.com>
30799 PR tree-optimization/77536
30800 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
30801 (tree_transform_and_unroll_loop): Use above function to compute the
30802 estimated niter of unrolled loop and use it when scaling profile.
30803 Also use count info rather than frequency if it's non-zero.
30804 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
30805 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
30806 (vect_transform_loop): Call above function.
30808 2017-02-27 Richard Biener <rguenther@suse.de>
30810 PR tree-optimization/45397
30811 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
30812 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
30813 (visit_nary_op): Add pattern matching for CSEing sign-changed
30814 or truncated operations with wider ones.
30816 2017-02-27 Richard Biener <rguenther@suse.de>
30818 PR tree-optimization/79690
30819 * tree-vect-stmts.c (vectorizable_store): Use vector type
30820 built from the DR with address-space.
30822 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
30824 * doc/invoke.texi (Optimize Options): Refine the description
30825 of asan-use-after-return.
30827 2017-02-25 Alan Modra <amodra@gmail.com>
30829 PR rtl-optimization/79584
30830 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
30831 base, not ad->base_term, the reg within base. Remove assertion
30832 that ad->base == ad->base_term. Replace gen_int_mode using
30833 bogus mode with const0_rtx.
30835 2017-02-25 Jakub Jelinek <jakub@redhat.com>
30837 PR middle-end/79396
30838 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
30839 FMA_EXPR like tcc_binary or tcc_unary.
30841 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
30844 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
30846 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
30847 (output_loc_operands): Handle DW_OP_call_ref and
30848 DW_OP_GNU_variable_value.
30849 (struct variable_value_struct): New type.
30850 (struct variable_value_hasher): Likewise.
30851 (variable_value_hash): New variable.
30852 (string_types): Remove.
30853 (copy_loc_descr): New function.
30854 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
30855 (prepend_loc_descr_to_each): New function.
30856 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
30857 instead of add_loc_descr_to_each if the first argument is single
30858 location list and the second has multiple.
30859 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
30860 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
30861 when looking for variable value which doesn't have other location info.
30862 (loc_list_from_tree): Formatting fix.
30863 (gen_array_type_die): Simplify DW_AT_string_length handling.
30864 (adjust_string_types): Remove.
30865 (gen_subprogram_die): Don't call adjust_string_types nor test/set
30866 string_types. Call resolve_variable_values.
30867 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
30868 (resolve_addr_in_expr): Likewise. Add A argument.
30869 (copy_deref_exprloc): Remove deref argument. Adjust for the
30870 original expression being DW_OP_GNU_variable_value with optionally
30871 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
30872 optionally after it.
30873 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
30874 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
30875 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
30876 (variable_value_hasher::hash, variable_value_hasher::equal): New
30878 (resolve_variable_value_in_expr, resolve_variable_value,
30879 resolve_variable_values, note_variable_value_in_expr,
30880 note_variable_value): New functions.
30881 (dwarf2out_early_finish): Call note_variable_value on all toplevel
30884 2017-02-24 Jakub Jelinek <jakub@redhat.com>
30887 * opts.h (handle_generated_option): Add GENERATED_P argument.
30888 * opts-common.c (handle_option): Adjust function comment.
30889 (handle_generated_option): Add GENERATED_P argument, pass it to
30891 (control_warning_option): Pass false to handle_generated_option
30893 * opts.c (maybe_default_option): Pass true to handle_generated_option
30895 * optc-gen.awk: Likewise.
30897 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30899 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
30900 a REG, look at the REG it is a SUBREG of.
30901 (splitter for cmpeqsi_t): Ditto.
30903 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30905 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
30906 the special USEs with the pattern of the insn, not the insn itself.
30908 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
30911 * doc/invoke.texi: Document -mload-store-pairs.
30913 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30914 Sandra Loosemore <sandra@codesourcery.com>
30916 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
30917 argument isn't a CONST_INT.
30918 (nios2_alternate_compare_const): Assert op is a CONST_INT.
30919 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
30920 (nios2_validate_compare): Bypass alternate compare logic if *op2
30921 is not a CONST_INT.
30922 (ldstwm_operation_p): Return false if first_base is not a REG or
30923 if first_offset is not a CONST_INT.
30925 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30927 * config/cris/cris.md: Use correct operand in a define_peephole2.
30929 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30931 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
30933 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30935 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
30936 this_insn if it is an INSN or JUMP_INSN.
30937 (force_offsettable): Look at base, not at addr.
30938 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
30939 on things that aren't necessarily CONST_INTs.
30941 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
30943 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
30944 -mfpmath=sse is the default also for x86-32 targets with SSE2
30945 instruction set when @option{-ffast-math} is enabled
30947 2017-02-24 Jeff Law <law@redhat.com>
30949 PR rtl-optimizatoin/79286
30950 * ira.c (update_equiv_regs): Drop may_trap_p exception to
30953 2017-02-24 Richard Biener <rguenther@suse.de>
30955 PR tree-optimization/79389
30956 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
30959 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
30961 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
30962 function comment to reflect reality.
30963 (loop_exits_before_overflow): Fix typo in function description.
30965 2017-02-24 Richard Biener <rguenther@suse.de>
30967 PR tree-optimization/79389
30968 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
30969 properly that a threading opportunity exists. Detect conditional
30970 copy/constant propagation opportunities.
30972 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
30974 * config/visium/visium.md (type): Add trap.
30975 (b): New mode attribute.
30976 (*btst): Rename into...
30977 (*btst<mode>): ...this and adjust.
30978 (*cbranchsi4_btst_insn): Rename into...
30979 (*cbranch<mode>4_btst_insn): ...this and adjust.
30980 (trap): New define_insn.
30982 2017-02-23 Jakub Jelinek <jakub@redhat.com>
30984 PR tree-optimization/79389
30985 * ifcvt.c (struct noce_if_info): Add rev_cond field.
30986 (noce_reversed_cond_code): New function.
30987 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
30988 reversed_comparison_code. Formatting fix.
30989 (noce_try_store_flag): Test rev_cond != NULL in addition to
30990 reversed_comparison_code.
30991 (noce_try_store_flag_constants): Likewise.
30992 (noce_try_store_flag_mask): Likewise.
30993 (noce_try_addcc): Use rev_cond if non-NULL instead of
30994 reversed_comparison_code.
30995 (noce_try_cmove_arith): Likewise. Formatting fixes.
30996 (noce_try_minmax, noce_try_abs): Clear rev_cond.
30997 (noce_find_if_block): Initialize rev_cond.
30998 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
30999 instead of false as last argument never attempt to reverse it
31002 2017-02-23 Bin Cheng <bin.cheng@arm.com>
31004 PR tree-optimization/79663
31005 * tree-predcom.c (combine_chains): Process refs in reverse order
31006 only for ZERO length chains, and add explaining comment.
31008 2017-02-23 Jeff Law <law@redhat.com>
31010 PR tree-optimization/79578
31011 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
31012 in call to operand_equal_p.
31014 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
31017 * config/i386/cpuid.h: Fix another undefined behavior.
31019 2017-02-23 Richard Biener <rguenther@suse.de>
31021 PR tree-optimization/79683
31022 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
31023 vector types for data-refs.
31025 2017-02-23 Martin Liska <mliska@suse.cz>
31027 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
31029 2017-02-23 Jakub Jelinek <jakub@redhat.com>
31031 PR middle-end/79665
31032 * internal-fn.c (get_range_pos_neg): Moved to ...
31033 * tree.c (get_range_pos_neg): ... here. No longer static.
31034 * tree.h (get_range_pos_neg): New prototype.
31035 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
31036 are known to be in between 0 and signed maximum inclusive, try to
31037 expand both unsigned and signed divmod and use the cheaper one from
31040 2017-02-22 Jeff Law <law@redhat.com>
31042 PR tree-optimization/79578
31043 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
31044 to compare base operands.
31046 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
31049 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
31050 gpc_reg_operand instead of fpr_reg_operand.
31052 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
31054 * config/mips/mips.c (mips_return_in_memory): Force FP
31055 vector types to be returned in memory for o32 ABI.
31057 2017-02-22 Jakub Jelinek <jakub@redhat.com>
31059 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
31060 instead of DW_TAG_member for static data member declarations and don't
31061 set no_linkage_name for static inline data members.
31062 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
31065 2017-02-22 Martin Liska <mliska@suse.cz>
31067 * doc/invoke.texi: Replace inequality signs with square brackets
31070 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
31073 * lra-constraints.c (simplify_operand_subreg): Handle
31074 WORD_REGISTER_OPERATIONS targets.
31076 2017-02-22 Jakub Jelinek <jakub@redhat.com>
31079 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
31080 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
31081 elimination by swapping fld*.
31083 2017-02-22 Richard Biener <rguenther@suse.de>
31085 PR tree-optimization/79673
31086 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
31087 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
31088 irrelevant address-space qualifiers and avoiding a
31089 ADDR_SPACE_CONVERT_EXPR from fold_convert.
31091 2017-02-22 Richard Biener <rguenther@suse.de>
31093 PR tree-optimization/79666
31094 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
31095 to not symbolically negate if that may introduce undefined
31098 2017-02-22 Martin Liska <mliska@suse.cz>
31101 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
31102 * data-streamer-out.c (streamer_write_gcov_count_stream):
31104 * value-prof.c (stream_out_histogram_value): Make assert more
31105 precise based on type of counter.
31107 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
31110 * config/i386/i386.md (standard_x87sse_constant_load splitter):
31111 Use nonimmediate_operand instead of memory_operand for operand 1.
31112 (float-extend standard_x87sse_constant_load splitter): Ditto.
31114 2017-02-21 Jeff Law <law@redhat.com>
31116 PR tree-optimization/79621
31117 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
31118 blocks with edges to themselves.
31120 2017-02-21 Jakub Jelinek <jakub@redhat.com>
31123 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
31124 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
31125 Use gimple_call_builtin_p.
31128 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
31129 on temporarily removed DEBUG_INSNs.
31131 PR tree-optimization/79649
31132 * tree-loop-distribution.c (classify_partition): Give up on
31133 non-generic address space loads/stores.
31135 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
31137 * doc/loop.texi (Loop manipulation): Remove nonexistent
31138 tree_ssa_loop_version from the documentation.
31139 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
31141 2017-02-21 Jakub Jelinek <jakub@redhat.com>
31144 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
31145 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
31146 * config/rs6000/rs6000.c: Include except.h.
31147 (rs6000_expand_split_stack_prologue): Call
31148 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
31150 2017-02-21 Martin Jambor <mjambor@suse.cz>
31153 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
31154 have been analyzed.
31156 2017-02-21 Martin Jambor <mjambor@suse.cz>
31158 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
31159 for backward compatibility only.
31160 * doc/invoke.texi (Option Summary): Remove all references to
31161 -fipa-cp-alignment.
31163 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
31167 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31169 * lra-constraints.c (curr_insn_transform): Handle
31170 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
31172 2017-02-21 Martin Liska <mliska@suse.cz>
31174 * config/i386/i386.opt: Replace -masm-dialect with -masm.
31176 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
31178 PR translation/79638
31179 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
31181 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
31184 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
31185 (arm_function_ok_for_sibcall): Return false for an indirect call by
31186 descriptor if all the argument registers are used.
31187 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
31188 alignment of the function.
31190 2017-02-21 Jakub Jelinek <jakub@redhat.com>
31192 PR tree-optimization/61441
31193 * simplify-rtx.c (simplify_const_unary_operation): For
31194 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
31195 the sNaN unmodified.
31197 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
31199 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
31200 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
31201 instead of SYSTEM_HEADER_DIR.
31203 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
31204 Martin Liška <mliska@suse.cz>
31206 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
31207 Fix typos and grammar, use active voice, and clarify.
31209 2017-02-20 Marek Polacek <polacek@redhat.com>
31211 PR middle-end/79537
31212 * gimplify.c (gimplify_expr): Handle unused *&&L;.
31215 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
31217 2017-02-20 Jakub Jelinek <jakub@redhat.com>
31220 * config/i386/i386.c (ix86_expand_builtin): Handle
31221 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
31222 ix86_builtins_isa[fcode].isa as a requirement of those
31223 flags and any other flag in the bitmask.
31224 (ix86_init_mmx_sse_builtins): Use 0 instead of
31225 ~OPTION_MASK_ISA_64BIT as mask.
31226 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
31227 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
31228 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
31229 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
31231 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31234 * lra-constraints.c (split_reg): Check requested split mode
31235 is supported by the register.
31237 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31239 * lra-constraints.c (simplify_operand_subreg): Remove early
31242 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31245 * lra-constraints.c (curr_insn_transform): Tighten condition
31246 for converting SUBREG reloads from OP_OUT to OP_INOUT.
31248 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31251 * lra-constraints.c (curr_insn_transform): Handle
31252 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
31254 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
31257 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
31259 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
31261 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
31264 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
31267 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31269 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
31272 2017-02-18 Jakub Jelinek <jakub@redhat.com>
31274 * final.c (last_columnnum, override_columnnum): New variables.
31275 (final_start_function): Set last_columnnum, pass it to begin_prologue
31276 hook and pass 0 to dwarf2out_begin_prologue.
31277 (final_scan_insn): Update override_columnnum. Pass last_columnnum
31278 to source_line debug hook.
31279 (notice_source_line): Compute last_columnnum and for debug_column_info
31280 return true on column changes.
31281 * debug.h (struct gcc_debug_hooks): Add column argument to
31282 source_line and begin_prologue hooks.
31283 (debug_nothing_int_charstar_int_bool): Remove prototype.
31284 (debug_nothing_int_int_charstar,
31285 debug_nothing_int_int_charstar_int_bool): New prototypes.
31286 (dwarf2out_begin_prologue): Add column argument.
31287 * debug.c (do_nothing_debug_hooks): Adjust source_line and
31288 begin_prologue hooks.
31289 (debug_nothing_int_charstar_int_bool): Remove.
31290 (debug_nothing_int_int_charstar,
31291 debug_nothing_int_int_charstar_int_bool): New functions.
31292 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
31293 through to dwarf2out_source_line.
31294 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
31295 (dwarf2out_source_line): Add column argument, emit it if requested.
31296 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
31298 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
31299 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
31300 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
31301 through to dwarf2out_begin_prologue.
31302 (vmsdbgout_source_line): Add column argument, pass it through to
31303 dwarf2out_source_line.
31304 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
31305 dbxout_source_line caller.
31306 (dbxout_source_line): Add column argument.
31308 * common.opt (gno-column-info, gcolumn-info): New options.
31309 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
31310 (check_die): Also test for multiple DW_AT_decl_column attributes.
31311 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
31312 DW_AT_decl_column if requested.
31313 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
31315 (gen_variable_die): Likewise.
31316 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
31317 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
31320 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
31321 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
31322 (ix86_handle_option): Handle OPT_m3dnowa.
31323 * doc/invoke.texi (-m3dnowa): Document.
31324 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
31325 -m3dnowa instead of -m3dnow -march=athlon.
31328 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
31329 instead of gcc_assert for K, r and R code checks. Formatting fixes.
31331 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31334 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
31335 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
31336 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
31337 generator for vsx_xxpermdi_<mode>_be.
31338 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
31339 force big-endian semantics.
31340 (vsx_xxpermdi_<mode>_be): New define_expand with same
31341 implementation as previous version of vsx_xxpermdi_<mode>.
31343 2017-02-17 Jakub Jelinek <jakub@redhat.com>
31345 PR tree-optimization/79327
31346 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
31347 variable, its initialization and use.
31349 2017-02-17 Julia Koval <julia.koval@intel.com>
31351 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
31352 (OPTION_MASK_ISA_PKU_UNSET): New.
31353 (ix86_handle_option): Handle -mrdpid.
31354 * config/i386/cpuid.h (bit_RDPID): New.
31355 * config/i386/driver-i386.c (host_detect_local_cpu):
31356 Detect RDPID feature.
31357 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
31358 * config/i386/i386-c.c (ix86_target_macros_internal):
31360 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
31361 (ix86_valid_target_attribute_inner_p): Add "rdpid".
31362 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
31363 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
31364 * config/i386/i386.md (define_insn "rdpid"): New.
31365 * config/i386/i386.opt Add -mrdpid.
31366 * config/i386/immintrin.h (_rdpid_u32): New.
31368 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
31370 PR rtl-optimization/79541
31371 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
31372 instead of transforming it into USE.
31374 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
31376 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
31377 If HONOR_SNANS (SFmode) force the input to a register.
31378 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
31379 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
31380 an frsp or similar insn.
31382 2017-02-17 Martin Liska <mliska@suse.cz>
31384 PR rtl-optimization/79577
31385 * params.def (selsched-max-sched-times): Increase minimum to 1.
31387 2017-02-17 Martin Liska <mliska@suse.cz>
31389 PR rtl-optimization/79574
31390 * gcse.c (want_to_gcse_p): Prevent integer overflow.
31392 2017-02-17 Martin Liska <mliska@suse.cz>
31394 PR tree-optimization/79529
31395 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
31396 ssa_defined_default_def_p to handle cases which are implicitly
31398 * tree-ssa.c (ssa_defined_default_def_p): New function.
31399 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
31400 which are implicitly defined.
31401 * tree-ssa.h (ssa_defined_default_def_p): Declare.
31403 2017-02-17 Richard Biener <rguenther@suse.de>
31405 PR middle-end/79576
31406 * params.def (max-ssa-name-query-depth): Limit to 10.
31408 2017-02-17 Richard Biener <rguenther@suse.de>
31410 PR tree-optimization/79552
31411 * tree-ssa-structalias.c (visit_loadstore): Properly verify
31414 2017-02-17 Richard Biener <rguenther@suse.de>
31417 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
31419 2017-02-17 Marek Polacek <polacek@redhat.com>
31421 PR middle-end/79536
31422 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
31423 (fold_negate_expr): New wrapper.
31425 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
31427 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
31428 Correct terminology and de-emphasize pre-standard behavior.
31430 2017-02-16 Alan Modra <amodra@gmail.com>
31432 PR rtl-optimization/79286
31433 * ira.c (def_dominates_uses): New function.
31434 (update_equiv_regs): Don't create an equivalence for insns that
31435 may trap where the register def does not dominate the use.
31437 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
31439 PR rtl-optimization/78127
31440 * lra.c (lra): Call lra_eliminate before finish the loop after
31443 2017-02-16 Richard Biener <rguenther@suse.de>
31445 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
31447 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
31448 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
31449 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
31450 (isl_val_int_from_wi): New function.
31451 (extract_affine_gmp): Rename to ...
31452 (extract_affine_wi): ... this, take a widest_int.
31453 (extract_affine_int): Just wrap extract_affine_wi.
31454 (add_param_constraints): Use isl_val_int_from_wi.
31455 (add_loop_constraints): Likewise, and extract_affine_wi.
31457 2017-02-15 Jeff Law <law@redhat.com>
31459 PR middle-end/79521
31460 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
31461 ira_init_register_move_cost_if_necessary.
31463 2017-02-15 Martin Sebor <msebor@redhat.com>
31465 PR middle-end/32003
31466 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
31467 removed in a prior commit.
31469 2017-02-15 Bin Cheng <bin.cheng@arm.com>
31471 PR tree-optimization/79347
31472 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
31473 counters during peeling.
31475 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
31477 * Makefile.in (site.exp): Remove "set ISLVER".
31479 2017-02-15 Jakub Jelinek <jakub@redhat.com>
31482 * real.c (real_from_integer): Call real_convert even for decimal.
31484 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
31487 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
31489 2017-02-14 Andrew Pinski <apinski@cavium.com>
31491 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
31492 cores and change the partno/implementer to be correct.
31493 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
31494 the 'B" as the implementer.
31495 * config/aarch64/aarch64-tune.md: Regenerate.
31497 2017-02-14 Carl Love <cel@us.ibm.com>
31499 * config/rs6000/rs6000.c: Add case statement entry to make the
31500 xvcvuxdsp built-in argument unsigned.
31501 * config/rs6000/vsx.md: Fix the source and return operand types so they
31502 match the instruction definitions from the ISA document. Fix typo
31503 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
31506 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
31509 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
31510 member early_clobber_alts.
31511 * lra-lives.c (reg_early_clobber_p): New.
31512 (process_bb_lives): Use it.
31513 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
31514 (debug_operand_data): Initialize early_clobber_alts.
31515 (setup_operand_alternative): Set up early_clobber_alts.
31516 (collect_non_operand_hard_regs): Ditto. Pass early clobber
31517 alternatives to new_insn_reg.
31518 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
31520 (lra_update_insn_regno_info): Pass the new arg.
31522 2017-02-14 Jakub Jelinek <jakub@redhat.com>
31524 PR middle-end/79505
31525 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
31526 (new_oacc_loop_raw): Don't clear already cleared fields.
31529 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
31530 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
31531 _mm512_prefetch_i64gather_ps): New inline functions and macros.
31533 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
31536 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
31538 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
31541 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
31542 the extra instruction to the right place to store 128-bit constant
31545 2017-02-14 Martin Sebor <msebor@redhat.com>
31547 PR middle-end/79448
31548 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
31549 warning for strings of unknown length.
31551 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
31553 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
31555 2017-02-14 Jeff Law <law@redhat.com>
31558 * ira-costs.c (scan_one_insn): Initialize register move costs
31559 for pseudos seen in USE/CLOBBER insns.
31561 PR tree-optimization/79095
31562 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
31563 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
31564 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
31565 if the operands are known to be not equal, then the resulting range
31567 (intersect_ranges): If the new range is ~[0,0] and the old range is
31568 wide, then prefer ~[0,0].
31569 * tree-vrp.c (overflow_comparison_p_1): New function.
31570 (overflow_comparison_p): New function.
31571 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
31572 if NAME is used in an overflow test.
31573 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
31574 overflow check that can be expressed as an equality test, then adjust
31575 ops to be that equality test.
31577 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31579 * config/s390/s390-builtin-types.def: Remove flags argument.
31580 * config/s390/s390.c (s390_init_builtins): Likewise.
31582 2017-02-14 Martin Liska <mliska@suse.cz>
31584 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
31585 vector. Fix trailing white spaces.
31587 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
31589 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
31592 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31594 PR rtl-optimization/68664
31595 * config/arm/arm.c (arm_sched_can_speculate_insn):
31596 New function. Declare prototype.
31597 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
31599 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31601 PR rtl-optimization/68664
31602 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
31604 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
31606 2017-02-14 Amit Pawar <amit.pawar@amd.com>
31608 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
31609 max skip bytes for function, loop and jump.
31611 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31613 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
31614 ABS_EXPR for gimple dump.
31616 2017-02-14 Jakub Jelinek <jakub@redhat.com>
31619 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
31621 PR tree-optimization/79408
31622 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
31623 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
31624 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
31625 also if rhs1 is INTEGER_CST.
31627 2017-02-14 Richard Biener <rguenther@suse.de>
31629 PR middle-end/79432
31630 * tree-into-ssa.c (insert_phi_nodes): When the function can
31631 have abnormal edges rewrite SSA names with broken use-def
31632 dominance out of SSA and register them for PHI insertion.
31634 2017-02-13 Martin Sebor <msebor@redhat.com>
31636 PR middle-end/79496
31637 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
31638 clearing info.nowrite flag when snprintf size argument is a range.
31640 2017-02-13 Jakub Jelinek <jakub@redhat.com>
31642 * cprop.c (cprop_jump): Add missing space in string literal.
31643 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
31644 (get_constraint_for_component_ref): Likewise.
31645 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
31646 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
31647 * lra-constraints.c (process_alt_operands): Likewise.
31648 * ipa-inline.c (inline_small_functions): Likewise.
31649 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
31650 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
31651 * trans-mem.c (diagnose_tm_1_op): Likewise.
31652 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
31653 (grid_parallel_clauses_gridifiable): Likewise.
31655 * config/nvptx/mkoffload.c (process): Add space in between
31658 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
31659 "MOD4_SSE_REGS" and "ALL_REGS".
31661 * spellcheck.c (test_data): Add , in between "foo" and "food".
31663 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31666 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
31667 boundary crossing check and subsequent code generation agree.
31669 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31671 * config/aarch64/aarch64.c (has_memory_op): Delete.
31672 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
31675 2017-02-13 Jakub Jelinek <jakub@redhat.com>
31677 PR rtl-optimization/79388
31678 PR rtl-optimization/79450
31679 * combine.c (distribute_notes): When removing TEM_INSN for which
31680 corresponding dest has last value recorded, invalidate that last
31683 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31685 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
31686 of explicit '@'. Add missing assembly comment marker on branch costs
31689 2017-02-13 Nathan Sidwell <nathan@acm.org>
31691 * gengtype-lex.l (<in_struct>): Add '/'.
31693 2017-02-13 Martin Liska <mliska@suse.cz>
31696 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
31698 2017-02-13 Richard Biener <rguenther@suse.de>
31700 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
31702 * configure: Re-generate.
31703 * config.in: Likewise.
31704 * graphite-dependences.c: Simplify as if
31705 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
31706 * graphite-isl-ast-to-gimple.c: Likewise.
31707 * graphite-optimize-isl.c: Likewise.
31708 * graphite-poly.c: Likewise.
31709 * graphite-sese-to-poly.c: Likewise.
31710 * graphite.h: Likewise.
31711 * toplev.c: Include isl/version.h and use isl_version () for
31712 printing the ISL version.
31713 * doc/install.texi: Update ISL requirement.
31715 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
31717 * doc/standards.texi (Standards): Update reference to
31720 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
31722 * doc/extend.texi (Named Address Spaces): sourceware.org now
31724 * doc/install.texi (Binaries): Ditto.
31727 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
31729 * doc/cpp.texi: Replace "stringify"/"stringification" with C
31730 standard terminology "stringize"/"stringizing" throughout.
31731 * doc/cppinternals.texi: Likewise.
31733 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
31735 * doc/extend.texi: Fix some spelling mistakes and typos.
31736 * doc/invoke.texi: Likewise.
31738 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
31741 * params.def (inline-min-speedup) Change from 10 to 8.
31743 2017-02-11 Jakub Jelinek <jakub@redhat.com>
31745 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
31748 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
31751 * ipa-inline-analysis.c (get_minimal_bb): New function.
31752 (record_modified): Use it.
31753 (remap_edge_change_prob): Handle also ancestor functions.
31755 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
31757 * doc/contrib.texi (Contributors): Remove broken link into
31758 the Mauve CVS repository.
31760 2017-02-11 Jakub Jelinek <jakub@redhat.com>
31762 PR middle-end/79454
31763 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
31764 result computation whenever lhs doesn't have vector mode, not
31765 just when it has BLKmode.
31767 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
31769 * doc/makefile.texi (profiledbootstrap): Refer to the
31770 installation instructions only in textual form.
31772 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31775 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
31777 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
31779 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
31780 (Specific): Update mingw-w64 reference.
31782 (Specific): Remove broken link to Renesas RX processor.
31784 2017-02-10 Richard Biener <rguenther@suse.de>
31786 * toplev.c (process_options): Do not mention obsolete graphite
31787 options when printing sorry message about missing graphite support.
31788 Mention -floop-nest-optimize.
31790 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
31792 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
31793 (vtst_p16): Likewise.
31794 (vtstq_p8): Likewise.
31795 (vtstq_p16): Likewise.
31797 (vtstq_p64): Likewise.
31798 * config/arm/arm_neon.h (vgetq_lane_p64): New.
31799 (vset_lane_p64): New.
31800 (vsetq_lane_p64): New.
31802 2017-02-10 Jakub Jelinek <jakub@redhat.com>
31804 PR tree-optimization/79411
31805 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
31806 stmt operands are SSA_NAMEs used in abnormal phis.
31807 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
31810 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
31813 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
31816 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
31818 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
31820 2017-02-09 Jakub Jelinek <jakub@redhat.com>
31822 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
31826 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
31827 not arbitrary TREE_CONSTANT.
31830 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
31831 "omp declare target link" attribute unless is_global_var.
31832 * omp-offload.c (find_link_var_op): Likewise.
31834 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
31835 Chung-Lin Tang <cltang@codesourcery.com>
31837 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
31839 (gimplify_adjust_omp_clauses): Don't delete TILE.
31840 (gimplify_omp_for): Deal with TILE.
31841 * internal-fn.c (expand_GOACC_TILE): New function.
31842 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
31844 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
31845 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
31847 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
31848 avoid DIV for outermost collapse var.
31849 (expand_oacc_for): Insert tile element loop as needed. Adjust.
31850 Remove out of date comments, fix whitespace.
31851 * omp-general.c (omp_extract_for_data): Deal with tiling.
31852 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
31853 adjust OLF_DIM_BASE value.
31854 (struct omp_for_data): Add tiling field.
31855 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
31856 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
31857 for auto loops. Remove default auto determining, moved to
31858 oacc_loop_fixed_partitions.
31859 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
31860 stmts, add e_mask field.
31861 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
31862 (oacc_thread_numbers): Use oacc_dim_call.
31863 (oacc_xform_tile): New.
31864 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
31865 (finish_oacc_loop): Adjust for ifns vector.
31866 (oacc_loop_discover_walk): Append loop abstraction sites to list,
31867 add case for GOACC_TILE fns.
31868 (oacc_loop_xform_loop): Delete.
31869 (oacc_loop_process): Iterate over call list directly, and add
31870 handling for GOACC_TILE fns.
31871 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
31873 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
31874 vector partitioning to outer loops. Assign 2 partitions to loops
31875 when available. Add TILE handling.
31876 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
31877 (execite_oacc_device_lower): Process GOACC_TILE fns,
31878 ignore unknown specs.
31879 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
31880 * tree.c (omp_clause_num_ops): Adjust TILE ops.
31881 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
31883 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
31885 * configure.ac (ACX_BUGURL): Update.
31886 * configure: Regenerate.
31888 2017-02-09 Richard Biener <rguenther@suse.de>
31890 PR tree-optimization/69823
31891 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
31892 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
31894 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
31896 * config/arc/arc-c.def: Add __NPS400__ definition.
31897 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
31898 (TARGET_NPS400): Define.
31900 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
31902 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
31904 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
31905 pointer, arch_info.
31906 (arc_cpu_types): Fill the arch_info field with a pointer into the
31907 arc_arch_types table.
31908 (arc_selected_cpu): Declare.
31909 * config/arc/arc.c (arc_selected_cpu): Make global.
31910 (arc_selected_arch): Delete.
31911 (arc_base_cpu): Delete.
31912 (arc_override_options): Remove references to deleted variables,
31913 update access to arch information.
31914 (ARC_OPT): Update access to arch information.
31915 (ARC_OPTX): Likewise.
31916 * config/arc/arc.h (arc_base_cpu): Remove declaration.
31917 (TARGET_ARC600): Update access to arch information.
31918 (TARGET_ARC601): Likewise.
31919 (TARGET_ARC700): Likewise.
31920 (TARGET_EM): Likewise.
31921 (TARGET_HS): Likewise.
31922 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
31925 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
31928 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
31929 condition/operands for integer GE/LE/GEU/LEU operations.
31931 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
31933 PR translation/79397
31934 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
31937 2017-02-08 Martin Jambor <mjambor@suse.cz>
31940 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
31941 whether allocation happened.
31942 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
31943 nothing was allocated.
31945 2017-02-08 Jakub Jelinek <jakub@redhat.com>
31947 PR tree-optimization/79408
31948 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
31949 constant, but SSA_NAME with a known integer range, use the minimum
31950 of that range instead of op1 to determine if modulo can be replaced
31951 with its first operand.
31953 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31955 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
31957 2017-02-08 Richard Biener <rguenther@suse.de>
31959 PR tree-optimization/71824
31960 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
31961 Check all loops contained in the merged region.
31963 2017-02-07 Andrew Pinski <apinski@cavium.com>
31965 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
31967 2017-02-07 Andrew Pinski <apinski@cavium.com>
31969 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
31970 (thunderxt88): Likewise.
31971 (thunderxt81): Disable LSE and change v8.1 to v8.
31972 (thunderxt83): Likewise.
31974 2017-02-07 Jakub Jelinek <jakub@redhat.com>
31975 Richard Biener <rguenther@suse.de>
31977 PR middle-end/79399
31978 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
31979 type from int to size_t.
31980 * ira-costs.c (struct_costs_size): Change type from int to size_t.
31982 2017-02-07 Jakub Jelinek <jakub@redhat.com>
31984 PR rtl-optimization/79386
31985 * cprop.c (bypass_conditional_jumps): Initialize
31986 bypass_last_basic_block already before splitting bbs after
31987 unconditional traps...
31988 (bypass_conditional_jumps): ... rather than here.
31991 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
31992 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
31993 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
31994 fix -masm=intel patterns.
31996 2017-02-07 Richard Biener <rguenther@suse.de>
31998 PR tree-optimization/79256
31999 PR middle-end/79278
32000 * builtins.c (get_object_alignment_2): Use min_align_of_type
32001 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
32002 and ADJUST_FIELD_ALIGN.
32004 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
32006 * doc/tm.texi: Regenerate.
32007 * stor-layout.c (layout_decl): Adjust.
32008 (update_alignment_for_field): Likewise.
32009 (place_field): Likewise.
32010 (min_align_of_type): Likewise.
32011 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
32012 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
32013 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
32014 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
32015 * config/frv/frv.c (frv_adjust_field_align): Likewise.
32016 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
32017 * config/i386/i386.c (x86_field_alignment): Likewise.
32018 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
32019 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
32020 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
32021 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
32022 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
32023 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
32027 2017-01-30 Richard Biener <rguenther@suse.de>
32029 PR tree-optimization/79256
32030 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
32031 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
32034 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
32036 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
32037 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
32038 builtins to SImode and emit a zero-extend, if necessary.
32040 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
32042 * docs/invoke.texi (RISC-V Options): Alphabetize.
32044 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
32046 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
32049 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
32051 * config/riscv/riscv.c: New file.
32052 * common/config/riscv/riscv-common.c: Likewise.
32053 * config.gcc: Likewise.
32054 * config/riscv/constraints.md: Likewise.
32055 * config/riscv/elf.h: Likewise.
32056 * config/riscv/generic.md: Likewise.
32057 * config/riscv/linux.h: Likewise.
32058 * config/riscv/multilib-generator: Likewise.
32059 * config/riscv/peephole.md: Likewise.
32060 * config/riscv/pic.md: Likewise.
32061 * config/riscv/predicates.md: Likewise.
32062 * config/riscv/riscv-builtins.c: Likewise.
32063 * config/riscv/riscv-c.c: Likewise.
32064 * config/riscv/riscv-ftypes.def: Likewise.
32065 * config/riscv/riscv-modes.def: Likewise.
32066 * config/riscv/riscv-opts.h: Likewise.
32067 * config/riscv/riscv-protos.h: Likewise.
32068 * config/riscv/riscv.h: Likewise.
32069 * config/riscv/riscv.md: Likewise.
32070 * config/riscv/riscv.opt: Likewise.
32071 * config/riscv/sync.md: Likewise.
32072 * config/riscv/t-elf-multilib: Likewise.
32073 * config/riscv/t-linux: Likewise.
32074 * config/riscv/t-linux-multilib: Likewise.
32075 * config/riscv/t-riscv: Likewise.
32076 * configure.ac: Likewise.
32077 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
32078 Waterman as RISC-V maintainers.
32079 * doc/install.texi: Add RISC-V entries.
32080 * doc/invoke.texi: Add RISC-V options section.
32081 * doc/md.texi: Add RISC-V constraints section.
32082 * configure: Regenerated.
32084 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
32087 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
32088 false values to be constant vectors with all 0 or all 1 bits set.
32089 (vcondu<mode><mode>): Likewise.
32090 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
32092 (fpmask_comparison_operator): Update comment.
32093 (vecint_comparison_operator): New predicate.
32094 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
32095 vector conditionals when the true and false values are constant
32096 vectors with all 0 bits or all 1 bits set.
32098 2017-02-06 Martin Sebor <msebor@redhat.com>
32100 PR tree-optimization/79376
32101 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
32103 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
32105 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
32106 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
32107 to simplify split condition.
32109 2017-02-06 Jakub Jelinek <jakub@redhat.com>
32111 * omp-expand.c (oxpand_omp_atomic_fetch_op,
32112 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
32115 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
32117 PR rtl-optimization/68664
32118 * target.def (can_speculate_insn): New hook.
32119 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
32120 * doc/tm.texi: Regenerate.
32121 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
32122 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
32123 (rs6000_sched_can_speculate_insn): New function.
32125 2017-02-06 Jakub Jelinek <jakub@redhat.com>
32127 PR tree-optimization/79284
32128 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
32129 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
32130 vectorizable_mask_load_store, vectorizable_operation,
32131 vect_is_simple_cond, get_same_sized_vectype): Use it instead
32132 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
32133 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
32134 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
32135 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
32136 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
32137 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
32138 is_gimple_assign (stmt). Replace another such test with
32139 is_gimple_assign (stmt).
32141 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
32144 * config/avr/avr.c (rtl-iter.h): Include it.
32145 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
32146 (avr_legitimate_combined_insn): ...and implementation.
32148 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
32150 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
32151 * config/s390/s390.c (s390_const_operand_ok)
32152 (s390_canonicalize_comparison, s390_extract_part)
32153 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
32154 (s390_contiguous_bitmask_p, s390_rtx_costs)
32155 (legitimize_pic_address): Likewise.
32156 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
32157 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
32158 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
32159 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
32160 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
32162 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
32164 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
32165 REGNO($0) == REGNO($1).
32167 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32169 * config/s390/linux.h(SIZE_TYPE): Add comment.
32171 2017-02-06 Julian Brown <julian@codesourcery.com>
32172 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32173 Virendra Pathak <virendra.pathak@broadcom.com>
32175 * config/aarch64/aarch64-cores.def: Change the scheduler
32177 * config/aarch64/aarch64.md: Include thunderx2t99.md.
32178 * config/aarch64/thunderx2t99.md: New file.
32180 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
32182 * doc/standards.texi (Go Language): Update link to language
32185 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
32187 * tree-eh.c (lower_resx): Sanitize profile.
32188 (cleanup_empty_eh_move_lp): Likewise.
32190 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
32193 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
32195 * cfgloopmanip.h (loop_version): Update prototype.
32196 * modulo-sched.c (sms_schedule): Update call of loop_version.
32197 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
32198 * tree-parloops.c (gen_parallel_loop): Likewise.
32199 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
32200 * tree-ssa-loop-split.c (split_loop): Likewise.
32201 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
32202 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
32204 2017-02-05 Martin Liska <mliska@suse.cz>
32207 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
32209 (print_operand_address): Initialize a struct to zero.
32211 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
32213 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
32214 garbage collector only in textual form.
32216 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
32218 * doc/extend.texi (x86 specific memory model extensions for
32219 transactional memory): Simplify a phrase.
32221 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
32224 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
32225 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
32226 (atomic_storedi_1): Likewise.
32228 2017-02-04 Jakub Jelinek <jakub@redhat.com>
32230 PR tree-optimization/79338
32231 * tree-parloops.c (gather_scalar_reductions): Don't call
32232 vect_analyze_loop_form for loop->inner before destroying loop's
32235 2017-02-03 Martin Sebor <msebor@redhat.com>
32237 PR tree-optimization/79327
32238 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
32239 when precision has resulted in leading zeros.
32240 (format_integer): Adjust the likely counter to assume an unknown
32241 argument that may be zero is non-zero.
32243 2017-02-03 Jason Merrill <jason@redhat.com>
32246 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
32247 avoid copying non-taken branch.
32249 2017-02-03 Jakub Jelinek <jakub@redhat.com>
32251 PR tree-optimization/79340
32252 * tree-vect-loop.c (vectorizable_reduction): Release
32253 vec_defs elements after safe_splicing them into other vectors.
32256 PR tree-optimization/79327
32257 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
32258 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
32260 (format_integer): Use wide_int_to_tree instead of build_int_cst
32261 + to_?hwi. If argmin is NULL, just set argmin and argmax to
32262 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
32263 of shortest and longest sequence.
32265 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
32267 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
32268 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
32270 2017-02-03 Walter Lee <walt@tilera.com>
32273 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
32274 after initial stackframe link reg save.
32275 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
32277 2017-02-03 Jakub Jelinek <jakub@redhat.com>
32280 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
32281 wu for stxssp alternative.
32283 2017-02-03 Martin Sebor <msebor@redhat.com>
32285 PR tree-optimization/79352
32286 * gimple-fold.c (get_range_strlen): Add argument.
32287 (get_range_strlen): Change return type to bool.
32288 (get_maxval_strlen): Pass in a dummy argument.
32289 * gimple-fold.h (get_range_strlen): Change return type to bool.
32290 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
32291 * tree.h (array_at_struct_end_p): Add argument.
32292 * tree.c (array_at_struct_end_p): Handle it.
32294 2017-02-03 Martin Liska <mliska@suse.cz>
32297 * multiple_target.c (create_dispatcher_calls): Redirect edge
32298 from a caller of a dispatcher.
32299 (expand_target_clones): Make the clones local.
32300 (ipa_target_clone): Do both target clones and resolvers.
32301 (ipa_dispatcher_calls): Remove the pass.
32302 (pass_dispatcher_calls::gate): Likewise.
32303 (make_pass_dispatcher_calls): Likewise.
32304 * passes.def (pass_target_clone): Put as very first IPA early
32307 2017-02-03 Martin Liska <mliska@suse.cz>
32309 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
32310 in case of a function with ifunc attribute.
32312 2017-02-03 Martin Liska <mliska@suse.cz>
32314 * cgraph.c (cgraph_node::dump): Dump function version info.
32315 * symtab.c (symtab_node::dump_base): Add missing new line.
32317 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
32319 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
32320 (ifcombine_ifandif): Use it.
32322 2017-02-03 Martin Liska <mliska@suse.cz>
32324 * doc/invoke.texi: Document default value for
32325 use-after-scope-direct-emission-threshold.
32327 2017-02-03 Martin Liska <mliska@suse.cz>
32329 PR tree-optimization/79339
32330 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
32331 (format_floating): Likewise.
32333 2017-02-03 Martin Liska <mliska@suse.cz>
32336 * ipa-prop.c (ipa_node_params_t::insert): Remove current
32338 (ipa_node_params_t::remove): Likewise.
32339 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
32340 initialization from removed ipa_node_params_t::insert.
32341 (ipa_node_params::~ipa_node_params): Move from removed
32342 ipa_node_params_t::release.
32343 * symbol-summary.h (symbol_summary::m_released): New member.
32344 Do not release a summary twice. Do not allow to call finalizer
32345 for types of a summary that live in GGC memory.
32347 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32349 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
32352 2017-02-02 Martin Sebor <msebor@redhat.com>
32354 PR middle-end/79275
32355 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
32356 (format_string): Tighten up the range of output for non-constant
32357 strings and correct the expected range for wide non-constant strings.
32359 2017-02-02 Martin Sebor <msebor@redhat.com>
32361 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
32363 PR middle-end/32003
32364 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
32366 (-fdump-tree-@var): Add to index and document how to come up
32367 with pass-specific option and dump file names.
32368 (-fdump-passes): Clarify where to look for output.
32370 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
32372 PR middle-end/77445
32373 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
32374 statistics of the analyzed path; allow threading for speed when
32375 any of BBs along the path are optimized for speed.
32377 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
32379 PR middle-end/78468
32380 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
32381 settings of the virtual registers.
32384 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
32386 * explow.c (get_dynamic_stack_size): Take known alignment of stack
32387 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
32390 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32392 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
32393 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
32395 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32397 * config/s390/s390.md: Add missing comments with the expanded
32399 * config/s390/vector.md: Likewise.
32400 * config/s390/vx-builtins.md: Likewise.
32402 2017-02-02 Jakub Jelinek <jakub@redhat.com>
32405 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
32406 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
32407 conditions on a single line.
32409 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32411 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
32412 __S390_VX__ to __VX__.
32414 2017-02-01 Andrew Pinski <apinski@cavium.com>
32416 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
32417 stmt_info to record_stmt_cost.
32418 (vect_get_known_peeling_cost): Pass stmt_info if known to
32420 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
32421 cpu_vector_cost field into
32422 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
32423 field into vec_int_stmt_cost and vec_fp_stmt_cost.
32424 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
32425 splitting of scalar_stmt_cost and vec_stmt_cost.
32426 (thunderx_vector_cost): Likewise.
32427 (cortexa57_vector_cost): LIkewise.
32428 (exynosm1_vector_cost): Likewise.
32429 (xgene1_vector_cost): Likewise.
32430 (thunderx2t99_vector_cost): Improve after the splitting of the two
32432 (aarch64_builtin_vectorization_cost): Update for the splitting of
32433 scalar_stmt_cost and vec_stmt_cost.
32435 2017-02-01 Torvald Riegel <triegel@redhat.com>
32436 Richard Henderson <rth@redhat.com>
32438 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
32439 conditional on existance of a fast atomic load.
32440 * optabs-query.c (can_atomic_load_p): New function.
32441 * optabs-query.h (can_atomic_load_p): Declare it.
32442 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
32443 no fast atomic load is available for the particular size of access.
32444 (expand_atomic_compare_and_swap): Likewise.
32445 (expand_atomic_load): Likewise.
32446 (expand_atomic_store): Likewise.
32447 (expand_atomic_fetch_op): Likewise.
32448 * testsuite/lib/target-supports.exp
32449 (check_effective_target_sync_int_128): Remove x86 because it provides
32450 no fast atomic load.
32451 (check_effective_target_sync_int_128_runtime): Likewise.
32453 2017-02-01 Richard Biener <rguenther@suse.de>
32455 * graphite.c: Include tree-vectorizer.h for find_loop_location.
32456 (graphite_transform_loops): Provide opt-info for optimized nests.
32457 * tree-parloop.c (parallelize_loops): Provide opt-info for
32458 parallelized loops.
32460 2017-02-01 Richard Biener <rguenther@suse.de>
32462 PR middle-end/79315
32463 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
32464 was not set before.
32466 2017-02-01 Richard Biener <rguenther@suse.de>
32468 PR tree-optimization/71824
32469 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
32470 Verify the loops are valid in the merged SESE region.
32471 (scop_detection::can_represent_loop_1): Check analyzing the
32472 evolution of the number of iterations in the region succeeds.
32474 2017-01-31 Ian Lance Taylor <iant@golang.org>
32476 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
32477 REG_ARGS_SIZE note to 32-bit push insns and call insn.
32479 2017-01-31 David Malcolm <dmalcolm@redhat.com>
32481 PR preprocessor/79210
32482 * input.c (get_substring_ranges_for_loc): Replace line_width
32483 assertion with error-handling.
32485 2017-01-31 Richard Biener <rguenther@suse.de>
32487 PR tree-optimization/77318
32488 * graphite-sese-to-poly.c (extract_affine): Fix assert.
32489 (create_pw_aff_from_tree): Take loop parameter.
32490 (add_condition_to_pbb): Pass loop of the condition to
32491 create_pw_aff_from_tree.
32493 2017-01-31 Jakub Jelinek <jakub@redhat.com>
32495 * config/s390/s390.c (s390_asan_shadow_offset): New function.
32496 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
32498 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
32502 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
32504 (convert_int_to_float128): Likewise.
32505 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
32506 (convert_int_to_float128): Likewise.
32507 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
32508 (UNSPEC_IEEE128_CONVERT): Likewise.
32509 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
32510 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
32511 Use local variables for IBM extended format.
32512 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
32513 (fix_trunc<mode>si2_fprs): Likewise.
32514 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
32515 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
32516 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
32517 to know that we can now have integers of all sizes in vector
32519 (fix<uns>_<mode>di2_hw): Likewise.
32520 (float<uns>_<mode>si2_hw): Likewise.
32521 (fix_<mode>si2_hw): Likewise.
32522 (fixuns_<mode>si2_hw): Likewise.
32523 (float<uns>_<mode>di2_hw): Likewise.
32524 (float_<mode>di2_hw): Likewise.
32525 (float_<mode>si2_hw): Likewise.
32526 (floatuns_<mode>di2_hw): Likewise.
32527 (floatuns_<mode>si2_hw): Likewise.
32528 (xscvqp<su>wz_<mode>): Delete, no longer used.
32529 (xscvqp<su>dz_<mode>): Likewise.
32530 (xscv<su>dqp_<mode>): Likewise.
32531 (ieee128_mfvsrd_64bit): Likewise.
32532 (ieee128_mfvsrd_32bit): Likewise.
32533 (ieee128_mfvsrwz): Likewise.
32534 (ieee128_mtvsrw): Likewise.
32535 (ieee128_mtvsrd_64bit): Likewise.
32536 (ieee128_mtvsrd_32bit): Likewise.
32538 2017-01-31 Martin Liska <mliska@suse.cz>
32541 * ipa-prop.c (ipa_free_all_node_params): Call release method
32542 instead of ~sumbol_summary to not to trigger double times
32545 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
32547 PR tree-optimization/71691
32548 * bitmap.h (class auto_bitmap): New.
32549 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
32550 is_maybe_undefined instead of ssa_undefined_value_p.
32552 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32554 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
32555 __S390_ARCH_LEVEL__ to __ARCH__.
32557 2017-01-31 Jakub Jelinek <jakub@redhat.com>
32559 PR tree-optimization/79267
32560 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
32561 if should_remove_lhs_p is true.
32563 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
32566 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
32567 (add_alignment_attribute): New.
32568 (base_type_die): Add alignment attribute.
32569 (subrange_type_die): Likewise.
32570 (modified_type_die): Likewise.
32571 (gen_array_type_die): Likewise.
32572 (gen_descr_array_type_die: Likewise.
32573 (gen_enumeration_type_die): Likewise.
32574 (gen_subprogram_die): Likewise.
32575 (gen_variable_die): Likewise.
32576 (gen_field_die): Likewise.
32577 (gen_ptr_to_mbr_type_die): Likewise.
32578 (gen_struct_or_union_type_die): Likewise.
32579 (gen_subroutine_type_die): Likewise.
32580 (gen_typedef_die): Likewise.
32581 (base_type_cmp): Compare alignment attribute.
32583 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32586 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
32587 (setb_unsigned) New pattern for setb with CCUNS.
32588 * config/rs6000/rs6000.c (expand_block_compare): Use a different
32589 subfc./subfe sequence to avoid overflow problems. Generate a
32590 shorter sequence with cmpld/setb for power9.
32591 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
32592 for generating subfc. instruction.
32593 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
32594 now uses this instruction.
32596 2017-01-30 Ian Lance Taylor <iant@google.com>
32599 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
32600 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
32602 2017-01-30 Martin Sebor <msebor@redhat.com>
32604 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
32605 Move constant to the right of a relational operator.
32606 (get_mpfr_format_length, format_character, format_string): Ditto.
32607 (should_warn_p, maybe_warn): Same.
32609 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
32611 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
32614 * asan.c (get_translation_unit_decl): Remove function.
32615 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
32617 2017-01-30 Martin Liska <mliska@suse.cz>
32619 PR gcov-profile/79259
32620 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
32621 -fprofile-generate.
32623 2017-01-30 Martin Liska <mliska@suse.cz>
32626 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
32627 Initialize variables with NULL value.
32629 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
32632 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
32634 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
32636 2017-01-30 Richard Biener <rguenther@suse.de>
32638 PR tree-optimization/79276
32639 * tree-vrp.c (process_assert_insertions): Properly adjust common
32640 when removing a duplicate.
32642 2017-01-30 Richard Biener <rguenther@suse.de>
32644 PR tree-optimization/79256
32645 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
32646 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
32648 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
32650 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
32653 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
32654 ("*r<noxa>sbg_<mode>_sll_bitmask")
32655 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
32656 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
32657 Use contiguous_bitmask_nowrap_operand.
32659 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32662 * config/rs6000/altivec.h (vec_xl): Revise #define.
32663 (vec_xst): Likewise.
32665 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
32667 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
32669 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
32671 PR rtl-optimization/79194
32672 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
32673 traps before call to bypass_conditional_jumps.
32675 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
32677 PR tree-optimization/71374
32678 * lra-constraints.c (check_conflict_input_operands): New.
32679 (match_reload): Use it.
32681 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
32684 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
32685 account to calculate conflict_set.
32687 2017-01-27 Bin Cheng <bin.cheng@arm.com>
32689 PR rtl-optimization/78559
32690 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
32691 other_insn in combine.
32693 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
32695 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
32696 uint16_type_node for BT_UINT16.
32698 2017-01-27 David Malcolm <dmalcolm@redhat.com>
32700 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
32701 "RTL Tests" to menu.
32702 (GIMPLE Tests): New node.
32703 (RTL Tests): New node.
32705 2017-01-27 Richard Biener <rguenther@suse.de>
32707 PR tree-optimization/79245
32708 * tree-loop-distribution.c (distribute_loop): Apply cost
32709 modeling also to detected patterns.
32711 2017-01-27 Richard Biener <rguenther@suse.de>
32713 PR tree-optimization/71433
32714 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
32715 (compare_assert_loc): New function.
32716 (process_assert_insertions): Sort and optimize assert locations
32717 to remove duplicates and push down identical assertions on
32718 edges to their destination block.
32720 2017-01-27 Richard Biener <rguenther@suse.de>
32722 PR tree-optimization/79244
32723 * tree-vrp.c (remove_range_assertions): Forcefully propagate
32724 out SSA names even if abnormal.
32726 2017-01-27 Jakub Jelinek <jakub@redhat.com>
32728 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
32729 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
32730 instead of MPFR_RNDN.
32732 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
32735 * arm.c (arm_option_override): Don't call build_target_option_node
32736 until after doing all option overrides.
32737 (arm_valid_target_attribute_tree): Likewise.
32739 2017-01-27 Martin Liska <mliska@suse.cz>
32741 * doc/invoke.texi (-fprofile-arcs): Document profiling support
32742 for {cd}tors and C++ {cd}tors.
32744 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
32746 * config/s390/s390.md ("*setmem_long_and")
32747 ("*setmem_long_and_31z"): Use zero_extend instead of and.
32749 2017-01-26 Martin Sebor <msebor@redhat.com>
32751 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
32754 2017-01-26 Martin Sebor <msebor@redhat.com>
32756 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
32757 HAVE_DFmode before using XFmode or DFmode.
32758 (parse_directive): Avoid using the z length modifier to avoid
32759 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
32761 PR middle-end/78703
32762 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
32763 to accept adjustment as an array.
32764 (get_int_range): New function.
32765 (struct directive): Make width and prec arrays.
32766 (directive::set_width, directive::set_precision): Call get_int_range.
32767 (format_integer, format_floating): Handle width and precision ranges.
32768 (format_string, parse_directive): Same.
32770 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32773 * dwarf2out.c (generate_skeleton_bottom_up): For children with
32774 comdat_type_p set, just clone them, but keep the children in the
32778 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
32779 which have direct callers with -fvar-tracking-assignments enabled
32781 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
32782 inside of type units.
32784 2017-01-26 Martin Sebor <msebor@redhat.com>
32786 PR middle-end/78703
32787 * gimple-ssa-sprintf.c (struct result_range): Add likely and
32789 (struct format_result): Replace number_chars, number_chars_min,
32790 and number_chars_max with a single member of struct result_range.
32792 (format_result::operator+=): Adjust.
32793 (struct fmtresult): Remove bounded. Handle likely and unlikely
32795 (fmtresult::adjust_for_width_or_precision): New function.
32796 (fmtresult:type_max_digits): New function.
32797 (bytes_remaining): Handle likely and unlikely counters.
32798 (min_bytes_remaining): Remove.
32799 (format_percent): Simplify.
32800 (format_integer, format_floating): Set likely and unlikely counters.
32801 (get_string_length, format_character, format_string): Same.
32802 (format_plain, should_warn_p): New function.
32803 (maybe_warn): Call should_warn_p. Update diagnostic messages
32804 and handle those for all directives, including plain strings.
32805 (format_directive): Handle likely and unlikely counters.
32806 Remove unnecessary quoting from diagnostics. Add an informational
32808 (add_bytes): Remove.
32809 (pass_sprintf_length::compute_format_length): Simplify.
32810 (try_substitute_return_value): Handle likely and unlikely counters.
32812 2017-01-26 Carl Love <cel@us.ibm.com>
32814 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
32815 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
32817 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
32820 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
32821 endianess for subregs into account.
32822 * lra-constraints.c (lra_constraints): Do risky transformations
32823 always on the first iteration.
32824 * lra-lives.c (check_pseudos_live_through_calls): Add arg
32825 last_call_used_reg_set.
32826 (process_bb_lives): Define and use last_call_used_reg_set.
32827 * lra.c (lra): Always continue after lra_constraints on the first
32830 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
32832 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
32834 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
32836 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32838 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
32839 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
32840 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
32841 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
32842 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
32843 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
32844 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
32845 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
32846 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
32848 2017-01-26 Marek Polacek <polacek@redhat.com>
32851 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
32852 for the third operand.
32854 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32856 PR middle-end/79236
32857 * omp-low.c (struct omp_context): Add simt_stmt field.
32858 (scan_omp_for): Return omp_context *.
32859 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
32860 context to the _simt_ SIMD stmt.
32861 (lower_omp_for): For combined SIMD with sibling _simt_
32862 SIMD, make sure to use the same decls in _looptemp_
32863 clauses as in the sibling.
32865 2017-01-26 David Sherwood <david.sherwood@arm.com>
32867 PR middle-end/79212
32868 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
32871 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32874 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
32875 emit fld b; fld a; if possible.
32877 * brig-builtins.def: Update copyright years.
32878 * config/arm/arm_acle_builtins.def: Update copyright years.
32880 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
32883 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
32884 constraint instead of o for the stxsd instruction.
32886 2017-01-25 Carl Love <cel@us.ibm.com>
32888 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
32889 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
32891 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
32893 * doc/invoke.texi (C++ Dialect Options): Fix typo.
32895 2017-01-25 Richard Biener <rguenther@suse.de>
32897 PR tree-optimization/69264
32898 * target.def (vector_alignment_reachable): Improve documentation.
32899 * doc/tm.texi: Regenerate.
32900 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
32902 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
32903 earlier changes with respect to TYPE_USER_ALIGN.
32904 (vector_alignment_reachable_p): Likewise. Improve dumping.
32906 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32909 * config/arm/arm.md (xordi3): Force constant operand into a register
32912 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32914 * doc/invoke.texi (-fstore-merging): Correct default optimization
32915 levels at which it is enabled.
32916 (-O): Move -fstore-merging from list to...
32919 2017-01-25 Richard Biener <rguenther@suse.de>
32922 * omp-expand.c: Include debug.h.
32923 (expand_omp_taskreg): Make sure to generate early debug before
32924 outlining anything from a function.
32925 (expand_omp_target): Likewise.
32926 (grid_expand_target_grid_body): Likewise.
32928 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
32931 * asan.c (get_translation_unit_decl): New function.
32932 (asan_add_global): Extract modules file name from globals
32933 TRANSLATION_UNIT_DECL name.
32935 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
32938 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
32939 for long calls with APCS frame and VFP.
32941 2017-01-24 David Malcolm <dmalcolm@redhat.com>
32943 * cfg.c (original_copy_tables_initialized_p): New function.
32944 * cfg.h (original_copy_tables_initialized_p): New decl.
32945 * cfgrtl.c (relink_block_chain): Guard the call to
32946 free_original_copy_tables with a call to
32947 original_copy_tables_initialized_p.
32948 * cgraph.h (symtab_node::native_rtl_p): New decl.
32949 * cgraphunit.c (symtab_node::native_rtl_p): New function.
32950 (symtab_node::needed_p): Don't assert for early assembly output
32951 for __RTL functions.
32952 (cgraph_node::finalize_function): Set "force_output" for __RTL
32954 (cgraph_node::analyze): Bail out early for __RTL functions.
32955 (analyze_functions): Update assertion to support __RTL functions.
32956 (cgraph_node::expand): Bail out early for __RTL functions.
32957 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
32959 * function.h (struct function): Update comment for field
32961 * gimple-expr.c: Include "tree-pass.h".
32962 (gimple_has_body_p): Return false for __RTL functions.
32963 * Makefile.in (OBJS): Add run-rtl-passes.o.
32964 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
32966 (gcc::pass_manager::get_clean_slate): New accessor.
32967 * passes.c: Include "insn-addr.h".
32968 (should_skip_pass_p): Add logging. Update logic for running
32969 "expand" to be compatible with both __GIMPLE and __RTL. Guard
32970 property-provider override so it is only done for gimple passes.
32972 (skip_pass): New function.
32973 (execute_one_pass): Call skip_pass when skipping passes.
32974 * read-md.c (md_reader::read_char): Support filtering
32975 the input to a subset of line numbers.
32976 (md_reader::md_reader): Initialize fields
32977 m_first_line and m_last_line.
32978 (md_reader::read_file_fragment): New function.
32979 * read-md.h (md_reader::read_file_fragment): New decl.
32980 (md_reader::m_first_line): New field.
32981 (md_reader::m_last_line): New field.
32982 * read-rtl-function.c (function_reader::create_function): Only
32983 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
32984 curr_properties. Set DECL_INITIAL to a dummy block.
32985 (read_rtl_function_body_from_file_range): New function.
32986 * read-rtl-function.h (read_rtl_function_body_from_file_range):
32988 * run-rtl-passes.c: New file.
32989 * run-rtl-passes.h: New file.
32991 2017-01-24 Jeff Law <law@redhat.com>
32993 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
32996 2017-01-24 Bin Cheng <bin.cheng@arm.com>
32998 PR tree-optimization/79159
32999 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
33000 (record_nonwrapping_iv): Improve boundary using above function if no
33001 value range information.
33003 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
33004 Martin Jambor <mjambor@suse.cz>
33006 * brig-builtins.def: New file.
33007 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
33008 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
33009 (DEF_HSAIL_SAT_BUILTIN): Likewise.
33010 (DEF_HSAIL_INTR_BUILTIN): Likewise.
33011 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
33012 * builtin-types.def (BT_INT8): New.
33013 (BT_INT16): Likewise.
33014 (BT_UINT8): Likewise.
33015 (BT_UINT16): Likewise.
33016 (BT_FN_ULONG): Likewise.
33017 (BT_FN_UINT_INT): Likewise.
33018 (BT_FN_UINT_ULONG): Likewise.
33019 (BT_FN_UINT_LONG): Likewise.
33020 (BT_FN_UINT_PTR): Likewise.
33021 (BT_FN_ULONG_PTR): Likewise.
33022 (BT_FN_INT8_FLOAT): Likewise.
33023 (BT_FN_INT16_FLOAT): Likewise.
33024 (BT_FN_UINT32_FLOAT): Likewise.
33025 (BT_FN_UINT16_FLOAT): Likewise.
33026 (BT_FN_UINT8_FLOAT): Likewise.
33027 (BT_FN_UINT64_FLOAT): Likewise.
33028 (BT_FN_UINT16_UINT32): Likewise.
33029 (BT_FN_UINT32_UINT16): Likewise.
33030 (BT_FN_UINT16_UINT16_UINT16): Likewise.
33031 (BT_FN_INT_PTR_INT): Likewise.
33032 (BT_FN_UINT_PTR_UINT): Likewise.
33033 (BT_FN_LONG_PTR_LONG): Likewise.
33034 (BT_FN_ULONG_PTR_ULONG): Likewise.
33035 (BT_FN_VOID_UINT64_UINT64): Likewise.
33036 (BT_FN_UINT8_UINT8_UINT8): Likewise.
33037 (BT_FN_INT8_INT8_INT8): Likewise.
33038 (BT_FN_INT16_INT16_INT16): Likewise.
33039 (BT_FN_INT_INT_INT): Likewise.
33040 (BT_FN_UINT_FLOAT_UINT): Likewise.
33041 (BT_FN_FLOAT_UINT_UINT): Likewise.
33042 (BT_FN_ULONG_UINT_UINT): Likewise.
33043 (BT_FN_ULONG_UINT_PTR): Likewise.
33044 (BT_FN_ULONG_ULONG_ULONG): Likewise.
33045 (BT_FN_UINT_UINT_UINT): Likewise.
33046 (BT_FN_VOID_UINT_PTR): Likewise.
33047 (BT_FN_UINT_UINT_PTR: Likewise.
33048 (BT_FN_UINT32_UINT64_PTR): Likewise.
33049 (BT_FN_INT_INT_UINT_UINT): Likewise.
33050 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
33051 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
33052 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
33053 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
33054 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
33055 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
33056 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
33057 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
33058 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
33059 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
33060 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
33061 * doc/frontends.texi: List BRIG FE.
33062 * doc/install.texi (Testing): Add BRIG tesring requirements.
33063 * doc/invoke.texi (Overall Options): Mention BRIG.
33064 * doc/standards.texi (Standards): Doucment BRIG HSA version.
33066 2017-01-24 Richard Biener <rguenther@suse.de>
33068 PR translation/79208
33069 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
33071 2017-01-24 Martin Jambor <mjambor@suse.cz>
33074 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
33075 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
33076 and known_contexts.
33078 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
33080 PR middle-end/79123
33081 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
33082 casts from signed to unsigned really don't have a range.
33084 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
33086 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
33087 GMP_RNDx for compatiblity.
33089 2017-01-24 Martin Liska <mliska@suse.cz>
33092 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
33093 that would prevent us to call alloca with -1 as argument.
33095 2017-01-24 Jakub Jelinek <jakub@redhat.com>
33097 * dwarf2out.c (output_compilation_unit_header, output_file_names):
33098 Avoid -Wformat-security warning.
33100 2017-01-23 Andrew Pinski <apinski@cavium.com>
33102 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
33105 2017-01-23 Martin Sebor <msebor@redhat.com>
33107 PR middle-end/78703
33108 * gimple-ssa-sprintf.c (warn_level): New global.
33109 (format_integer): Use it here and throughout the rest of the file.
33110 Use the same switch to compute sign as base.
33111 (maybe_warn): New function.
33112 (format_directive): Factor out warnings into maybe_warn.
33113 Add debugging output. Use warn_level.
33114 (add_bytes): Use warn_level.
33115 (pass_sprintf_length::compute_format_length): Add debugging output.
33116 (try_substitute_return_value): Same.
33117 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
33119 PR middle-end/78703
33120 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
33121 (struct fmtresult, format_integer, format_floating): Adjust.
33122 (fmtresult::fmtresult): Set max correctly in two argument ctor.
33123 (get_string_length, format_string,format_directive): Same.
33124 (pass_sprintf_length::compute_format_length): Same.
33125 (try_substitute_return_value): Simplify slightly.
33127 PR middle-end/78703
33128 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
33129 (fmtresult::operator+=): Outlined.
33130 (struct fmtresult): Add ctors.
33131 (struct conversion_spec): Rename...
33132 (struct directive): ...to this. Add and remove data members.
33133 (directive::set_width, directive::set_precision): New functions.
33134 (format_percent): Use fmtresult ctor.
33135 (get_width_and_precision): Remove.
33136 (format_integer): Make naming changes. Avoid computing width and
33138 (format_floating): Same. Adjust indentation.
33139 (format_character, format_none): New functions.
33140 (format_string): Moved character handling to format_character.
33141 (format_directive): Remove arguments, change return type.
33142 (parse_directive): New function.
33143 (pass_sprintf_length::compute_format_length): Move directive
33144 parsing to parse_directive.
33146 2017-01-23 Jakub Jelinek <jakub@redhat.com>
33148 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
33149 (assign_assembler_name_if_needed): ... this.
33150 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
33151 (assign_assembler_name_if_needed): ... this.
33152 (free_lang_data_in_cgraph): Adjust callers.
33153 * cgraphunit.c (cgraph_node::analyze): Likewise.
33154 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
33157 2017-01-23 Richard Biener <rguenther@suse.de>
33159 PR tree-optimization/79088
33160 PR tree-optimization/79188
33161 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
33162 resetting loop bounds after last path deletion. Reset loop
33163 bounds of the target loop, make code match the comments.
33164 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
33165 Make sure loops need no fixups.
33167 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
33169 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
33170 exponent support with double type for first argument.
33171 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
33172 type returned by __builtin_vec_extract_sig,
33173 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
33174 functions from "vector int" to "vector unsigned int" or from
33175 "vector long long int" to "vector unsigned long long int".
33176 Changed type returned by __builtin_vec_extract_exp,
33177 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
33178 functions from "vector int" to "vector unsigned int" or from
33179 "vector long long int" to "vector unsigned long long int".
33180 Changed return type of __builtin_vec_test_data_class,
33181 __builtin_vec_test_data_class_sp, and
33182 __builtin_vec_test_data_class_dp from "vector int" to
33183 "vector bool int" or from "vector long long int" to "vector bool
33184 long long int" and changed second argument type from "unsigned
33185 int" to "int". Added new overloaded function forms "vector float
33186 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
33187 "vector float __builtin_vec_insert_exp_sp (vector float, vector
33188 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
33189 double, vector unsigned long long int)" and "vector double
33190 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
33191 long int)". Changed return type of
33192 __builtin_scalar_test_data_class and
33193 __builtin_scalar_test_data_class_sp and
33194 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
33195 int" and changed second argument from "unsigned int" to "int".
33196 Changed type returned by __builtin_scalar_test_neg,
33197 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
33198 from "int" to "bool int". Added new overloaded function form
33199 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
33200 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
33201 exponent double-precision with floating point first argument.
33202 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
33203 documentation of scalar_test_data_class, scalar_test_neg,
33204 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
33205 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
33206 vec_test_data_class built-in functions to reflect refinements in
33207 their type signatures.
33209 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
33211 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
33213 (aarch64_elf_asm_destructor): Likewise.
33215 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
33217 PR rtl-optimization/78634
33218 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
33219 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
33220 * ifcvt.c (noce_try_cmove): Add missing cost check.
33222 PR rtl-optimization/71724
33223 * combine.c (if_then_else_cond): Look for situations where it is
33224 beneficial to undo the work of one of the recursive calls.
33226 2017-01-23 Bin Cheng <bin.cheng@arm.com>
33228 PR tree-optimization/70754
33229 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
33230 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
33231 combined stmt before it if not NULL.
33232 (combine_chains): Process refs reversely and compute dominance point
33235 2017-01-23 Martin Liska <mliska@suse.cz>
33237 PR tree-optimization/79196
33238 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
33239 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
33241 (strlen_optimize_stmt): Call the renamed function.
33243 2017-01-23 Michael Matz <matz@suse.de>
33245 PR tree-optimization/78384
33246 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
33248 2017-01-23 Richard Biener <rguenther@suse.de>
33250 PR tree-optimization/79186
33251 * tree-vrp.c (register_new_assert_for): Make sure we've seen
33252 both incoming edges before moving an assert.
33254 2017-01-23 Martin Jambor <mjambor@suse.cz>
33256 * ipa-prop.c (load_from_param_1): Removed.
33257 (load_from_unmodified_param): Bits from load_from_param_1 put back
33259 (load_from_param): Removed.
33260 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
33261 with stmt. Reverted back to use of load_from_unmodified_param.
33263 2017-01-23 Martin Jambor <mjambor@suse.cz>
33266 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
33267 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
33268 field a pointer to garbage collected vector, mark lattices and
33269 ipcp_orig_node with GTY((skip)).
33270 (ipa_get_param_count): Adjust to descriptors being a pointer.
33271 (ipa_get_param): Likewise.
33272 (ipa_get_type): Likewise.
33273 (ipa_get_param_move_cost): Likewise.
33274 (ipa_set_param_used): Likewise.
33275 (ipa_get_controlled_uses): Likewise.
33276 (ipa_set_controlled_uses): Likewise.
33277 (ipa_is_param_used): Likewise.
33278 (ipa_node_params_t): Move into garbage collector. New methods insert
33280 (ipa_node_params_sum): Annotate wth GTY(()).
33281 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
33283 (ipa_load_from_parm_agg): Adjust declaration.
33284 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
33285 * ipa-profile.c (ipa_profile): Likewise.
33286 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
33287 (ipa_populate_param_decls): Make descriptors parameter garbage
33289 (ipa_dump_param): Adjust to descriptors being a pointer.
33290 (ipa_alloc_node_params): Likewise.
33291 (ipa_initialize_node_params): Likewise.
33292 (load_from_param_1): Make descriptors parameter garbage collected.
33293 (load_from_unmodified_param): Likewise.
33294 (load_from_param): Likewise.
33295 (ipa_load_from_parm_agg): Likewise.
33296 (ipa_node_params::~ipa_node_params): Removed.
33297 (ipa_free_all_node_params): Remove call to delete operator.
33298 (ipa_node_params_t::insert): New.
33299 (ipa_node_params_t::remove): Likewise.
33300 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
33301 copy known_csts and known_contexts vectors.
33302 (ipa_read_node_info): Adjust to descriptors being a pointer.
33303 (ipcp_modif_dom_walker): Make m_descriptors field garbage
33305 (ipcp_transform_function): Make descriptors variable garbage
33308 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
33310 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
33311 * config/i386/avx512dqintrin.h: Ditto.
33312 * config/i386/avx512fintrin.h: Ditto.
33313 * config/i386/i386.c: Handle new builtins.
33314 * config/i386/i386-builtin.def: Add new builtins.
33315 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
33316 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
33318 2017-01-23 Jakub Jelinek <jakub@redhat.com>
33319 Martin Liska <mliska@suse.cz>
33321 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
33322 * asan.c (asan_expand_poison_ifn): Support stores and use
33323 appropriate ASAN report function.
33324 * internal-fn.c (expand_ASAN_POISON_USE): New function.
33325 * internal-fn.def (ASAN_POISON_USE): Declare.
33326 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
33327 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
33328 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
33329 ASAN_POISON calls w/o LHS.
33330 * tree-ssa.c (execute_update_addresses_taken): Create clobber
33331 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
33332 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
33333 * gimplify.c (asan_poison_variables): Add attribute
33334 use_after_scope_memory to variables that really needs to live
33336 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
33337 having the attribute.
33339 2017-01-23 Martin Liska <mliska@suse.cz>
33341 * asan.c (create_asan_shadow_var): New function.
33342 (asan_expand_poison_ifn): Likewise.
33343 * asan.h (asan_expand_poison_ifn): New declaration.
33344 * internal-fn.c (expand_ASAN_POISON): Likewise.
33345 * internal-fn.def (ASAN_POISON): New builtin.
33346 * sanopt.c (pass_sanopt::execute): Expand
33347 asan_expand_poison_ifn.
33348 * tree-inline.c (copy_decl_for_dup_finish): Make function
33350 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
33351 * tree-ssa.c (is_asan_mark_p): New function.
33352 (execute_update_addresses_taken): Rewrite local variables
33353 (identified just by use-after-scope as addressable) into SSA.
33355 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
33357 * doc/install.texi (Specific): opensource.apple.com uses https
33358 now. Remove trailing slash.
33360 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
33362 * README.Portability: Remove note on an Irix compatibility issue.
33364 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
33366 * gcov.c (INCLUDE_ALGORITHM): Define.
33367 (INCLUDE_VECTOR): Define.
33368 No longer include <vector> and <algorithm> directly.
33370 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
33372 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
33374 * doc/invoke.texi (Code Gen Options): Ditto.
33376 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
33379 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
33381 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
33383 rtl-optimization/79125
33384 * cprop.c (local_cprop_pass): Handle cases where we make an
33385 unconditional trap.
33387 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
33391 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
33392 read from, for big endian.
33394 2017-01-20 Jiong Wang <jiong.wang@arm.com>
33396 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
33397 register pauth builtins for LP64 only.
33399 2017-01-20 Marek Polacek <polacek@redhat.com>
33402 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
33405 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
33407 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
33409 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
33410 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
33411 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
33413 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33416 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
33417 in big-endian mode when they are not a single duplicated value.
33419 2017-01-20 Richard Biener <rguenther@suse.de>
33421 * BASE-VER: Bump to 7.0.1.
33423 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
33425 * omp-low.c (omplow_simd_context): New struct. Use it...
33426 (lower_rec_simd_input_clauses): ...here and...
33427 (lower_rec_input_clauses): ...here to hold common data. Adjust all
33428 references to idx, lane, max_vf, is_simt.
33430 2017-01-20 Graham Markall <graham.markall@embecosm.com>
33432 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
33435 2017-01-20 Martin Jambor <mjambor@suse.cz>
33437 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
33438 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
33440 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
33441 (GTFILES): Rename hsa.c to hsa-common.c.
33442 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
33443 * hsa-dump.c: Likewise.
33444 * hsa-gen.c: Likewise.
33445 * hsa-regalloc.c: Likewise.
33446 * ipa-hsa.c: Likewise.
33447 * omp-expand.c: Likewise.
33448 * omp-low.c: Likewise.
33449 * toplev.c: Likewise.
33451 2017-01-20 Marek Polacek <polacek@redhat.com>
33454 * doc/invoke.texi: Document -Wduplicated-branches.
33455 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
33456 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
33457 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
33458 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
33459 return 0 only when not OEP_LEXICOGRAPHIC.
33460 (fold_build_cleanup_point_expr): Use the expression
33461 location when building CLEANUP_POINT_EXPR.
33462 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
33463 * tree.c (add_expr): Handle error_mark_node.
33465 2017-01-20 Martin Liska <mliska@suse.cz>
33468 * tree-profile.c (init_ic_make_global_vars): Do not call
33470 (gimple_init_gcov_profiler): Likewise.
33472 2017-01-20 Martin Liska <mliska@suse.cz>
33475 * cgraph.h (maybe_create_reference): Remove argument and
33477 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
33479 * ipa-cp.c (create_specialized_node): Likewise.
33480 * symtab.c (symtab_node::maybe_create_reference): Handle
33481 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
33483 2017-01-20 Martin Liska <mliska@suse.cz>
33485 * read-rtl-function.c (function_reader::create_function): Use
33486 build_decl instread of build_decl_stat.
33488 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
33490 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
33491 * config/i386/avx512dqintrin.h: Ditto.
33492 * config/i386/avx512fintrin.h: Ditto.
33493 * config/i386/i386-builtin-types.def: Add new types.
33494 * config/i386/i386.c: Handle new types.
33495 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
33496 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
33497 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
33498 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
33499 (__builtin_ia32_kshiftridi): New.
33500 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
33502 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
33506 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
33507 define to rs6000_init_stack_protect_guard.
33508 (rs6000_init_stack_protect_guard): New function.
33510 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
33511 Yunqiang Su <yunqiang.su@imgtec.com>
33513 * config.gcc (supported_defaults): Add madd4.
33514 (with_madd4): Add validation.
33515 (all_defaults): Add madd4.
33516 * config/mips/mips.opt (mmadd4): New option.
33517 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
33519 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
33521 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
33522 (ISA_HAS_FUSED_MADD4): Likewise.
33523 * doc/invoke.texi (-mmadd4): Document the new option.
33524 * doc/install.texi (--with-madd4): Document the new option.
33526 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33528 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
33529 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
33530 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
33531 (aarch64_init_pauth_hint_builtins): New.
33532 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
33533 (aarch64_expand_builtin): Expand new builtins.
33535 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33537 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
33538 * combine-stack-adj.c (no_unhandled_cfa): Handle
33539 REG_CFA_TOGGLE_RA_MANGLE.
33540 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
33541 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
33542 info for return address signing.
33543 (aarch64_expand_epilogue): Likewise.
33545 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33547 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
33548 * config/aarch64/aarch64-protos.h
33549 (aarch64_return_address_signing_enabled): New declaration.
33550 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
33552 (aarch64_expand_prologue): Sign return address before it's pushed onto
33554 (aarch64_expand_epilogue): Authenticate return address fetched from
33556 (aarch64_override_options): Sanity check for ILP32 and ISA level.
33557 (aarch64_attributes): New function attributes for "sign-return-address".
33558 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
33559 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
33560 ("*do_return"): Generate combined instructions according to key index.
33561 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
33562 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
33564 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
33565 * config/aarch64/aarch64.opt (msign-return-address=): New.
33566 * doc/extend.texi (AArch64 Function Attributes): Documents
33567 "sign-return-address=".
33568 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
33570 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
33572 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
33573 overall option summary.
33575 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33577 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
33578 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
33579 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
33580 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
33582 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
33584 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
33585 -mpower9-minmax by default for -mcpu=power9.
33586 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
33587 128-bit floating point.
33589 2017-01-20 Alan Modra <amodra@gmail.com>
33591 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
33592 optimizing for size.
33594 2017-01-20 Alan Modra <amodra@gmail.com>
33597 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
33598 for strcmp and strncmp from corresponding builtin decl.
33600 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
33602 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
33603 instead of i386/rtems-64.h.
33604 * config/i386/rtems-64.h: Remove.
33606 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
33610 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
33612 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
33614 2017-01-19 Tamar Christina <tamar.christina@arm.com>
33616 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
33617 Change int to HOST_WIDE_INT.
33618 * config/aarch64/aarch64-protos.h
33619 (aarch64_simd_gen_const_vector_dup): Likewise.
33620 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
33622 2017-01-19 David Malcolm <dmalcolm@redhat.com>
33624 * langhooks-def.h (lhd_type_for_size): New decl.
33625 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
33626 * langhooks.c (lhd_type_for_size): New function, taken from
33629 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
33631 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
33632 define_bypass for CR latency.
33633 (power9-cracked-alu): Update bypass latency and remove power9-branch.
33634 (power9-alu2): Add define_bypass for CR latency.
33636 (power9-mul): Update insn latency.
33637 (power9-mul-compare): Update insn latency, bypass latency and remove
33640 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33642 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
33644 * config/aarch64/aarch64.md
33645 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
33646 aarch64_nopcrelative_literal_loads.
33647 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
33649 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
33651 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
33652 TARGET_LOONGSON_3A.
33653 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
33655 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
33658 * config.gcc (supported_defaults): Add lxc1-sxc1.
33659 (with_lxc1_sxc1): Add validation.
33660 (all_defaults): Add lxc1-sxc1.
33661 * config/mips/mips.opt (mlxc1-sxc1): New option.
33662 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
33664 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
33665 __mips_no_lxc1_sxc1.
33666 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
33667 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
33668 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
33670 2017-01-19 Richard Biener <rguenther@suse.de>
33672 PR tree-optimization/72488
33673 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
33674 sure to restore SSA info.
33675 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
33677 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
33679 PR rtl-optimization/79121
33680 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
33681 of the inner type when shifting an extended value.
33683 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
33686 * symtab.c (symtab_node::equal_address_to): Fix comparing of
33687 interposable aliases.
33689 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
33692 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
33693 Use the evmergelohi instruction.
33694 (mov_si<mode>_e500_subreg4_2_le): Likewise.
33695 (mov_sitf_e500_subreg8_2_be): Likewise.
33696 (mov_sitf_e500_subreg12_2_le): Likewise.
33697 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
33698 (mov_si<mode>_e500_subreg4_2_be): Likewise.
33699 (mov_sitf_e500_subreg8_2_le): Likewise.
33700 (mov_sitf_e500_subreg12_2_be): Likewise.
33702 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33704 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
33705 attribute from vecsimple to vecperm.
33706 (altivec_vbpermq2): Likewise.
33708 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33711 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
33713 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33714 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
33715 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
33716 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
33717 case where N arg is SIZE_MAX.
33718 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
33719 (cmpstrsi): Add pattern.
33721 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
33723 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
33724 __builtin_vec_revb builtins.
33725 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
33726 built-in functions to support generation of the ISA 3.0 XXBR<x>
33727 vector byte reverse instructions.
33728 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
33729 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
33730 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
33731 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
33732 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
33733 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
33734 (P9V_BUILTIN_VEC_REVB): Likewise.
33735 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
33736 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
33737 (p9_xxbrq_v16qi): Likewise.
33738 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
33739 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
33740 (p9_xxbrh_v8hi): Likewise.
33741 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
33742 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
33743 vec_revb built-in functions.
33745 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
33747 PR rtl-optimization/78952
33748 * config/i386/i386.md (any_extract): New code iterator.
33749 (*insvqi_2): Use any_extract for source operand.
33750 (*insvqi_3): Use any_shiftrt for source operand.
33752 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
33754 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
33756 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
33758 2017-01-18 Matthias Klose <doko@ubuntu.com>
33760 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
33762 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33764 * config/rs6000/altivec.h (vec_bperm): Change #define.
33765 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
33766 (altivec_vbpermq2): New define_insn.
33767 (altivec_vbpermd): Likewise.
33768 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
33769 function interface.
33770 (VBPERMD): Likewise.
33771 (VBPERM): New polymorphic function interface.
33772 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
33773 Add entries for P9V_BUILTIN_VEC_VBPERM.
33774 * doc/extend.texi: Add interfaces for vec_bperm.
33776 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33778 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
33779 first letter of error messages.
33780 (s390_resolve_overloaded_builtin): Likewise.
33781 * config/s390/s390.c (s390_expand_builtin): Likewise.
33782 (s390_invalid_arg_for_unprototyped_fn): Likewise.
33783 (s390_valid_target_attribute_inner_p): Likewise.
33784 * config/s390/s390.md ("tabort"): Likewise.
33786 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
33788 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
33789 (ISA_AVOID_DIV_HILO): New macro.
33790 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
33791 (ISA_HAS_DDIV): Likewise.
33793 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33795 * doc/invoke.texi (fabi-version): Correct number of occurrences.
33797 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33799 * doc/invoke.texi (fabi-version): Spelling fix.
33801 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33804 * doc/invoke.texi (fabi-version): Mention mangling fix for
33807 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33810 * doc/invoke.texi (fabi-version): Document discriminator mangling.
33812 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
33815 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
33816 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
33818 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
33820 (stack_protect_test): Ditto.
33821 * config/rs6000/rs6000.opt (mstack-protector-guard=,
33822 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
33824 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
33825 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
33826 -mstack-protector-guard-offset=.
33827 (RS/6000 and PowerPC Options): Ditto.
33829 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
33831 * config/i386/i386.h (MASK_CLASS_P): New define.
33832 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
33833 there are no registers from different register sets also when
33834 mask registers are used. Update function comment.
33835 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
33836 to (*k/*r) and (*k/*km) alternatives.
33838 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
33840 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
33841 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
33842 (EH_RETURN_HANDLER_RTX): New define.
33843 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
33844 Force frame pointer in EH return functions.
33845 (aarch64_expand_epilogue): Add barrier for eh_return.
33846 (aarch64_final_eh_return_addr): Remove.
33847 (aarch64_eh_return_handler_rtx): New function.
33848 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
33850 (aarch64_eh_return_handler_rtx): New prototype.
33852 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33854 * config/rs6000/altivec.h (vec_rlmi): New #define.
33855 (vec_vrlnm): Likewise.
33856 (vec_rlnm): Likewise.
33857 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
33858 (UNSPEC_VRLNM): Likewise.
33859 (VIlong): New mode iterator.
33860 (altivec_vrl<VI_char>mi): New define_insn.
33861 (altivec_vrl<VI_char>nm): Likewise.
33862 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
33864 (VRLDNM): Likewise.
33865 (RLNM): New polymorphic function entry.
33866 (VRLWMI): New monomorphic function entry.
33867 (VRLDMI): Likewise.
33868 (RLMI): New polymorphic function entry.
33869 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
33870 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
33871 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
33874 2017-01-17 Jakub Jelinek <jakub@redhat.com>
33877 * dwarf2out.c (field_byte_offset): Restore the
33878 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
33879 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
33880 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
33883 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
33886 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
33888 2017-01-17 Jakub Jelinek <jakub@redhat.com>
33891 * dwarf2out.c (add_data_member_location_attribute): For constant
33892 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
33893 instead of DW_AT_data_member_location, DW_AT_bit_offset and
33894 DW_AT_byte_size attributes.
33896 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
33898 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
33899 after forcing to constant memory when the code model is medium.
33901 2017-01-17 Julia Koval <julia.koval@intel.com>
33904 * config/i386/avx512fintrin.h
33905 (_mm512_i32gather_ps): Change __addr type to void const*.
33906 (_mm512_mask_i32gather_ps): Ditto.
33907 (_mm512_i32gather_pd): Ditto.
33908 (_mm512_mask_i32gather_pd): Ditto.
33909 (_mm512_i64gather_ps): Ditto.
33910 (_mm512_mask_i64gather_ps): Ditto.
33911 (_mm512_i64gather_pd): Ditto.
33912 (_mm512_mask_i64gather_pd): Ditto.
33913 (_mm512_i32gather_epi32): Ditto.
33914 (_mm512_mask_i32gather_epi32): Ditto.
33915 (_mm512_i32gather_epi64): Ditto.
33916 (_mm512_mask_i32gather_epi64): Ditto.
33917 (_mm512_i64gather_epi32): Ditto.
33918 (_mm512_mask_i64gather_epi32): Ditto.
33919 (_mm512_i64gather_epi64): Ditto.
33920 (_mm512_mask_i64gather_epi64): Ditto.
33921 (_mm512_i32scatter_ps): Change __addr type to void*.
33922 (_mm512_mask_i32scatter_ps): Ditto.
33923 (_mm512_i32scatter_pd): Ditto.
33924 (_mm512_mask_i32scatter_pd): Ditto.
33925 (_mm512_i64scatter_ps): Ditto.
33926 (_mm512_mask_i64scatter_ps): Ditto.
33927 (_mm512_i64scatter_pd): Ditto.
33928 (_mm512_mask_i64scatter_pd): Ditto.
33929 (_mm512_i32scatter_epi32): Ditto.
33930 (_mm512_mask_i32scatter_epi32): Ditto.
33931 (_mm512_i32scatter_epi64): Ditto.
33932 (_mm512_mask_i32scatter_epi64): Ditto.
33933 (_mm512_i64scatter_epi32): Ditto.
33934 (_mm512_mask_i64scatter_epi32): Ditto.
33935 (_mm512_i64scatter_epi64): Ditto.
33936 (_mm512_mask_i64scatter_epi64): Ditto.
33937 * config/i386/avx512pfintrin.h
33938 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
33939 (_mm512_mask_prefetch_i32gather_ps): Ditto.
33940 (_mm512_mask_prefetch_i64gather_pd): Ditto.
33941 (_mm512_mask_prefetch_i64gather_ps): Ditto.
33942 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
33943 (_mm512_prefetch_i32scatter_ps): Ditto.
33944 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
33945 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
33946 (_mm512_prefetch_i64scatter_pd): Ditto.
33947 (_mm512_prefetch_i64scatter_ps): Ditto.
33948 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
33949 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
33950 * config/i386/avx512vlintrin.h
33951 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
33952 (_mm_mmask_i32gather_ps): Ditto.
33953 (_mm256_mmask_i32gather_pd): Ditto.
33954 (_mm_mmask_i32gather_pd): Ditto.
33955 (_mm256_mmask_i64gather_ps): Ditto.
33956 (_mm_mmask_i64gather_ps): Ditto.
33957 (_mm256_mmask_i64gather_pd): Ditto.
33958 (_mm_mmask_i64gather_pd): Ditto.
33959 (_mm256_mmask_i32gather_epi32): Ditto.
33960 (_mm_mmask_i32gather_epi32): Ditto.
33961 (_mm256_mmask_i32gather_epi64): Ditto.
33962 (_mm_mmask_i32gather_epi64): Ditto.
33963 (_mm256_mmask_i64gather_epi32): Ditto.
33964 (_mm_mmask_i64gather_epi32): Ditto.
33965 (_mm256_mmask_i64gather_epi64): Ditto.
33966 (_mm_mmask_i64gather_epi64): Ditto.
33967 (_mm256_i32scatter_ps): Change __addr type to void*.
33968 (_mm256_mask_i32scatter_ps): Ditto.
33969 (_mm_i32scatter_ps): Ditto.
33970 (_mm_mask_i32scatter_ps): Ditto.
33971 (_mm256_i32scatter_pd): Ditto.
33972 (_mm256_mask_i32scatter_pd): Ditto.
33973 (_mm_i32scatter_pd): Ditto.
33974 (_mm_mask_i32scatter_pd): Ditto.
33975 (_mm256_i64scatter_ps): Ditto.
33976 (_mm256_mask_i64scatter_ps): Ditto.
33977 (_mm_i64scatter_ps): Ditto.
33978 (_mm_mask_i64scatter_ps): Ditto.
33979 (_mm256_i64scatter_pd): Ditto.
33980 (_mm256_mask_i64scatter_pd): Ditto.
33981 (_mm_i64scatter_pd): Ditto.
33982 (_mm_mask_i64scatter_pd): Ditto.
33983 (_mm256_i32scatter_epi32): Ditto.
33984 (_mm256_mask_i32scatter_epi32): Ditto.
33985 (_mm_i32scatter_epi32): Ditto.
33986 (_mm_mask_i32scatter_epi32): Ditto.
33987 (_mm256_i32scatter_epi64): Ditto.
33988 (_mm256_mask_i32scatter_epi64): Ditto.
33989 (_mm_i32scatter_epi64): Ditto.
33990 (_mm_mask_i32scatter_epi64): Ditto.
33991 (_mm256_i64scatter_epi32): Ditto.
33992 (_mm256_mask_i64scatter_epi32): Ditto.
33993 (_mm_i64scatter_epi32): Ditto.
33994 (_mm_mask_i64scatter_epi32): Ditto.
33995 (_mm256_i64scatter_epi64): Ditto.
33996 (_mm256_mask_i64scatter_epi64): Ditto.
33997 (_mm_i64scatter_epi64): Ditto.
33998 (_mm_mask_i64scatter_epi64): Ditto.
33999 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
34000 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
34001 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
34002 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
34003 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
34004 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
34005 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
34006 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
34007 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
34008 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
34009 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
34010 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
34011 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
34012 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
34013 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
34014 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
34015 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
34016 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
34017 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
34018 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
34019 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
34020 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
34021 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
34022 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
34023 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
34024 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
34025 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
34026 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
34027 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
34028 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
34029 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
34030 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
34031 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
34032 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
34033 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
34034 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
34035 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
34036 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
34037 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
34038 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
34039 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
34040 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
34041 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
34042 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
34043 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
34044 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
34045 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
34046 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
34047 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
34048 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
34049 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
34050 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
34051 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
34052 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
34053 definitions accordingly.
34055 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
34056 Kuan-Lin Chen <kuanlinchentw@gmail.com>
34059 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
34062 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
34065 * ira-conflicts.c (ira_build_conflicts): Update total conflict
34066 hard regs for inner regno.
34068 2017-01-17 Martin Liska <mliska@suse.cz>
34071 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
34072 assumption and add comment.
34074 2017-01-17 Nathan Sidwell <nathan@acm.org>
34076 * ipa-visibility.c (localize_node): New function, broken out of ...
34077 (function_and_variable_visibility): ... here. Call it.
34079 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
34081 PR middle-end/77445
34082 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
34083 correctly set frequency of oudgoing edge.
34084 (duplicate_thread_path): Fix profile updating.
34086 2017-01-17 Jakub Jelinek <jakub@redhat.com>
34089 * configure.ac: Add GCC_BASE_VER.
34090 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
34091 version from BASE-VER file.
34092 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
34093 (gcc.o): Depend on $(BASEVER).
34094 * common.opt (dumpfullversion): New option.
34095 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
34096 * doc/invoke.texi: Document -dumpfullversion.
34097 * doc/install.texi: Document --with-gcc-major-version-only.
34098 * configure: Regenerated.
34100 2017-01-17 Richard Biener <rguenther@suse.de>
34102 PR tree-optimization/71433
34103 * tree-vrp.c (register_new_assert_for): Merge same asserts
34104 on all incoming edges.
34105 (process_assert_insertions_for): Handle insertions at the
34108 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
34110 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
34111 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
34113 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
34116 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
34119 2017-01-17 Alan Modra <amodra@gmail.com>
34122 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
34123 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
34124 symbolic stack limit when pic.
34126 2017-01-16 Martin Sebor <msebor@redhat.com>
34128 PR tree-optimization/78608
34129 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
34131 2017-01-16 Jeff Law <law@redhat.com>
34134 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
34135 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
34136 for several include directories that may be relative to sysroot.
34137 * config/i386/x-mingw32 (gplus_includedir): Define.
34138 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
34139 (native_system_includedir): Likewise.
34140 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
34141 override if TARGET_SYSTEM_ROOT is defined.
34142 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
34144 PR tree-optimization/79090
34145 PR tree-optimization/33562
34146 PR tree-optimization/61912
34147 PR tree-optimization/77485
34148 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
34149 and computed trims into the dump file.
34151 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
34153 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
34155 2017-01-16 Jakub Jelinek <jakub@redhat.com>
34158 * gimplify.c (gimplify_init_constructor): If want_value and
34159 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
34163 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
34164 sequence. Formatting fixes.
34165 (doloop_optimize): Formatting fixes.
34168 * gcc.c (debug_level_greater_than_spec_func): New function.
34169 (static_spec_functions): Add debug-level-gt spec function.
34170 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
34172 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
34173 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
34174 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
34175 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
34176 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
34177 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
34179 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
34181 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
34182 QImode fixups to general and mask registers only.
34184 2017-01-16 Carl Love <cel@us.ibm.com>
34186 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
34187 for built-in functions
34188 vector signed char vec_nabs (vector signed char)
34189 vector signed short vec_nabs (vector signed short)
34190 vector signed int vec_nabs (vector signed int)
34191 vector signed long long vec_nabs (vector signed long long)
34192 vector float vec_nabs (vector float)
34193 vector double vec_nabs (vector double)
34194 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
34196 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
34197 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
34198 * doc/extend.texi: Update the documentation file for the new built-in
34201 2017-01-16 Martin Sebor <msebor@redhat.com>
34203 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
34206 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34208 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
34209 UNSPEC_VSX__XXSPLTD to require special splat handling.
34211 2017-01-16 David Malcolm <dmalcolm@redhat.com>
34214 * system.h: Poison strndup.
34216 2017-01-16 Alan Modra <amodra@gmail.com>
34219 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
34222 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
34224 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
34226 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
34228 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
34229 call recog here. Assert that INSN_CODE (insn) is non-negative.
34231 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
34234 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
34236 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
34237 in the currently scheduled RTL fragment.
34239 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
34241 PR rtl-optimization/78751
34242 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
34245 2017-01-14 Jeff Law <law@redhat.com>
34247 PR tree-optimization/79090
34248 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
34249 variable length stores.
34250 (compute_trims): Delete dead assignment to *trim_tail.
34251 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
34254 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
34256 PR rtl-optimization/78626
34257 PR rtl-optimization/78727
34258 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
34259 of a block, and split such blocks after everything else is finished.
34261 2017-01-14 Alan Modra <amodra@gmail.com>
34264 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
34265 target legitimate_combined_insn.
34266 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
34267 (rs6000_legitimate_combined_insn): New function.
34268 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
34270 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
34271 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
34272 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
34274 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
34276 * doc/frontends.texi (G++ and GCC): Remove references to Java.
34278 2017-01-13 Jeff Law <law@redhat.com>
34280 PR tree-optimization/33562
34281 PR tree-optimization/61912
34282 PR tree-optimization/77485
34283 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
34285 (delete_dead_assignment): Likewise.
34286 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
34287 statement to delete_dead_call and delete_dead_assignment.
34289 2017-01-13 David Malcolm <dmalcolm@redhat.com>
34292 * substring-locations.c (format_warning_va): Strengthen case 1 so
34293 that both endpoints of the substring must be within the format
34294 range for just the substring to be printed.
34296 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
34298 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
34299 * config/i386/i386.c (ix86_target_string): Add missing options
34300 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
34301 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
34302 flags_other and ix86_target_other to flags2_other. Display unknown
34304 (ix86_valid_target_attribute_inner_p): Add missing options and
34305 reorder options by implied ISAs, as in ix86_target_string.
34307 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
34309 * hash-table.h (hash_table::too_empty_p): New function.
34310 (hash_table::expand): Use it.
34311 (hash_table::traverse): Likewise.
34312 (hash_table::empty_slot): Use sizeof (value_type) instead of
34313 sizeof (PTR) to convert bytes to elements. Shrink the table
34314 if the current size is excessive for the current number of
34317 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
34319 * ira-costs.c (record_reg_classes): Break from the inner loop
34320 early once alt_fail is known to be true. Update outer loop
34321 handling accordingly.
34323 2017-01-13 Jeff Law <law@redhat.com>
34325 * tree-ssa-dse.c (decrement_count): New function.
34326 (increment_start_addr, maybe_trim_memstar_call): Likewise.
34327 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
34328 when we know the partially dead statement is a mem* function.
34330 PR tree-optimization/61912
34331 PR tree-optimization/77485
34332 * tree-ssa-dse.c: Include expr.h.
34333 (maybe_trim_constructor_store): New function.
34334 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
34336 PR tree-optimization/33562
34337 PR tree-optimization/61912
34338 PR tree-optimization/77485
34339 * doc/invoke.texi: Document new dse-max-object-size param.
34340 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
34341 * tree-ssa-dse.c: Include params.h.
34342 (dse_store_status): New enum.
34343 (initialize_ao_ref_for_dse): New, partially extracted from
34345 (valid_ao_ref_for_dse, normalize_ref): New.
34346 (setup_live_bytes_from_ref, compute_trims): Likewise.
34347 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
34348 (maybe_trim_partially_dead_store): Likewise.
34349 (maybe_trim_complex_store): Likewise.
34350 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
34351 Track what bytes live from the original store. Return tri-state
34352 for dead, partially dead or live.
34353 (dse_dom_walker): Add constructor, destructor and new private members.
34354 (delete_dead_call, delete_dead_assignment): New extracted from
34356 (dse_optimize_stmt): Make a member of dse_dom_walker.
34357 Use initialize_ao_ref_for_dse.
34359 PR tree-optimization/33562
34360 PR tree-optimization/61912
34361 PR tree-optimization/77485
34362 * sbitmap.h (bitmap_count_bits): Prototype.
34363 (bitmap_clear_range, bitmap_set_range): Likewise.
34364 * sbitmap.c (bitmap_clear_range): New function.
34365 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
34367 2017-01-13 Martin Liska <mliska@suse.cz>
34370 * function.c (set_cfun): Add new argument force.
34371 * function.h (set_cfun): Likewise.
34372 * ipa-inline-transform.c (inline_call): Use the function when
34373 strict alising from is dropped for function we inline to.
34375 2017-01-13 Richard Biener <rguenther@suse.de>
34377 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
34378 for dumping GIMPLE INTEGER_CSTs.
34380 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
34382 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
34383 to 201112L since C++17.
34385 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
34388 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
34389 if -fsanitize=kernel-address is present.
34391 2017-01-13 Richard Biener <rguenther@suse.de>
34393 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
34394 as _Literal ( type ) number in case usual suffixes do not
34395 preserve all information.
34397 2017-01-13 Richard Biener <rguenther@suse.de>
34399 PR tree-optimization/77283
34400 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
34401 and ssa-iterators.h.
34402 (is_feasible_trace): Implement a cost model based on joiner
34405 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
34408 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
34409 char or short to __float128/_Float128 directly.
34411 2017-01-12 Martin Sebor <msebor@redhat.com>
34413 to -Wformat-overflow.
34414 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
34415 (min_bytes_remaining): Same.
34416 (get_string_length): Same.
34417 (format_string): Same.
34418 (format_directive): Same.
34420 (pass_sprintf_length::handle_gimple_call): Same.
34422 2017-01-12 Jakub Jelinek <jakub@redhat.com>
34424 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
34425 info.nowrite calls with no lhs that can't throw. Return bool
34426 whether gsi_remove has been called or not.
34427 (pass_sprintf_length::handle_gimple_call): Return bool whether
34428 try_substitute_return_value called gsi_remove. Formatting fix.
34429 (pass_sprintf_length::execute): Don't use gsi_remove if
34430 handle_gimple_call returned true.
34433 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
34434 be removed due to side-effects, don't remove following barrier nor
34435 turn the successor edge into fallthru edge.
34437 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34440 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
34441 element-reversing loads and stores as not swappable.
34443 2017-01-12 Nathan Sidwell <nathan@acm.org>
34444 Nicolai Stange <nicstange@gmail.com>
34446 * combine.c (try_combine): Don't ignore result of overlap checking
34447 loop. Combine overlap & asm check into single loop.
34449 2017-01-12 Richard Biener <rguenther@suse.de>
34451 * tree-pretty-print.c (dump_generic_node): Provide -gimple
34452 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
34454 2017-01-12 Richard Biener <rguenther@suse.de>
34456 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
34457 and TS_TARGET_OPTION directly derive from TS_BASE.
34458 * tree-core.h (tree_optimization_option): Derive from tree_base.
34459 (tree_target_option): Likewise.
34461 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
34463 * config/i386/i386.c (memory_address_length): Increase len
34464 only when rip_relative_addr_p returns false.
34466 2017-01-11 Julia Koval <julia.koval@intel.com>
34468 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
34469 (OPTION_MASK_ISA_SGX_SET): New.
34470 (ix86_handle_option): Handle OPT_msgx.
34471 * config.gcc: Added sgxintrin.h.
34472 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
34473 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
34474 * config/i386/i386.c (ix86_target_string): Add -msgx.
34476 (ix86_option_override_internal): Handle new options.
34477 (ix86_valid_target_attribute_inner_p): Add sgx.
34478 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
34479 * config/i386/i386.opt: Add msgx.
34480 * config/i386/sgxintrin.h: New file.
34481 * config/i386/x86intrin.h: Add sgxintrin.h.
34483 2017-01-11 Jakub Jelinek <jakub@redhat.com>
34486 * fold-const.c (maybe_nonzero_address): Return 1 for function
34488 (tree_single_nonzero_warnv_p): Don't handle function local objects
34492 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
34495 2017-01-11 David Malcolm <dmalcolm@redhat.com>
34498 * opts.c: Include "spellcheck.h"
34499 (struct string_fragment): New struct.
34500 (struct edit_distance_traits<const string_fragment &>): New
34502 (get_closest_sanitizer_option): New function.
34503 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
34505 2017-01-11 Jakub Jelinek <jakub@redhat.com>
34507 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
34509 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
34510 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
34511 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
34512 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
34513 for initial die_offset if dwarf_split_debug_info.
34514 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
34515 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
34517 (output_skeleton_debug_sections): Formatting fix. Use
34518 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
34519 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
34521 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
34523 * config/arm/cortex-a53.md: Add bypasses for
34524 cortex_a53_r2f_cvt.
34525 (cortex_a53_r2f): Only use for transfers.
34526 (cortex_a53_f2r): Likewise.
34527 (cortex_a53_r2f_cvt): Add reservation for conversions.
34528 (cortex_a53_f2r_cvt): Likewise.
34530 2017-01-11 Tamar Christina <tamar.christina@arm.com>
34532 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
34533 to all inlined functions, change static to extern.
34535 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
34538 * config/arm/arm.c (legitimize_pic_address): Handle reference to
34540 (arm_assemble_integer): Likewise.
34542 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
34544 * config.gcc: Use new awk script to check CPU, FPU and architecture
34545 parameters for --with-... options.
34546 * config/arm/parsecpu.awk: New file
34547 * config/arm/arm-cpus.in: New file.
34548 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
34550 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
34552 * config/arm/t-arm: Update dependency rules.
34553 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
34554 of processing .def files.
34555 * config/arm/genopt.sh: Deleted.
34556 * config/arm/gentune.sh: Deleted.
34557 * config/arm/arm-cores.def: Deleted.
34558 * config/arm/arm-arches.def: Deleted.
34559 * config/arm/arm-fpus.def: Deleted.
34560 * config/arm/arm-tune.md: Regenerated.
34561 * config/arm/arm-tables.opt: Regenerated.
34562 * config/arm/arm-cpu.h: New generated file.
34563 * config/arm/arm-cpu-data.h: New generated file.
34564 * config/arm/arm-cpu-cdata.h: New generated file.
34566 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
34569 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
34571 (input_varpool_node): Unpack dynamically_initialized bit.
34573 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
34575 PR rtl-optimization/79032
34576 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
34577 the alignment of the adjusted memory reference against that of MODE,
34578 instead of the alignment of the original memory reference.
34580 2017-01-11 Martin Jambor <mjambor@suse.cz>
34582 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
34584 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
34585 decorated functions.
34587 2017-01-11 Richard Biener <rguenther@suse.de>
34589 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
34590 set range/nonnull info for PHI results. Do not set it on
34591 stmts marked for removal.
34593 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
34595 * expr.c (store_field): In the bitfield case, fetch the return value
34596 from the registers before applying a single big-endian adjustment.
34597 Always do a final load for a BLKmode value not larger than a word.
34599 2017-01-10 David Malcolm <dmalcolm@redhat.com>
34602 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
34603 that we correctly handle column numbers greater than
34604 LINE_MAP_MAX_COLUMN_NUMBER.
34606 2017-01-10 Martin Sebor <msebor@redhat.com>
34608 PR middle-end/78245
34609 * gimple-ssa-sprintf.c (get_destination_size): Call
34610 {init,fini}object_sizes.
34611 * tree-object-size.c (addr_object_size): Adjust.
34612 (pass_through_call): Adjust.
34613 (pass_object_sizes::execute): Adjust.
34614 * tree-object-size.h (fini_object_sizes): Declare.
34616 2017-01-10 Martin Sebor <msebor@redhat.com>
34618 PR tree-optimization/78775
34619 * builtins.c (get_size_range): Move...
34620 * calls.c: ...to here.
34621 (alloc_max_size): Accept zero argument.
34622 (operand_signed_p): Remove.
34623 (maybe_warn_alloc_args_overflow): Call get_size_range.
34624 * calls.h (get_size_range): Declare.
34626 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
34628 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
34629 from TI's devices.csv file as of September 2016.
34630 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
34632 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
34634 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
34635 * doc/invoke.texi: Likewise.
34636 * doc/md.texi: Likewise.
34637 * doc/objc.texi: Likewise.
34639 2017-01-10 Joshua Conner <joshconner@google.com>
34641 * config/arm/fuchsia-elf.h: New file.
34642 * config/fuchsia.h: New file.
34643 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
34644 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
34646 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
34648 2016-01-10 Richard Biener <rguenther@suse.de>
34650 PR tree-optimization/79034
34651 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
34652 Propagate out degenerate PHIs in the joiner.
34654 2017-01-10 Martin Liska <mliska@suse.cz>
34656 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
34657 (sort_congruence_classes_by_decl_uid): Likewise.
34658 (sort_congruence_class_groups_by_decl_uid): Likewise.
34659 (sem_item_optimizer::merge_classes): Sort class, groups in these
34660 classes and members in the groups by DECL_UID of declarations.
34661 This would make merge operations stable.
34663 2017-01-10 Martin Liska <mliska@suse.cz>
34665 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
34666 usage of m_classes_vec.
34667 (sem_item_optimizer::~sem_item_optimizer): Likewise.
34668 (sem_item_optimizer::get_group_by_hash): Likewise.
34669 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
34670 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
34671 (sem_item_optimizer::verify_classes): Likewise.
34672 (sem_item_optimizer::process_cong_reduction): Likewise.
34673 (sem_item_optimizer::dump_cong_classes): Likewise.
34674 (sem_item_optimizer::merge_classes): Likewise.
34675 * ipa-icf.h (congruence_class_hash): Rename from
34676 congruence_class_group_hash. Remove declaration of m_classes_vec.
34678 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
34680 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
34681 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
34682 * config.gcc: Add avx512vpopcntdqintrin.h.
34683 * config/i386/avx512vpopcntdqintrin.h: New.
34684 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
34685 * config/i386/i386-builtin-types.def: Add new types.
34686 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
34687 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
34688 __builtin_ia32_vpopcountq_v8di_mask): New.
34689 * config/i386/i386-c.c (ix86_target_macros_internal): Define
34690 __AVX512VPOPCNTDQ__.
34691 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
34692 (PTA_AVX512VPOPCNTDQ): Define.
34693 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
34694 TARGET_AVX512VPOPCNTDQ_P): Define.
34695 * config/i386/i386.opt: Add mavx512vpopcntdq.
34696 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
34697 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
34699 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34701 PR middle-end/77484
34702 * predict.def (PRED_CALL): Set to 67.
34704 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
34706 * expr.c (store_field): In the bitfield case, if the value comes from
34707 a function call and is of an aggregate type returned in registers, do
34708 not modify the field mode; extract the value in all cases if the mode
34709 is BLKmode and the size is not larger than a word.
34711 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
34714 * config/i386/cpuid.h: Fix undefined behavior.
34716 2017-01-04 Jeff Law <law@redhat.com>
34718 PR tree-optimization/79007
34719 PR tree-optimization/67955
34720 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
34721 conservative for pt.null when flag_non_call_exceptions is on.
34723 2017-01-09 Jakub Jelinek <jakub@redhat.com>
34725 PR translation/79019
34726 PR translation/79020
34727 * params.def (PARAM_INLINE_MIN_SPEEDUP,
34728 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
34729 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
34731 * config/avr/avr.opt (maccumulate-args): Likewise.
34732 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
34733 * common.opt (freport-bug): Likewise.
34734 * cif-code.def (CIF_FINAL_ERROR): Likewise.
34735 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
34736 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
34737 translatable string.
34738 * config/i386/i386.c (function_value_32): Likewise.
34739 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
34740 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
34742 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
34743 * common/config/msp430/msp430-common.c (msp430_handle_option):
34745 * symtab.c (symtab_node::verify_base): Likewise.
34746 * opts.c (set_debug_level): Likewise.
34747 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
34748 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
34749 missing whitespace to translatable strings.
34750 * config/avr/avr.md (bswapsi2): Fix typo in comment.
34751 * config/sh/superh.h: Likewise.
34752 * config/i386/xopintrin.h: Likewise.
34753 * config/i386/znver1.md: Likewise.
34754 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
34755 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
34756 * double-int.h (struct double_int): Likewise.
34757 * double-int.c (div_and_round_double): Likewise.
34758 * wide-int.cc: Likewise.
34759 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
34760 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
34761 * cfgcleanup.c (crossjumps_occured): Renamed to ...
34762 (crossjumps_occurred): ... this.
34763 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
34766 PR tree-optimization/78899
34767 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
34768 returning bool return struct loop *, NULL for failure and the new
34770 (versionable_outer_loop_p): Don't version outer loop if it has
34771 dont_vectorized bit set.
34772 (tree_if_conversion): When versioning outer loop, ensure
34773 tree_if_conversion is performed also on the inner loop of the
34774 non-vectorizable outer loop copy.
34775 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
34776 LOOP_VECTORIZED in inner loop of the scalar outer loop and
34777 prevent vectorization of it.
34778 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
34779 the outer loop vectorization of the non-scalar version is attempted
34780 before vectorization of the inner loop in scalar version. If
34781 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
34782 vectorization of its inner loop.
34783 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
34784 has 2 inner loops, rename also on edges from bb whose single pred
34785 is outer_loop->header. Fix typo in function comment.
34787 2017-01-09 Martin Sebor <msebor@redhat.com>
34790 * asan.c (asan_emit_stack_protection): Increase local buffer size
34791 to avoid snprintf truncation warning.
34793 2017-01-09 Andrew Pinski <apinski@cavium.com>
34795 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
34796 to reference thunderx2t99 for the tuning structure
34797 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
34799 (thunderx2t99_extra_costs): This.
34800 * config/aarch64/aarch64-tune.md: Regenerate.
34801 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
34802 (vulcan_addrcost_table): This.
34803 (vulcan_regmove_cost): Rename to ...
34804 (thunderx2t99_regmove_cost): This.
34805 (vulcan_vector_cost): Rename to ...
34806 (thunderx2t99_vector_cost): this.
34807 (vulcan_branch_cost): Rename to ...
34808 (thunderx2t99_branch_cost): This.
34809 (vulcan_tunings): Rename to ...
34810 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
34811 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
34813 2017-01-09 Martin Jambor <mjambor@suse.cz>
34817 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
34818 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
34819 (propagate_vr_accross_jump_function): Use the above function for all
34820 value range computations for pass-through jump functions and type
34821 converasion from explicit value range values.
34822 (ipcp_propagate_stage): Do not attempt to deduce types of formal
34823 parameters from TYPE_ARG_TYPES.
34824 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
34825 (ipa_write_node_info): Stream type of the actual argument.
34826 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
34828 2017-01-09 Martin Liska <mliska@suse.cz>
34831 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
34832 (lookup_compiler): Do not show error message with have_E.
34834 2017-01-09 Jakub Jelinek <jakub@redhat.com>
34836 PR tree-optimization/78938
34837 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
34838 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
34839 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
34840 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
34843 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34845 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
34848 2017-01-09 Richard Biener <rguenther@suse.de>
34850 PR tree-optimization/78997
34851 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
34852 name condition properly.
34854 2017-01-09 Richard Biener <rguenther@suse.de>
34857 * dwarf2out.c (is_cxx): New overload with context.
34858 (is_naming_typedef_decl): Use it.
34860 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
34862 * invoke.texi (Option Summary): Correct spacing in option lists
34863 and add line breaks to fix over-long lines.
34865 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
34867 PR middle-end/17660
34869 * extend.texi (Common Variable Attributes): Add xref to GCC
34870 Internals manual to explain mode attribute keywords.
34872 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
34875 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
34876 and Preprocessor Options.
34877 (Options for Linking): Document -pthread here....
34878 (RS/6000 and PowerPC Options): ...not here.
34879 (Solaris 2 Options): ...or here.
34880 * doc/cppopts.texi: Document -pthread.
34882 2017-01-08 Martin Sebor <msebor@redhat.com>
34884 PR middle-end/77708
34885 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
34886 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
34887 New member functions.
34888 (format_directive): Used them.
34890 (pass_sprintf_length::handle_gimple_call): Same.
34891 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
34892 to avoid truncation for any argument.
34893 (extract_affine_mul): Same.
34894 * tree.c (get_file_function_name): Same.
34896 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34898 PR middle-end/77484
34899 * predict.def (PRED_INDIR_CALL): Set to 86.
34901 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
34903 PR preprocessor/54124
34904 * doc/cppopts.texi: Reformat -d subtable to list the full name
34905 of the options. Add cross-reference to the docs for the general
34906 compiler -d options.
34907 * doc/invoke.texi (Developer Options): Add cross-reference to the
34908 preprocessor-specific -d option documentation.
34910 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
34912 PR preprocessor/13498
34913 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
34914 redudant material, and reflect new command-line options.
34915 (System Headers): Likewise.
34917 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
34919 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
34920 -isystem, and -idirafter. Copy-edit.
34921 * doc/cppopts.texi: Copy-edit. Remove contradiction about
34922 default for -ftrack-macro-expansion. Delete obsolete and
34923 badly-formatted implementation details about -fdebug-cpp output.
34924 * doc/cppwarnopts.texi: Copy-edit.
34926 2017-01-07 David Malcolm <dmalcolm@redhat.com>
34929 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
34930 that the transition from a max line width >= 1<<10 to narrower
34931 lines works correctly.
34933 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
34935 * doc/options.texi (PerFunction): New.
34936 * opt-functions.awk (switch_flags): Map both Optimization and
34937 PerFunction to CL_OPTIMIZATION.
34938 * opth-gen.awk: Test for PerFunction flag along with
34940 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
34941 it only when the latter is present. Skip those that don't in
34942 the hash function generator.
34943 * common.opt (fvar-tracking): Mark as PerFunction instead of
34945 (fvar-tracking-assignments): Likewise.
34946 (fvar-tracking-assignments-toggle): Likewise.
34947 (fvar-tracking-uninit): Likewise.
34949 2017-01-07 Jakub Jelinek <jakub@redhat.com>
34951 PR translation/79018
34952 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
34955 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
34958 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
34959 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
34960 TARGET_LONG_JUMP_TABLE_OFFSETS.
34961 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
34962 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
34963 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
34964 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
34965 * config/m68k/m68k.md (tablejump expander): Likewise.
34966 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
34967 TARGET_LONG_JUMP_TABLE_OFFSETS.
34968 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
34969 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
34971 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
34972 David Holsgrove <david.holsgrove@xilinx.com>
34974 * common/config/microblaze/microblaze-common.c
34975 (TARGET_EXCEPT_UNWIND_INFO): Remove.
34976 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
34978 * config/microblaze/microblaze.c (microblaze_must_save_register)
34979 (microblaze_expand_epilogue, microblaze_return_addr): Handle
34981 (microblaze_eh_return): New function.
34982 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
34983 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
34984 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
34985 * config/microblaze/microblaze.md (eh_return): New pattern.
34987 2017-01-06 Jakub Jelinek <jakub@redhat.com>
34989 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
34990 GCC_DIAGNOSTIC_STRINGIFY): Define.
34992 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
34994 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34996 * config/arm/arm.md (<mcrr>): New.
34998 * config/arm/arm.c (arm_arch5te): New.
34999 (arm_option_override): Set arm_arch5te.
35000 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
35002 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
35003 (arm_mcrr_qualifiers): ... this. New.
35004 (MRRC_QUALIFIERS): Define to...
35005 (arm_mrrc_qualifiers): ... this. New.
35006 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
35008 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
35009 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
35010 (MRRCI, mrrc, MRRC): New.
35011 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
35012 VUNSPEC_MRRC2): New.
35014 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35016 * config/arm/arm.md (<mcr>): New.
35018 * config/arm/arm.c (arm_coproc_builtin_available): Add
35019 support for mcr, mrc, mcr2 and mrc2.
35020 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
35021 (arm_mcr_qualifiers): ... this. New.
35022 (MRC_QUALIFIERS): Define to ...
35023 (arm_mrc_qualifiers): ... this. New.
35024 (MCR_QUALIFIERS): Define to ...
35025 (arm_mcr_qualifiers): ... this. New.
35026 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
35028 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
35029 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
35030 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
35031 VUNSPEC_MRC2): New.
35033 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35035 * config/arm/arm.md (*ldc): New.
35039 * config/arm/arm.c (arm_coproc_builtin_available): Add
35040 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
35041 (arm_coproc_ldc_stc_legitimate_address): New.
35042 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
35043 'qualifier_const_pointer'.
35044 (LDC_QUALIFIERS): Define to...
35045 (arm_ldc_qualifiers): ... this. New.
35046 (STC_QUALIFIERS): Define to...
35047 (arm_stc_qualifiers): ... this. New.
35048 * config/arm/arm-protos.h
35049 (arm_coproc_ldc_stc_legitimate_address): New.
35050 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
35051 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
35052 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
35053 stc2, stcl, stc2l): New.
35054 * config/arm/constraints.md (Uz): New.
35055 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
35056 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
35057 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
35058 VUNSPEC_STC2L): New.
35060 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35062 * config/arm/arm.md (<cdp>): New.
35063 * config/arm/arm.c (neon_const_bounds): Rename this ...
35064 (arm_const_bounds): ... this.
35065 (arm_coproc_builtin_available): New.
35066 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
35067 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
35068 (CDP_QUALIFIERS): Define to...
35069 (arm_cdp_qualifiers): ... this. New.
35071 (arm_expand_builtin_args): Add case for 6 arguments.
35072 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
35073 (arm_const_bounds): ... this.
35074 (arm_coproc_builtin_available): New.
35075 * config/arm/arm_acle.h (__arm_cdp): New.
35077 * config/arm/arm_acle_builtins.def (cdp): New.
35079 * config/arm/iterators.md (CDPI,CDP,cdp): New.
35080 * config/arm/neon.md: Rename all 'neon_const_bounds' to
35081 'arm_const_bounds'.
35082 * config/arm/types.md (coproc): New.
35083 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
35084 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
35085 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
35086 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
35088 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35090 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
35091 (UBINOP_QUALIFIERS): New.
35093 (acle_builtin_data): New. Change comment.
35094 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
35095 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
35096 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
35097 arm_acle_builtins.def.
35098 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
35099 (arm_init_acle_builtins): New.
35100 (CRC32_BUILTIN): Remove.
35101 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
35102 crc32cb, crc32ch and crc32cw.
35103 (arm_init_crc32_builtins): Remove.
35104 (arm_init_builtins): Use arm_init_acle_builtins rather
35105 than arm_init_crc32_builtins.
35106 (arm_expand_acle_builtin): New.
35107 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
35108 * config/arm/arm_acle_builtins.def: New.
35110 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35112 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
35113 (arm_builtin_datum): ... this.
35114 (arm_init_neon_builtin): Rename to ...
35115 (arm_init_builtin): ... this. Add a new parameters PREFIX
35116 and USE_SIG_IN_NAME.
35117 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
35118 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
35119 'arm_builtin_datum'.
35120 (arm_init_vfp_builtins): Likewise.
35121 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
35122 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
35123 (arm_expand_neon_args): Rename to ...
35124 (arm_expand_builtin_args): ... this. Rename builtin_arg
35125 enum values and differentiate between ARG_BUILTIN_MEMORY
35126 and ARG_BUILTIN_NEON_MEMORY.
35127 (arm_expand_neon_builtin_1): Rename to ...
35128 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
35129 values, arm_expand_builtin_args and add bool parameter NEON.
35130 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
35131 (arm_expand_vfp_builtin): Likewise.
35132 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
35134 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35136 PR middle-end/77484
35137 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
35138 * predict.c (tree_estimate_probability_bb): Reverse direction of
35139 polymorphic call predictor.
35141 2017-01-06 David Malcolm <dmalcolm@redhat.com>
35143 * passes.c (execute_one_pass): Split out pass-skipping logic into...
35144 (determine_pass_name_match): ...this new function and...
35145 (should_skip_pass_p): ...this new function.
35147 2017-01-06 Nathan Sidwell <nathan@acm.org>
35149 * ipa-visibility.c (function_and_variable_visibility): Reformat
35150 comments and long lines. Remove extrneous if.
35151 * symtab.c (symtab_node::make_decl_local): Fix code format.
35152 (symtab_node::set_section_for_node): Fix comment typo.
35154 2017-01-06 Martin Liska <mliska@suse.cz>
35157 * lra-constraints.c: Rename invariant to lra_invariant.
35158 * predict.c (set_even_probabilities): Initialize e to NULL.
35160 2017-01-05 Martin Sebor <msebor@redhat.com>
35162 PR tree-optimization/78910
35163 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
35164 (format_integer): Correct off-by-one error in the handling
35165 of precision with negative numbers in signed conversions..
35167 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
35169 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
35171 2017-01-05 Jakub Jelinek <jakub@redhat.com>
35173 PR tree-optimization/71016
35174 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
35175 factor_out_conditional_conversion. Formatting fix.
35176 (factor_out_conditional_conversion): Add cond_stmt argument.
35177 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
35178 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
35181 2017-01-05 David Malcolm <dmalcolm@redhat.com>
35183 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
35184 read-rtl-function.o, and selftest-rtl.o.
35185 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
35186 (selftest::aarch64_test_loading_full_dump): New function.
35187 (selftest::aarch64_run_selftests): New function.
35188 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
35189 selftest::aarch64_run_selftests.
35190 * config/i386/i386.c
35191 (selftest::ix86_test_loading_dump_fragment_1): New function.
35192 (selftest::ix86_test_loading_call_insn): New function.
35193 (selftest::ix86_test_loading_full_dump): New function.
35194 (selftest::ix86_test_loading_unspec): New function.
35195 (selftest::ix86_run_selftests): Call the new functions.
35196 * emit-rtl.c (maybe_set_max_label_num): New function.
35197 * emit-rtl.h (maybe_set_max_label_num): New decl.
35198 * function.c (instantiate_decls): Guard call to
35199 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
35200 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
35202 * gensupport.c (gen_reader::gen_reader): Pass "false"
35203 for new "compact" param of rtx_reader.
35204 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
35205 rather than an empty string for NULL strings.
35206 * read-md.c: Potentially include config.h rather than bconfig.h.
35207 Wrap include of errors.h with #ifdef GENERATOR_FILE.
35208 (have_error): New global, copied from errors.c.
35209 (md_reader::read_name): Rename to...
35210 (md_reader::read_name_1): ...this, adding "out_loc" param,
35211 and converting "missing name or number" to returning false, rather
35213 (md_reader::read_name): Reimplement in terms of read_name_1.
35214 (md_reader::read_name_or_nil): New function.
35215 (md_reader::read_string): Handle "(nil)" by returning NULL.
35216 (md_reader::md_reader): Add new param "compact".
35217 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
35218 (md_reader::read_file): New method.
35219 * read-md.h (md_reader::md_reader): Add new param "compact".
35220 (md_reader::read_file): New method.
35221 (md_reader::is_compact): New accessor.
35222 (md_reader::read_name): Convert return type from void to file_location.
35223 (md_reader::read_name_or_nil): New decl.
35224 (md_reader::read_name_1): New decl.
35225 (md_reader::m_compact): New field.
35226 (noop_reader::noop_reader): Pass "false" for new "compact" param
35228 (rtx_reader::rtx_reader): Add new "compact" param.
35229 (rtx_reader::read_rtx_operand): Make virtual and convert return
35230 type from void to rtx.
35231 (rtx_reader::read_until): New decl.
35232 (rtx_reader::handle_any_trailing_information): New virtual function.
35233 (rtx_reader::postprocess): New virtual function.
35234 (rtx_reader::finalize_string): New virtual function.
35235 (rtx_reader::m_in_call_function_usage): New field.
35236 (rtx_reader::m_reuse_rtx_by_id): New field.
35237 * read-rtl-function.c: New file.
35238 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
35239 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
35240 (selftest::verify_three_block_rtl_cfg): New decl.
35241 * read-rtl-function.h: New file.
35242 * read-rtl.c: Potentially include config.h rather than bconfig.h.
35243 For host, include function.h, memmodel.h, and emit-rtl.h.
35244 (one_time_initialization): New function.
35245 (struct compact_insn_name): New struct.
35246 (compact_insn_names): New array.
35247 (find_code): Handle insn codes in compact dumps.
35248 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
35249 (bind_subst_iter_and_attr): Likewise.
35250 (add_condition_to_string): Likewise.
35251 (add_condition_to_rtx): Likewise.
35252 (apply_attribute_uses): Likewise.
35253 (add_current_iterators): Likewise.
35254 (apply_iterators): Likewise.
35255 (initialize_iterators): Guard usage of apply_subst_iterator with
35256 #ifdef GENERATOR_FILE.
35257 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
35258 (md_reader::read_mapping): Likewise.
35259 (add_define_attr_for_define_subst): Likewise.
35260 (add_define_subst_attr): Likewise.
35261 (read_subst_mapping): Likewise.
35262 (check_code_iterator): Likewise.
35263 (rtx_reader::read_rtx): Likewise. Move one-time initialization
35265 (one_time_initialization): New function.
35266 (rtx_reader::read_until): New method.
35267 (read_flags): New function.
35268 (parse_reg_note_name): New function.
35269 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
35270 Handle reuse_rtx ids.
35271 Wrap iterator lookup within #ifdef GENERATOR_FILE.
35272 Add parsing support for RTL dumps, mirroring the special-cases in
35273 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
35274 values, and calling handle_any_trailing_information.
35275 (rtx_reader::read_rtx_operand): Convert return type from void
35276 to rtx, returning return_rtx. Handle case 'e'. Call
35277 finalize_string on XSTR and XTMPL fields.
35278 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
35279 "(nil)" values were omitted. Call the postprocess vfunc on the
35281 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
35282 class ctor. Initialize m_in_call_function_usage. Call
35283 one_time_initialization.
35284 * rtl-tests.c (selftest::test_uncond_jump): Call
35285 set_new_first_and_last_insn.
35286 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
35287 * selftest-rtl.c: New file.
35288 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
35289 (selftest::get_insn_by_uid): New decl.
35290 * selftest-run-tests.c (selftest::run_tests): Call
35291 read_rtl_function_c_tests.
35292 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
35293 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
35296 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
35298 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
35299 operands in a special way. Assert that pos+len <= mode precision.
35301 2017-01-05 Jakub Jelinek <jakub@redhat.com>
35303 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
35304 3 argument Alias with unlimited for the negative form.
35305 (fno-vect-cost-model): Removed.
35307 2017-01-05 Martin Liska <mliska@suse.cz>
35309 * hsa-gen.c (gen_hsa_divmod): New function.
35310 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
35312 2017-01-05 Martin Liska <mliska@suse.cz>
35315 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
35318 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35320 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
35321 small constant length operands.
35323 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35325 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
35326 between loop iterations.
35328 2017-01-05 Martin Liska <mliska@suse.cz>
35331 * gimplify.c (gimplify_decl_expr): Compare to
35332 asan_poisoned_variables instread of checking flags.
35333 (gimplify_target_expr): Likewise.
35334 (gimplify_expr): Likewise.
35335 (gimplify_function_tree): Conditionally initialize
35336 asan_poisoned_variables.
35338 2017-01-04 Jeff Law <law@redhat.com>
35340 PR tree-optimizatin/78812
35341 * rtl.h (contains_mem_rtx_p): Prototype.
35342 * ifcvt.c (containts_mem_rtx_p): Move from here to...
35343 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
35344 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
35345 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
35346 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
35348 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35350 * input.c (assert_char_at_range): Default-initialize actual_range.
35352 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35354 * df-scan.c (df_ref_create_structure): Make regno unsigned,
35355 to match the caller.
35357 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35359 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
35360 insns after final jump in test to emit dummy move.
35362 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35364 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
35365 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
35367 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35369 * multiple_target.c (create_dispatcher_calls): Init e_next.
35370 * tree-ssa-loop-split.c (split_loop): Init border.
35371 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
35374 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
35379 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
35380 (altivec_register_operand): Do not return true if the operand
35381 contains a SUBREG mixing SImode and SFmode.
35382 (vsx_register_operand): Likewise.
35383 (vsx_reg_sfsubreg_ok): New predicate.
35384 (vfloat_operand): Do not return true if the operand contains a
35385 SUBREG mixing SImode and SFmode.
35386 (vint_operand): Likewise.
35387 (vlogical_operand): Likewise.
35388 (gpc_reg_operand): Likewise.
35389 (int_reg_operand): Likewise.
35390 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
35391 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
35392 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
35394 (rs6000_emit_move_si_sf_subreg): New helper function.
35395 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
35396 fixup SUBREGs involving SImode and SFmode.
35397 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
35398 numbers for the new peephole2 optimization.
35399 (peephole2 for SFmode unions): New peephole2 to optimize cases in
35400 the GLIBC math library that do AND/IOR/XOR operations on single
35401 precision floating point.
35402 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
35403 target macros to say whether we need to avoid SUBREGs mixing
35405 (TARGET_ALLOW_SF_SUBREG): Likewise.
35406 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
35407 (UNSPEC_SI_FROM_SF): Likewise.
35408 (iorxor): Change spacing.
35409 (and_ior_xor): New iterator for AND, IOR, and XOR.
35410 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
35411 (movdi_from_sf_zero_ext): Likewise.
35412 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
35413 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
35414 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
35415 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
35416 (fms<mode>4): Likewise.
35417 (fnma<mode>4): Likewise.
35418 (fnms<mode>4): Likewise.
35419 (nfma<mode>4): Likewise.
35420 (nfms<mode>4): Likewise.
35422 2017-01-04 Marek Polacek <polacek@redhat.com>
35425 * doc/invoke.texi: Document -Wpointer-compare.
35427 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35429 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
35432 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
35433 descriptions for -gdwarf-5 and emit them as uleb128 instead of
35436 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
35439 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
35440 documentation of the powerpc_popcntb_ok attribute.
35441 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
35442 code to issue warning messages if a requested CPU configuration is
35443 not supported by the binary (assembler and loader) toolchain.
35444 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
35445 made to define a built-in function that has been disabled.
35446 (paired_init_builtins): Add assertion to prevent ICE if attempt is
35447 made to define a built-in function that has been disabled.
35448 (altivec_init_builtins): Add comment explaining why definition
35449 of the DST built-in functions is not preceded by an assertion
35450 check. Add assertions to prevent ICE if attempts are made to
35451 define an altivec predicate or an abs* built-in function that has
35453 (htm_init_builtins): Add comment explaining why definition of the
35454 htm built-in functions is not preceded by an assertion check.
35456 2017-01-04 Jeff Law <law@redhat.com>
35458 PR tree-optimizatin/67955
35459 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
35460 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
35461 the points-to solution does not include pt_null. Use DECL_PT_UID
35464 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
35466 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
35467 Use gen_int_mode instead of gen_lopwart for const_int operands.
35469 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35471 PR tree-optimization/71563
35472 * match.pd: Simplify X << Y into X if Y is known to be 0 or
35473 out of range value - has low bits known to be zero.
35475 2017-01-04 Alan Modra <amodra@gmail.com>
35477 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
35478 * configure: Regenerate.
35479 * config.in: Regenerate.
35481 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35484 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
35485 a substring of the message, but strcmp with the whole message. Ifdef
35486 ENABLE_NLS, translate the message first using dgettext.
35488 2017-01-03 Jeff Law <law@redhat.com>
35490 PR tree-optimizatin/78856
35491 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
35492 (mark_threaded_blocks): Remove code to truncate thread paths that
35493 cross multiple loop headers. Instead invalidate the cached loop
35494 iteration information and handle case of a thread path walking
35495 into an irreducible region.
35497 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
35500 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
35501 assertions. Add support for doing the signbit if the IEEE 128-bit
35502 floating point value is in a GPR.
35503 * config/rs6000/rs6000.md (Fsignbit): Delete.
35504 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
35505 Update the length attribute if the value is in a GPR.
35506 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
35507 the sign or zero extension instruction, since the value is always 0/1.
35508 (signbit<mode>2_dm2): Delete using <Fsignbit>.
35511 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
35512 extracting SImode to a GPR register so that we can generate a
35513 store, limit the vector to be in a traditional Altivec register
35514 for the vextuwrx instruction.
35516 2017-01-03 Ian Lance Taylor <iant@google.com>
35518 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
35520 2017-01-03 Martin Sebor <msebor@redhat.com>
35522 PR tree-optimization/78696
35523 * gimple-ssa-sprintf.c (format_floating): Correct handling of
35524 precision. Use MPFR for %f for greater fidelity. Correct handling
35526 (pass_sprintf_length::compute_format_length): Set width and precision
35527 specified by asrerisk to void_node for vararg functions.
35528 (try_substitute_return_value): Adjust dump output.
35530 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
35532 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
35534 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
35536 * doc/invoke.texi (SPARC options): Document -mlra as the default.
35537 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
35538 -mlra/-mno-lra was passed to the compiler.
35540 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
35542 PR rtl-optimization/65618
35543 * emit-rtl.c (try_split): Move initialization of "before" and
35544 "after" to just before the call to emit_insn_after_setloc.
35546 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
35548 * doc/md.texi (Standard Names): Remove reference to Java frontend.
35550 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
35552 * dwarf2out.c (gen_enumeration_type_die): When
35553 -gno-strict-dwarf, add a DW_AT_encoding attribute.
35555 2017-01-03 Jakub Jelinek <jakub@redhat.com>
35557 PR tree-optimization/78965
35558 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
35559 Change first argument from const call_info & to call_info &. For %n
35560 set info.nowrite to false.
35562 PR middle-end/78901
35563 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
35564 possibly throwing calls.
35566 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
35567 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
35568 and fns handling, rather than in a separate case SSA_NAME.
35570 2017-01-02 Jeff Law <law@redhat.com>
35572 * config/darwin-driver.c (darwin_driver_init): Const-correctness
35573 fixes for first_period and second_period variables.
35575 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
35578 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
35579 (*insvqi_1): New insn pattern.
35580 (*insvqi_1_mem_rex64): Ditto.
35581 (*insvqi_2): Ditto.
35582 (*insvqi_3): Rename from *insvqi.
35584 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
35586 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
35588 * doc/cfg.texi (Edges): Remove reference to Java.
35589 (Maintaining the CFG): Ditto.
35591 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35593 PR middle-end/77674
35594 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
35595 transparent aliases.
35597 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35599 PR middle-end/77484
35600 * predict.def (PRED_CALL): Update hitrate.
35601 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
35602 * predict.c (tree_estimate_probability_bb): Split CALL predictor
35603 into direct/indirect/polymorphic variants.
35605 2017-01-01 Jakub Jelinek <jakub@redhat.com>
35607 Update copyright years.
35609 * gcc.c (process_command): Update copyright notice dates.
35610 * gcov-dump.c (print_version): Ditto.
35611 * gcov.c (print_version): Ditto.
35612 * gcov-tool.c (print_version): Ditto.
35613 * gengtype.c (create_file): Ditto.
35614 * doc/cpp.texi: Bump @copying's copyright year.
35615 * doc/cppinternals.texi: Ditto.
35616 * doc/gcc.texi: Ditto.
35617 * doc/gccint.texi: Ditto.
35618 * doc/gcov.texi: Ditto.
35619 * doc/install.texi: Ditto.
35620 * doc/invoke.texi: Ditto.
35622 Copyright (C) 2017 Free Software Foundation, Inc.
35624 Copying and distribution of this file, with or without modification,
35625 are permitted in any medium without royalty provided the copyright
35626 notice and this notice are preserved.