1 2017-09-19 Jeff Law <law@redhat.com>
3 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
4 nothing for stack adjustments with REG_STACK_CHECK.
5 * sched-deps.c (parse_add_or_inc): Reject insns with
6 REG_STACK_CHECK from dependency breaking.
7 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
8 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
9 * reg-notes.def (STACK_CHECK): New note.
11 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
12 (ix86_expand_prologue): Dump stack clash info as needed.
13 Call ix86_adjust_stack_and_probe_stack_clash as needed.
15 * function.c (dump_stack_clash_frame_info): New function.
16 * function.h (dump_stack_clash_frame_info): Prototype.
17 (enum stack_clash_probes): New enum.
19 * config/alpha/alpha.c (alpha_expand_prologue): Also check
20 flag_stack_clash_protection.
21 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
22 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
23 (arm_frame_pointer_required): Likewise.
24 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
25 (ia64_expand_prologue): Likewise.
26 * config/mips/mips.c (mips_expand_prologue): Likewise.
27 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
28 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
29 (sparc_flat_expand_prologue): Likewise.
30 * config/spu/spu.c (spu_expand_prologue): Likewise.
32 * explow.c: Include "params.h".
33 (anti_adjust_stack_and_probe_stack_clash): New function.
34 (get_stack_check_protect): Likewise.
35 (compute_stack_clash_protection_loop_data): Likewise.
36 (emit_stack_clash_protection_loop_start): Likewise.
37 (emit_stack_clash_protection_loop_end): Likewise.
38 (allocate_dynamic_stack_space): Use get_stack_check_protect.
39 Use anti_adjust_stack_and_probe_stack_clash.
40 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
41 (emit_stack_clash_protection_loop_start): Likewise.
42 (emit_stack_clash_protection_loop_end): Likewise.
43 * rtl.h (get_stack_check_protect): Prototype.
44 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
45 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
46 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
48 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
50 * doc/tm.texi: Rebuilt.
51 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
52 get_stack_check_protect.
53 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
54 * config/arm/arm.c (arm_expand_prologue): Likewise.
55 (arm_frame_pointer_required): Likewise.
56 * config/i386/i386.c (ix86_expand_prologue): Likewise.
57 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
58 * config/mips/mips.c (mips_expand_prologue): Likewise.
59 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
60 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
61 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
62 (sparc_flat_expand_prologue): Likewise.
64 * common.opt (-fstack-clash-protection): New option.
65 * flag-types.h (enum stack_check_type): Note difference between
66 -fstack-check= and -fstack-clash-protection.
67 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
68 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
69 * toplev.c (process_options): Issue warnings/errors for cases
70 not handled with -fstack-clash-protection.
71 * doc/invoke.texi (-fstack-clash-protection): Document new option.
72 (-fstack-check): Note additional problem with -fstack-check=generic.
73 Note that -fstack-check is primarily for Ada and refer users
74 to -fstack-clash-protection for stack-clash-protection.
75 Document new params for stack clash protection.
77 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
79 * config/i386/i386.md (*scc_bt<mode>): New insn_and_split pattern.
80 (*scc_bt<mode>_1): Ditto.
81 (*scc_bt<mode>_mask): Ditto.
83 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
85 * config/i386/i386.c (ix86_split_long_move): Do not handle
86 address used for LEA in a special way.
88 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
90 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
92 2017-09-19 Martin Sebor <msebor@redhat.com>
95 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
96 of incompatible types.
98 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
100 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
101 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
102 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
103 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
105 2017-09-19 Richard Biener <rguenther@suse.de>
107 PR tree-optimization/82244
108 * tree-vrp.c (remove_range_assertions): Do not propagate
109 a constant to abnormals but replace the assert with a copy.
111 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
113 PR rtl-optimization/57878
114 PR rtl-optimization/68988
115 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
116 avoidance test involving non_reload_pseudos. Move frequency test
117 below the general fragmentation avoidance test.
119 2017-09-19 Richard Biener <rguenther@suse.de>
121 PR tree-optimization/69728
122 * graphite-sese-to-poly.c (schedule_error): New global.
123 (add_loop_schedule): Handle empty domain by failing the
125 (build_original_schedule): Handle schedule_error.
127 2017-09-19 Richard Biener <rguenther@suse.de>
129 * graphite-scop-detection.c (scop_detection::can_represent_loop):
130 Do not iterate to sibling loops but only to siblings of inner
133 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
136 * config/m68k/m68k.md (moveq feeding equality comparison): Check
137 that the registers are different.
139 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
141 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
142 to processor_model and "amdfam17h" to arch_names_table.
143 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
145 2017-09-18 Jakub Jelinek <jakub@redhat.com>
148 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
150 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
151 Alan Hayward <alan.hayward@arm.com>
152 David Sherwood <david.sherwood@arm.com>
154 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
156 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
158 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
159 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
160 vect_schedule_slp_instance.
161 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
162 Update call to vect_slp_analyze_node_operations. Simplify return
164 (vect_slp_analyze_bb_1): Update call accordingly.
165 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
166 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
167 (vect_schedule_slp): Update call accordingly.
169 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
170 Alan Hayward <alan.hayward@arm.com>
171 David Sherwood <david.sherwood@arm.com>
173 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
174 with types that aren't in fact scalar.
176 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
178 * tree-vect-slp.c (vect_record_max_nunits): New function,
180 (vect_build_slp_tree_1): ...here.
181 (vect_build_slp_tree_2): Call it for phis too.
183 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
185 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
186 to vect_get_vec_def_for_operand when getting the mask operand.
188 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
189 Alan Hayward <alan.hayward@arm.com>
190 David Sherwood <david.sherwood@arm.com>
192 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
195 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
196 Alan Hayward <alan.hayward@arm.com>
197 David Sherwood <david.sherwood@arm.com>
199 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
200 calculation for vector booleans.
202 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
203 Alan Hayward <alan.hayward@arm.com>
204 David Sherwood <david.sherwood@arm.com>
206 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
208 (vect_transform_stmt): ...here.
209 (vect_analyze_stmt): Use it instead of calling
210 vectorizable_live_operation directly.
212 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
214 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
215 non-SIMT targets in acc vector loops.
217 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
219 * configure.ac: Add arc and check if assembler supports gdwarf2.
220 * configure: Regenerate.
222 2017-09-18 Richard Biener <rguenther@suse.de>
224 PR tree-optimization/82220
225 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
226 epilogue niters from the min_profitable_iters compute.
228 2017-09-18 Jakub Jelinek <jakub@redhat.com>
231 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
232 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
233 (ix86_init_pic_reg): Revert 2017-09-01 changes.
235 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
238 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
239 switching to a new text section.
241 2017-09-18 Richard Biener <rguenther@suse.de>
243 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
245 (build_alias_set): Reject aliases with no access function.
247 2017-09-18 Richard Biener <rguenther@suse.de>
249 PR tree-optimization/79622
250 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
252 (build_cross_bb_scalars_use): Likewise.
254 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
256 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
258 2017-09-18 Alan Modra <amodra@gmail.com>
261 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
262 stack_pointer_rtx for count 0. Update comments. Break up
263 large rtl expression.
265 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
267 config/i386/i386.c: (xlogue_layout::STUB_NAME_MAX_LEN): Increase to 20
269 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
270 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
271 or avx version of the stub.
273 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
276 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
277 compute the minimum stack alignment. Also update preferred stack
278 boundary for leaf functions.
280 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
282 PR tree-optimization/82228
283 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
286 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
288 * common/config/nds32/nds32-common.c
289 (nds32_option_optimization_table): Refine formatting.
290 (nds32_option_optimization_table): Use -fsched-pressure and
291 -fomit-frame-pointer for specific optimization level.
293 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
295 * config/nds32/nds32.c: Refine formatting and comments.
296 * config/nds32/nds32.h: Likewise.
297 * config/nds32/nds32.md: Likewise.
298 * config/nds32/nds32-cost.c: Likewise.
299 * config/nds32/nds32-isr.c: Likewise.
300 * config/nds32/nds32-md-auxiliary.c: Likewise.
301 * config/nds32/nds32-multiple.md: Likewise.
302 * config/nds32/nds32-predicates.c: Likewise.
304 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
305 Jakub Jelinek <jakub@redhat.com>
307 Add support for -std=c++2a.
308 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
310 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
312 2017-09-15 Steve Ellcey <sellcey@cavium.com>
315 * doc/extend.texi (Common Function Attributes): Add
316 references to ARM, AArch64, and S/390 specific attributes.
317 (Function Specific Option Pragmas): Add AArch64 and S/390
318 to list of back ends that support the target pragma.
320 2017-09-15 Nathan Sidwell <nathan@acm.org>
322 * doc/standards.texi: Fix C++17 description. Update URLs for
325 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
327 * common.opt (Wcast-align=strict): New warning option.
328 * doc/invoke.texi: Document -Wcast-align=strict.
330 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
332 * cgraph.h (cgraph_thunk_info): Add comments.
333 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
334 assert for VIRTUAL_* arguments stricter.
336 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
338 PR tree-optimization/71026
339 * match.pd: Move RDIV patterns from fold-const.c
340 * fold-const.c (distribute_real_division): Removed.
341 (fold_binary_loc): Remove calls to distribute_real_divison.
343 2017-09-15 Jakub Jelinek <jakub@redhat.com>
345 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
346 c++1z and gnu++1z as deprecated. Change other references to
347 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
348 Change -Wc++1z-compat to -Wc++17-compat.
349 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
350 * dwarf2out.c (highest_c_language): Handle C++17.
351 (gen_compile_unit_die): Likewise.
353 2017-09-15 Jakub Jelinek <jakub@redhat.com>
355 PR rtl-optimization/82192
356 * combine.c (make_extraction): Don't look through non-paradoxical
357 SUBREGs or TRUNCATE if pos + len is or might be bigger than
360 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
361 Alan Hayward <alan.hayward@arm.com>
362 David Sherwood <david.sherwood@arm.com>
364 * target.def (function_arg_offset): New hook.
365 * targhooks.h (default_function_arg_offset): Declare.
366 * targhooks.c (default_function_arg_offset): New function.
367 * function.c (locate_and_pad_parm): Use
368 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
369 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
370 (TARGET_FUNCTION_ARG_OFFSET): ...this.
371 * doc/tm.texi: Regenerate.
372 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
373 * config/spu/spu.c (spu_function_arg_offset): New function.
374 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
375 * system.h (FUNCTION_ARG_OFFSET): Poison.
377 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
378 Alan Hayard <alan.hayward@arm.com>
379 David Sherwood <david.sherwood@arm.com>
381 * target.def (truly_noop_truncation): New hook.
382 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
383 than TRULY_NOOP_TRUNCATION.
384 * hooks.h (hook_bool_uint_uint_true): Declare.
385 * hooks.c (hook_bool_uint_uint_true): New function.
386 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
387 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
388 * doc/tm.texi: Regenerate.
389 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
390 rather than TRULY_NOOP_TRUNCATION in comments.
391 (simplify_comparison): Likewise.
392 (record_truncated_value): Likewise.
393 * expmed.c (extract_bit_field_1): Likewise.
394 (extract_split_bit_field): Likewise.
395 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
396 instead of TRULY_NOOP_TRUNCATION.
397 * function.c (assign_parm_setup_block): Likewise.
398 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
399 * rtlhooks.c: Include target.h.
400 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
401 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
402 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
403 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
404 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
405 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
406 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
407 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
408 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
409 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
410 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
411 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
412 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
413 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
414 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
415 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
416 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
417 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
418 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
419 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
420 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
421 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
422 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
423 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
424 * config/mips/mips.c (mips_truly_noop_truncation): New function.
425 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
426 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
427 rather than TRULY_NOOP_TRUNCATION in comments.
428 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
429 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
430 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
431 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
432 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
433 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
434 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
435 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
436 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
437 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
438 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
439 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
440 rather than TRULY_NOOP_TRUNCATION in comments.
441 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
442 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
443 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
444 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
445 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
446 TRULY_NOOP_TRUNCATION condition.
447 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
448 (TRULY_NOOP_TRUNCATION): Delete.
449 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
450 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
451 * config/spu/spu.c (spu_truly_noop_truncation): New function.
452 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
453 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
454 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
455 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
456 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
457 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
458 rather than TRULY_NOOP_TRUNCATION in comments.
459 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
460 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
461 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
462 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
463 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
464 * system.h (TRULY_NOOP_TRUNCATION): Poison.
466 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
469 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
470 (*cmp_ior): Likewise.
471 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
472 (*ior_scc_scc_cmp): Likewise.
473 (*and_scc_scc): Likewise.
474 (*and_scc_scc_cmp): Likewise.
476 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
477 Alan Hayard <alan.hayward@arm.com>
478 David Sherwood <david.sherwood@arm.com>
480 * target.def (can_change_mode_class): New hook.
481 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
482 (hard_regno_nregs): Likewise.
483 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
484 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
485 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
486 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
487 (LOAD_EXTEND_OP): Update accordingly.
488 * doc/tm.texi: Regenerate.
489 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
490 CANNOT_CHANGE_MODE_CLASS.
491 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
492 (REG_CAN_CHANGE_MODE_P): ...this new macro.
493 * combine.c (simplify_set): Update accordingly.
494 * emit-rtl.c (validate_subreg): Likewise.
495 * recog.c (general_operand): Likewise.
496 * regcprop.c (mode_change_ok): Likewise.
497 * reload1.c (choose_reload_regs): Likewise.
498 (inherit_piecemeal_p): Likewise.
499 * rtlanal.c (simplify_subreg_regno): Likewise.
500 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
501 instead of CANNOT_CHANGE_MODE_CLASS.
502 (reload_cse_simplify_operands): Likewise.
503 * reload.c (push_reload): Use targetm.can_change_mode_class
504 instead of CANNOT_CHANGE_MODE_CLASS.
505 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
506 REG_CANNOT_CHANGE_MODE_P.
507 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
508 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
509 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
510 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
511 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
512 (arm_can_change_mode_class): New function.
513 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
514 than CANNOT_CHANGE_MODE_CLASS in comments.
515 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
516 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
517 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
518 (ix86_can_change_mode_class): ...this new function, inverting the
519 sense of the return value.
520 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
521 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
522 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
523 (ia64_can_change_mode_class): New function.
524 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
525 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
526 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
527 (m32c_can_change_mode_class): ...this new function, inverting the
528 sense of the return value.
529 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
530 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
531 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
532 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
533 (mips_can_change_mode_class): ...this new function, inverting the
534 sense of the return value.
535 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
536 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
537 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
538 (msp430_can_change_mode_class): New function.
539 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
540 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
541 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
542 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
543 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
544 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
545 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
546 (pa_cannot_change_mode_class): Replace with...
547 (pa_can_change_mode_class): ...this new function, inverting the
548 sense of the return value.
549 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
550 than CANNOT_CHANGE_MODE_CLASS in comments.
551 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
552 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
553 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
554 (pdp11_cannot_change_mode_class): Replace with...
555 (pdp11_can_change_mode_class): ...this new function, inverting the
556 sense of the return value.
557 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
558 * config/powerpcspe/powerpcspe-protos.h
559 (rs6000_cannot_change_mode_class_ptr): Delete.
560 * config/powerpcspe/powerpcspe.c
561 (rs6000_cannot_change_mode_class_ptr): Delete.
562 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
563 (rs6000_option_override_internal): Assign to
564 targetm.can_change_mode_class instead of
565 rs6000_cannot_change_mode_class_ptr.
566 (rs6000_cannot_change_mode_class): Replace with...
567 (rs6000_can_change_mode_class): ...this new function, inverting the
568 sense of the return value.
569 (rs6000_debug_cannot_change_mode_class): Replace with...
570 (rs6000_debug_can_change_mode_class): ...this new function.
571 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
572 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
573 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
574 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
575 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
577 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
578 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
579 (rs6000_option_override_internal): Assign to
580 targetm.can_change_mode_class instead of
581 rs6000_cannot_change_mode_class_ptr.
582 (rs6000_cannot_change_mode_class): Replace with...
583 (rs6000_can_change_mode_class): ...this new function, inverting the
584 sense of the return value.
585 (rs6000_debug_cannot_change_mode_class): Replace with...
586 (rs6000_debug_can_change_mode_class): ...this new function.
587 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
588 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
589 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
590 (s390_can_change_mode_class): ...this new function, inverting the
591 sense of the return value.
592 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
593 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
594 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
595 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
596 (sh_cannot_change_mode_class): Replace with...
597 (sh_can_change_mode_class): ...this new function, inverting the
598 sense of the return value.
599 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
600 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
601 (sparc_can_change_mode_class): New function.
602 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
603 * config/spu/spu.c (spu_can_change_mode_class): New function.
604 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
605 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
606 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
607 (visium_can_change_mode_class): New function.
608 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
610 2017-09-15 Richard Biener <rguenther@suse.de>
612 PR tree-optimization/82217
613 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
614 but not undefined case.
616 2017-09-15 Jakub Jelinek <jakub@redhat.com>
619 * postreload.c (reload_cse_simplify_operands): Skip
620 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
622 2017-09-15 Richard Biener <rguenther@suse.de>
624 PR tree-optimization/68823
625 * graphite-scop-detection.c (build_alias_set): If we have a
626 possible dependence check whether we can handle them by just
627 looking at the DRs DR_ACCESS_FNs.
628 (build_scops): If build_alias_set fails, fail the SCOP.
630 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
632 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
633 to support float128 built-in functions that require the ISA 3.0
635 (BU_FLOAT128_3_HW): Likewise.
636 (SQRTF128): Add support for the IEEE 128-bit square root and fma
640 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
641 support for built-in functions that need the ISA 3.0 IEEE 128-bit
642 floating point instructions.
643 (rs6000_invalid_builtin): Likewise.
644 (rs6000_builtin_mask_names): Likewise.
645 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
646 (RS6000_BTM_FLOAT128_HW): Likewise.
647 (RS6000_BTM_COMMON): Likewise.
648 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
650 * doc/extend.texi (RS/6000 built-in functions): Document the
651 IEEE 128-bit floating point square root and fused multiply-add
654 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
656 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
657 reg (r2) isn't in the set of registers defined in the prologue.
659 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
660 Alan Hayward <alan.hayward@arm.com>
661 David Sherwood <david.sherwood@arm.com>
663 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
664 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
665 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
666 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
667 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
669 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
670 max_vectorization_factor.
671 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
673 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
674 Alan Hayward <alan.hayward@arm.com>
675 David Sherwood <david.sherwood@arm.com>
677 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
678 (vect_worthwhile_without_simd_p): Declare.
679 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
680 (vectorizable_reduction): Use it.
681 * tree-vect-stmts.c (vectorizable_shift): Likewise.
682 (vectorizable_operation): Likewise.
684 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
685 Alan Hayward <alan.hayward@arm.com>
686 David Sherwood <david.sherwood@arm.com>
688 * tree-vectorizer.h (vect_get_num_copies): New function.
689 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
690 * tree-vect-loop.c (vectorizable_reduction): Likewise.
691 (vectorizable_induction): Likewise.
692 (vectorizable_live_operation): Likewise.
693 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
694 (vectorizable_bswap): Likewise.
695 (vectorizable_call): Likewise.
696 (vectorizable_conversion): Likewise.
697 (vectorizable_assignment): Likewise.
698 (vectorizable_shift): Likewise.
699 (vectorizable_operation): Likewise.
700 (vectorizable_store): Likewise.
701 (vectorizable_load): Likewise.
702 (vectorizable_condition): Likewise.
703 (vectorizable_comparison): Likewise.
704 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
706 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
707 Alan Hayward <alan.hayward@arm.com>
708 David Sherwood <david.sherwood@arm.com>
710 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
713 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
714 Alan Hayward <alan.hayward@arm.com>
715 David Sherwood <david.sherwood@arm.com>
717 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
718 an inline wrapper that provides a location.
719 (gimple_build_vector): Likewise.
720 * gimple-fold.c (gimple_build_vector_from_val): New function.
721 (gimple_build_vector): Likewise.
722 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
723 functions to build the initial value. Always return a gimple value.
724 (get_initial_defs_for_reduction): Likewise. Only compute
726 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
727 vect_init_vector on the results from get_initial_def(s)_for_reduction.
728 (vectorizable_induction): Use gimple_build_vector rather than
731 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
732 Alan Hayward <alan.hayward@arm.com>
733 David Sherwood <david.sherwood@arm.com>
735 * target.h (vec_perm_indices): New typedef.
736 (auto_vec_perm_indices): Likewise.
737 * optabs-query.h: Include target.h
738 (can_vec_perm_p): Take a vec_perm_indices *.
739 * optabs-query.c (can_vec_perm_p): Likewise.
740 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
741 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
742 * tree-vect-generic.c (lower_vec_perm): Likewise.
743 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
744 (vect_grouped_load_supported): Likewise.
745 (vect_shift_permute_load_chain): Likewise.
746 (vect_permute_store_chain): Use auto_vec_perm_indices.
747 (vect_permute_load_chain): Likewise.
748 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
749 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
750 Update uses of can_vec_perm_p.
751 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
752 mode with a number of elements. Take a vec_perm_indices *.
753 (vect_create_epilog_for_reduction): Update accordingly.
754 Use auto_vec_perm_indices.
755 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
756 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
757 (vect_transform_slp_perm_load): Likewise.
758 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
759 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
760 (vect_gen_perm_mask_checked): Likewise.
761 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
762 (vect_gen_perm_mask_checked): Likewise.
763 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
764 (vectorizable_store): Likewise.
765 (vectorizable_load): Likewise.
766 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
767 (vectorizable_bswap): Likewise.
769 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
770 Alan Hayward <alan.hayward@arm.com>
771 David Sherwood <david.sherwood@arm.com>
773 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
774 * tree.c (build_vector): Likewise.
775 (build_vector_from_ctor): Update accordingly.
776 (build_vector_from_val): Likewise.
777 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
778 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
779 * tree-vect-generic.c (add_rshift): Likewise.
780 (expand_vector_divmod): Likewise.
781 (optimize_vector_constructor): Likewise.
782 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
783 (vect_transform_slp_perm_load): Likewise.
784 (vect_schedule_slp_instance): Likewise.
785 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
786 (vectorizable_call): Likewise.
787 (vect_gen_perm_mask_any): Likewise. Add elements in order.
788 * expmed.c (make_tree): Likewise.
789 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
790 a vector passed to build_vector.
791 (fold_convert_const): Likewise.
792 (exact_inverse): Likewise.
793 (fold_ternary_loc): Likewise.
794 (fold_relational_const): Likewise.
795 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
796 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
797 (const_unop): Likewise. Store the reduction accumulator in a
798 variable rather than an array.
799 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
800 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
801 the new vector, rather than constructing it after the input arrays.
802 (native_interpret_vector): Use auto_vec<tree> when building
803 a vector passed to build_vector. Add elements in order.
804 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
805 auto_vec<tree> when building a vector passed to build_vector.
806 (vect_create_epilog_for_reduction): Likewise.
807 (vectorizable_induction): Likewise.
808 (get_initial_def_for_reduction): Likewise. Fix indentation of
810 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
812 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
813 passed to build_vector.
815 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
816 Alan Hayward <alan.hayward@arm.com>
817 David Sherwood <david.sherwood@arm.com>
819 * tree-core.h (tree_base::u): Add an "nelts" field.
820 (tree_vector): Use VECTOR_CST_NELTS as the length.
821 * tree.c (tree_size): Likewise.
822 (make_vector): Initialize VECTOR_CST_NELTS.
823 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
824 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
825 TYPE_VECTOR_SUBPARTS.
826 * expr.c (const_vector_mask_from_tree): Consistently use "units"
827 as the number of units, setting it from VECTOR_CST_NELTS.
828 (const_vector_from_tree): Likewise.
829 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
830 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
831 (fold_negate_expr_1): Likewise.
832 (fold_convert_const): Likewise.
833 (const_binop): Likewise. Differentiate the number of output and
835 (const_unop): Likewise.
836 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
837 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
838 in cases that did the opposite.
840 2017-09-14 Richard Biener <rguenther@suse.de>
842 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
845 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
847 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
849 2017-09-14 Jakub Jelinek <jakub@redhat.com>
852 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
853 if and where to split a bb, except for splitting before debug insn
854 sequences followed by non-label real insn. Delete debug insns
855 in between basic blocks.
857 * combine.c (make_compound_operation_int): Formatting fixes.
859 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
860 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
861 * config/netbsd.h (LINK_EH_SPEC): Likewise.
862 * config/sol2.h (LINK_EH_SPEC): Likewise.
863 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
864 * config/s390/linux.h (LINK_SPEC): Likewise.
865 * config/freebsd.h (LINK_EH_SPEC): Likewise.
866 * config/openbsd.h (LINK_EH_SPEC): Likewise.
867 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
868 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
869 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
870 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
871 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
872 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
874 2017-09-13 Jakub Jelinek <jakub@redhat.com>
876 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
878 (ENDFILE_LINUX_SPEC): Likewise.
879 (LINK_EH_SPEC): Likewise.
880 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
881 (LINK_OS_LINUX_SPEC32): Likewise.
882 (LINK_OS_LINUX_SPEC64): Likewise.
883 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
884 (LINK_OS_LINUX_SPEC): Likewise.
886 2017-09-13 Martin Liska <mliska@suse.cz>
889 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
890 CASE_HIGH is NULL_TREE.
892 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
893 Alan Hayward <alan.hayward@arm.com>
894 David Sherwood <david.sherwood@arm.com>
896 * target.def (secondary_memory_needed): New hook.
897 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
898 instead of SECONDARY_MEMORY_NEEDED.
899 (secondary_memory_needed_mode): Likewise.
900 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
901 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
902 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
903 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
904 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
905 * doc/tm.texi: Regenerate.
906 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
907 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
908 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
909 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
910 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
911 * config/i386/i386.c (inline_secondary_memory_needed): Put the
912 mode argument first and change the reg_class arguments to reg_class_t.
913 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
914 Make static. Update the call to inline_secondary_memory_needed.
915 (ix86_register_move_cost): Update the call to
916 inline_secondary_memory_needed.
917 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
918 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
920 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
921 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
923 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
924 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
925 * config/mips/mips.c (mips_secondary_memory_needed): Make static
926 and match hook interface. Add comment from mips.h.
927 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
928 * config/mmix/mmix.md (truncdfsf2): Refer to
929 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
931 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
932 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
933 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
934 (pa_secondary_memory_needed): New function.
935 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
936 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
937 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
938 (pdp11_secondary_memory_needed): Make static and match hook interface.
939 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
940 * config/powerpcspe/powerpcspe-protos.h
941 (rs6000_secondary_memory_needed_ptr): Delete.
942 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
944 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
945 (rs6000_option_override_internal): Assign to
946 targetm.secondary_memory_needed rather than
947 rs6000_secondary_memory_needed_ptr.
948 (rs6000_secondary_memory_needed): Match hook interface.
949 (rs6000_debug_secondary_memory_needed): Likewise.
950 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
951 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
952 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
953 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
954 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
955 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
957 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
958 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
959 (rs6000_option_override_internal): Assign to
960 targetm.secondary_memory_needed rather than
961 rs6000_secondary_memory_needed_ptr.
962 (rs6000_secondary_memory_needed): Match hook interface.
963 (rs6000_debug_secondary_memory_needed): Likewise.
964 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
965 * config/s390/s390.c (s390_secondary_memory_needed): New function.
966 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
967 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
968 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
969 (sparc_secondary_memory_needed): New function.
970 * lra-constraints.c (check_and_process_move): Refer to
971 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
973 (curr_insn_transform): Likewise.
974 (process_alt_operands): Use targetm.secondary_memory_needed
975 instead of TARGET_SECONDARY_MEMORY_NEEDED.
976 (check_secondary_memory_needed_p): Likewise.
977 (choose_split_class): Likewise.
978 * reload.c: Unconditionally include code that was previously
979 conditional on SECONDARY_MEMORY_NEEDED.
980 (push_secondary_reload): Use targetm.secondary_memory_needed
981 instead of TARGET_SECONDARY_MEMORY_NEEDED.
982 (push_reload): Likewise.
983 * reload1.c: Unconditionally include code that was previously
984 conditional on SECONDARY_MEMORY_NEEDED.
985 (choose_reload_regs): Use targetm.secondary_memory_needed
986 instead of TARGET_SECONDARY_MEMORY_NEEDED.
987 (gen_reload): Likewise.
988 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
990 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
991 Alan Hayward <alan.hayward@arm.com>
992 David Sherwood <david.sherwood@arm.com>
994 * target.def (secondary_memory_needed_mode): New hook:
995 * targhooks.c (default_secondary_memory_needed_mode): Declare.
996 * targhooks.h (default_secondary_memory_needed_mode): New function.
997 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
998 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
999 * doc/tm.texi: Regenerate.
1000 * lra-constraints.c (check_and_process_move): Use
1001 targetm.secondary_memory_needed_mode instead of
1002 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
1003 (curr_insn_transform): Likewise.
1004 * reload.c (get_secondary_mem): Likewise.
1005 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
1006 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
1008 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
1009 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
1010 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
1011 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
1012 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
1014 * config/powerpcspe/powerpcspe-protos.h
1015 (rs6000_secondary_memory_needed_mode): Delete.
1016 * config/powerpcspe/powerpcspe.c
1017 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
1018 (rs6000_secondary_memory_needed_mode): Make static.
1019 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
1020 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
1022 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
1024 (rs6000_secondary_memory_needed_mode): Make static.
1025 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
1026 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
1027 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
1028 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
1029 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
1031 (sparc_secondary_memory_needed_mode): New function.
1032 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
1034 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
1036 * config/aarch64/constraints.md (Umq): New constraint.
1037 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
1039 (mov<mode>): Update condition.
1041 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1043 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
1044 when bitposition is the same.
1046 2017-09-13 Richard Biener <rguenther@suse.de>
1048 * dwarf2out.c (output_die_symbol): Remove.
1049 (output_die): Do not output a DIEs symbol.
1051 2017-09-13 Richard Biener <rguenther@suse.de>
1054 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
1055 default-def to avoid breaking iterator update with the weird
1056 interaction with cgraph_update_edges_for_call_stmt_node.
1058 2017-09-13 Richard Biener <rguenther@suse.de>
1060 * tree-cfg.c (verify_gimple_assign_binary): Add verification
1061 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
1062 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
1063 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
1065 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
1067 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
1068 Disable pc relative literal load irrespective of
1069 TARGET_FIX_ERR_A53_84341 for default.
1071 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
1073 * config/sparc/sparc.c (output_return): Output the source location of
1074 the insn in the delay slot, if any.
1075 (output_sibcall): Likewise.
1077 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
1080 * common.opt (-static-pie): New alias.
1081 (shared): Negate static-pie.
1082 (-no-pie): Update help text.
1084 (static-pie): New option.
1085 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
1086 -static-pie support.
1087 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
1088 (LINK_EH_SPEC): Likewise.
1089 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
1090 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
1091 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
1092 * gcc.c (LINK_COMMAND_SPEC): Likewise.
1093 (init_gcc_specs): Likewise.
1094 (init_spec): Likewise.
1095 (display_help): Update help message for -pie.
1096 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
1099 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
1101 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
1102 (movdi_aarch64): Likewise.
1103 (movti_aarch64): Likewise.
1105 2017-09-12 Simon Wright <simon@pushface.org>
1108 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
1109 calculation of the minor version, always output as 0.
1111 2017-09-12 Jakub Jelinek <jakub@redhat.com>
1114 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
1115 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
1116 on it early, rather than manual conversion late. For
1117 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
1118 instead of performing manual conversion.
1120 2017-09-12 Carl Love <cel@us.ibm.com>
1122 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
1123 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
1125 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
1126 VMULOSW): Add definitions.
1127 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
1128 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
1129 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
1130 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
1131 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
1133 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
1135 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
1137 (movti_aarch64): Likewise.
1138 (movdf_aarch64): Likewise.
1139 (movtf_aarch64): Likewise.
1140 (load_pairdi): Likewise.
1141 (store_pairdi): Likewise.
1142 (load_pairdf): Likewise.
1143 (store_pairdf): Likewise.
1144 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
1145 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
1146 (ldr_got_small_<mode>): Likewise.
1147 (ldr_got_small_28k_<mode>): Likewise.
1148 (ldr_got_tiny): Likewise.
1149 * config/aarch64/iterators.md (ldst_sz): New.
1150 (ldpstp_sz): Likewise.
1151 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
1153 (thunderx_load): Split load_8 to load_16.
1154 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
1156 (thunderx2t99_storepair_basic): Split store_8 to store_16.
1157 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
1158 (xgene1_store_pair): Split store_8 to store_16.
1159 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
1160 (falkor_st_0_st_sd): Split store_8 to store_16.
1162 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
1164 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
1165 and store1/2/3/4 to store_4/8/12/16.
1166 * config/aarch64/aarch64.md: Update for rename.
1167 * config/arm/arm.md: Likewise.: Likewise.
1168 * config/arm/arm.c: Likewise.
1169 * config/arm/thumb1.md: Likewise.
1170 * config/arm/thumb2.md: Likewise.
1171 * config/arm/vfp.md: Likewise.
1172 * config/arm/arm-generic.md: Likewise.
1173 * config/arm/arm1020e.md: Likewise.
1174 * config/arm/arm1026ejs.md: Likewise.
1175 * config/arm/arm1136jfs.md: Likewise.
1176 * config/arm/arm926ejs.md: Likewise.
1177 * config/arm/cortex-a15.md: Likewise.
1178 * config/arm/cortex-a17.md: Likewise.
1179 * config/arm/cortex-a5.md: Likewise.
1180 * config/arm/cortex-a53.md: Likewise.
1181 * config/arm/cortex-a57.md: Likewise.
1182 * config/arm/cortex-a7.md: Likewise.
1183 * config/arm/cortex-a8.md: Likewise.
1184 * config/arm/cortex-a9.md: Likewise.
1185 * config/arm/cortex-m4.md: Likewise.
1186 * config/arm/cortex-m7.md: Likewise.
1187 * config/arm/cortex-r4.md: Likewise.
1188 * config/arm/exynos-m1.md: Likewise.
1189 * config/arm/fa526.md: Likewise.
1190 * config/arm/fa606te.md: Likewise.
1191 * config/arm/fa626te.md: Likewise.
1192 * config/arm/fa726te.md: Likewise.
1193 * config/arm/fmp626.md: Likewise.
1194 * config/arm/iwmmxt.md: Likewise.
1195 * config/arm/ldmstm.md: Likewise.
1196 * config/arm/marvell-pj4.md: Likewise.
1197 * config/arm/xgene1.md: Likewise.
1198 * config/aarch64/thunderx.md: Likewise.
1199 * config/aarch64/thunderx2t99.md: Likewise.
1200 * config/aarch64/falkor.md: Likewise.
1202 2017-09-12 Martin Liska <mliska@suse.cz>
1204 * attribs.c (private_lookup_attribute): New function.
1205 * attribs.h (private_lookup_attribute): Declared here.
1206 (lookup_attribute): Called from this place.
1208 2017-09-12 Richard Biener <rguenther@suse.de>
1210 PR tree-optimization/82157
1211 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
1212 stmts with side-effects.
1214 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1215 Alan Hayward <alan.hayward@arm.com>
1216 David Sherwood <david.sherwood@arm.com>
1218 * target.def (hard_regno_nregs): New hook.
1219 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
1220 * targhooks.h (default_hard_regno_nregs): Declare.
1221 * targhooks.c (default_hard_regno_nregs): New function.
1222 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
1223 (TARGET_HARD_REGNO_NREGS): ...this hook.
1224 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
1225 (CLASS_MAX_NREGS): Likewise.
1226 * doc/tm.texi: Regenerate.
1227 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
1228 instead of HARD_REGNO_NREGS.
1229 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
1230 HARD_REGNO_NREGS in the comment.
1231 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
1232 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
1233 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
1234 Return an unsigned int.
1235 (TARGET_HARD_REGNO_NREGS): Redefine.
1236 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
1237 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
1238 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
1239 (arc_hard_regno_nregs): New function.
1240 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
1241 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
1242 (arm_hard_regno_nregs): New function.
1243 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
1244 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
1245 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
1246 (TARGET_HARD_REGNO_NREGS): Redefine.
1247 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
1248 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
1249 (HARD_REGNO_NREGS): Delete.
1250 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
1251 (cr16_hard_regno_nregs): New function.
1252 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
1253 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
1254 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
1255 (cris_hard_regno_nregs): New function.
1256 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
1257 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
1258 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
1259 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
1260 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
1261 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
1262 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
1263 (frv_hard_regno_nregs): Make static. Take and return an
1265 (frv_class_max_nregs): Remove outdated copy of documentation.
1266 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
1267 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
1268 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
1269 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
1270 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
1271 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
1272 (TARGET_HARD_REGNO_NREGS): Redefine.
1273 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
1274 (CLASS_MAX_NREGS): Update comment.
1275 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
1276 (ia64_hard_regno_nregs): New function.
1277 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
1278 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
1279 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
1280 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
1281 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
1283 (m32c_hard_regno_nregs): Likewise. Make static.
1284 (TARGET_HARD_REGNO_NREGS): Redefine.
1285 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
1286 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
1287 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
1288 (m68k_hard_regno_nregs): New function.
1289 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
1290 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
1291 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
1292 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
1293 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
1294 Take and return an unsigned int.
1295 (TARGET_HARD_REGNO_NREGS): Redefine.
1296 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
1297 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
1298 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
1299 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
1300 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
1301 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
1302 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
1303 (msp430_hard_regno_nregs): Make static. Take and return an
1305 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
1306 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
1307 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
1308 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
1309 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
1310 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
1311 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
1312 (TARGET_HARD_REGNO_NREGS): Redefine.
1313 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
1314 (PA_HARD_REGNO_NREGS): ...this.
1315 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
1316 (PA_HARD_REGNO_NREGS): ...this.
1317 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
1318 (pa_hard_regno_nregs): New function.
1319 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
1320 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
1321 (pdp11_hard_regno_nregs): New function.
1322 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
1323 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
1324 (rs6000_hard_regno_nregs_hook): New function.
1325 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
1326 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
1327 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
1328 Take and return an unsigned int. Move earlier in file.
1329 (TARGET_HARD_REGNO_NREGS): Redefine.
1330 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
1331 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
1332 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
1333 (rl78_hard_regno_nregs): Make static. Take and return an
1335 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
1336 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
1337 (rs6000_hard_regno_nregs_hook): New function.
1338 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
1339 * config/rx/rx.c (rx_hard_regno_nregs): New function.
1340 (TARGET_HARD_REGNO_NREGS): Redefine.
1341 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
1342 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
1343 instead of HARD_REGNO_NREGS.
1344 (s390_hard_regno_nregs): New function.
1345 (s390_hard_regno_mode_ok): Add comment from s390.h.
1346 (TARGET_HARD_REGNO_NREGS): Redefine.
1347 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
1348 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
1349 (sh_hard_regno_nregs): New function.
1350 (sh_pass_in_reg_p): Use it.
1351 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
1352 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
1353 (sparc_hard_regno_nregs): New function.
1354 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
1355 * config/spu/spu.c (spu_hard_regno_nregs): New function.
1356 (spu_function_arg_advance): Use it, supplying a valid register number.
1357 (TARGET_HARD_REGNO_NREGS): Redefine.
1358 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
1359 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
1360 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
1361 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
1362 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
1363 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
1364 (CLASS_MAX_NREGS): Remove copy of old documentation.
1365 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
1366 (visium_hard_regno_nregs): New function.
1367 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
1368 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
1369 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
1370 xtensa_hard_regno_nregs): New function.
1371 * system.h (HARD_REGNO_NREGS): Poison.
1373 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1375 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
1376 hard_regno_nregs instead of HARD_REGNO_NREGS.
1377 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
1378 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
1379 (c6x_expand_epilogue): Likewise.
1380 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
1381 (frv_read_iacc_argument): Likewise.
1382 * config/sh/sh.c: Include regs.h.
1383 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
1384 (regs_used): Likewise.
1385 (output_stack_adjust): Likewise.
1386 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
1387 * expmed.c: Include regs.h.
1388 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
1389 * ree.c: Include regs.h.
1390 (combine_reaching_defs): Use hard_regno_nregs instead of
1392 (add_removable_extension): Likewise.
1394 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1396 * regs.h (hard_regno_nregs): Turn into a function.
1397 (end_hard_regno): Update accordingly.
1398 * caller-save.c (setup_save_areas): Likewise.
1399 (save_call_clobbered_regs): Likewise.
1400 (replace_reg_with_saved_mem): Likewise.
1401 (insert_restore): Likewise.
1402 (insert_save): Likewise.
1403 * combine.c (can_change_dest_mode): Likewise.
1404 (move_deaths): Likewise.
1405 (distribute_notes): Likewise.
1406 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
1407 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
1408 (rs6000_split_multireg_move): Likewise.
1409 (rs6000_register_move_cost): Likewise.
1410 (rs6000_memory_move_cost): Likewise.
1411 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
1412 (rs6000_split_multireg_move): Likewise.
1413 (rs6000_register_move_cost): Likewise.
1414 (rs6000_memory_move_cost): Likewise.
1415 * cselib.c (cselib_reset_table): Likewise.
1416 (cselib_lookup_1): Likewise.
1417 * emit-rtl.c (set_mode_and_regno): Likewise.
1418 * function.c (aggregate_value_p): Likewise.
1419 * ira-color.c (setup_profitable_hard_regs): Likewise.
1420 (check_hard_reg_p): Likewise.
1421 (calculate_saved_nregs): Likewise.
1422 (assign_hard_reg): Likewise.
1423 (improve_allocation): Likewise.
1424 (calculate_spill_cost): Likewise.
1425 * ira-emit.c (modify_move_list): Likewise.
1426 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
1427 (ira_hard_reg_in_set_p): Likewise.
1428 * ira.c (setup_reg_mode_hard_regset): Likewise.
1429 (clarify_prohibited_class_mode_regs): Likewise.
1430 (check_allocation): Likewise.
1431 * lra-assigns.c (find_hard_regno_for_1): Likewise.
1432 (lra_setup_reg_renumber): Likewise.
1433 (setup_try_hard_regno_pseudos): Likewise.
1434 (spill_for): Likewise.
1435 (assign_hard_regno): Likewise.
1436 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
1437 * lra-constraints.c (in_class_p): Likewise.
1438 (lra_constraint_offset): Likewise.
1439 (simplify_operand_subreg): Likewise.
1440 (lra_constraints): Likewise.
1441 (split_reg): Likewise.
1442 (split_if_necessary): Likewise.
1443 (invariant_p): Likewise.
1444 (inherit_in_ebb): Likewise.
1445 * lra-lives.c (process_bb_lives): Likewise.
1446 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
1447 (get_hard_regs): Likewise.
1448 (do_remat): Likewise.
1449 * lra-spills.c (assign_spill_hard_regs): Likewise.
1450 * mode-switching.c (create_pre_exit): Likewise.
1451 * postreload.c (reload_combine_recognize_pattern): Likewise.
1452 * recog.c (peep2_find_free_register): Likewise.
1453 * regcprop.c (kill_value_regno): Likewise.
1454 (set_value_regno): Likewise.
1455 (copy_value): Likewise.
1456 (maybe_mode_change): Likewise.
1457 (find_oldest_value_reg): Likewise.
1458 (copyprop_hardreg_forward_1): Likewise.
1459 * regrename.c (check_new_reg_p): Likewise.
1460 (regrename_do_replace): Likewise.
1461 * reload.c (push_reload): Likewise.
1462 (combine_reloads): Likewise.
1463 (find_dummy_reload): Likewise.
1464 (operands_match_p): Likewise.
1465 (find_reloads): Likewise.
1466 (find_equiv_reg): Likewise.
1467 (reload_adjust_reg_for_mode): Likewise.
1468 * reload1.c (count_pseudo): Likewise.
1469 (count_spilled_pseudo): Likewise.
1470 (find_reg): Likewise.
1471 (clear_reload_reg_in_use): Likewise.
1472 (free_for_value_p): Likewise.
1473 (allocate_reload_reg): Likewise.
1474 (choose_reload_regs): Likewise.
1475 (reload_adjust_reg_for_temp): Likewise.
1476 (emit_reload_insns): Likewise.
1477 (delete_output_reload): Likewise.
1478 * rtlanal.c (subreg_get_info): Likewise.
1479 * sched-deps.c (sched_analyze_reg): Likewise.
1480 * sel-sched.c (init_regs_for_mode): Likewise.
1481 (mark_unavailable_hard_regs): Likewise.
1482 (choose_best_reg_1): Likewise.
1483 (verify_target_availability): Likewise.
1484 * valtrack.c (dead_debug_insert_temp): Likewise.
1485 * var-tracking.c (track_loc_p): Likewise.
1486 (emit_note_insn_var_location): Likewise.
1487 * varasm.c (make_decl_rtl): Likewise.
1488 * reginfo.c (choose_hard_reg_mode): Likewise.
1489 (init_reg_modes_target): Refer directly to
1490 this_target_regs->x_hard_regno_nregs.
1492 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1494 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
1495 instead of hard_regno_nregs.
1497 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1499 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
1500 end_hard_regno instead of hard_regno_nregs.
1501 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
1502 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
1503 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
1504 * ira-color.c (improve_allocation): Likewise.
1505 * lra-assigns.c (find_hard_regno_for_1): Likewise.
1506 * lra-lives.c (mark_regno_live): Likewise.
1507 (mark_regno_dead): Likewise.
1508 * lra-remat.c (operand_to_remat): Likewise.
1509 * lra.c (collect_non_operand_hard_regs): Likewise.
1510 * postreload.c (reload_combine_note_store): Likewise.
1511 (move2add_valid_value_p): Likewise.
1512 * reload.c (regno_clobbered_p): Likewise.
1514 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1516 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
1518 * config/v850/v850.c (v850_reorg): Likewise.
1519 * reload.c (refers_to_regno_for_reload_p): Likewise.
1520 (find_equiv_reg): Likewise.
1521 * reload1.c (reload_reg_reaches_end_p): Likewise.
1523 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1525 * caller-save.c (add_used_regs): Use REG_NREGS instead of
1527 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
1528 * config/arm/arm.c (output_move_neon): Likewise.
1529 (arm_attr_length_move_neon): Likewise.
1530 (neon_split_vcombine): Likewise.
1531 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
1532 (c6x_mark_reg_written): Likewise.
1533 (c6x_dwarf_register_span): Likewise.
1534 * config/i386/i386.c (ix86_save_reg): Likewise.
1535 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
1536 (rws_access_reg): Likewise.
1537 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
1538 * mode-switching.c (create_pre_exit): Likewise.
1539 * ree.c (combine_reaching_defs): Likewise.
1540 (add_removable_extension): Likewise.
1541 * regcprop.c (find_oldest_value_reg): Likewise.
1542 (copyprop_hardreg_forward_1): Likewise.
1543 * reload.c (reload_inner_reg_of_subreg): Likewise.
1544 (push_reload): Likewise.
1545 (combine_reloads): Likewise.
1546 (find_dummy_reload): Likewise.
1547 (reload_adjust_reg_for_mode): Likewise.
1548 * reload1.c (find_reload_regs): Likewise.
1549 (forget_old_reloads_1): Likewise.
1550 (reload_reg_free_for_value_p): Likewise.
1551 (reload_adjust_reg_for_temp): Likewise.
1552 (emit_reload_insns): Likewise.
1553 (delete_output_reload): Likewise.
1554 * sel-sched.c (choose_best_reg_1): Likewise.
1555 (choose_best_pseudo_reg): Likewise.
1557 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1558 Alan Hayward <alan.hayward@arm.com>
1559 David Sherwood <david.sherwood@arm.com>
1561 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
1562 * target.def (slow_unaligned_access): New hook.
1563 * targhooks.h (default_slow_unaligned_access): Declare.
1564 * targhooks.c (default_slow_unaligned_access): New function.
1565 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
1566 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
1567 * doc/tm.texi: Regenerate.
1568 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
1569 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
1570 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
1572 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
1573 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
1575 (rs6000_slow_unaligned_access): New function.
1576 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
1577 (expand_block_compare): Likewise.
1578 (expand_strn_compare): Likewise.
1579 (rs6000_rtx_costs): Likewise.
1580 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
1581 (riscv_slow_unaligned_access): Likewise.
1582 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
1583 (riscv_slow_unaligned_access_p): ...this and make static.
1584 (riscv_option_override): Update accordingly.
1585 (riscv_slow_unaligned_access): New function.
1586 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
1587 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
1588 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
1589 (rs6000_slow_unaligned_access): New function.
1590 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
1591 (rs6000_rtx_costs): Likewise.
1592 * config/rs6000/rs6000-string.c (expand_block_compare)
1593 (expand_strn_compare): Use targetm.slow_unaligned_access instead
1594 of SLOW_UNALIGNED_ACCESS.
1595 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
1596 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
1597 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
1598 of SLOW_UNALIGNED_ACCESS.
1599 * expmed.c (simple_mem_bitfield_p): Likewise.
1600 * expr.c (alignment_for_piecewise_move): Likewise.
1601 (emit_group_load_1): Likewise.
1602 (emit_group_store): Likewise.
1603 (copy_blkmode_from_reg): Likewise.
1604 (emit_push_insn): Likewise.
1605 (expand_assignment): Likewise.
1606 (store_field): Likewise.
1607 (expand_expr_real_1): Likewise.
1608 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
1609 * lra-constraints.c (simplify_operand_subreg): Likewise.
1610 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
1611 * gimple-ssa-store-merging.c: Likewise in block comment at start
1613 * tree-ssa-strlen.c: Include target.h.
1614 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
1615 of SLOW_UNALIGNED_ACCESS.
1616 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
1618 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1620 PR rtl-optimization/82185
1621 * expmed.c (emit_store_flag_int): Only test tem if it has been
1624 2017-09-12 Richard Biener <rguenther@suse.de>
1627 * match.pd ((FTYPE) N CMP CST): Fix typo.
1629 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
1631 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
1633 (mips_near_type_p): Add 'short_call' attribute as a synonym
1635 * doc/extend.texi (short_call): Document new function attribute.
1637 2017-09-12 Jakub Jelinek <jakub@redhat.com>
1640 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
1641 assertion check that in the condition.
1642 (get_atomic_generic_size): Likewise. Before testing if parameter
1643 has pointer type, if it has array type, call for C++
1644 default_conversion to perform array-to-pointer conversion.
1646 2017-09-12 Richard Biener <rguenther@suse.de>
1648 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
1649 for operations we cannot scalarize.
1651 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
1653 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
1654 vectors heap vectors. Clean up comments.
1655 Make visited_bbs a reference.
1656 (profitable_jump_thread_path): Make GC
1657 vectors heap vectors. Clean up comments.
1659 (convert_and_register_jump_thread_path): Make GC vectors heap
1661 (check_subpath_and_update_thread_path): Same. Clean up comments.
1662 Make visited_bbs a reference.
1663 (handle_phi): Abstract common code to to
1664 register_jump_thread_path_if_profitable.
1665 Rename VAR_BB to DEF_BB.
1667 Make GC vectors heap vectors.
1668 Make visited_bbs a reference.
1669 (handle_assignment): Same.
1670 (register_jump_thread_path_if_profitable): New.
1671 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
1673 Make GC vectors heap vectors. Clean up comments.
1674 Make visited_bbs a reference.
1675 (find_jump_threads_backwards): Make visited_bbs live in the stack.
1676 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
1679 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
1682 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
1683 words of E_DImode object are reachable by xtensa_uimm8x4 access.
1685 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
1687 Revert r251800 and r251799.
1689 2017-09-11 Martin Jambor <mjambor@suse.cz>
1692 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
1693 arguments in advance.
1694 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
1695 use it to find predecessor edges.
1696 (naive_outof_ssa): Collect vector of predecessors.
1698 2017-09-08 Jason Merrill <jason@redhat.com>
1700 PR c++/70029 - ICE with ref-qualifier and -flto
1701 * langhooks.h (struct lang_hooks_for_types): Add
1702 copy_lang_qualifiers.
1703 * attribs.c (build_type_attribute_qual_variant): Use it.
1704 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
1706 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
1707 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
1709 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
1712 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
1713 (*mulsi3_sp64): New instruction.
1714 (mulsi3): New expander.
1716 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
1718 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
1720 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1722 * sancov.c: Include memmodel.h.
1724 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
1727 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
1730 2017-09-07 Carl Love <cel@us.ibm.com>
1732 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
1733 the sldi instruction.
1735 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
1737 * sancov.c: Include tm_p.h.
1739 2017-09-07 Jakub Jelinek <jakub@redhat.com>
1742 * output.h (switch_to_other_text_partition): New declaration.
1743 * varasm.c (switch_to_other_text_partition): New function.
1744 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
1745 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
1746 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
1747 to the other text partition before emitting LCL label and switch back
1748 after emitting the word after it.
1750 2017-09-07 Richard Biener <rguenther@suse.de>
1752 * passes.def (pass_split_crit_edges): Remove instance before PRE.
1753 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
1754 critical edges here, after loop init.
1755 (pass_data_pre): Remove PROP_no_crit_edges flags.
1756 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
1757 for valueization of call args to avoid leaking VN_TOP.
1758 (visit_use): Assert we do not visit default defs.
1759 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
1760 Use error_mark_node to more easily detect leaking VN_TOP.
1761 All default-defs are varying, not VN_TOP. Mark them visited.
1762 (run_scc_vn): Make code match comment.
1764 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
1766 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
1767 OPTION_MASK_FLOAT128_KEYWORD.
1768 (POWERPC_MASKS): Likewise.
1769 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
1770 support for the -mfloat128-type option, and make -mfloat128
1771 default on PowerPC Linux systems. Define or undefine
1772 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
1773 Define __float128 to be __ieee128 if IEEE 128-bit support is
1774 enabled, or undefine it.
1775 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
1776 Delete defining __FLOAT128_TYPE__.
1777 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
1778 -mfloat128-type option and make -mfloat128 default on PowerPC
1780 (TARGET_FLOAT128_TYPE): Likewise.
1781 (-mfloat128-type): Likewise.
1782 * config/rs6000/rs6000.c (rs6000_option_override_internal):
1783 Delete the -mfloat128-type option and make -mfloat128 default on
1784 PowerPC Linux systems. Always use __ieee128 to be the keyword for
1785 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
1786 128-bit floating point is enabled. Change tests from using
1787 -mfloat128-type to -mfloat128.
1788 (rs6000_mangle_type): Use the correct mangling for the __float128
1789 type even if normal long double is restricted to 64-bits.
1790 (floatn_mode): Enable the _Float128 type by default on VSX Linux
1792 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
1793 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
1794 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
1796 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
1797 documentation for -mfloat128.
1799 2017-09-06 Olivier Hainque <hainque@adacore.com>
1801 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
1803 2017-09-06 Wish Wu <wishwu007@gmail.com>
1804 Jakub Jelinek <jakub@redhat.com>
1806 * asan.c (initialize_sanitizer_builtins): Add
1807 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
1808 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
1809 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
1810 BT_FN_VOID_UINT64_PTR variables.
1811 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
1812 (BT_FN_VOID_UINT16_UINT16): Likewise.
1813 (BT_FN_VOID_UINT32_UINT32): Likewise.
1814 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
1815 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
1816 (BT_FN_VOID_UINT64_PTR): Likewise.
1817 * common.opt (flag_sanitize_coverage): New variable.
1818 (fsanitize-coverage=trace-pc): Remove.
1819 (fsanitize-coverage=): Add.
1820 * flag-types.h (enum sanitize_coverage_code): New enum.
1821 * fold-const.c (fold_range_test): Disable non-short-circuit
1822 optimization if flag_sanitize_coverage.
1823 (fold_truth_andor): Likewise.
1824 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
1825 * opts.c (COVERAGE_SANITIZER_OPT): Define.
1826 (coverage_sanitizer_opts): New array.
1827 (get_closest_sanitizer_option): Add OPTS argument, handle also
1828 OPT_fsanitize_coverage_.
1829 (parse_sanitizer_options): Adjusted to also handle
1830 OPT_fsanitize_coverage_.
1831 (common_handle_option): Add OPT_fsanitize_coverage_.
1832 * sancov.c (instrument_comparison, instrument_switch): New function.
1833 (sancov_pass): Add trace-cmp support.
1834 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
1835 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
1836 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
1837 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
1838 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
1839 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
1840 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
1841 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
1842 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
1843 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
1845 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
1847 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
1848 error. Only quit immediately if parsing is complete.
1849 (BEGIN): Initialize fatal_err and parse_done.
1850 (begin fpu, end fpu): Check number of arguments.
1851 (begin arch, end arch): Likewise.
1852 (begin cpu, end cpu): Likewise.
1853 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
1854 (optalias): Likewise.
1856 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
1858 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
1859 * config/arm/arm-isa.h: Delete. Move definitions to ...
1860 * arm-cpus.in: ... here. Use new feature and fgroup values.
1861 * config/arm/arm.c (arm_option_override): Use lower case for feature
1863 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
1864 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
1865 * config/arm/parsecpu.awk (END): Add new command 'isa'.
1867 (print_isa_bits_for): New function.
1868 (gen_isa): New function.
1869 (gen_comm_data): Use print_isa_bits_for.
1870 (define feature): New keyword.
1871 (define fgroup): New keyword.
1872 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
1873 (arm-isa.h): Add rule to generate file.
1874 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
1875 case for feature bit names.
1877 2017-09-06 Richard Biener <rguenther@suse.de>
1879 * tree-ssa-pre.c (NECESSARY): Remove.
1880 (create_expression_by_pieces): Do not touch pass-local flags.
1881 (insert_into_preds_of_block): Likewise.
1882 (do_pre_regular_insertion): Likewise.
1883 (eliminate_insert): Likewise.
1884 (eliminate_dom_walker::before_dom_children): Likewise.
1885 (fini_eliminate): Do not look at inserted_exprs.
1886 (mark_operand_necessary): Remove.
1887 (remove_dead_inserted_code): Replace with simple work-list
1888 algorithm based on inserted_exprs and SSA uses.
1889 (pass_pre::execute): Re-order fini_eliminate and
1890 remove_dead_inserted_code.
1892 2017-09-06 Olivier Hainque <hainque@adacore.com>
1894 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
1895 for VxWorks 7. Adjust surrounding comments.
1897 2017-09-06 Richard Biener <rguenther@suse.de>
1899 * gimple-ssa-strength-reduction.c
1900 (find_candidates_dom_walker::before_dom_children): Also allow
1903 2017-09-06 Richard Biener <rguenther@suse.de>
1905 PR tree-optimization/82108
1906 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
1907 for gap in the non-permutation SLP case.
1909 2017-09-06 Martin Jambor <mjambor@suse.cz>
1911 PR tree-optimization/82078
1912 * tree-sra.c (sort_and_splice_var_accesses): Move call to
1913 add_access_to_work_queue...
1914 (build_accesses_from_assign): ...here.
1915 (propagate_all_subaccesses): Make sure racc is the group
1916 representative, if there is one.
1918 2017-09-06 Jakub Jelinek <jakub@redhat.com>
1921 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
1924 2017-09-06 Richard Biener <rguenther@suse.de>
1926 * gimple-ssa-strength-reduction.c
1927 (find_candidates_dom_walker::before_doom_children): Use a
1928 type and not a mode check.
1930 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
1933 * config/arm/predicates.md (arm_general_adddi_operand): Create new
1935 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
1937 2017-09-05 Jeff Law <law@redhat.com>
1939 PR tree-optimization/64910
1940 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
1941 cases where we have 3 or more operands.
1943 2017-09-05 Jakub Jelinek <jakub@redhat.com>
1946 * omp-low.c (lower_omp_for): Recompute tree invariant if
1947 gimple_omp_for_initial/final is ADDR_EXPR.
1950 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
1951 into gimple val before gimplification fo the COND_EXPR.
1953 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
1955 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
1956 REGION_COPY argument.
1957 (thread_through_all_blocks): Remove unused argument to
1958 duplicate_thread_path.
1960 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1961 Alan Hayward <alan.hayward@arm.com>
1962 David Sherwood <david.sherwood@arm.com>
1964 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
1965 Take a scalar_mode rather than a machine_mode.
1966 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
1967 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
1968 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
1969 (aarch64_gen_adjusted_ldpstp): Likewise.
1970 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
1972 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1973 Alan Hayward <alan.hayward@arm.com>
1974 David Sherwood <david.sherwood@arm.com>
1976 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
1977 Take a scalar_int_mode instead of a machine_mode.
1978 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
1979 (aarch64_output_scalar_simd_mov_immediate): Likewise.
1980 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
1981 (aarch64_simd_attr_length_rglist): Delete.
1982 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
1983 a scalar_int_mode instead of a machine_mode.
1984 (aarch64_add_offset): Likewise.
1985 (aarch64_internal_mov_immediate): Likewise
1986 (aarch64_add_constant_internal): Likewise.
1987 (aarch64_add_constant): Likewise.
1988 (aarch64_movw_imm): Likewise.
1989 (aarch64_rtx_arith_op_extract_p): Likewise.
1990 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
1991 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
1992 Remove assert that the mode isn't a vector.
1993 (aarch64_output_scalar_simd_mov_immediate): Likewise.
1994 (aarch64_expand_mov_immediate): Update calls after above changes.
1995 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
1996 (aarch64_and_bitmask_imm): Check for scalar integer modes.
1997 (aarch64_move_imm): Likewise.
1998 (aarch64_can_const_movi_rtx_p): Likewise.
1999 (aarch64_strip_extend): Likewise.
2000 (aarch64_extr_rtx_p): Likewise.
2001 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
2002 a CONST_INT when the mode parameter is VOIDmode.
2003 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
2005 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2007 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
2008 * stor-layout.c (bitwise_mode_for_mode): Likewise.
2009 (bitwise_type_for_mode): Update accordingly.
2011 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2013 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
2014 * stor-layout.c (mode_for_size_tree): Likewise.
2015 (mode_for_array): Update accordingly.
2016 (layout_decl): Likewise.
2017 (compute_record_mode): Likewise. Only set the mode once.
2019 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2021 * target.def (get_mask_mode): Change return type to opt_mode.
2023 * doc/tm.texi: Regenerate.
2024 * targhooks.h (default_get_mask_mode): Return an opt_mode.
2025 * targhooks.c (default_get_mask_mode): Likewise.
2026 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
2027 * optabs-query.c (can_vec_mask_load_store_p): Update use of
2028 targetm.get_mask_mode.
2029 * tree.c (build_truth_vector_type): Likewise.
2031 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2033 * machmode.h (mode_for_vector): Return an opt_mode.
2034 * stor-layout.c (mode_for_vector): Likewise.
2035 (mode_for_int_vector): Update accordingly.
2036 (layout_type): Likewise.
2037 * config/i386/i386.c (emit_memmov): Likewise.
2038 (ix86_expand_set_or_movmem): Likewise.
2039 (ix86_expand_vector_init): Likewise.
2040 (ix86_get_mask_mode): Likewise.
2041 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
2043 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
2044 * expmed.c (extract_bit_field_1): Likewise.
2045 * expr.c (expand_expr_real_2): Likewise.
2046 * optabs-query.c (can_vec_perm_p): Likewise.
2047 (can_vec_mask_load_store_p): Likewise.
2048 * optabs.c (expand_vec_perm): Likewise.
2049 * targhooks.c (default_get_mask_mode): Likewise.
2050 * tree-vect-stmts.c (vectorizable_store): Likewise.
2051 (vectorizable_load): Likewise.
2052 (get_vectype_for_scalar_type_and_size): Likewise.
2054 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2056 * machmode.h (mode_for_int_vector): New function.
2057 * stor-layout.c (mode_for_int_vector): Likewise.
2058 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
2059 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
2060 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
2061 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
2062 (s390_expand_vcond): Likewise.
2064 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2066 * machmode.h (opt_machine_mode): New type.
2067 (opt_mode<T>): Allow construction from anything that can be
2069 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
2070 (mode_for_size): Return an opt_machine_mode.
2071 * stor-layout.c (mode_for_size): Likewise.
2072 (mode_for_size_tree): Update call accordingly.
2073 (bitwise_mode_for_mode): Likewise.
2074 (make_fract_type): Likewise.
2075 (make_accum_type): Likewise.
2076 * caller-save.c (replace_reg_with_saved_mem): Update call
2078 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
2079 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
2080 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
2081 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
2082 * expmed.c (extract_bit_field_1): Likewise.
2083 * reload.c (get_secondary_mem): Likewise.
2084 * varasm.c (assemble_integer): Likewise.
2085 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
2088 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2090 * machmode.h (decimal_float_mode_for_size): New function.
2091 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
2092 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
2093 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
2094 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
2095 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
2096 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
2098 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2100 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
2101 (get_builtin_sync_mode): Likewise.
2102 (expand_ifn_atomic_compare_exchange): Likewise.
2103 (expand_builtin_atomic_clear): Likewise.
2104 (expand_builtin_atomic_test_and_set): Likewise.
2105 (fold_builtin_atomic_always_lock_free): Likewise.
2106 * calls.c (compute_argument_addresses): Likewise.
2107 (emit_library_call_value_1): Likewise.
2108 (store_one_arg): Likewise.
2109 * combine.c (combine_instructions): Likewise.
2110 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
2111 * config/arm/arm.c (arm_function_value): Likewise.
2112 (aapcs_allocate_return_reg): Likewise.
2113 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
2114 * config/i386/i386.c (construct_container): Likewise.
2115 (ix86_gimplify_va_arg): Likewise.
2116 (ix86_expand_sse_cmp): Likewise.
2117 (emit_memmov): Likewise.
2118 (emit_memset): Likewise.
2119 (expand_small_movmem_or_setmem): Likewise.
2120 (ix86_expand_pextr): Likewise.
2121 (ix86_expand_pinsr): Likewise.
2122 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
2123 * config/microblaze/microblaze.c (microblaze_block_move_straight):
2125 * config/mips/mips.c (mips_function_value_1) Likewise.
2126 (mips_block_move_straight): Likewise.
2127 (mips_expand_ins_as_unaligned_store): Likewise.
2128 * config/powerpcspe/powerpcspe.c
2129 (rs6000_darwin64_record_arg_advance_flush): Likewise.
2130 (rs6000_darwin64_record_arg_flush): Likewise.
2131 * config/rs6000/rs6000.c
2132 (rs6000_darwin64_record_arg_advance_flush): Likewise.
2133 (rs6000_darwin64_record_arg_flush): Likewise.
2134 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
2135 (sparc_function_value_1): Likewise.
2136 * config/spu/spu.c (adjust_operand): Likewise.
2137 (spu_emit_branch_or_set): Likewise.
2138 (arith_immediate_p): Likewise.
2139 * emit-rtl.c (gen_lowpart_common): Likewise.
2140 * expr.c (expand_expr_real_1): Likewise.
2141 * function.c (assign_parm_setup_block): Likewise.
2142 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
2143 * reload1.c (alter_reg): Likewise.
2144 * stor-layout.c (mode_for_vector): Likewise.
2145 (layout_type): Likewise.
2147 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2149 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
2150 (spu_convert_move): Likewise.
2151 * lower-subreg.c (resolve_simple_move): Likewise.
2153 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2156 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
2157 define_insn to a define_expand.
2158 (altivec_vsum2sws_direct): New define_insn.
2159 (altivec_vsumsws): Convert from a define_insn to a define_expand.
2161 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
2163 * config/arm/arm.c (arm_option_params_internal): Improve setting of
2166 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
2171 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
2172 realign stack if stack alignment needed is less than incoming
2175 2017-09-05 Marek Polacek <polacek@redhat.com>
2178 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
2181 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
2183 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
2185 2017-09-05 Richard Biener <rguenther@suse.de>
2187 PR tree-optimization/82084
2188 * fold-const.c (can_native_encode_string_p): Handle wide characters.
2190 2017-09-05 Richard Biener <rguenther@suse.de>
2192 PR tree-optimization/82102
2193 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
2195 2017-09-05 Martin Liska <mliska@suse.cz>
2197 PR tree-optimization/82032
2198 * tree-cfg.c (generate_range_test): New function.
2199 * tree-cfg.h (generate_range_test): Declared here.
2200 * tree-cfgcleanup.c (convert_single_case_switch): New function.
2201 (cleanup_control_expr_graph): Use it.
2202 * tree-switch-conversion.c (try_switch_expansion): Remove
2204 (emit_case_nodes): Use generate_range_test.
2206 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
2209 * config/i386/i386.md (*<btsc><mode>_mask): Add
2210 TARGET_USE_BT to insn constraint.
2211 (*btr<mode>_mask): Ditto.
2213 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
2215 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
2216 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
2218 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2221 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
2222 TARGET_NEON and TARGET_IWMMXT.
2223 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
2224 TARGET_NEON and TARGET_IWMMXT.
2225 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
2227 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
2229 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
2230 (ix86_rewrite_tls_address): Ditto.
2231 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
2232 (ix86_rewrite_tls_address_1): Ditto.
2233 (ix86_rewrite_tls_address): Ditto.
2234 * config/i386/predicates.md (tls_address_pattern): New predicate.
2235 * config/i386/i386.md (TLS address splitter): New splitter.
2237 2017-09-04 Richard Biener <rguenther@suse.de>
2239 PR tree-optimization/82084
2240 * fold-const.h (can_native_encode_string_p): Declare.
2241 * fold-const.c (can_native_encode_string_p): Factor out from ...
2242 (native_encode_string): ... here.
2243 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
2244 vectorizing stores from constants we later cannot handle.
2246 2017-09-04 Marek Polacek <polacek@redhat.com>
2249 * doc/invoke.texi: Update -Wtautological-compare documentation.
2251 2017-09-04 Jeff Law <law@redhat.com>
2253 PR tree-optimization/64910
2254 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
2255 swap the first and last operand if the last is a constant.
2257 2017-09-04 Marek Polacek <polacek@redhat.com>
2260 * convert.c (do_narrow): When sanitizing signed integer overflows,
2261 bail out for signed types.
2262 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
2264 2017-09-04 Richard Biener <rguenther@suse.de>
2266 PR tree-optimization/82060
2267 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
2268 Move devirtualization after stmt folding and before EH/AB/noreturn
2269 cleanup to get the stmt refs canonicalized. Use a bool instead
2270 of gimple_modified_p since that doesn't work for NOPs. Schedule
2271 NOPs generated by folding for removal.
2273 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2274 Alan Hayward <alan.hayward@arm.com>
2275 David Sherwood <david.sherwood@arm.com>
2277 * coretypes.h (pad_direction): New enum.
2278 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
2279 (FUNCTION_ARG_PADDING): Likewise.
2280 * target.def (function_arg_padding): New hook.
2281 * targhooks.h (default_function_arg_padding): Declare.
2282 * targhooks.c (default_function_arg_padding): New function.
2283 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
2284 (TARGET_FUNCTION_ARG_PADDING): ...this.
2285 * doc/tm.texi: Regenerate.
2286 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
2287 instead of direction.
2288 (compute_argument_addresses): Likewise.
2289 (load_register_parameters): Likewise.
2290 (emit_library_call_value_1): Likewise.
2291 (store_one_arg): Use targetm.calls.function_arg_padding instead
2292 of FUNCTION_ARG_PADDING.
2293 (must_pass_in_stack_var_size_or_pad): Likewise.
2294 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
2295 (emit_group_store): Likewise.
2296 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
2297 instead of FUNCTION_ARG_PADDING.
2298 (emit_push_insn): Likewise, and propagate enum change throughout
2300 * function.h (direction): Delete.
2301 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
2303 * function.c (assign_parm_find_stack_rtl): Likewise.
2304 (assign_parm_setup_block_p): Likewise.
2305 (assign_parm_setup_block): Likewise.
2306 (gimplify_parameters): Likewise.
2307 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
2308 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
2310 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
2311 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
2312 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
2313 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
2314 (aarch64_function_arg_padding): ...this new function.
2315 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
2316 (TARGET_FUNCTION_ARG_PADDING): Redefine.
2317 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
2318 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
2319 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
2320 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2321 (arm_pad_arg_upward): Replace with...
2322 (arm_function_arg_padding): ...this new function.
2323 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
2325 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
2326 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
2327 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2328 (ia64_hpux_function_arg_padding): Replace with...
2329 (ia64_function_arg_padding): ...this new function. Use pad_direction
2330 instead of direction. Check for TARGET_HPUX.
2331 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
2332 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2333 (iq2000_function_arg_padding): New function.
2334 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
2335 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
2336 (mips_function_arg_padding): ...this new function.
2337 (mips_pad_reg_upward): Update accordingly.
2338 (TARGET_FUNCTION_ARG_PADDING): Redefine.
2339 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
2340 targetm.calls.function_arg_padding.
2341 (FUNCTION_ARG_PADDING): Delete.
2342 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
2343 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
2344 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
2345 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
2346 (nios2_block_reg_padding): Return pad_direction instead of direction.
2347 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
2348 instead of direction.
2349 (nios2_function_arg_padding): Likewise. Make static.
2350 (TARGET_FUNCTION_ARG_PADDING): Redefine.
2351 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
2352 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
2353 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
2354 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2355 (pa_function_arg_padding): Make static. Return pad_direction instead
2357 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
2358 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
2359 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
2360 instead of direction. Use targetm.calls.function_arg_padding.
2361 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
2362 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
2363 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
2364 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
2365 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
2367 (function_arg_padding): Rename to...
2368 (rs6000_function_arg_padding): ...this. Make static. Return
2369 pad_direction instead of direction.
2370 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
2371 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
2372 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
2373 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
2374 instead of direction. Use targetm.calls.function_arg_padding.
2375 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
2376 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
2377 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
2378 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
2379 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2380 (function_arg_padding): Rename to...
2381 (rs6000_function_arg_padding): ...this. Make static. Return
2382 pad_direction instead of direction.
2383 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
2384 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
2385 * config/s390/s390.c (s390_function_arg_padding): New function.
2386 (TARGET_FUNCTION_ARG_PADDING): Redefine.
2387 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
2388 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
2389 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2390 (function_arg_padding): Rename to...
2391 (sparc_function_arg_padding): ...this. Make static. Return
2392 pad_direction instead of direction.
2393 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
2394 * config/spu/spu.c (spu_function_arg_padding): New function.
2395 (TARGET_FUNCTION_ARG_PADDING): Redefine.
2396 * system.h (FUNCTION_ARG_PADDING): Poison.
2398 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2399 Alan Hayward <alan.hayward@arm.com>
2400 David Sherwood <david.sherwood@arm.com>
2402 * target.def (modes_tieable_p): New hook.
2403 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
2404 (TARGET_MODES_TIEABLE_P): ...this.
2405 * doc/tm.texi.in: Regenerate.
2406 * hooks.h (hook_bool_mode_mode_true): Declare.
2407 * hooks.c (hook_bool_mode_mode_true): New function.
2408 * combine.c (subst): Use targetm.modes_tieable_p instead of
2410 * dse.c (find_shift_sequence): Likewise.
2411 * expmed.c (extract_low_bits): Likewise.
2412 * lower-subreg.c: Include target.h.
2413 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
2415 * rtlanal.c (rtx_cost): Likewise.
2416 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
2417 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
2418 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
2419 (TARGET_MODES_TIEABLE_P): Redefine.
2420 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
2421 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
2422 (TARGET_MODES_TIEABLE_P): Redefine.
2423 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
2424 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
2425 (arc_modes_tieable_p): New function.
2426 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
2427 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
2428 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
2429 (arm_modes_tieable_p): Make static.
2430 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
2431 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
2432 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
2433 (TARGET_MODES_TIEABLE_P): Redefine.
2434 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
2435 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
2436 (TARGET_MODES_TIEABLE_P): Redefine.
2437 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
2438 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
2439 (cr16_modes_tieable_p): New function.
2440 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
2441 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
2442 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
2443 (TRULY_NOOP_TRUNCATION): Update comment.
2444 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
2445 (TRULY_NOOP_TRUNCATION): Update comment.
2446 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
2447 (frv_modes_tieable_p): New function.
2448 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
2449 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
2450 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
2451 (TARGET_MODES_TIEABLE_P): Redefine.
2452 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
2453 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
2454 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
2455 (TARGET_MODES_TIEABLE_P): Redefine.
2456 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
2457 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
2458 (ia64_modes_tieable_p): New function.
2459 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
2460 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
2461 (iq2000_modes_tieable_p): New function.
2462 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
2463 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
2464 (lm32_modes_tieable_p): New function.
2465 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
2466 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
2467 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
2468 (TARGET_MODES_TIEABLE_P): Redefine.
2469 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
2470 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
2471 (m32r_modes_tieable_p): New function.
2472 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
2473 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
2474 (m68k_modes_tieable_p): New function.
2475 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
2476 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
2477 (mcore_modes_tieable_p): New function.
2478 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
2479 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
2481 (TARGET_MODES_TIEABLE_P): Redefine.
2482 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
2483 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
2484 * config/mips/mips.c (mips_modes_tieable_p): Make static.
2485 (TARGET_MODES_TIEABLE_P): Redefine.
2486 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
2487 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
2488 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
2489 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
2490 (mn10300_modes_tieable_p): ...this and make static.
2491 (TARGET_MODES_TIEABLE_P): Redefine.
2492 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
2493 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
2494 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
2495 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
2496 (msp430_modes_tieable_p): Make static.
2497 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
2498 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
2499 (TARGET_MODES_TIEABLE_P): Redefine.
2500 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
2501 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
2502 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
2503 (TARGET_MODES_TIEABLE_P): Redefine.
2504 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
2505 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
2506 * config/pa/pa.c (pa_modes_tieable_p): Make static.
2507 (TARGET_MODES_TIEABLE_P): Redefine.
2508 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
2509 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
2510 (pdp11_modes_tieable_p): New function.
2511 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
2512 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
2513 (rs6000_modes_tieable_p): New function.
2514 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
2515 * config/powerpcspe/powerpcspe.md: Update comment.
2516 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
2517 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
2518 (TARGET_MODES_TIEABLE_P): Redefine.
2519 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
2520 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
2521 (rl78_modes_tieable_p): New function.
2522 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
2523 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
2524 (rs6000_modes_tieable_p): New function.
2525 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
2526 * config/rs6000/rs6000.md: Update comment.
2527 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
2528 * config/rx/rx.c (rx_modes_tieable_p): New function.
2529 (TARGET_MODES_TIEABLE_P): Redefine.
2530 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
2531 * config/s390/s390.c (s390_modes_tieable_p): New function.
2532 (TARGET_MODES_TIEABLE_P): Redefine.
2533 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
2534 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
2535 (sh_modes_tieable_p): New function.
2536 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
2537 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
2538 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
2539 (sparc_modes_tieable_p): Make static.
2540 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
2541 * config/spu/spu.c (spu_modes_tieable_p): New function.
2542 (TARGET_MODES_TIEABLE_P): Redefine.
2543 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
2544 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
2545 (TARGET_MODES_TIEABLE_P): Redefine.
2546 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
2547 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
2548 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
2549 * config/v850/v850.c (v850_modes_tieable_p): New function.
2550 (TARGET_MODES_TIEABLE_P): Redefine.
2551 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
2552 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
2553 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
2554 (visium_modes_tieable_p): New function.
2555 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
2556 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
2557 (xtensa_modes_tieable_p): New function.
2558 * system.h (MODES_TIEABLE_P): Poison.
2560 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2561 Alan Hayward <alan.hayward@arm.com>
2562 David Sherwood <david.sherwood@arm.com>
2564 * target.def (hard_regno_mode_ok): New hook.
2565 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
2566 (TARGET_HARD_REGNO_MODE_OK): ...this.
2567 * doc/tm.texi.in: Regenerate.
2568 * hooks.h (hook_bool_uint_mode_true): Declare.
2569 * hooks.c (hook_bool_uint_mode_true): New function.
2570 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
2572 * genpreds.c (write_insn_preds_c): Add an include of target.h.
2573 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
2574 instead of HARD_REGNO_MODE_OK.
2575 * caller-save.c: Include target.h.
2576 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
2578 * combine.c (can_combine_p): Likewise.
2579 (combinable_i3pat): Likewise.
2580 (can_change_dest_mode): Likewise.
2581 * expr.c (init_expr_target): Likewise.
2582 (convert_move): Likewise.
2583 (convert_modes): Likewise.
2584 * ira.c (setup_prohibited_class_mode_regs): Likewise.
2585 (setup_prohibited_mode_move_regs): Likewise.
2586 * ira.h (target_ira): Likewise.
2587 * lra-assigns.c (find_hard_regno_for_1): Likewise.
2588 * lra-constraints.c (process_alt_operands): Likewise.
2589 (split_reg): Likewise.
2590 * recog.c (peep2_find_free_register): Likewise.
2591 * ree.c (combine_reaching_defs): Likewise.
2592 * regcprop.c (maybe_mode_change): Likewise.
2593 * reginfo.c (init_reg_sets_1): Likewise.
2594 (choose_hard_reg_mode): Likewise.
2595 (simplifiable_subregs): Likewise.
2596 * regrename.c (check_new_reg_p): Likewise.
2597 * reload.c (find_valid_class): Likewise.
2598 (find_valid_class_1): Likewise.
2599 (reload_inner_reg_of_subreg): Likewise.
2600 (push_reload): Likewise.
2601 (combine_reloads): Likewise.
2602 (find_dummy_reload): Likewise.
2603 (find_reloads): Likewise.
2604 * reload1.c (find_reg): Likewise.
2605 (set_reload_reg): Likewise.
2606 (allocate_reload_reg): Likewise.
2607 (choose_reload_regs): Likewise.
2608 (reload_adjust_reg_for_temp): Likewise.
2609 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
2610 (simplify_subreg_regno): Likewise.
2611 * sel-sched.c (init_regs_for_mode): Likewise.
2612 * varasm.c (make_decl_rtl): Likewise.
2613 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
2614 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
2616 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
2617 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
2618 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2619 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
2620 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
2621 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2622 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
2623 (arc_mode_class): Delete.
2624 (HARD_REGNO_MODE_OK): Delete.
2625 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2626 (arc_hard_regno_mode_ok): Rename old array to...
2627 (arc_hard_regno_mode_ok_modes): ...this.
2628 (arc_conditional_register_usage): Update accordingly.
2629 (arc_mode_class): Make static.
2630 (arc_hard_regno_mode_ok): New function.
2631 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
2632 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
2633 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2634 (arm_hard_regno_mode_ok): Make static.
2635 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
2637 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
2638 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
2639 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
2641 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2642 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
2643 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
2644 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
2645 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
2646 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2647 * config/bfin/predicates.md (valid_reg_operand): Use
2648 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
2649 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
2650 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
2651 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2652 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
2653 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
2654 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2655 (cr16_hard_regno_mode_ok): Make static and return a bool.
2656 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
2657 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2658 (cris_hard_regno_mode_ok): New function.
2659 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
2660 (epiphany_mode_class): Delete.
2661 (HARD_REGNO_MODE_OK): Delete.
2662 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
2663 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2664 (hard_regno_mode_ok): Rename to...
2665 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
2666 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
2667 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
2669 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
2670 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
2671 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2672 (frv_hard_regno_mode_ok): Make static and return a bool.
2673 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
2675 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
2676 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
2677 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
2678 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
2680 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2681 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
2682 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
2683 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
2685 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2686 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
2687 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2688 (ia64_hard_regno_mode_ok): New function.
2689 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
2690 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2691 (iq2000_hard_regno_mode_ok): New function.
2692 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
2693 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2694 (lm32_hard_regno_mode_ok): New function.
2695 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
2696 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
2697 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
2698 instead of HARD_REGNO_MODE_OK.
2699 (m32c_hard_regno_ok): Rename to...
2700 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
2701 (m32c_cannot_change_mode_class): Update accordingly.
2702 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2703 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
2704 (m32r_mode_class): Delete.
2705 (HARD_REGNO_MODE_OK): Delete.
2706 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2707 (m32r_hard_regno_mode_ok): Rename to...
2708 (m32r_hard_regno_modes): ...this.
2709 (m32r_mode_class): Make static.
2710 (m32r_hard_regno_mode_ok): New function.
2711 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
2712 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
2713 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2714 (m68k_hard_regno_mode_ok): Make static.
2715 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
2716 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2717 (mcore_hard_regno_mode_ok): New function.
2718 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
2719 (HARD_REGNO_MODE_OK): Delete.
2720 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
2722 (microblaze_hard_regno_mode_ok_p): ...this and make static.
2723 (microblaze_hard_regno_mode_ok): New function.
2724 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2725 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
2726 (mips_hard_regno_mode_ok): Delete.
2727 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
2728 (mips_hard_regno_mode_ok_p): ...this and make static.
2729 (mips_hard_regno_mode_ok_p): Rename to...
2730 (mips_hard_regno_mode_ok_uncached): ...this.
2731 (mips_hard_regno_mode_ok): New function.
2732 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
2733 of HARD_REGNO_MODE_OK.
2734 (mips_option_override): Update after above name changes.
2735 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2736 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
2737 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
2738 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
2739 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
2740 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2741 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
2742 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
2743 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
2744 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2745 (msp430_hard_regno_mode_ok): Make static and return a bool.
2746 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
2747 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
2748 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
2750 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2751 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
2752 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
2753 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
2754 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
2755 (PA_HARD_REGNO_MODE_OK): ...this
2756 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
2757 (PA_HARD_REGNO_MODE_OK): ...this.
2758 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2759 (pa_hard_regno_mode_ok): New function.
2760 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
2761 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2762 (pdp11_hard_regno_mode_ok): New function.
2763 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
2764 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
2766 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
2768 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2769 (rs6000_hard_regno_mode_ok): Rename to...
2770 (rs6000_hard_regno_mode_ok_uncached): ...this.
2771 (rs6000_init_hard_regno_mode_ok): Update accordingly.
2772 (rs6000_hard_regno_mode_ok): New function.
2773 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
2774 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
2775 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
2776 (riscv_hard_regno_mode_ok): ...this and make static.
2777 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2778 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
2779 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
2780 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2781 (rl78_hard_regno_mode_ok): Make static and return bool.
2782 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
2783 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
2785 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
2786 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2787 (rs6000_hard_regno_mode_ok): Rename to...
2788 (rs6000_hard_regno_mode_ok_uncached): ...this.
2789 (rs6000_init_hard_regno_mode_ok): Update accordingly.
2790 (rs6000_hard_regno_mode_ok): New function.
2791 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
2792 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
2793 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2794 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
2795 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
2796 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
2797 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2798 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
2799 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
2800 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2801 (sh_hard_regno_mode_ok): Make static.
2802 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
2803 instead of HARD_REGNO_MODE_OK.
2804 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
2805 (sparc_mode_class): Delete.
2806 (HARD_REGNO_MODE_OK): Delete.
2807 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2808 (hard_regno_mode_classes): Make static.
2809 (sparc_mode_class): Likewise.
2810 (sparc_hard_regno_mode_ok): New function.
2811 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
2812 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
2813 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
2815 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2816 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
2817 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
2818 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
2819 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
2820 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2821 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
2822 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
2823 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2824 (visium_hard_regno_mode_ok): New function.
2825 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
2826 instead of HARD_REGNO_MODE_OK.
2827 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
2828 (HARD_REGNO_MODE_OK): Delete.
2829 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
2830 (xtensa_hard_regno_mode_ok_p): ...this and make static.
2831 (xtensa_option_override): Update accordingly.
2832 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2833 (xtensa_hard_regno_mode_ok): New function.
2834 * system.h (HARD_REGNO_MODE_OK): Poison.
2836 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2837 Alan Hayward <alan.hayward@arm.com>
2838 David Sherwood <david.sherwood@arm.com>
2840 * target.def (hard_regno_call_part_clobbered): New hook.
2841 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
2842 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
2843 * doc/tm.texi: Regenerate.
2844 * hooks.h (hook_bool_uint_mode_false): Declare.
2845 * hooks.c (hook_bool_uint_mode_false): New function.
2846 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2847 * cselib.c (cselib_process_insn): Use
2848 targetm.hard_regno_call_part_clobbered instead of
2849 HARD_REGNO_CALL_PART_CLOBBERED.
2850 * ira-conflicts.c (ira_build_conflicts): Likewise.
2851 * ira-costs.c (ira_tune_allocno_costs): Likewise.
2852 * lra-constraints.c (need_for_call_save_p): Likewise.
2853 * lra-lives.c: Include target.h.
2854 (check_pseudos_live_through_calls): Use
2855 targetm.hard_regno_call_part_clobbered instead of
2856 HARD_REGNO_CALL_PART_CLOBBERED.
2857 * regcprop.c: Include target.h.
2858 (copyprop_hardreg_forward_1): Use
2859 targetm.hard_regno_call_part_clobbered instead of
2860 HARD_REGNO_CALL_PART_CLOBBERED.
2861 * reginfo.c (choose_hard_reg_mode): Likewise.
2862 * regrename.c (check_new_reg_p): Likewise.
2863 * reload.c (find_equiv_reg): Likewise.
2864 * reload1.c (emit_reload_insns): Likewise.
2865 * sched-deps.c (deps_analyze_insn): Likewise.
2866 * sel-sched.c (init_regs_for_mode): Likewise.
2867 (mark_unavailable_hard_regs): Likewise.
2868 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
2869 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2870 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
2872 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2873 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2874 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
2876 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
2878 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2879 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2880 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
2882 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2883 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2884 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
2886 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2887 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
2889 * config/powerpcspe/powerpcspe.c
2890 (rs6000_hard_regno_call_part_clobbered): New function.
2891 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2892 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2893 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
2895 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2896 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2897 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
2899 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2900 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2901 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
2903 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2904 Alan Hayward <alan.hayward@arm.com>
2905 David Sherwood <david.sherwood@arm.com>
2907 * rtl.h (subreg_memory_offset): Declare.
2908 * emit-rtl.c (subreg_memory_offset): New function.
2909 * expmed.c (store_bit_field_1): Use it.
2910 * expr.c (undefined_operand_subword_p): Likewise.
2911 * simplify-rtx.c (simplify_subreg): Likewise.
2913 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
2915 PR rtl-optimization/57448
2918 * optabs.c (expand_atomic_load): Place compiler memory barriers if
2919 using atomic_load pattern.
2920 (expand_atomic_store): Likewise.
2922 2017-09-04 Jakub Jelinek <jakub@redhat.com>
2925 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
2926 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
2927 handling. Use replace_call_with_value with NULL instead of
2928 gsi_replace, unlink_stmt_vdef and release_defs.
2930 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
2933 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
2935 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2938 * rtl.h (emit_library_call_value_1): Declare.
2939 (emit_library_call): Replace declaration with a series of overloads.
2940 Remove the parameter count argument.
2941 (emit_library_call_value): Likewise.
2942 * calls.c (emit_library_call_value_1): Make global. Replace varargs
2943 with an "rtx_mode_t *".
2944 (emit_library_call_value): Delete.
2945 (emit_library_call): Likewise.
2946 * asan.c (asan_emit_stack_protection): Update calls accordingly.
2947 (asan_emit_allocas_unpoison): Likewise.
2948 * builtins.c (expand_builtin_powi): Likewise.
2949 (expand_asan_emit_allocas_unpoison): Likewise.
2950 * cfgexpand.c (expand_main_function): Likewise.
2951 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
2952 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
2953 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
2954 * config/arm/arm.c (arm_trampoline_init): Likewise.
2955 (arm_call_tls_get_addr): Likewise.
2956 (arm_expand_divmod_libfunc): Likewise.
2957 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
2958 (smulsi3_highpart): Likewise.
2959 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
2960 (c6x_expand_compare): Likewise.
2961 (c6x_expand_movmem): Likewise.
2962 * config/frv/frv.c (frv_trampoline_init): Likewise.
2963 * config/i386/i386.c (ix86_trampoline_init): Likewise.
2964 (ix86_expand_divmod_libfunc): Likewise.
2965 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
2966 (ia64_expand_compare): Likewise.
2967 (ia64_profile_hook): Likewise.
2968 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
2969 (nonlocal_goto): Likewise.
2970 (restore_stack_nonlocal): Likewise.
2971 * config/m32r/m32r.c (block_move_call): Likewise.
2972 (m32r_trampoline_init): Likewise.
2973 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
2974 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
2975 (m68k_call_m68k_read_tp): Likewise.
2976 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
2977 (microblaze_expand_divide): Likewise.
2978 * config/mips/mips.h (mips_args): Likewise.
2979 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
2980 (MIPS_ICACHE_SYNC): Likewise.
2981 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
2982 (nios2_trampoline_init): Likewise.
2983 * config/pa/pa.c (hppa_tls_call): Likewise.
2984 (pa_trampoline_init): Likewise.
2985 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
2986 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
2987 (expand_strn_compare): Likewise.
2988 (rs6000_generate_compare): Likewise.
2989 (rs6000_expand_float128_convert): Likewise.
2990 (output_profile_hook): Likewise.
2991 (rs6000_trampoline_init): Likewise.
2992 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
2993 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
2994 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
2995 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
2996 (rs6000_generate_compare): Likewise.
2997 (rs6000_expand_float128_convert): Likewise.
2998 (output_profile_hook): Likewise.
2999 (rs6000_trampoline_init): Likewise.
3000 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
3001 * config/sh/sh.c (sh_trampoline_init): Likewise.
3002 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
3003 (sparc_emit_float_lib_cmp): Likewise.
3004 (sparc32_initialize_trampoline): Likewise.
3005 (sparc64_initialize_trampoline): Likewise.
3006 (sparc_profile_hook): Likewise.
3007 * config/spu/spu.c (ea_load_store): Likewise.
3008 * config/spu/spu.md (floatunssidf2): Likewise.
3009 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
3010 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
3011 * config/visium/visium.c (expand_block_move_4): Likewise.
3012 (expand_block_move_2): Likewise.
3013 (expand_block_move_1): Likewise.
3014 (expand_block_set_4): Likewise.
3015 (expand_block_set_2): Likewise.
3016 (expand_block_set_1): Likewise.
3017 (visium_trampoline_init): Likewise.
3018 (visium_profile_hook): Likewise.
3019 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
3020 (xtensa_setup_frame_addresses): Likewise.
3021 (xtensa_trampoline_init): Likewise.
3022 * except.c (sjlj_emit_function_enter): Likewise.
3023 (sjlj_emit_function_exit): Likewise.
3024 * explow.c (allocate_dynamic_stack_space): Likewise.
3025 (probe_stack_range): Likewise.
3026 * expr.c (convert_mode_scalar): Likewise.
3027 * optabs.c (expand_binop): Likewise.
3028 (expand_twoval_binop_libfunc): Likewise.
3029 (expand_unop): Likewise.
3030 (prepare_cmp_insn): Likewise.
3031 (prepare_float_lib_cmp): Likewise.
3032 (expand_float): Likewise.
3033 (expand_fix): Likewise.
3034 (expand_fixed_convert): Likewise.
3035 (maybe_emit_sync_lock_test_and_set): Likewise.
3036 (expand_atomic_compare_and_swap): Likewise.
3037 (expand_mem_thread_fence): Likewise.
3038 (expand_atomic_fetch_op): Likewise.
3040 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
3042 * doc/generic.texi (OpenACC): Adjust URL.
3043 * doc/invoke.texi (C Dialect Options): Ditto.
3045 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
3047 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
3048 predicate for operand 1. Add (m,<S>) constraint.
3049 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
3050 Prevent memory operand 1 with register operand 2.
3052 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
3054 PR rtl-optimization/82024
3055 * combine.c (try_combine): If the combination result is a PARALLEL,
3056 and we only need to retain the SET in there that would be placed
3057 at I2, check that we can place that at I3 instead, before doing so.
3059 2017-09-01 Jakub Jelinek <jakub@redhat.com>
3062 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
3064 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
3065 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
3068 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
3069 Jeff Law <law@redhat.com>
3071 * varasm.c (bss_initializer_p): Do not put constants into .bss
3072 (categorize_decl_for_section): Handle bss_initializer_p returning
3073 false when DECL_INITIAL is NULL.
3075 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3078 * config/s390/s390.c (s390_can_inline_p): New function.
3080 2017-09-01 Jeff Law <law@redhat.com>
3082 PR tree-optimization/82052
3083 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
3084 Always initialize the returned slot after a hash table miss
3085 when INSERT is true.
3087 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
3089 * config/s390/s390.md (mem_signal_fence): Remove.
3090 * doc/md.texi (mem_signal_fence): Remove.
3091 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
3093 * target-insns.def (mem_signal_fence): Remove.
3095 2017-09-01 Jakub Jelinek <jakub@redhat.com>
3098 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
3101 * asan.c (create_odr_indicator): Strip name encoding from assembler
3102 name before appending it after __odr_asan_.
3104 2017-09-01 Martin Liska <mliska@suse.cz>
3106 PR tree-optimization/82059
3107 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
3108 frequency only when an edge is redirected.
3110 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
3112 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
3113 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
3114 (arc_conditional_register_usage): Remove ARC600 lp_count
3116 (arc_file_start): Emit Tag_ARC_CPU_variation.
3117 (arc_can_use_doloop_p): New conditions to use ZOLs.
3118 (hwloop_fail): New function.
3119 (hwloop_optimize): Likewise.
3120 (hwloop_pattern_reg): Likewise.
3121 (arc_doloop_hooks): New struct, to be used with reorg_loops.
3122 (arc_reorg_loops): New function, calls reorg_loops.
3123 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
3124 (arc600_corereg_hazard): Remove ZOL checking, case handled by
3126 (arc_loop_hazard): Remove function, functionality moved into
3128 (arc_hazard): Remove arc_loop_hazard call.
3129 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
3130 into hwloop_optimize.
3131 (arc_label_align): Remove ZOL handling.
3132 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
3133 * config/arc/arc.md (doloop_begin): Remove pattern.
3134 (doloop_begin_i): Likewise.
3135 (doloop_end_i): Likewise.
3136 (doloop_fallback): Likewise.
3137 (doloop_fallback_m): Likewise.
3138 (doloop_end): Reimplement expand.
3139 (arc_lp): New pattern for LP instruction.
3140 (loop_end): New pattern.
3141 (loop_fail): Likewise.
3142 (decrement_and_branch_until_zero): Likewise.
3143 * config/arc/arc.opt (mlpc-width): New option.
3144 * doc/invoke.texi (mlpc-width): Document option.
3146 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
3148 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
3149 (arc_ccfsm_advance): Fix checking for delay slots.
3150 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
3152 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
3154 * config/arc/arc.md (movqi_insn): Add stores to save constant long
3156 (movhi_insn): Update store instruction constraint which are saving
3157 6-bit short immediates.
3158 (movsi_insn): Consider also short scaled load operations.
3159 (zero_extendhisi2_i): Use Usd constraint instead of T.
3160 (extendhisi2_i): Add q constraint.
3161 (arc_clzsi2): Add type and length attributes.
3162 (arc_ctzsi2): Likewise.
3163 * config/arc/constraints.md (Usc): Update constraint, the
3164 assembler can parse two relocations for a single instruction.
3166 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
3168 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
3169 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
3171 2017-08-31 Olivier Hainque <hainque@adacore.com>
3173 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
3174 match as powerpc-wrs-vxworks*.
3176 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
3178 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
3179 register constraint for by-element operand.
3180 (aarch64_mls_elt_merge<mode>): Likewise.
3182 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
3184 * config/arc/arc.c (arc_can_follow_jump): Check for short
3187 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
3189 * config.gcc: Use g.opt for arc.
3190 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
3191 functionality moved to ...
3192 (legitimate_scaled_address_p): New function, ...here.
3193 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
3194 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
3195 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
3197 (arc_override_options): Handle G option.
3198 (arc_output_pic_addr_const): Correct function definition.
3199 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
3200 (arc_decl_anon_ns_mem_p): Delete.
3201 (arc_in_small_data_p): Overhaul this function to take into
3202 consideration the value given via G option.
3203 (arc_rewrite_small_data_1): Renamed and corrected old
3204 arc_rewrite_small_data function.
3205 (arc_rewrite_small_data): New function.
3206 (small_data_pattern): Don't use pic_offset_table_rtx.
3207 * config/arc/arc.h (CC1_SPEC): Recognize G option.
3208 * config/arc/simdext.md (movmisalignv2hi): Use
3209 prepare_move_operands function.
3211 (movmisalign*): Likewise.
3212 * doc/invoke.texi (ARC options): Document -G option.
3214 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
3216 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
3218 * config/arc/arc.c (arc_print_operand): Output scalled address for
3219 sdata whenever is possible.
3220 (arc_in_small_data_p): Allow sdata for 64bit datum when double
3221 load/stores are available.
3222 (compact_sda_memory_operand): Check for the alignment required by
3223 code density instructions.
3224 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
3226 * config/arc/constraints.md (Usd): Update constraint.
3227 (Us0): New constraint.
3228 (Usc): Update constraint.
3230 2017-08-31 Richard Biener <rguenther@suse.de>
3233 * dwarf2out.c (dwarf2out_early_global_decl): Process each
3236 2017-08-31 Tamar Christina <tamar.christina@arm.com>
3238 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
3239 Resize type_signature.
3241 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
3242 Alan Hayward <alan.hayward@arm.com>
3243 David Sherwood <david.sherwood@arm.com>
3245 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
3246 subregs whose inner modes can be stored in GPRs.
3247 (aarch64_classify_index): Likewise.
3249 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
3250 Alan Hayward <alan.hayward@arm.com>
3251 David Sherwood <david.sherwood@arm.com>
3253 * config/aarch64/iterators.md (V_cmp_result): Rename to...
3254 (V_INT_EQUIV): ...this.
3255 (v_cmp_result): Rename to...
3256 (v_int_equiv): ...this.
3257 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
3258 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
3259 (copysign<mode>3): Likewise.
3260 (aarch64_simd_bsl<mode>_internal): Likewise.
3261 (aarch64_simd_bsl<mode>): Likewise.
3262 (vec_cmp<mode><mode>): Likewise.
3263 (vcond<mode><mode>): Likewise.
3264 (vcond<v_cmp_mixed><mode>): Likewise.
3265 (vcondu<mode><v_cmp_mixed>): Likewise.
3266 (aarch64_cm<optab><mode>): Likewise.
3267 (aarch64_cmtst<mode>): Likewise.
3268 (aarch64_fac<optab><mode>): Likewise.
3269 (vec_perm_const<mode>): Likewise.
3270 (vcond_mask_<mode><v_cmp_result>): Rename to...
3271 (vcond_mask_<mode><v_int_equiv>): ...this.
3272 (vec_cmp<mode><v_cmp_result>): Rename to...
3273 (vec_cmp<mode><v_int_equiv>): ...this.
3275 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
3276 Alan Hayward <alan.hayward@arm.com>
3277 David Sherwood <david.sherwood@arm.com>
3279 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
3281 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
3282 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
3283 (UNSPEC_LD4_DREG): New unspecs.
3284 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
3285 (aarch64_ld2<mode>_dreg_be): Replace with...
3286 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
3288 (aarch64_ld3<mode>_dreg_le)
3289 (aarch64_ld3<mode>_dreg_be): Replace with...
3290 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
3292 (aarch64_ld4<mode>_dreg_le)
3293 (aarch64_ld4<mode>_dreg_be): Replace with...
3294 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
3297 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3299 PR tree-optimization/81987
3300 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
3301 insert an initializer in a location not dominated by the stride
3304 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
3306 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
3307 on the entire header of the finally block in the fallthru case.
3309 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
3311 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
3313 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
3315 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
3316 rs6000_emit_move_from_cr and call renamed function.
3317 (rs6000_emit_prologue): Call renamed functions.
3318 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
3319 movesi_from_cr, remove volatile CRs.
3321 2017-08-30 Jon Beniston <jon@beniston.com>
3322 Richard Biener <rguenther@suse.de>
3324 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
3325 of VECTOR_MODE_P check.
3326 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
3327 element vector types.
3329 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3331 * df.h (df_read_modify_subreg_p): Remove in favor of...
3332 * rtl.h (read_modify_subreg_p): ...this new function. Take a
3333 const_rtx instead of an rtx.
3334 * cprop.c (local_cprop_find_used_regs): Update accordingly.
3335 * df-problems.c (df_word_lr_mark_ref): Likewise.
3336 * ira-lives.c (mark_pseudo_reg_live): Likewise.
3337 (mark_pseudo_reg_dead): Likewise.
3338 (mark_ref_dead): Likewise.
3339 * reginfo.c (init_subregs_of_mode): Likewise.
3340 * sched-deps.c (sched_analyze_1): Likewise.
3341 * df-scan.c (df_def_record_1): Likewise.
3342 (df_uses_record): Likewise.
3343 (df_read_modify_subreg_p): Remove in favor of...
3344 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
3345 const_rtx instead of an rtx.
3347 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3348 Alan Hayward <alan.hayward@arm.com>
3349 David Sherwood <david.sherwood@arm.com>
3351 * rtl.h (partial_subreg_p): New function.
3352 * caller-save.c (save_call_clobbered_regs): Use it.
3353 * calls.c (expand_call): Likewise.
3354 * combine.c (combinable_i3pat): Likewise.
3355 (simplify_set): Likewise.
3356 (make_extraction): Likewise.
3357 (make_compound_operation_int): Likewise.
3358 (gen_lowpart_or_truncate): Likewise.
3359 (force_to_mode): Likewise.
3360 (make_field_assignment): Likewise.
3361 (reg_truncated_to_mode): Likewise.
3362 (record_truncated_value): Likewise.
3363 (move_deaths): Likewise.
3364 * cse.c (record_jump_cond): Likewise.
3365 (cse_insn): Likewise.
3366 * cselib.c (cselib_lookup_1): Likewise.
3367 * expmed.c (extract_bit_field_using_extv): Likewise.
3368 * function.c (assign_parm_setup_reg): Likewise.
3369 * ifcvt.c (noce_convert_multiple_sets): Likewise.
3370 * ira-build.c (create_insn_allocnos): Likewise.
3371 * lra-coalesce.c (merge_pseudos): Likewise.
3372 * lra-constraints.c (match_reload): Likewise.
3373 (simplify_operand_subreg): Likewise.
3374 (curr_insn_transform): Likewise.
3375 * lra-lives.c (process_bb_lives): Likewise.
3376 * lra.c (new_insn_reg): Likewise.
3377 (lra_substitute_pseudo): Likewise.
3378 * regcprop.c (mode_change_ok): Likewise.
3379 (maybe_mode_change): Likewise.
3380 (copyprop_hardreg_forward_1): Likewise.
3381 * reload.c (push_reload): Likewise.
3382 (find_reloads): Likewise.
3383 (find_reloads_subreg_address): Likewise.
3384 * reload1.c (alter_reg): Likewise.
3385 (eliminate_regs_1): Likewise.
3386 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3388 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
3390 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
3391 back to if statements, including unpack.
3393 2017-08-30 Martin Liska <mliska@suse.cz>
3396 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
3398 (func_checker::compare_asm_inputs_outputs): ... this function.
3399 (func_checker::compare_gimple_asm): Use the function to compare
3400 also ASM constrains.
3401 * ipa-icf-gimple.h: Rename the function.
3403 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3404 Alan Hayward <alan.hayward@arm.com>
3405 David Sherwood <david.sherwood@arm.com>
3407 * coretypes.h (complex_mode): New type.
3408 * gdbhooks.py (build_pretty_printer): Handle it.
3409 * machmode.h (complex_mode): New class.
3410 (complex_mode::includes_p): New function.
3411 (is_complex_int_mode): Likewise.
3412 (is_complex_float_mode): Likewise.
3413 * genmodes.c (get_mode_class): Handle complex mode classes.
3414 * function.c (expand_function_end): Use is_complex_int_mode.
3416 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3417 Alan Hayward <alan.hayward@arm.com>
3418 David Sherwood <david.sherwood@arm.com>
3420 * coretypes.h (scalar_mode_pod): New typedef.
3421 * gdbhooks.py (build_pretty_printer): Handle it.
3422 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
3423 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
3424 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
3425 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
3428 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3429 Alan Hayward <alan.hayward@arm.com>
3430 David Sherwood <david.sherwood@arm.com>
3432 * machmode.h (mode_for_vector): Take a scalar_mode instead
3434 * stor-layout.c (mode_for_vector): Likewise.
3435 * explow.c (promote_mode): Use as_a <scalar_mode>.
3436 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
3438 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3439 Alan Hayward <alan.hayward@arm.com>
3440 David Sherwood <david.sherwood@arm.com>
3442 * target.def (preferred_simd_mode): Take a scalar_mode
3443 instead of a machine_mode.
3444 * targhooks.h (default_preferred_simd_mode): Likewise.
3445 * targhooks.c (default_preferred_simd_mode): Likewise.
3446 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
3447 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
3448 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
3449 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
3450 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
3451 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
3452 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
3453 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
3455 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
3456 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
3457 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
3458 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
3459 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
3460 * doc/tm.texi: Regenerate.
3461 * optabs-query.c (can_vec_mask_load_store_p): Return false for
3464 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3465 Alan Hayward <alan.hayward@arm.com>
3466 David Sherwood <david.sherwood@arm.com>
3468 * target.def (scalar_mode_supported_p): Take a scalar_mode
3469 instead of a machine_mode.
3470 * targhooks.h (default_scalar_mode_supported_p): Likewise.
3471 * targhooks.c (default_scalar_mode_supported_p): Likewise.
3472 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
3473 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
3474 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
3475 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
3476 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
3477 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
3478 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
3479 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
3480 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
3481 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
3482 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
3483 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
3485 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
3486 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
3487 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
3488 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
3489 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
3491 * doc/tm.texi: Regenerate.
3493 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3494 Alan Hayward <alan.hayward@arm.com>
3495 David Sherwood <david.sherwood@arm.com>
3497 * coretypes.h (opt_scalar_mode): New typedef.
3498 * gdbhooks.py (build_pretty_printers): Handle it.
3499 * machmode.h (mode_iterator::get_2xwider): Add overload for
3501 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
3503 * expr.c (convert_mode_scalar): Likewise.
3504 * omp-low.c (omp_clause_aligned_alignment): Likewise.
3505 * optabs.c (expand_float): Likewise.
3506 (expand_fix): Likewise.
3507 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
3509 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3510 Alan Hayward <alan.hayward@arm.com>
3511 David Sherwood <david.sherwood@arm.com>
3513 * optabs.c (expand_float): Explicitly check for scalars before
3514 using a branching expansion.
3515 (expand_fix): Likewise.
3517 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3518 Alan Hayward <alan.hayward@arm.com>
3519 David Sherwood <david.sherwood@arm.com>
3521 * expr.c (convert_mode): Split scalar handling out into...
3522 (convert_mode_scalar): ...this new function. Treat the modes
3525 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3526 Alan Hayward <alan.hayward@arm.com>
3527 David Sherwood <david.sherwood@arm.com>
3529 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
3531 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
3533 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3534 Alan Hayward <alan.hayward@arm.com>
3535 David Sherwood <david.sherwood@arm.com>
3537 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
3538 rather than a machine_mode.
3539 (fixed_from_string): Likewise.
3540 (fixed_convert): Likewise.
3541 (fixed_convert_from_int): Likewise.
3542 (fixed_convert_from_real): Likewise.
3543 (real_convert_from_fixed): Likewise.
3544 * fixed-value.c (fixed_from_double_int): Likewise.
3545 (fixed_from_string): Likewise.
3546 (fixed_convert): Likewise.
3547 (fixed_convert_from_int): Likewise.
3548 (fixed_convert_from_real): Likewise.
3549 (real_convert_from_fixed): Likewise.
3550 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
3552 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3553 Alan Hayward <alan.hayward@arm.com>
3554 David Sherwood <david.sherwood@arm.com>
3556 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
3557 of separate mode class checks. Do not allow vector modes here.
3558 (immed_wide_int_const): Use as_a <scalar_mode>.
3559 * explow.c (trunc_int_for_mode): Likewise.
3560 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
3561 (wi::shwi): Likewise.
3562 (wi::min_value): Likewise.
3563 (wi::max_value): Likewise.
3564 * dwarf2out.c (loc_descriptor): Likewise.
3565 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
3568 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3569 Alan Hayward <alan.hayward@arm.com>
3570 David Sherwood <david.sherwood@arm.com>
3572 * tree.h (SCALAR_TYPE_MODE): New macro.
3573 * expr.c (expand_expr_addr_expr_1): Use it.
3574 (expand_expr_real_2): Likewise.
3575 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
3576 (fold_convert_const_fixed_from_int): Likewise.
3577 (fold_convert_const_fixed_from_real): Likewise.
3578 (native_encode_fixed): Likewise
3579 (native_encode_complex): Likewise
3580 (native_encode_vector): Likewise.
3581 (native_interpret_fixed): Likewise.
3582 (native_interpret_real): Likewise.
3583 (native_interpret_complex): Likewise.
3584 (native_interpret_vector): Likewise.
3585 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
3586 (simd_clone_adjust_argument_types): Likewise.
3587 (simd_clone_init_simd_arrays): Likewise.
3588 (simd_clone_adjust): Likewise.
3589 * stor-layout.c (layout_type): Likewise.
3590 * tree.c (build_minus_one_cst): Likewise.
3591 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
3592 * tree-inline.c (estimate_move_cost): Likewise.
3593 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
3594 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
3595 (vectorizable_reduction): Likewise.
3596 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
3597 (vect_recog_mixed_size_cond_pattern): Likewise.
3598 (check_bool_pattern): Likewise.
3599 (adjust_bool_pattern): Likewise.
3600 (search_type_for_mask_1): Likewise.
3601 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
3602 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
3603 (vectorizable_load): Likewise.
3604 (vectorizable_store): Likewise.
3605 * ubsan.c (ubsan_encode_value): Likewise.
3606 * varasm.c (output_constant): Likewise.
3608 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3609 Alan Hayward <alan.hayward@arm.com>
3610 David Sherwood <david.sherwood@arm.com>
3612 * coretypes.h (scalar_mode): New class.
3613 * machmode.h (scalar_mode): Likewise.
3614 (scalar_mode::includes_p): New function.
3615 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
3616 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
3617 * genmodes.c (get_mode_class): Handle remaining scalar modes.
3618 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
3619 * expmed.c (store_bit_field_1): Likewise.
3620 (extract_bit_field_1): Likewise.
3621 * expr.c (write_complex_part): Likewise.
3622 (read_complex_part): Likewise.
3623 (emit_move_complex_push): Likewise.
3624 (expand_expr_real_2): Likewise.
3625 * function.c (assign_parm_setup_reg): Likewise.
3626 (assign_parms_unsplit_complex): Likewise.
3627 * optabs.c (expand_binop): Likewise.
3628 * rtlanal.c (subreg_get_info): Likewise.
3629 * simplify-rtx.c (simplify_immed_subreg): Likewise.
3630 * varasm.c (output_constant_pool_2): Likewise.
3632 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3633 Alan Hayward <alan.hayward@arm.com>
3634 David Sherwood <david.sherwood@arm.com>
3636 * expmed.c (extract_high_half): Use scalar_int_mode and remove
3638 (expmed_mult_highpart_optab): Likewise.
3639 (expmed_mult_highpart): Likewise.
3641 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3642 Alan Hayward <alan.hayward@arm.com>
3643 David Sherwood <david.sherwood@arm.com>
3645 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
3646 instead of a machine_mode.
3647 (builtin_memset_read_str): Likewise.
3648 * builtins.c (c_readstr): Likewise.
3649 (builtin_memcpy_read_str): Likewise.
3650 (builtin_strncpy_read_str): Likewise.
3651 (builtin_memset_read_str): Likewise.
3652 (builtin_memset_gen_str): Likewise.
3653 (expand_builtin_signbit): Use scalar_int_mode for local variables.
3654 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
3655 instead of a machine_mode.
3656 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
3658 (make_extraction): Likewise.
3659 (try_widen_shift_mode): Take and return scalar_int_modes instead
3661 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
3662 a scalar_int_mode instead of a machine_mode.
3663 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
3664 (avr_addr_space_pointer_mode): Likewise.
3665 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
3666 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
3667 (msp430_unwind_word_mode): Likewise.
3668 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
3669 (spu_addr_space_pointer_mode): Likewise.
3670 (spu_addr_space_address_mode): Likewise.
3671 (spu_libgcc_cmp_return_mode): Likewise.
3672 (spu_libgcc_shift_count_mode): Likewise.
3673 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
3674 (rl78_addr_space_pointer_mode): Likewise.
3675 (fl78_unwind_word_mode): Likewise.
3676 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
3678 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
3679 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
3680 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
3681 (mips_valid_pointer_mode): Likewise.
3682 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
3683 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
3684 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
3686 (ft32_addr_space_address_mode): Likewise.
3687 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
3688 scalar_int_mode instead of a machine_mode.
3689 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
3691 (m32c_addr_space_address_mode): Likewise.
3692 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
3693 (rs6000_eh_return_filter_mode): Likewise.
3694 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
3695 (rs6000_eh_return_filter_mode): Likewise.
3696 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
3697 (s390_libgcc_shift_count_mode): Likewise.
3698 (s390_unwind_word_mode): Likewise.
3699 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
3701 * target.def (mode_rep_extended): Likewise.
3702 (valid_pointer_mode): Likewise.
3703 (addr_space.valid_pointer_mode): Likewise.
3704 (eh_return_filter_mode): Return a scalar_int_mode rather than
3706 (libgcc_cmp_return_mode): Likewise.
3707 (libgcc_shift_count_mode): Likewise.
3708 (unwind_word_mode): Likewise.
3709 (addr_space.pointer_mode): Likewise.
3710 (addr_space.address_mode): Likewise.
3711 * doc/tm.texi: Regenerate.
3712 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
3714 (do_jump): Use scalar_int_mode for local variables.
3715 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
3716 rather than a machine_mode.
3717 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
3718 (scompare_loc_descriptor_wide): Likewise.
3719 (scompare_loc_descriptor_narrow): Likewise.
3720 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
3722 * except.c (sjlj_emit_dispatch_table): Likewise.
3723 (expand_builtin_eh_copy_values): Likewise.
3724 * explow.c (convert_memory_address_addr_space_1): Likewise.
3725 Take a scalar_int_mode rather than a machine_mode.
3726 (convert_memory_address_addr_space): Take a scalar_int_mode rather
3727 than a machine_mode.
3728 (memory_address_addr_space): Use scalar_int_mode for local variables.
3729 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
3730 rather than a machine_mode.
3731 * expmed.c (mask_rtx): Likewise.
3732 (init_expmed_one_conv): Likewise.
3733 (expand_mult_highpart_adjust): Likewise.
3734 (extract_high_half): Likewise.
3735 (expmed_mult_highpart_optab): Likewise.
3736 (expmed_mult_highpart): Likewise.
3737 (expand_smod_pow2): Likewise.
3738 (expand_sdiv_pow2): Likewise.
3739 (emit_store_flag_int): Likewise.
3740 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
3742 (extract_low_bits): Likewise.
3743 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
3745 * expr.c (pieces_addr::adjust): Likewise.
3746 (can_store_by_pieces): Likewise.
3747 (store_by_pieces): Likewise.
3748 (clear_by_pieces_1): Likewise.
3749 (expand_expr_addr_expr_1): Likewise.
3750 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
3751 (expand_expr_real_1): Likewise.
3752 (try_casesi): Likewise.
3753 * final.c (shorten_branches): Likewise.
3754 * fold-const.c (fold_convert_const_int_from_fixed): Change the
3755 type of "mode" to machine_mode.
3756 * internal-fn.c (expand_arith_overflow_result_store): Take a
3757 scalar_int_mode rather than a machine_mode.
3758 (expand_mul_overflow): Use scalar_int_mode for local variables.
3759 * loop-doloop.c (doloop_modify): Likewise.
3760 (doloop_optimize): Likewise.
3761 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
3762 than a machine_mode.
3763 (expand_doubleword_shift_condmove): Likewise.
3764 (expand_doubleword_shift): Likewise.
3765 (expand_doubleword_clz): Likewise.
3766 (expand_doubleword_popcount): Likewise.
3767 (expand_doubleword_parity): Likewise.
3768 (expand_absneg_bit): Use scalar_int_mode for local variables.
3769 (prepare_float_lib_cmp): Likewise.
3770 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
3771 rather than a machine_mode.
3772 (convert_memory_address_addr_space): Likewise.
3773 (get_mode_bounds): Likewise.
3774 (get_address_mode): Return a scalar_int_mode rather than a
3776 * rtlanal.c (get_address_mode): Likewise.
3777 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
3778 than a machine_mode.
3779 * targhooks.c (default_mode_rep_extended): Likewise.
3780 (default_valid_pointer_mode): Likewise.
3781 (default_addr_space_valid_pointer_mode): Likewise.
3782 (default_eh_return_filter_mode): Return a scalar_int_mode rather
3783 than a machine_mode.
3784 (default_libgcc_cmp_return_mode): Likewise.
3785 (default_libgcc_shift_count_mode): Likewise.
3786 (default_unwind_word_mode): Likewise.
3787 (default_addr_space_pointer_mode): Likewise.
3788 (default_addr_space_address_mode): Likewise.
3789 * targhooks.h (default_eh_return_filter_mode): Likewise.
3790 (default_libgcc_cmp_return_mode): Likewise.
3791 (default_libgcc_shift_count_mode): Likewise.
3792 (default_unwind_word_mode): Likewise.
3793 (default_addr_space_pointer_mode): Likewise.
3794 (default_addr_space_address_mode): Likewise.
3795 (default_mode_rep_extended): Take a scalar_int_mode rather than
3797 (default_valid_pointer_mode): Likewise.
3798 (default_addr_space_valid_pointer_mode): Likewise.
3799 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
3801 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
3802 rather than a machine_mode.
3803 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
3804 for local variables.
3805 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
3806 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
3807 than a machine_mode.
3809 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3810 Alan Hayward <alan.hayward@arm.com>
3811 David Sherwood <david.sherwood@arm.com>
3813 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
3814 the mode argument to scalar_int_mode.
3815 (do_jump_by_parts_zero_rtx): Likewise.
3816 (do_jump_by_parts_equality_rtx): Likewise.
3817 (do_jump_by_parts_greater): Take a mode argument.
3818 (do_jump_by_parts_equality): Likewise.
3819 (do_jump_1): Update calls accordingly.
3821 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3822 Alan Hayward <alan.hayward@arm.com>
3823 David Sherwood <david.sherwood@arm.com>
3825 * is-a.h (safe_dyn_cast): New function.
3826 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
3827 (jump_table_for_label): Likewise.
3828 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
3829 instead of an rtx_insn *.
3830 (shorten_branches): Use dyn_cast instead of LABEL_P and
3831 JUMP_TABLE_DATA_P. Use jump_table_for_label and
3832 rtx_jump_table_data::get_data_mode.
3833 (final_scan_insn): Likewise.
3835 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3836 Alan Hayward <alan.hayward@arm.com>
3837 David Sherwood <david.sherwood@arm.com>
3839 * combine.c (try_combine): Use is_a <scalar_int_mode> when
3840 trying to combine a full-register integer set with a subreg
3843 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3844 Alan Hayward <alan.hayward@arm.com>
3845 David Sherwood <david.sherwood@arm.com>
3847 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
3848 that is always either address_mode or pointer_mode.
3850 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3851 Alan Hayward <alan.hayward@arm.com>
3852 David Sherwood <david.sherwood@arm.com>
3854 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
3855 when the two are known to be equal.
3857 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3858 Alan Hayward <alan.hayward@arm.com>
3859 David Sherwood <david.sherwood@arm.com>
3861 * simplify-rtx.c (simplify_const_unary_operation): Use
3862 is_a <scalar_int_mode> instead of checking for a nonzero
3863 precision. Forcibly convert op_mode to a scalar_int_mode
3864 in that case. More clearly differentiate the operand and
3865 result modes and use the former when deciding what the value
3866 of a count-bits operation should be. Use is_int_mode instead
3867 of checking for a MODE_INT. Remove redundant check for whether
3868 this mode has a zero precision.
3870 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3871 Alan Hayward <alan.hayward@arm.com>
3872 David Sherwood <david.sherwood@arm.com>
3874 * optabs.c (widen_leading): Change the type of the mode argument
3875 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
3876 (widen_bswap): Likewise.
3877 (expand_parity): Likewise.
3878 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
3879 (expand_ffs): Likewise.
3880 (epand_unop): Check for scalar integer modes before calling the
3883 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3884 Alan Hayward <alan.hayward@arm.com>
3885 David Sherwood <david.sherwood@arm.com>
3887 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
3889 (expand_expr_real_1): Update call accordingly.
3891 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3892 Alan Hayward <alan.hayward@arm.com>
3893 David Sherwood <david.sherwood@arm.com>
3895 * expmed.c (store_bit_field_using_insv): Add op0_mode and
3896 value_mode arguments. Use scalar_int_mode internally.
3897 (store_bit_field_1): Rename the new integer mode from imode
3898 to op0_mode and use it instead of GET_MODE (op0). Update calls
3899 to store_split_bit_field, store_bit_field_using_insv and
3900 store_fixed_bit_field.
3901 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
3902 Use scalar_int_mode internally. Use a bit count rather than a mode
3903 when calculating the largest bit size for get_best_mode.
3904 Update calls to store_split_bit_field and store_fixed_bit_field_1.
3905 (store_fixed_bit_field_1): Add mode and value_mode arguments.
3906 Remove assertion that OP0 has a scalar integer mode.
3907 (store_split_bit_field): Add op0_mode and value_mode arguments.
3908 Update calls to extract_fixed_bit_field.
3909 (extract_bit_field_using_extv): Add an op0_mode argument.
3910 Use scalar_int_mode internally.
3911 (extract_bit_field_1): Rename the new integer mode from imode to
3912 op0_mode and use it instead of GET_MODE (op0). Update calls to
3913 extract_split_bit_field, extract_bit_field_using_extv and
3914 extract_fixed_bit_field.
3915 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
3916 to extract_split_bit_field and extract_fixed_bit_field_1.
3917 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
3918 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
3920 (extract_split_bit_field): Add an op0_mode argument. Update call
3921 to extract_fixed_bit_field.
3923 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3924 Alan Hayward <alan.hayward@arm.com>
3925 David Sherwood <david.sherwood@arm.com>
3927 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
3928 * explow.c (hard_function_value): Likewise.
3929 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
3930 convert_to_mode call outside the loop.
3931 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
3932 for the mode iterator. Require the mode specified by max_pieces
3934 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
3936 (copy_blkmode_to_reg): Likewise.
3937 (set_storage_via_setmem): Likewise.
3938 * optabs.c (prepare_cmp_insn): Likewise.
3939 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
3940 * stor-layout.c (finish_bitfield_representative): Likewise.
3942 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3943 Alan Hayward <alan.hayward@arm.com>
3944 David Sherwood <david.sherwood@arm.com>
3946 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
3947 * expr.c (convert_move): Use them.
3948 (convert_modes): Likewise.
3949 (store_expr_with_bounds): Likewise.
3951 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3952 Alan Hayward <alan.hayward@arm.com>
3953 David Sherwood <david.sherwood@arm.com>
3955 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
3956 parameter for the mode of "x". Remove the "known_x", "known_mode"
3957 and "known_ret" arguments. Change the type of the mode argument
3959 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
3960 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
3961 (reg_num_sign_bit_copies_for_combine): Likewise.
3962 * rtlanal.c (nonzero_bits1): Likewise.
3963 (num_sign_bit_copies1): Likewise.
3964 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
3965 (reg_num_sign_bit_copies_general): Likewise.
3966 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
3967 (reg_nonzero_bits_general): Likewise.
3969 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3970 Alan Hayward <alan.hayward@arm.com>
3971 David Sherwood <david.sherwood@arm.com>
3973 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
3974 than in subroutines. Return 1 for non-integer modes.
3975 (cached_num_sign_bit_copies): Change the type of the mode parameter
3977 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
3978 classes. Handle CONST_INT_P first and then check whether X also
3979 has a scalar integer mode. Check the same thing for inner registers
3980 of a SUBREG and for values that are being extended or truncated.
3982 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3983 Alan Hayward <alan.hayward@arm.com>
3984 David Sherwood <david.sherwood@arm.com>
3986 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
3987 in subroutines. Return the mode mask for non-integer modes.
3988 (cached_nonzero_bits): Change the type of the mode parameter
3990 (nonzero_bits1): Likewise. Remove early exit for other mode
3991 classes. Handle CONST_INT_P first and then check whether X
3992 also has a scalar integer mode.
3994 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3995 Alan Hayward <alan.hayward@arm.com>
3996 David Sherwood <david.sherwood@arm.com>
3998 * expr.c (widest_int_mode_for_size): Make the comment match the code.
3999 Return a scalar_int_mode and assert that the size is greater than
4001 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
4002 (op_by_pieces_d::op_by_pieces_d): Likewise.
4003 (op_by_pieces_d::run): Likewise.
4004 (can_store_by_pieces): Likewise.
4006 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4007 Alan Hayward <alan.hayward@arm.com>
4008 David Sherwood <david.sherwood@arm.com>
4010 * combine.c (extract_left_shift): Add a mode argument and update
4012 (make_compound_operation_int): Change the type of the mode parameter
4013 to scalar_int_mode and update the call to extract_left_shift.
4015 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4016 Alan Hayward <alan.hayward@arm.com>
4017 David Sherwood <david.sherwood@arm.com>
4019 * combine.c (simplify_and_const_int): Change the type of the mode
4020 parameter to scalar_int_mode.
4021 (simplify_and_const_int_1): Likewise. Update recursive call.
4023 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4024 Alan Hayward <alan.hayward@arm.com>
4025 David Sherwood <david.sherwood@arm.com>
4027 * combine.c (simplify_compare_const): Check that the mode is a
4028 scalar_int_mode (rather than VOIDmode) before testing its
4030 (simplify_comparison): Move COMPARISON_P handling out of the
4031 loop and restrict the latter part of the loop to scalar_int_modes.
4032 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
4033 and when considering SUBREG_REGs. Use is_int_mode instead of
4034 checking GET_MODE_CLASS against MODE_INT.
4036 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4037 Alan Hayward <alan.hayward@arm.com>
4038 David Sherwood <david.sherwood@arm.com>
4040 * combine.c (try_widen_shift_mode): Move check for equal modes to...
4041 (simplify_shift_const_1): ...here. Use scalar_int_mode for
4042 shift_unit_mode and for modes involved in scalar shifts.
4044 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4045 Alan Hayward <alan.hayward@arm.com>
4046 David Sherwood <david.sherwood@arm.com>
4048 * combine.c (force_int_to_mode): New function, split out from...
4049 (force_to_mode): ...here. Keep xmode up-to-date and use it
4050 instead of GET_MODE (x).
4052 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4053 Alan Hayward <alan.hayward@arm.com>
4054 David Sherwood <david.sherwood@arm.com>
4056 * optabs-query.h (extraction_insn::struct_mode): Change type to
4057 opt_scalar_int_mode and update comment.
4058 (extraction_insn::field_mode): Change type to scalar_int_mode.
4059 (extraction_insn::pos_mode): Likewise.
4060 * combine.c (make_extraction): Update accordingly.
4061 * optabs-query.c (get_traditional_extraction_insn): Likewise.
4062 (get_optab_extraction_insn): Likewise.
4063 * recog.c (simplify_while_replacing): Likewise.
4064 * expmed.c (narrow_bit_field_mem): Change the type of the mode
4065 parameter to opt_scalar_int_mode.
4067 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4068 Alan Hayward <alan.hayward@arm.com>
4069 David Sherwood <david.sherwood@arm.com>
4071 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
4072 to a scalar_int_mode instead of a machine_mode.
4073 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
4074 (get_best_mode): Return a boolean and use a pointer argument to store
4075 the selected mode. Replace the limit mode parameter with a bit limit.
4076 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
4077 for the values returned by bit_field_mode_iterator::next_mode.
4078 (store_bit_field): Update call to get_best_mode.
4079 (store_fixed_bit_field): Likewise.
4080 (extract_fixed_bit_field): Likewise.
4081 * expr.c (optimize_bitfield_assignment_op): Likewise.
4082 * fold-const.c (optimize_bit_field_compare): Likewise.
4083 (fold_truth_andor_1): Likewise.
4084 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
4085 Update for new type of m_mode.
4086 (get_best_mode): As above.
4088 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4089 Alan Hayward <alan.hayward@arm.com>
4090 David Sherwood <david.sherwood@arm.com>
4092 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
4093 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
4094 (store_bit_field): Check is_a <scalar_int_mode> before calling
4095 strict_volatile_bitfield_p.
4096 (extract_bit_field): Likewise.
4098 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4099 Alan Hayward <alan.hayward@arm.com>
4100 David Sherwood <david.sherwood@arm.com>
4102 * target.def (cstore_mode): Return a scalar_int_mode.
4103 * doc/tm.texi: Regenerate.
4104 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
4105 * targhooks.h (default_cstore_mode): Likewise.
4106 * targhooks.c (default_cstore_mode): Likewise, using a forced
4108 * expmed.c (emit_cstore): Expect the target of the cstore to be
4111 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4112 Alan Hayward <alan.hayward@arm.com>
4113 David Sherwood <david.sherwood@arm.com>
4115 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
4117 (niter_desc): Likewise mode.
4118 (iv_analyze): Add a mode parameter.
4120 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
4121 and change its type to scalar_int_mode.
4122 * loop-iv.c: Update commentary at head of file.
4123 (iv_constant): Pass the mode paraeter before the rtx it describes
4124 and change its type to scalar_int_mode. Remove VOIDmode handling.
4125 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
4126 (iv_extend): Likewise.
4127 (shorten_into_mode): Likewise.
4128 (iv_add): Use scalar_int_mode.
4129 (iv_mult): Likewise.
4130 (iv_shift): Likewise.
4131 (canonicalize_iv_subregs): Likewise.
4132 (get_biv_step_1): Pass the outer_mode parameter before the rtx
4133 it describes and change its mode to scalar_int_mode. Also change
4134 the type of the returned inner_mode to scalar_int_mode.
4135 (get_biv_step): Likewise, turning outer_mode from a pointer
4136 into a direct parameter. Update call to get_biv_step_1.
4137 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
4138 iv_constant and get_biv_step.
4139 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
4140 and change its type to scalar_int_mode. Don't initialise iv->mode
4141 to VOIDmode and remove later checks for its still being VOIDmode.
4142 Update calls to iv_analyze_op and iv_analyze_expr. Check
4143 is_a <scalar_int_mode> when changing the mode under consideration.
4144 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
4145 Update call to iv_analyze_expr.
4146 (iv_analyze_op): Add a mode parameter. Reject subregs whose
4147 inner register is not also a scalar_int_mode. Update call to
4149 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
4150 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
4151 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
4152 separate mode class checks. Update calls to iv_analyze. Remove
4153 fix-up of VOIDmodes after iv_analyze_biv.
4154 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
4155 don't have a scalar_int_mode. Update call to biv_p.
4157 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4158 Alan Hayward <alan.hayward@arm.com>
4159 David Sherwood <david.sherwood@arm.com>
4161 * cfgexpand.c (convert_debug_memory_address): Use
4162 as_a <scalar_int_mode>.
4163 * combine.c (expand_compound_operation): Likewise.
4164 (make_extraction): Likewise.
4165 (change_zero_ext): Likewise.
4166 (simplify_comparison): Likewise.
4167 * cse.c (cse_insn): Likewise.
4168 * dwarf2out.c (minmax_loc_descriptor): Likewise.
4169 (mem_loc_descriptor): Likewise.
4170 (loc_descriptor): Likewise.
4171 * expmed.c (init_expmed_one_mode): Likewise.
4172 (synth_mult): Likewise.
4173 (emit_store_flag_1): Likewise.
4174 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
4175 of a comparison with size.
4176 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
4177 (reduce_to_bit_field_precision): Likewise.
4178 * function.c (expand_function_end): Likewise.
4179 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
4180 * loop-doloop.c (doloop_modify): Likewise.
4181 * optabs.c (expand_binop): Likewise.
4182 (expand_unop): Likewise.
4183 (expand_copysign_absneg): Likewise.
4184 (prepare_cmp_insn): Likewise.
4185 (maybe_legitimize_operand): Likewise.
4186 * recog.c (const_scalar_int_operand): Likewise.
4187 * rtlanal.c (get_address_mode): Likewise.
4188 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4189 (simplify_cond_clz_ctz): Likewise.
4190 * tree-nested.c (get_nl_goto_field): Likewise.
4191 * tree.c (build_vector_type_for_mode): Likewise.
4192 * var-tracking.c (use_narrower_mode): Likewise.
4194 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4195 Alan Hayward <alan.hayward@arm.com>
4196 David Sherwood <david.sherwood@arm.com>
4198 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
4199 * builtins.c (expand_builtin_signbit): Use it.
4200 * cfgexpand.c (expand_debug_expr): Likewise.
4201 * dojump.c (do_jump): Likewise.
4202 (do_compare_and_jump): Likewise.
4203 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
4204 * expmed.c (make_tree): Likewise.
4205 * expr.c (expand_expr_real_2): Likewise.
4206 (expand_expr_real_1): Likewise.
4207 (try_casesi): Likewise.
4208 * fold-const-call.c (fold_const_call_ss): Likewise.
4209 * fold-const.c (unextend): Likewise.
4210 (extract_muldiv_1): Likewise.
4211 (fold_single_bit_test): Likewise.
4212 (native_encode_int): Likewise.
4213 (native_encode_string): Likewise.
4214 (native_interpret_int): Likewise.
4215 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
4216 * internal-fn.c (expand_addsub_overflow): Likewise.
4217 (expand_neg_overflow): Likewise.
4218 (expand_mul_overflow): Likewise.
4219 (expand_arith_overflow): Likewise.
4220 * match.pd: Likewise.
4221 * stor-layout.c (layout_type): Likewise.
4222 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
4223 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
4224 * tree-ssanames.c (get_range_info): Likewise.
4225 * tree-switch-conversion.c (array_value_type) Likewise.
4226 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
4227 (vect_recog_divmod_pattern): Likewise.
4228 (vect_recog_mixed_size_cond_pattern): Likewise.
4229 * tree-vrp.c (extract_range_basic): Likewise.
4230 (simplify_float_conversion_using_ranges): Likewise.
4231 * tree.c (int_fits_type_p): Likewise.
4232 * ubsan.c (instrument_bool_enum_load): Likewise.
4233 * varasm.c (mergeable_string_section): Likewise.
4234 (narrowing_initializer_constant_valid_p): Likewise.
4235 (output_constant): Likewise.
4237 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4238 Alan Hayward <alan.hayward@arm.com>
4239 David Sherwood <david.sherwood@arm.com>
4241 * machmode.h (NARROWEST_INT_MODE): New macro.
4242 * expr.c (alignment_for_piecewise_move): Use it instead of
4243 GET_CLASS_NARROWEST_MODE (MODE_INT).
4244 (push_block): Likewise.
4245 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
4247 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
4249 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4250 Alan Hayward <alan.hayward@arm.com>
4251 David Sherwood <david.sherwood@arm.com>
4253 * postreload.c (move2add_valid_value_p): Change the type of the
4254 mode parameter to scalar_int_mode.
4255 (move2add_use_add2_insn): Add a mode parameter and use it instead
4257 (move2add_use_add3_insn): Likewise.
4258 (reload_cse_move2add): Update accordingly.
4260 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4261 Alan Hayward <alan.hayward@arm.com>
4262 David Sherwood <david.sherwood@arm.com>
4264 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
4266 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
4267 * optabs.c (expand_unop): Likewise.
4269 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4270 Alan Hayward <alan.hayward@arm.com>
4271 David Sherwood <david.sherwood@arm.com>
4273 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
4274 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
4275 (popcount_loc_descriptor): Likewise.
4276 (bswap_loc_descriptor): Likewise.
4277 (rotate_loc_descriptor): Likewise.
4278 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
4279 calling the functions above.
4281 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4282 Alan Hayward <alan.hayward@arm.com>
4283 David Sherwood <david.sherwood@arm.com>
4285 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
4287 (try_combine): Likewise.
4288 (simplify_if_then_else): Likewise.
4289 * cse.c (cse_insn): Likewise.
4290 * dwarf2out.c (mem_loc_descriptor): Likewise.
4291 * emit-rtl.c (gen_lowpart_common): Likewise.
4292 * simplify-rtx.c (simplify_truncation): Likewise.
4293 (simplify_binary_operation_1): Likewise.
4294 (simplify_const_relational_operation): Likewise.
4295 (simplify_ternary_operation): Likewise.
4296 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
4298 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4299 Alan Hayward <alan.hayward@arm.com>
4300 David Sherwood <david.sherwood@arm.com>
4302 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
4303 * reload.c (push_reload): Likewise.
4304 (find_reloads): Likewise.
4306 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4307 Alan Hayward <alan.hayward@arm.com>
4308 David Sherwood <david.sherwood@arm.com>
4310 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
4311 (make_compound_operation_int): Likewise.
4312 (change_zero_ext): Likewise.
4313 * expr.c (convert_move): Likewise.
4314 (convert_modes): Likewise.
4315 * fwprop.c (forward_propagate_subreg): Likewise.
4316 * loop-iv.c (get_biv_step_1): Likewise.
4317 * optabs.c (widen_operand): Likewise.
4318 * postreload.c (move2add_valid_value_p): Likewise.
4319 * recog.c (simplify_while_replacing): Likewise.
4320 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4321 (simplify_binary_operation_1): Likewise. Remove redundant
4322 mode equality check.
4324 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4325 Alan Hayward <alan.hayward@arm.com>
4326 David Sherwood <david.sherwood@arm.com>
4328 * combine.c (combine_simplify_rtx): Add checks for
4329 is_a <scalar_int_mode>.
4330 (simplify_if_then_else): Likewise.
4331 (make_field_assignment): Likewise.
4332 (simplify_comparison): Likewise.
4333 * ifcvt.c (noce_try_bitop): Likewise.
4334 * loop-invariant.c (canonicalize_address_mult): Likewise.
4335 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4337 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4338 Alan Hayward <alan.hayward@arm.com>
4339 David Sherwood <david.sherwood@arm.com>
4341 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
4342 is_a <scalar_int_mode> instead of != BLKmode.
4344 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4345 Alan Hayward <alan.hayward@arm.com>
4346 David Sherwood <david.sherwood@arm.com>
4348 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
4349 instead of != VOIDmode.
4350 * combine.c (if_then_else_cond): Likewise.
4351 (change_zero_ext): Likewise.
4352 * dwarf2out.c (mem_loc_descriptor): Likewise.
4353 (loc_descriptor): Likewise.
4354 * rtlanal.c (canonicalize_condition): Likewise.
4355 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
4357 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4358 Alan Hayward <alan.hayward@arm.com>
4359 David Sherwood <david.sherwood@arm.com>
4361 * simplify-rtx.c (simplify_binary_operation_1): Use
4362 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
4364 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4365 Alan Hayward <alan.hayward@arm.com>
4366 David Sherwood <david.sherwood@arm.com>
4368 * wide-int.h (int_traits<unsigned char>) New class.
4369 (int_traits<unsigned short>) Likewise.
4370 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
4371 Use GET_MODE_UNIT_PRECISION and remove redundant test for
4373 * combine.c (set_nonzero_bits_and_sign_copies): Use
4374 is_a <scalar_int_mode>.
4375 (find_split_point): Likewise.
4376 (combine_simplify_rtx): Likewise.
4377 (simplify_logical): Likewise.
4378 (expand_compound_operation): Likewise.
4379 (expand_field_assignment): Likewise.
4380 (make_compound_operation): Likewise.
4381 (extended_count): Likewise.
4382 (change_zero_ext): Likewise.
4383 (simplify_comparison): Likewise.
4384 * dwarf2out.c (scompare_loc_descriptor): Likewise.
4385 (ucompare_loc_descriptor): Likewise.
4386 (minmax_loc_descriptor): Likewise.
4387 (mem_loc_descriptor): Likewise.
4388 (loc_descriptor): Likewise.
4389 * expmed.c (init_expmed_one_mode): Likewise.
4390 * lra-constraints.c (lra_constraint_offset): Likewise.
4391 * optabs.c (prepare_libcall_arg): Likewise.
4392 * postreload.c (move2add_note_store): Likewise.
4393 * reload.c (operands_match_p): Likewise.
4394 * rtl.h (load_extend_op): Likewise.
4395 * rtlhooks.c (gen_lowpart_general): Likewise.
4396 * simplify-rtx.c (simplify_truncation): Likewise.
4397 (simplify_unary_operation_1): Likewise.
4398 (simplify_binary_operation_1): Likewise.
4399 (simplify_const_binary_operation): Likewise.
4400 (simplify_const_relational_operation): Likewise.
4401 (simplify_subreg): Likewise.
4402 * stor-layout.c (bitwise_mode_for_mode): Likewise.
4403 * var-tracking.c (adjust_mems): Likewise.
4404 (prepare_call_arguments): Likewise.
4406 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4407 Alan Hayward <alan.hayward@arm.com>
4408 David Sherwood <david.sherwood@arm.com>
4410 * machmode.h (is_int_mode): New fuction.
4411 * combine.c (find_split_point): Use it.
4412 (combine_simplify_rtx): Likewise.
4413 (simplify_if_then_else): Likewise.
4414 (simplify_set): Likewise.
4415 (simplify_shift_const_1): Likewise.
4416 (simplify_comparison): Likewise.
4417 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
4418 * cse.c (notreg_cost): Likewise.
4419 (cse_insn): Likewise.
4420 * cselib.c (cselib_lookup_1): Likewise.
4421 * dojump.c (do_jump_1): Likewise.
4422 (do_compare_rtx_and_jump): Likewise.
4423 * dse.c (get_call_args): Likewise.
4424 * dwarf2out.c (rtl_for_decl_init): Likewise.
4425 (native_encode_initializer): Likewise.
4426 * expmed.c (emit_store_flag_1): Likewise.
4427 (emit_store_flag): Likewise.
4428 * expr.c (convert_modes): Likewise.
4429 (store_field): Likewise.
4430 (expand_expr_real_1): Likewise.
4431 * fold-const.c (fold_read_from_constant_string): Likewise.
4432 * gimple-ssa-sprintf.c (get_format_string): Likewise.
4433 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
4434 * optabs.c (expand_binop): Likewise.
4435 (expand_unop): Likewise.
4436 (expand_abs_nojump): Likewise.
4437 (expand_one_cmpl_abs_nojump): Likewise.
4438 * simplify-rtx.c (mode_signbit_p): Likewise.
4439 (val_signbit_p): Likewise.
4440 (val_signbit_known_set_p): Likewise.
4441 (val_signbit_known_clear_p): Likewise.
4442 (simplify_relational_operation_1): Likewise.
4443 * tree.c (vector_type_mode): Likewise.
4445 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4446 Alan Hayward <alan.hayward@arm.com>
4447 David Sherwood <david.sherwood@arm.com>
4449 * machmode.h (smallest_mode_for_size): Fix formatting.
4450 (smallest_int_mode_for_size): New function.
4451 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
4452 instead of smallest_mode_for_size.
4453 * combine.c (make_extraction): Likewise.
4454 * config/arc/arc.c (arc_expand_movmem): Likewise.
4455 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
4456 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
4457 * config/s390/s390.c (s390_expand_insv): Likewise.
4458 * config/sparc/sparc.c (assign_int_registers): Likewise.
4459 * config/spu/spu.c (spu_function_value): Likewise.
4460 (spu_function_arg): Likewise.
4461 * coverage.c (get_gcov_type): Likewise.
4462 (get_gcov_unsigned_t): Likewise.
4463 * dse.c (find_shift_sequence): Likewise.
4464 * expmed.c (store_bit_field_1): Likewise.
4465 * expr.c (convert_move): Likewise.
4466 (store_field): Likewise.
4467 * internal-fn.c (expand_arith_overflow): Likewise.
4468 * optabs-query.c (get_best_extraction_insn): Likewise.
4469 * optabs.c (expand_twoval_binop_libfunc): Likewise.
4470 * stor-layout.c (layout_type): Likewise.
4471 (initialize_sizetypes): Likewise.
4472 * targhooks.c (default_get_mask_mode): Likewise.
4473 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
4475 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4476 Alan Hayward <alan.hayward@arm.com>
4477 David Sherwood <david.sherwood@arm.com>
4479 * machmode.h (opt_mode::else_blk): New function.
4480 (int_mode_for_mode): Declare.
4481 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
4482 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
4484 * cfgexpand.c (expand_debug_expr): Likewise.
4485 * combine.c (gen_lowpart_or_truncate): Likewise.
4486 (gen_lowpart_for_combine): Likewise.
4487 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
4488 * config/avr/avr.c (avr_to_int_mode): Likewise.
4489 (avr_out_plus_1): Likewise.
4490 (avr_out_plus): Likewise.
4491 (avr_out_round): Likewise.
4492 * config/i386/i386.c (ix86_split_to_parts): Likewise.
4493 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
4494 (s390_expand_vcond): Likewise.
4495 * config/spu/spu.c (spu_split_immediate): Likewise.
4496 (spu_expand_mov): Likewise.
4497 * dse.c (get_stored_val): Likewise.
4498 * expmed.c (store_bit_field_1): Likewise.
4499 (convert_extracted_bit_field): Use int_mode_for_mode instead of
4501 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
4502 (extract_low_bits): Likewise.
4503 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
4504 handling rather than repeating the check.
4505 (emit_group_store): Likewise.
4506 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
4507 * optabs.c (expand_absneg_bit): Likewise.
4508 (expand_copysign_absneg): Likewise.
4509 (expand_copysign_bit): Likewise.
4510 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
4511 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
4512 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
4513 * var-tracking.c (prepare_call_arguments): Likewise.
4514 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
4515 int_mode_for_mode instead of mode_for_size.
4516 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
4518 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4519 Alan Hayward <alan.hayward@arm.com>
4520 David Sherwood <david.sherwood@arm.com>
4522 * machmode.h (int_mode_for_size): New function.
4523 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
4524 instead of mode_for_size.
4525 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
4527 * combine.c (expand_field_assignment): Use int_mode_for_size
4528 instead of mode_for_size.
4529 (make_extraction): Likewise.
4530 (simplify_shift_const_1): Likewise.
4531 (simplify_comparison): Likewise.
4532 * dojump.c (do_jump): Likewise.
4533 * dwarf2out.c (mem_loc_descriptor): Likewise.
4534 * emit-rtl.c (init_derived_machine_modes): Likewise.
4535 * expmed.c (flip_storage_order): Likewise.
4536 (convert_extracted_bit_field): Likewise.
4537 * expr.c (copy_blkmode_from_reg): Likewise.
4538 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
4539 * internal-fn.c (expand_mul_overflow): Likewise.
4540 * lower-subreg.c (simple_move): Likewise.
4541 * optabs-libfuncs.c (init_optabs): Likewise.
4542 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4543 * tree.c (vector_type_mode): Likewise.
4544 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
4545 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
4546 * tree-vect-generic.c (expand_vector_parallel): Likewise.
4547 * tree-vect-stmts.c (vectorizable_load): Likewise.
4548 (vectorizable_store): Likewise.
4550 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4551 Alan Hayward <alan.hayward@arm.com>
4552 David Sherwood <david.sherwood@arm.com>
4554 * coretypes.h (pod_mode): New type.
4555 (scalar_int_mode_pod): New typedef.
4556 * machmode.h (pod_mode): New class.
4557 (int_n_data_t::m): Change type to scalar_int_mode_pod.
4558 * genmodes.c (emit_mode_int_n): Update accordingly.
4559 * lower-subreg.h (target_lower_subreg): Change type to
4560 scalar_int_mode_pod.
4561 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
4562 scalar_int_mode_pod.
4564 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4565 Alan Hayward <alan.hayward@arm.com>
4566 David Sherwood <david.sherwood@arm.com>
4568 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
4569 machine_mode to scalar_int_mode.
4570 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
4571 (rs6000_option_override_internal): Remove cast to int.
4572 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
4573 machine_mode to scalar_int_mode.
4574 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
4575 (rs6000_option_override_internal): Remove cast to int.
4576 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
4577 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
4579 * config/s390/s390.c (s390_expand_builtin): Likewise.
4580 * coretypes.h (scalar_int_mode): New type.
4581 (opt_scalar_int_mode): New typedef.
4582 * machmode.h (scalar_int_mode): New class.
4583 (scalar_int_mode::includes_p): New function.
4584 (byte_mode): Change type to scalar_int_mode.
4585 (word_mode): Likewise.
4586 (ptr_mode): Likewise.
4587 * emit-rtl.c (byte_mode): Likewise.
4588 (word_mode): Likewise.
4589 (ptr_mode): Likewise.
4590 (init_derived_machine_modes): Update accordingly.
4591 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
4592 and MODE_PARTIAL_INT.
4593 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
4594 opt_scalar_int_mode.
4596 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4597 Alan Hayward <alan.hayward@arm.com>
4598 David Sherwood <david.sherwood@arm.com>
4600 * target.def (libgcc_floating_mode_supported_p): Take a
4602 * doc/tm.texi: Regenerate.
4603 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
4605 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
4606 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
4609 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4610 Alan Hayward <alan.hayward@arm.com>
4611 David Sherwood <david.sherwood@arm.com>
4613 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
4614 * doc/tm.texi: Regenerate.
4615 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
4616 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
4617 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
4618 * targhooks.h (default_floatn_mode): Likewise.
4619 * targhooks.c (default_floatn_mode): Likewise.
4620 * tree.c (build_common_tree_nodes): Update accordingly.
4622 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4623 Alan Hayward <alan.hayward@arm.com>
4624 David Sherwood <david.sherwood@arm.com>
4626 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
4627 (mode_iterator::iterate_p): Likewise.
4628 (mode_iterator::get_wider): Likewise.
4629 * expr.c (init_expr_target): Use opt_scalar_float_mode.
4631 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4632 Alan Hayward <alan.hayward@arm.com>
4633 David Sherwood <david.sherwood@arm.com>
4635 * coretypes.h (opt_scalar_float_mode): New typedef.
4636 * machmode.h (float_mode_for_size): New function.
4637 * emit-rtl.c (double_mode): Delete.
4638 (init_emit_once): Use float_mode_for_size.
4639 * stor-layout.c (layout_type): Likewise.
4640 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
4642 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4643 Alan Hayward <alan.hayward@arm.com>
4644 David Sherwood <david.sherwood@arm.com>
4646 * output.h (assemble_real): Take a scalar_float_mode.
4647 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
4648 * config/arm/arm.md (consttable_4): Likewise.
4649 (consttable_8): Likewise.
4650 (consttable_16): Likewise.
4651 * config/mips/mips.md (consttable_float): Likewise.
4652 * config/s390/s390.c (s390_output_pool_entry): Likewise.
4653 * varasm.c (assemble_real): Take a scalar_float_mode.
4654 (output_constant_pool_2): Update accordingly.
4655 (output_constant): Likewise.
4657 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4658 Alan Hayward <alan.hayward@arm.com>
4659 David Sherwood <david.sherwood@arm.com>
4661 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
4662 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
4663 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
4664 (native_encode_real): Likewise.
4665 (native_interpret_real): Likewise.
4666 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
4667 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
4669 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4670 Alan Hayward <alan.hayward@arm.com>
4671 David Sherwood <david.sherwood@arm.com>
4673 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
4674 <scalar_float_mode>. Simplify.
4675 (gen_extend_conv_libfunc): Likewise.
4677 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4678 Alan Hayward <alan.hayward@arm.com>
4679 David Sherwood <david.sherwood@arm.com>
4681 * coretypes.h (scalar_float_mode): New type.
4682 * machmode.h (mode_traits::from_int): Use machine_mode if
4683 USE_ENUM_MODES is defined.
4684 (is_a): New function.
4686 (dyn_cast): Likewise.
4687 (scalar_float_mode): New class.
4688 (scalar_float_mode::includes_p): New function.
4689 (is_float_mode): Likewise.
4690 * gdbhooks.py (MachineModePrinter): New class.
4691 (build_pretty_printer): Use it for scalar_float_mode.
4692 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
4693 (format_helper::format_helper): Turn into a template.
4694 * genmodes.c (get_mode_class): New function.
4695 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
4696 or machine_mode if none.
4697 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
4698 as_a <scalar_float_mode>.
4699 * dwarf2out.c (mem_loc_descriptor): Likewise.
4700 (insert_float): Likewise.
4701 (add_const_value_attribute): Likewise.
4702 * simplify-rtx.c (simplify_immed_subreg): Likewise.
4703 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
4704 (expand_unop): Update accordingly.
4705 (expand_abs_nojump): Likewise.
4706 (expand_copysign_absneg): Take a scalar_float_mode.
4707 (expand_copysign_bit): Likewise.
4708 (expand_copysign): Update accordingly.
4710 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4711 Alan Hayward <alan.hayward@arm.com>
4712 David Sherwood <david.sherwood@arm.com>
4714 * coretypes.h (opt_mode): New class.
4715 * machmode.h (opt_mode): Likewise.
4716 (opt_mode::else_void): New function.
4717 (opt_mode::require): Likewise.
4718 (opt_mode::exists): Likewise.
4719 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
4720 (GET_MODE_2XWIDER_MODE): Likewise.
4721 (mode_iterator::get_wider): Update accordingly.
4722 (mode_iterator::get_2xwider): Likewise.
4723 (mode_iterator::get_known_wider): Likewise, turning into a template.
4724 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
4725 forcing a wider mode to exist.
4726 * config/cr16/cr16.h (LONG_REG_P): Likewise.
4727 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
4728 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
4729 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
4730 * lower-subreg.c (init_lower_subreg): Likewise.
4731 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
4732 on the final iteration.
4733 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
4734 a wider mode exists before asking for a move pattern.
4735 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
4736 forcing a wider mode to exist.
4737 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
4738 returning false if no such mode exists.
4739 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
4740 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
4741 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
4742 Avoid checking for a MODE_INT if we already know the mode is not a
4744 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
4745 forcing a wider mode to exist.
4746 (expmed_mult_highpart_optab): Likewise.
4747 (expmed_mult_highpart): Likewise.
4748 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
4750 * lto-streamer-in.c (lto_input_mode_table): Likewise.
4751 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
4752 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
4753 * internal-fn.c (expand_mul_overflow): Update use of
4754 GET_MODE_2XWIDER_MODE.
4755 * omp-low.c (omp_clause_aligned_alignment): Likewise.
4756 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
4757 GET_MODE_WIDER_MODE.
4758 (convert_plusminus_to_widen): Likewise.
4759 * tree-switch-conversion.c (array_value_type): Likewise.
4760 * var-tracking.c (emit_note_insn_var_location): Likewise.
4761 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
4762 Return false inside rather than outside the loop if no wider mode
4764 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
4765 and GET_MODE_2XWIDER_MODE
4766 (can_compare_p): Use else_void.
4767 * gdbhooks.py (OptMachineModePrinter): New class.
4768 (build_pretty_printer): Use it for opt_mode.
4770 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4771 Alan Hayward <alan.hayward@arm.com>
4772 David Sherwood <david.sherwood@arm.com>
4774 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
4775 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
4777 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4778 Alan Hayward <alan.hayward@arm.com>
4779 David Sherwood <david.sherwood@arm.com>
4781 * machmode.h (mode_traits): New structure.
4782 (get_narrowest_mode): New function.
4783 (mode_iterator::start): Likewise.
4784 (mode_iterator::iterate_p): Likewise.
4785 (mode_iterator::get_wider): Likewise.
4786 (mode_iterator::get_known_wider): Likewise.
4787 (mode_iterator::get_2xwider): Likewise.
4788 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
4789 (FOR_EACH_MODE): Likewise.
4790 (FOR_EACH_MODE_FROM): Likewise.
4791 (FOR_EACH_MODE_UNTIL): Likewise.
4792 (FOR_EACH_WIDER_MODE): Likewise.
4793 (FOR_EACH_2XWIDER_MODE): Likewise.
4794 * builtins.c (expand_builtin_strlen): Use new mode iterators.
4795 * combine.c (simplify_comparison): Likewise
4796 * config/i386/i386.c (type_natural_mode): Likewise.
4797 * cse.c (cse_insn): Likewise.
4798 * dse.c (find_shift_sequence): Likewise.
4799 * emit-rtl.c (init_derived_machine_modes): Likewise.
4800 (init_emit_once): Likewise.
4801 * explow.c (hard_function_value): Likewise.
4802 * expmed.c (extract_fixed_bit_field_1): Likewise.
4803 (extract_bit_field_1): Likewise.
4804 (expand_divmod): Likewise.
4805 (emit_store_flag_1): Likewise.
4806 * expr.c (init_expr_target): Likewise.
4807 (convert_move): Likewise.
4808 (alignment_for_piecewise_move): Likewise.
4809 (widest_int_mode_for_size): Likewise.
4810 (emit_block_move_via_movmem): Likewise.
4811 (copy_blkmode_to_reg): Likewise.
4812 (set_storage_via_setmem): Likewise.
4813 (compress_float_constant): Likewise.
4814 * omp-low.c (omp_clause_aligned_alignment): Likewise.
4815 * optabs-query.c (get_best_extraction_insn): Likewise.
4816 * optabs.c (expand_binop): Likewise.
4817 (expand_twoval_unop): Likewise.
4818 (expand_twoval_binop): Likewise.
4819 (widen_leading): Likewise.
4820 (widen_bswap): Likewise.
4821 (expand_parity): Likewise.
4822 (expand_unop): Likewise.
4823 (prepare_cmp_insn): Likewise.
4824 (prepare_float_lib_cmp): Likewise.
4825 (expand_float): Likewise.
4826 (expand_fix): Likewise.
4827 (expand_sfix_optab): Likewise.
4828 * postreload.c (move2add_use_add2_insn): Likewise.
4829 * reg-stack.c (reg_to_stack): Likewise.
4830 * reginfo.c (choose_hard_reg_mode): Likewise.
4831 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
4832 * stor-layout.c (mode_for_size): Likewise.
4833 (smallest_mode_for_size): Likewise.
4834 (mode_for_vector): Likewise.
4835 (finish_bitfield_representative): Likewise.
4836 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
4837 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
4838 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
4839 * var-tracking.c (prepare_call_arguments): Likewise.
4841 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4842 Alan Hayward <alan.hayward@arm.com>
4843 David Sherwood <david.sherwood@arm.com>
4845 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
4846 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
4847 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
4848 * machmode.h (mode_size): Move earlier in file.
4849 (mode_precision): Likewise.
4850 (mode_inner): Likewise.
4851 (mode_nunits): Likewise.
4852 (mode_unit_size): Likewise.
4853 (unit_unit_precision): Likewise.
4854 (mode_wider): Likewise.
4855 (mode_2xwider): Likewise.
4856 (machine_mode): New class.
4857 (mode_to_bytes): New function.
4858 (mode_to_bits): Likewise.
4859 (mode_to_precision): Likewise.
4860 (mode_to_inner): Likewise.
4861 (mode_to_unit_size): Likewise.
4862 (mode_to_unit_precision): Likewise.
4863 (mode_to_nunits): Likewise.
4864 (GET_MODE_SIZE): Use mode_to_bytes.
4865 (GET_MODE_BITSIZE): Use mode_to_bits.
4866 (GET_MODE_PRECISION): Use mode_to_precision.
4867 (GET_MODE_INNER): Use mode_to_inner.
4868 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
4869 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
4870 (GET_MODE_NUNITS): Use mode_to_nunits.
4871 * system.h (ALWAYS_INLINE): New macro.
4872 * config/powerpcspe/powerpcspe-c.c
4873 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
4874 int for arg1_mode and arg2_mode.
4876 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4877 Alan Hayward <alan.hayward@arm.com>
4878 David Sherwood <david.sherwood@arm.com>
4880 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
4881 Prefix mode names with E_ in case statements.
4882 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4883 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
4884 (aarch64_split_simd_move): Likewise.
4885 (aarch64_gen_storewb_pair): Likewise.
4886 (aarch64_gen_loadwb_pair): Likewise.
4887 (aarch64_gen_store_pair): Likewise.
4888 (aarch64_gen_load_pair): Likewise.
4889 (aarch64_get_condition_code_1): Likewise.
4890 (aarch64_constant_pool_reload_icode): Likewise.
4891 (get_rsqrte_type): Likewise.
4892 (get_rsqrts_type): Likewise.
4893 (get_recpe_type): Likewise.
4894 (get_recps_type): Likewise.
4895 (aarch64_gimplify_va_arg_expr): Likewise.
4896 (aarch64_simd_container_mode): Likewise.
4897 (aarch64_emit_load_exclusive): Likewise.
4898 (aarch64_emit_store_exclusive): Likewise.
4899 (aarch64_expand_compare_and_swap): Likewise.
4900 (aarch64_gen_atomic_cas): Likewise.
4901 (aarch64_emit_bic): Likewise.
4902 (aarch64_emit_atomic_swap): Likewise.
4903 (aarch64_emit_atomic_load_op): Likewise.
4904 (aarch64_evpc_trn): Likewise.
4905 (aarch64_evpc_uzp): Likewise.
4906 (aarch64_evpc_zip): Likewise.
4907 (aarch64_evpc_ext): Likewise.
4908 (aarch64_evpc_rev): Likewise.
4909 (aarch64_evpc_dup): Likewise.
4910 (aarch64_gen_ccmp_first): Likewise.
4911 (aarch64_gen_ccmp_next): Likewise.
4912 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
4913 (alpha_emit_xfloating_libcall): Likewise.
4914 (emit_insxl): Likewise.
4915 (alpha_arg_type): Likewise.
4916 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
4917 (arc_preferred_simd_mode): Likewise.
4918 (arc_secondary_reload): Likewise.
4919 (get_arc_condition_code): Likewise.
4920 (arc_print_operand): Likewise.
4921 (arc_legitimate_constant_p): Likewise.
4922 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4923 * config/arc/arc.md (casesi_load): Likewise.
4924 (casesi_compact_jump): Likewise.
4925 * config/arc/predicates.md (proper_comparison_operator): Likewise.
4926 (cc_use_register): Likewise.
4927 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4928 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
4929 (arm_init_iwmmxt_builtins): Likewise.
4930 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
4931 (neon_expand_vector_init): Likewise.
4932 (arm_attr_length_move_neon): Likewise.
4933 (maybe_get_arm_condition_code): Likewise.
4934 (arm_emit_vector_const): Likewise.
4935 (arm_preferred_simd_mode): Likewise.
4936 (arm_output_iwmmxt_tinsr): Likewise.
4937 (thumb1_output_casesi): Likewise.
4938 (thumb2_output_casesi): Likewise.
4939 (arm_emit_load_exclusive): Likewise.
4940 (arm_emit_store_exclusive): Likewise.
4941 (arm_expand_compare_and_swap): Likewise.
4942 (arm_evpc_neon_vuzp): Likewise.
4943 (arm_evpc_neon_vzip): Likewise.
4944 (arm_evpc_neon_vrev): Likewise.
4945 (arm_evpc_neon_vtrn): Likewise.
4946 (arm_evpc_neon_vext): Likewise.
4947 (arm_validize_comparison): Likewise.
4948 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
4949 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
4950 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
4951 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
4952 (c6x_preferred_simd_mode): Likewise.
4953 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
4954 (epiphany_rtx_costs): Likewise.
4955 * config/epiphany/predicates.md (proper_comparison_operator):
4957 * config/frv/frv.c (condexec_memory_operand): Likewise.
4958 (frv_emit_move): Likewise.
4959 (output_move_single): Likewise.
4960 (output_condmove_single): Likewise.
4961 (frv_hard_regno_mode_ok): Likewise.
4962 (frv_matching_accg_mode): Likewise.
4963 * config/h8300/h8300.c (split_adds_subs): Likewise.
4964 (h8300_rtx_costs): Likewise.
4965 (h8300_print_operand): Likewise.
4966 (compute_mov_length): Likewise.
4967 (output_logical_op): Likewise.
4968 (compute_logical_op_length): Likewise.
4969 (compute_logical_op_cc): Likewise.
4970 (h8300_shift_needs_scratch_p): Likewise.
4971 (output_a_shift): Likewise.
4972 (compute_a_shift_length): Likewise.
4973 (compute_a_shift_cc): Likewise.
4974 (expand_a_rotate): Likewise.
4975 (output_a_rotate): Likewise.
4976 * config/i386/i386.c (classify_argument): Likewise.
4977 (function_arg_advance_32): Likewise.
4978 (function_arg_32): Likewise.
4979 (function_arg_64): Likewise.
4980 (function_value_64): Likewise.
4981 (ix86_gimplify_va_arg): Likewise.
4982 (ix86_legitimate_constant_p): Likewise.
4983 (put_condition_code): Likewise.
4984 (split_double_mode): Likewise.
4985 (ix86_avx256_split_vector_move_misalign): Likewise.
4986 (ix86_expand_vector_logical_operator): Likewise.
4987 (ix86_split_idivmod): Likewise.
4988 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
4989 (ix86_build_const_vector): Likewise.
4990 (ix86_build_signbit_mask): Likewise.
4991 (ix86_match_ccmode): Likewise.
4992 (ix86_cc_modes_compatible): Likewise.
4993 (ix86_expand_branch): Likewise.
4994 (ix86_expand_sse_cmp): Likewise.
4995 (ix86_expand_sse_movcc): Likewise.
4996 (ix86_expand_int_sse_cmp): Likewise.
4997 (ix86_expand_vec_perm_vpermi2): Likewise.
4998 (ix86_expand_vec_perm): Likewise.
4999 (ix86_expand_sse_unpack): Likewise.
5000 (ix86_expand_int_addcc): Likewise.
5001 (ix86_split_to_parts): Likewise.
5002 (ix86_vectorize_builtin_gather): Likewise.
5003 (ix86_vectorize_builtin_scatter): Likewise.
5004 (avx_vpermilp_parallel): Likewise.
5005 (inline_memory_move_cost): Likewise.
5006 (ix86_tieable_integer_mode_p): Likewise.
5007 (x86_maybe_negate_const_int): Likewise.
5008 (ix86_expand_vector_init_duplicate): Likewise.
5009 (ix86_expand_vector_init_one_nonzero): Likewise.
5010 (ix86_expand_vector_init_one_var): Likewise.
5011 (ix86_expand_vector_init_concat): Likewise.
5012 (ix86_expand_vector_init_interleave): Likewise.
5013 (ix86_expand_vector_init_general): Likewise.
5014 (ix86_expand_vector_set): Likewise.
5015 (ix86_expand_vector_extract): Likewise.
5016 (emit_reduc_half): Likewise.
5017 (ix86_emit_i387_round): Likewise.
5018 (ix86_mangle_type): Likewise.
5019 (ix86_expand_round_sse4): Likewise.
5020 (expand_vec_perm_blend): Likewise.
5021 (canonicalize_vector_int_perm): Likewise.
5022 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
5023 (expand_vec_perm_1): Likewise.
5024 (expand_vec_perm_interleave3): Likewise.
5025 (expand_vec_perm_even_odd_pack): Likewise.
5026 (expand_vec_perm_even_odd_1): Likewise.
5027 (expand_vec_perm_broadcast_1): Likewise.
5028 (ix86_vectorize_vec_perm_const_ok): Likewise.
5029 (ix86_expand_vecop_qihi): Likewise.
5030 (ix86_expand_mul_widen_hilo): Likewise.
5031 (ix86_expand_sse2_abs): Likewise.
5032 (ix86_expand_pextr): Likewise.
5033 (ix86_expand_pinsr): Likewise.
5034 (ix86_preferred_simd_mode): Likewise.
5035 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
5036 * config/i386/sse.md (*andnot<mode>3): Likewise.
5037 (<mask_codefor><code><mode>3<mask_name>): Likewise.
5038 (*<code><mode>3): Likewise.
5039 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
5040 (ia64_expand_atomic_op): Likewise.
5041 (ia64_arg_type): Likewise.
5042 (ia64_mode_to_int): Likewise.
5043 (ia64_scalar_mode_supported_p): Likewise.
5044 (ia64_vector_mode_supported_p): Likewise.
5045 (expand_vec_perm_broadcast): Likewise.
5046 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
5047 (iq2000_function_arg_advance): Likewise.
5048 (iq2000_function_arg): Likewise.
5049 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
5050 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
5051 (m68k_libcall_value): Likewise.
5052 (m68k_function_value): Likewise.
5053 (sched_attr_op_type): Likewise.
5054 * config/mcore/mcore.c (mcore_output_move): Likewise.
5055 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
5057 (microblaze_function_arg): Likewise.
5058 * config/mips/mips.c (mips16_build_call_stub): Likewise.
5059 (mips_print_operand): Likewise.
5060 (mips_mode_ok_for_mov_fmt_p): Likewise.
5061 (mips_vector_mode_supported_p): Likewise.
5062 (mips_preferred_simd_mode): Likewise.
5063 (mips_expand_vpc_loongson_even_odd): Likewise.
5064 (mips_expand_vec_unpack): Likewise.
5065 (mips_expand_vi_broadcast): Likewise.
5066 (mips_expand_vector_init): Likewise.
5067 (mips_expand_vec_reduc): Likewise.
5068 (mips_expand_msa_cmp): Likewise.
5069 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
5070 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
5071 (cc_flags_for_mode): Likewise.
5072 * config/msp430/msp430.c (msp430_print_operand): Likewise.
5073 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
5074 (nds32_output_casesi_pc_relative): Likewise.
5075 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
5076 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
5077 (nvptx_gen_unpack): Likewise.
5078 (nvptx_gen_pack): Likewise.
5079 (nvptx_gen_shuffle): Likewise.
5080 (nvptx_gen_wcast): Likewise.
5081 (nvptx_preferred_simd_mode): Likewise.
5082 * config/pa/pa.c (pa_secondary_reload): Likewise.
5083 * config/pa/predicates.md (base14_operand): Likewise.
5084 * config/powerpcspe/powerpcspe-c.c
5085 (altivec_resolve_overloaded_builtin): Likewise.
5086 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
5088 (rs6000_preferred_simd_mode): Likewise.
5089 (output_vec_const_move): Likewise.
5090 (rs6000_expand_vector_extract): Likewise.
5091 (rs6000_split_vec_extract_var): Likewise.
5092 (reg_offset_addressing_ok_p): Likewise.
5093 (rs6000_legitimate_offset_address_p): Likewise.
5094 (rs6000_legitimize_address): Likewise.
5095 (rs6000_emit_set_const): Likewise.
5096 (rs6000_const_vec): Likewise.
5097 (rs6000_emit_move): Likewise.
5098 (spe_build_register_parallel): Likewise.
5099 (rs6000_darwin64_record_arg_recurse): Likewise.
5100 (swap_selector_for_mode): Likewise.
5101 (spe_init_builtins): Likewise.
5102 (paired_init_builtins): Likewise.
5103 (altivec_init_builtins): Likewise.
5104 (do_load_for_compare): Likewise.
5105 (rs6000_generate_compare): Likewise.
5106 (rs6000_expand_float128_convert): Likewise.
5107 (emit_load_locked): Likewise.
5108 (emit_store_conditional): Likewise.
5109 (rs6000_output_function_epilogue): Likewise.
5110 (rs6000_handle_altivec_attribute): Likewise.
5111 (rs6000_function_value): Likewise.
5112 (emit_fusion_gpr_load): Likewise.
5113 (emit_fusion_p9_load): Likewise.
5114 (emit_fusion_p9_store): Likewise.
5115 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
5116 (fusion_gpr_mem_load): Likewise.
5117 (fusion_addis_mem_combo_load): Likewise.
5118 (fusion_addis_mem_combo_store): Likewise.
5119 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
5120 (fusion_gpr_mem_load): Likewise.
5121 (fusion_addis_mem_combo_load): Likewise.
5122 (fusion_addis_mem_combo_store): Likewise.
5123 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5125 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
5126 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
5127 (rs6000_preferred_simd_mode): Likewise.
5128 (output_vec_const_move): Likewise.
5129 (rs6000_expand_vector_extract): Likewise.
5130 (rs6000_split_vec_extract_var): Likewise.
5131 (reg_offset_addressing_ok_p): Likewise.
5132 (rs6000_legitimate_offset_address_p): Likewise.
5133 (rs6000_legitimize_address): Likewise.
5134 (rs6000_emit_set_const): Likewise.
5135 (rs6000_const_vec): Likewise.
5136 (rs6000_emit_move): Likewise.
5137 (rs6000_darwin64_record_arg_recurse): Likewise.
5138 (swap_selector_for_mode): Likewise.
5139 (paired_init_builtins): Likewise.
5140 (altivec_init_builtins): Likewise.
5141 (rs6000_expand_float128_convert): Likewise.
5142 (emit_load_locked): Likewise.
5143 (emit_store_conditional): Likewise.
5144 (rs6000_output_function_epilogue): Likewise.
5145 (rs6000_handle_altivec_attribute): Likewise.
5146 (rs6000_function_value): Likewise.
5147 (emit_fusion_gpr_load): Likewise.
5148 (emit_fusion_p9_load): Likewise.
5149 (emit_fusion_p9_store): Likewise.
5150 * config/rx/rx.c (rx_gen_move_template): Likewise.
5151 (flags_from_mode): Likewise.
5152 * config/s390/predicates.md (s390_alc_comparison): Likewise.
5153 (s390_slb_comparison): Likewise.
5154 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
5155 (s390_vector_mode_supported_p): Likewise.
5156 (s390_cc_modes_compatible): Likewise.
5157 (s390_match_ccmode_set): Likewise.
5158 (s390_canonicalize_comparison): Likewise.
5159 (s390_emit_compare_and_swap): Likewise.
5160 (s390_branch_condition_mask): Likewise.
5161 (s390_rtx_costs): Likewise.
5162 (s390_secondary_reload): Likewise.
5163 (__SECONDARY_RELOAD_CASE): Likewise.
5164 (s390_expand_cs): Likewise.
5165 (s390_preferred_simd_mode): Likewise.
5166 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
5167 * config/sh/sh.c (sh_print_operand): Likewise.
5168 (dump_table): Likewise.
5169 (sh_secondary_reload): Likewise.
5170 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
5171 * config/sh/sh.md (casesi_worker_1): Likewise.
5172 (casesi_worker_2): Likewise.
5173 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
5174 (fcc_comparison_operator): Likewise.
5175 * config/sparc/sparc.c (sparc_expand_move): Likewise.
5176 (emit_soft_tfmode_cvt): Likewise.
5177 (sparc_preferred_simd_mode): Likewise.
5178 (output_cbranch): Likewise.
5179 (sparc_print_operand): Likewise.
5180 (sparc_expand_vec_perm_bmask): Likewise.
5181 (vector_init_bshuffle): Likewise.
5182 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
5183 (spu_vector_mode_supported_p): Likewise.
5184 (spu_expand_insv): Likewise.
5185 (spu_emit_branch_or_set): Likewise.
5186 (spu_handle_vector_attribute): Likewise.
5187 (spu_builtin_splats): Likewise.
5188 (spu_builtin_extract): Likewise.
5189 (spu_builtin_promote): Likewise.
5190 (spu_expand_sign_extend): Likewise.
5191 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
5192 (tilegx_simd_int): Likewise.
5193 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
5194 (tilepro_simd_int): Likewise.
5195 * config/v850/v850.c (const_double_split): Likewise.
5196 (v850_print_operand): Likewise.
5197 (ep_memory_offset): Likewise.
5198 * config/vax/vax.c (vax_rtx_costs): Likewise.
5199 (vax_output_int_move): Likewise.
5200 (vax_output_int_add): Likewise.
5201 (vax_output_int_subtract): Likewise.
5202 * config/visium/predicates.md (visium_branch_operator): Likewise.
5203 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
5204 (visium_print_operand_address): Likewise.
5205 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
5206 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
5207 (xtensa_expand_conditional_branch): Likewise.
5208 (xtensa_copy_incoming_a7): Likewise.
5209 (xtensa_output_literal): Likewise.
5210 * dfp.c (decimal_real_maxval): Likewise.
5211 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
5213 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5214 Alan Hayward <alan.hayward@arm.com>
5215 David Sherwood <david.sherwood@arm.com>
5217 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
5218 (mode_nunits_inline): Likewise.
5219 (mode_inner_inline): Likewise.
5220 (mode_unit_size_inline): Likewise.
5221 (mode_unit_precision_inline): Likewise.
5222 (emit_insn_modes_h): Likewise. Also emit a #define of the
5224 (emit_mode_wider): Add an E_ prefix to mode names.
5225 (emit_mode_complex): Likewise.
5226 (emit_mode_inner): Likewise.
5227 (emit_mode_adjustments): Likewise.
5228 (emit_mode_int_n): Likewise.
5229 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
5230 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
5231 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
5232 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
5233 (CRC32_BUILTIN, ENTRY): Likewise.
5234 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
5235 (aarch64_pop_regs): Likewise.
5236 (aarch64_process_components): Likewise.
5237 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
5238 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
5239 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
5240 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
5241 * config/arm/arm.c (arm_init_libfuncs): Likewise.
5242 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
5244 * config/i386/i386-builtin.def (pcmpestr): Likewise.
5245 (pcmpistr): Likewise.
5246 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
5247 * config/mmix/mmix.c (mmix_output_condition): Likewise.
5248 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
5250 * config/rl78/rl78.c (mduc_regs): Likewise.
5251 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
5252 (htm_expand_builtin): Likewise.
5253 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
5254 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
5255 * config/xtensa/xtensa.c (print_operand): Likewise.
5256 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
5257 (NUM_MODE_VECTOR_INT): Likewise.
5258 * genoutput.c (null_operand): Likewise.
5259 (output_operand_data): Likewise.
5260 * genrecog.c (print_parameter_value): Likewise.
5261 * lra.c (debug_operand_data): Likewise.
5263 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5265 * dwarf2out.c (scompare_loc_descriptor_wide)
5266 (scompare_loc_descriptor_narrow): New functions, split out from...
5267 (scompare_loc_descriptor): ...here.
5268 * expmed.c (emit_store_flag_int): New function, split out from...
5269 (emit_store_flag): ...here.
5271 2017-08-30 Richard Biener <rguenther@suse.de>
5273 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
5274 (dwarf2out_early_finish): Move setting of AT_pubnames from
5275 early debug output to early finish.
5277 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
5279 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
5280 and -mdata-region to the assembler.
5282 2017-08-30 Richard Biener <rguenther@suse.de>
5284 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
5286 (gen_subprogram_die): Add DW_AT_object_pointer only early.
5287 (dwarf2out_early_global_decl): Only generate a DIE for the
5288 abstract origin if it doesn't already exist or is a declaration DIE.
5289 (resolve_addr): Do not add the linkage name twice when
5290 generating a stub DIE for the DW_TAG_GNU_call_site target.
5292 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5294 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5295 Use machine_mode rather than int for arg1_mode.
5297 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
5300 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
5301 that the second argument of the built-in functions to unpack
5302 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
5303 switch statement instead a lot of if statements.
5304 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
5305 Allow 64-bit values to be in Altivec registers as well as
5306 traditional floating point registers.
5307 (pack<mode>, FMOVE128_VSX iterator): Likewise.
5309 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
5311 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
5312 MAX_REGS_PER_ADDRESS == 1.
5314 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
5316 * config/i386/i386.opt (flag_fentry): Do not init to -1.
5317 * config/i386/i386.c (ix86_option_override_internal): Simplify
5318 setting of opts->x_flag_entry.
5320 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5321 Jakub Jelinek <jakub@redhat.com>
5322 Richard Biener <rguenther@suse.de>
5324 PR tree-optimization/81503
5325 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
5326 folded constant fits in the target type; reorder tests for clarity.
5328 2017-08-29 Martin Liska <mliska@suse.cz>
5330 * passes.def: Include pass_lower_switch.
5331 * stmt.c (dump_case_nodes): Remove and move to
5332 tree-switch-conversion.
5333 (case_values_threshold): Likewise.
5334 (expand_switch_as_decision_tree_p): Likewise.
5335 (emit_case_decision_tree): Likewise.
5336 (expand_case): Likewise.
5337 (balance_case_nodes): Likewise.
5338 (node_has_low_bound): Likewise.
5339 (node_has_high_bound): Likewise.
5340 (node_is_bounded): Likewise.
5341 (emit_case_nodes): Likewise.
5342 (struct simple_case_node): New struct.
5343 (add_case_node): Remove.
5344 (emit_case_dispatch_table): Use vector instead of case_list.
5345 (reset_out_edges_aux): Remove.
5346 (compute_cases_per_edge): Likewise.
5347 (expand_case): Build list of simple_case_node.
5348 (expand_sjlj_dispatch_table): Use it.
5349 * tree-switch-conversion.c (struct case_node): Moved from
5350 stmt.c and adjusted.
5351 (emit_case_nodes): Likewise.
5352 (node_has_low_bound): Likewise.
5353 (node_has_high_bound): Likewise.
5354 (node_is_bounded): Likewise.
5355 (case_values_threshold): Likewise.
5356 (reset_out_edges_aux): Likewise.
5357 (compute_cases_per_edge): Likewise.
5358 (add_case_node): Likewise.
5359 (dump_case_nodes): Likewise.
5360 (balance_case_nodes): Likewise.
5361 (expand_switch_as_decision_tree_p): Likewise.
5362 (emit_jump): Likewise.
5363 (emit_case_decision_tree): Likewise.
5364 (try_switch_expansion): Likewise.
5365 (do_jump_if_equal): Likewise.
5366 (emit_cmp_and_jump_insns): Likewise.
5367 (fix_phi_operands_for_edge): New function.
5368 (record_phi_operand_mapping): Likewise.
5369 (class pass_lower_switch): New pass.
5370 (pass_lower_switch::execute): New function.
5371 (make_pass_lower_switch): Likewise.
5372 (conditional_probability):
5373 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
5374 * tree-pass.h: Add make_pass_lower_switch.
5376 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
5379 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
5382 2017-08-29 Richard Biener <rguenther@suse.de>
5384 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
5385 we do not add a DW_AT_inline attribute twice.
5386 (gen_subprogram_die): Remove code setting DW_AT_inline on
5387 DECL_ABSTRACT_P nodes.
5389 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
5391 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
5392 calls to internal functions.
5393 (gimplify_modify_expr): Likewise.
5394 * tree-call-cdce.c (use_internal_fn): Likewise.
5395 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
5396 (convert_to_divmod): Set the nothrow flag.
5397 * tree-if-conv.c (predicate_mem_writes): Likewise.
5398 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
5399 (vectorizable_call): Likewise.
5400 (vectorizable_store): Likewise.
5401 (vectorizable_load): Likewise.
5402 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
5403 (vect_recog_mask_conversion_pattern): Likewise.
5405 2017-08-29 Martin Liska <mliska@suse.cz>
5408 * gcc.c (driver_handle_option): Add new argument.
5409 * opts-common.c (handle_option): Pass
5410 target_option_override_hook.
5411 * opts-global.c (lang_handle_option): Add new option.
5412 (set_default_handlers): Add new argument.
5413 (decode_options): Likewise.
5414 * opts.c (target_handle_option): Likewise.
5415 (common_handle_option): Call target_option_override_hook.
5416 * opts.h (struct cl_option_handler_func): Add hook for
5417 target option override.
5418 (struct cl_option_handlers): Likewise.
5419 (set_default_handlers): Add new argument.
5420 (decode_options): Likewise.
5421 (common_handle_option): Likewise.
5422 (target_handle_option): Likewise.
5423 * toplev.c (toplev::main): Pass targetm.target_option.override
5426 2017-08-29 Richard Biener <rguenther@suse.de>
5427 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
5429 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
5430 life to the active subtree.
5432 2017-08-28 Jeff Law <law@redhat.com>
5434 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
5435 derive_equivalences.
5436 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
5438 (edge_info::derive_equivalences): New private member function
5440 * tree-ssa-dom.c (class edge_info): Changed from a struct
5441 to a class. Add ctor/dtor, methods and data members.
5442 (edge_info::edge_info): Renamed from allocate_edge_info.
5443 Initialize additional members.
5444 (edge_info::~edge_info): New.
5445 (free_dom_edge_info): Delete the edge info.
5446 (record_edge_info): Use new class & associated member functions.
5447 Tighten forms for testing for edge equivalences.
5448 (record_temporary_equivalences): Iterate over the simple
5449 equivalences rather than assuming there's only one per edge.
5450 (cprop_into_successor_phis): Iterate over the simple
5451 equivalences rather than assuming there's only one per edge.
5452 (optimize_stmt): Use operand_equal_p rather than pointer
5453 equality for mini-DSE code.
5455 2017-08-28 Nathan Sidwell <nathan@acm.org>
5457 * gcc.c (execute): Fold SIGPIPE handling into switch
5458 statement. Adjust internal error message.
5460 2017-08-28 Richard Biener <rguenther@suse.de>
5463 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
5464 Do nothing for removed DIEs.
5466 2017-08-28 Richard Biener <rguenther@suse.de>
5468 PR tree-optimization/81977
5469 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
5472 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
5475 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
5476 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
5477 using targetm.gen_mem_thread_fence.
5479 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
5482 * config/i386/i386.md (*<btsc><mode>): Change operand 2
5483 predicate to register_operand. Reorder operands.
5484 (*btr<mode>): Ditto.
5485 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
5486 (*btr<mode>_mask): Ditto.
5488 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
5490 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
5491 * config/rs6000/xmmintrin.h: New file.
5492 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
5494 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5497 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
5498 parameter and_insn and return it.
5499 (recombine_lvx_pattern): Insert a copy to ensure availability of
5500 the base register of the copied masking operation at the point of
5501 the instruction replacement.
5502 (recombine_stvx_pattern): Likewise.
5504 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
5506 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
5507 undocumented switches.
5508 (-mpower9-dform-vector): Likewise.
5509 (-mpower9-dform): Likewise.
5510 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
5511 comments to delete references to -mpower9-dform* switches.
5512 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
5513 Delete reference to -mpower9-dform* switches, test for
5514 -mpower9-vector instead.
5515 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
5516 (OTHER_P9_VECTOR_MASKS): Likewise.
5517 (POWERPC_MASKS): Likewise.
5518 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
5519 tests against -mpower9-dform* to -mpower9-vector. Delete code
5520 that checked for -mpower9-dform* consistancy with other options.
5521 Add test for -mpower9-misc to enable other power9 switches.
5522 (rs6000_init_hard_regno_mode_ok): Likewise.
5523 (rs6000_option_override_internal): Likewise.
5524 (rs6000_emit_prologue): Likewise.
5525 (rs6000_emit_epilogue): Likewise.
5526 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
5527 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
5528 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
5530 (emit_fusion_p9_store): Likewise.
5531 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
5532 resetting these macros if the assembler does not support ISA 3.0
5534 (TARGET_P9_DFORM_VECTOR): Likewise.
5535 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
5536 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
5538 2017-08-25 Alan Modra <amodra@gmail.com>
5540 PR rtl-optimization/81747
5541 * cse.c (cse_extended_basic_block): Don't attempt to record
5542 equivalences for degenerate conditional branches that branch
5543 to their fall-through.
5545 2017-08-24 Martin Sebor <msebor@redhat.com>
5548 * gimple-fold.c (size_must_be_zero_p): New function.
5549 (gimple_fold_builtin_memory_op): Call it.
5551 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
5553 * config/rs6000/mm_malloc.h: New file.
5555 2017-08-24 Bin Cheng <bin.cheng@arm.com>
5557 PR tree-optimization/81913
5558 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
5559 analysis when either IVs in condition can wrap.
5561 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
5563 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
5564 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
5566 2017-08-24 Richard Biener <rguenther@suse.de>
5569 * targhooks.c (default_target_can_inline_p): Properly
5570 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
5571 is present and always compare.
5572 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
5573 infer -mfpmath=sse from TARGET_SSE_P.
5574 (ix86_can_inline_p): Properly use target_option_default_node when
5575 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
5577 2017-08-24 Richard Biener <rguenther@suse.de>
5580 * dwarf2out.c (output_die): Handle flag_generate_offload like
5582 (output_comp_unit): Likewise.
5583 (gen_array_type_die): Likewise.
5584 (dwarf2out_early_finish): Likewise.
5585 (note_variable_value_in_expr): Likewise.
5586 (dwarf2out_finish): Likewise. Adjust assert.
5587 * cgraphunit.c (symbol_table::compile): Move setting of
5588 flag_generate_offload earlier ...
5589 (symbol_table::finalize_compilation_unit): ... here, before
5590 early debug finalization.
5592 2017-08-24 Richard Biener <rguenther@suse.de>
5594 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
5595 and ipa-fnsummary.h.
5596 (ix86_can_inline_p): When ix86_fpmath flags do not match
5597 check whether the callee uses FP math at all.
5599 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
5602 * tree-ssanames.c (get_nonzero_bits): Use element_precision
5603 instead of TYPE_PRECISION.
5605 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
5606 Alan Hayward <alan.hayward@arm.com>
5607 David Sherwood <david.sherwood@arm.com>
5609 * combine.c (make_extraction): Use subreg_offset_from_lsb.
5611 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
5613 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
5615 (ix86_frame::stack_realign_allocate): New field.
5616 (struct machine_frame_state): Modify comments.
5617 (machine_frame_state::sp_realigned_fp_end): New field.
5618 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
5620 (sp_valid_at): Add assertion to assure no attempt to access invalid
5621 offset of a realigned stack.
5622 (fp_valid_at): Likewise.
5623 (choose_baseaddr): Modify comments.
5624 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
5625 ix86_expand_prologue.
5626 (ix86_expand_prologue): Modify stack realignment and allocation.
5627 (ix86_expand_epilogue): Modify comments.
5628 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
5629 avx2_runtime, avx512f, and avx512f_runtime.
5631 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
5633 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
5634 (mstackrealign): Do not init to -1.
5635 * config/i386/i386.c (ix86_option_override_internal):
5636 Check opts_set, not opts when setting default value of
5637 opts->x_ix86_force_align_arg_pointer.
5639 2017-08-23 Richard Biener <rguenther@suse.de>
5641 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
5642 lang_hooks.decl_printable_name.
5643 * print-rtl-function.c (print_rtx_function): Likewise.
5644 * tree-pretty-print.c (dump_function_header): Likewise.
5646 2017-08-23 Richard Biener <rguenther@suse.de>
5649 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
5650 -g0 at compile-time.
5652 2017-08-23 Tamar Christina <tamar.christina@arm.com>
5655 * doc/sourcebuild.texi (Other hardware attributes):
5658 2017-08-23 Tamar Christina <tamar.christina@arm.com>
5661 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
5662 Fix single-use check.
5664 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
5666 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
5668 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
5670 * doc/install.texi: Modify to add more details on running selected
5673 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
5675 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
5676 is combined with -mabi=ms.
5677 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
5680 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5682 PR tree-optimization/81488
5683 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
5684 and cached_basis fields.
5685 (MAX_SPREAD): New constant.
5686 (alloc_cand_and_find_basis): Initialize new fields.
5687 (clear_visited): New function.
5688 (create_phi_basis_1): Rename from create_phi_basis, set visited
5689 and cached_basis fields.
5690 (create_phi_basis): New wrapper function.
5691 (phi_add_costs_1): Rename from phi_add_costs, add spread
5692 parameter, set visited field, short-circuit when limits reached.
5693 (phi_add_costs): New wrapper function.
5694 (record_phi_increments_1): Rename from record_phi_increments, set
5696 (record_phi_increments): New wrapper function.
5697 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
5698 (phi_incr_cost): New wrapper function.
5699 (all_phi_incrs_profitable_1): Rename from
5700 all_phi_incrs_profitable, set visited field.
5701 (all_phi_incrs_profitable): New wrapper function.
5703 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
5704 Alan Hayward <alan.hayward@arm.com>
5705 David Sherwood <david.sherwood@arm.com>
5707 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
5708 that takes the outer and inner modes.
5709 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
5710 comparison as the canonical test for a paradoxical subreg.
5711 * combine.c (simplify_set): Use paradoxical_subreg_p.
5712 (make_extraction): Likewise.
5713 (force_to_mode): Likewise.
5714 (rtx_equal_for_field_assignment_p): Likewise.
5715 (gen_lowpart_for_combine): Likewise.
5716 (simplify_comparison): Likewise.
5717 * cse.c (equiv_constant): Likewise.
5718 * expmed.c (store_bit_field_1): Likewise.
5719 * final.c (alter_subreg): Likewise.
5720 * fwprop.c (propagate_rtx): Likewise.
5721 (forward_propagate_subreg): Likewise.
5722 * ira-conflicts.c (ira_build_conflicts): Likewise.
5723 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
5724 * lra-constraints.c (curr_insn_transform): Likewise.
5725 (split_reg): Likewise.
5726 * lra-eliminations.c (move_plus_up): Likewise.
5727 (lra_eliminate_regs_1): Likewise.
5728 * recog.c (general_operand): Likewise.
5729 * ree.c (combine_reaching_defs): Likewise.
5730 * reload.c (push_reload): Likewise.
5731 (find_reloads): Likewise.
5732 * reload1.c (elimination_effects): Likewise.
5733 (compute_reload_subreg_offset): Likewise.
5734 (choose_reload_regs): Likewise.
5735 * rtlanal.c (subreg_lsb_1): Likewise.
5736 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5737 (simplify_subreg): Likewise.
5738 * var-tracking.c (track_loc_p): Likewise.
5739 * emit-rtl.c (byte_lowpart_offset): Likewise.
5740 (paradoxical_subreg_p): Delete out-of-line definition.
5742 2017-08-22 Jeff Law <law@redhat.com>
5744 PR tree-optimization/81741
5745 PR tree-optimization/71947
5746 * tree-ssa-dom.c: Include tree-inline.h.
5747 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
5748 equivalences if one is more expensive to compute than the other.
5749 * tree-ssa-scopedtables.h (class const_or_copies): Make
5750 record_const_or_copy_raw method private.
5751 (class avail_exprs_stack): New method simplify_binary_operation.
5752 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
5753 avail_exprs_stack::simplify_binary_operation as needed.
5754 (avail_exprs_stack::simplify_binary_operation): New function.
5756 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
5758 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
5759 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
5760 (DOT_SYMBOLS): Likewise.
5761 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
5762 (RELOCATABLE_NEEDS_FIXUP): Likewise.
5763 (RS6000_ABI_NAME): Likewise.
5764 (TARGET_CMODEL): Likewise.
5765 (TOC_SECTION_ASM_OP): Likewise.
5766 (SET_CMODEL): New macro.
5767 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
5769 2017-08-22 Richard Biener <rguenther@suse.de>
5771 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
5772 to allow for free-lang-data replacements similar to verify_type_variant.
5774 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
5775 Alan Hayward <alan.hayward@arm.com>
5776 David Sherwood <david.sherwood@arm.com>
5778 * config/aarch64/aarch64.md (casesi): Use DImode rather than
5779 VOIDmode for the LABEL_REF.
5781 2017-08-22 Richard Biener <rguenther@suse.de>
5783 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
5785 2017-08-22 Richard Biener <rguenther@suse.de>
5787 * common.opt (feliminate-dwarf2-dups): Ignore.
5788 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
5789 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
5790 same_die_p_wrap, compute_section_prefix,
5791 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
5792 (comdat_symbol_id, comdat_symbol_number): Likewise.
5793 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
5795 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
5796 (output_die): Mark unreachable path unreachable.
5797 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
5798 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
5799 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
5800 (dwarf2out_early_finish): Likewise.
5802 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
5804 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
5806 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
5809 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
5810 not VAR_P. Filter attribute warnings with OPT_Wattributes.
5811 (avr_attribute_table) <io, io_low, address>: Initialize
5812 .decl_required with true.
5814 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
5816 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
5817 undocumented debugging options.
5818 (-mvsx-scalar-double): Likewise.
5819 (-mallow-df-permute): Likewise.
5820 (-mvectorize-builtins): Likewise.
5821 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
5822 (rs6000_builtin_vectorized_function): Likewise.
5823 (rs6000_builtin_md_vectorized_function): Likewise.
5824 (rs6000_opt_vars): Likewise.
5826 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
5829 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
5830 (*btrq_imm): Rename from *btrq.
5831 (*btcq_imm): Rename from *btcq.
5832 (btsc): New code attribute.
5833 (*<btsc><mode>): New insn pattern.
5834 (*btr<mode>): Ditto.
5835 (*<btsc><mode>_mask): New insn_and_split pattern.
5836 (*btr<mode>_mask): Ditto.
5838 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5839 Alan Hayward <alan.hayward@arm.com>
5840 David Sherwood <david.sherwood@arm.com>
5842 * function.c (pad_below): Simplify padding calculation.
5844 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5845 Alan Hayward <alan.hayward@arm.com>
5846 David Sherwood <david.sherwood@arm.com>
5848 * target.def (function_prologue): Remove frame size argument.
5849 (function_epilogue): Likewise.
5850 * doc/tm.texi: Regenerate.
5851 * final.c (final_start_function): Update call to function_prologue.
5852 (final_end_function): Update call to function_epilogue.
5853 (default_function_pro_epilogue): Remove frame size argument.
5854 * output.h (default_function_pro_epilogue): Likewise.
5855 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
5856 (arm_output_function_prologue): Likewise.
5857 * config/frv/frv.c (frv_function_prologue): Likewise.
5858 (frv_function_epilogue): Likewise.
5859 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
5860 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
5861 (ia64_output_function_epilogue): Likewise.
5862 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
5863 (m32r_output_function_epilogue): Likewise.
5864 * config/microblaze/microblaze.c (microblaze_function_prologue)
5865 (microblaze_function_epilogue): Likewise.
5866 * config/mips/mips.c (mips_output_function_prologue): Likewise.
5867 (mips_output_function_epilogue): Likewise.
5868 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
5869 (mmix_target_asm_function_epilogue): Likewise.
5870 * config/msp430/msp430.c (msp430_start_function): Likewise.
5871 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
5872 (nds32_asm_function_epilogue): Likewise.
5873 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
5874 * config/pa/pa.c (pa_output_function_prologue): Likewise.
5875 (pa_output_function_epilogue): Likewise.
5876 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
5877 (rs6000_output_function_epilogue): Likewise.
5878 * config/rl78/rl78.c (rl78_start_function): Likewise.
5879 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
5880 (rs6000_output_function_epilogue): Likewise.
5881 * config/rx/rx.c (rx_output_function_prologue): Likewise.
5882 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
5883 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
5884 (sparc_asm_function_epilogue): Likewise.
5886 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5888 * tree.h (type_has_mode_precision_p): New function.
5889 * convert.c (convert_to_integer_1): Use it.
5890 * expr.c (expand_expr_real_2): Likewise.
5891 (expand_expr_real_1): Likewise.
5892 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
5893 * match.pd: Likewise.
5894 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
5895 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
5896 * tree-tailcall.c (process_assignment): Likewise.
5897 * tree-vect-loop.c (vectorizable_reduction): Likewise.
5898 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
5899 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
5900 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
5901 (vectorizable_assignment): Likewise.
5902 (vectorizable_shift): Likewise.
5903 (vectorizable_operation): Likewise.
5904 * tree-vrp.c (register_edge_assert_for_2): Likewise.
5906 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
5908 * match.pd: Add pow (C, x) simplification.
5910 2017-08-21 Richard Biener <rguenther@suse.de>
5912 PR tree-optimization/81900
5913 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
5914 for blocks with abnormal predecessors.
5915 (compute_antic): Do not set visited flag prematurely.
5917 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
5920 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
5922 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5924 * stor-layout.h (vector_type_mode): Move to...
5925 * tree.h (vector_type_mode): ...here.
5926 * stor-layout.c (vector_type_mode): Move to...
5927 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
5929 2017-08-21 Richard Biener <rguenther@suse.de>
5931 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
5932 register_external_die hooks.
5933 (debug_false_tree_charstarstar_uhwistar): Declare.
5934 (debug_nothing_tree_charstar_uhwi): Likewise.
5935 * debug.c (do_nothing_debug_hooks): Adjust.
5936 (debug_false_tree_charstarstar_uhwistar): New do nothing.
5937 (debug_nothing_tree_charstar_uhwi): Likewise.
5938 * dbxout.c (dbx_debug_hooks): Adjust.
5939 (xcoff_debug_hooks): Likewise.
5940 * sdbout.c (sdb_debug_hooks): Likewise.
5941 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
5942 * dwarf2out.c (macinfo_label_base): New global.
5943 (dwarf2out_register_external_die): New function for the
5944 register_external_die hook.
5945 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
5946 (dwarf2_debug_hooks): Use them.
5947 (dwarf2_lineno_debug_hooks): Adjust.
5948 (struct die_struct): Add with_offset flag.
5949 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
5950 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
5951 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
5952 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
5953 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
5954 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
5955 defining section names for the early LTO debug variants.
5956 (reset_indirect_string): New helper.
5957 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
5958 (print_dw_val): Add support for offsetted symbol references.
5959 (get_ultimate_context): Split out from is_cxx.
5960 (is_cxx): Use get_ultimate_context.
5961 (is_fortran): Add decl overload.
5962 (compute_comp_unit_symbol): Split out worker from
5963 compute_section_prefix.
5964 (compute_section_prefix): Call compute_comp_unit_symbol and
5965 set comdat_type_p here.
5966 (output_die): Skip DIE symbol output for the LTO added one.
5967 Handle DIE symbol references with offset.
5968 (output_comp_unit): Guard section name mangling properly.
5969 For LTO debug sections emit a symbol at the section beginning
5970 which we use to refer to its DIEs.
5971 (add_abstract_origin_attribute): For DIEs registered via
5972 dwarf2out_register_external_die directly refer to the early
5973 DIE rather than indirectly through the shadow one we created.
5974 Remove obsolete call to dwarf2out_abstract_function for
5975 non-function/block origins.
5976 (gen_array_type_die): When generating early LTO debug do
5977 not emit DW_AT_string_length.
5978 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
5979 late when in LTO. As suggested place a gcc_unreachable for
5980 the DECL_ABSTRACT_P case.
5981 (gen_subprogram_die): Avoid another specification DIE
5982 for early built declarations/definitions for the late LTO case.
5983 (gen_variable_die): Add type references for late duplicated VLA dies
5985 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
5986 we have the abstract instance already.
5987 (process_scope_var): Adjust decl DIE contexts in LTO which
5988 first puts them in limbo.
5989 (gen_decl_die): Do not generate type DIEs late apart from
5990 types for VLAs or for decls we do not yet have a DIE. Do not
5991 call dwarf2out_abstract_function late.
5992 (dwarf2out_early_global_decl): Make sure to create DIEs
5993 for abstract instances of a decl first.
5994 (dwarf2out_late_global_decl): Adjust comment.
5995 (output_macinfo_op): With multiple macro sections use
5996 macinfo_label_base to distinguish labels.
5997 (output_macinfo): Likewise. Update macinfo_label_base.
5998 Pass in the line info label.
5999 (note_variable_value_in_expr): When generating LTO resolve
6000 all variable values here by generating DIEs as needed.
6001 (init_sections_and_labels): Add early LTO debug flag parameter
6002 and generate different sections and names if set. Add generation
6003 counter for the labels so we can have multiple of them.
6004 (reset_dies): Helper to allow DIEs to be output multiple times.
6005 (dwarf2out_finish): When outputting DIEs to the fat part of an
6006 LTO object first reset DIEs.
6007 (dwarf2out_early_finish): Output early DIEs when generating LTO.
6008 (modified_type_die): Check for decl_ultimate_origin being self
6010 (gen_type_die_with_usage): Likewise.
6011 (gen_typedef_die): Allow decl_ultimate_origin being self.
6012 (set_decl_abstract_flags): Remove.
6013 (set_block_abstract_flags): Likewise.
6014 (dwarf2out_abstract_function): Treat the early generated DIEs
6015 as the abstract copy and only add DW_AT_inline and
6016 DW_AT_artificial here and call set_decl_origin_self.
6017 If the DIE has an abstract origin don't do anything.
6018 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
6019 if we have none yet (Go fails to build one, PR78628).
6020 (variably_modified_type_p): Prevent endless recursion for Ada
6021 cyclic pointer types.
6022 * lto-streamer-in.c: Include debug.h.
6023 (dref_queue): New global.
6024 (lto_read_tree_1): Stream in DIE references.
6025 (lto_input_tree): Register DIE references.
6026 (input_function): Stream DECL_DEBUG_ARGS.
6027 * lto-streamer-out.c: Include debug.h.
6028 (lto_write_tree_1): Output DIE references.
6029 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
6030 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
6031 (output_function): Stream DECL_DEBUG_ARGS.
6032 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
6033 Stream DECL_ABSTRACT_ORIGIN.
6034 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
6035 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
6036 DECL_CONTEXT for file-scope decls.
6037 * lto-streamer.h (struct dref_entry): Declare.
6038 (dref_queue): Likewise.
6039 * cfgexpand.c (pass_expand::execute): Do not call the
6040 outlining_inline_function hook here.
6041 * lto-wrapper.c (debug_obj): New global.
6042 (tool_cleanup): Unlink it if required.
6043 (debug_objcopy): New function.
6044 (run_gcc): Handle early debug sections in the IL files by
6045 extracting them to separate files, partially linkin them and
6046 feeding the result back as result to the linker.
6047 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
6048 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
6049 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
6050 sections into a separate segment.
6051 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
6053 (darwin_asm_dwarf_section): Likewise.
6054 (darwin_asm_output_dwarf_offset): Likewise.
6055 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
6057 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
6058 Alan Hayward <alan.hayward@arm.com>
6059 David Sherwood <david.sherwood@arm.com>
6061 * read-md.h (md_reader::record_potential_iterator_use): Replace
6062 pointer argument with an rtx and an index.
6063 * read-rtl.c (iterator_group::apply_iterator): Likewise.
6064 (apply_mode_iterator): Likewise.
6065 (apply_code_iterator): Likewise.
6066 (apply_int_iterator): Likewise.
6067 (apply_subst_iterator): Likewise.
6068 (record_iterator_use): Likewise.
6069 (record_attribute_use): Likewise.
6070 (md_reader::record_potential_iterator_use): Likewise. Update calls
6071 to record_iterator_use and apply_iterator.
6072 (iterator_use): Replace ptr with x and index.
6073 (attribute_use): Likewise.
6074 (apply_attribute_uses): Update calls to apply_iterator.
6075 (apply_iterators): Likewise. Update initialization of iterator_use.
6076 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
6077 and record_potential_iterator_use.
6078 (rtx_reader::read_rtx_operand): Likewise.
6080 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
6081 Alan Hayward <alan.hayward@arm.com>
6082 David Sherwood <david.sherwood@arm.com>
6084 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
6087 2017-08-21 Richard Biener <rguenther@suse.de>
6090 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
6091 at struct end conservatively when comparing common bases.
6093 2017-08-21 Richard Biener <rguenther@suse.de>
6095 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
6096 (mem_ref_in_stmt): Remove.
6097 (determine_max_movement): Use ref index to get at the reference.
6098 (invariantness_dom_walker::before_dom_children): Deal with
6099 lim data already initialized.
6100 (gather_mem_refs_stmt): Initialize lim data and record ref index.
6102 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
6104 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
6105 (TARGET_ISA_ROUND): Ditto.
6106 (TARGET_ROUND): Ditto.
6107 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
6108 * config/i386/i386.md: Ditto.
6109 * config/i386/sse.md: Ditto.
6110 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
6111 with OPTION_MASK_ISA_SSE4_1.
6113 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
6116 * doc/extend.texi (x86 Built-in Functions): Correct the name of
6117 __builtin_ia32_lzcnt_u16.
6119 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
6122 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
6123 (rs6000_set_current_function): Rewrite function to use it.
6125 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
6128 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
6129 and TYPE_WARN_IF_NOT_ALIGN.
6130 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
6131 (handle_warn_if_not_align): New.
6132 (place_union_field): Call handle_warn_if_not_align.
6133 (place_field): Call handle_warn_if_not_align.
6134 Copy TYPE_WARN_IF_NOT_ALIGN.
6135 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
6136 (layout_type): Likewise.
6137 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
6139 (tree_decl_common): Add warn_if_not_align.
6140 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
6141 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
6142 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
6143 (DECL_WARN_IF_NOT_ALIGN): Likewise.
6144 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
6145 * doc/extend.texi: Document warn_if_not_aligned attribute.
6146 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
6148 2017-08-17 Martin Liska <mliska@suse.cz>
6151 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
6152 (get_data_dependence): Use it as pointer type.
6153 (distribute_loop): Likewise.
6155 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6157 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
6158 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
6159 (p8_vmrgow_<mode>_direct): New define_insn.
6160 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
6161 handle endianness for vmrgew and vmrgow permute patterns.
6163 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
6165 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
6166 * config/rs6000/rs6000-cpus.def: Remove comment.
6167 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
6168 (POWERPC_MASKS): Likewise.
6169 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
6170 use of TARGET_VSX_TIMODE.
6171 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
6172 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
6173 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
6174 (rs6000_option_override_internal): Remove dead code.
6175 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
6176 (rs6000_legitimize_reload_address): Likewise.
6177 (rs6000_legitimate_address_p): Likewise.
6178 (rs6000_opt_masks): Delete "vsx-timode".
6179 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
6180 from function comment.
6181 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
6182 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
6183 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
6185 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
6186 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
6187 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
6190 (define_peephole2): Likewise.
6192 2017-08-17 Martin Sebor <msebor@redhat.com>
6195 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
6196 past the end of an array.
6197 (test_pp_format): Add test cases.
6199 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
6201 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
6202 missing ECF_NOTHROW flags.
6204 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
6207 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
6208 operands residing in integer registers.
6209 (*vsx_le_perm_load_<mode>): Likewise.
6210 (*vsx_le_perm_store_<mode>): Likewise.
6211 (define_peephole2): Add peepholes to optimize the above.
6213 2017-08-17 Marek Polacek <polacek@redhat.com>
6216 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
6217 to mimic what shorten_compare did. Change the return type to bool.
6218 (fold_cond_expr_with_comparison): Update call to
6219 operand_equal_for_comparison_p.
6220 (fold_ternary_loc): Likewise.
6222 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
6224 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
6226 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
6228 2017-08-17 Richard Biener <rguenther@suse.de>
6230 * tree-ssa-structalias.c (solve_graph): When propagating
6231 to successors update the graphs succ edges and avoid duplicate work.
6233 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
6236 * config/i386/i386.c (ix86_option_override_internal): Save target
6237 specific options after ix86_stack_protector_guard_reg was changed.
6239 2017-08-17 Richard Biener <rguenther@suse.de>
6241 PR tree-optimization/81827
6242 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
6243 (new_var_info): Initialize it conservatively.
6244 (get_call_vi): Mark register vars.
6245 (new_scalar_tmp_constraint_exp): Likewise.
6246 (handle_rhs_call): Likewise.
6247 (handle_const_call): Likewise.
6248 (create_function_info_for): Likewise.
6249 (solve_constraints): Sort varinfos to separate register from
6250 non-register vars to pack points-to solution bitmaps during
6253 2017-08-17 Marek Polacek <polacek@redhat.com>
6255 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
6257 2017-08-17 Richard Biener <rguenther@suse.de>
6259 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
6260 to true when overflow is undefined and we saturated the result.
6262 2017-08-17 Alan Modra <amodra@gmail.com>
6265 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
6266 Don't use store multiple if only one reg needs saving.
6267 (interesting_frame_related_regno): New function.
6268 (rs6000_frame_related): Don't emit frame info for regs that
6270 (rs6000_emit_epilogue): Likewise.
6272 2017-08-16 Nathan Sidwell <nathan@acm.org>
6274 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
6275 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
6276 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
6277 * tree.c (free_lang_data_in_type): Use else-if chain. Always
6278 clear TYPE_LANG_1. Remove obsolete member-function stripping.
6279 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
6280 (verify_type): Adjust for TYPE_BINFO move.
6281 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
6282 process TYPE_BINFO directly.
6283 (hash_tree): Likewise.
6284 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
6286 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
6289 2017-08-16 David Malcolm <dmalcolm@redhat.com>
6291 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
6293 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
6296 * config/i386/i386.md (*anddi_1_btr): Change predicates of
6297 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
6298 Add ix86_binary_operator_ok to insn constraint.
6299 (*iordi_1_bts): Ditto.
6300 (*xordi_1_btc): Ditto.
6301 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
6302 Update corresponding peephole2 pattern.
6306 2017-08-16 Bin Cheng <bin.cheng@arm.com>
6308 PR tree-optimization/81832
6309 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
6310 copy loop header which has IFN_LOOP_DIST_ALIAS call.
6312 2017-08-16 Marek Polacek <polacek@redhat.com>
6315 * fold-const.c (fold_indirect_ref_1): Restore original behavior
6316 regarding size_zero_node.
6318 2017-08-16 Martin Liska <mliska@suse.cz>
6321 * config.gcc: Respect previously set extra_objs in case
6324 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
6326 PR tree-optimization/81835
6327 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
6328 the phi SSA_NAME. Check that the condition in a COND_EXPR does
6329 not depend on the phi.
6331 2017-08-16 Alan Modra <amodra@gmail.com>
6333 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
6336 2017-08-16 Alan Modra <amodra@gmail.com>
6338 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
6339 (save_reg_p): ..into this. Update all callers.
6340 (first_reg_to_save): Simplify.
6342 2017-08-16 Alan Modra <amodra@gmail.com>
6344 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
6347 2017-08-15 Joseph Myers <joseph@codesourcery.com>
6351 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
6352 constant count if that count is less than 32.
6354 2017-08-15 Nathan Sidwell <nathan@acm.org>
6356 * gcc.c (execute): Emit friendlier message if inferior is killed
6357 by an external cause.
6359 2017-08-15 Richard Biener <rguenther@suse.de>
6361 PR tree-optimization/81790
6362 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
6363 CONSTRUCTORs from simplifying and VN.
6365 2017-08-14 Martin Sebor <msebor@redhat.com>
6367 * builtin-attrs.def: Add comments.
6369 2017-08-14 Martin Sebor <msebor@redhat.com>
6372 * doc/extend.texi (attribute nonstring): Document new attribute.
6374 2017-08-14 Martin Sebor <msebor@redhat.com>
6377 * tree-diagnostic.c (default_tree_printer): Handle %G.
6378 * gimple-pretty-print.h (percent_G_format): Declare new function.
6379 * gimple-pretty-print.c (percent_G_format): Define.
6380 * tree-pretty-print.c (percent_K_format): Add argument.
6382 2017-08-14 Martin Sebor <msebor@redhat.com>
6384 PR translation/79998
6385 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
6386 Remove a stray space.
6388 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
6391 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
6392 (*iordi_1_bts): Ditto.
6393 (*xordi_1_btc): Ditto.
6395 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6398 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
6399 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6401 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
6402 quoted strings, and make more translator-friendly.
6403 (darwin_rs6000_override_options): Likewise.
6404 (rs6000_option_override_internal): Likewise.
6405 (rs6000_return_in_memory): Fix overlong line.
6406 (init_cmulative_args): Use quoted strings, and make more
6407 translator-friendly.
6408 (rs6000_pass_by_reference): Fix overlong line.
6409 (def_builtin): Use quoted strings.
6410 (altivec_expand_predicate_builtin): Use quoted strings, and make
6411 more translator-friendly.
6412 (htm_expand_builtin): Use quoted strings.
6413 (cpu_expand_builtin): Use quoted strings, and make more
6414 translator-friendly.
6415 (altivec_expand_builtin): Likewise.
6416 (paired_expand_predicate_builtin): Likewise.
6417 (rs6000_invalid_builtin): Likewise.
6418 (builtin_function_type): Use quoted strings.
6419 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
6420 more translator-friendly.
6421 (rs6000_trampoline_init): Likewise.
6422 (rs6000_handle_altivec_attribute): Likewise.
6423 (rs6000_inner_target_options): Use quoted strings.
6424 (rs6000_disable_incompatible_switches): Likewise.
6425 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
6426 strings, and make more translator-friendly.
6427 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
6429 2017-08-14 Bin Cheng <bin.cheng@arm.com>
6431 PR tree-optimization/81799
6432 * tree-loop-distribution.c (version_loop_by_alias_check): Force
6433 cond_expr to simple gimple operand.
6435 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
6438 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
6440 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
6444 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
6445 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
6446 TARGET_GASISR_PROLOGUES.
6447 * config/avr/avr.c (avr_option_override): Same.
6448 (avr_pass_pre_proep::execute): Same.
6450 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
6453 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
6454 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
6456 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
6458 * config/i386/i386.md (*load_tp_<mode>): Redefine as
6459 define_insn_and_split. Split to a memory load from 0 in
6460 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
6461 using PTR mode iterator.
6462 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
6463 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
6464 (*add_tp_<mode>): Redefine as define_insn_and_split.
6465 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
6466 address space. Merge with *add_tp_x32 using PTR mode iterator.
6467 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
6468 Split to an add with a memory load from 0 in
6469 DEFAULT_TLS_SEG_REG address space.
6471 2017-08-12 Andrew Pinski <apinski@cavium.com>
6473 * config/aarch64/aarch64-option-extensions.def (rdma):
6474 Fix feature string to what Linux prints out in /proc/cpuinfo.
6476 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
6479 * dwarf2out.c (modified_type_die): For C typedef types that have
6480 an ultimate origin, process the ultimate origin instead of the
6482 (gen_typedef_die): Assert that input DECLs have no ultimate
6484 (gen_type_die_with_usage): For typedef variants that have an
6485 ultimate origin, just call gen_decl_die on the original DECL.
6486 (process_scope_var): Avoid creating DIEs for local typedefs and
6487 concrete static variables.
6489 2017-08-12 Alan Modra <amodra@gmail.com>
6493 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
6494 match gnu-user.h startfile.
6495 (ENDFILE_LINUX_SPEC): Similarly.
6497 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
6500 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
6502 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
6504 2017-08-11 Tamar Christina <tamar.christina@arm.com>
6505 * config/aarch64/aarch64.md (mov<mode>): Change.
6506 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
6507 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
6508 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
6510 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
6512 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
6513 for storage order barriers.
6515 2017-08-11 Martin Liska <mliska@suse.cz>
6518 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
6519 variables of void type.
6521 2017-08-11 Martin Liska <mliska@suse.cz>
6523 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
6524 TARGET_SUPPORTS_ALIASES.
6525 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
6526 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
6527 (optimize_weakref): Likewise.
6528 * symtab.c (symtab_node::noninterposable_alias): Likewise.
6529 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
6530 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
6532 2017-08-11 Martin Liska <mliska@suse.cz>
6535 * config/i386/i386.c (make_resolver_func): Do complete
6536 refactoring of the function.
6538 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
6541 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
6542 * config/i386/i386.c (ix86_stack_protect_guard): Use
6543 ix86_stack_protect_guard_symbol_str to generate varible declaration.
6544 * doc/invoke.texi (x86 Options): Document
6545 -mstack-protector-guard-symbol= option.
6547 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
6549 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
6550 * config/i386/i386.c (ix86_split_stack_guard): New function.
6551 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
6552 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
6553 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
6554 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
6555 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
6556 (split_stack_space_check): Call ix86_split_stack_guard.
6558 2017-08-10 Martin Sebor <msebor@redhat.com>
6560 * print-tree.c (print_node): Print location using the established
6562 Replace spaces with colons.
6563 (debug_raw, debug): Ditto.
6565 2017-08-10 Martin Sebor <msebor@redhat.com>
6568 * pretty-print.c (pp_format): Correct the handling of %s precision.
6570 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
6573 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
6575 (ix86_finalize_stack_frame_flags): This. Also clear
6576 frame_pointer_needed if -fno-omit-frame-pointer is used without
6578 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
6579 with ix86_finalize_stack_frame_flags.
6580 (ix86_expand_epilogue): Likewise.
6581 (ix86_expand_split_stack_prologue): Likewise.
6582 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
6584 2017-08-10 Martin Liska <mliska@suse.cz>
6587 * c-attribs.c (handle_target_attribute):
6588 Report warning for an empty string argument of target attribute.
6590 2017-08-09 Jakub Jelinek <jakub@redhat.com>
6593 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
6595 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
6596 or DECL_NONLOCAL labels.
6597 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
6598 or DECL_NONLOCAL labels here.
6600 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
6602 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
6603 to indicate when early gimple folding has been disabled.
6604 (rs6000_gimple_fold_builtin): Add debug content.
6605 (rs6000_invalid_builtin): Fix whitespace.
6606 (rs6000_expand_builtin): Fix whitespace.
6607 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
6609 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
6612 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
6613 SAVE_MULTIPLE if not all the registers that saves, should be saved.
6615 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
6617 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
6618 (qdf24xx): Likewise.
6619 * config/aarch64/aarch64.md: Include falkor.md.
6620 * config/aarch64/falkor.md: New.
6622 2017-08-09 Marek Polacek <polacek@redhat.com>
6625 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
6626 * diagnostic.c (emit_diagnostic): Add a comment.
6627 (emit_diagnostic_valist): New function.
6629 2017-08-09 Marek Polacek <polacek@redhat.com>
6632 * input.c (make_location): New overload.
6633 * input.h (make_location): Declare.
6635 2017-08-08 Alan Modra <amodra@gmail.com>
6636 H.J. Lu <hongjiu.lu@intel.com>
6639 * gcc.c (NO_PIE_SPEC): Delete.
6640 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
6642 (LINK_PIE_SPEC): ..to here.
6643 (LINK_COMMAND_SPEC): Support -no-pie.
6644 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
6645 chain of crtbegin*.o selection, update for PIE_SPEC changes and
6647 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
6648 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
6649 (ENDFILE_CRTEND_SPEC): Similarly.
6651 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
6654 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
6655 (mstack-protector-guard-offset=): Ditto.
6656 * config/i386/i386.c (ix86_option_override): Handle
6657 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
6659 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
6660 ix86_stack_protect_guard_offset variables.
6661 (TARGET_STACK_PROTECT_GUARD): Always define.
6662 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
6663 and -mstack-protector-guard-offset= options.
6665 2017-08-08 Bin Cheng <bin.cheng@arm.com>
6667 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
6668 boundary case for the last candidate.
6670 2017-08-08 Bin Cheng <bin.cheng@arm.com>
6672 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
6673 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
6675 2017-08-08 Tamar Christina <tamar.christina@arm.com>
6678 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
6679 * config/aarch64/aarch64-builtins.c
6680 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
6681 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
6682 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
6684 2017-08-08 Tamar Christina <tamar.christina@arm.com>
6685 Andrew Pinski <pinskia@gmail.com>
6688 * internal-fn.def (XORSIGN): New.
6689 * optabs.def (xorsign_optab): New.
6690 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
6691 (convert_expand_mult_copysign): New.
6692 (pass_optimize_widening_mul::execute): Call
6693 convert_expand_mult_copysign.
6695 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6697 PR tree-optimization/81354
6698 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
6699 Insert on edges rather than explicitly creating landing pads.
6700 (analyze_candidates_and_replace): Commit edge inserts.
6702 2017-08-08 Richard Biener <rguenther@suse.de>
6705 * tree-ssa-loop-niter.c: Include tree-dfa.h.
6706 (expand_simple_operations): Also look through ADDR_EXPRs with
6707 MEM_REF bases treating them as POINTER_PLUS_EXPR.
6709 2017-08-08 Richard Biener <rguenther@suse.de>
6711 PR tree-optimization/81723
6712 * tree-vect-slp.c (struct bst_traits): New hash traits.
6713 (bst_fail): New global.
6714 (vect_build_slp_tree_2): New worker, split out from ...
6715 (vect_build_slp_tree): ... this now wrapping it with using
6716 bst_fail set to cache SLP tree build fails. Properly handle
6718 (vect_analyze_slp_instance): Allocate and free bst_fail.
6720 2017-08-08 Martin Liska <mliska@suse.cz>
6723 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
6724 LABEL_DECLs that can be from a different function.
6726 2017-08-08 Bin Cheng <bin.cheng@arm.com>
6728 PR tree-optimization/81744
6729 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
6730 loop's number of iterations.
6732 2017-08-08 Martin Liska <mliska@suse.cz>
6734 * asan.c: Include header files.
6735 * attribs.c (build_decl_attribute_variant): New function moved
6737 (build_type_attribute_qual_variant): Likewise.
6738 (cmp_attrib_identifiers): Likewise.
6739 (simple_cst_list_equal): Likewise.
6740 (omp_declare_simd_clauses_equal): Likewise.
6741 (attribute_value_equal): Likewise.
6742 (comp_type_attributes): Likewise.
6743 (build_type_attribute_variant): Likewise.
6744 (lookup_ident_attribute): Likewise.
6745 (remove_attribute): Likewise.
6746 (merge_attributes): Likewise.
6747 (merge_type_attributes): Likewise.
6748 (merge_decl_attributes): Likewise.
6749 (merge_dllimport_decl_attributes): Likewise.
6750 (handle_dll_attribute): Likewise.
6751 (attribute_list_equal): Likewise.
6752 (attribute_list_contained): Likewise.
6753 * attribs.h (lookup_attribute): New function moved from tree.[ch].
6754 (lookup_attribute_by_prefix): Likewise.
6755 * bb-reorder.c: Include header files.
6756 * builtins.c: Likewise.
6757 * calls.c: Likewise.
6758 * cfgexpand.c: Likewise.
6759 * cgraph.c: Likewise.
6760 * cgraphunit.c: Likewise.
6761 * convert.c: Likewise.
6762 * dwarf2out.c: Likewise.
6763 * final.c: Likewise.
6764 * fold-const.c: Likewise.
6765 * function.c: Likewise.
6766 * gimple-expr.c: Likewise.
6767 * gimple-fold.c: Likewise.
6768 * gimple-pretty-print.c: Likewise.
6769 * gimple.c: Likewise.
6770 * gimplify.c: Likewise.
6771 * hsa-common.c: Likewise.
6772 * hsa-gen.c: Likewise.
6773 * internal-fn.c: Likewise.
6774 * ipa-chkp.c: Likewise.
6775 * ipa-cp.c: Likewise.
6776 * ipa-devirt.c: Likewise.
6777 * ipa-fnsummary.c: Likewise.
6778 * ipa-inline.c: Likewise.
6779 * ipa-visibility.c: Likewise.
6781 * lto-cgraph.c: Likewise.
6782 * omp-expand.c: Likewise.
6783 * omp-general.c: Likewise.
6784 * omp-low.c: Likewise.
6785 * omp-offload.c: Likewise.
6786 * omp-simd-clone.c: Likewise.
6787 * opts-global.c: Likewise.
6788 * passes.c: Likewise.
6789 * predict.c: Likewise.
6790 * sancov.c: Likewise.
6791 * sanopt.c: Likewise.
6792 * symtab.c: Likewise.
6793 * toplev.c: Likewise.
6794 * trans-mem.c: Likewise.
6795 * tree-chkp.c: Likewise.
6796 * tree-eh.c: Likewise.
6797 * tree-into-ssa.c: Likewise.
6798 * tree-object-size.c: Likewise.
6799 * tree-parloops.c: Likewise.
6800 * tree-profile.c: Likewise.
6801 * tree-ssa-ccp.c: Likewise.
6802 * tree-ssa-live.c: Likewise.
6803 * tree-ssa-loop.c: Likewise.
6804 * tree-ssa-sccvn.c: Likewise.
6805 * tree-ssa-structalias.c: Likewise.
6806 * tree-ssa.c: Likewise.
6807 * tree-streamer-in.c: Likewise.
6808 * tree-vectorizer.c: Likewise.
6809 * tree-vrp.c: Likewise.
6811 * ubsan.c: Likewise.
6812 * varasm.c: Likewise.
6813 * varpool.c: Likewise.
6814 * tree.c: Remove functions moved to attribs.[ch].
6816 * config/aarch64/aarch64.c: Add attrs.h header file.
6817 * config/alpha/alpha.c: Likewise.
6818 * config/arc/arc.c: Likewise.
6819 * config/arm/arm.c: Likewise.
6820 * config/avr/avr.c: Likewise.
6821 * config/bfin/bfin.c: Likewise.
6822 * config/c6x/c6x.c: Likewise.
6823 * config/cr16/cr16.c: Likewise.
6824 * config/cris/cris.c: Likewise.
6825 * config/darwin.c: Likewise.
6826 * config/epiphany/epiphany.c: Likewise.
6827 * config/fr30/fr30.c: Likewise.
6828 * config/frv/frv.c: Likewise.
6829 * config/ft32/ft32.c: Likewise.
6830 * config/h8300/h8300.c: Likewise.
6831 * config/i386/winnt.c: Likewise.
6832 * config/ia64/ia64.c: Likewise.
6833 * config/iq2000/iq2000.c: Likewise.
6834 * config/lm32/lm32.c: Likewise.
6835 * config/m32c/m32c.c: Likewise.
6836 * config/m32r/m32r.c: Likewise.
6837 * config/m68k/m68k.c: Likewise.
6838 * config/mcore/mcore.c: Likewise.
6839 * config/microblaze/microblaze.c: Likewise.
6840 * config/mips/mips.c: Likewise.
6841 * config/mmix/mmix.c: Likewise.
6842 * config/mn10300/mn10300.c: Likewise.
6843 * config/moxie/moxie.c: Likewise.
6844 * config/msp430/msp430.c: Likewise.
6845 * config/nds32/nds32-isr.c: Likewise.
6846 * config/nds32/nds32.c: Likewise.
6847 * config/nios2/nios2.c: Likewise.
6848 * config/nvptx/nvptx.c: Likewise.
6849 * config/pa/pa.c: Likewise.
6850 * config/pdp11/pdp11.c: Likewise.
6851 * config/powerpcspe/powerpcspe.c: Likewise.
6852 * config/riscv/riscv.c: Likewise.
6853 * config/rl78/rl78.c: Likewise.
6854 * config/rx/rx.c: Likewise.
6855 * config/s390/s390.c: Likewise.
6856 * config/sh/sh.c: Likewise.
6857 * config/sol2.c: Likewise.
6858 * config/sparc/sparc.c: Likewise.
6859 * config/spu/spu.c: Likewise.
6860 * config/stormy16/stormy16.c: Likewise.
6861 * config/tilegx/tilegx.c: Likewise.
6862 * config/tilepro/tilepro.c: Likewise.
6863 * config/v850/v850.c: Likewise.
6864 * config/vax/vax.c: Likewise.
6865 * config/visium/visium.c: Likewise.
6866 * config/xtensa/xtensa.c: Likewise.
6868 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
6871 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
6872 constraints since the -mupper-regs-* switches have been
6874 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
6875 into a vector from a double word element that was extracted from
6876 another vector, and eliminate extra XXPERMDI instructions.
6877 (vsx_concat_<mode>_2): Likewise.
6878 (vsx_concat_<mode>_3): Likewise.
6879 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
6880 concat to allow optimizing inserts from previous extracts.
6882 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
6884 * config/i386/i386.c (ix86_stack_protect_guard): Generate
6885 memory reference to a SSP offset in TLS address space.
6886 (ix86_print_operand) <case '@'>: Remove.
6887 (ix86_print_operand_punct_valid_p): Remove '@' code.
6888 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
6890 (stack_tls_protect_set_<mode>): Remove.
6891 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
6892 (stack_tls_protect_test_<mode>): Remove.
6893 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
6895 2017-08-07 Olivier Hainque <hainque@adacore.com>
6898 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
6900 2017-08-07 Douglas Rupp <rupp@adacore.com>
6902 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
6903 variable was referenced as multidir in command.
6905 2017-08-07 Jakub Jelinek <jakub@redhat.com>
6908 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
6911 2017-08-07 Martin Liska <mliska@suse.cz>
6913 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
6914 * config/rl78/rl78.c: Add include of attribs.h.
6915 * config/sh/sh.c: Likewise.
6916 * config/v850/v850.c: Likewise.
6918 2017-08-07 Tom de Vries <tom@codesourcery.com>
6921 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
6923 2017-08-07 Martin Liska <mliska@suse.cz>
6925 * config/mips/mips.c: Include attribs.h.
6927 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
6930 * doc/invoke.texi: Document change in behvaior for -Ofast for
6933 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
6935 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
6937 (aarch64_pop_regs): Likewise.
6938 (aarch64_gen_load_pair): Likewise.
6939 (aarch64_save_callee_saves): Likewise.
6940 (aarch64_restore_callee_saves): Likewise.
6942 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
6944 * config/i386/i386.c: Revert the last change.
6946 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
6949 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
6951 (ix86_finalize_stack_frame_flags): This. Also clear
6952 frame_pointer_needed if -fno-omit-frame-pointer is used without
6954 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
6955 with ix86_finalize_stack_frame_flags.
6956 (ix86_expand_epilogue): Likewise.
6957 (ix86_expand_split_stack_prologue): Likewise.
6959 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
6962 * config/i386/i386.c (get_builtin_code_for_version): Set priority
6963 to P_AES for Westmere.
6965 2017-08-07 Jonathan Yong <10walls@gmail.com>
6967 * config/i386/mingw.opt (fset-stack-executable): Removed.
6968 * config/i386/cygming.opt (fset-stack-executable): Moved
6970 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
6972 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
6974 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
6976 2017-08-07 Marek Polacek <polacek@redhat.com>
6979 * fold-const.c (fold_indirect_ref_1): Check type_domain.
6981 2017-08-07 Martin Liska <mliska@suse.cz>
6983 * attribs.h (canonicalize_attr_name): New function.
6984 (cmp_attribs): Move from c-format.c and adjusted.
6985 (is_attribute_p): Moved from tree.h.
6986 * tree-inline.c: Add new includes.
6987 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
6988 (private_is_attribute_p): Remove.
6989 (private_lookup_attribute): Likewise.
6990 (private_lookup_attribute_by_prefix): Simplify.
6991 (remove_attribute): Use is_attribute_p.
6992 * tree.h: Remove removed declarations.
6994 2017-08-07 Jakub Jelinek <jakub@redhat.com>
6997 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
6998 instead of computing it in the function. Formatting fix.
6999 (expand_case): Don't rely on default_edge being the first edge,
7000 clear it if removing it, pass default_edge to
7001 emit_case_dispatch_table.
7002 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
7005 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
7007 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
7008 insn in the function, emit NOP after the insn.
7010 2017-08-06 Tom de Vries <tom@codesourcery.com>
7012 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
7015 2017-08-06 Tom de Vries <tom@codesourcery.com>
7017 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
7020 2017-08-04 Yury Gribov <tetra2005@gmail.com>
7022 PR tree-optimization/57371
7023 * match.pd: New pattern.
7025 2017-08-04 Marek Polacek <polacek@redhat.com>
7028 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
7029 perform the computation in offset_int.
7031 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
7033 PR tree-optimization/81136
7034 * tree-vectorizer.h: Include tree-hash-traits.h.
7035 (vec_base_alignments): New typedef.
7036 (vec_info): Add a base_alignments field.
7037 (vect_record_base_alignments): Declare.
7038 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
7040 (DR_IS_CONDITIONAL_IN_STMT): New macro.
7041 (create_data_ref): Add an is_conditional_in_stmt argument.
7042 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
7043 the is_conditional_in_stmt field.
7044 (data_ref_loc): Add an is_conditional_in_stmt field.
7045 (get_references_in_stmt): Set the is_conditional_in_stmt field.
7046 (find_data_references_in_stmt): Update call to create_data_ref.
7047 (graphite_find_data_references_in_stmt): Likewise.
7048 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
7049 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
7050 (vect_record_base_alignment): New function.
7051 (vect_record_base_alignments): Likewise.
7052 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
7053 for nested statements even if we fail to compute a misalignment.
7054 Use pooled base alignments for unconditional references.
7055 (vect_find_same_alignment_drs): Compare base addresses instead
7057 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
7058 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
7060 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
7062 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
7063 Add an explicit name for the enum. Use auto_vec for slp_instances
7065 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
7067 (_bb_vec_info): Add a constructor and destructor.
7068 (vinfo_for_stmt): Return NULL for uids of -1 as well.
7069 (destroy_loop_vec_info): Delete.
7070 (vect_destroy_datarefs): Likewise.
7071 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
7072 (vec_info::vec_info): New function.
7073 (vec_info::~vec_info): Likewise.
7074 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
7075 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
7076 destroy_loop_vec_info.
7077 * tree-vect-loop.c (new_loop_vec_info): Replace with...
7078 (_loop_vec_info::_loop_vec_info): ...this.
7079 (destroy_loop_vec_info): Replace with...
7080 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
7081 the stmt_vec_infos. Leave handling of vec_info information to its
7082 destructor. Remove explicit vector releases.
7083 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
7084 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
7085 * tree-vect-slp.c (new_bb_vec_info): Replace with...
7086 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
7087 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
7088 (destroy_bb_vec_info): Replace with...
7089 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
7090 information to its destructor.
7091 (vect_slp_analyze_bb_1): Use new and delete instead of
7092 new_bb_vec_info and destroy_bb_vec_info.
7093 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
7096 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
7098 * tree-data-ref.h (subscript): Add access_fn field.
7099 (data_dependence_relation): Add could_be_independent_p.
7100 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
7101 (same_access_functions): Move to tree-data-ref.c.
7102 * tree-data-ref.c (ref_contains_union_access_p): New function.
7103 (access_fn_component_p): Likewise.
7104 (access_fn_components_comparable_p): Likewise.
7105 (dr_analyze_indices): Add a reference to access_fn_component_p.
7106 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
7108 (constant_access_functions): Likewise.
7109 (add_other_self_distances): Likewise.
7110 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
7111 (initialize_data_dependence_relation): Use XCNEW and remove
7112 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
7113 of access functions that have the same type. Allow the
7114 subsequence to end with different bases in some circumstances.
7115 Record the chosen access functions in SUB_ACCESS_FN.
7116 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
7117 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
7118 (subscript_dependence_tester_1): Likewise dra and drb.
7119 (build_classic_dist_vector): Update calls accordingly.
7120 (subscript_dependence_tester): Likewise.
7121 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
7122 DDR_COULD_BE_INDEPENDENT_P.
7123 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
7124 comp_alias_ddrs instead of may_alias_ddrs.
7125 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
7127 (vect_analyze_data_ref_dependence): Use it if
7128 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
7129 distance vectors if that fails.
7130 (dependence_distance_ge_vf): New function.
7131 (vect_prune_runtime_alias_test_list): Use it. Don't clear
7132 LOOP_VINFO_MAY_ALIAS_DDRS.
7134 2017-08-04 Richard Biener <rguenther@suse.de>
7137 * fold-const.c (fold_binary_loc): Properly restrict
7138 minus_var0 && minus_var1 case when associating undefined overflow
7141 2017-08-04 Tom de Vries <tom@codesourcery.com>
7143 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
7145 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7147 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7148 Don't start diagnostic messages with a capital letter.
7149 * config/rs6000/rs6000.c (rs6000_option_override_internal):
7151 (rs6000_invalid_builtin): Likewise.
7152 (rs6000_trampoline_init): Likewise.
7154 2017-08-03 Jakub Jelinek <jakub@redhat.com>
7157 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
7158 after setting changeable df flags.
7160 2017-08-03 Richard Biener <rguenther@suse.de>
7162 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
7163 up if the use is in USE - X.
7165 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
7167 * toplev.c (dumpfile.h): New include.
7168 (internal_error_reentered): New static function. Use it...
7169 (internal_error_function): ...here to handle reentered internal_error.
7171 2017-08-03 Richard Biener <rguenther@suse.de>
7174 * fold-const.c (split_tree): Add minus_var and minus_con
7175 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
7176 here but always use minus_*.
7177 (associate_trees): Assert we never associate with MINUS_EXPR
7178 and NULL first operand. Do not recurse for PLUS_EXPR operands
7179 when associating as MINUS_EXPR either.
7180 (fold_binary_loc): Track minus_var and minus_con.
7182 2017-08-03 Tom de Vries <tom@codesourcery.com>
7185 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
7186 ACCEL_COMPILER, apply finish_options on
7187 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
7189 2017-08-03 Tom de Vries <tom@codesourcery.com>
7192 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
7193 function_entry_patch_area_size > 0.
7195 2017-08-03 Jakub Jelinek <jakub@redhat.com>
7198 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
7199 instead of 10??LU, perform unit multiplication in wide_int,
7200 don't change alloc_object_size_limit if the limit is larger
7203 PR tree-optimization/81655
7204 PR tree-optimization/81588
7205 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
7206 the case when ranges[i].low and high are 1 for unsigned type with
7210 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
7211 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
7213 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
7215 * tree-vrp.h: Add include guard.
7217 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
7220 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
7221 (ud2): New insn pattern.
7222 * config/i386/i386.c (ix86_expand_epilogue):
7223 For naked functions, generate ud2 instead of trap insn.
7225 2017-08-02 Marek Polacek <polacek@redhat.com>
7228 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
7230 2017-08-02 Tom de Vries <tom@codesourcery.com>
7231 Cesar Philippidis <cesar@codesourcery.com>
7233 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
7234 Add missing edge probabilities.
7236 2017-08-02 Tamar Christina <tamar.christina@arm.com>
7238 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
7241 2017-08-02 Jakub Jelinek <jakub@redhat.com>
7244 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
7245 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
7246 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
7247 prologue_seq sequences - if any.
7249 2017-08-02 Richard Biener <rguenther@suse.de>
7251 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
7252 via vectors if supported, integer extracts via punning if supported
7253 or otherwise vector extracts.
7255 2017-08-02 Richard Biener <rguenther@suse.de>
7257 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
7259 (bitmap_insert_into_set): ... this.
7261 2017-08-02 Richard Biener <rguenther@suse.de>
7263 PR tree-optimization/81633
7265 2015-08-17 Alan Hayward <alan.hayward@arm.com>
7267 PR tree-optimization/71752
7268 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
7270 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
7272 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
7273 (machine_function::call_ms2sysv_pad_out): Remove field.
7274 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
7275 (ix86_compute_frame_layout): Likewise.
7277 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
7280 * config/i386/i386.c (ix86_set_func_type): Disallow naked
7281 attribute with interrupt attribute.
7283 2017-08-01 Andrew Pinski <apinski@cavium.com>
7285 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
7286 BIT_INSERT_EXPR's operand 1
7287 to see if the types precision matches.
7289 2017-08-01 Martin Liska <mliska@suse.cz>
7292 * builtins.c (expand_builtin_memcpy_args): Remove.
7293 (expand_builtin_memcpy): Call newly added function
7294 expand_builtin_memory_copy_args.
7295 (expand_builtin_memcpy_with_bounds): Likewise.
7296 (expand_builtin_mempcpy): Remove last argument.
7297 (expand_builtin_mempcpy_with_bounds): Likewise.
7298 (expand_builtin_memory_copy_args): New function created from
7299 expand_builtin_mempcpy_args with small modifications.
7300 (expand_builtin_mempcpy_args): Remove.
7301 (expand_builtin_stpcpy): Remove unused argument.
7302 (expand_builtin): Likewise.
7303 (expand_builtin_with_bounds): Likewise.
7305 2017-08-01 Martin Liska <mliska@suse.cz>
7308 Make mempcpy more optimal (PR middle-end/70140).
7310 2017-08-01 Jakub Jelinek <jakub@redhat.com>
7313 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
7314 __builtin_vec_cmpne verify both arguments are compatible vectors
7315 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
7316 verify arg1_type is a pointer or array type. For __builtin_vec_st,
7317 move computation of aligned to after checking the argument types.
7321 * config/rs6000/vsx.md (vextract_fp_from_shorth,
7322 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
7325 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
7326 Doug Rupp <rupp@adacore.com>
7327 Olivier Hainque <hainque@adacore.com>
7329 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
7330 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
7332 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
7333 for TARGET_OS_CPP_BUILTIN.
7334 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
7335 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
7337 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
7338 passing required abi options to the assembler for EABI configurations.
7339 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
7340 of .text.hot and .text.unlikely sections for kernel modules when
7341 using ARM style exceptions.
7342 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
7343 options. Add EXTRA_CC1_SPEC.
7344 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
7346 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
7348 (ARM_TARGET2_DWARF_FORMAT): Define.
7349 * config/arm/t-vxworks: Adjust multilib control to removal of the
7350 Diab command line options.
7352 2017-08-01 Martin Liska <mliska@suse.cz>
7354 PR gcov-profile/81561
7355 * gcov.c (unblock): Make unblocking safe as we need to preserve
7356 index correspondence of blocks and block_lists.
7358 2017-08-01 Richard Biener <rguenther@suse.de>
7360 PR tree-optimization/81181
7361 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
7362 (compute_antic): ... end of iteration here.
7364 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
7366 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
7367 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
7368 (ftree-slp-vectorize): Likewise.
7369 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
7370 can no longer be set independent of flag_tree_loop_vectorize.
7371 * omp-general.c (emp_max_vf): Likewise.
7372 * opts.c (enable_fdo_optimizations): Remove references to
7373 flag_tree_vectorize, these are now implicit.
7374 (common_handle_option): Remove handling for OPT_ftree_vectorize,
7375 and leave it for the options machinery.
7377 2017-08-01 Martin Liska <mliska@suse.cz>
7380 * builtins.c (expand_builtin_memcpy_args): Remove.
7381 (expand_builtin_memcpy): Call newly added function
7382 expand_builtin_memory_copy_args.
7383 (expand_builtin_memcpy_with_bounds): Likewise.
7384 (expand_builtin_mempcpy): Remove last argument.
7385 (expand_builtin_mempcpy_with_bounds): Likewise.
7386 (expand_builtin_memory_copy_args): New function created from
7387 expand_builtin_mempcpy_args with small modifications.
7388 (expand_builtin_mempcpy_args): Remove.
7389 (expand_builtin_stpcpy): Remove unused argument.
7390 (expand_builtin): Likewise.
7391 (expand_builtin_with_bounds): Likewise.
7393 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
7396 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
7397 print "ds:" only for immediates in generic address space.
7399 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
7402 * config/i386/i386.c (ix86_funciton_naked): New prototype.
7403 (ix86_function_ok_for_sibcall): Return false for naked functions.
7405 2017-08-01 Richard Biener <rguenther@suse.de>
7407 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
7408 (compute_antic): Seed worklist with exit block predecessors.
7409 * cfganal.c (dfs_find_deadend): For a cycle return the source
7410 of the edge closing it.
7412 2017-08-01 Tamar Christina <tamar.christina@arm.com>
7414 * config/aarch64/aarch64.c
7415 (aarch64_can_const_movi_rtx_p): Move 0 check.
7417 2017-08-01 Bin Cheng <bin.cheng@arm.com>
7419 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
7420 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
7422 * match.pd: Ditto in address comparison pattern.
7424 2017-08-01 Bin Cheng <bin.cheng@arm.com>
7426 PR tree-optimization/81627
7427 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
7428 closed ssa form for store-store chain.
7430 2017-08-01 Bin Cheng <bin.cheng@arm.com>
7432 PR tree-optimization/81620
7433 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
7434 for store-store chain.
7436 2017-08-01 Jakub Jelinek <jakub@redhat.com>
7438 PR tree-optimization/81588
7439 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
7440 ranges[i].in_p, invert comparison code ccode. For >/>=,
7441 swap rhs1 and rhs2 and comparison code unconditionally,
7442 for </<= don't do that. Don't swap rhs1/rhs2 again if
7443 ranges[i].in_p, instead invert comparison code ccode if
7444 opcode or oe->rank is BIT_IOR_EXPR.
7447 * optabs.def (vec_extract_optab, vec_init_optab): Change from
7448 a direct optab to conversion optab.
7449 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
7450 with GET_MODE_INNER as last argument instead of optab_handler.
7451 * expmed.c (extract_bit_field_1): Likewise. Use vector from
7452 vector extraction if possible and optab is available.
7453 * expr.c (store_constructor): Use convert_optab_handler instead
7454 of optab_handler. Use vector initialization from smaller
7455 vectors if possible and optab is available.
7456 * tree-vect-stmts.c (vectorizable_load): Likewise.
7457 * doc/md.texi (vec_extract, vec_init): Document that the optabs
7459 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
7460 of vec_init from half-sized vectors with the same element mode.
7461 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
7462 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
7463 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
7464 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
7465 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
7466 after mode in gen_vec_extract* calls.
7467 (vec_extract<mode>): Renamed to ...
7468 (vec_extract<mode><ssescalarmodelower>): ... this.
7469 (vec_extract<mode><ssehalfvecmodelower>): New expander.
7470 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
7471 element mode after mode in gen_vec_init* calls.
7472 (VEC_INIT_HALF_MODE): New mode iterator.
7473 (vec_init<mode>): Renamed to ...
7474 (vec_init<mode><ssescalarmodelower>): ... this.
7475 (vec_init<mode><ssehalfvecmodelower>): New expander.
7476 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
7477 (vec_extractv2sfsf): ... this.
7478 (vec_initv2sf): Renamed to ...
7479 (vec_initv2sfsf): ... this.
7480 (vec_extractv2si): Renamed to ...
7481 (vec_extractv2sisi): ... this.
7482 (vec_initv2si): Renamed to ...
7483 (vec_initv2sisi): ... this.
7484 (vec_extractv4hi): Renamed to ...
7485 (vec_extractv4hihi): ... this.
7486 (vec_initv4hi): Renamed to ...
7487 (vec_initv4hihi): ... this.
7488 (vec_extractv8qi): Renamed to ...
7489 (vec_extractv8qiqi): ... this.
7490 (vec_initv8qi): Renamed to ...
7491 (vec_initv8qiqi): ... this.
7492 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
7493 (vec_init<mode>): Renamed to ...
7494 (vec_init<mode><VEC_base_l>): ... this.
7495 (vec_extract<mode>): Renamed to ...
7496 (vec_extract<mode><VEC_base_l>): ... this.
7497 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
7498 (vec_initv2sfsf): ... this.
7499 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
7500 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
7501 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
7502 element mode after mode in gen_vec_init* calls.
7503 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
7504 (vec_init<mode><Vel>): ... this.
7505 (vec_extract<mode>): Renamed to ...
7506 (vec_extract<mode><Vel>): ... this.
7507 * config/aarch64/iterators.md (Vel): New mode attribute.
7508 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
7509 Add element mode after mode in gen_vec_extract* calls.
7510 * config/s390/vector.md (non_vec_l): New mode attribute.
7511 (vec_extract<mode>): Renamed to ...
7512 (vec_extract<mode><non_vec_l>): ... this.
7513 (vec_init<mode>): Renamed to ...
7514 (vec_init<mode><non_vec_l>): ... this.
7515 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
7516 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
7518 * config/arm/iterators.md (V_elem_l): New mode attribute.
7519 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
7520 (vec_extract<mode><V_elem_l>): ... this.
7521 (vec_extractv2di): Renamed to ...
7522 (vec_extractv2didi): ... this.
7523 (vec_init<mode>): Renamed to ...
7524 (vec_init<mode><V_elem_l>): ... this.
7525 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
7526 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
7527 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
7528 Add element mode after gen_vec_extract* calls.
7529 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
7530 (vec_init<mode><unitmode>): ... this.
7531 (vec_extract<mode>): Renamed to ...
7532 (vec_extract<mode><unitmode>): ... this.
7533 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
7534 (vec_init<mode><unitmode>): ... this.
7535 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
7536 (vec_initv2sfsf): ... this.
7537 (vec_extractv2sf): Renamed to ...
7538 (vec_extractv2sfsf): ... this.
7539 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
7540 Add element mode after gen_vec_extract* calls.
7541 * config/mips/mips.md (unitmode): New mode iterator.
7542 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
7543 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
7544 * config/spu/spu.md (inner_l): New mode attribute.
7545 (vec_init<mode>): Renamed to ...
7546 (vec_init<mode><inner_l>): ... this.
7547 (vec_extract<mode>): Renamed to ...
7548 (vec_extract<mode><inner_l>): ... this.
7549 * config/sparc/sparc.md (veltmode): New mode iterator.
7550 (vec_init<VMALL:mode>): Renamed to ...
7551 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
7552 * config/ia64/vect.md (vec_initv2si): Renamed to ...
7553 (vec_initv2sisi): ... this.
7554 (vec_initv2sf): Renamed to ...
7555 (vec_initv2sfsf): ... this.
7556 (vec_extractv2sf): Renamed to ...
7557 (vec_extractv2sfsf): ... this.
7558 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
7559 (vec_init<mode>): Renamed to ...
7560 (vec_init<mode><VEC_base_l>): ... this.
7561 (vec_extract<mode>): Renamed to ...
7562 (vec_extract<mode><VEC_base_l>): ... this.
7563 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
7564 (vec_initv2sfsf): ... this.
7565 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
7566 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
7567 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
7568 gen_vec_init* calls.
7570 2017-08-01 Richard Biener <rguenther@suse.de>
7572 PR tree-optimization/81297
7573 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
7574 TREE_OVERFLOW from INTEGER_CSTs.
7576 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
7578 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
7580 2017-07-31 Carl Love <cel@us.ibm.com>
7582 * config/rs6000/rs6000-c: Add support for built-in functions
7583 vector signed char vec_xl_be (signed long long, signed char *);
7584 vector unsigned char vec_xl_be (signed long long, unsigned char *);
7585 vector signed int vec_xl_be (signed long long, signed int *);
7586 vector unsigned int vec_xl_be (signed long long, unsigned int *);
7587 vector signed long long vec_xl_be (signed long long, signed long long *);
7588 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
7589 vector signed short vec_xl_be (signed long long, signed short *);
7590 vector unsigned short vec_xl_be (signed long long, unsigned short *);
7591 vector double vec_xl_be (signed long long, double *);
7592 vector float vec_xl_be (signed long long, float *);
7593 * config/rs6000/altivec.h (vec_xl_be): Add #define.
7594 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
7595 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
7597 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
7598 (altivec_expand_builtin): Add switch statement to call
7599 altivec_expand_xl_be for each builtin.
7600 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
7601 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
7602 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
7603 __builtin_vsx_le_be_v16qi.
7604 * doc/extend.texi: Update the built-in documentation file for the
7605 new built-in functions.
7607 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
7610 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
7612 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
7614 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7616 * config.gcc: Add z14.
7617 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
7618 CPU model numbers for z13s and z14.
7619 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
7621 * config/s390/s390-opts.h (enum processor_type): Rename
7622 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
7623 * config/s390/s390.c (processor_table): Add field for CPU name to
7624 be passed to Binutils.
7625 (s390_asm_output_machine_for_arch): Use the new field in
7626 processor_table for Binutils.
7627 (s390_expand_builtin): Replace arch12 with z14.
7628 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
7629 (s390_get_sched_attrmask): Likewise.
7630 (s390_get_unit_mask): Likewise.
7631 * config/s390/s390.opt: Add z14 to processor_type enum.
7633 2017-07-31 Martin Jambor <mjambor@suse.cz>
7636 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
7637 regardless of optimization level.
7639 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
7640 Martin Liska <mliska@suse.cz>
7642 * predict.def: Remove old comment and adjust probability.
7643 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
7646 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
7649 * config/i386/i386.c (ix86_function_naked): New function.
7650 (ix86_can_use_return_insn_p): Return false for naked functions.
7651 (ix86_expand_prologue): Skip prologue for naked functions.
7652 (ix86_expand_epilogue): Skip epilogue for naked functions
7653 and emit trap instruction.
7654 (ix86_warn_func_return): New function.
7655 (ix86_attribute_table): Add "naked" attribute specification.
7656 (TARGET_WARN_FUNC_RETURN): Define.
7657 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
7659 2017-07-31 Martin Liska <mliska@suse.cz>
7661 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
7662 (dump_gimple_bb_header): Always dump BB info.
7663 (pp_cfg_jump): Do not append info about BB when dumping a jump.
7665 2017-07-31 Martin Liska <mliska@suse.cz>
7668 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
7669 also with current_function_decl non-null equality.
7671 2017-07-31 Jakub Jelinek <jakub@redhat.com>
7674 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
7675 change type to the element type, instead add eltype variable and
7676 use it where we are interested in the element type.
7678 PR tree-optimization/81603
7679 * ipa-polymorphic-call.c
7680 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
7681 offset arithmetic in offset_int, bail out if the resulting bit offset
7682 doesn't fit into shwi.
7684 2017-07-31 Martin Liska <mliska@suse.cz>
7686 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
7687 (gimplify_save_expr): Fix comment.
7689 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
7692 * config/i386/i386.c (ix86_function_arg): Update arguments for
7694 (ix86_compute_frame_layout): Set the initial stack offset to
7695 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
7696 INCOMING_FRAME_SP_OFFSET.
7697 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
7698 stack before exception handler returns.
7699 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
7700 the 'ERROR_CODE' for exception handler.
7702 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
7704 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
7705 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
7706 (ASM_OUTPUT_REG_POP): Ditto.
7707 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
7708 instead of asm_fprintf to output pure string.
7710 2017-07-29 Jakub Jelinek <jakub@redhat.com>
7712 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
7713 to imported_module_or_decl hook.
7714 (debug_nothing_tree_tree_tree_bool): Remove.
7715 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
7716 * debug.c (do_nothing_debug_hooks): Use
7717 debug_nothing_tree_tree_tree_bool_bool instead of
7718 debug_nothing_tree_tree_tree_bool.
7719 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
7720 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
7721 * sdbout.c (sdb_debug_hooks): Likewise.
7722 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
7723 (gen_namespace_die): Add DW_AT_export_symbols attribute if
7725 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
7726 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
7727 attribute, don't add anything.
7729 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7731 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
7732 (fold_build2_stat_loc): Likewise.
7733 (fold_build3_stat_loc): Likewise.
7734 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
7735 (fold_build1_loc): Remove macro.
7736 (fold_build2_loc): Likewise.
7737 (fold_build3_loc): Likewise.
7739 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7741 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
7742 (gimple_build_debug_bind_source_stat): Likewise.
7743 * gimple.h (gimple_build_debug_bind): Remove macro.
7744 (gimple_build_debug_bind_source): Likewise.
7746 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7748 * bitmap.c (bitmap_alloc): Adjust.
7749 (bitmap_gc_alloc): Likewise.
7750 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
7752 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7754 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
7755 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
7756 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
7757 (bitmap_gc_alloc_stat): Likewise.
7758 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
7760 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7762 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
7763 * rtl.h (shallow_copy_rtx): Remove macro.
7765 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7767 * emit-rtl.c (gen_raw_REG): Adjust.
7768 * gengenrtl.c (gendef): Likewise.
7769 * rtl.c (rtx_alloc_stat): Remove _stat from name.
7770 * rtl.h (rtx_alloc): Remove macro.
7772 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7774 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
7775 (build_tree_list_stat): Likewise.
7776 * tree.h (build_tree_list): Remove macro.
7777 (build_tree_list_vec): Likewise.
7779 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7781 * tree.c (make_vector_stat): Remove _stat from name.
7782 (build_vector_stat): Likewise.
7783 * tree.h (make_vector_stat): Remove macro.
7784 (build_vector_stat): Likewise.
7786 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7788 * tree.h (build_var_debug_value): Remove prototype.
7790 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7792 * tree.c (tree_cons_stat): Remove _stat from name.
7793 * tree.h (tree_cons): Remove macro.
7795 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7797 * tree.c (build_vl_exp_stat): Remove _stat from name.
7798 * tree.h (build_vl_exp): Remove macro.
7800 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7802 * tree.c (build_decl_stat): Remove _stat from name.
7803 * tree.h (build_decl): Remove macro.
7805 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7807 * gimple.c (gimple_build_with_ops_stat): Adjust.
7808 (gimple_alloc_stat): Remove _stat from name.
7809 * gimple.h (gimple_alloc): Remove macro.
7811 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7813 * tree.c (make_tree_vec_stat): Remove _stat from name.
7814 (grow_tree_vec_stat): Likewise.
7815 * tree.h (make_tree_vec_stat): Adjust prototype.
7816 (grow_tree_vec_stat): Likewise.
7817 (make_tree_vec): Remove macro.
7818 (grow_tree_vec): Likewise.
7820 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7822 * fold-const.c (fold_build1_stat_loc): Adjust.
7823 (fold_build2_stat_loc): Likewise.
7824 (fold_build3_stat_loc): Likewise.
7825 * tree.c (build0_stat): Remove _stat from name.
7826 (build1_stat): Likewise.
7827 (build2_stat): Likewise.
7828 (build3_stat): Likewise.
7829 (build4_stat): Likewise.
7830 (build5_stat): Likewise.
7831 * tree.h (build1_loc): Remove macro, and rename _stat function
7833 (build2_loc): Likewise.
7834 (build3_loc): Likewise.
7835 (build4_loc): Likewise.
7836 (build5_loc): Likewise.
7838 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7840 * tree.c (make_int_cst_stat): Remove _stat from name.
7841 * tree.h (make_int_cst_stat): Adjust prototype.
7842 (make_int_cst): Remove macro.
7844 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7846 * tree.c (make_tre_binfo_stat): Remove _stat from name.
7847 * tree.h (make_tree_binfo_stat): Adjust prototype.
7848 (make_tree_binfo): Remove.
7850 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7852 * tree.c (copy_node_stat): Rename to copy_node.
7853 (build_distinct_type_copy): Adjust.
7854 * tree.h (copy_node_stat): Adjust prototype.
7855 (copy_node): Remove macro.
7857 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7859 * tree.c (make_node_stat): rename to make_node.
7860 (build_tree_list_stat): Adjust.
7861 (build0_stat): Likewise.
7862 (build2_stat): Likewise.
7863 (build3_stat): Likewise.
7864 (build4_stat): Likewise.
7865 (build5_stat): Likewise.
7866 (build_decl_stat): Likewise.
7867 * tree.h (make_node_stat): Adjust prototype.
7868 (make_node): remove macro.
7870 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
7872 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
7873 (PPC_FEATURE2_SCV): Likewise.
7874 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
7876 2017-07-28 Tamar Christina <tamar.christina@arm.com>
7878 * config/aarch64/aarch64.c
7879 (aarch64_internal_mov_immediate): Add new special pattern.
7880 * config/aarch64/aarch64.md (*movdi_aarch64):
7881 Add reg/32bit const mov case.
7883 2017-07-28 Tamar Christina <tamar.christina@arm.com>
7884 Richard Sandiford <richard.sandiford@linaro.org>
7886 * config/aarch64/aarch64.md (mov<mode>): Generalize.
7887 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
7888 Add integer and movi cases.
7889 (movi-split-hf-df-sf split, fp16): New.
7890 (enabled): Added TARGET_FP_F16INST.
7891 * config/aarch64/iterators.md (GPF_HF): New.
7892 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
7894 2017-07-28 Tamar Christina <tamar.christina@arm.com>
7896 * config/aarch64/aarch64.c
7897 (aarch64_simd_container_mode): Add prototype.
7898 (aarch64_expand_mov_immediate): Add HI support.
7899 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
7900 (aarch64_can_const_movi_rtx_p): New.
7901 (aarch64_preferred_reload_class):
7902 Remove restrictions of using FP registers for certain SIMD operations.
7903 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
7904 (aarch64_valid_floating_const): Add integer move validation.
7905 (aarch64_simd_imm_scalar_p): Remove.
7906 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
7907 (aarch64_legitimate_constant_p): Expand list of supported cases.
7908 * config/aarch64/aarch64-protos.h
7909 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
7910 (aarch64_reinterpret_float_as_int): New.
7911 (aarch64_simd_imm_scalar_p): Remove.
7912 * config/aarch64/constraints.md (Uvi): New.
7913 (Dd): Split into Ds and new Dd.
7914 * config/aarch64/aarch64.md (*movsi_aarch64):
7916 (*movdi_aarch64): Add SIMD mov case.
7918 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7920 * tree-predcom.c: (struct chain): Handle store-store chain in which
7921 stores for elimination only store loop invariant values.
7922 (execute_pred_commoning_chain): Ditto.
7923 (prepare_initializers_chain_store_elim): Ditto.
7924 (prepare_finalizers): Ditto.
7925 (is_inv_store_elimination_chain): New function.
7926 (initialize_root_vars_store_elim_1): New function.
7928 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7930 * tree-predcom.c: Revise general description of the pass.
7931 (enum chain_type): New enum type for store elimination.
7932 (struct chain): New field supporting store elimination.
7933 (struct component): Ditto.
7934 (dump_chain): Dump store-stores chain.
7935 (release_chain): Release resources.
7936 (split_data_refs_to_components): Compute and create component
7937 contains only stores for elimination.
7938 (get_chain_last_ref_at): New function.
7939 (make_invariant_chain): Initialization.
7940 (make_rooted_chain): Specify chain type in parameter and record it.
7941 (add_looparound_copies): Skip for store-stores chain.
7942 (determine_roots_comp): Compute type of chain and pass it to
7944 (initialize_root_vars_store_elim_2): New function.
7945 (finalize_eliminated_stores): New function.
7946 (remove_stmt): Handle store for elimination.
7947 (execute_pred_commoning_chain): Execute predictive commoning on
7949 (determine_unroll_factor): Skip unroll for store-stores chain.
7950 (prepare_initializers_chain_store_elim): New function.
7951 (prepare_initializers_chain): Hanlde store-store chain.
7952 (prepare_finalizers_chain, prepare_finalizers): New function.
7953 (tree_predictive_commoning_loop): Return integer value indicating
7954 if loop is unrolled or lcssa form is corrupted.
7955 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
7957 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7959 * tree-predcom.c (initialize_root): Delete.
7960 (execute_pred_commoning_chain): Initialize root vars and replace
7961 reference of non-combined chain directly, rather than call above
7964 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7966 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
7967 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
7969 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7971 * tree-predcom.c (struct chain): New field init_seq.
7972 (release_chain): Release init_seq.
7973 (prepare_initializers_chain): Record intialization stmts in above
7975 (insert_init_seqs): New function.
7976 (tree_predictive_commoning_loop): Call insert_init_seqs.
7978 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7980 * tree-predcom.c (determine_roots_comp): Skip trivial components.
7982 2017-07-28 Richard Biener <rguenther@suse.de>
7984 * match.pd: Remove superfluous :c.
7985 * genmatch.c (simplify::id): Add member.
7986 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
7988 (current_id): New global.
7989 (dt_node::parent): Move from ...
7990 (dt_operand::parent): ... here. Add for_id member.
7991 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
7992 (decision_tree::find_node): Relax order requirement when
7993 merging DT_TRUE nodes to ones inbetween the current simplify
7994 and the one we try to merge with. Add diagnostic whenever
7995 we need to enforce pattern order by not merging.
7996 (decision_tree::insert): Set current_id.
7997 (decision_tree::print_node): Dump parent node and for_id.
7998 (parser::last_id): Add member.
7999 (parser::push_simplify): Assign unique id.
8000 (parser::parser): Initialize last_id.
8002 2017-07-28 Martin Liska <mliska@suse.cz>
8005 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
8006 gimple_build_debug_bind.
8008 2017-07-28 Richard Biener <rguenther@suse.de>
8010 PR tree-optimization/81502
8011 * match.pd: Add pattern combining BIT_INSERT_EXPR with
8013 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
8015 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
8016 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
8017 for BIT_FIELD_REF args.
8018 * fold-const.c (make_bit_field_ref): Likewise.
8019 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
8021 2017-07-28 Jakub Jelinek <jakub@redhat.com>
8024 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
8025 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
8026 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
8027 Or it into SANITIZER_UNDEFINED.
8028 * ubsan.c: Include gimple-fold.h and varasm.h.
8029 (ubsan_expand_ptr_ifn): New function.
8030 (instrument_pointer_overflow): New function.
8031 (maybe_instrument_pointer_overflow): New function.
8032 (instrument_object_size): Formatting fix.
8033 (pass_ubsan::execute): Call instrument_pointer_overflow
8034 and maybe_instrument_pointer_overflow.
8035 * internal-fn.c (expand_UBSAN_PTR): New function.
8036 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
8037 * sanitizer.def (__ubsan_handle_pointer_overflow,
8038 __ubsan_handle_pointer_overflow_abort): New builtins.
8039 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
8040 * internal-fn.def (UBSAN_PTR): New internal function.
8041 * opts.c (sanitizer_opts): Add pointer-overflow.
8042 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
8043 * fold-const.c (build_range_check): Compute pointer range check in
8044 integral type if pointer arithmetics would be needed. Formatting
8047 2017-07-28 Martin Liska <mliska@suse.cz>
8050 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
8051 parameters that are of a variable-length.
8053 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
8055 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
8057 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
8058 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
8059 (CRT_CALL_STATIC_FUNCTION): Likewise.
8060 (ASM_DEFAULT_SPEC): New define.
8061 (ASM_SPEC32): Likewise.
8062 (ASM_SPEC64): Likewise.
8063 (ASM_SPEC_COMMON): Likewise.
8064 (ASM_SPEC): Likewise.
8065 (INVALID_64BIT): Likewise.
8066 (LINK_OS_DEFAULT_SPEC): Likewise.
8067 (LINK_OS_SPEC32): Likewise.
8068 (LINK_OS_SPEC64): Likewise.
8069 (POWERPC_LINUX): Likewise.
8070 (PTRDIFF_TYPE): Likewise.
8071 (RESTORE_FP_PREFIX): Likewise.
8072 (RESTORE_FP_SUFFIX): Likewise.
8073 (SAVE_FP_PREFIX): Likewise.
8074 (SAVE_FP_SUFFIX): Likewise.
8075 (SIZE_TYPE): Likewise.
8076 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
8077 (TARGET_64BIT): Likewise.
8078 (TARGET_64BIT): Likewise.
8079 (TARGET_AIX): Likewise.
8080 (WCHAR_TYPE_SIZE): Likewise.
8081 (WCHAR_TYPE): Undefine.
8082 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
8083 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
8084 (CPP_OS_RTEMS_SPEC): Delete.
8085 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
8086 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
8088 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
8090 2017-07-28 Jakub Jelinek <jakub@redhat.com>
8092 PR tree-optimization/81578
8093 * tree-parloops.c (build_new_reduction): Bail out if
8094 reduction_code isn't one of the standard OpenMP reductions.
8095 Move the details printing after that decision.
8097 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
8099 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
8100 related to reload_in_progress.
8101 (splat_input_operand): Likewise.
8102 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
8104 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
8106 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
8107 (TARGET_INSTANTIATE_DECLS): Likewise.
8108 (legitimate_indexed_address_p): Delete reload_in_progress code.
8109 (rs6000_debug_legitimate_address_p): Likewise.
8110 (rs6000_eliminate_indexed_memrefs): Likewise.
8111 (rs6000_emit_le_vsx_store): Likewise.
8112 (rs6000_emit_move_si_sf_subreg): Likewise.
8113 (rs6000_emit_move): Likewise.
8114 (register_to_reg_type): Likewise.
8115 (rs6000_pre_atomic_barrier): Likewise.
8116 (rs6000_machopic_legitimize_pic_address): Likewise.
8117 (rs6000_allocate_stack_temp): Likewise.
8118 (rs6000_address_for_fpconvert): Likewise.
8119 (rs6000_address_for_altivec): Likewise.
8120 (rs6000_secondary_memory_needed_rtx): Delete function.
8121 (rs6000_check_sdmode): Likewise.
8122 (rs6000_alloc_sdmode_stack_slot): Likewise.
8123 (rs6000_instantiate_decls): Likewise.
8124 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
8125 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
8126 Delete reload_in_progress.
8127 (*vec_reload_and_plus_<mptrsize>): Likewise.
8128 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
8129 (vsx_div_v2di): Likewise.
8130 (vsx_udiv_v2di): Likewise.
8132 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
8134 * config/rs6000/rs6000.opt (mlra): Replace with stub.
8135 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
8136 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
8137 (rs6000_debug_reg_global): Delete print of LRA status.
8138 (rs6000_option_override_internal): Delete dead LRA related code.
8139 (rs6000_lra_p): Delete function.
8140 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
8142 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
8144 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
8145 * config/riscv/rtems.h: New file.
8147 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8148 Sudakshina Das <sudi.das@arm.com>
8150 * config/aarch64/aarch64.md
8151 (define_split for and<mode>3nr_compare): Move
8152 non aarch64_logical_operand to a register.
8153 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
8154 register immediate operand to a register.
8155 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
8157 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
8160 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
8162 2017-07-27 Richard Biener <rguenther@suse.de>
8164 PR tree-optimization/81573
8165 PR tree-optimization/81494
8166 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
8167 multi defuse cycle case.
8169 2017-07-27 Richard Biener <rguenther@suse.de>
8171 PR tree-optimization/81571
8172 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
8175 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
8177 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
8178 earlier and only if MASK_FPU is set. Adjust formatting.
8180 2017-07-27 Martin Liska <mliska@suse.cz>
8182 * opt-functions.awk: Add validation of value of Init.
8183 * optc-gen.awk: Pass new argument.
8185 2017-07-27 Martin Liska <mliska@suse.cz>
8187 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
8188 Fix wrong condition.
8190 2017-07-27 Martin Liska <mliska@suse.cz>
8192 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
8193 BBs and edges seen by autoFDO.
8195 2017-07-27 Richard Biener <rguenther@suse.de>
8197 PR tree-optimization/81502
8198 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
8199 with incompatible but same sized type.
8200 (execute_update_addresses_taken): Likewise.
8202 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
8204 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
8205 flag_tree_loop_vectorize rather than flag_tree_vectorize.
8207 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8210 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
8211 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
8212 Change s_operand to memory_operand.
8214 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
8216 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
8217 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
8218 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
8219 Emit instructions rather than returning an expression. Handle TFmode
8220 and KFmode by casting to TImode.
8221 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
8222 (rs6000_emit_le_vsx_store): Likewise.
8223 * config/rs6000/vsx.md (VSX_TI): New iterator.
8224 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
8225 (*vsx_le_undo_permute_<mode>): Likewise.
8226 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
8227 emit the split sequence.
8228 (*vsx_le_perm_store_<mode>): Likewise.
8230 2017-07-27 Jakub Jelinek <jakub@redhat.com>
8232 PR tree-optimization/81555
8233 PR tree-optimization/81556
8234 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
8235 if true, force CHANGED for the recursive invocation.
8236 (reassociate_bb): Remember original length of ops array, pass
8237 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
8239 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
8240 attributes for noipa attribute. For naked attribute use
8241 lookup_attribute first before lookup_attribute_spec.
8242 * final.c (rest_of_handle_final): Disable IPA RA for functions with
8244 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
8245 for functions with noipa attribute.
8246 (cgraph_externally_visible_p): Return true for functions with noipa
8248 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
8249 for functions with noipa attribute.
8250 * doc/extend.texi: Document noipa function attribute.
8251 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
8252 also for functions with noipa attribute.
8253 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
8255 2017-07-26 Andrew Pinski <apinski@cavium.com>
8257 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
8258 vec_unalign_load_cost and vec_unalign_store_cost.
8260 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
8262 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
8263 -mvsx-small-integer option.
8264 (ISA_3_0_MASKS_IEEE): Likewise.
8265 (OTHER_VSX_VECTOR_MASKS): Likewise.
8266 (POWERPC_MASKS): Likewise.
8267 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
8268 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
8269 code, only testing for DImode being allowed in non-VSX floating
8271 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
8272 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
8274 (rs6000_option_override_internal): Delete -mvsx-small-integer.
8275 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
8276 TARGET_P8_VECTOR test.
8277 (rs6000_secondary_reload_simple_move): Likewise.
8278 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
8279 since TARGET_P9_VECTOR was already tested.
8280 (rs6000_opt_masks): Remove -mvsx-small-integer.
8281 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
8282 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
8284 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
8285 test for TARGET_VEXTRACTUB was used, and that uses
8287 (p9 extract splitter): Likewise.
8288 (vsx_extract_<mode>_di_p9): Likewise.
8289 (vsx_extract_<mode>_store_p9): Likewise.
8290 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
8291 for TARGET_P9_VECTOR was used. Delete code that is now dead with
8292 the elimination of TARGET_VSX_SMALL_INTEGER.
8293 (vsx_extract_<mode>_p8): Likewise.
8294 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
8295 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
8296 (vsx_set_<mode>_p9): Likewise.
8297 (vsx_set_v4sf_p9): Likewise.
8298 (vsx_set_v4sf_p9_zero): Likewise.
8299 (vsx_insert_extract_v4sf_p9): Likewise.
8300 (vsx_insert_extract_v4sf_p9_2): Likewise.
8301 * config/rs6000/rs6000.md (sign extend splitter): Change
8302 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
8303 (floatsi<mode>2_lfiwax_mem): Likewise.
8304 (floatunssi<mode>2_lfiwzx_mem): Likewise.
8305 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
8306 since a test for TARGET_P9_VECTOR was used.
8307 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
8308 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
8309 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
8310 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
8311 TARGET_P8_VECTOR test.
8312 (fix_trunc<mode>si2_stfiwx): Likewise.
8313 (fix_trunc<mode>si2_internal): Likewise.
8314 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
8315 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
8317 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
8318 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
8319 TARGET_P8_VECTOR test.
8320 (fixuns_trunc<mode>si2_stfiwx): Likewise.
8321 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
8322 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
8324 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
8325 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
8326 since a test for TARGET_P9_VECTOR was used.
8327 (splitter for loading small constants): Likewise.
8329 2017-07-26 Andrew Pinski <apinski@cavium.com>
8331 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
8334 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
8337 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
8338 (fp_valid_at): Likewise.
8340 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
8342 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
8343 (qdf24xx_addrcost_table): Likewise.
8344 (cortexa57_tunings): Update to use generic_branch_cost.
8345 (cortexa72_tunings): Likewise.
8346 (cortexa73_tunings): Likewise.
8347 (qdf24xx_tunings): Likewise.
8349 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
8351 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
8352 (thunderx2t99_branch_cost): Likewise.
8353 (cortexa35_tunings): Update to use generic_branch_cost.
8354 (cortexa53_tunings): Likewise.
8355 (cortexa57_tunings): Likewise.
8356 (cortexa72_tunings): Likewise.
8357 (cortexa73_tunings): Likewise.
8358 (thunderx2t99_tunings): Likewise.
8360 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
8362 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
8363 (sparc_option_override): Honour MASK_FSMULD.
8364 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
8365 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
8366 * config/sparc/sparc.opt (mfsmuld): New option.
8367 * doc/invoke.texi (mfsmuld): Document option.
8369 2017-07-26 Marek Polacek <polacek@redhat.com>
8372 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
8374 2017-07-26 Richard Biener <rguenther@suse.de>
8376 * gimple-match-head.c (do_valueize): Return OP if valueize
8378 (get_def): New helper to get at the def stmt of a SSA name
8380 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
8381 do_valueize to get at the def stmt.
8382 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
8384 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
8387 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
8389 2017-07-26 Martin Liska <mliska@suse.cz>
8392 * function.c (expand_function_start): Make expansion of
8393 nonlocal_goto_save_area after parm_birth_insn.
8395 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
8397 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
8398 from all CPU target flags enable members.
8400 2017-07-26 Richard Biener <rguenther@suse.de>
8402 * genmatch.c (dt_simplify::gen): Make iterator vars const.
8403 (decision_tree::gen): Make 'type' const.
8404 (write_predicate): Likewise.
8406 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
8408 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
8409 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
8410 (rs6000_option_override_internal): Likewise.
8411 (rs6000_expand_vector_set): Likewise.
8412 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
8413 (TARGET_UPPER_REGS_SF): Likewise.
8414 (TARGET_UPPER_REGS_DI): Likewise.
8415 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
8416 (TARGET_DIRECT_MOVE_64BIT): Likewise.
8417 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
8418 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
8419 (Splitters for DI constants in Altivec registers): Likewise.
8420 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
8421 (vsx_set_v4sf_p9): Likewise.
8422 (vsx_set_v4sf_p9_zero): Likewise.
8423 (vsx_insert_extract_v4sf_p9): Likewise.
8424 (vsx_insert_extract_v4sf_p9_2): Likewise.
8426 2017-07-25 Carl Love <cel@us.ibm.com>
8428 * doc/extend.texi: Update the built-in documentation file for the
8429 existing built-in functions
8430 vector signed char vec_cnttz (vector signed char);
8431 vector unsigned char vec_cnttz (vector unsigned char);
8432 vector signed short vec_cnttz (vector signed short);
8433 vector unsigned short vec_cnttz (vector unsigned short);
8434 vector signed int vec_cnttz (vector signed int);
8435 vector unsigned int vec_cnttz (vector unsigned int);
8436 vector signed long long vec_cnttz (vector signed long long);
8437 vector unsigned long long vec_cnttz (vector unsigned long long);
8439 2017-07-25 Andrew Pinski <apinski@cavium.com>
8441 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
8442 accesses where the use is for the first operand of a BIT_INSERT.
8444 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
8447 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
8448 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
8450 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
8452 * config/i386/gstabs.h: Delete.
8453 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
8455 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
8457 * config/i386/i386.c (ix86_decompose_address): Do not check for
8458 register RTX when looking at index_reg or base_reg.
8459 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
8461 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
8463 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
8464 to update EH info here.
8466 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
8468 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
8470 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
8472 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
8474 2017-07-25 Torsten Duwe <duwe@suse.de>
8476 * common.opt: Introduce -fpatchable-function-entry
8477 command line option, and its variables function_entry_patch_area_size
8478 and function_entry_patch_area_start.
8479 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
8480 including a two-value parser.
8481 * target.def (print_patchable_function_entry): New target hook.
8482 * targhooks.h (default_print_patchable_function_entry): New function.
8483 * targhooks.c (default_print_patchable_function_entry): Likewise.
8484 * toplev.c (process_options): Switch off IPA-RA if
8485 patchable function entries are being generated.
8486 * varasm.c (assemble_start_function): Look at the
8487 patchable-function-entry command line switch and current
8488 function attributes and maybe generate NOP instructions by
8489 calling the print_patchable_function_entry hook.
8490 * doc/extend.texi: Document patchable_function_entry attribute.
8491 * doc/invoke.texi: Document -fpatchable_function_entry
8492 command line option.
8493 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
8495 * doc/tm.texi: Re-generate.
8497 2017-07-25 Jakub Jelinek <jakub@redhat.com>
8500 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
8501 TARGET_AVX512DQ rather than TARGET_AVX512BW.
8503 2017-07-25 Tamar Christina <tamar.christina@arm.com>
8505 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
8507 2017-07-25 Richard Biener <rguenther@suse.de>
8509 PR tree-optimization/81455
8510 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
8511 not walk in cycles when looking for guards.
8513 2017-07-25 Richard Biener <rguenther@suse.de>
8515 PR tree-optimization/81529
8516 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
8517 when optimizing backedge uses.
8519 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
8521 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
8523 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
8524 to dl_section_ref. On AIX, append an expression to subtract
8525 the size of the section length to dl_section_ref.
8527 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
8529 * configure.ac: If any of the config.* scripts fail, exit 1.
8530 * configure: Regenerate.
8532 2017-07-25 Richard Biener <rguenther@suse.de>
8535 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
8536 of immediate uses, be more verbose on errors.
8538 2017-07-25 Richard Biener <rguenther@suse.de>
8540 PR tree-optimization/81510
8541 * tree-vect-loop.c (vect_is_simple_reduction): When the
8542 reduction stmt is not inside the loop bail out.
8544 2017-07-25 Richard Biener <rguenther@suse.de>
8546 PR tree-optimization/81303
8547 * tree-vect-loop-manip.c (vect_loop_versioning): Build
8548 profitability check against LOOP_VINFO_NITERSM1.
8550 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
8552 * domwalk.c (cmp_bb_postorder): Simplify.
8553 (sort_bbs_postorder): New function. Use it...
8554 (dom_walker::walk): ...here to optimize common cases.
8556 2017-07-25 Martin Liska <mliska@suse.cz>
8559 * ipa-visibility.c (function_and_variable_visibility): Make the
8560 redirection just on target that supports aliasing.
8561 Fix GNU coding style.
8563 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
8566 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
8568 * config/arm/rtems.h: Likewise.
8569 * config/bfin/rtems.h: Likewise.
8570 * config/i386/rtemself.h: Likewise.
8571 * config/lm32/rtems.h: Likewise.
8572 * config/m32c/rtems.h: Likewise.
8573 * config/m68k/rtemself.h: Likewise.
8574 * config/microblaze/rtems.h: Likewise.
8575 * config/mips/rtems.h: Likewise.
8576 * config/moxie/rtems.h: Likewise.
8577 * config/nios2/rtems.h: Likewise.
8578 * config/powerpcspe/rtems.h: Likewise.
8579 * config/rs6000/rtems.h: Likewise.
8580 * config/rtems.h: Likewise.
8581 * config/sh/rtems.h: Likewise.
8582 * config/sh/rtemself.h: Likewise.
8583 * config/sparc/rtemself.h: Likewise.
8585 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
8588 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
8589 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
8590 * tree-ssa-structalias.c (alias_get_name): Same.
8592 2017-07-25 Bin Cheng <bin.cheng@arm.com>
8595 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
8596 instructions if no du chain is found.
8598 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
8600 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
8602 2017-07-25 Richard Biener <rguenther@suse.de>
8605 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
8608 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
8610 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
8612 (ISA_2_7_MASKS_SERVER): Likewise.
8613 (ISA_3_0_MASKS_IEEE): Likewise.
8614 (OTHER_P8_VECTOR_MASKS): Likewise.
8615 (OTHER_VSX_VECTOR_MASKS): Likewise.
8616 (POWERPC_MASKS): Likewise.
8617 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
8618 duplicate list of options.
8619 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
8620 explicit -mupper-regs options.
8621 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
8622 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
8623 alias for -mupper-regs-df.
8624 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
8625 (rs6000_init_hard_regno_mode_ok): Likewise.
8626 (rs6000_option_override_internal): Likewise.
8627 (rs6000_opt_masks): Likewise.
8628 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
8629 options in terms of whether -mvsx or -mpower8-vector was used.
8630 (TARGET_UPPER_REGS_DI): Likewise.
8631 (TARGET_UPPER_REGS_SF): Likewise.
8632 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
8633 -mupper-regs-* options.
8635 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
8637 * passes.c (emergency_dump_function): Print some empty lines and a
8638 header before the RTL dump.
8640 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
8642 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
8644 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
8647 * config/aarch64/aarch64.c (aarch64_classify_symbol):
8648 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
8650 2017-07-24 Carl Love <cel@us.ibm.com>
8652 * config/rs6000/rs6000-c.c: Add support for built-in functions
8653 vector float vec_extract_fp32_from_shorth (vector unsigned short);
8654 vector float vec_extract_fp32_from_shortl (vector unsigned short);
8655 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
8656 vec_extract_fp_from_shortl): Add defines for the two builtins.
8657 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
8658 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
8660 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
8661 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
8662 * doc/extend.texi: Update the built-in documentation file for the
8663 new built-in function.
8665 2017-07-24 Jakub Jelinek <jakub@redhat.com>
8668 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
8670 * doc/generic.texi: Likewise.
8671 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
8672 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
8674 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
8676 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
8677 (aarch64_mls_elt_merge<mode>): Likewise.
8679 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
8681 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
8682 having __cxa_atexit.
8684 2017-07-23 Michael Collison <michael.collison@arm.com>
8686 * config/arm/arm.c (arm_option_override): Deprecate
8687 use of -mstructure-size-boundary.
8688 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
8689 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
8691 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8694 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
8695 Reduce cost estimate for direct moves.
8697 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
8700 * config/i386/i386.c (ix86_option_override_internal): Disable
8701 BMI, BMI2 and TBM instructions for -m16.
8703 2017-07-21 Carl Love <cel@us.ibm.com>
8705 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8706 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
8707 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
8708 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
8709 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
8710 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
8711 VMULOSW): New enum "unspec" values.
8712 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
8713 altivec_vmulosw): New patterns.
8714 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
8715 VMULOSW): Add definitions.
8717 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
8719 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
8720 (qdf24xx): Likewise.
8721 * config/aarch64/aarch64-options-extensions.def (rdma); New.
8722 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
8723 (AARCH64_FL_V8_1): Renumber.
8724 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
8725 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
8726 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
8727 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
8728 rdma to feature modifiers list.
8730 2017-07-21 Yury Gribov <tetra2005@gmail.com>
8733 * ipa-visibility (function_and_variable_visibility): Convert
8734 recursive PLT call to direct call if appropriate.
8736 2017-07-21 Andrew Pinski <apinski@cavium.com>
8738 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
8739 operand 1 to see if the types precision matches.
8740 * fold-const.c (operand_equal_p): Likewise.
8742 2017-07-21 Richard Biener <rguenther@suse.de>
8744 PR tree-optimization/81303
8745 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
8746 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
8747 (vect_peeling_hash_get_lowest_cost): Adjust.
8748 (vect_enhance_data_refs_alignment): Likewise. Use
8749 vect_get_peeling_costs_all_drs to compute the penalty for no
8750 peeling to match up costs.
8752 2017-07-21 Richard Biener <rguenther@suse.de>
8754 PR tree-optimization/81500
8755 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
8756 we didn't identify a reduction path.
8758 2017-07-21 Tom de Vries <tom@codesourcery.com>
8759 Cesar Philippidis <cesar@codesourcery.com>
8761 PR gcov-profile/81442
8762 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
8765 2017-07-21 Tom de Vries <tom@codesourcery.com>
8768 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
8770 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
8771 nvptx_override_options_after_change.
8773 2017-07-21 Ulrich Drepper <drepper@redhat.com>
8775 * dwarf2out.c (output_file_names): Avoid double testing for
8778 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
8780 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
8782 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
8784 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
8786 (try_crossjump_to_edge): Do not punt on partitioned functions.
8788 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
8790 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
8791 Put all BBs reachable only via paths crossing cold region to cold
8793 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
8795 2016-07-21 Richard Biener <rguenther@suse.de>
8797 PR tree-optimization/81303
8798 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
8799 into account prologue and epilogue iterations when raising
8800 min_profitable_iters to sth at least covering one vector iteration.
8802 2017-07-21 Tamar Christina <tamar.christina@arm.com>
8804 * config/arm/arm.c (arm_test_cpu_arch_dat):
8807 2017-07-20 Nathan Sidwell <nathan@acm.org>
8809 Remove TYPE_METHODS.
8810 * tree.h (TYPE_METHODS): Delete.
8811 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
8812 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
8813 (dbxout_type_methods): Scan TYPE_FIELDS.
8814 (dbxout_type): Don't check TYPE_METHODS here.
8815 * function.c (use_register_for_decl): Always ignore register for
8816 class types when not optimizing.
8817 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
8818 * tree.c (free_lang_data_in_type): Stitch out member functions and
8819 templates from TYPE_FIELDS.
8820 (build_distinct_type_copy, verify_type_variant,
8821 verify_type): Member fns are on TYPE_FIELDS.
8822 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
8823 * tree-pretty-print.c (dump_generic_node): Likewise.
8825 2017-07-20 Jakub Jelinek <jakub@redhat.com>
8828 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
8829 V2TImode and V4TImode.
8830 (ix86_expand_vector_extract): Likewise.
8831 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
8832 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
8833 (ssescalarmode): Handle V4TImode and V2TImode.
8834 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
8835 (*vec_extractv2ti, *vec_extractv4ti): New insns.
8836 (VEXTRACTI128_MODE): New mode iterator.
8837 (splitter for *vec_extractv?ti first element): New.
8838 (VEC_INIT_MODE): New mode iterator.
8839 (vec_init<mode>): Consolidate 3 expanders into one using
8840 VEC_INIT_MODE mode iterator.
8842 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
8844 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
8845 non_spilled_static_chain_regno_p.
8847 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
8849 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
8851 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
8853 * bb-reorder.c (connect_traces): Allow copying of blocks within
8856 2017-07-20 Richard Biener <rguenther@suse.de>
8858 * gimple.h (gimple_phi_result): Add gphi * overload.
8859 (gimple_phi_result_ptr): Likewise.
8860 (gimple_phi_arg): Likewise. Adjust index assert to only
8861 allow actual argument accesses rather than all slots available
8863 (gimple_phi_arg_def): Add gphi * overload.
8864 * tree-phinodes.c (make_phi_node): Initialize only actual
8866 (resize_phi_node): Clear memory not covered by old node,
8867 do not initialize excess argument slots.
8868 (reserve_phi_args_for_new_edge): Initialize new argument slot
8871 2017-07-20 Bin Cheng <bin.cheng@arm.com>
8873 PR tree-optimization/81388
8875 2016-07-21 Bin Cheng <bin.cheng@arm.com>
8877 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
8878 by removing computation of may_be_zero.
8880 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
8881 Tom de Vries <tom@codesourcery.com>
8884 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
8885 when gimple level profile disagrees with what RTL expander did.
8887 2017-07-20 Richard Biener <rguenther@suse.de>
8889 PR tree-optimization/61171
8890 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
8891 (vect_analyze_stmt): Add slp instance parameter.
8892 (vectorizable_reduction): Likewise.
8893 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
8894 (vect_is_simple_reduction): Deal with chains not detected
8895 as SLP reduction chain, specifically not properly associated
8896 chains containing a mix of plus/minus.
8897 (get_reduction_op): Remove.
8898 (get_initial_defs_for_reduction): Simplify, pass in whether
8899 this is a reduction chain, pass in the SLP node for the PHIs.
8900 (vect_create_epilog_for_reduction): Get the SLP instance as
8902 (vectorizable_reduction): Get the SLP instance as arg.
8903 During analysis remember the SLP node with the PHIs in the
8904 instance. Simplify getting at the vectorized reduction PHIs.
8905 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
8906 through SLP instance.
8907 (vect_slp_analyze_operations): Likewise.
8908 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
8909 (vect_transform_stmt): Likewise.
8911 2017-07-20 Tom de Vries <tom@codesourcery.com>
8913 PR tree-optimization/81489
8914 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
8915 read of phi arg location to before loop that modifies phi.
8917 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
8919 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
8922 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
8925 * except.c (execute): Fix ordering issue.
8927 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
8929 PR rtl-optimization/81423
8930 * combine.c (make_compound_operation_int): Don't try to optimize
8931 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
8933 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
8935 PR rtl-optimization/81423
8936 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
8937 with a constant that is -1 in the truncated to mode.
8939 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
8941 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
8942 (determine_unlikely_bbs): ... here.
8943 * predict.h (propagate_unlikely_bbs_forward): Declare.
8944 * cfgexpand.c (pass_expand::execute): Use it.
8945 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
8947 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
8948 propagate_unlikely_bbs_forward.
8950 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
8953 * except.c (maybe_add_nop_after_section_switch): New function.
8956 2017-07-19 Tom de Vries <tom@codesourcery.com>
8958 * gimple.h (gimple_phi_set_arg): Make assert more strict.
8960 2017-07-19 Tom de Vries <tom@codesourcery.com>
8962 * gimple.h (gimple_phi_arg): Make assert more strict.
8964 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
8966 * config.gcc (powerpc*-*-*): Add mmintrin.h.
8967 * config/rs6000/mmintrin.h: New file.
8968 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
8970 2017-07-19 Jakub Jelinek <jakub@redhat.com>
8972 PR tree-optimization/81346
8973 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
8975 2017-07-19 Tom de Vries <tom@codesourcery.com>
8977 * config/nvptx/nvptx.md (VECIM): Add V2DI.
8979 2017-07-19 Tom de Vries <tom@codesourcery.com>
8981 * config/nvptx/nvptx-modes.def: Add V2DImode.
8982 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
8983 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
8984 (nvptx_output_mov_insn): Handle lack of mov.b128.
8985 (nvptx_print_operand): Handle 'H' and 'L' codes.
8986 (nvptx_vector_mode_supported): Allow V2DImode.
8987 (nvptx_preferred_simd_mode): New function.
8988 (nvptx_data_alignment): New function.
8989 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
8990 nvptx_preferred_simd_mode.
8991 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
8993 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
8995 2017-07-19 Tom de Vries <tom@codesourcery.com>
8997 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
8998 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
8999 (nvptx_vector_mode_supported): New function. Allow V2SImode.
9000 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
9001 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
9002 (mov<VECIM>_insn): New define_insn.
9003 (define_expand "mov<VECIM>): New define_expand.
9005 2017-07-19 Tom de Vries <tom@codesourcery.com>
9007 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
9009 2017-07-19 Jakub Jelinek <jakub@redhat.com>
9011 PR tree-optimization/81346
9012 * fold-const.h (fold_div_compare, range_check_type): Declare.
9013 * fold-const.c (range_check_type): New function.
9014 (build_range_check): Use range_check_type.
9015 (fold_div_compare): No longer static, rewritten into
9016 a match.pd helper function.
9017 (fold_comparison): Don't call fold_div_compare here.
9018 * match.pd (X / C1 op C2): New optimization using fold_div_compare
9021 2017-07-19 Nathan Sidwell <nathan@acm.org>
9023 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
9024 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
9025 * tree.c (find_decls_types_r, verify_type): Use
9026 TYPE_{MIN,MAX}_VALUE_RAW.
9027 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
9028 (hash_tree): Likewise.
9029 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
9031 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
9034 2017-07-18 Tom de Vries <tom@codesourcery.com>
9037 * omp-expand.c (expand_omp_for_static_chunk): Handle
9038 equal-argument loop exit phi.
9040 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
9043 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
9044 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
9045 operand 2 predicate.
9046 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
9047 operand 2 predicate.
9048 (ror,rol -> rorx splitters): Use const_int_operand as
9049 operand 2 predicate.
9051 2017-06-18 Richard Biener <rguenther@suse.de>
9053 PR tree-optimization/81410
9054 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
9055 the gap in the ! slp_perm SLP case after each group.
9057 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
9060 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
9063 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
9066 * predict.c (set_even_probabilities): Cleanup; do not affect
9067 probabilities that are already known.
9068 (combine_predictions_for_bb): Call even when count is set.
9070 2017-07-18 Nathan Sidwell <nathan@acm.org>
9072 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
9075 2017-07-18 Bin Cheng <bin.cheng@arm.com>
9078 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
9079 optimization for loop niter analysis.
9081 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
9084 * config/avr/avr.c (avr_optimize_casesi): Don't use
9085 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
9087 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
9089 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
9090 body_cost_vec from _vect_peel_extended_info.
9091 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
9092 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
9095 2017-07-18 Bin Cheng <bin.cheng@arm.com>
9097 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
9099 2017-07-18 Richard Biener <rguenther@suse.de>
9101 PR tree-optimization/80620
9102 PR tree-optimization/81403
9103 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
9104 info when re-using a VN table entry.
9106 2017-07-18 Richard Biener <rguenther@suse.de>
9108 PR tree-optimization/81418
9109 * tree-vect-loop.c (vectorizable_reduction): Properly compute
9110 vectype_in. Verify that with lane-reducing reduction operations
9111 we have a single def-use cycle.
9113 2017-07-17 Carl Love <cel@us.ibm.com>
9115 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
9117 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9118 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
9119 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
9120 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
9121 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
9122 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
9123 VMULOSW): New enum "unspec" values.
9124 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
9125 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
9126 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
9127 altivec_vmulosw): New patterns.
9128 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
9129 VMULOSW): Add definitions.
9130 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
9132 * config/alpha/alpha.c: Include predict.h.
9134 2017-07-17 Yury Gribov <tetra2005@gmail.com>
9136 * tree-vrp.c (compare_assert_loc): Fix comparison function
9137 to return predictable results.
9139 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
9141 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
9144 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
9145 * doc/invoke.texi (mexpand-adddi): Update text.
9147 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
9149 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
9150 that also clobbers the CC register. The old expand code is moved
9152 (*arc_clzsi2): ... here.
9153 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
9154 the CC register. The old expand code is moved to ...
9155 (arc_ctzsi2): ... here.
9157 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
9159 * config/arc/arc.opt (mindexed-loads): Use initial value
9160 TARGET_INDEXED_LOADS_DEFAULT.
9161 (mauto-modify-reg): Use initial value
9162 TARGET_AUTO_MODIFY_REG_DEFAULT.
9163 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
9164 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
9165 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
9166 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
9168 2017-07-17 Martin Liska <mliska@suse.cz>
9171 * opts.c (finish_options): Do not allow -fgnu-tm
9172 w/ -fsanitize={kernel-,}address. Say sorry.
9174 2017-07-17 Bin Cheng <bin.cheng@arm.com>
9177 * tree-loop-distribution.c (classify_partition): Only assert on
9178 numer of iterations.
9179 (merge_dep_scc_partitions): Delete prameter. Update function call.
9180 (distribute_loop): Remove code handling loop with unknown niters.
9181 (pass_loop_distribution::execute): Skip loop with unknown niters.
9183 2017-07-17 Bin Cheng <bin.cheng@arm.com>
9186 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
9187 function sort_partitions_by_post_order.
9189 2017-07-17 Bin Cheng <bin.cheng@arm.com>
9191 PR tree-optimization/81374
9192 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
9193 the max index of basic blocks, rather than number of basic blocks.
9195 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
9197 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
9199 (arc_legitimate_pic_operand_p): Likewise.
9200 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
9202 (arc_needs_pcl_p): Likewise.
9203 (arc_legitimate_pc_offset_p): Likewise.
9204 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
9205 function is also used in constrains.md.
9206 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
9207 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
9208 PLUS. Only return true/false in known cases, otherwise assert.
9209 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
9210 is already called in arc_legitimate_constant_p.
9211 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
9213 (LEGITIMATE_PIC_OPERAND_P): Use
9214 arc_raw_symbolic_reference_mentioned_p function.
9215 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
9220 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
9221 Andrew Burgess <andrew.burgess@embecosm.com>
9223 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
9224 (arc_return_address_register): New function.
9225 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
9226 (arc_handle_fndecl_attribute): Add naked attribute.
9227 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
9228 (TARGET_WARN_FUNC_RETURN): Likewise.
9229 (arc_allocate_stack_slots_for_args): New function.
9230 (arc_warn_func_return): Likewise.
9231 (machine_function): Change type fn_type.
9232 (arc_compute_function_type): Consider new naked function type,
9233 change function return type.
9234 (arc_must_save_register): Adapt to handle new
9235 arc_compute_function_type's return type.
9236 (arc_expand_prologue): Likewise.
9237 (arc_expand_epilogue): Likewise.
9238 (arc_return_address_regs): Delete.
9239 (arc_return_address_register): New function.
9240 (arc_epilogue_uses): Use above function.
9241 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
9242 (arc_function_type): Change encoding, add naked type.
9243 (ARC_INTERRUPT_P): Change to handle the new encoding.
9244 (ARC_FAST_INTERRUPT_P): Likewise.
9245 (ARC_NORMAL_P): Define.
9246 (ARC_NAKED_P): Likewise.
9247 (arc_compute_function_type): Delete prototype.
9248 * config/arc/arc.md (in_ret_delay_slot): Use
9249 arc_return_address_register function.
9250 (simple_return): Likewise.
9251 (p_return_i): Likewise.
9253 2017-07-17 Jakub Jelinek <jakub@redhat.com>
9255 PR tree-optimization/81428
9256 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
9257 can't be built for those types.
9259 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
9261 Remove stuff dead since r239246.
9263 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
9264 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
9265 (avr_inform_devices): Remove dead stuff.
9267 2017-07-17 Tamar Christina <tamar.christina@arm.com>
9269 * config/arm/arm_neon.h: Fix softp typo.
9271 2017-07-17 Jakub Jelinek <jakub@redhat.com>
9273 PR tree-optimization/81365
9274 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
9275 aggregate moves onto bb predecessor edges, make sure there are no
9276 loads that could alias the lhs in between the start of bb and the
9279 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
9282 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
9283 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
9284 [LSHIFTRT, outer_code = TRUNCATE]: Same.
9286 2017-07-17 Jakub Jelinek <jakub@redhat.com>
9288 PR tree-optimization/81396
9289 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
9290 (init_symbolic_number): Initialize it to 1.
9291 (perform_symbolic_merge): Add n_ops from both operands into the new
9293 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
9294 without base_addr as useless if they need more than one operation.
9295 (bswap_replace): Handle !bswap case for NULL base_addr.
9297 2017-07-17 Tom de Vries <tom@codesourcery.com>
9300 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
9303 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
9305 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
9306 conditional builtin define __FIX_LEON3FT_B2BST.
9308 2017-07-17 Daniel Cederman <cederman@gaisler.com>
9310 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
9311 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
9314 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
9316 PR rtl-optimization/81424
9317 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
9318 to remove potential trapping from operands if -fnon-call-exceptions.
9320 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9322 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
9323 profile_proability for scalling.
9324 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
9326 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9328 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
9330 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9332 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
9333 fixpoint arithmetics.
9335 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9337 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
9338 fixpoint arithmetics.
9340 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9342 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
9343 fixpoint arithmetics.
9345 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9347 * profile-count.h (profile_probability::from_reg_br_prob_note,
9348 profile_probability::to_reg_br_prob_note): New functions.
9349 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
9350 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
9351 * predict.c (probability_reliable_p): Update.
9352 (edge_probability_reliable_p): Update.
9353 (br_prob_note_reliable_p): Update.
9354 (invert_br_probabilities): Update.
9355 (add_reg_br_prob_note): New function.
9356 (combine_predictions_for_insn): Update.
9357 * asan.c (asan_clear_shadow): Update.
9358 * cfgbuild.c (compute_outgoing_frequencies): Update.
9359 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
9360 (update_br_prob_note): Update.
9361 (rtl_verify_edges): Update.
9362 (purge_dead_edges): Update.
9363 (fixup_reorder_chain): Update.
9364 * emit-rtl.c (try_split): Update.
9365 * ifcvt.c (cond_exec_process_insns): Update.
9366 (cond_exec_process_if_block): Update.
9367 (dead_or_predicable): Update.
9368 * internal-fn.c (expand_addsub_overflow): Update.
9369 (expand_neg_overflow): Update.
9370 (expand_mul_overflow): Update.
9371 * loop-doloop.c (doloop_modify): Update.
9372 * loop-unroll.c (compare_and_jump_seq): Update.
9373 * optabs.c (emit_cmp_and_jump_insn_1): Update.
9374 * predict.h: Update.
9375 * reorg.c (mostly_true_jump): Update.
9377 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
9378 * config/alpha/alpha.c (emit_unlikely_jump): Update.
9379 * config/arc/arc.c: (emit_unlikely_jump): Update.
9380 * config/arm/arm.c: (emit_unlikely_jump): Update.
9381 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
9382 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
9383 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
9384 (ix86_print_operand): Update.
9385 (ix86_split_fp_branch): Update.
9386 (predict_jump): Update.
9387 * config/ia64/ia64.c (ia64_print_operand): Update.
9388 * config/mmix/mmix.c (mmix_print_operand): Update.
9389 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
9390 (rs6000_expand_split_stack_prologue): Update.
9391 * config/rs6000/rs6000.c: Update.
9392 * config/s390/s390.c (s390_expand_vec_strlen): Update.
9393 (s390_expand_vec_movstr): Update.
9394 (s390_expand_cs_tdsi): Update.
9395 (s390_expand_split_stack_prologue): Update.
9396 * config/sh/sh.c (sh_print_operand): Update.
9397 (expand_cbranchsi4): Update.
9398 (expand_cbranchdi4): Update.
9399 * config/sparc/sparc.c (output_v9branch): Update.
9400 * config/spu/spu.c (get_branch_target): Update.
9401 (ea_load_store_inline): Update.
9402 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
9403 * config/tilepro/tilepro.c: Update.
9405 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
9407 * gimplify.c (mostly_copy_tree_r): Revert latest change.
9408 (gimplify_save_expr): Likewise.
9410 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
9412 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
9414 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
9416 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
9418 * timevar.def (TV_IPA_FNSUMMARY): Define.
9420 2017-07-16 Daniel Cederman <cederman@gaisler.com>
9422 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
9423 to back store errata sensitive sequence from being generated.
9424 (sqrtdf2_fix): Likewise.
9426 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
9428 * tree-ssa-threadupdate.c (compute_path_counts,
9429 update_joiner_offpath_counts): Use profile_probability.
9431 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
9434 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
9436 * config/arm/arm-c.c (arm_cpu_builtins): Define
9437 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
9439 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
9441 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9442 array entries to represent __ieee128 versions of the
9443 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
9444 scalar_extract_sig, and scalar_insert_exp built-in functions.
9445 (altivec_resolve_overloaded_builtin): Add special case handling
9446 for the __builtin_scalar_insert_exp function, as represented by
9447 the P9V_BUILTIN_VEC_VSIEDP constant.
9448 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
9449 exponent support for __ieee128 argument.
9450 (VSESQP): Add scalar extract signature support for __ieee128
9452 (VSTDCNQP): Add scalar test negative support for __ieee128
9454 (VSIEQP): Add scalar insert exponent support for __int128 argument
9455 with __ieee128 result.
9456 (VSIEQPF): Add scalar insert exponent support for __ieee128
9457 argument with __ieee128 result.
9458 (VSTDCQP): Add scalar test data class support for __ieee128
9460 (VSTDCNQP): Add overload support for scalar test negative with
9462 (VSTDCQP): Add overload support for scalar test data class
9464 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
9466 (UNSPEC_VSX_SIEXPQP): New constant.
9467 (xsxexpqp): New insn for VSX scalar extract exponent quad
9469 (xsxsigqp): New insn for VSX scalar extract significand quad
9471 (xsiexpqpf): New insn for VSX scalar insert exponent quad
9472 precision with floating point argument.
9473 (xststdcqp): New expand for VSX scalar test data class quad
9475 (xststdcnegqp): New expand for VSX scalar test negative quad
9477 (xststdcqp): New insn to match expansions for VSX scalar test data
9478 class quad precision and VSX scalar test negative quad precision.
9479 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
9480 special case operand checking to enforce that second operand of
9481 VSX scalar test data class with quad precision argument is a 7-bit
9483 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
9484 prototypes and descriptions of __ieee128 versions of
9485 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
9486 scalar_test_data_class, and scalar_test_neg built-in functions.
9488 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9490 PR tree-optimization/81162
9491 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
9492 replace a negate with an add.
9494 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
9496 * doc/invoke.texi (arm/-mcpu): Document +crypto.
9498 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
9500 * config/arm/arm-c.c (arm_cpu_builtins): Define
9501 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
9503 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
9505 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
9506 (armv8-r): Set ARM Cortex-R52 as default CPU.
9507 * config/arm/arm-tables.opt: Regenerate.
9508 * config/arm/arm-tune.md: Regenerate.
9509 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
9511 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
9512 extension for -mcpu=cortex-r52.
9514 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
9516 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
9517 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
9518 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
9519 (fp-armv8): Define it as FP_ARMv8 only.
9520 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
9521 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
9523 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
9524 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
9525 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
9526 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
9527 than TARGET_FPU_ARMV8.
9528 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
9529 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
9530 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
9532 * config/arm/neon.md (neon_vrint): Likewise.
9533 (neon_vcvt): Likewise.
9534 (neon_<fmaxmin_op><mode>): Likewise.
9535 (<fmaxmin><mode>3): Likewise.
9536 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
9537 * config/arm/predicates.md (arm_cond_move_operator): Check against
9538 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
9540 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
9542 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
9545 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9547 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
9548 loop in comment with memset.
9550 2017-07-14 Martin Liska <mliska@suse.cz>
9552 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
9553 * dwarf2out.c (is_java): Remove the function.
9554 (output_pubname): Remove usage of the function.
9555 (lower_bound_default): Remove usage of DW_LANG_Java.
9556 (gen_compile_unit_die): Likewise.
9557 * gcc.c: Remove compiler defaults for .java and .zip files.
9558 * gimple-expr.c (remove_suffix): Change as there's no longer
9559 extension than 4-letter one.
9560 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
9561 (gimplify_save_expr): Likewise.
9562 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
9563 as it's possible even for other languages than Java.
9564 * langhooks.h (struct lang_hooks): Remove Java from a comment.
9565 * lto-opts.c (lto_write_options): Remove reference to Java.
9566 * opts.c (strip_off_ending): Update file extension handling.
9567 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
9568 * tree-eh.c (lower_resx): Likewise.
9569 * tree.c (free_lang_data_in_type): Remove dead code.
9570 (find_decls_types_r): Likewise.
9571 (build_common_builtin_nodes): Remove Java from a comment.
9572 (verify_type): Remove dead code.
9573 * varasm.c (assemble_external): Remove Java from a comment.
9575 2017-07-14 Martin Liska <mliska@suse.cz>
9577 * opts.c (finish_options): Add quotes.
9578 (common_handle_option): Likewise.
9580 2017-07-14 Martin Liska <mliska@suse.cz>
9582 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
9583 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
9585 * dwarf2out.c (lower_bound_default): Do not handle
9587 (gen_compile_unit_die): Likewise.
9588 * gcc.c: Remove default extension binding for GNU Pascal.
9589 * stmt.c: Remove Pascal language from a comment.
9590 * xcoffout.c: Likewise.
9592 2017-07-13 David Malcolm <dmalcolm@redhat.com>
9595 * diagnostic-show-locus.c (fixit_cmp): New function.
9596 (layout::layout): Sort m_fixit_hints.
9597 (column_range::column_range): Assert that the values are valid.
9598 (struct char_span): New struct.
9599 (correction::overwrite): New method.
9600 (struct source_line): New struct.
9601 (line_corrections::add_hint): Add assertions. Reimplement memcpy
9602 calls in terms of classes source_line and char_span, and
9603 correction::overwrite.
9604 (selftest::test_overlapped_fixit_printing_2): New function.
9605 (selftest::diagnostic_show_locus_c_tests): Call it.
9607 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
9609 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
9610 early if there is no lhs.
9612 2017-07-13 Martin Liska <mliska@suse.cz>
9614 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
9615 (gen_reference_type_die): Likewise.
9616 * stor-layout.c: Remove Pascal-related comment.
9618 2017-07-13 Martin Liska <mliska@suse.cz>
9620 * opts.c (finish_options): Add quotes to error messages.
9621 (parse_sanitizer_options): Likewise.
9623 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
9625 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
9627 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
9629 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
9631 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
9633 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
9635 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
9637 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
9640 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
9641 provides the hardware capability bits, define the macro
9642 __BUILTIN_CPU_SUPPORTS__.
9643 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
9644 if GLIBC does not provide the hardware capability bits. Add a
9645 gcc_unreachable call if the built-in cpu function is neither
9646 __builtin_cpu_is nor __builtin_cpu_supports.
9647 (rs6000_get_function_versions_dispatcher): Change the warning
9648 that an old GLIBC is used which does not export the capability
9649 bits to be an error.
9650 * doc/extend.texi (target_clones attribute): Document the
9651 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
9652 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
9653 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
9654 the macros defined by GCC if the newer GLIBC is available.
9656 2017-07-12 Jeff Law <law@redhat.com>
9658 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
9659 remaining includes slightly.
9660 * config/riscv/riscv-builtins.c: Include profile-count.h.
9662 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
9665 * config/avr/avr.c (avr_set_current_function): In diagnostic
9666 messages: Quote keywords and (parts of) identifiers.
9667 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
9670 2017-07-12 Carl Love <cel@us.ibm.com>
9672 * config/rs6000/rs6000-c.c: Add support for built-in functions
9673 vector bool char vec_revb (vector bool char);
9674 vector bool short vec_revb (vector short char);
9675 vector bool int vec_revb (vector bool int);
9676 vector bool long long vec_revb (vector bool long long);
9677 * doc/extend.texi: Update the built-in documentation file for the
9678 new built-in functions.
9680 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9682 * config/s390/s390.md: Remove movcc splitter.
9684 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9686 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
9687 load/store on condition.
9689 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
9692 * config/avr/avr.c (avr_encode_section_info)
9693 [progmem && !TREE_READONLY]: Error if progmem object needs
9696 2017-07-11 Michael Collison <michael.collison@arm.com>
9698 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
9701 2017-07-11 Carl Love <cel@us.ibm.com>
9703 * config/rs6000/rs6000-c.c: Add support for builtins
9704 vector unsigned int vec_parity_lsbb (vector signed int);
9705 vector unsigned int vec_parity_lsbb (vector unsigned int);
9706 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
9707 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
9708 vector unsigned long long vec_parity_lsbb (vector signed long long);
9709 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
9710 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
9711 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
9712 * doc/extend.texi: Update the built-in documentation file for the
9713 new built-in functions.
9715 2017-07-11 David Malcolm <dmalcolm@redhat.com>
9717 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
9718 (layout::m_primary_loc): New field.
9719 (layout::layout): Initialize new field. Move location filtering
9720 logic from here to...
9721 (layout::maybe_add_location_range): ...this new method. Add
9722 support for filtering to just the lines already specified by other
9724 (layout::will_show_line_p): New method.
9725 (gcc_rich_location::add_location_if_nearby): New method.
9726 (selftest::test_add_location_if_nearby): New test function.
9727 (selftest::diagnostic_show_locus_c_tests): Call it.
9728 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
9731 2017-07-11 Tom de Vries <tom@codesourcery.com>
9733 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
9734 (bb_first_real_insn): New function.
9735 (nvptx_single): Add extra initialization of broadcasted condition
9738 2017-07-11 Nathan Sidwell <nathan@acm.org>
9740 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
9742 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
9744 * doc/extend.texi (AVR Function Attributes): Remove weblink to
9745 Binutils doc as TEXI will mess them up.
9746 * doc/invoke.texi (AVR Options): Same here.
9748 2017-07-11 Daniel Cederman <cederman@gaisler.com>
9750 * config/sparc/sparc.opt (mfix-ut700): New option.
9751 (mfix-gr712rc): Likewise.
9752 (sparc_fix_b2bst): New variable.
9753 * doc/invoke.texi (SPARC options): Document them.
9754 (ARM options): Fix warnings.
9755 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
9756 instructions to prevent sequences that can trigger the store-store
9757 errata for certain LEON3FT processors.
9758 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
9759 (sparc_option_override): Set sparc_fix_b2bst appropriately.
9760 * config/sparc/sparc.md (fix_b2bst): New attribute.
9761 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
9763 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
9766 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
9768 (*rsqrtsf2_sse): Ditto.
9770 (div<mode>3): Macroize insn from divdf3 and divsf3
9771 using MODEF mode iterator.
9773 2017-07-10 Martin Sebor <msebor@redhat.com>
9775 PR tree-optimization/80397
9776 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
9777 instead of testing for equality to INTEGER_TYPE.
9779 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
9781 * config.gcc: Remove uclibc from arc target spec.
9783 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
9785 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
9787 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
9790 * lto-wrapper.c (remove_option): New function.
9791 (merge_and_complain): Merge PIC/PIE options more realistically.
9793 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
9795 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
9799 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
9800 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
9801 * config.in: Regenerate.
9802 * configure: Regenerate.
9803 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
9804 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
9805 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
9806 (TARGET_GASISR_PROLOGUES): ...target mask.
9807 * common/config/avr/avr-common.c
9808 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
9809 Set -mgas-isr-prologues.
9810 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
9811 INSERT_PASS_BEFORE for it.
9812 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
9813 * config/avr/avr.c (avr_option_override)
9814 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
9815 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
9816 (avr_attribute_table) <no_gccisr>: Add new function attribute.
9817 (avr_set_current_function) <is_no_gccisr>: Init machine field.
9818 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
9820 (make_avr_pass_pre_proep): New function.
9821 (emit_push_sfr) <treg>: Add argument to function and use it
9823 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
9824 and set machine->gasisr.yes.
9825 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
9826 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
9827 __gcc_isr.n_pushed to .L__stack_usage.
9828 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
9829 (avr_asm_final_postscan_insn): ...this new static function.
9830 * config/avr/avr.h (machine_function)
9831 <is_no_gccisr, use_L__stack_usage>: New fields.
9832 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
9833 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
9834 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
9835 (gasisr, *gasisr): New expander and insn.
9836 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
9837 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
9838 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
9840 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
9842 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
9845 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
9847 Move jump-tables out of .text again.
9850 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
9851 (ASM_OUTPUT_ADDR_VEC): New function.
9852 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
9853 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
9854 INSN_ADDRESSes as asm comment.
9855 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
9856 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
9857 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
9858 * config/avr/avr.md (*tablejump): Adjust comment.
9859 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
9860 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
9862 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
9863 (avr_output_addr_vec): New proto.
9864 (avr_log_t) <insn_addresses>: New field.
9866 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
9869 * config/i386/i386.c (ix86_function_arg_advance): Set
9870 outgoing_args_on_stack to true if there are outgoing arguments
9872 (ix86_function_arg): Likewise.
9873 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
9874 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
9875 * config/i386/i386.h (machine_function): Add
9876 outgoing_args_on_stack.
9878 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
9880 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
9882 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
9884 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
9886 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
9887 (REAL_H): Remove $(MACHMODE_H).
9888 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
9890 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
9891 $(MACHMODE_H) and double-int.h.
9892 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
9894 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
9897 2017-07-07 Andrew Pinski <apinski@cavium.com>
9899 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
9900 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
9902 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
9904 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
9905 Add warning if GCC was not configured to link against a GLIBC that
9906 exports the hardware capability bits.
9907 (make_resolver_func): Make resolver function private and not a
9908 COMDAT function. Create the name with clone_function_name instead
9909 of make_unique_name.
9912 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
9913 correct operand in doing the split.
9915 2017-07-07 Carl Love <cel@us.ibm.com>
9917 * config/rs6000/rs6000-c: Add support for built-in function
9918 vector unsigned short vec_pack_to_short_fp32 (vector float,
9920 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
9921 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
9922 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
9923 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
9924 (convert_4f32_8i16): Add define_expand.
9925 * doc/extend.texi: Update the built-in documentation file for the
9926 new built-in function.
9928 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9930 * config/sparc/m8.md: New file.
9931 * config/sparc/sparc.md: Include m8.md.
9933 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9935 * config/sparc/sparc.opt: New option -mvis4b.
9936 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
9937 (sparc_option_override): Handle VIS4B.
9938 (enum sparc_builtins): Define
9939 SPARC_BUILTIN_DICTUNPACK{8,16,32},
9940 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
9941 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
9942 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
9943 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
9944 (check_constant_argument): New function.
9945 (sparc_vis_init_builtins): Define builtins
9946 __builtin_vis_dictunpack{8,16,32},
9947 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
9948 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
9949 __builtin_vis_fpcmpde{8,16,32}shl and
9950 __builtin_vis_fpcmpur{8,16,32}shl.
9951 (sparc_expand_builtin): Check that the constant operands to
9952 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
9953 constant and in range.
9954 * config/sparc/sparc-c.c (sparc_target_macros): Handle
9956 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
9957 (SPARC_IMM5_P): Likewise.
9958 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
9959 (enabled): Handle vis4b.
9960 (UNSPEC_DICTUNPACK): New unspec.
9961 (UNSPEC_FPCMPSHL): Likewise.
9962 (UNSPEC_FPUCMPSHL): Likewise.
9963 (UNSPEC_FPCMPDESHL): Likewise.
9964 (UNSPEC_FPCMPURSHL): Likewise.
9965 (cpu_feature): New CPU feature `vis4b'.
9966 (dictunpack{8,16,32}): New insns.
9967 (FPCSMODE): New mode iterator.
9968 (fpcscond): New code iterator.
9969 (fpcsucond): Likewise.
9970 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
9971 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
9972 (fpcmpde{8,16,32}{si,di}shl): Likewise.
9973 (fpcmpur{8,16,32}{si,di}shl): Likewise.
9974 * config/sparc/constraints.md: Define constraints `q' for unsigned
9975 2-bit integer constants and `t' for unsigned 5-bit integer
9977 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
9979 (imm5_operand_dictunpack16): Likewise.
9980 (imm5_operand_dictunpack32): Likewise.
9981 (imm2_operand): Likewise.
9982 * doc/invoke.texi (SPARC Options): Document -mvis4b.
9983 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
9984 ditunpack* and fpcmp*shl builtins.
9986 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9988 * config.gcc: Handle m8 in --with-{cpu,tune} options.
9989 * config.in: Add HAVE_AS_SPARC6 define.
9990 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
9992 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
9994 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
9995 (CPP_CPU_SPEC): Handle m8.
9996 (ASM_CPU_SPEC): Likewise.
9997 * config/sparc/sparc-opts.h (enum processor_type): Add
9999 * config/sparc/sparc.c (m8_costs): New struct.
10000 (sparc_option_override): Handle TARGET_CPU_m8.
10001 (sparc32_initialize_trampoline): Likewise.
10002 (sparc64_initialize_trampoline): Likewise.
10003 (sparc_issue_rate): Likewise.
10004 (sparc_register_move_cost): Likewise.
10005 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
10006 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
10007 (ASM_CPU64_DEFAULT_SPEC): Likewise.
10008 (CPP_CPU_SPEC): Handle M8.
10009 (ASM_CPU_SPEC): Likewise.
10010 (AS_M8_FLAG): Define.
10011 * config/sparc/sparc.md: Add m8 to the cpu attribute.
10012 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
10013 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
10015 * configure: Regenerate.
10016 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
10019 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10021 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
10023 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
10024 ("*movdi_insn_sp32"): Do not set v3pipe.
10025 ("*movsi_insn"): Likewise.
10026 ("*movdi_insn_sp64"): Likewise.
10027 ("*movsf_insn"): Likewise.
10028 ("*movdf_insn_sp32"): Likewise.
10029 ("*movdf_insn_sp64"): Likewise.
10030 ("*zero_extendsidi2_insn_sp64"): Likewise.
10031 ("*sign_extendsidi2_insn"): Likewise.
10032 ("*mov<VM32:mode>_insn"): Likewise.
10033 ("*mov<VM64:mode>_insn_sp64"): Likewise.
10034 ("*mov<VM64:mode>_insn_sp32"): Likewise.
10035 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
10036 ("<vlop:code><VL:mode>3"): Likewise.
10037 ("*not_<vlop:code><VL:mode>3"): Likewise.
10038 ("*nand<VL:mode>_vis"): Likewise.
10039 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
10040 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
10041 ("one_cmpl<VL:mode>2"): Likewise.
10042 ("faligndata<VM64:mode>_vis"): Likewise.
10043 ("alignaddrsi_vis"): Likewise.
10044 ("alignaddrdi_vis"): Likweise.
10045 ("alignaddrlsi_vis"): Likewise.
10046 ("alignaddrldi_vis"): Likewise.
10047 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
10048 ("bmaskdi_vis"): Likewise.
10049 ("bmasksi_vis"): Likewise.
10050 ("bshuffle<VM64:mode>_vis"): Likewise.
10051 ("cmask8<P:mode>_vis"): Likewise.
10052 ("cmask16<P:mode>_vis"): Likewise.
10053 ("cmask32<P:mode>_vis"): Likewise.
10054 ("pdistn<P:mode>_vis"): Likewise.
10055 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
10057 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10059 * config/sparc/sparc.md ("subtype"): New insn attribute.
10060 ("*wrgsr_sp64"): Set insn subtype.
10061 ("*rdgsr_sp64"): Likewise.
10062 ("alignaddrsi_vis"): Likewise.
10063 ("alignaddrdi_vis"): Likewise.
10064 ("alignaddrlsi_vis"): Likewise.
10065 ("alignaddrldi_vis"): Likewise.
10066 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
10067 ("fexpand_vis"): Likewise.
10068 ("fpmerge_vis"): Likewise.
10069 ("faligndata<VM64:mode>_vis"): Likewise.
10070 ("bshuffle<VM64:mode>_vis"): Likewise.
10071 ("cmask8<P:mode>_vis"): Likewise.
10072 ("cmask16<P:mode>_vis"): Likewise.
10073 ("cmask32<P:mode>_vis"): Likewise.
10074 ("fchksm16_vis"): Likewise.
10075 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
10076 ("fmean16_vis"): Likewise.
10077 ("fp<plusminus_insn>64_vis"): Likewise.
10078 ("<plusminus_insn>v8qi3"): Likewise.
10079 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
10080 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
10081 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
10082 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
10083 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
10084 ("*movqi_insn"): Likewise.
10085 ("*movhi_insn"): Likewise.
10086 ("*movsi_insn"): Likewise.
10087 ("movsi_pic_gotdata_op"): Likewise.
10088 ("*movdi_insn_sp32"): Likewise.
10089 ("*movdi_insn_sp64"): Likewise.
10090 ("movdi_pic_gotdata_op"): Likewise.
10091 ("*movsf_insn"): Likewise.
10092 ("*movdf_insn_sp32"): Likewise.
10093 ("*movdf_insn_sp64"): Likewise.
10094 ("*zero_extendhisi2_insn"): Likewise.
10095 ("*zero_extendqihi2_insn"): Likewise.
10096 ("*zero_extendqisi2_insn"): Likewise.
10097 ("*zero_extendqidi2_insn"): Likewise.
10098 ("*zero_extendhidi2_insn"): Likewise.
10099 ("*zero_extendsidi2_insn_sp64"): Likewise.
10100 ("ldfsr"): Likewise.
10101 ("prefetch_64"): Likewise.
10102 ("prefetch_32"): Likewise.
10103 ("tie_ld32"): Likewise.
10104 ("tie_ld64"): Likewise.
10105 ("*tldo_ldub_sp32"): Likewise.
10106 ("*tldo_ldub1_sp32"): Likewise.
10107 ("*tldo_ldub2_sp32"): Likewise.
10108 ("*tldo_ldub_sp64"): Likewise.
10109 ("*tldo_ldub1_sp64"): Likewise.
10110 ("*tldo_ldub2_sp64"): Likewise.
10111 ("*tldo_ldub3_sp64"): Likewise.
10112 ("*tldo_lduh_sp32"): Likewise.
10113 ("*tldo_lduh1_sp32"): Likewise.
10114 ("*tldo_lduh_sp64"): Likewise.
10115 ("*tldo_lduh1_sp64"): Likewise.
10116 ("*tldo_lduh2_sp64"): Likewise.
10117 ("*tldo_lduw_sp32"): Likewise.
10118 ("*tldo_lduw_sp64"): Likewise.
10119 ("*tldo_lduw1_sp64"): Likewise.
10120 ("*tldo_ldx_sp64"): Likewise.
10121 ("*mov<VM32:mode>_insn"): Likewise.
10122 ("*mov<VM64:mode>_insn_sp64"): Likewise.
10123 ("*mov<VM64:mode>_insn_sp32"): Likewise.
10125 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10127 * config/sparc/sparc.md ("type"): New insn type viscmp.
10128 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
10130 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
10131 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
10132 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
10133 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
10135 ("n7_vis_logical_11cycle"): Likewise.
10136 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
10137 * config/sparc/niagara2.md ("niag3_vis": Likewise.
10138 * config/sparc/niagara.md ("niag_vis"): Likewise.
10139 * config/sparc/ultra3.md ("us3_fga"): Likewise.
10140 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
10142 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10144 * config/sparc/sparc.md: New instruction type `bmask'.
10145 (bmaskdi_vis): Use the `bmask' type.
10146 (bmasksi_vis): Likewise.
10147 * config/sparc/ultra3.md (us3_array): Likewise.
10148 * config/sparc/niagara7.md (n7_array): Likewise.
10149 * config/sparc/niagara4.md (n4_array): Likewise.
10150 * config/sparc/niagara2.md (niag2_vis): Likewise.
10151 (niag3_vis): Likewise.
10152 * config/sparc/niagara.md (niag_vis): Likewise.
10154 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
10156 * ipa-comdats.c: Remove optimize check from gate.
10157 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
10158 for functions not optimized.
10159 (ipa_fn_summary_read): Skip optimize check.
10160 (ipa_fn_summary_write): Likewise.
10161 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
10163 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
10165 (can_inline_edge_p): Check flag_pcc_struct_return for match.
10166 (check_callers): Give up on caller which is not optimized.
10167 (inline_small_functions): Likewise.
10168 (ipa_inline): Do not give up when not optimizing.
10169 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
10170 away unoptimizes cdtors.
10171 (whole_program_function_and_variable_visibility): Do
10172 ipa_discover_readonly_nonaddressable_vars in LTO mode.
10173 * ipa.c (process_references): Do not check optimize.
10174 (symbol_table::remove_unreachable_nodes): Update optimize check.
10175 (set_writeonly_bit): Update optimize check.
10176 (pass_ipa_cdtor_merge::gate): Do not check optimize.
10177 (pass_ipa_single_use::gate): Remove.
10179 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
10181 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
10182 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
10183 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
10184 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
10185 permute_load, permute_store, adjust_extract, adjust_splat,
10186 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
10187 replace_swap_with_copy, dump_swap_insn_table,
10188 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
10189 recombine_lvx_pattern, recombine_stvx_pattern,
10190 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
10191 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
10192 to file rs6000-p8swap.c.
10193 * config/rs6000/rs6000-p8swap.c: New file.
10194 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
10195 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
10196 and rs6000*-*-* targets.
10198 2017-07-06 David Malcolm <dmalcolm@redhat.com>
10200 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
10202 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
10204 * lto-wrapper.c (merge_and_complain): Do not merge
10205 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
10206 fsigned_zeros, ftrapping_math, fwrapv.
10207 (append_compiler_options): Do not track these options.
10208 (append_linker_options): Likewie
10210 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
10212 * cgraphunit.c (cgraph_node::finalize_function): When
10213 !flag_toplevel_reorde set no_reorder flag.
10214 (varpool_node::finalize_decl): Likewise.
10215 (symbol_table::compile): Drop no toplevel reorder path.
10217 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
10219 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
10220 edges; zero probability is not better than uninitialized.
10222 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
10224 * asan.h (asan_sanitize_allocas_p): Declare.
10225 * asan.c (asan_sanitize_allocas_p): New function.
10226 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
10227 (handle_builtin_alloca): Likewise.
10228 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
10229 if !asan_sanitize_allocas_p.
10230 * params.def (asan-instrument-allocas): Add new option.
10231 * params.h (ASAN_PROTECT_ALLOCAS): Define.
10232 * opts.c (common_handle_option): Disable allocas sanitization for
10235 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
10237 * asan.c: Include gimple-fold.h.
10238 (get_last_alloca_addr): New function.
10239 (handle_builtin_stackrestore): Likewise.
10240 (handle_builtin_alloca): Likewise.
10241 (asan_emit_allocas_unpoison): Likewise.
10242 (get_mem_refs_of_builtin_call): Add new parameter, remove const
10243 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
10244 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
10245 (instrument_builtin_call): Pass gimple iterator to
10246 get_mem_refs_of_builtin_call.
10247 (last_alloca_addr): New global.
10248 * asan.h (asan_emit_allocas_unpoison): Declare.
10249 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
10250 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
10251 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
10252 if function calls alloca.
10253 * gimple-fold.c (replace_call_with_value): Remove static keyword.
10254 * gimple-fold.h (replace_call_with_value): Declare.
10255 * internal-fn.c: Include asan.h.
10256 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
10257 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
10259 2017-07-06 David Malcolm <dmalcolm@redhat.com>
10261 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
10262 (C_SELFTEST_FLAGS): New.
10263 (CPP_SELFTEST_FLAGS): New.
10264 (SELFTEST_DEPS): New, from deps of s-selftest.
10265 (C_SELFTEST_DEPS): New, from deps of s-selftest.
10266 (CPP_SELFTEST_DEPS): New.
10267 (selftest): Add dependency on s-selftest-c++.
10268 (s-selftest): Rename to...
10269 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
10270 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
10271 than SELFTEST_FLAGS.
10272 (selftest-gdb): Rename to...
10273 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
10275 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
10276 (selftest-valgrind): Rename to...
10277 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
10279 (selftest-valgrind): Reintroduce as an alias for
10280 selftest-c-valgrind.
10281 (s-selftest-c++): New.
10282 (selftest-c++-gdb): New.
10283 (selftest-c++-valgrind): New.
10285 2017-07-06 Olivier Hainque <hainque@adacore.com>
10287 * gcc.c (process_command): When deciding if undefined variables
10288 should be ignored when processing specs, accept "gcc -v" as well.
10290 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
10292 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
10293 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
10295 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
10297 * config/arm/arm-cpus.in (armv8-r): Add new entry.
10298 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
10299 * config/arm/arm-tables.opt: Regenerate.
10300 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
10302 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
10304 2017-07-06 Carl Love <cel@us.ibm.com>
10306 * ChangeLog: Clean up from mid air collision
10308 2017-07-06 Carl Love <cel@us.ibm.com>
10310 * config/rs6000/rs6000-c.c: Add support for built-in functions
10311 vector signed int vec_subc (vector signed int, vector signed int);
10312 vector signed __int128 vec_subc (vector signed __int128,
10313 vector signed __int128);
10314 vector unsigned __int128 vec_subc (vector unsigned __int128,
10315 vector unsigned __int128);
10316 vector signed int vec_sube (vector signed int, vector signed int,
10317 vector signed int);
10318 vector unsigned int vec_sube (vector unsigned int,
10319 vector unsigned int,
10320 vector unsigned int);
10321 vector signed __int128 vec_sube (vector signed __int128,
10322 vector signed __int128,
10323 vector signed__int128);
10324 vector unsigned __int128 vec_sube (vector unsigned __int128,
10325 vector unsigned __int128,
10326 vector unsigned __int128);
10327 vector signed int vec_subec (vector signed int, vector signed int,
10328 vector signed int);
10329 vector unsigned int vec_subec (vector unsigned int,
10330 vector unsigned int,
10331 vector unsigned int);
10332 vector signed __int128 vec_subec (vector signed __int128,
10333 vector signed __int128,
10334 vector signed__int128);
10335 vector unsigned __int128 vec_subec (vector unsigned __int128,
10336 vector unsigned __int128,
10337 vector unsigned __int128);
10338 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
10339 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
10340 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
10341 BU_ALTIVEC_OVERLOAD_X definitions.
10342 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
10343 * doc/extend.texi: Update the built-in documentation file for the new
10344 built-in functions.
10346 2017-07-06 David Malcolm <dmalcolm@redhat.com>
10349 * diagnostic-show-locus.c (layout::layout): Use start and finish
10350 spelling location for the start and finish of each range.
10351 * genmatch.c (linemap_client_expand_location_to_spelling_point):
10352 Add unused aspect param.
10353 * input.c (expand_location_1): Add "aspect" param, and use it
10354 to access the correct part of the location.
10355 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
10357 (expand_location_to_spelling_point): Likewise.
10358 (linemap_client_expand_location_to_spelling_point): Add "aspect"
10359 param, and pass it to expand_location_1.
10361 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
10363 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
10364 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
10365 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
10366 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
10367 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
10368 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
10369 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
10370 _mm_maskz_getmant_ss): New intrinsics.
10371 (__builtin_ia32_getexpss128_mask): Changed to ...
10372 __builtin_ia32_getexpss128_round ... this.
10373 (__builtin_ia32_getexpsd128_mask): Changed to ...
10374 __builtin_ia32_getexpsd128_round ... this.
10375 * config/i386/i386-builtin-types.def
10376 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
10377 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
10378 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
10379 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
10380 __builtin_ia32_getmantss_mask_round): New builtins.
10381 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
10382 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
10383 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
10384 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
10385 * config/i386/sse.md
10386 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
10387 avx512f_sgetexp<mode><mask_scalar_name>
10388 <round_saeonly_scalar_name> ... this.
10389 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
10390 %0, %1, %2<round_saeonly_op3>}): Changed to ...
10391 vgetexp<ssescalarmodesuffix>
10392 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
10393 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
10394 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
10395 avx512f_vgetmant<mode><mask_scalar_name>
10396 <round_saeonly_scalar_name> ... this.
10397 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
10398 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
10399 vgetmant<ssescalarmodesuffix>
10400 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
10401 %0<mask_scalar_operand4>, %1, %2
10402 <round_saeonly_scalar_mask_op4>, %3} ... this.
10403 * config/i386/subst.md (mask_scalar_operand4,
10404 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
10405 round_saeonly_scalar_nimm_predicate): New subst attributes.
10407 2017-07-06 Julia Koval <julia.koval@intel.com>
10409 * config/i386/i386.c (ix86_erase_embedded_rounding):
10410 Remove code for old rounding pattern.
10412 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
10414 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
10416 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
10418 * doc/sourcebuild.texi (Test Directives, Variants of
10419 dg-require-support): Add documentation for dg-require-stack-check.
10421 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
10423 * config/i386/subst.md (mask_scalar, round_scalar,
10424 round_saeonly_scalar): New meta-templates.
10425 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
10426 round_scalar_mask_operand3, round_scalar_mask_op3,
10427 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
10428 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
10429 round_saeonly_scalar_constraint,
10430 round_saeonly_scalar_prefix): New subst attribute.
10431 * config/i386/sse.md
10432 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
10433 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
10434 <round_scalar_name> ... this.
10435 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
10436 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
10437 <round_scalar_name> ... this.
10438 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
10439 <sse>_vm<code><mode>3<mask_scalar_name>
10440 <round_saeonly_scalar_name> ... this.
10441 (v<plusminus_mnemonic><ssescalarmodesuffix>
10442 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
10443 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
10444 v<plusminus_mnemonic><ssescalarmodesuffix>
10445 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
10446 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
10447 (v<multdiv_mnemonic><ssescalarmodesuffix>
10448 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
10449 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
10450 v<multdiv_mnemonic><ssescalarmodesuffix>
10451 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
10452 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
10453 (v<maxmin_float><ssescalarmodesuffix>
10454 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
10455 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
10456 v<maxmin_float><ssescalarmodesuffix>
10457 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
10458 %0<mask_scalar_operand3>, %1, %<iptr>2
10459 <round_saeonly_scalar_mask_op3>} ... this.
10461 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
10463 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
10464 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
10466 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
10467 Alan Hayward <alan.hayward@arm.com>
10468 David Sherwood <david.sherwood@arm.com>
10470 * combine.c (simplify_if_then_else): Remove "enum" before
10472 * compare-elim.c (can_eliminate_compare): Likewise.
10473 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
10475 (aarch64_lookup_simd_builtin_type): Likewise.
10476 (aarch64_simd_builtin_type): Likewise.
10477 (aarch64_init_simd_builtin_types): Likewise.
10478 (aarch64_simd_expand_args): Likewise.
10479 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
10481 (aarch64_reverse_mask): Likewise.
10482 (aarch64_simd_emit_reg_reg_move): Likewise.
10483 (aarch64_gen_adjusted_ldpstp): Likewise.
10484 (aarch64_ccmp_mode_to_code): Likewise.
10485 (aarch64_operands_ok_for_ldpstp): Likewise.
10486 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10487 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
10489 (aarch64_min_divisions_for_recip_mul): Likewise.
10490 (aarch64_reassociation_width): Likewise.
10491 (aarch64_get_condition_code_1): Likewise.
10492 (aarch64_simd_emit_reg_reg_move): Likewise.
10493 (aarch64_simd_attr_length_rglist): Likewise.
10494 (aarch64_reverse_mask): Likewise.
10495 (aarch64_operands_ok_for_ldpstp): Likewise.
10496 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10497 (aarch64_gen_adjusted_ldpstp): Likewise.
10498 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
10500 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
10501 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
10502 (arm_lookup_simd_builtin_type): Likewise.
10503 (arm_simd_builtin_type): Likewise.
10504 (arm_init_simd_builtin_types): Likewise.
10505 (arm_expand_builtin_args): Likewise.
10506 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
10507 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
10508 (ft32_setup_incoming_varargs): Likewise.
10509 (ft32_function_arg): Likewise.
10510 (ft32_function_arg_advance): Likewise.
10511 (ft32_pass_by_reference): Likewise.
10512 (ft32_arg_partial_bytes): Likewise.
10513 (ft32_valid_pointer_mode): Likewise.
10514 (ft32_addr_space_pointer_mode): Likewise.
10515 (ft32_addr_space_legitimate_address_p): Likewise.
10516 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
10518 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
10519 (ix86_emit_outlined_ms2sysv_restore): Likewise.
10520 (iamcu_alignment): Likewise.
10521 (canonicalize_vector_int_perm): Likewise.
10522 (ix86_noce_conversion_profitable_p): Likewise.
10523 (ix86_mpx_bound_mode): Likewise.
10524 (ix86_operands_ok_for_move_multiple): Likewise.
10525 * config/microblaze/microblaze-protos.h
10526 (microblaze_expand_conditional_branch_reg): Likewise.
10527 * config/microblaze/microblaze.c
10528 (microblaze_expand_conditional_branch_reg): Likewise.
10529 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
10531 (rs6000_reassociation_width): Likewise.
10532 (rs6000_invalid_binary_op): Likewise.
10533 (fusion_p9_p): Likewise.
10534 (emit_fusion_p9_load): Likewise.
10535 (emit_fusion_p9_store): Likewise.
10536 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
10538 (riscv_hard_regno_mode_ok_p): Likewise.
10539 (riscv_address_insns): Likewise.
10540 (riscv_split_symbol): Likewise.
10541 (riscv_legitimize_move): Likewise.
10542 (riscv_function_value): Likewise.
10543 (riscv_hard_regno_nregs): Likewise.
10544 (riscv_expand_builtin): Likewise.
10545 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
10546 (riscv_build_integer): Likewise.
10547 (riscv_split_integer): Likewise.
10548 (riscv_legitimate_constant_p): Likewise.
10549 (riscv_cannot_force_const_mem): Likewise.
10550 (riscv_regno_mode_ok_for_base_p): Likewise.
10551 (riscv_valid_base_register_p): Likewise.
10552 (riscv_valid_offset_p): Likewise.
10553 (riscv_valid_lo_sum_p): Likewise.
10554 (riscv_classify_address): Likewise.
10555 (riscv_legitimate_address_p): Likewise.
10556 (riscv_address_insns): Likewise.
10557 (riscv_load_store_insns): Likewise.
10558 (riscv_force_binary): Likewise.
10559 (riscv_split_symbol): Likewise.
10560 (riscv_force_address): Likewise.
10561 (riscv_legitimize_address): Likewise.
10562 (riscv_move_integer): Likewise.
10563 (riscv_legitimize_const_move): Likewise.
10564 (riscv_legitimize_move): Likewise.
10565 (riscv_address_cost): Likewise.
10566 (riscv_subword): Likewise.
10567 (riscv_output_move): Likewise.
10568 (riscv_canonicalize_int_order_test): Likewise.
10569 (riscv_emit_int_order_test): Likewise.
10570 (riscv_function_arg_boundary): Likewise.
10571 (riscv_pass_mode_in_fpr_p): Likewise.
10572 (riscv_pass_fpr_single): Likewise.
10573 (riscv_pass_fpr_pair): Likewise.
10574 (riscv_get_arg_info): Likewise.
10575 (riscv_function_arg): Likewise.
10576 (riscv_function_arg_advance): Likewise.
10577 (riscv_arg_partial_bytes): Likewise.
10578 (riscv_function_value): Likewise.
10579 (riscv_pass_by_reference): Likewise.
10580 (riscv_setup_incoming_varargs): Likewise.
10581 (riscv_print_operand): Likewise.
10582 (riscv_elf_select_rtx_section): Likewise.
10583 (riscv_save_restore_reg): Likewise.
10584 (riscv_for_each_saved_reg): Likewise.
10585 (riscv_register_move_cost): Likewise.
10586 (riscv_hard_regno_mode_ok_p): Likewise.
10587 (riscv_hard_regno_nregs): Likewise.
10588 (riscv_class_max_nregs): Likewise.
10589 (riscv_memory_move_cost): Likewise.
10590 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
10591 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
10592 (rl78_addr_space_address_mode): Likewise.
10593 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10595 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
10596 (rs6000_reassociation_width): Likewise.
10597 (rs6000_invalid_binary_op): Likewise.
10598 (fusion_p9_p): Likewise.
10599 (emit_fusion_p9_load): Likewise.
10600 (emit_fusion_p9_store): Likewise.
10601 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
10602 (ok_for_simple_move_operands): Likewise.
10603 (ok_for_simple_move_strict_operands): Likewise.
10604 (ok_for_simple_arith_logic_operands): Likewise.
10605 (visium_legitimize_reload_address): Likewise.
10606 (visium_select_cc_mode): Likewise.
10607 (output_cbranch): Likewise.
10608 (visium_split_double_move): Likewise.
10609 (visium_expand_copysign): Likewise.
10610 (visium_expand_int_cstore): Likewise.
10611 (visium_expand_fp_cstore): Likewise.
10612 * config/visium/visium.c (visium_pass_by_reference): Likewise.
10613 (visium_function_arg): Likewise.
10614 (visium_function_arg_advance): Likewise.
10615 (visium_libcall_value): Likewise.
10616 (visium_setup_incoming_varargs): Likewise.
10617 (visium_legitimate_constant_p): Likewise.
10618 (visium_legitimate_address_p): Likewise.
10619 (visium_legitimize_address): Likewise.
10620 (visium_secondary_reload): Likewise.
10621 (visium_register_move_cost): Likewise.
10622 (visium_memory_move_cost): Likewise.
10623 (prepare_move_operands): Likewise.
10624 (ok_for_simple_move_operands): Likewise.
10625 (ok_for_simple_move_strict_operands): Likewise.
10626 (ok_for_simple_arith_logic_operands): Likewise.
10627 (visium_function_value_1): Likewise.
10628 (rtx_ok_for_offset_p): Likewise.
10629 (visium_legitimize_reload_address): Likewise.
10630 (visium_split_double_move): Likewise.
10631 (visium_expand_copysign): Likewise.
10632 (visium_expand_int_cstore): Likewise.
10633 (visium_expand_fp_cstore): Likewise.
10634 (visium_split_cstore): Likewise.
10635 (visium_select_cc_mode): Likewise.
10636 (visium_split_cbranch): Likewise.
10637 (output_cbranch): Likewise.
10638 (visium_print_operand_address): Likewise.
10639 * expmed.c (flip_storage_order): Likewise.
10640 * expmed.h (emit_cstore): Likewise.
10641 (flip_storage_order): Likewise.
10642 * genrecog.c (validate_pattern): Likewise.
10643 * hsa-gen.c (gen_hsa_addr): Likewise.
10644 * internal-fn.c (expand_arith_overflow): Likewise.
10645 * ira-color.c (allocno_copy_cost_saving): Likewise.
10646 * lra-assigns.c (find_hard_regno_for_1): Likewise.
10647 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
10648 (process_invariant_for_inheritance): Likewise.
10649 * lra-eliminations.c (move_plus_up): Likewise.
10650 * omp-low.c (lower_oacc_reductions): Likewise.
10651 * simplify-rtx.c (simplify_subreg): Likewise.
10652 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
10653 (TARGET_CHKP_BOUND_MODE): Likewise..
10654 * targhooks.c (default_chkp_bound_mode): Likewise.
10655 (default_setup_incoming_vararg_bounds): Likewise.
10656 * targhooks.h (default_chkp_bound_mode): Likewise.
10657 (default_setup_incoming_vararg_bounds): Likewise.
10658 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
10659 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
10660 (have_whole_vector_shift): Likewise.
10661 * tree-vect-stmts.c (vectorizable_load): Likewise.
10662 * doc/tm.texi: Regenerate.
10664 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
10666 Graceful degrade if Binutils PR21472 is not available.
10669 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
10670 .rodata in flash test fails.
10671 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
10672 * confgure: Regenerate.
10673 * config.in: Regenerate.
10674 * config/avr/avr.c (avr_asm_named_section)
10675 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
10676 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
10677 (avr_asm_init_sections): Same.
10679 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
10681 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
10682 (fma<VH:mode>4_intrinsic): Likewise.
10683 (*fmsub<VCVTF:mode>4): Likewise.
10684 (*fmsub<VH:mode>4_intrinsic): Likewise.
10686 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
10689 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
10690 Don't depend on "optimize > 0".
10691 (out_movhi_r_mr, out_movqi_mr_r): Same.
10692 (out_movhi_mr_r, out_movqi_r_mr): Same.
10693 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
10694 io_address_operand on "optimize > 0".
10696 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10698 * tree-loop-distribution.c: Add general explanantion on the pass.
10699 (generate_loops_for_partition): Mark distributed loop.
10700 (pg_add_dependence_edges): New parameter. Handle alias data
10701 dependence specially and record it in the parameter if asked.
10702 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
10703 (init_partition_graph_vertices, add_partition_graph_edge): New.
10704 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
10705 (free_partition_graph_vdata, build_partition_graph): New.
10706 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
10707 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
10708 (data_ref_segment_size, latch_dominated_by_data_ref): New.
10709 (compute_alias_check_pairs, version_loop_by_alias_check): New.
10710 (version_for_distribution_p, finalize_partitions): New.
10711 (distribute_loop): Handle alias data dependence specially. Factor
10712 out loop fusion code as functions and call these functions.
10714 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10716 * tree-loop-distribution.c (classify_partition): New parameter and
10717 better handle reduction statement.
10718 (rdg_build_partitions): Revise comment.
10719 (distribute_loop): Compute statements in all partitions and pass it
10720 to classify_partition.
10722 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10724 * tree-loop-distribution.c (enum partition_type): New.
10725 (struct partition): New field type.
10726 (partition_merge_into): Add parameter. Update partition type.
10727 (data_dep_in_cycle_p, update_type_for_merge): New functions.
10728 (build_rdg_partition_for_vertex): Compute partition type.
10729 (rdg_build_partitions): Dump partition type.
10730 (distribute_loop): Update calls to partition_merge_into.
10732 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10734 * tree-loop-distribution.c (struct ddr_hasher): New.
10735 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
10737 (classify_partition): Call get_data_dependence.
10738 (pg_add_dependence_edges): Ditto.
10739 (distribute_loop): Release data dependence hash table.
10741 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10743 * tree-loop-distribution.c (ref_base_address): Delete.
10744 (similar_memory_accesses): Rename ...
10745 (share_memory_accesses): ... to this. Check if partitions access
10746 the same memory reference.
10747 (distribute_loop): Call share_memory_accesses.
10749 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10751 * tree-loop-distribution.c (struct partition): New field recording
10752 its data reference.
10753 (partition_alloc, partition_free): Init and release data refs.
10754 (partition_merge_into): Merge data refs.
10755 (build_rdg_partition_for_vertex): Collect data refs for partition.
10756 (pg_add_dependence_edges): Change parameters from vector to bitmap.
10758 (distribute_loop): Remve data refs from vertice data of partition
10761 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10763 * tree-loop-distribution.c (params.h): Include header file.
10764 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
10765 (datarefs_vec): New global var.
10766 (create_rdg_vertices): Use datarefs_vec directly.
10767 (free_rdg): Don't free data references.
10768 (build_rdg): Update use. Don't free data references.
10769 (distribute_loop): Compute global variable for data references.
10770 Bail out if there are too many data references.
10772 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10774 * tree-loop-distribution.c (loop_nest): New global var.
10775 (build_rdg): Use loop directly, rather than loop nest.
10776 (pg_add_dependence_edges): Remove loop nest parameter. Use global
10778 (distribute_loop): Compute global variable loop nest. Update use.
10780 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10782 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
10783 (partition_merge_into): New parameter. Dump reason for fusion.
10784 (distribute_loop): Update use of partition_merge_into.
10786 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10788 * tree-loop-distribution.c (bb_top_order_index): New.
10789 (bb_top_order_index_size, bb_top_order_cmp): New.
10790 (stmts_from_loop): Use topological order.
10791 (pass_loop_distribution::execute): Compute and release topological
10792 order for basic blocks.
10794 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10796 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
10799 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10801 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
10802 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
10803 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
10804 * internal-fn.def (LOOP_DIST_ALIAS): New.
10805 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
10806 (fold_loop_internal_call): ... this.
10807 (vect_loop_dist_alias_call): New function.
10808 (set_uid_loop_bbs): Call fold_loop_internal_call.
10809 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
10812 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
10815 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
10816 Require dead FLAGS_REG at the beginning of a peephole.
10818 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
10821 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
10822 arguments in the call to __builtin_ia32_sbb_u32.
10823 (_subborrow_u64): Swap _X and _Y arguments in the call to
10824 __builtin_ia32_sbb_u64.
10826 2017-07-04 Jakub Jelinek <jakub@redhat.com>
10829 * tree-vrp.c (compare_assert_loc): Turn into a function template
10830 with stable template parameter. Only test if a->e is NULL,
10831 !a->e == !b->e has been verified already. Use e == NULL or
10832 e != NULL instead of e or ! e tests. If stable is true, don't use
10833 iterative_hash_expr, on the other side allow a or b or both NULL
10834 and sort the NULLs last.
10835 (process_assert_insertions): Sort using compare_assert_loc<false>
10836 instead of compare_assert_loc, later sort using
10837 compare_assert_loc<true> before calling process_assert_insertions_for
10838 in a loop. Use break instead of continue once seen NULL pointer.
10840 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
10842 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
10843 Cortex-R7 and Cortex-R8 processors.
10845 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
10847 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
10848 uninitialized while src is not.
10850 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
10852 * common/config/arm/arm-common.c: Adjust include path for
10854 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
10855 (arm-cpu.h): Create in build directory. Adjust dependency rules.
10856 (arm-cpu-data.h): Likewise.
10857 (arm-cpu-cdata.h): Likewise.
10858 * config/arm/arm-cpu.h: Delete.
10859 * config/arm/arm-cpu-cdata.h: Delete.
10860 * config/arm/arm-cpu-data.h: Delete.
10862 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
10864 * config/arm/arm-cpus.in (cortex-a55): New.
10865 (cortex-a75): Likewise.
10866 (cortex-a75.cortex-a55): Likewise.
10867 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
10869 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
10870 * config/arm/arm-cpu-cdata.h: Regenerate.
10871 * config/arm/arm-cpu-data.h: Regenerate.
10872 * config/arm/arm-cpu.h: Regenerate.
10873 * config/arm/arm-tables.opt: Regenerate.
10874 * config/arm/arm-tune.md: Regenerate.
10876 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
10878 * haifa-sched.c (sched_create_recovery_edges): Update profile.
10880 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
10882 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
10885 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
10887 PR tree-optimization/81292
10888 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
10889 full_string_p, also call adjust_related_strinfos if the adjustment
10890 is simple, otherwise invalidate related strinfos.
10892 2017-07-04 Martin Liska <mliska@suse.cz>
10895 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
10896 newly created variable as DECL_IGNORED_P.
10898 2017-07-04 Martin Liska <mliska@suse.cz>
10901 * ipa-inline.c (inline_small_functions):
10902 Use xstrdup_for_dump.
10904 2017-07-04 Tom de Vries <tom@codesourcery.com>
10906 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
10908 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
10911 * config/darwin.c (darwin_function_switched_text_sections):
10914 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
10916 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
10918 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
10920 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
10922 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10924 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
10925 min_profitable_iters, and th as inclusive lower bounds.
10926 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
10927 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
10928 for min_profitable_iters and min_profitable_estimate.
10929 (vect_transform_loop): Treat th as an inclusive lower bound.
10930 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
10932 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
10935 * config/darwin.c (darwin_function_switched_text_sections):
10936 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
10937 in two pieces, and suppress the use of buf.
10939 2017-07-03 Nathan Sidwell <nathan@acm.org>
10941 * hash-table.h (hash_table_mod1): Fix indentation.
10943 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10945 PR middle-end/81290
10946 * predict.c (force_edge_cold): Be more careful about propagation
10948 * profile-count.h (profile_probability::guessed,
10949 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
10951 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
10953 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
10955 * doc/invoke.texi (rcpc architecture extension): Document it.
10957 2017-07-03 Richard Biener <rguenther@suse.de>
10959 PR tree-optimization/60510
10960 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
10961 the scalar reduction PHI and use it.
10962 (vectorizable_reduction): Properly guard the single_defuse_cycle
10963 path for non-SLP reduction chains where we cannot use it.
10964 Rework reduc_def/index and vector type deduction. Rework
10965 vector operand gathering during reduction op code-gen.
10966 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
10967 chains dissolve the chain and leave it to non-SLP reduction
10970 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10972 * tree-data-ref.h (dr_alignment): Declare.
10973 * tree-data-ref.c (dr_alignment): New function.
10974 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
10975 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
10977 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
10979 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10981 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
10982 and base_misalignment fields.
10983 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
10984 * tree-data-ref.c: Include builtins.h.
10985 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
10986 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
10987 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
10988 * tree-vect-data-refs.c: Include tree-cfg.h.
10989 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
10990 fields instead of calculating an alignment here.
10991 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
10992 innermost_loop_behavior fields.
10994 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10996 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
10998 (DR_STEP_ALIGNMENT): New macro.
10999 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
11000 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
11001 (create_data_ref): Print it.
11002 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
11003 to tell whether the step preserves vector (mis)alignment.
11004 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
11005 Move the check for an integer step and generalise to all INTEGER_CST.
11006 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
11007 Print the outer step alignment.
11009 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11011 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
11012 with offset_alignment.
11013 (DR_ALIGNED_TO): Delete.
11014 (DR_OFFSET_ALIGNMENT): New macro.
11015 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
11016 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
11017 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
11018 (create_data_ref): Likewise.
11019 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
11020 (vect_analyze_data_refs): Likewise.
11021 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
11022 creating dummy innermost behavior.
11024 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11026 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
11027 with a "innermost_loop_behavior *" and refeence tree.
11028 * tree-data-ref.c (dr_analyze_innermost): Likewise.
11029 (create_data_ref): Update call accordingly.
11030 * tree-predcom.c (find_looparound_phi): Likewise.
11032 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11034 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
11035 fields with dr_wrt_vec_loop.
11036 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
11037 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
11038 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
11039 (vect_dr_behavior): New function.
11040 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
11041 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
11042 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
11043 track whether the step preserves the misalignment.
11044 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
11045 Use vect_dr_behavior.
11046 (vect_setup_realignment): Update call accordingly.
11047 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
11048 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
11049 call to vect_create_addr_base_for_vector_ref.
11050 (vect_create_cond_for_align_checks): Likewise.
11051 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
11052 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
11053 (vect_recog_mask_conversion_pattern): Likewise.
11054 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
11055 (new_stmt_vec_info): Remove redundant zeroing.
11057 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
11059 * common/config/arm/arm-common.c (arm_be8_option): New function.
11060 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
11061 (ISA_ARMv6): Add isa_bit_be8.
11062 * config/arm/arm.h (arm_be8_option): Add prototype.
11063 (BE8_SPEC_FUNCTION): New define.
11064 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
11065 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
11066 (mlittle-endian): Similarly.
11067 (mbe8, mbe32): New options.
11068 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
11069 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
11071 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11073 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
11075 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11077 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
11078 (cleanup_tree_cfg_bb): Use it.
11079 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
11081 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
11083 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11086 * loop-doloop.c (add_test): Update profile.
11088 2017-07-03 Martin Liska <mliska@suse.cz>
11091 * sanopt.c (rewrite_usage_of_param): New function.
11092 (sanitize_rewrite_addressable_params): Likewise.
11093 (pass_sanopt::execute): Call rewrite_usage_of_param.
11095 2017-07-03 Richard Biener <rguenther@suse.de>
11097 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
11098 back to using VIEW_CONVERT_EXPR.
11100 2017-07-03 Martin Liska <mliska@suse.cz>
11103 * doc/extend.texi: Document when a resolver function is
11104 generated for target_clones.
11106 2017-07-03 Martin Liska <mliska@suse.cz>
11108 * asan.c (asan_emit_stack_protection): Unpoison just red zones
11109 and shadow memory of auto variables which are subject of
11110 use-after-scope sanitization.
11111 (asan_expand_mark_ifn): Add do set only when is_poison.
11113 2016-07-03 Richard Biener <rguenther@suse.de>
11115 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
11117 (vect_force_simple_reduction): Record reduction def -> phi mapping.
11118 (vectorizable_reduction): Perform reduction PHI creation when
11119 visiting a reduction PHI and adjust and simplify code generation
11120 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
11121 (vect_transform_loop): Visit reduction PHIs.
11122 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
11123 defs into the SLP tree.
11124 (vect_build_slp_tree): Reduction defs terminate the recursion.
11125 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
11127 (vect_get_vec_defs_for_stmt_copy): Export.
11128 (vect_get_vec_defs): Likewise.
11129 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
11131 (vect_get_vec_defs_for_stmt_copy): Declare.
11132 (vect_get_vec_defs): Likewise.
11134 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11136 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
11137 parameter with a "loop" parameter and use it instead of the
11138 loop containing DR_STMT. Don't check simple_iv when doing
11139 BB analysis. Describe the two analysis modes in the comment.
11141 2017-07-03 Tom de Vries <tom@codesourcery.com>
11143 PR tree-optimization/69468
11144 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
11145 (find_same_succ_bb): Handle ignore_edge_flags.
11147 2017-07-03 Tom de Vries <tom@codesourcery.com>
11149 PR tree-optimization/81192
11150 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
11152 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
11154 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
11156 2017-07-03 Tom de Vries <tom@codesourcery.com>
11158 PR tree-optimization/81192
11159 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
11160 BB_SAME_SUCC (bb) == NULL.
11162 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11164 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
11167 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11169 * dumpfile.c: Include profile-count.h
11170 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
11172 (insert_cond_bb): Update profile.
11173 * tree-cfg.h (insert_cond_bb): Update prototype.
11174 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
11175 * tree-dump.c: Do not include tree-cfg.
11177 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11179 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
11181 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11183 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
11186 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11188 * tree-complex.c (expand_complex_div_wide): update profile.
11190 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
11191 Alan Hayward <alan.hayward@arm.com>
11192 David Sherwood <david.sherwood@arm.com>
11194 * Makefile.in (MACHMODE_H): Remove insn-modes.h
11195 (CORETYPES_H): New define.
11196 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
11197 (insn-modes-inline.h, s-modes-inline-h): New rules.
11198 (generated_files): Add insn-modes-inline.h.
11199 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
11200 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
11201 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
11202 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
11203 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
11204 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
11205 (build/gencodes.o, build/genconditions.o): Likewise.
11206 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
11207 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
11208 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
11209 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
11210 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
11211 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
11212 * coretypes.h: Include everything up to real.h for generators.
11213 Include insn-modes.h first. Include wide-int-print.h after
11214 wide-int.h. Include insn-modes-inline.h and then machmode.h.
11215 * machmode.h: Don't include insn-modes.h here.
11216 * function-tests.c: Remove includes of signop.h, machmode.h,
11217 double-int.h and wide-int.h.
11219 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
11221 * optc-save-gen.awk: Likewise.
11222 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
11223 * godump.c: Remove include of wide-int-print.h.
11224 * pretty-print.h: Likewise.
11225 * wide-int-print.cc: Likewise.
11226 * wide-int.cc: Likewise.
11227 * hash-map-tests.c: Remove include of signop.h.
11228 * hash-set-tests.c: Likewise.
11229 * rtl-tests.c: Likewise.
11230 * mkconfig.sh: Remove include of machmode.h.
11231 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
11233 (emit_insn_modes_inline_h): ...this new function. Emit the code
11234 into an insn-modes-inline.h header file, adding appropriate
11235 include guards and end comments.
11236 (emit_insn_modes_c_header): Remove include of machmode.h.
11237 (emit_min_insn_modes_c_header): Include coretypes.h rather than
11239 (main): Handle -i flag and call emit_insn_modes_inline_h when
11242 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
11244 * tree-ssa-strlen.c (strinfo): Rename the length field to
11245 nonzero_chars. Add a full_string_p field.
11246 (compare_nonzero_chars, zero_length_string_p): New functions.
11247 (get_addr_stridx): Add an offset_out parameter.
11248 Use compare_nonzero_chars.
11249 (get_stridx): Update accordingly. Use compare_nonzero_chars.
11250 (new_strinfo): Update after above changes to strinfo.
11251 (set_endptr_and_length): Set full_string_p.
11252 (get_string_length): Update after above changes to strinfo.
11253 (unshare_strinfo): Update call to new_strinfo.
11254 (maybe_invalidate): Likewise.
11255 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
11256 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
11257 as a uhwi instead of an shwi. Update after above changes to
11258 strinfo and new_strinfo.
11259 (zero_length_string): Assert that chainsi contains full strings.
11260 Use zero_length_string_p. Update call to new_strinfo.
11261 (adjust_related_strinfos): Update after above changes to strinfo.
11262 Copy full_string_p from origsi.
11263 (adjust_last_stmt): Use zero_length_string_p.
11264 (handle_builtin_strlen): Update after above changes to strinfo and
11265 new_strinfo. Install the lhs as the string length if the previous
11266 entry didn't describe a full string.
11267 (handle_builtin_strchr): Update after above changes to strinfo
11269 (handle_builtin_strcpy): Likewise.
11270 (handle_builtin_strcat): Likewise.
11271 (handle_builtin_malloc): Likewise.
11272 (handle_pointer_plus): Likewise.
11273 (handle_builtin_memcpy): Likewise. Track nonzero characters
11274 that aren't necessarily followed by a nul terminator.
11275 (handle_char_store): Likewise.
11277 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
11279 PR tree-optimization/80769
11280 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
11281 for malloc and calloc. Document the new invariant that all related
11282 strinfos have delayed lengths or none do.
11283 (verify_related_strinfos): Move earlier in file.
11284 (set_endptr_and_length): New function, split out from...
11285 (get_string_length): ...here. Also set the lengths of related
11287 (zero_length_string): Assert that chainsi has known (rather than
11289 (adjust_related_strinfos): Likewise.
11291 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
11293 PR tree-optimization/81136
11294 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
11295 assert that two references with the same misalignment have the same
11296 compile-time misalignment if those compile-time misalignments
11299 2017-07-01 Andi Kleen <ak@linux.intel.com>
11301 * print-tree.c (print_node): Print all attributes.
11303 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11305 * cfg.c (scale_bbs_frequencies): New function.
11306 * cfg.h (scale_bbs_frequencies): Declare it.
11307 * cfgloopanal.c (single_likely_exit): Cleanup.
11308 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
11310 (scale_loop_profile): Likewise.
11311 (loop_version): Likewise.
11312 (create_empty_loop_on_edge): Update.
11313 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
11314 scale_loop_frequencies, scale_loop_profile, loopify,
11315 loop_version): Update prototypes.
11316 * modulo-sched.c (sms_schedule): Update.
11317 * predict.c (unlikely_executed_edge_p): Also check probability.
11318 (probably_never_executed_edge_p): Fix typo.
11319 * tree-if-conv.c (version_loop_for_if_conversion): Update.
11320 * tree-parloops.c (gen_parallel_loop): Update.
11321 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
11322 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
11323 * tree-ssa-loop-split.c (split_loop): Update.
11324 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
11325 * tree-vect-loop-manip.c (vect_do_peeling): Update.
11326 (vect_loop_versioning): Update.
11327 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
11329 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11331 * trans-mem.c (split_bb_make_tm_edge): Update profile.
11333 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11335 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
11336 to keep profile consistent.
11338 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11340 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
11341 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
11342 * profile-count.h (max_safe_multiplier): Make unsigned.
11343 (profile_count::guessed_zero): New.
11345 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11347 * bb-reorder.c (fix_up_crossing_landing_pad,
11348 fix_crossing_conditional_branches): Use make_single_succ_edge
11349 to keep profile consistent.
11351 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11353 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
11356 2017-07-01 Jakub Jelinek <jakub@redhat.com>
11359 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
11360 the right scopes, make sure cond_jump isn't preserved between multiple
11361 iterations. Search for fallthru edge whenever there are 3+ edges and
11362 use find_fallthru_edge for it.
11364 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
11366 Patch by Alexander Monakov <amonakov@ispras.ru>
11367 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
11368 probabilities consistently.
11370 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
11372 * pa.c (pa_expand_compare_and_swap_loop): Update call of
11373 emit_cmp_and_jump_insns.
11375 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
11378 * tree-inline.c (expand_call_inline): Combine profile statuses.
11380 2017-06-30 Andrew Pinski <apinski@cavium.com>
11382 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
11383 fold_stmt returned true.
11385 2017-06-30 Nathan Sidwell <nathan@acm.org>
11387 * ggc.h (empty_string): Delete.
11388 * cfgexpand.c (expand_asm_stmt): Use plain "".
11389 * optabs.c (expand_asm_memory_barrier): Likewise.
11390 * stringpool.c (empty_string): Delete.
11391 (digit_vector, digit_string): Delete.
11392 (ggc_alloc_string): Use plain "", don't optimize single digit
11393 strings. Use ggc_alloc_atomic.
11395 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
11397 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
11398 comparison set and one other set, use the cost of the non-comparison
11401 2017-06-30 Nathan Sidwell <nathan@acm.org>
11403 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
11406 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
11408 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
11409 loops. Remove now unneeded calls to gimple_switch_set_label() that
11410 just set removed labels to NULL_TREE.
11412 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
11414 * tree-ssanames.c (set_range_info_raw): Abstract from ...
11415 (set_range_info): ...here. Only call set_range_info_raw if domain
11417 (set_nonzero_bits): Call set_range_info_raw.
11418 * tree-ssanames.h (set_range_info_raw): New.
11420 2017-06-30 Jakub Jelinek <jakub@redhat.com>
11423 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
11424 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
11425 of nonimmediate_operand and <store_mask_constraint> instead of m
11426 for the input operand. For V8FI iterator, always split if input
11427 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
11428 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
11429 <store_mask_predicate> instead of register_operand and
11430 <store_mask_constraint> instead of v for the input operand. Make
11431 sure both operands aren't MEMs for if not <mask_applied>.
11433 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
11435 * lto-wrapper.c (copy_file) Close both file descriptors before
11438 2017-06-30 Martin Liska <mliska@suse.cz>
11441 * multiple_target.c (create_dispatcher_calls): Make ifunc
11442 also for function that don't have calls or are not referenced.
11444 2017-06-30 Richard Biener <rguenther@suse.de>
11446 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
11447 analyze the first scalar stmt. Move vector type computation
11448 for the BB case here from ...
11449 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
11450 live operation processing in the SLP case properly.
11452 2017-06-30 Richard Biener <rguenther@suse.de>
11454 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
11456 2017-06-30 Martin Liska <mliska@suse.cz>
11459 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
11460 before BUILT_IN_UNWIND_RESUME when ASAN is used.
11462 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
11464 * doc/invoke.texi (AArch64): Add missing options and remove redundant
11467 2017-06-30 Richard Biener <rguenther@suse.de>
11469 PR tree-optimization/81249
11470 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
11471 condition reduction result to original scalar type.
11473 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11475 * profile-count.h (enum profile_quality): Fix typos and whitespace
11478 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11480 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
11481 type for branch probabilities.
11483 2017-06-29 Julian Brown <julian@codesourcery.com>
11484 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11486 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
11487 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
11488 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
11489 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
11491 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11493 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
11494 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
11495 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
11496 CC usage from generic code to here.
11497 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
11498 CC usage into the target macros.
11500 2017-06-29 Maya Rashish <coypu@sdf.org>
11502 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
11505 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
11507 * arm/arm-builtins.c: Include profile-count.h
11508 * except.c (sjlj_emit_function_enter): Use
11509 profile_probability::unlikely.
11511 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
11513 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
11514 and tocrel_offset be pointer args rather than implicitly using
11516 (legitimate_constant_pool_address_p, rs6000_emit_move,
11517 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
11518 tocrel_offset and use in toc_relative_expr_p call.
11519 (print_operand, print_operand_address): Use static tocrel_base_oac
11520 and tocrel_offset_oac.
11521 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
11524 2017-06-29 Maya Rashish <coypu@sdf.org>
11526 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
11528 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
11530 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
11531 objects, take into account only the alignment of 'op0' and 'mode1' if
11534 2017-06-29 Steve Ellcey <sellcey@cavium.com>
11536 * ccmp.c (ccmp_tree_comparison_p): New function.
11537 (ccmp_candidate_p): Update to use above function.
11538 (get_compare_parts): New function.
11539 (expand_ccmp_next): Update to use new functions.
11540 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
11542 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
11543 take mode as argument.
11544 * ccmp.h (expand_ccmp_expr): Add mode as argument.
11545 * expr.c (expand_expr_real_1): Pass mode as argument.
11547 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
11549 * combine.c (combine_instructions): Print insns to dump_file, together
11552 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
11554 * asan.c (asan_emit_stack_protection): Update.
11555 (create_cond_insert_point): Update.
11556 * auto-profile.c (afdo_propagate_circuit): Update.
11557 * basic-block.h (struct edge_def): Turn probability to
11558 profile_probability.
11559 (EDGE_FREQUENCY): Update.
11560 * bb-reorder.c (find_traces_1_round): Update.
11561 (better_edge_p): Update.
11562 (sanitize_hot_paths): Update.
11563 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
11564 (make_single_succ_edge): Update.
11565 (check_bb_profile): Update.
11566 (dump_edge_info): Update.
11567 (update_bb_profile_for_threading): Update.
11568 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
11569 probabilitycount to 0.
11570 * cfgbuild.c (compute_outgoing_frequencies): Update.
11571 * cfgcleanup.c (try_forward_edges): Update.
11572 (outgoing_edges_match): Update.
11573 (try_crossjump_to_edge): Update.
11574 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
11575 (expand_gimple_tailcall): Update.
11576 (construct_init_block): Use make_single_succ_edge.
11577 (construct_exit_block): Use make_single_succ_edge.
11578 * cfghooks.c (verify_flow_info): Update.
11579 (redirect_edge_succ_nodup): Update.
11580 (split_edge): Update.
11581 (account_profile_record): Update.
11582 * cfgloopanal.c (single_likely_exit): Update.
11583 * cfgloopmanip.c (scale_loop_profile): Update.
11584 (set_zero_probability): Remove.
11585 (duplicate_loop_to_header_edge): Update.
11586 * cfgloopmanip.h (loop_version): Update prototype.
11587 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
11588 (force_nonfallthru_and_redirect): Update.
11589 (update_br_prob_note): Update.
11590 (rtl_verify_edges): Update.
11591 (purge_dead_edges): Update.
11592 (rtl_lv_add_condition_to_bb): Update.
11593 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
11594 * cgraphunit.c (init_lowered_empty_function): Update.
11595 (cgraph_node::expand_thunk): Update.
11596 * cilk-common.c: Include profile-count.h
11597 * dojump.c (inv): Remove.
11598 (jumpifnot): Update.
11599 (jumpifnot_1): Update.
11600 (do_jump_1): Update.
11602 (do_jump_by_parts_greater_rtx): Update.
11603 (do_compare_rtx_and_jump): Update.
11604 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
11605 do_jump_1. do_compare_rtx_and_jump): Update prototype.
11606 * dwarf2cfi.c: Include profile-count.h
11607 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
11608 (sjlj_emit_dispatch_table): Likewise.
11609 * explow.c: Include profile-count.h
11610 * expmed.c (emit_store_flag_force): Update.
11611 (do_cmp_and_jump): Update.
11612 * expr.c (compare_by_pieces_d::generate): Update.
11613 (compare_by_pieces_d::finish_mode): Update.
11614 (emit_block_move_via_loop): Update.
11615 (store_expr_with_bounds): Update.
11616 (store_constructor): Update.
11617 (expand_expr_real_2): Update.
11618 (expand_expr_real_1): Update.
11619 * expr.h (try_casesi, try_tablejump): Update prototypes.
11620 * gimple-pretty-print.c (dump_probability): Update.
11621 (dump_profile): New.
11622 (dump_gimple_label): Update.
11623 (dump_gimple_bb_header): Update.
11624 * graph.c (draw_cfg_node_succ_edges): Update.
11625 * hsa-gen.c (convert_switch_statements): Update.
11626 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
11627 (find_if_case_1): Update.
11628 (find_if_case_2): Update.
11629 * internal-fn.c (expand_arith_overflow_result_store): Update.
11630 (expand_addsub_overflow): Update.
11631 (expand_neg_overflow): Update.
11632 (expand_mul_overflow): Update.
11633 (expand_vector_ubsan_overflow): Update.
11634 * ipa-cp.c (good_cloning_opportunity_p): Update.
11635 * ipa-split.c (split_function): Use make_single_succ_edge.
11636 * ipa-utils.c (ipa_merge_profiles): Update.
11637 * loop-doloop.c (add_test): Update.
11638 (doloop_modify): Update.
11639 * loop-unroll.c (compare_and_jump_seq): Update.
11640 (unroll_loop_runtime_iterations): Update.
11641 * lra-constraints.c (lra_inheritance): Update.
11642 * lto-streamer-in.c (input_cfg): Update.
11643 * lto-streamer-out.c (output_cfg): Update.
11644 * mcf.c (adjust_cfg_counts): Update.
11645 * modulo-sched.c (sms_schedule): Update.
11646 * omp-expand.c (expand_omp_for_init_counts): Update.
11647 (extract_omp_for_update_vars): Update.
11648 (expand_omp_ordered_sink): Update.
11649 (expand_omp_for_ordered_loops): Update.
11650 (expand_omp_for_generic): Update.
11651 (expand_omp_for_static_nochunk): Update.
11652 (expand_omp_for_static_chunk): Update.
11653 (expand_cilk_for): Update.
11654 (expand_omp_simd): Update.
11655 (expand_omp_taskloop_for_outer): Update.
11656 (expand_omp_taskloop_for_inner): Update.
11657 * omp-simd-clone.c (simd_clone_adjust): Update.
11658 * optabs.c (expand_doubleword_shift): Update.
11659 (expand_abs): Update.
11660 (emit_cmp_and_jump_insn_1): Update.
11661 (expand_compare_and_swap_loop): Update.
11662 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
11663 * predict.c (predictable_edge_p): Update.
11664 (edge_probability_reliable_p): Update.
11665 (set_even_probabilities): Update.
11666 (combine_predictions_for_insn): Update.
11667 (combine_predictions_for_bb): Update.
11668 (propagate_freq): Update.
11669 (estimate_bb_frequencies): Update.
11670 (force_edge_cold): Update.
11671 * profile-count.c (profile_count::dump): Add missing space into dump.
11672 (profile_count::debug): Add newline.
11673 (profile_count::differs_from_p): Explicitly convert to unsigned.
11674 (profile_count::stream_in): Update.
11675 (profile_probability::dump): New member function.
11676 (profile_probability::debug): New member function.
11677 (profile_probability::differs_from_p): New member function.
11678 (profile_probability::differs_lot_from_p): New member function.
11679 (profile_probability::stream_in): New member function.
11680 (profile_probability::stream_out): New member function.
11681 * profile-count.h (profile_count_quality): Rename to ...
11682 (profile_quality): ... this one.
11683 (profile_probability): New.
11684 (profile_count): Update.
11685 * profile.c (compute_branch_probabilities): Update.
11686 * recog.c (peep2_attempt): Update.
11687 * sched-ebb.c (schedule_ebbs): Update.
11688 * sched-rgn.c (find_single_block_region): Update.
11689 (compute_dom_prob_ps): Update.
11690 (schedule_region): Update.
11691 * sel-sched-ir.c (compute_succs_info): Update.
11692 * stmt.c (struct case_node): Update.
11693 (do_jump_if_equal): Update.
11694 (get_outgoing_edge_probs): Update.
11695 (conditional_probability): Update.
11696 (emit_case_dispatch_table): Update.
11697 (expand_case): Update.
11698 (expand_sjlj_dispatch_table): Update.
11699 (emit_case_nodes): Update.
11700 * targhooks.c: Update.
11701 * tracer.c (better_p): Update.
11702 (find_best_successor): Update.
11703 * trans-mem.c (expand_transaction): Update.
11704 * tree-call-cdce.c: Update.
11705 * tree-cfg.c (gimple_split_edge): Upate.
11706 (move_sese_region_to_fn): Upate.
11707 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
11708 * tree-eh.c (lower_resx): Upate.
11709 (cleanup_empty_eh_move_lp): Upate.
11710 * tree-if-conv.c (version_loop_for_if_conversion): Update.
11711 * tree-inline.c (copy_edges_for_bb): Update.
11712 (copy_cfg_body): Update.
11713 * tree-parloops.c (gen_parallel_loop): Update.
11714 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
11715 (gimple_gen_time_profiler): Update.
11716 * tree-ssa-dce.c (remove_dead_stmt): Update.
11717 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
11718 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
11719 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
11720 (unloop_loops): Update.
11721 (try_peel_loop): Update.
11722 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
11723 * tree-ssa-loop-split.c (connect_loops): Update.
11724 (split_loop): Update.
11725 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
11726 (hoist_guard): Update.
11727 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
11728 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
11729 (value_replacement): Update.
11730 * tree-ssa-reassoc.c (branch_fixup): Update.
11731 * tree-ssa-tail-merge.c (replace_block_by): Update.
11732 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
11733 (create_edge_and_update_destination_phis): Update.
11734 (compute_path_counts): Update.
11735 (recompute_probabilities): Update.
11736 (update_joiner_offpath_counts): Update.
11737 (freqs_to_counts_path): Update.
11738 (duplicate_thread_path): Update.
11739 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
11740 (struct switch_conv_info): Update.
11741 (gen_inbound_check): Update.
11742 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
11743 (vect_do_peeling): Update.
11744 (vect_loop_versioning): Update.
11745 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
11746 (optimize_mask_stores): Update.
11747 * ubsan.c (ubsan_expand_null_ifn): Update.
11748 * value-prof.c (gimple_divmod_fixed_value): Update.
11749 (gimple_divmod_fixed_value_transform): Update.
11750 (gimple_mod_pow2): Update.
11751 (gimple_mod_pow2_value_transform): Update.
11752 (gimple_mod_subtract): Update.
11753 (gimple_mod_subtract_transform): Update.
11754 (gimple_ic): Update.
11755 (gimple_stringop_fixed_value): Update.
11756 (gimple_stringops_transform): Update.
11757 * value-prof.h: Update.
11759 2017-06-29 Carl Love <cel@us.ibm.com>
11761 * config/rs6000/rs6000-c.c: Add support for built-in functions
11762 vector signed int vec_signed (vector float);
11763 vector signed long long vec_signed (vector double);
11764 vector signed int vec_signed2 (vector double, vector double);
11765 vector signed int vec_signede (vector double);
11766 vector signed int vec_signedo (vector double);
11767 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
11768 instruction generator.
11769 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
11770 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
11771 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
11773 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
11774 vunsignede_v2df): Add define_expands.
11775 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
11776 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
11777 VEC_UNSIGNEDO): Add definitions.
11778 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
11779 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
11780 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
11781 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
11782 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
11783 * config/rs6000/altivec.h (vec_signed, vec_signed2,
11784 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
11785 vec_unsignede, vec_unsignedo): Add builtin defines.
11786 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
11788 * doc/extend.texi: Update the built-in documentation file for the
11789 new built-in functions.
11791 2017-06-29 Richard Biener <rguenther@suse.de>
11793 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
11794 reduction chains to LOOP_VINFO_REDUCTIONS.
11795 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
11796 SLP reductions after processing reduction chains.
11798 2017-06-29 Nathan Sidwell <nathan@acm.org>
11800 * builtins.c (fold_builtin_FUNCTION): Use
11801 lang_hooks.decl_printable_name.
11803 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
11805 PR middle-end/81194
11806 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
11807 with only one label.
11808 * stmt.c (expand_case): Assert NCASES is greater than one.
11810 2017-06-29 Richard Biener <rguenther@suse.de>
11812 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
11814 (group_case_labels): Likewise.
11815 (find_taken_edge): Push sanity checking on val to workers...
11816 (find_taken_edge_cond_expr): ... here
11817 (find_taken_edge_switch_expr): ... and here, handle cases
11818 with just a default label.
11819 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
11820 (group_case_labels): Likewise.
11821 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
11822 group_case_labels does anything cleanup the CFG again.
11824 2017-06-29 Bin Cheng <bin.cheng@arm.com>
11826 PR tree-optimization/81196
11827 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
11828 exit condition comparing two IVs.
11830 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
11832 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
11833 profile to the dummy entry at the end of the list of architectures.
11834 * config/arm/arm-cpu-cdata.h: Regenerated.
11836 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11837 Michael Collison <michael.collison@arm.com>
11840 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
11842 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
11843 (*aarch64_reg_<mode>3_minus_mask): New pattern.
11844 (*aarch64_<optab>_reg_di3_mask2): New pattern.
11845 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
11846 of shift when the shift amount is masked with constant equal to
11847 the size of the mode.
11848 * config/aarch64/predicates.md (subreg_lowpart_operator): New
11851 2017-06-29 Martin Liska <mliska@suse.cz>
11853 * config/i386/i386.opt: Change range from [1,5] to [0,5].
11855 2017-06-29 Yury Gribov <tetra2005@gmail.com>
11858 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
11861 (edge_growth_cache_entry::edge_growth_cache_entry): New
11863 (reset_edge_growth_cache): Update to use constructor.
11865 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
11867 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
11868 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
11869 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
11871 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
11873 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
11874 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
11876 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
11878 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
11879 (*-linux-uclibc*): Add t-uclibc tmake_file.
11880 * config/t-musl: New.
11881 * config/t-uclibc: New.
11883 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
11885 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
11887 (gen_comm_data): Emit architectural setting of arch_prof.
11888 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
11890 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
11891 (armv8-m.base, armv8-m.main): Likewise.
11892 * arm-protos.h (arm_build_target): Add profile field.
11893 (arch_option): Likewise.
11894 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
11896 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
11897 arm_active_target.profile.
11899 2017-06-28 Richard Biener <rguenther@suse.de>
11901 PR middle-end/81227
11902 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
11903 TYPE_OVERFLOW_WRAPS.
11904 * match.pd (negate_expr_p): Likewise.
11905 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
11906 fold_build2, not fold_binary.
11908 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
11910 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
11911 Convert memory address to Pmode.
11912 (aarch64_print_operand): Assert MEM operands are always Pmode.
11914 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
11917 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
11918 Remove redundant if.
11919 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
11920 * config/arm/aarch-common-protos.h
11921 (aarch_forward_to_shift_is_not_shifted_re): Remove.
11922 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
11924 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
11927 * multiple_target.c (create_dispatcher_calls): Set the default
11928 clone to be static, not public.
11930 2017-06-28 Richard Biener <rguenther@suse.de>
11932 * tree-vect-loop.c (vectorizable_reduction): Move special
11933 cond reduction IV var creation ...
11934 (vect_create_epilog_for_reduction): ... here. Remove induction_index
11935 parameter. Use STMT_VINFO_VECTYPE.
11936 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
11939 2017-06-28 Martin Liska <mliska@suse.cz>
11942 * ipa-visibility.c (non_local_p): Handle visibility.
11944 2017-06-28 Martin Liska <mliska@suse.cz>
11947 * common.opt: Add IntegerRange to various options.
11948 * opt-functions.awk (integer_range_info): New function.
11949 * optc-gen.awk: Add integer_range_info to cl_options struct.
11950 * opts-common.c (decode_cmdline_option): Handle
11951 CL_ERR_INT_RANGE_ARG.
11952 (cmdline_handle_error): Likewise.
11953 * opts.c (print_filtered_help): Show valid interval in
11954 when --help is provided.
11955 * opts.h (struct cl_option): Add range_min and range_max fields.
11956 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
11958 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
11960 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
11961 (x * C EQ/NE y * C): New transformation.
11963 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
11965 * genmultilib (combination_space): Accept '+' in option names.
11967 2017-06-28 Martin Liska <mliska@suse.cz>
11970 * asan.c (instrument_derefs): Bail out inner references
11971 that are hard register variables.
11973 2017-06-28 Jakub Jelinek <jakub@redhat.com>
11976 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
11977 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
11979 2017-06-28 Richard Biener <rguenther@suse.de>
11981 * tree-vectorizer.h (vect_get_vec_defs): Remove.
11982 (vect_get_slp_defs): Adjust.
11983 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
11985 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
11987 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
11988 get_initial_defs_for_reduction instead of vect_get_vec_defs.
11989 (vectorizable_reduction): Adjust.
11990 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
11992 (vect_get_slp_defs): Likewise.
11993 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
11994 (vectorizable_bswap): Adjust.
11995 (vectorizable_call): Likewise.
11996 (vectorizable_conversion): Likewise.
11997 (vectorizable_assignment): Likewise.
11998 (vectorizable_shift): Likewise.
11999 (vectorizable_operation): Likewise.
12000 (vectorizable_store): Likewise.
12001 (vectorizable_condition): Likewise.
12002 (vectorizable_comparison): Likewise.
12004 2017-06-28 Michael Collison <michael.collison@arm.com>
12007 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
12009 (arm_gen_movmemqi): Removed unused variable 'i'.
12010 Convert 'for' loop into 'while' loop.
12011 (arm_expand_prologue): Remove last unnecessary set of insn.
12012 (thumb_pop): Remove unused variable 'pushed_words'.
12013 (thumb_exit): Remove last unnecessary set of regs_to_pop.
12015 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12017 * config/s390/predicates.md: Use s390_rel_address_ok_p.
12018 * config/s390/s390-protos.h: Add prototype of
12019 s390_rel_address_ok_p.
12020 * config/s390/s390.c (s390_got_symbol): New function.
12021 (s390_rel_address_ok_p): New function.
12022 (legitimize_pic_address): Use s390_rel_address_ok_p.
12023 (s390_load_got): Use s390_got_symbol.
12024 (s390_option_override): Issue error if
12025 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
12026 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
12028 * config/s390/s390.opt: New option mpic-data-is-text-relative.
12030 2017-06-27 Andrew Pinski <apinski@cavium.com>
12032 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
12033 (X * copysign (1.0, X)): New pattern.
12034 (X * copysign (1.0, -X)): New pattern.
12035 (copysign (-1.0, CST)): New pattern.
12037 2017-06-27 Joseph Myers <joseph@codesourcery.com>
12039 * genmultilib (combination_space): Remove variable.
12040 Validate reuse rules against regular expression for any sequence
12041 of multilib options in any order.
12043 2017-06-27 Michael Collison <michael.collison@arm.com>
12045 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
12046 call aarch64_split_simd_combine.
12047 * (aarch64_combine_internal<mode>): Delete pattern.
12048 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
12049 Allow register and subreg operands.
12051 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
12053 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
12054 specific need, just fallback on defaults.
12055 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
12057 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
12058 Olivier Hainque <hainque@adacore.com>
12060 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
12062 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
12063 targets. Pick a default if no particular attempt applied.
12064 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
12067 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
12069 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
12070 (x86_64-wrs-vxworks7): Likewise.
12072 2017-06-27 Marek Polacek <polacek@redhat.com>
12075 * ubsan.c (instrument_null): Check get_base_address's result for null.
12077 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
12079 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
12081 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
12083 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
12084 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
12085 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
12086 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
12087 New function types.
12088 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
12089 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
12090 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
12091 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
12092 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
12093 BUILT_IN_FEUPDATEENV): New builtins.
12094 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
12095 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
12096 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
12097 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
12099 (builtin_structptr_types): Adjust size.
12100 * tree.c (builtin_structptr_types): Add four entries.
12102 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
12103 Olivier Hainque <hainque@adacore.com>
12105 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
12106 (TLS_SYM): New local macro, forcing reference to __tls__ on
12107 link command lines for VxWorks 7 RTPs, triggering initialization
12109 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
12110 OS features TLS support, true for RTPs on VxWorks 7.
12111 * config/vxworks.c (vxworks_override_options): Setup emutls
12114 2017-06-27 Jakub Jelinek <jakub@redhat.com>
12116 * predict.c (test_prediction_value_range): Use -1U instead of -1
12117 to avoid narrowing conversion warning.
12118 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
12119 to avoid narrowing conversion warning.
12120 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
12122 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
12124 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
12126 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
12127 64bit configurations.
12128 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
12129 (SIZE_TYPE): Likewise.
12130 * config/vxworks.c (vxworks_emutls_var_fields): Use
12131 long_unsigned_type_node instead of unsigned_type_node as the offset
12132 field type, which is "pointer" mode in emutls.c.
12134 2017-06-27 Jakub Jelinek <jakub@redhat.com>
12137 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
12139 PR middle-end/81207
12140 * gimple-fold.c (replace_call_with_call_and_fold): Handle
12141 gimple_vuse copying separately from gimple_vdef copying.
12143 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
12145 * value-prof.c (free_hist): Remove call to memset and the enclosing if
12148 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
12149 Olivier Hainque <hainque@adacore.com>
12151 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
12152 for all vxworks7 targets.
12153 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
12154 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
12155 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
12156 variations for VX6/VX7 and 32/64bits later on in ...
12157 (VXWORKS_LIB_SPEC): Leverage new macros.
12158 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
12159 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
12161 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
12163 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
12164 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
12166 2017-06-26 Carl Love <cel@us.ibm.com>
12168 * config/rs6000/rs6000-c.c: Add support for built-in functions
12169 vector bool char vec_reve (vector bool char);
12170 vector signed char vec_reve (vector signed char);
12171 vector unsigned char vec_reve (vector unsigned char);
12172 vector bool int vec_reve (vector bool int);
12173 vector signed int vec_reve (vector signed int);
12174 vector unsigned int vec_reve (vector unsigned int);
12175 vector bool long long vec_reve (vector bool long long);
12176 vector signed long long vec_reve (vector signed long long);
12177 vector unsigned long long vec_reve (vector unsigned long long);
12178 vector bool short vec_reve (vector bool short);
12179 vector signed short vec_reve (vector signed short);
12180 vector double vec_reve (vector double);
12181 vector float vec_reve (vector float);
12182 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
12183 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
12184 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
12185 (altivec_vreve): New pattern.
12186 * config/rs6000/altivec.h (vec_reve): New define.
12187 * doc/extend.texi (vec_rev): Update the built-in documentation file
12188 for the new built-in functions.
12190 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12192 PR tree-optimization/71815
12193 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
12195 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
12197 (slsr_process_phi): Likewise.
12198 (replace_uncond_cands_and_profitable_phis): Don't replace a
12199 multiply candidate with a stride of 1 (copy or cast).
12200 (phi_incr_cost): Call uses_consumed_by_stmt rather than
12202 (lowest_cost_path): Likewise.
12203 (total_savings): Likewise.
12205 2017-06-26 Richard Biener <rguenther@suse.de>
12208 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
12209 Use def_builtin_pure for all gather builtins.
12211 2017-06-26 Richard Biener <rguenther@suse.de>
12213 PR tree-optimization/81203
12214 * tree-tailcall.c (find_tail_calls): Do not move stmts into
12215 non-dominating BBs.
12217 2017-06-26 Marek Polacek <polacek@redhat.com>
12220 * doc/invoke.texi: Document -Wmultistatement-macros.
12222 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
12224 * doc/sourcebuild.texi (ARM-specific attributes): Document new
12225 arm_neon_ok_no_float_abi effective target.
12227 2017-06-26 Richard Biener <rguenther@suse.de>
12229 PR tree-optimization/80928
12230 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
12231 (copy_bbs): Set BB_DUPLICATED flag early.
12232 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
12234 (execute_on_shrinking_pred): Likewise.
12235 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
12236 BB_DUPLICATED blocks.
12237 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
12238 iterate over all PHIs considering removal of *gsi.
12240 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
12242 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
12245 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
12247 * config/rs6000/rs6000-string.c: (expand_block_clear,
12248 do_load_for_compare, select_block_compare_mode,
12249 compute_current_alignment, expand_block_compare,
12250 expand_strncmp_align_check, expand_strn_compare,
12251 expand_block_move, rs6000_output_load_multiple)
12252 Move functions related to string/block move/compare
12253 to a separate file.
12254 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
12255 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
12256 for this function which is now used in two files.
12257 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
12258 * config.gcc: Add rs6000-string.o to extra_objs for
12259 targets powerpc*-*-* and rs6000*-*-*.
12261 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
12264 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
12265 32-bit, since indexed is not valid for DImode.
12266 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
12267 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
12268 (define_peephole2 for Altivec d-form load): Add 32-bit support.
12269 (define_peephole2 for Altivec d-form store): Likewise.
12272 * multiple_target.c (create_dispatcher_calls): Only create the
12273 dispatcher call if the function is the default clone of a
12274 versioned function.
12276 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
12278 PR middle-end/80902
12279 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
12280 a call, force the call to not be a tail call.
12282 2017-06-23 Jeff Law <law@redhat.com>
12284 * doc/contrib.texi: Add entry for Steven Pemberton's work on
12287 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
12289 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
12290 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
12291 handling for early expansion of vector shifts (sl,sr,sra,rl).
12292 (builtin_function_type): Add vector shift right instructions
12293 to the unsigned argument list.
12295 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
12297 rtl-optimizatoin/79286
12298 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
12299 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
12300 trap. PIC register plus a const unspec without offset can never trap.
12302 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
12304 * tree.h (builtin_structptr_type): New type.
12305 (builtin_structptr_types): Declare new array.
12306 * tree.c (builtin_structptr_types): New array.
12307 (free_lang_data, build_common_tree_nodes): Use it.
12309 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
12312 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
12315 2017-06-22 Matt Turner <mattst88@gmail.com>
12317 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
12318 Lake models to skylake case. Assume skylake for unknown
12319 models with clflushopt.
12321 2017-06-22 Jeff Law <law@redhat.com>
12323 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
12324 frame sizes that do not satisfy aarch64_uimm12_shift.
12326 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
12328 * profile-count.h (apply_probability,
12329 apply_scale, probability_in): Fix checks for zero.
12331 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12333 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
12334 * doc/cppdiropts.texi (-I @var{dir}): Document it.
12336 2016-06-22 Richard Biener <rguenther@suse.de>
12338 * tree-vect-loop.c (vect_model_reduction_cost): Handle
12339 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
12340 REDUC_MAX_EXPR support.
12341 (vectorizable_reduction): Likewise.
12342 (vect_create_epilog_for_reduction): Likewise.
12344 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
12346 * match.pd (A / (1 << B) -> A >> B): New.
12347 * generic-match-head.c: Include optabs-tree.h.
12348 * gimple-match-head.c: Likewise.
12349 * optabs-tree.h (target_supports_op_p): New.
12350 * optabs-tree.c (target_supports_op_p): New.
12352 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12354 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
12355 $gcc_cv_ld --help output.
12356 (gcc_cv_ld_demangle): Likewise.
12357 (gcc_cv_ld_eh_frame_hdr): Likewise.
12358 (gcc_cv_ld_pie): Likewise.
12359 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
12360 (gcc_cv_ld_buildid): Likewise.
12361 (gcc_cv_ld_sysroot): Likewise.
12362 (ld_bndplt_support): Likewise.
12363 (ld_pushpopstate_support): Likewise.
12364 * configure: Regenerate.
12365 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
12367 2017-06-21 Jakub Jelinek <jakub@redhat.com>
12370 * config/i386/sse.md (round<mode>2): Renumber match_dup and
12371 operands indexes to avoid gap between operands and match_dups.
12373 2017-06-21 Andrew Pinski <apinski@cavium.com>
12375 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
12376 Increment Arith_shift and Arith_shift_reg by 1.
12377 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
12379 * config/aarch64/aarch64.c (thunderx_tunings): Enable
12380 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
12381 (aarch64_strip_extend): Add new argument and test for it.
12382 (aarch64_cheap_mult_shift_p): New function.
12383 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
12384 add a cost if it is true.
12385 Update calls to aarch64_strip_extend.
12386 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
12388 2017-06-21 Andrew Pinski <apinski@cavium.com>
12390 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
12392 (thunderxt88): Likewise.
12393 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
12394 (thunderx_prefetch_tune): New variable.
12395 (thunderx2t99_prefetch_tune): Update for the correct values.
12396 (thunderxt88_tunings): New variable.
12397 (thunderx_tunings): Use thunderx_prefetch_tune instead of
12398 generic_prefetch_tune.
12399 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
12401 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12403 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
12404 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
12405 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
12406 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
12407 (aarch64_atomic_cas<mode>, GPI): Likewise.
12409 2017-06-21 Martin Liska <mliska@suse.cz>
12411 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
12412 statements on cold and hot labels.
12413 * predict.c (tree_estimate_probability_bb): Remove the
12414 prediction from this place.
12416 2017-06-21 Martin Liska <mliska@suse.cz>
12418 PR tree-optimization/79489
12419 * gimplify.c (maybe_add_early_return_predict_stmt): New
12421 (gimplify_return_expr): Call the function.
12422 * predict.c (tree_estimate_probability_bb): Remove handling
12424 * predict.def: Update comment about early return predictor.
12425 * gimple-predict.h (is_gimple_predict): New function.
12426 * predict.def: Change default value of early return to 66.
12427 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
12429 * passes.def: Put pass_strip_predict_hints to the beginning of
12432 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
12434 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
12435 FUNCTION_DECL declarations.
12436 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
12438 (dwaf2out_decl): Likewise.
12439 * godump.c (go_early_global_decl): Skip call to the real debug hook
12440 for FUNCTION_DECL declarations.
12441 * passes.c (rest_of_decl_compilation): Skip call to the
12442 early_global_decl debug hook for FUNCTION_DECL declarations, unless
12443 -fdump-go-spec is passed.
12445 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
12447 * config/i386/i386.c (struct builtin_isa): New field pure_p.
12448 Reorder for compactness.
12449 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
12450 (def_builtin_pure, def_builtin_pure2): New functions.
12451 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
12453 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
12455 * match.pd (nop_convert): New predicate.
12456 ((A +- CST1) +- CST2): Allow some NOP conversions.
12458 2017-06-21 Jakub Jelinek <jakub@redhat.com>
12461 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
12462 with ctors/dtors if GOVD_SHARED is set.
12464 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
12466 * config/aarch64/aarch64.md (movti_aarch64):
12467 Emit mov rather than orr.
12468 (movtf_aarch64): Likewise.
12469 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
12470 Emit mov rather than orr.
12472 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
12474 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
12475 Swap alternatives, make integer dup more expensive.
12477 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
12479 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
12480 Return true for non-tls symbols.
12482 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
12484 * config/aarch64/aarch64-cores.def (cortex-a55): New.
12485 (cortex-a75): Likewise.
12486 (cortex-a75.cortex-a55): Likewise.
12487 * config/aarch64/aarch64-tune.md: Regenerate.
12488 * doc/invoke.texi (-mtune): Document new values for -mtune.
12490 2017-06-21 Tom de Vries <tom@codesourcery.com>
12492 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
12493 stack_size feature.
12494 (Effective-Target Keywords, Other attributes): Suggest using
12495 dg-add-options stack_size feature to get stack limit in stack_size
12496 effective target documentation.
12498 2017-06-21 Julian Brown <julian@codesourcery.com>
12499 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
12501 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
12502 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
12503 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
12505 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
12506 attribute type list for neon_multiply.
12507 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
12508 attribute type list for neon_multiply.
12509 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
12510 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
12511 attribute type list for neon_multiply.
12512 * config/arm/types.md (crypto_pmull): Add.
12513 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
12514 attribute type list.
12516 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
12518 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
12521 2017-06-20 Jakub Jelinek <jakub@redhat.com>
12523 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
12524 to make sure not to dereference a NULL cost_classes_ptr pointer.
12526 2017-06-20 Carl Love <cel@us.ibm.com>
12528 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12529 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
12530 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
12531 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
12532 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
12533 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
12534 VMULOSW): New enum "unspec" values.
12535 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
12536 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
12537 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
12538 altivec_vmulosw): New patterns.
12539 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
12540 VMULOSW): Add definitions.
12542 2017-06-20 Julia Koval <julia.koval@intel.com>
12544 * config/i386/i386.c: Fix rounding expand for new pattern.
12545 * config/i386/subst.md: Fix pattern (parallel -> unspec).
12547 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
12549 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
12550 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
12552 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
12554 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
12557 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
12559 * config/aarch64/aarch64-cores.def: Rearrange to sort by
12560 architecture, then by implementer ID.
12561 * config/aarch64/aarch64-tune.md: Regenerate.
12563 2017-06-20 Richard Biener <rguenther@suse.de>
12565 PR middle-end/81097
12566 * fold-const.c (split_tree): Fold to type before negating.
12568 2017-06-20 David Malcolm <dmalcolm@redhat.com>
12570 * diagnostic-show-locus.c
12571 (selftest::test_fixit_deletion_affecting_newline): New function.
12572 (selftest::diagnostic_show_locus_c_tests): Call it.
12574 2017-06-20 Andreas Schwab <schwab@suse.de>
12577 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
12580 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
12582 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
12583 __ARM_FEATURE_COPROC according to support.
12585 2017-06-20 Jakub Jelinek <jakub@redhat.com>
12587 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
12588 Rewritten to avoid overflow for > 32-bit pointers.
12591 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
12592 by removing enum keyword.
12593 (ubsan_type_descriptor): Likewise. Formatting fix.
12596 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
12597 splitter): Require TARGET_SSE2 in the condition.
12599 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
12602 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
12603 for doing vector set of SFmode on ISA 3.0.
12604 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
12605 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
12607 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
12608 SFmode value into a V4SF variable that was extracted from another
12609 V4SF variable without converting the element to double precision
12610 and back to single precision vector format.
12611 (vsx_insert_extract_v4sf_p9_2): Likewise.
12613 2017-06-19 Jakub Jelinek <jakub@redhat.com>
12615 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
12616 in UWHI to avoid undefined overflow.
12619 * ubsan.h (enum ubsan_encode_value_phase): New.
12620 (ubsan_encode_value): Change second argument to
12621 enum ubsan_encode_value_phase with default value of
12622 UBSAN_ENCODE_VALUE_GENERIC.
12623 * ubsan.c (ubsan_encode_value): Change second argument to
12624 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
12625 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
12626 create_tmp_var_raw instead of create_tmp_var and use a
12628 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
12629 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
12630 ubsan_encode_value callers.
12633 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
12634 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
12635 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
12637 2017-06-19 Richard Biener <rguenther@suse.de>
12639 PR middle-end/81118
12640 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
12641 estimates if we changed anything.
12643 2017-06-19 Richard Biener <rguenther@suse.de>
12645 PR tree-optimization/80887
12646 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
12647 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
12648 simplified lookups, then reset mprts_hook.
12649 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
12651 (try_to_simplify): Likewise.
12653 2017-06-19 Martin Liska <mliska@suse.cz>
12656 * gimplify.c (gimplify_switch_expr):
12657 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
12659 2017-06-19 Martin Liska <mliska@suse.cz>
12661 * doc/install.texi: Document that PGO runs in 4 stages.
12663 2017-06-19 Martin Liska <mliska@suse.cz>
12666 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
12667 to dispatcher function name.
12668 * multiple_target.c (replace_function_decl): New function.
12669 (create_dispatcher_calls): Redirect both edges and references.
12671 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
12673 * profile-count.c (profile_count::dump): Dump quality.
12674 (profile_count::differs_from_p): Update for unsigned val.
12675 * profile-count.h (profile_count_quality): New enum.
12676 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
12678 2017-06-19 Richard Biener <rguenther@suse.de>
12680 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
12681 struct function as arg.
12682 (estimate_numbers_of_iterations): Export overload with loop arg.
12683 (free_numbers_of_iterations_estimates_loop): Use an overload of
12684 free_numbers_of_iterations_estimates instead.
12685 * tree-cfg.c (remove_bb): Adjust.
12686 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
12687 * tree-parloops.c (gen_parallel_loop): Likewise.
12688 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
12690 (tree_unroll_loops_completely): Likewise.
12691 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
12692 Use an overload instead and export.
12693 (estimated_loop_iterations): Adjust.
12694 (max_loop_iterations): Likewise.
12695 (likely_max_loop_iterations): Likewise.
12696 (estimate_numbers_of_iterations): Take struct function as arg
12698 (loop_exits_before_overflow): Adjust.
12699 (free_numbers_of_iterations_estimates_loop): Use an overload.
12700 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
12701 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
12703 2017-06-19 Richard Biener <rguenther@suse.de>
12706 * ipa-prop.c (find_constructor_constant_at_offset): Handle
12707 RANGE_EXPR conservatively.
12709 2017-06-16 Carl Love <cel@us.ibm.com>
12711 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12712 definitions for vec_float, vec_float2, vec_floato,
12713 vec_floate built-ins.
12714 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
12715 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
12717 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
12718 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
12719 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
12720 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
12721 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
12722 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
12723 vec_floato): Add builtin defines.
12724 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
12725 Update the built-in documentation file for the new built-in
12728 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12730 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
12731 (mthumb): Mark as the negative of -marm.
12733 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12735 * doc/invoke.texi (ARM Options, -mcpu): Document supported
12737 (ARM Options, -mtune): Document that this accepts the same
12738 extension options as -mcpu.
12739 (ARM Options, -mfpu): Document addition of -mfpu=auto.
12741 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12743 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
12744 permitted extensions.
12746 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12748 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
12749 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
12750 (armv8-m.main): Add option +nodsp.
12751 * config/arm/arm-cpu-cdata.h: Regenerated.
12753 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12755 * config/arm/t-fuchsia: New file.
12756 * config.gcc (arm*-*-fuchsia*): Use it.
12758 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12760 * config/arm/t-symbian: Rewrite for new option infrastructure.
12762 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12764 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
12765 (MULTILIB_REQUIRED): Likewise.
12767 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12769 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
12770 (MULTILIB_RESUE): Likewise.
12771 (MULTILIB_MATCHES): Likewise.
12772 (MULTLIB_REQUIRED): Likewise.
12774 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12776 * config/arm/t-rtems: Rewrite for new option framework.
12778 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12780 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
12781 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
12782 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
12783 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
12784 * config/arm/t-multilib: ... here.
12785 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
12786 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
12787 armv7-a and armv8*-a when A-profile libraries have not been built.
12788 * config/arm/t-rmprofile: Rewrite.
12790 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12792 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
12793 with a backslash. Remove the backslash after substituting unescaped
12795 * doc/fragments.texi (MULTILIB_REUSE): Document it.
12797 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12799 * config.gcc: (arm*-*-*): When building a-profile libraries, force
12800 the driver to pass through the default setting of -mfloat-abi.
12801 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
12803 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
12804 (all_feat_combs): New rule.
12805 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
12807 * config/arm/t-aprofile: Rewrite.
12809 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12811 * config/arm/arm.h (FPUTYPE_AUTO): Define.
12812 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
12813 fpu is not specified by the user/command-line.
12814 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
12815 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
12816 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
12817 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
12818 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
12819 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
12821 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12823 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
12824 * config/arm/t-arm-elf: Rewritten.
12826 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12828 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
12829 have some floating-point instructions.
12830 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
12831 (TARGET_MAYBE_HARD_FLOAT): New macro.
12832 * config/arm/arm-builtins.c (arm_init_builtins): Use
12833 TARGET_MAYBE_HARD_FLOAT.
12834 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
12836 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12838 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
12839 (configargs.h): Include it.
12840 (arm_print_hint_for_fpu_option): New function.
12841 (arm_parse_fpu_option): New function.
12842 (candidate_extension): New class.
12843 (arm_canon_for_multilib): New function.
12844 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
12845 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
12846 (ARCH_CANONICAL_SPECS): New macro.
12847 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
12849 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12851 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
12852 are set after handling multilib fragments. Set target_cpu_default2
12855 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12857 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
12859 (arm*-*-*): Set target_cpu_default2 to a quoted string.
12860 * config/arm/parsecpu.awk (check_cpu): Validate any extension
12862 (check_arch): Likewise.
12863 * config/arm/arm.c (arm_configure_build_target): Handle
12864 TARGET_CPU_DEFAULT being a string constant. Scan any feature
12865 options in the default.
12867 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12869 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
12870 when an option is an alias of another.
12871 * config/arm/parsecpu.awk (optalias): New parser token.
12872 (gen_comm_data): Mark non-alias options as such. Emit entries
12873 for extension aliases.
12874 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
12875 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
12876 (armv6kz, armv6zk, armv6t2): Likewise.
12877 (armv7): Make vfpv3-d16 an alias.
12878 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
12880 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
12881 Sort in canonical order.
12882 (armv8-a): Sort in canonical order.
12883 (armv8.1-a, armv8.2-a): Likewise.
12884 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
12886 (cortex-a9): Sort in canonical order.
12887 * config/arm/arm.c (selftests.h): Include it.
12888 (arm_test_cpu_arch_data): New function.
12889 (arm_run_self_tests): New function.
12890 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
12891 (targetm): Move declaration to the end of the file.
12892 * arm-cpu-cdata.h: Regenerated.
12894 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12896 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
12897 call to target_mode_check describing the type of option passed.
12898 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
12899 (arm_target_thumb_only): Use arm_parse_arch_option_name or
12900 arm_parse_cpu_option_name to match parameters against list of
12902 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
12903 arm_arch_core_flags data structure.
12904 * config/arm/arm-cpu_cdata.h: Regenerated.
12906 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12908 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
12910 (arm_print_hint_for_cpu_option): Likewise.
12911 (arm_print_hint_for_arch_option): Likewise.
12912 (arm_parse_cpu_option_name): Likewise.
12913 (arm_parse_arch_option_name): Likewise.
12914 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
12915 of entries in the all_fpus list.
12916 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
12917 (arm_parse_cpu_option_name): Declare.
12918 (arm_parse_arch_option_name): Declare.
12919 (arm_parse_option_features): Declare.
12920 (arm_intialize_isa): Declare.
12921 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
12923 (gen_comm_data): ... here. Make definitions non-static.
12924 * config/arm/arm-cpu-data.h: Regenerated.
12925 * config/arm/arm-cpu-cdata.h: Regenerated.
12927 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12929 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
12930 (cpu_arch_extension): New structure.
12931 (cpu_arch_option, arch_option, cpu_option): New structures.
12932 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
12933 architecture types.
12934 (gen_data): Generate new format data tables.
12935 * config/arm/arm.c (cpu_tune): New structure.
12936 (cpu_option, processors): Delete.
12937 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
12938 (arm_print_hint_for_cpu_option): ... this and ...
12939 (arm_print_hint_for_arch_option): ... this.
12940 (arm_parse_arch_cpu_name): Delete. Replace with ...
12941 (arm_parse_cpu_option_name): ... this and ...
12942 (arm_parse_arch_option_name): ... this.
12943 (arm_unrecognized_feature): Change type of target parameter to
12945 (arm_parse_arch_cpu_features): Delete. Replace with ...
12946 (arm_parse_option_features): ... this.
12947 (arm_configure_build_target): Rework to use new configuration data
12949 (arm_print_tune_info): Rework for new configuration data tables.
12950 * config/arm/arm-cpu-data.h: Regenerated.
12951 * config/arm/arm-cpu.h: Regenerated.
12953 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12955 * Makefile.in (OBJS): Move sbitmap.o from here ...
12956 (OBJS-libcommon): ... to here.
12958 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12960 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
12961 (ISA_ALL_CRYPTO): New macro.
12962 (ISA_ALL_SIMD): New macro
12963 (ISA_ALL_FP): New macro.
12964 * config/arm/arm.c (fpu_bitlist): Update initializer.
12965 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
12967 (arm9e): Add fpu. Add option for nofp
12968 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
12969 (arm926ej-s, arm1026ej-s): Likewise.
12970 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
12971 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
12972 neon-fp16, neon-vfpv4, nofp and nosimd.
12973 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
12974 (cortex-a8): Add fpu. Add option for nofp.
12975 (cortex-a9): Add fpu. Add options for nosimd and nofp.
12976 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
12977 (cortex-r4f): Add fpu.
12978 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
12979 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
12981 (cortex-r8): Likewise.
12982 (cortex-m4): Add fpu. Add option for nofp.
12983 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
12984 (cortex-a17.cortex-a7): Likewise.
12985 (cortex-a32): Add fpu. Add options for crypto and nofp.
12986 (cortex-a35, cortex-a53): Likewise.
12987 (cortex-a57): Add fpu. Add option for crypto.
12988 (cortex-a72, cortex-a73): Likewise.
12989 (exynos-m1): Likewise.
12990 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
12991 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
12992 (cortex-m33): Add fpu. Add option for nofp.
12993 * config/arm/arm-cpu-cdata.h: Regenerated
12994 * config/arm/arm-cpu-data.h: Regenerated.
12996 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12998 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
12999 (armv5te, armv5tej): Likewise.
13000 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
13001 (armv7): Add options fp and vfpv3-d16.
13002 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
13003 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
13005 (armv7ve): Likewise.
13006 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
13007 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
13008 (armv8-a): Add nocrypto option.
13009 (armv8.1-a, armv8.2-a): Likewise.
13010 (armv8-m.main): add options fp, fp.dp and nofp.
13012 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13014 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
13016 (armv8-a+crc): Delete.
13017 (armv8.1-a): Add options simd, crypto and nofp.
13018 (armv8.2-a): Add options fp16, simd, crypto and nofp.
13019 (armv8.2-a+fp16): Delete.
13020 (armv8-m.main): Add option dsp.
13021 (armv8-m.main+dsp): Delete.
13022 (cortex-a8): Add fpu. Add nofp option.
13023 (cortex-a9): Add fpu. Add nofp and nosimd options.
13024 * config/arm/parsecpu.awk (gen_data): Generate option tables and
13025 link to main cpu and architecture data structures.
13026 (gen_comm_data): Only put isa attributes from the main architecture
13028 (option): New statement for architecture and CPU entries.
13029 * arm.c (struct cpu_option): New structure.
13030 (struct processors): Add entry for options.
13031 (arm_unrecognized_feature): New function.
13032 (arm_parse_arch_cpu_name): Ignore any characters after the first
13034 (arm_parse_arch_cpu_feature): New function.
13035 (arm_configure_build_target): Separate out any CPU and architecture
13036 features and parse separately. Don't error out if -mfpu=auto is
13037 used with only an architecture string.
13038 (arm_print_asm_arch_directives): New function.
13039 (arm_file_start): Call it.
13040 * config/arm/arm-cpu-cdata.h: Regenerated.
13041 * config/arm/arm-cpu-data.h: Likewise.
13042 * config/arm/arm-tables.opt: Likewise.
13044 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13046 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
13047 assembler when it is not -mfpu=auto.
13049 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13051 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
13052 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
13053 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
13054 (ASM_CPU_SPEC): Rewrite.
13055 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
13056 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
13057 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
13058 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
13059 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
13060 copied string is NUL-terminated. Also strip any characters prefixed
13062 (arm_rewrite_selected_arch): New function.
13063 (arm_rewrite_march): New function.
13065 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13067 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
13068 (x_arm_cpu_string, x_arm_tune_string): Likewise.
13069 (march, mcpu, mtune): Convert to string-based options.
13070 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
13071 (arm_parse_arch_cpu_name): New function.
13072 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
13073 identify selected architecture or CPU.
13074 (arm_option_save): New function.
13075 (TARGET_OPTION_SAVE): Redefine.
13076 (arm_option_restore): Restore string options.
13077 (arm_option_print): Print string options.
13079 2017-06-16 Martin Sebor <msebor@redhat.com>
13081 PR tree-optimization/80933
13082 PR tree-optimization/80934
13083 * builtins.c (fold_builtin_3): Do not handle bcmp here.
13084 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
13085 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
13086 (gimple_fold_builtin): Call them.
13088 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13090 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
13091 as unlikely; update profile.
13093 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13095 * predict.c (force_edge_cold): Handle declaring edges impossible
13098 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13100 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
13102 (try_unroll_loop_completely): Fix reporting.
13104 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13106 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
13108 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
13111 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
13112 if given a non-constant argument for an intrinsic which requires a
13115 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13117 * profile.c (compare_freqs): New function.
13118 (branch_prob): Sort edge list.
13119 (find_spanning_tree): Assume that the list is priority sorted.
13121 2017-06-16 Richard Biener <rguenther@suse.de>
13123 PR tree-optimization/81090
13124 * passes.def (pass_record_bounds): Remove.
13125 * tree-pass.h (make_pass_record_bounds): Likewise.
13126 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
13127 make_pass_record_bounds): Likewise.
13128 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
13129 not free niter estimates at the beginning but at the end.
13130 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
13132 2017-06-16 Richard Biener <rguenther@suse.de>
13134 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
13135 initializer to workaround ICE in host GCC 4.8.
13137 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13139 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
13141 (clone_inlined_nodes): Update.
13143 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
13145 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
13146 prefetch settings, and enable prefetching by default at -O3.
13148 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
13150 * config/aarch64/aarch64.c (aarch64_override_options_internal):
13151 Set flag_prefetch_loop_arrays according to tuning data.
13153 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
13155 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
13156 New tune structure.
13157 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
13158 [Unrelated to main purpose of the patch] Place the pointer field last
13159 to enable type checking errors when tune structure are wrongly merged.
13160 * config/aarch64/aarch64.c (generic_prefetch_tune,)
13161 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
13162 (thunderx2t99_prefetch_tune): New tune constants.
13163 (tune_params *_tunings): Update all tunings (no functional change).
13164 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
13165 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
13166 from tunings structures.
13168 2017-06-16 Jakub Jelinek <jakub@redhat.com>
13171 * ubsan.c (instrument_null): Add T argument, use it instead
13172 of computing it based on IS_LHS.
13173 (instrument_object_size): Likewise.
13174 (pass_ubsan::execute): Adjust instrument_null and
13175 instrument_object_size callers to pass gimple_get_lhs or
13176 gimple_assign_rhs1 result to it. Use instrument_null instead of
13177 calling get_base_address and instrument_mem_ref. Handle
13178 aggregate call arguments for object-size sanitization.
13180 2017-06-16 Yury Gribov <tetra2005@gmail.com>
13182 PR tree-optimization/81089
13183 * tree-vrp.c (is_masked_range_test): Validate operands of
13186 2017-06-15 Martin Sebor <msebor@redhat.com>
13189 * dumpfile.c (dump_register): Avoid calling memset to initialize
13190 a class with a default ctor.
13191 * gcc.c (struct compiler): Remove const qualification.
13192 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
13193 * hash-table.h: Ditto.
13194 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
13196 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
13197 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
13199 * params.h (struct param_info): Make struct members non-const.
13200 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
13201 with default initialization.
13202 * vec.h (vec_copy_construct, vec_default_construct): New helper
13204 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
13205 with vec_copy_construct.
13206 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
13207 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
13208 * doc/invoke.texi (-Wclass-memaccess): Document.
13210 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
13212 * emit-rtl.h (is_leaf): Update comment about local
13213 register allocator.
13215 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13218 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
13219 for a variable to have a section before checking if the section has a
13221 Set section to.persistent if persistent attribute is set.
13222 Warn if .persistent attribute is used on an automatic variable.
13224 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
13226 PR rtl-optimization/80474
13227 * reorg.c (update_block): Do not ignore instructions in a delay slot.
13229 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
13231 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
13234 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
13236 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
13237 (casesi): Emit bounds checking as RTL.
13238 (casesi_internal_mips16_<mode>): Remove bounds checking.
13240 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
13242 * config/xtensa/xtensa.c (xtensa_option_override): Append
13243 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
13244 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
13245 xtensa_doloop_hooks): Define unconditionally.
13246 (xtensa_reorg_loops): Only call reorg_loops in the presence of
13248 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
13249 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
13250 for it in xtensa_option_override.
13251 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
13252 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
13254 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
13256 * doc/cppopts.texi: Document '-' special value to -MF.
13258 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
13260 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
13261 (cortex_a53_fconst): Likewise.
13262 (cortex_a53_fpmul): Likewise.
13263 (cortex_a53_f_load_64): Likewise.
13264 (cortex_a53_f_load_many): Likewise.
13265 (cortex_a53_advsimd_alu): Likewise.
13266 (cortex_a53_advsimd_alu_q): Likewise.
13267 (cortex_a53_advsimd_mul): Likewise.
13268 (cortex_a53_advsimd_mul_q): Likewise.
13269 (fpmac bypass): Add new bypass for fpmac-fpmac case.
13270 Add missing fmul, r2f_cvt and fconst cases.
13272 2017-06-14 Richard Biener <rguenther@suse.de>
13274 PR middle-end/81088
13275 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
13277 (fold_binary_loc): When associating do not treat pre-existing
13278 TREE_OVERFLOW on literal constants as a reason to allow
13279 TREE_OVERFLOW on associated literal constants.
13281 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
13283 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
13284 (MASK_FEATURES): New macro.
13285 * config/sparc/sparc.c (sparc_option_override): Remove the special
13286 handling of -mfpu and generalize it to all MASK_FEATURES switches.
13288 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
13290 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
13291 a division of 0 if non-call exceptions are enabled.
13293 2017-06-14 Andrew Pinski <apinski@cavium.com>
13294 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
13297 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
13298 Improve vector initialization code gen for only variable case.
13300 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
13302 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
13304 2017-06-14 Richard Biener <rguenther@suse.de>
13306 PR tree-optimization/81083
13307 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
13310 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13312 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
13313 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
13314 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
13315 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
13316 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
13317 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
13319 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13321 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
13322 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
13324 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13326 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
13328 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13330 * config/rs6000/t-rtems: Don't handle SPE.
13332 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13334 * config/rs6000/t-linux: Don't handle SPE.
13336 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13338 * config/rs6000/eabispe.h: Delete file.
13340 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13342 * config/rs6000/t-spe: Delete file.
13344 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13346 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
13347 (rs6000_legitimate_offset_address_p): Return false for anything in
13348 V2SImode or V2SFmode.
13350 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13352 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
13353 except V2SF and V2SI. Rearrange the vector modes, and add comments.
13354 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
13356 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
13357 (rs6000_legitimate_offset_address_p): Ditto.
13358 (rs6000_emit_move): Ditto.
13359 (rs6000_init_builtins): Remove V4HI_type_node.
13361 2017-06-13 Martin Liska <mliska@suse.cz>
13364 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
13365 (gate_asan): Likewise.
13366 * asan.h (asan_no_sanitize_address_p): Remove the function.
13367 (sanitize_flags_p): New function.
13368 * builtins.def: Fix coding style.
13369 * common.opt: Use renamed enum value.
13370 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
13371 * doc/extend.texi: Document no_sanitize attribute.
13372 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
13373 to SANITIZE_UNDEFINED_NONDEFAULT.
13374 * gcc.c (sanitize_spec_function): Use the renamed enum value.
13375 * gimple-fold.c (optimize_atomic_compare_exchange_p):
13376 Use sanitize_flags_p.
13377 * gimplify.c (gimplify_function_tree): Likewise.
13378 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
13379 * opts.c (parse_no_sanitize_attribute): New function.
13380 (common_handle_option): Use renamed enum value.
13381 * opts.h (parse_no_sanitize_attribute): Declare.
13382 * tree.c (sanitize_flags_p): New function.
13383 * tree.h: Declared here.
13384 * tsan.c: Use sanitize_flags_p.
13385 * ubsan.c (ubsan_expand_null_ifn): Likewise.
13386 (instrument_mem_ref): Likewise.
13387 (instrument_bool_enum_load): Likewise.
13388 (do_ubsan_in_current_function): Remove the function.
13389 (pass_ubsan::execute): Use sanitize_flags_p.
13390 * ubsan.h: Remove do_ubsan_in_current_function
13391 * tree-cfg.c (print_no_sanitize_attr_value): New function.
13392 (dump_function_to_file): Use it here.
13394 2017-06-13 Martin Jambor <mjambor@suse.cz>
13396 PR tree-optimization/80803
13397 PR tree-optimization/81063
13398 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
13399 (propagate_subaccesses_across_link): Enqueue subtree whenever
13400 necessary instead of relying on the caller.
13402 2017-06-13 Martin Jambor <mjambor@suse.cz>
13404 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
13405 that have a first_link.
13406 (sort_and_splice_var_accesses): Do not check first_link before
13408 (subtree_mark_written_and_enqueue): Likewise.
13409 (propagate_all_subaccesses): Likewise and do not stop at first
13410 parent with a first_link.
13412 2017-06-13 Martin Jambor <mjambor@suse.cz>
13414 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
13417 2017-06-13 Yury Gribov <tetra2005@gmail.com>
13419 * match.pd: New pattern.
13421 2017-06-13 Yury Gribov <tetra2005@gmail.com>
13423 * tree-vrp.c (is_masked_range_test): New function.
13424 (register_edge_assert_for): Determine ranges for
13427 2017-06-13 Yury Gribov <tetra2005@gmail.com>
13429 PR tree-optimization/67328
13430 * fold-const.c (maskable_range_p): New function.
13431 (build_range_check): Generate bittests if possible.
13433 2017-06-13 Martin Liska <mliska@suse.cz>
13435 * gimple-pretty-print.c (dump_probability): Add new argument.
13436 (dump_edge_probability): Dump both probability and count.
13437 (dump_gimple_label): Likewise.
13438 (dump_gimple_bb_header): Likewise.
13440 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
13443 * config/avr/avr-devices.c: Fix indentation.
13444 * config/avr/gen-avr-mmcu-specs.c: Dito.
13446 2017-06-13 Richard Biener <rguenther@suse.de>
13448 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
13449 instead get vector type from stmt_info.
13450 (vectorizable_reduction): Adjust. Remove dead code.
13452 2017-06-13 Richard Biener <rguenther@suse.de>
13454 PR middle-end/81065
13455 * fold-const.c (extract_muldiv_1): Remove bogus distribution
13456 case of C * (x * C2 + C3).
13457 (fold_addr_of_array_ref_difference): Properly fold index difference.
13459 2017-06-12 David S. Miller <davem@davemloft.net>
13462 * config/sparc/sparc.md (return expander): Emit frame blockage if
13463 function uses alloca.
13465 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
13467 * combine.c (make_field_assignment): Check len rather than the mode
13468 precision when calling force_to_mode.
13470 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
13472 Support multilibs and devices that see flash in RAM address range.
13475 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
13476 (avr_mcu_t) <flash_pm_offset>: New field.
13477 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
13478 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
13479 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
13480 (AVR_TINY_PM_OFFSET): Remove macro.
13481 * config/avr/avr.opt (-mshort-calls): New option.
13482 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
13483 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
13484 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
13485 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
13486 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
13487 instead of avr_arch->have_jmp_call.
13488 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
13489 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
13490 avr_arch->flash_pm_offset to define.
13491 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
13492 new field flash_pm_offset. Add entry for avrxmega3.
13493 (avr_texinfo): Add entry for avrxmega3.
13494 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
13495 attiny212, attiny214,
13496 attiny412, attiny414, attiny416, attiny417,
13497 attiny814, attiny816, attiny817,
13498 attiny1614, attiny1616, attiny1617,
13499 attiny3214, attiny3216, attiny3217.
13500 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
13501 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
13502 (avr_print_operand_address) [AVR_TINY]: Same.
13503 (avr_asm_init_sections) <readonly_data_section>: Only patch
13504 callback if avr_arch->flash_pm_offset = 0.
13505 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
13506 for rodata if avr_arch->flash_pm_offset != 0.
13507 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
13508 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
13509 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
13510 (m_options): Append opt_rcall.
13511 (m_dirnames): Append dir_rcall.
13512 * config/avr/t-multilib: Regenerate.
13514 * configure.ac [target=avr]: Check whether avrxmega3 default
13515 linker description file works as needed.
13516 * configure: Regenerate.
13517 * doc/avr-mmcu.texi: Regenerate.
13518 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
13519 <__AVR_ARCH__>: Document avrxmega3 and 103.
13520 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
13521 <__AVR_SHORT_CALLS__>: Document it.
13522 <__AVR_PM_BASE_ADDRESS__>: Document it.
13523 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
13524 (AVR Variable Attributes) <progmem>: Document this is
13525 not needed for avrxmega3.
13526 (AVR Named Address Spaces) <__flash>: Dito.
13528 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
13530 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
13532 2017-06-12 Doug Rupp <rupp@adacore.com>
13534 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
13535 Append vxworks-stdint.h to the tm_file list.
13536 * config/vxworks-stdint.h: New file.
13538 2017-06-12 Martin Liska <mliska@suse.cz>
13540 PR tree-optimization/81041
13541 * tree-profile.c (gimple_gen_ic_func_profiler):
13542 Create an extra BB in profile-generate
13543 (gimple_gen_time_profiler): Likewise.
13545 2017-06-12 Jakub Jelinek <jakub@redhat.com>
13547 PR tree-optimization/81003
13548 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
13549 (update_range_test): Use it instead of force_gimple_operand_gsi.
13551 2017-06-12 Richard Biener <rguenther@suse.de>
13553 PR tree-optimization/81053
13554 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
13555 with backedge value not defined in loop. Simplify def stmt
13558 2017-06-11 Tom de Vries <tom@codesourcery.com>
13561 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
13563 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
13564 nvptx_cannot_force_const_mem.
13566 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
13568 * opts.c (finish_options): Move test for flag_split_stack after
13569 it has been initialized.
13571 2017-06-11 Jason Merrill <jason@redhat.com>
13573 * tree.h (id_equal): New.
13574 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
13575 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
13576 instead of strcmp of IDENTIFIER_POINTER.
13578 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
13580 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
13581 (mark_all_inlined_calls_cdtor): Fix formating.
13582 (inline_transform): Rescale profile before inlining.
13584 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
13586 * cgraph.h (cgraph_edge::clone): Update prototype.
13587 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
13588 (cgraph_node::create_clone): Update.
13589 (cgraph_node::create_version_clone): Update.
13590 * tree-inline.c (copy_bb): Update.
13591 (expand_call_inline): Update.
13593 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
13595 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
13596 factored out from ...
13597 (rs6000_emit_prologue): ... here.
13599 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
13601 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
13602 factored out from ...
13603 (rs6000_emit_prologue): ... here.
13605 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
13607 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
13609 (handle_missing_profiles): Fix computation of tp_first_run.
13610 (counts_to_freqs): Do not touch freqs when count is 0.
13612 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
13614 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
13617 2017-06-10 Tom de Vries <tom@codesourcery.com>
13619 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
13620 attributes): Document signal effective target.
13622 2017-06-10 Tom de Vries <tom@codesourcery.com>
13624 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
13625 Document effective target stack_size.
13627 2017-06-09 David Malcolm <dmalcolm@redhat.com>
13629 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
13630 to the edit_context if they can be auto-applied.
13632 2017-06-9 Ian Lance Taylor <iant@golang.org>
13634 * opts.c (finish_options): If -fsplit-stack, disable implicit
13635 -forder-blocks-and-partition.
13636 * doc/invoke.texi (Optimize Options): Document that when using
13637 -fsplit-stack -forder-blocks-and-partition is not implicitly
13640 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
13642 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
13643 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
13644 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
13645 * builtins.def (abort, trap, unreachable): Declare cold.
13646 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
13647 * tree-core.h (ECF_COLD): New.
13648 * tree.c (set_call_expr_flags): Handle ECF_COLD.
13649 (build_common_builtin_nodes): Mark unreachable and abort as cold.
13651 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
13653 * predict.c (unlikely_executed_stmt_p): Cleanup.
13655 2017-06-09 Richard Biener <rguenther@suse.de>
13657 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
13658 model if the ref is always written to.
13660 2017-06-09 Tamar Christina <tamar.christina@arm.com>
13662 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
13664 2017-06-09 Tamar Christina <tamar.christina@arm.com>
13666 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
13669 2017-06-09 Tom de Vries <tom@codesourcery.com>
13672 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
13673 "target cannot support label values" when encountering LABEL_REF.
13675 2017-06-09 Martin Liska <mliska@suse.cz>
13677 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
13678 (gimple_gen_ic_func_profiler): Emit direct comparison
13679 of __gcov_indirect_call_callee with NULL.
13680 (gimple_gen_time_profiler): Change probability from
13681 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
13683 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
13685 * profile.c (edge_gcov_counts): Turn to pointer.
13686 (compute_branch_probabilities, compute_branch_probabilities): Update.
13687 (branch_prob): Do not clear edge_gcov_count.
13688 * profile.h (edge_gcov_counts): Turn to pointer.
13689 (edge_gcov_count): Update.
13691 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
13693 * gimple.h (gimple_check_failed): Mark cold.
13695 2017-06-09 Richard Biener <rguenther@suse.de>
13697 PR tree-optimization/66623
13698 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
13699 refactor check_reduction into two parts, properly computing
13700 whether we have to check reduction validity for outer loop
13703 2017-06-09 Richard Biener <rguenther@suse.de>
13705 PR tree-optimization/79483
13706 * graphite-scop-detection.c (order): New global.
13707 (get_order): Compute bb to order mapping that satisfies code
13708 generation constraints.
13709 (cmp_pbbs): New helper.
13710 (build_scops): Start domwalk at entry block, sort generated
13713 2017-06-09 Richard Biener <rguenther@suse.de>
13715 PR middle-end/81007
13716 * ipa-polymorphic-call.c
13717 (ipa_polymorphic_call_context::restrict_to_inner_class):
13718 Skip FIELD_DECLs with error_mark_node type.
13719 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
13722 2017-06-09 Martin Liska <mliska@suse.cz>
13724 * predict.c (struct branch_predictor): New struct.
13725 (test_prediction_value_range): New test.
13726 (predict_c_tests): New function.
13727 * selftest-run-tests.c (selftest::run_tests): Run the function.
13728 * selftest.h: Declare new tests.
13730 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
13733 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
13734 gen_add3_insn did not fail.
13735 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
13736 r0, construct that number in a temporary reg and add that reg to r0.
13737 If asked to put the result in r0 as well, fail.
13739 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
13741 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13742 for early expansion of vec_eqv.
13744 2017-06-08 Jakub Jelinek <jakub@redhat.com>
13746 PR middle-end/81005
13747 * ubsan.c (instrument_null): Avoid pointless code temporary.
13748 (pass_ubsan::execute): Instrument aggregate arguments of calls.
13750 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
13754 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
13757 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
13760 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13762 * predict.c (unlikely_executed_edge_p): Move ahead.
13763 (probably_never_executed_edge_p): Use it.
13765 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
13767 PR middle-end/79988
13768 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
13769 gimple_call_builtin_p call.
13771 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13773 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
13774 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
13775 rtl_check_failed_type2, rtl_check_failed_code1,
13776 rtl_check_failed_code2, rtl_check_failed_code_mode,
13777 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
13778 rtvec_check_failed_bounds, rtl_check_failed_flag,
13779 _fatal_insn_not_found, _fatal_insn): Likewise.
13780 * tree.h (tree_contains_struct_check_failed,
13781 tree_check_failed, tree_not_check_failed,
13782 tree_class_check_failed, tree_range_check_failed,
13783 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
13784 tree_vec_elt_check_failed, phi_node_elt_check_failed,
13785 tree_operand_check_failed, omp_clause_check_failed,
13786 omp_clause_operand_check_failed, omp_clause_range_check_failed):
13789 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13791 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
13792 flag_branch_probabilities.
13793 * ipa-inline.c (edge_badness): Likewise.
13794 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
13795 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
13796 * predict.c (maybe_hot_frequency_p): Likewise.
13797 (probably_never_executed): Likewise.
13798 * sched-ebb.c (schedule_ebbs): Likewise.
13799 * sched-rgn.c (find_single_block_region): Likewise.
13800 * tracer.c (tail_duplicate): Likewise.
13802 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13804 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
13805 longer requires x_flag_profile_use.
13807 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13809 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
13810 instead of flag_reorder_blocks_and_partition.
13811 * dbxout.c (dbxout_function_end): Likewise.
13812 * dwarf2out.c (gen_subprogram_die): Likewise.
13813 * haifa-sched.c (sched_create_recovery_edges): Likewise.
13814 * hw-doloop.c (reorg_loops): Likewise.
13815 * varasm.c (assemble_start_function,
13816 assemble_end_function): Likewise.
13817 (decide_function_section): Do not check for
13818 flag_reorder_blocks_and_partition.
13820 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
13822 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
13824 (chkp_get_hard_register_fake_addr_expr): Ditto.
13825 (chkp_build_addr_expr): Add check for hard reg case.
13826 (chkp_parse_array_and_component_ref): Ditto.
13827 (chkp_find_bounds_1): Ditto.
13828 (chkp_process_stmt): Don't generate bounds store for
13831 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13833 * predict.c (maybe_hot_bb_p): Do not check profile status.
13834 (maybe_hot_edge_p): Likewise.
13835 (probably_never_executed): Check for zero counts even if profile
13837 (unlikely_executed_edge_p): New function.
13838 (unlikely_executed_stmt_p): New function.
13839 (unlikely_executed_bb_p): New function.
13840 (set_even_probabilities): Use unlikely predicates.
13841 (combine_predictions_for_bb): Likewise.
13842 (predict_paths_for_bb): Likewise.
13843 (predict_paths_leading_to_edge): Likewise.
13844 (determine_unlikely_bbs): New function.
13845 (estimate_bb_frequencies): Use it.
13846 (compute_function_frequency): Use zero counts even if profile is
13848 * profile-count.h: Fix typo.
13850 2017-08-08 Julia Koval <julia.koval@intel.com>
13852 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
13853 _mm512_mask_cvtsepi16_storeu_epi8,
13854 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
13855 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
13856 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
13857 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
13858 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
13859 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
13860 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
13861 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
13862 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
13863 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
13864 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
13865 __builtin_ia32_pmovuswb256mem_mask,
13866 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
13867 __builtin_ia32_pmovwb512mem_mask): New builtins.
13869 2017-08-08 Julia Koval <julia.koval@intel.com>
13871 PR target/73350,80862
13872 * config/i386/subst.md (round): Fix round pattern.
13873 * config/i386/i386.c (ix86_erase_embedded_rounding):
13874 Fix erasing rounding for the fixed pattern.
13876 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13878 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
13880 2017-06-08 Martin Liska <mliska@suse.cz>
13882 PR gcov-profile/80911
13883 * gcov.c (block_info::block_info): New constructor.
13885 2017-06-07 Carl Love <cel@us.ibm.com>
13887 * config/rs6000/rs6000-c: The return type of the following
13888 built-in functions was implemented as int not long long. Fix sign
13889 of return value for the unsigned version of vec_mulo and vec_mule.
13890 vector unsigned long long vec_bperm (vector unsigned long long,
13891 vector unsigned char)
13892 vector signed long long vec_mule (vector signed int,
13894 vector unsigned long long vec_mule (vector unsigned int,
13895 vector unsigned int)
13896 vector signed long long vec_mulo (vector signed int,
13898 vector unsigned long long vec_mulo (vector unsigned int,
13899 vector unsigned int)
13900 * doc/extend.texi: Fix the documentation for the built-in
13903 2017-06-07 Carl Love <cel@us.ibm.com>
13906 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
13909 2017-06-07 Carl Love <cel@us.ibm.com>
13911 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
13912 support, Generate doublehv for signed int/float for BE case only.
13914 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
13916 * doc/invoke.texi (mcx16): Rewrite.
13918 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13920 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
13921 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
13922 *mov<mode>_softfloat, and an anonymous splitter): Use
13923 nonimmediate_operand instead of rs6000_nonimmediate_operand.
13925 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13927 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
13929 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
13930 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
13931 (rs6000_debug_reg_global): Adjust.
13932 (rs6000_init_hard_regno_mode_ok): Adjust.
13933 (rs6000_dbx_register_number): Adjust.
13934 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
13935 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
13936 Remove SPE_ACC and SPEFSCR.
13937 (REG_ALLOC_ORDER): Ditto.
13938 (FRAME_POINTER_REGNUM): Change to 111.
13939 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
13940 (REG_CLASS_NAMES): Ditto.
13941 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
13942 (REGISTER_NAMES): Ditto.
13943 (ADDITIONAL_REG_NAMES): Ditto.
13944 (rs6000_reg_names): Ditto.
13945 * config/rs6000/rs6000.md: Renumber some register number
13948 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13950 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
13952 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
13953 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
13955 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
13956 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
13957 Delete the SPE high registers.
13958 (REG_ALLOC_ORDER): Ditto.
13959 (enum reg_class): Remove SPE_HIGH_REGS.
13960 (REG_CLASS_NAMES): Ditto.
13961 (REG_CLASS_CONTENTS): Delete the SPE high registers.
13962 (REGISTER_NAMES): Ditto.
13963 (rs6000_reg_names): Ditto.
13964 * doc/tm.texi.in: Remove SPE as example.
13965 * doc/tm.texi: Regenerate.
13967 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13969 * config/rs6000/8540.md (ppc8540_brinc): Delete.
13970 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
13971 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
13972 * config/rs6000/rs6000.md (type): Remove "brinc".
13974 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13976 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
13977 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
13978 * config/rs6000/linuxspe.h: Delete file.
13979 * config/rs6000/rs6000.md: Don't include spe.md.
13980 * config/rs6000/spe.h: Delete file.
13981 * config/rs6000/spe.md: Delete file.
13982 * config/rs6000/t-rs6000: Remove spe.md.
13984 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13986 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
13987 (reg_or_none500mem_operand): Delete.
13988 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
13989 instead of reg_or_none500mem_operand.
13991 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13993 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
13994 handling of SPE flags.
13995 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
13997 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13999 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
14001 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
14002 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
14003 paired_divv2sf3): Similar.
14004 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
14005 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
14006 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
14008 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
14009 Rename the paired_* instruction patterns.
14010 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
14012 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
14013 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
14014 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
14015 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
14016 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
14017 PAIRED_VECTOR_MODE.
14018 (struct machine_function): Delete field spe_insn_chain_scanned_p.
14019 (spe_func_has_64bit_regs_p): Delete.
14020 (spe_expand_predicate_builtin): Delete.
14021 (spe_expand_evsel_builtin): Delete.
14022 (TARGET_DWARF_REGISTER_SPAN): Do not define.
14023 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
14024 (invalid_e500_subreg): Delete.
14025 (rs6000_legitimize_address): Always force_reg op2 as well, for
14026 paired single memory accesses.
14027 (rs6000_member_type_forces_blk): Delete.
14028 (rs6000_spe_function_arg): Delete.
14029 (rs6000_expand_unop_builtin): Delete SPE handling.
14030 (rs6000_expand_binop_builtin): Ditto.
14031 (spe_expand_stv_builtin): Delete.
14032 (bdesc_2arg_spe): Delete.
14033 (spe_expand_builtin): Delete.
14034 (spe_expand_predicate_builtin): Delete.
14035 (spe_expand_evsel_builtin): Delete.
14036 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
14037 (spe_init_builtins): Delete.
14038 (spe_func_has_64bit_regs_p): Delete.
14039 (savres_routine_name): Delete "info" parameter. Adjust callers.
14040 (rs6000_emit_stack_reset): Ditto.
14041 (rs6000_dwarf_register_span): Delete.
14042 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
14043 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
14044 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
14046 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
14048 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
14049 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
14050 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
14051 mulv2sf3, divv2sf3): Delete expanders.
14053 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14055 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
14057 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14059 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
14060 * config/rs6000/rs6000.c: Ditto.
14062 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14064 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
14065 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
14066 comparison_operator.
14068 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14070 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
14071 * config/rs6000/rs6000.opt: Ditto.
14072 * config/rs6000/t-rtems: Ditto.
14074 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14076 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
14077 TARGET_E500_SINGLE by 0, simplify.
14078 * config/rs6000/rs6000.c: Ditto.
14079 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
14080 (spe_build_register_parallel): Delete.
14081 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
14082 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
14083 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
14084 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
14085 (E500_CONVERT): Delete.
14086 * config/rs6000/spe.md: Remove many patterns and all define_constants.
14088 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14090 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
14091 * config/rs6000/dfp.md: Ditto.
14092 (negdd2, *negdd2_fpr): Merge.
14093 (absdd2, *absdd2_fpr): Merge.
14094 (negtd2, *negtd2_fpr): Merge.
14095 (abstd2, *abstd2_fpr): Merge.
14096 * config/rs6000/e500.h: Delete file.
14097 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
14098 TARGET_FPRS by 1 and simplify.
14099 * config/rs6000/rs6000-c.c: Ditto.
14100 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
14101 TARGET_DF_SPE by 0.
14102 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
14104 * config/rs6000/rs6000.md: Ditto.
14105 (floatdidf2, *floatdidf2_fpr): Merge.
14106 (move_from_CR_gt_bit): Delete.
14107 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
14108 (E500_CR_IOR_COMPARE): Delete.
14109 (All patterns that require !TARGET_FPRS): Delete.
14110 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
14112 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14114 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
14116 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14118 * graphds.c (add_edge): Intitialize edge's attached data.
14119 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
14120 pointer parameter. Call pointed function on each edge during
14121 graph traversing. Skip traversing the edge when the function
14123 (graphds_dfs, graphds_scc): Ditto.
14124 (for_each_edge): New parameter. Pass the new parameter to callback
14126 * graphds.h (skip_edge_callback): New function pointer type.
14127 (graphds_dfs, graphds_scc): New function pointer parameter.
14128 (graphds_edge_callback, for_each_edge): New parameter.
14130 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14132 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
14133 out code checking if runtime alias check is possible to below ...
14134 Call the new function.
14135 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
14136 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
14138 2017-06-07 Marek Polacek <polacek@redhat.com>
14141 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
14142 TYPE_OVERFLOW_WRAPS check.
14144 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14146 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
14147 if versioning is required.
14148 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
14149 peeling with the check for versioning.
14151 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14153 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
14154 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
14155 Set true to new parameter if new ssa variable is defined.
14156 (vect_gen_vector_loop_niters): Refactor. Set range information
14157 for the new vector loop bound variable.
14158 (vect_do_peeling): Ditto.
14160 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14162 * tree-affine.c (ssa.h): Include header file.
14163 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
14164 has wrapping overflow behavior.
14166 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14168 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
14170 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14172 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
14173 (tree_to_aff_combination): ... here.
14175 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14177 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
14178 reg_pressure model function.
14179 (ivopts_global_cost_for_size): Delete.
14180 (determine_set_costs, iv_ca_recount_cost): Call new model function
14181 ivopts_estimate_reg_pressure.
14183 2017-06-07 Tamar Christina <tamar.christina@arm.com>
14185 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
14186 expensive than udiv. Remove floating point cases from mod.
14188 2017-06-07 Tamar Christina <tamar.christina@arm.com>
14190 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
14191 Increase idiv cost.
14193 2017-06-07 Tamar Christina <tamar.christina@arm.com>
14195 * config/aarch64/aarch64.md
14196 (copysignsf3): Fix mask generation.
14198 2017-06-07 Jakub Jelinek <jakub@redhat.com>
14200 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
14202 (class dump_manager): Add register_dumps method.
14203 * dumpfile.c: Include langhooks.h.
14204 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
14205 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
14206 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
14207 (dump_manager::dump_register): Start with 512 entries instead of 32.
14208 (dump_manager::register_dumps): New method.
14209 * toplev.c (general_init): Instead of invoking register_dumps
14210 langhook, invoke register_dumps method on the dump manager.
14211 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
14214 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
14216 * doc/md.texi: Clarify the restrictions on a define_insn condition.
14217 Say that # requires an associated define_split to exist, and that
14218 the define_split must be suitable for use after register allocation.
14220 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14222 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
14223 (compute_outgoing_frequencies): Also initialize zero counts.
14224 (find_many_sub_basic_blocks): Do not produce uninitialized profile
14225 around loops; preserve more of profile when nothing changes.
14227 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
14229 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
14231 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
14232 * config/arm/arm-cpu-cdata.h: Regenerate.
14233 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
14234 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
14235 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
14236 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
14237 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
14239 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
14241 * config/arm/t-rmprofile: Likewise.
14242 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
14244 2017-06-06 David S. Miller <davem@davemloft.net>
14247 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
14248 blockage if function uses alloca.
14250 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
14252 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
14253 New "uid" fields to hold pretty-print IDs of group and ref.
14254 Memory references are now identified as <group_id>:<ref_id>
14255 instead of using [random] addresses.
14256 (dump_mem_details): Simplify, no functional change.
14257 (dump_mem_ref): Simplify and make output more concise.
14258 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
14259 (find_or_create_group): Initialize group uid.
14260 (record_ref): Initialize ref uid. Improve debug output.
14261 (prune_group_by_reuse, should_issue_prefetch_p,)
14262 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
14263 (mark_nontemporal_store, determine_loop_nest_reuse):
14264 Improve debug output.
14266 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
14268 * dbgcnt.def (prefetch): New debug counter.
14269 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
14270 (schedule_prefetches): Stop issueing prefetches if debug counter
14273 2017-06-06 Tom de Vries <tom@codesourcery.com>
14275 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
14276 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
14278 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14280 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
14281 Use aarch64_reg_or_zero predicate for operand 4.
14282 (aarch64_compare_and_swap<mode> define_insn_and_split):
14283 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
14284 (aarch64_store_exclusive<mode>): Likewise for operand 2.
14286 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
14288 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
14289 (arm_compute_save_core_reg_mask): This.
14290 (thumb1_compute_save_reg_mask): Rename into ...
14291 (thumb1_compute_save_core_reg_mask): This.
14292 (arm_compute_save_reg0_reg12_mask): Adapt comment.
14293 (arm_compute_frame_layout): Likewise.
14295 2017-06-06 Richard Biener <rguenther@suse.de>
14297 PR tree-optimization/80974
14298 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
14299 keep or clear leaders SSA info.
14301 2017-06-06 Tom de Vries <tom@codesourcery.com>
14303 * config/nvptx/nvptx.c (split_mode_p): New function.
14304 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
14306 2017-06-06 Tom de Vries <tom@codesourcery.com>
14308 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
14310 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14313 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
14316 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14318 * shrink-wrap.c (handle_simple_exit): Update profile.
14319 (try_shrink_wrapping): Upate profile.
14321 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14323 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
14324 (tree_guess_outgoing_edge_probabilities): New.
14325 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
14326 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
14328 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14330 * ipa-split.c (split_function): Initialize return bb profile.
14332 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14334 * profile.c (compute_branch_probabilities): Also initialize
14335 EXIT_BLOCK profile.
14337 2017-06-06 Richard Biener <rguenther@suse.de>
14339 PR tree-optimization/80928
14340 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
14341 (vect_analyze_loop_operations): Properly guard analysis for
14343 (vect_transform_loop): Likewise.
14344 (vect_analyze_loop_2): Also reset SLP type on PHIs.
14345 (vect_model_induction_cost): Do not cost for pure SLP.
14346 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
14347 of induction in inner loop vectorization.
14348 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
14349 (vect_get_and_check_slp_defs): Handle vect_induction_def.
14350 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
14352 (vect_analyze_slp_cost_1): Cost induction.
14353 (vect_detect_hybrid_slp_stmts): Handle PHIs.
14354 (vect_get_slp_vect_defs): Likewise.
14355 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
14356 (vect_transform_stmt): Handle SLP reductions.
14357 * tree-vectorizer.h (vectorizable_induction): Adjust.
14359 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
14361 * config/rs6000/rs6000.c (make_resolver_func): Update
14362 init_lowered_empty_function call.
14364 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
14366 * doc/invoke.texi: Document the -fprofile-abs-path option.
14367 * common.opt (fprofile-abs-path): New option.
14368 * gcov-io.h (gcov_write_filename): Declare.
14369 * gcov-io.c (gcov_write_filename): New function.
14370 * coverage.c (coverage_begin_function): Use gcov_write_filename.
14371 * profile.c (output_location): Likewise.
14373 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
14375 * shring-wrap.c: Revert accidental commit.
14377 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
14379 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
14381 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
14383 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
14385 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
14386 profile in callgraph edge.
14387 * profile-count.h (apply_probability): If THIS is 0, then result is 0
14388 (apply_scale): Likewise.
14389 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
14390 Also scale profile when inlining function with zero profile.
14391 (initialize_cfun): Update exit block profile even when it is zero.
14392 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
14393 when profile is read.
14395 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
14397 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
14398 (CLONE_*): New constants to define the processors we can generate
14399 code for with the target_clone attribute.
14400 (rs6000_clone_map): New array to identify which clone processors
14401 the current program is running on.
14402 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
14403 target_clone attribute.
14404 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
14405 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
14406 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
14407 (cpu_expand_builtin): Add support for target_clone attribute.
14408 (rs6000_valid_attribute_p): Allow "default" attribute.
14409 (get_decl_name): New debug function to simplify printing the
14410 current function name in debugging statements.
14411 (rs6000_clone_priority): New functions to support the target_clone
14412 attribute, and be able to generate code to switch between ISA 2.05
14413 through ISA 3.0 (power6 through power9).
14414 (rs6000_compare_version_priority): Likewise.
14415 (rs6000_get_function_versions_dispatcher): Likewise.
14416 (make_resolver_func): Likewise.
14417 (add_condition_to_bb): Likewise.
14418 (dispatch_function_versions): Likewise.
14419 (rs6000_generate_version_dispatcher_body): Likewise.
14420 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
14421 (fusion_gpr_load_p): Fix a spacing issue.
14422 * doc/extend.texi (Common Function Attributes): Document that the
14423 PowerPC supports the target_clone attribute.
14425 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
14427 * config/arm/arm.h: explain F symbol found in description of ARM
14428 register allocation in its legend.
14430 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
14432 * config/mips/frame-header-opt.c: Include profile-count.h.
14433 * config/riscv/riscv.c: Include profile-count.h
14435 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
14437 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
14439 (sm_set_flag_if_changed): Add bbs field.
14440 (execute_sm_if_changed_flag_set): Pass BBS.
14441 (execute_sm): Update.
14443 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14445 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
14448 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14450 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
14451 (peephole2): New peephole2 to emit the above.
14452 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
14454 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14456 * config/aarch64/aarch64.c (define_peephole2 above
14457 *sub_<shift>_<mode>): New peephole.
14459 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
14461 * config/i386/i386.c (make_resolver_func): Update.
14462 * Makefile.in: Add profile-count.h and profile-count.o
14463 * auto-profile.c (afdo_indirect_call): Update to new API.
14464 (afdo_set_bb_count): Update.
14465 (afdo_propagate_edge): Update.
14466 (afdo_propagate_circuit): Update.
14467 (afdo_calculate_branch_prob): Update.
14468 (afdo_annotate_cfg): Update.
14469 * basic-block.h: Include profile-count.h
14470 (struct edge_def): Turn count to profile_count.
14471 (struct basic_block_def): Likewie.
14472 (REG_BR_PROB_BASE): Move to profile-count.h
14473 (RDIV): Move to profile-count.h
14474 * bb-reorder.c (max_entry_count): Turn to profile_count.
14475 (find_traces): Update.
14476 (rotate_loop):Update.
14477 (connect_traces):Update.
14478 (sanitize_hot_paths):Update.
14479 * bt-load.c (migrate_btr_defs): Update.
14480 * cfg.c (RDIV): Remove.
14481 (init_flow): Use alloc_block.
14482 (alloc_block): Uninitialize count.
14483 (unchecked_make_edge): Uninitialize count.
14484 (check_bb_profile): Update.
14485 (dump_edge_info): Update.
14486 (dump_bb_info): Update.
14487 (update_bb_profile_for_threading): Update.
14488 (scale_bbs_frequencies_int): Update.
14489 (scale_bbs_frequencies_gcov_type): Update.
14490 (scale_bbs_frequencies_profile_count): New.
14491 * cfg.h (update_bb_profile_for_threading): Update.
14492 (scale_bbs_frequencies_profile_count): Declare.
14493 * cfgbuild.c (compute_outgoing_frequencies): Update.
14494 (find_many_sub_basic_blocks): Update.
14495 * cfgcleanup.c (try_forward_edges): Update.
14496 (try_crossjump_to_edge): Update.
14497 * cfgexpand.c (expand_gimple_tailcall): Update.
14498 (construct_exit_block): Update.
14499 * cfghooks.c (verify_flow_info): Update.
14500 (dump_bb_for_graph): Update.
14501 (split_edge): Update.
14502 (make_forwarder_block): Update.
14503 (duplicate_block): Update.
14504 (account_profile_record): Update.
14505 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
14506 (get_estimated_loop_iterations): Update.
14507 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
14508 (single_likely_exit): Update.
14509 * cfgloopmanip.c (scale_loop_profile): Update.
14511 (set_zero_probability): Update.
14512 (lv_adjust_loop_entry_edge): Update.
14513 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
14514 (purge_dead_edges): Update.
14515 (rtl_account_profile_record): Update.
14516 * cgraph.c (cgraph_node::create): Uninitialize count.
14517 (symbol_table::create_edge): Uninitialize count.
14518 (cgraph_update_edges_for_call_stmt_node): Update.
14519 (cgraph_edge::dump_edge_flags): Update.
14520 (cgraph_node::dump): Update.
14521 (cgraph_edge::maybe_hot_p): Update.
14522 * cgraph.h: Include profile-count.h
14523 (create_clone), create_edge, create_indirect_edge): Update.
14524 (cgraph_node): Turn count to profile_count.
14525 (cgraph_edge0: Likewise.
14526 (make_speculative, clone): Update.
14527 (create_edge): Update.
14528 (init_lowered_empty_function): Update.
14529 * cgraphclones.c (cgraph_edge::clone): Update.
14530 (duplicate_thunk_for_node): Update.
14531 (cgraph_node::create_clone): Update.
14532 * cgraphunit.c (cgraph_node::analyze): Update.
14533 (cgraph_node::expand_thunk): Update.
14534 * final.c (dump_basic_block_info): Update.
14535 * gimple-streamer-in.c (input_bb): Update.
14536 * gimple-streamer-out.c (output_bb): Update.
14537 * graphite.c (print_global_statistics): Update.
14538 (print_graphite_scop_statistics): Update.
14539 * hsa-brig.c: Include basic-block.h.
14540 * hsa-dump.c: Include basic-block.h.
14541 * hsa-gen.c (T sum_slice): Update.
14542 (convert_switch_statements):Update.
14543 * hsa-regalloc.c: Include basic-block.h.
14544 * ipa-chkp.c (chkp_produce_thunks): Update.
14545 * ipa-cp.c (struct caller_statistics): Update.
14546 (init_caller_stats): Update.
14547 (gather_caller_stats): Update.
14548 (ipcp_cloning_candidate_p): Update.
14549 (good_cloning_opportunity_p): Update.
14550 (get_info_about_necessary_edges): Update.
14551 (dump_profile_updates): Update.
14552 (update_profiling_info): Update.
14553 (update_specialized_profile): Update.
14554 (perhaps_add_new_callers): Update.
14555 (decide_about_value): Update.
14556 (ipa_cp_c_finalize): Update.
14557 * ipa-devirt.c (struct odr_type_warn_count): Update.
14558 (struct decl_warn_count): Update.
14559 (struct final_warning_record): Update.
14560 (possible_polymorphic_call_targets): Update.
14561 (ipa_devirt): Update.
14562 * ipa-fnsummary.c (redirect_to_unreachable): Update.
14563 * ipa-icf.c (sem_function::merge): Update.
14564 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
14565 * ipa-inline.c (compute_uninlined_call_time): Update.
14566 (compute_inlined_call_time): Update.
14567 (want_inline_small_function_p): Update.
14568 (want_inline_self_recursive_call_p): Update.
14569 (edge_badness): Update.
14570 (lookup_recursive_calls): Update.
14571 (recursive_inlining): Update.
14572 (inline_small_functions): Update.
14573 (dump_overall_stats): Update.
14574 (dump_inline_stats): Update.
14575 * ipa-profile.c (ipa_profile_generate_summary): Update.
14576 (ipa_propagate_frequency): Update.
14577 (ipa_profile): Update.
14578 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
14579 * ipa-utils.c (ipa_merge_profiles): Update.
14580 * loop-doloop.c (doloop_modify): Update.
14581 * loop-unroll.c (report_unroll): Update.
14582 (unroll_loop_runtime_iterations): Update.
14583 * lto-cgraph.c (lto_output_edge): Update.
14584 (lto_output_node): Update.
14585 (input_node): Update.
14586 (input_edge): Update.
14587 (merge_profile_summaries): Update.
14588 * lto-streamer-in.c (input_cfg): Update.
14589 * lto-streamer-out.c (output_cfg): Update.
14590 * mcf.c (create_fixup_graph): Update.
14591 (adjust_cfg_counts): Update.
14592 (sum_edge_counts): Update.
14593 * modulo-sched.c (sms_schedule): Update.
14594 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
14595 * predict.c (maybe_hot_count_p): Update.
14596 (probably_never_executed): Update.
14597 (dump_prediction): Update.
14598 (combine_predictions_for_bb): Update.
14599 (propagate_freq): Update.
14600 (handle_missing_profiles): Update.
14601 (counts_to_freqs): Update.
14602 (rebuild_frequencies): Update.
14603 (force_edge_cold): Update.
14604 * predict.h: Include profile-count.h
14605 (maybe_hot_count_p, counts_to_freqs): UPdate.
14606 * print-rtl-function.c: Do not include cfg.h
14607 * print-rtl.c: Include basic-block.h
14608 * profile-count.c: New file.
14609 * profile-count.h: New file.
14610 * profile.c (is_edge_inconsistent): Update.
14611 (correct_negative_edge_counts): Update.
14612 (is_inconsistent): Update.
14613 (set_bb_counts): Update.
14614 (read_profile_edge_counts): Update.
14615 (compute_frequency_overlap): Update.
14616 (compute_branch_probabilities): Update; Initialize and deinitialize
14618 (branch_prob): Update.
14619 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
14620 (edge_gcov_count): New.
14621 (bb_gcov_count): New.
14622 * shrink-wrap.c (try_shrink_wrapping): Update.
14623 * tracer.c (better_p): Update.
14624 * trans-mem.c (expand_transaction): Update.
14625 (ipa_tm_insert_irr_call): Update.
14626 (ipa_tm_insert_gettmclone_call): Update.
14627 * tree-call-cdce.c: Update.
14628 * tree-cfg.c (gimple_duplicate_sese_region): Update.
14629 (gimple_duplicate_sese_tail): Update.
14630 (gimple_account_profile_record): Update.
14631 (execute_fixup_cfg): Update.
14632 * tree-inline.c (copy_bb): Update.
14633 (copy_edges_for_bb): Update.
14634 (initialize_cfun): Update.
14635 (freqs_to_counts): Update.
14636 (copy_cfg_body): Update.
14637 (expand_call_inline): Update.
14638 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
14639 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
14640 (try_unroll_loop_completely): Update.
14641 (try_peel_loop): Update.
14642 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
14643 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
14644 * tree-ssa-loop-split.c (connect_loops): Update.
14645 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
14646 * tree-ssa-reassoc.c (branch_fixup): Update.
14647 * tree-ssa-tail-merge.c (replace_block_by): Update.
14648 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
14649 (compute_path_counts): Update.
14650 (update_profile): Update.
14651 (recompute_probabilities): Update.
14652 (update_joiner_offpath_counts): Update.
14653 (estimated_freqs_path): Update.
14654 (freqs_to_counts_path): Update.
14655 (clear_counts_path): Update.
14656 (ssa_fix_duplicate_block_edges): Update.
14657 (duplicate_thread_path): Update.
14658 * tree-switch-conversion.c (case_bit_test_cmp): Update.
14659 (struct switch_conv_info): Update.
14660 * tree-tailcall.c (decrease_profile): Update.
14661 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
14662 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
14663 * value-prof.c (check_counter): Update.
14664 (gimple_divmod_fixed_value): Update.
14665 (gimple_mod_pow2): Update.
14666 (gimple_mod_subtract): Update.
14667 (gimple_ic_transform): Update.
14668 (gimple_stringop_fixed_value): Update.
14669 * value-prof.h (gimple_ic): Update.
14671 2017-06-02 Carl Love <cel@us.ibm.com>
14673 * config/rs6000/rs6000-c: Add support for built-in functions
14674 vector double vec_doublee (vector signed int);
14675 vector double vec_doublee (vector unsigned int);
14676 vector double vec_doublee (vector float);
14677 vector double vec_doubleh (vector signed int);
14678 vector double vec_doubleh (vector unsigned int);
14679 vector double vec_doubleh (vector float);
14680 vector double vec_doublel (vector signed int);
14681 vector double vec_doublel (vector unsigned int);
14682 vector double vec_doublel (vector float);
14683 vector double vec_doubleo (vector signed int);
14684 vector double vec_doubleo (vector unsigned int);
14685 vector double vec_doubleo (vector float);.
14686 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
14687 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
14689 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
14690 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
14691 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
14693 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
14694 vec_doublel, vec_doubleh.
14695 * doc/extend.texi: Update the built-in documentation file for the
14696 new built-in functions.
14698 2017-06-02 David Malcolm <dmalcolm@redhat.com>
14701 * ipa-inline-analysis.c (free_growth_caches): Set
14702 edge_removal_hook_holder to NULL after removing it.
14704 2017-06-02 Sudakshina Das <sudi.das@arm.com>
14706 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
14707 comparision with zero.
14709 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
14710 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14711 for early expansion of vec_min and vec_max builtins.
14712 (builtin_function_type): Add min/max unsigned variants to those
14713 identified as having unsigned arguments.
14715 2017-06-02 Olivier Hainque <hainque@adacore.com>
14717 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
14719 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14721 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
14722 Use VALL_F16 iterator rather than VALL.
14724 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14726 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
14727 Emit CBNZ inside loop when doing a strong exchange and comparing
14728 against zero. Generate the CC flags after the loop.
14730 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
14732 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
14733 (dl_section_ref): New.
14734 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
14735 On AIX, append an expression to subtract the size of the
14736 section length to dl_section_ref.
14738 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
14740 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14741 for early expansion of vector absolute builtins.
14743 2017-06-02 Richard Biener <rguenther@suse.de>
14745 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
14746 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
14748 2017-06-02 Richard Biener <rguenther@suse.de>
14750 PR tree-optimization/80948
14751 * tree-tailcall.c (find_tail_calls): Track stmts to move in
14752 stmt order as well.
14754 2017-06-02 Richard Biener <rguenther@suse.de>
14756 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
14758 * tree-vect-stmts.c (process_use): Do not mark backedge defs
14759 for inductions as relevant.
14761 2017-06-02 Richard Biener <rguenther@suse.de>
14763 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
14764 (vectorizable_induction): ... this. Remove dead code.
14766 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
14768 * builtins. (expand_builtin_alloca): Remove second parameter and
14769 infer its value from the first parameter instead.
14770 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
14772 2017-06-02 Jakub Jelinek <jakub@redhat.com>
14774 PR rtl-optimization/80903
14775 * loop-doloop.c (add_test): Unshare sequence.
14777 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
14779 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
14781 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
14783 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
14785 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
14786 xlogue_layout::get_instance, logue_layout::xlogue_layout,
14787 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
14788 (xlogue_layout::get_stub_rtx): Make static.
14789 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
14790 (xlogue_layout::compute_stub_managed_regs): Rename to...
14791 (xlogue_layout::count_stub_managed_regs): ...this.
14792 (xlogue_layout::is_stub_managed_reg): New function.
14793 (xlogue_layout::m_stub_names): Rename to...
14794 (xlogue_layout::s_stub_names): ...this, make static.
14795 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
14796 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
14797 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
14798 xlogue_layout::s_stub_names): Instantiate statics.
14799 (stub_managed_regs): Remove.
14800 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
14801 (disable_call_ms2sysv_xlogues): Rename to...
14802 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
14803 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
14805 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
14806 change after reload_completed.
14807 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
14809 (ix86_expand_prologue): Likewise.
14810 (ix86_expand_epilogue): Likewise.
14811 (ix86_expand_split_stack_prologue): Likewise.
14812 (ix86_compute_frame_layout): Remove frame parameter ...
14813 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
14814 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
14816 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
14817 (ix86_frame): Move from here ...
14818 * config/i386/i386.h (ix86_frame): ... to here.
14819 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
14820 complete ix86_frame data structure instead. Remove some_ld_name.
14822 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
14824 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
14825 symbols that hold a DECL_VALUE_EXPR.
14827 2017-06-01 Martin Jambor <mjambor@suse.cz>
14829 PR tree-optimization/80898
14830 * tree-sra.c (process_subtree_disqualification): Removed.
14831 (disqualify_candidate): Do not acll
14832 process_subtree_disqualification.
14833 (subtree_mark_written_and_enqueue): New function.
14834 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
14835 RHS has been disqualified and re-queue LHS if necessary. Apart
14836 from that, ignore disqualified RHS.
14838 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14840 * config/s390/s390.c (s390_emit_epilogue): Disable early return
14841 address fetch for z10 or later.
14843 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14845 * config/arc/arc.md (tst_movb): Add guard when splitting.
14847 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14849 * config/arc/arc.c (arc_can_eliminate): Test against
14850 arc_frame_pointer_needed.
14852 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14854 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
14855 to prevent store reordering.
14856 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
14857 (type): Add block type.
14858 (stack_tie): Define special instruction to be used in
14861 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14863 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
14864 constraint. It is not valid for the pattern.
14865 (noncommutative_binary_comparison): Likewise.
14867 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14869 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
14872 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14874 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
14875 be used by the reg-alloc.
14877 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14879 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
14880 reg-alloc when having mul64 or mul32x16 instructions.
14881 (mulsidi3): Likewise.
14882 (umulsidi3): Likewise.
14883 (mulsi32x16): New pattern.
14884 (mulsi64): Likewise.
14885 (mulsidi64): Likewise.
14886 (umulsidi64): Likewise.
14887 (MUL32x16_REG): Define.
14888 (mul64_600): Use MUL32x16_REG.
14889 (mac64_600): Likewise.
14890 (umul64_600): Likewise.
14891 (umac64_600): Likewise.
14893 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14895 * config/arc/arc.md (mulsi3_700): Make it commutative.
14897 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
14899 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
14901 (*sign_extendsidi2_insn): Likewise for movstosw.
14903 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
14905 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
14906 the type of the input discriminant value. Convert the
14907 discriminant value of signedness vary.
14909 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
14911 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
14912 Add to -Wall section.
14914 2017-06-01 Richard Biener <rguenther@suse.de>
14916 PR middle-end/66313
14917 * fold-const.c (fold_plusminus_mult_expr): If the factored
14918 factor may be zero use a wrapping type for the inner operation.
14919 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
14920 and handle moved defs.
14921 (process_assignment): Properly guard the unary op case. Return a
14922 tri-state indicating that moving the stmt before the call may allow
14923 to continue. Pass through to_move.
14924 (find_tail_calls): Handle moving unrelated defs before
14927 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
14930 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
14931 splitter result in the canonical way.
14933 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
14935 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
14936 also for 32bit target. Update insn attributes.
14937 (zero-extendsidi2 splitter): Allow all registers for operand 1.
14939 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
14941 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
14942 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
14943 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
14944 (_mm_maskz_min_ss): New intrinsics.
14946 2017-05-31 Martin Liska <mliska@suse.cz>
14948 * tree-vect-loop.c (vect_create_epilog_for_reduction):
14949 Change comment style to one we normally use.
14950 (vectorizable_reduction): Likewise.
14951 (vectorizable_induction): Likewise.
14952 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
14953 (vectorizable_call): Likewise.
14954 (vectorizable_simd_clone_call): Likewise.
14955 (vectorizable_conversion): Likewise.
14956 (vectorizable_assignment): Likewise.
14957 (vectorizable_shift): Likewise.
14958 (vectorizable_operation): Likewise.
14959 (vectorizable_store): Likewise.
14960 (vectorizable_load): Likewise.
14961 * tree-vectorizer.h: Likewise.
14963 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
14965 * passes.c (emergency_dump_function): New.
14966 * tree-pass.h (emergency_dump_function): Declare.
14967 * plugin.c (plugins_internal_error_function): Remove.
14968 * plugin.h (plugins_internal_error_function): Remove declaration.
14969 * toplev.c (internal_error_function): New static function. Use it...
14970 (general_init): ...here.
14972 2017-05-31 Graham Markall <graham.markall@embecosm.com>
14974 * config/arc/arc.c (arc_print_operand): Handle constant operands.
14975 (arc_rtx_costs): Add costs for new patterns.
14976 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
14977 * config/arc/predicates.md: Add _1_2_3_operand predicate.
14979 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
14981 * tree-ssa-strlen.c (get_next_strinfo): New function.
14982 (get_stridx_plus_constant): Use it.
14983 (zero_length_string): Likewise.
14984 (adjust_related_strinfos): Likewise.
14985 (adjust_last_stmt): Likewise.
14987 2017-05-31 Richard Biener <rguenther@suse.de>
14990 * config/i386/i386.c (ix86_expand_builtin): Remove assert
14991 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
14993 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
14995 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
14996 loop_vinfo argument and use of dependence distance vectors.
14997 Check instead whether the two references differ only in their
14998 initial value and assume that they have the same alignment if the
14999 difference is a multiple of the vector alignment.
15000 (vect_analyze_data_refs_alignment): Update call accordingly.
15002 2017-05-31 Martin Liska <mliska@suse.cz>
15005 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
15007 2017-05-31 Bin Cheng <bin.cheng@arm.com>
15009 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
15010 (create_intersect_range_checks): Move from ...
15011 * tree-data-ref.c (create_intersect_range_checks_index)
15012 (create_intersect_range_checks): ... to here.
15013 (create_runtime_alias_checks): New function factored from ...
15014 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
15015 here. Call above function.
15016 * tree-data-ref.h (create_runtime_alias_checks): New function.
15018 2017-05-31 Bin Cheng <bin.cheng@arm.com>
15020 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
15021 segment length for dr_b and compute it in wide_int.
15023 2017-05-31 Richard Biener <rguenther@suse.de>
15025 PR tree-optimization/80906
15026 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
15027 and pass through iv_map.
15028 (copy_bb_and_scalar_dependences): Adjust.
15029 (translate_pending_phi_nodes): Likewise.
15030 (copy_loop_close_phi_args): Handle code-generating IVs instead
15033 2017-05-30 David Malcolm <dmalcolm@redhat.com>
15035 * diagnostic-color.c (color_dict): Add "type-diff".
15036 (parse_gcc_colors): Update comment.
15037 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
15038 -fdiagnostics-show-template-tree and -fno-elide-type.
15039 (GCC_COLORS): Add type-diff to example.
15041 (-fdiagnostics-show-template-tree): New.
15042 (-fno-elide-type): New.
15043 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
15044 the pp_format_decoder callback. Call any m_format_postprocessor's
15046 (pretty_printer::pretty_printer): Initialize
15047 m_format_postprocessor.
15048 (pretty_printer::~pretty_printer): Delete any
15049 m_format_postprocessor.
15050 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
15051 (class format_postprocessor): New class.
15052 (struct pretty_printer::format_decoder): Document the new parameters.
15053 (struct pretty_printer::m_format_postprocessor): New field.
15054 * tree-diagnostic.c (default_tree_printer): Update for new
15055 bool and const char ** params.
15056 * tree-diagnostic.h (default_tree_printer): Likewise.
15058 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
15060 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
15061 (lwa_operand): Delete rs6000_gen_cell_microcode test.
15062 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
15063 rs6000_gen_cell_microcode code.
15064 (rs6000_final_prescan_insn): Delete.
15065 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
15066 "warn-cell-microcode" entries.
15067 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
15068 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
15069 throughout. Change cc_reg_not_micro_cr0_operand to
15070 cc_reg_not_cr0_operand throughout.
15071 (*extendhi<mode>2_noload): Delete.
15072 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
15073 (mwarn-cell-microcode): Delete.
15074 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
15075 -mgen-cell-microcode and -mwarn-cell-microcode.
15077 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
15080 * config/i386/constraints.md (Yd): New constraint.
15082 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
15083 and (?Yd, r) alternatives. Update insn attributes.
15084 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
15085 and (?*Yd, r) alternatives. Update insn attributes.
15086 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
15088 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
15090 * gimplify.c (gimplify_modify_expr): Don't create a
15091 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
15094 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
15096 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
15098 2017-05-30 Richard Biener <rguenther@suse.de>
15100 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
15101 and reduc_def fields.
15102 (STMT_VINFO_REDUC_TYPE): New define.
15103 (STMT_VINFO_REDUC_DEF): Likewise.
15104 (vect_force_simple_reduction): Adjust prototype.
15105 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
15106 (vect_is_simple_reduction): Remove check_reduction argument.
15107 (vect_force_simple_reduction): Adjust and set
15108 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
15109 (vectorizable_reduction): Do not re-do reduction analysis
15110 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
15111 * tree-parloops.c (gather_scalar_reductions): Adjust.
15113 2017-05-30 Richard Biener <rguenther@suse.de>
15115 PR middle-end/80901
15116 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
15119 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
15121 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
15122 Introduce unknown_misalignment parameter and remove vf.
15123 (vect_peeling_hash_get_lowest_cost):
15124 Pass unknown_misalignment parameter.
15125 (vect_enhance_data_refs_alignment):
15126 Fix unsupportable data ref treatment.
15128 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
15130 * tree-vect-data-refs.c (vect_get_data_access_cost):
15131 Workaround for SLP handling.
15132 (vect_enhance_data_refs_alignment):
15133 Compute costs for doing no peeling at all, compare to the best
15134 peeling costs so far and avoid peeling if cheaper.
15136 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
15138 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
15139 Return peeling info and set costs to zero for unlimited cost
15141 (vect_enhance_data_refs_alignment): Also inspect all datarefs
15142 with unknown misalignment. Compute and costs for unknown
15143 misalignment, compare them to the costs for known misalignment
15144 and choose the cheapest for peeling.
15146 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
15148 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
15149 (vect_get_peeling_costs_all_drs): Create function.
15150 (vect_peeling_hash_get_lowest_cost):
15151 Use vect_get_peeling_costs_all_drs.
15152 (vect_peeling_supportable): Create function.
15153 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
15155 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
15157 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
15158 DR_HAS_NEGATIVE_STEP.
15159 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
15160 (vect_enhance_data_refs_alignment): Use.
15161 (vect_duplicate_ssa_name_ptr_info): Use.
15162 * tree-vectorizer.h (dr_misalignment): Use.
15163 (known_alignment_for_access_p): Use.
15165 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
15168 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
15170 (has_section_name): New function.
15172 2017-05-30 Martin Liska <mliska@suse.cz>
15175 * auto-profile.c (get_function_decl_from_block): Fix
15178 2017-05-30 Richard Biener <rguenther@suse.de>
15180 PR middle-end/80876
15181 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
15183 2017-05-30 Martin Liska <mliska@suse.cz>
15185 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
15186 * dumpfile.h (struct dump_file_info): Remove ctors.
15188 2017-05-30 Martin Liska <mliska@suse.cz>
15190 * predict.def: Fix GNU coding style.
15192 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
15194 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
15195 Mark 'to' argument with ATTRIBUTE_UNUSED.
15197 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
15199 * config/xtensa/xtensa.c (xtensa_emit_call): Use
15200 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
15201 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
15204 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
15206 * doc/install.texi (Options specification): Restore entry of
15207 --enable-sjlj-exceptions.
15209 2017-05-27 Michael Eager <eager@eagercon.com>
15212 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
15214 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
15216 * config/microblaze/microblaze.h
15217 (FIXED_REGISTERS): Update in macro.
15218 (CALL_USED_REGISTERS): Update in macro.
15220 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
15222 * doc/install.texi: Add links to macOS binary distributions.
15224 2017-05-27 Jakub Jelinek <jakub@redhat.com>
15228 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
15230 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
15232 2017-05-26 Martin Liska <mliska@suse.cz>
15234 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
15236 2017-05-26 Martin Liska <mliska@suse.cz>
15238 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
15239 always leading ';; '.
15240 (dump_bb_info): Likewise.
15241 (brief_dump_cfg): Likewise.
15242 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
15243 * dumpfile.c: Remove usage of TDF_VERBOSE.
15244 * dumpfile.h (enum dump_kind): Likewise.
15245 (dump_gimple_bb_header): Do not use TDF_COMMENT.
15246 * print-tree.c (debug_verbose): Remove.
15247 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
15248 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
15249 * tree-diagnostic.c (default_tree_printer): Replace
15250 TDF_DIAGNOSTIC with TDF_SLIM.
15252 2017-05-26 Bin Cheng <bin.cheng@arm.com>
15254 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
15255 in parameter loop, rather than loop_vinfo.
15256 (create_intersect_range_checks): Ditto.
15257 (vect_create_cond_for_alias_checks): Update call to above functions.
15259 2017-05-26 Bin Cheng <bin.cheng@arm.com>
15261 PR tree-optimization/80815
15262 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
15263 for merging runtime alias checks. Handle negative DR_STEPs.
15265 2017-05-26 Bin Cheng <bin.cheng@arm.com>
15267 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
15269 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
15270 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
15271 out code pruning runtime alias checks.
15272 * tree-data-ref.c (prune_runtime_alias_test_list): New function
15273 factored out from above.
15274 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
15276 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
15278 (prune_runtime_alias_test_list): New decalaration.
15280 2017-05-26 Bin Cheng <bin.cheng@arm.com>
15282 * tree-vect-data-refs.c (compare_tree): Rename and move ...
15283 * tree-data-ref.c (data_ref_compare_tree): ... to here.
15284 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
15285 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
15286 (operator==, comp_dr_with_seg_len_pair): Ditto.
15287 (vect_prune_runtime_alias_test_list): Ditto.
15289 2017-05-26 Martin Liska <mliska@suse.cz>
15292 * params.def: Bound partial-inlining-entry-probability param.
15294 2017-05-26 Marek Polacek <polacek@redhat.com>
15297 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
15300 2017-05-26 Richard Biener <rguenther@suse.de>
15302 PR tree-optimization/80842
15303 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
15306 2017-05-26 Richard Biener <rguenther@suse.de>
15308 PR tree-optimization/80844
15309 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
15311 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
15313 * doc/md.texi (Machine Constraints): Update x86 family
15314 machine constraints section to match 'config/i386/constraints.md'.
15316 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
15318 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
15320 2017-05-25 Nathan Sidwell <nathan@acm.org>
15322 * doc/invoke.texi (--enable-languages): Update documentation.
15324 2017-05-25 Martin Liska <mliska@suse.cz>
15326 * dumpfile.c: Add TDF_FOLDING.
15327 * dumpfile.h (enum dump_kind): Likewise.
15328 * genmatch.c (dt_simplify::gen_1): Use it.
15330 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
15332 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
15334 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
15336 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
15337 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
15339 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
15341 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
15342 * match.pd (X == C): Rewrite it here.
15343 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
15344 with_certain_nonzero_bits2): New predicates.
15345 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
15347 2017-05-24 Nathan Sidwell <nathan@acm.org>
15349 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
15352 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
15355 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
15357 * config/powerpcspe: New port. Files are copied from the rs6000
15358 port, with "rs6000" in filenames replaced by "powerpcspe".
15360 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
15362 PR rtl-optimization/80754
15363 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
15365 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
15367 * config/sparc/sparc.md (length): Return the correct value for -mflat
15368 sibcalls to match output_sibcall.
15370 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
15374 * config/rs6000/rs6000.c (struct machine_function): Add new field
15376 (rs6000_get_separate_components): Init that field, use it.
15377 (rs6000_components_for_bb): Use the field.
15379 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
15381 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
15383 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
15385 PR middle-end/80823
15386 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
15388 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15391 * config/s390/s390.c (s390_check_qrst_address): Check incoming
15392 address against address_operand predicate.
15393 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
15395 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
15397 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
15398 parameters passed indirectly.
15400 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
15402 * config/i386/i386.md (*movdi_internal): Remove SSE4
15403 alternative 18 (?r, *v). Update insn attributes.
15404 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
15405 Update insn attributes.
15406 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
15407 Update insn attributes.
15408 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
15409 alternative 1 (r, v). Remove isa attribute.
15410 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
15411 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
15412 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
15414 2017-05-23 Tom de Vries <tom@codesourcery.com>
15416 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
15419 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
15421 * cgraphunit.c (symbol_table::process_new_functions): Update.
15422 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
15423 (inline_generate_summary): Rename to ...
15424 (ipa_fn_summary_generate): ... this one.
15425 (inline_read_summary): Rename to ...
15426 (ipa_fn_summary_read): ... this one.
15427 (inline_write_summary): Rename to ...
15428 (ipa_fn_summary_write): ... this one.
15429 (inline_free_summary): Rename to ...
15430 (ipa_free_fn_summary): ... this one.
15431 (pass_data_local_fn_summary, pass_local_fn_summary,
15432 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
15433 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
15434 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
15435 make_pass_ipa_fn_summary): New.
15436 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
15437 inline_write_summary, inline_free_summary): Remove.
15438 (ipa_free_fn_summary) : New.
15439 * ipa-inline.c (ipa_inline): Update.
15440 (pass_ipa_inline): Do not generate summaries.
15441 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
15443 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
15444 and add pass_ipa_fn_summary.
15445 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
15447 (make_pass_inline_parameters): Remove.
15449 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
15451 * omp-low.c (struct omp_context): Remove "default_kind" member.
15454 * omp-offload.c (execute_oacc_device_lower): Remove the
15455 parallelism dimensions function attributes for unparallelized
15456 OpenACC kernels constructs.
15458 2017-05-23 Martin Liska <mliska@suse.cz>
15460 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
15462 (cgraph_edge::make_speculative): Likewise.
15463 (cgraph_edge::resolve_speculation): Likewise.
15464 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
15465 (cgraph_node::dump): Likewise.
15466 * cgraph.h: Likewise.
15467 * cgraphunit.c (analyze_functions): Likewise.
15468 (symbol_table::compile): Likewise.
15469 * ipa-cp.c (print_all_lattices): Likewise.
15470 (determine_versionability): Likewise.
15471 (initialize_node_lattices): Likewise.
15472 (ipcp_verify_propagated_values): Likewise.
15473 (estimate_local_effects): Likewise.
15474 (update_profiling_info): Likewise.
15475 (create_specialized_node): Likewise.
15476 (perhaps_add_new_callers): Likewise.
15477 (decide_about_value): Likewise.
15478 (decide_whether_version_node): Likewise.
15479 (identify_dead_nodes): Likewise.
15480 (ipcp_store_bits_results): Likewise.
15481 * ipa-devirt.c (dump_targets): Likewise.
15482 (ipa_devirt): Likewise.
15483 * ipa-icf.c (sem_item::dump): Likewise.
15484 (sem_function::equals): Likewise.
15485 (sem_variable::equals): Likewise.
15486 (sem_item_optimizer::read_section): Likewise.
15487 (sem_item_optimizer::execute): Likewise.
15488 (congruence_class::dump): Likewise.
15489 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
15490 (dump_inline_summary): Likewise.
15491 (estimate_node_size_and_time): Likewise.
15492 (inline_analyze_function): Likewise.
15493 * ipa-inline-transform.c (inline_call): Likewise.
15494 * ipa-inline.c (report_inline_failed_reason): Likewise.
15495 (want_early_inline_function_p): Likewise.
15496 (edge_badness): Likewise.
15497 (update_edge_key): Likewise.
15498 (inline_small_functions): Likewise.
15499 * ipa-profile.c (ipa_profile): Likewise.
15500 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
15501 (ipa_make_edge_direct_to_target): Likewise.
15502 (remove_described_reference): Likewise.
15503 (ipa_impossible_devirt_target): Likewise.
15504 (propagate_controlled_uses): Likewise.
15505 (ipa_print_node_params): Likewise.
15506 (ipcp_transform_function): Likewise.
15507 * ipa-pure-const.c (pure_const_read_summary): Likewise.
15508 (propagate_pure_const): Likewise.
15509 * ipa-reference.c (generate_summary): Likewise.
15510 (read_write_all_from_decl): Likewise.
15511 (propagate): Likewise.
15512 (ipa_reference_read_optimization_summary): Likewise.
15513 * ipa-utils.c (ipa_merge_profiles): Likewise.
15514 * ipa.c (walk_polymorphic_call_targets): Likewise.
15515 (symbol_table::remove_unreachable_nodes): Likewise.
15516 (ipa_single_use): Likewise.
15517 * passes.c (execute_todo): Likewise.
15518 * predict.c (drop_profile): Likewise.
15519 * symtab.c (symtab_node::get_dump_name): New function.
15520 (symtab_node::dump_name): Likewise.
15521 (symtab_node::dump_asm_name): Likewise.
15522 (symtab_node::dump_references): Likewise.
15523 (symtab_node::dump_referring): Likewise.
15524 (symtab_node::dump_base): Likewise.
15525 (symtab_node::debug_symtab): Likewise.
15526 * tree-sra.c (convert_callers_for_node): Likewise.
15527 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
15528 * value-prof.c (init_node_map): Likewise.
15530 2017-05-23 Martin Liska <mliska@suse.cz>
15532 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
15533 and symtab_node::debug_symtab to symbol_table::debug.
15534 * cgraphunit.c (analyze_functions): Use the renamed function.
15535 (symbol_table::compile): Likewise.
15536 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
15537 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
15538 * passes.c (execute_todo): Likewise.
15539 * symtab.c (symbol_table::dump): New function.
15540 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
15542 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
15544 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
15545 that nonconst implies exec.
15547 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
15549 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
15550 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
15551 (inline_edge_summary_vec): Turn into ...
15552 (ipa_call_summaries): ... this one.
15553 (redirect_to_unreachable, edge_set_predicate,
15554 evaluate_properties_for_edge, inline_summary_alloc,
15555 reset_ipa_call_summary, reset_inline_summary,
15556 inline_summary_t::duplicate): Update.
15557 (inline_edge_duplication_hook): Turn to ...
15558 (ipa_call_summary_t::duplicate): ... this one.
15559 (inline_edge_removal_hook): Turn to ...
15560 (ipa_call_summary_t::remove): ... this one.
15561 (dump_inline_edge_summary): Turn to ...
15562 (dump_ipa_call_summary): ... this one.
15563 (estimate_function_body_sizes): Update.
15564 (inline_update_callee_summaries): Update.
15565 (remap_edge_change_prob): Update.
15566 (remap_edge_summaries): Update.
15567 (inline_merge_summary): Update.
15568 (do_estimate_edge_time): Update.
15569 (inline_generate_summary): Update.
15570 (inline_read_section): Update.
15571 (inline_read_summary): Update.
15572 (inline_free_summary): Update.
15573 * ipa-inline.c (can_inline_edge_p): Update.
15574 (compute_inlined_call_time): Update.
15575 (want_inline_small_function_p): Update.
15576 (edge_badness): Update.
15577 (early_inliner): Update.
15578 * ipa-inline.h (inline_edge_summary): Turn to ...
15579 (ipa_call_summary): ... this one.
15580 (ipa_call_summary_t): New class.
15581 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
15582 (ipa_call_summaries): New.
15583 (inline_edge_summary): Remove.
15584 (estimate_edge_growth): Update.
15585 * ipa-profile.c (ipa_propagate_frequency_1): Update.
15586 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
15587 * ipa-split.c (execute_split_functions): Update.
15588 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
15590 2017-05-23 Tom de Vries <tom@codesourcery.com>
15592 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
15593 attributes): Document rdrand effective target.
15595 2017-05-23 Tom de Vries <tom@codesourcery.com>
15597 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
15598 attributes): Sort alphabetically.
15600 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
15602 * config/avr/genmultilib.awk: Use gsub instead of gensub.
15604 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
15607 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
15608 V2DF/V2DI splat into two separate patterns, one that handles
15609 registers, and the other that only handles memory. Drop support
15610 for splatting from a GPR on ISA 2.07 and then splitting the
15611 splat into direct move and splat.
15612 (vsx_splat_<mode>_reg): Likewise.
15613 (vsx_splat_<mode>_mem): Likewise.
15615 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
15617 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
15619 2017-05-22 Jakub Jelinek <jakub@redhat.com>
15621 PR middle-end/80809
15622 * omp-low.c (finish_taskreg_remap): New function.
15623 (finish_taskreg_scan): If unit size of ctx->record_type
15624 is non-constant, unshare the size expression and replace
15625 decls in it with possible outer var refs.
15627 PR middle-end/80809
15628 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
15629 GOVD_SHARED rather than GOVD_PRIVATE with it.
15630 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
15631 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
15633 PR middle-end/80853
15634 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
15635 as last argument to build_outer_var_ref for pointer bases of array
15636 section reductions.
15638 2017-05-19 Martin Sebor <msebor@redhat.com>
15640 * print-tree.c (print_node): Print DECL_READ_P flag.
15642 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15644 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
15645 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
15646 * cgraph.c: Likewise.
15647 * cgraphunit.c: Likewise.
15648 * gengtype.c: Likewise.
15649 * ipa-cp.c: Likewise.
15650 * ipa-devirt.c: Likewise.
15651 * ipa-icf.c: Likewise.
15652 * ipa-predicate.c: Likewise.
15653 * ipa-profile.c: Likewise.
15654 * ipa-prop.c: Likewise.
15655 * ipa-split.c: Likewise.
15657 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
15658 edge_predicate_pool, dump_inline_hints,
15659 inline_summary::account_size_time, redirect_to_unreachable,
15660 edge_set_predicate, set_hint_predicate,
15661 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
15662 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
15663 inline_summary_t::remove, remap_hint_predicate_after_duplication,
15664 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
15665 ipa_call_summary_t::remove, initialize_growth_caches,
15666 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
15667 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
15668 mark_modified, unmodified_parm_1, unmodified_parm,
15669 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
15670 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
15671 compute_bb_predicates, will_be_nonconstant_expr_predicate,
15672 will_be_nonconstant_predicate, record_modified_bb_info,
15673 get_minimal_bb, record_modified, param_change_prob,
15674 phi_result_unknown_predicate, predicate_for_phi_result,
15675 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
15676 estimate_function_body_sizes, compute_inline_parameters,
15677 compute_inline_parameters_for_curren, pass_data_inline_parameters,
15678 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
15679 inline_update_callee_summaries, remap_edge_change_prob,
15680 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
15681 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
15682 inline_analyze_function, inline_summary_t::insert,
15683 inline_generate_summary, read_ipa_call_summary, inline_read_section,
15684 inline_read_summary, write_ipa_call_summary, inline_write_summary,
15685 inline_free_summary): Move to ipa-fnsummary.h
15686 (predicate_t): Remove.
15687 * ipa-fnsummary.c: New file.
15688 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
15689 (enum inline_hints_vals, inline_hints, agg_position_info,
15690 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
15691 inline_summaries, ipa_call_summary, ipa_call_summary_t,
15692 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
15693 dump_inline_summary, dump_inline_hints, inline_generate_summary,
15694 inline_read_summary, inline_write_summary, inline_free_summary,
15695 inline_analyze_function, initialize_inline_failed,
15696 inline_merge_summary, inline_update_overall_summary,
15697 compute_inline_parameters): Move to ipa-fnsummary.h
15698 * ipa-fnsummary.h: New file.
15699 * ipa-inline-transform.h: Include ipa-inline.h.
15700 * ipa-inline.c: LIkewise.
15702 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15704 * ipa-inline.c (edge_badness): Use inlined_time instead of
15705 inline_summaries->get.
15707 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15709 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
15711 2017-05-22 Nathan Sidwell <nathan@acm.org>
15713 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
15714 (fdump-lang): Document 'raw' option.
15715 * dumpfile.h (TDI_tu): Delete.
15716 * dumpfile.c (dump_files): Remove translation-unit.
15717 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
15719 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
15721 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
15722 command option from $(AWK) call.
15723 * config/avr/genmultilib.awk: Simplify and rewrite so that it
15724 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
15725 [FORMAT]: Remove handling of variable.
15726 * config/avr/t-multilib: Regenerate.
15728 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15730 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
15732 (dump_inline_summary): Do not print self_time.
15733 (estimate_function_body_sizes): Do not set self_time.
15734 (compute_inline_parameters): Likewise.
15735 (inline_read_section, inline_write_summary): Do not stream self_time.
15736 * ipa-inline.h (inline_summary): Drop self_time.
15738 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15740 * ipa-inline-analysis.c (account_size_time): Rename to ...
15741 (inline_summary::account_size_time): ... this one.
15742 (reset_ipa_call_summary): Turn to ...
15743 (ipa_call_summary::reset): ... this one.
15744 (reset_inline_summary): Turn to ...
15745 (inline_summary::reset): ... this one.
15746 (inline_summary_t::remove): Update.
15747 (inline_summary_t::duplicate): Update.
15748 (ipa_call_summary_t::remove): Update.
15749 (dump_inline_summary): Update.
15750 (estimate_function_body_sizes): Update.
15751 (compute_inline_parameters): Update.
15752 (estimate_node_size_and_time): Update.
15753 (inline_merge_summary): Update.
15754 (inline_update_overall_summary): Update.
15755 (inline_read_section): Update.
15756 (inline_write_summary): Update.
15757 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
15758 add account_size_time and reset member functions.
15759 (ipa_call_summary): Add reset function.
15760 * ipa-predicate.h (predicate::operator &): Constify.
15762 2017-05-22 Richard Biener <rguenther@suse.de>
15764 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
15766 2017-05-19 Jason Merrill <jason@redhat.com>
15768 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
15770 2017-05-19 Marek Polacek <polacek@redhat.com>
15773 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
15774 TYPE_OVERFLOW_WRAPS checks.
15776 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
15778 * tree-core.h (enum omp_clause_default_kind): Add
15779 "OMP_CLAUSE_DEFAULT_PRESENT".
15780 * tree-pretty-print.c (dump_omp_clause): Handle it.
15781 * gimplify.c (enum gimplify_omp_var_data): Add
15782 "GOVD_MAP_FORCE_PRESENT".
15783 (gimplify_adjust_omp_clauses_1): Map it to
15784 "GOMP_MAP_FORCE_PRESENT".
15785 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
15787 * gimplify.c (oacc_default_clause): Clarify.
15789 2017-05-19 Nathan Sidwell <nathan@acm.org>
15791 LANG_HOOK_REGISTER_DUMPS
15792 * toplev.c (general_init): Call register dump lang hook.
15793 * doc/invoke.texi: Document -fdump-lang option family.
15794 * dumpfile.c (dump_files): Remove class dump here.
15795 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
15796 * dumpfile.h (tree_dump_index): Remove TDI_class.
15797 * langhooks-def.h (lhd_register_dumps): Declare.
15798 (LANG_HOOKS_REGISTER_DUMPS): Define.
15799 (LANG_HOOKS_INITIALIZER): Add it.
15800 * langhooks.c (lhd_register_dumps): Define.
15801 * langhooks.h (struct lang_hooks): Add register_dumps.
15803 2017-05-19 Nathan Sidwell <nathan@acm.org>
15805 * context.h (context::set_passes): New.
15806 * context.c (context::context): Do not create pass manager.
15807 * toplev.c (general_init): Create pass manager here.
15809 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
15811 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
15812 use this splitter if two add or or instructions would also work for
15813 the constant we want to generate.
15815 2017-05-19 Richard Biener <rguenther@suse.de>
15818 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
15819 predicate evaluation.
15821 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
15823 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
15825 * ipa-inline.c (want_inline_small_function_p): Do not cast to
15828 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
15830 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
15831 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
15832 (inline_edge_summary_vec): Turn into ...
15833 (ipa_call_summaries): ... this one.
15834 (redirect_to_unreachable, edge_set_predicate,
15835 evaluate_properties_for_edge, inline_summary_alloc,
15836 reset_ipa_call_summary, reset_inline_summary,
15837 inline_summary_t::duplicate): Update.
15838 (inline_edge_duplication_hook): Turn to ...
15839 (ipa_call_summary_t::duplicate): ... this one.
15840 (inline_edge_removal_hook): Turn to ...
15841 (ipa_call_summary_t::remove): ... this one.
15842 (dump_inline_edge_summary): Turn to ...
15843 (dump_ipa_call_summary): ... this one.
15844 (estimate_function_body_sizes): Update.
15845 (inline_update_callee_summaries): Update.
15846 (remap_edge_change_prob): Update.
15847 (remap_edge_summaries): Update.
15848 (inline_merge_summary): Update.
15849 (do_estimate_edge_time): Update.
15850 (inline_generate_summary): Update.
15851 (inline_read_section): Update.
15852 (inline_read_summary): Update.
15853 (inline_free_summary): Update.
15854 * ipa-inline.c (can_inline_edge_p): Update.
15855 (compute_inlined_call_time): Update.
15856 (want_inline_small_function_p): Update.
15857 (edge_badness): Update.
15858 (early_inliner): Update.
15859 * ipa-inline.h (inline_edge_summary): Turn to ...
15860 (ipa_call_summary): ... this one.
15861 (ipa_call_summary_t): New class.
15862 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
15863 (ipa_call_summaries): New.
15864 (inline_edge_summary): Remove.
15865 (estimate_edge_growth): Update.
15866 * ipa-profile.c (ipa_propagate_frequency_1): Update.
15867 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
15868 * ipa-split.c (execute_split_functions): Update.
15869 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
15871 2017-05-19 Richard Biener <rguenther@suse.de>
15873 PR middle-end/80764
15874 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
15876 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
15878 * config/rs6000/rs6000.c (struct machine_function): Add field
15879 fpr_is_wrapped_separately.
15880 (rs6000_get_separate_components): Use 64 components. Handle the
15881 new FPR components.
15882 (rs6000_components_for_bb): Handle the FPR components.
15883 (rs6000_emit_prologue_components): Handle the FPR components.
15884 (rs6000_emit_epilogue_components): Handle the FPR components.
15885 (rs6000_set_handled_components): Handle the FPR components.
15886 (rs6000_emit_prologue): Don't output prologue code for those FPRs
15887 that are already separately shrink-wrapped.
15888 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
15889 that are already separately shrink-wrapped.
15891 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
15894 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
15897 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
15898 (define_peephole2 for Altivec d-form load): Add peepholes to catch
15899 cases where the register allocator uses a move and an offsettable
15900 memory operation to/from a FPR register on ISA 2.06/2.07.
15901 (define_peephole2 for Altivec d-form store): Likewise.
15903 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
15906 * config/i386/mmx.md (*mov<mode>_internal): Enable
15907 alternatives 11, 12, 13 and 14 also for 32bit targets.
15908 Remove alternatives 15, 16, 17 and 18.
15909 * config/i386/sse.md (vec_concatv2di): Change
15910 alternative (!x, *y) to (x, ?!*Yn).
15912 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
15914 * dumpfile.h (enum dump_kind): Remove stray comma.
15916 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
15918 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
15919 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
15920 predicate::num_conditions
15921 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
15922 (CHANGED): turn into predicate::changed.
15923 (agg_position_info): Move to ipa-predicate.h
15924 (add_condition, predicate::add_clause, predicate::operator &=,
15925 predicate::or_with, predicate::evaluate, predicate::probability,
15926 dump_condition, dump_clause, predicate::dump,
15927 predicate::remap_after_duplication, predicate::remap_after_inlining,
15928 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
15929 (evaluate_conditions_for_known_args): Update.
15930 (set_cond_stmt_execution_predicate): Update.
15931 * ipa-inline.h: Include ipa-predicate.h
15932 (condition, inline_param_summary, conditions, agg_position_info,
15933 predicate): Move to ipa-predicate.h
15934 * ipa-predicate.c: New file.
15935 * ipa-predicate.h: New file.
15937 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
15939 * final.c (leaf_function_p): Check we are not in a sequence.
15941 2017-05-18 Martin Liska <mliska@suse.cz>
15943 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
15944 * dumpfile.c (dump_register): Use new enum dump_kind.
15945 (get_dump_file_name): Likewise.
15946 (dump_enable_all): Likewise.
15947 (dump_switch_p_1): Likewise.
15948 (enable_rtl_dump_file): Remove usage of TDF_RTL.
15949 * dumpfile.h (enum dump_kind): New enum type.
15950 (struct dump_file_info): Create constructor and
15951 format fields and comments.
15952 * passes.c (pass_manager::register_one_dump_file):
15954 * statistics.c (statistics_early_init): Likewise.
15955 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
15956 TDF_TREE with TDF_SLIM.
15957 (gather_memory_references_ref): Likewise.
15959 2017-05-18 Martin Liska <mliska@suse.cz>
15961 * vec.h (struct vnull): Use it.
15963 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
15965 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
15966 (true_predicate, false_predicate, true_predicate_p,
15967 false_predicate_p): Remove.
15968 (single_cond_predicate, not_inlined_predicate): Turn to member function
15970 (add_condition): Update.
15971 (add_clause): Turn to...
15972 (predicate::add_clause): ... this one; update; allow passing NULL
15974 (and_predicates): Turn to ...
15975 (predicate::operator &=): ... this one.
15976 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
15977 (or_predicates): Turn to ...
15978 (predicate::or_with): ... this one.
15979 (evaluate_predicate): Turn to ...
15980 (predicate::evaluate): ... this one.
15981 (predicate_probability): Turn to ...
15982 (predicate::probability): ... this one.
15983 (dump_condition): Update.
15984 (dump_predicate): Turn to ...
15985 (predicate::dump): ... this one.
15986 (account_size_time): Update.
15987 (edge_set_predicate): Update.
15988 (set_hint_predicate): UPdate.
15989 (evaluate_conditions_for_known_args): Update.
15990 (evaluate_properties_for_edge): Update.
15991 (remap_predicate_after_duplication): Turn to...
15992 (predicate::remap_after_duplication): ... this one.
15993 (remap_hint_predicate_after_duplication): Update.
15994 (inline_summary_t::duplicate): UPdate.
15995 (dump_inline_edge_summary): Update.
15996 (dump_inline_summary): Update.
15997 (set_cond_stmt_execution_predicate): Update.
15998 (set_switch_stmt_execution_predicate): Update.
15999 (compute_bb_predicates): Update.
16000 (will_be_nonconstant_expr_predicate): Update.
16001 (will_be_nonconstant_predicate): Update.
16002 (phi_result_unknown_predicate): Update.
16003 (predicate_for_phi_result): Update.
16004 (array_index_predicate): Update.
16005 (estimate_function_body_sizes): Update.
16006 (estimate_node_size_and_time): Update.
16007 (estimate_ipcp_clone_size_and_time): Update.
16008 (remap_predicate): Rename to ...
16009 (predicate::remap_after_inlining): ... this one.
16010 (remap_hint_predicate): Update.
16011 (inline_merge_summary): Update.
16012 (inline_update_overall_summary): Update.
16013 (estimate_size_after_inlining): Update.
16014 (read_predicate): Rename to ...
16015 (predicate::stream_in): ... this one.
16016 (read_inline_edge_summary): Update.
16017 (write_predicate): Rename to ...
16018 (predicate::stream_out): ... this one.
16019 (write_inline_edge_summary): Update.
16020 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
16021 (clause_t): Turn to uint32_t
16022 (predicate): Turn to class; implement constructor and operators
16024 (size_time_entry): Update.
16025 (inline_summary): Update.
16026 (inline_edge_summary): Update.
16028 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
16030 * fold-const.c (fold_binary_loc): Move transformation...
16031 * match.pd (C - X CMP X): ... here.
16033 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
16035 * config/sparc/sparc.c (sparc_option_override): Set function
16036 alignment for -mcpu=niagara7 to 64 to match the I$ line.
16037 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
16039 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
16041 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
16043 2017-05-18 Marek Polacek <polacek@redhat.com>
16046 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
16047 (pass_ubsan::execute): Call gimple_assign_single_p instead of
16048 gimple_assign_load_p.
16050 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
16052 PR middle-end/80692
16053 * real.c (do_compare): Give decimal_do_compare preference over
16054 comparing just the signs.
16056 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
16058 * doc/md.texi (Canonicalization of Instructions): Describe the
16059 canonical form of instructions that inherently set a condition
16062 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
16064 PR middle-end/80775
16065 * tree-cfg.c: Move deletion of unreachable case statements to after
16066 the merging of consecutive case labels.
16068 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
16070 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
16071 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
16072 restoring of callee-saved registers.
16074 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
16076 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
16077 * config/visium/visium.c (single_set_and_flags): Likewise.
16078 * config/visium/visium.md (Substitutions): Likewise.
16080 2017-05-17 Martin Liska <mliska@suse.cz>
16082 * cfg.c: Introduce dump_flags_t type and
16083 use it instead of int type.
16085 * cfghooks.c: Likewise.
16086 * cfghooks.h (struct cfg_hooks): Likewise.
16087 * cfgrtl.c: Likewise.
16088 * cfgrtl.h: Likewise.
16089 * cgraph.c (cgraph_node::get_body): Likewise.
16090 * coretypes.h: Likewise.
16091 * domwalk.c: Likewise.
16092 * domwalk.h: Likewise.
16093 * dumpfile.c (struct dump_option_value_info): Likewise.
16094 (dump_enable_all): Likewise.
16095 (dump_switch_p_1): Likewise.
16096 (opt_info_switch_p): Likewise.
16097 * dumpfile.h (enum tree_dump_index): Likewise.
16098 (struct dump_file_info): Likewise.
16099 * genemit.c: Likewise.
16100 * generic-match-head.c: Likewise.
16101 * gengtype.c (open_base_files): Likewise.
16102 * gimple-pretty-print.c: Likewise.
16103 * gimple-pretty-print.h: Likewise.
16104 * graph.c (print_graph_cfg): Likewise.
16105 * graphite-scop-detection.c (dot_all_sese): Likewise.
16106 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
16107 * loop-unroll.c (report_unroll): Likewise.
16108 * passes.c (pass_manager::register_one_dump_file): Likewise.
16109 * print-tree.c: Likewise.
16110 * statistics.c: Likewise.
16111 * tree-cfg.c: Likewise.
16112 * tree-cfg.h: Likewise.
16113 * tree-dfa.c: Likewise.
16114 * tree-dfa.h: Likewise.
16115 * tree-dump.c (dump_function): Likewise.
16116 * tree-dump.h (struct dump_info): Likewise.
16117 * tree-pretty-print.c: Likewise.
16118 * tree-pretty-print.h: Likewise.
16119 * tree-ssa-live.c: Likewise.
16120 * tree-ssa-live.h: Likewise.
16121 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
16122 * tree-vect-loop.c: Likewise.
16123 * tree-vect-slp.c: Likewise.
16125 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
16126 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16128 PR tree-optimization/80457
16129 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
16130 of all arguments to a statement as scalar_to_vec operations.
16131 (vectorizable_call): Adjust call to vect_model_simple_cost for
16133 (vectorizable_conversion): Likewise.
16134 (vectorizable_assignment): Likewise.
16135 (vectorizable_shift): Likewise.
16136 (vectorizable_operation): Likewise.
16137 (vectorizable_comparison): Likewise.
16138 (vect_is_simple_cond): Record the def types for operands.
16139 (vectorizable_condition): Likewise, call vect_model_simple_cost.
16140 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
16141 for statement argument count.
16143 2017-05-16 Carl Love <cel@us.ibm.com>
16145 * config/rs6000/rs6000-c: Add support for built-in functions
16146 vector unsigned long long vec_bperm (vector unsigned long long,
16147 vector unsigned char)
16148 vector signed long long vec_mule (vector signed int,
16150 vector unsigned long long vec_mule (vector unsigned int,
16151 vector unsigned int)
16152 vector signed long long vec_mulo (vector signed int,
16154 vector unsigned long long vec_mulo (vector unsigned int,
16155 vector unsigned int)
16156 vector signed char vec_sldw (vector signed char,
16157 vector signed char,
16159 vector unsigned char vec_sldw (vector unsigned char,
16160 vector unsigned char,
16162 vector signed short vec_sldw (vector signed short,
16163 vector signed short,
16165 vector unsigned short vec_sldw (vector unsigned short,
16166 vector unsigned short,
16168 vector signed int vec_sldw (vector signed int,
16171 vector unsigned int vec_sldw (vector unsigned int,
16172 vector unsigned int,
16174 vector signed long long vec_sldw (vector signed long long,
16175 vector signed long long,
16177 vector unsigned long long vec_sldw (vector unsigned long long,
16178 vector unsigned long long,
16180 * config/rs6000/rs6000-c: Add support for built-in functions
16181 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
16182 * config/rs6000/altivec.h: Add defintion for vec_sldw.
16183 * doc/extend.texi: Update the built-in documentation for the
16184 new built-in functions.
16186 2017-05-16 Marek Polacek <polacek@redhat.com>
16190 * tree.c (save_expr): Don't fold the expression.
16192 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
16194 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
16195 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
16196 and (?*y,m). Update insn attributes.
16198 2017-05-16 Martin Liska <mliska@suse.cz>
16200 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
16201 flags argument of print_gimple_stmt, print_gimple_expr,
16202 print_generic_stmt and print_generic_expr.
16203 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
16204 * coretypes.h: Likewise.
16205 * except.c (dump_eh_tree): Likewise.
16206 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
16207 * gimple-pretty-print.h: Likewise.
16208 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
16209 (backprop::push_to_worklist): Likewise.
16210 (backprop::pop_from_worklist): Likewise.
16211 (backprop::process_use): Likewise.
16212 (backprop::intersect_uses): Likewise.
16213 (note_replacement): Likewise.
16214 * gimple-ssa-store-merging.c
16215 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
16216 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
16217 (pass_store_merging::execute): Likewise.
16218 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
16219 (ssa_base_cand_dump_callback): Likewise.
16220 (dump_incr_vec): Likewise.
16221 (replace_refs): Likewise.
16222 (replace_mult_candidate): Likewise.
16223 (create_add_on_incoming_edge): Likewise.
16224 (create_phi_basis): Likewise.
16225 (insert_initializers): Likewise.
16226 (all_phi_incrs_profitable): Likewise.
16227 (introduce_cast_before_cand): Likewise.
16228 (replace_one_candidate): Likewise.
16229 * gimplify.c (gimplify_expr): Likewise.
16230 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
16231 (set_rename): Likewise.
16232 (rename_uses): Likewise.
16233 (copy_loop_phi_nodes): Likewise.
16234 (add_close_phis_to_merge_points): Likewise.
16235 (copy_loop_close_phi_args): Likewise.
16236 (copy_cond_phi_args): Likewise.
16237 (graphite_copy_stmts_from_block): Likewise.
16238 (translate_pending_phi_nodes): Likewise.
16239 * graphite-poly.c (print_pdr): Likewise.
16240 (dump_gbb_cases): Likewise.
16241 (dump_gbb_conditions): Likewise.
16242 (print_scop_params): Likewise.
16243 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
16244 (build_cross_bb_scalars_use): Likewise.
16245 (gather_bbs::before_dom_children): Likewise.
16246 * hsa-dump.c (dump_hsa_immed): Likewise.
16247 * ipa-cp.c (print_ipcp_constant_value): Likewise.
16248 (get_replacement_map): Likewise.
16249 * ipa-inline-analysis.c (dump_condition): Likewise.
16250 (estimate_function_body_sizes): Likewise.
16251 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
16252 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
16253 * ipa-prop.c (ipa_dump_param): Likewise.
16254 (ipa_print_node_jump_functions_for_edge): Likewise.
16255 (ipa_modify_call_arguments): Likewise.
16256 (ipa_modify_expr): Likewise.
16257 (ipa_dump_param_adjustments): Likewise.
16258 (ipa_dump_agg_replacement_values): Likewise.
16259 (ipcp_modif_dom_walker::before_dom_children): Likewise.
16260 * ipa-pure-const.c (check_stmt): Likewise.
16261 (pass_nothrow::execute): Likewise.
16262 * ipa-split.c (execute_split_functions): Likewise.
16263 * omp-offload.c (dump_oacc_loop_part): Likewise.
16264 (dump_oacc_loop): Likewise.
16265 * trans-mem.c (tm_log_emit): Likewise.
16266 (tm_memopt_accumulate_memops): Likewise.
16267 (dump_tm_memopt_set): Likewise.
16268 (dump_tm_memopt_transform): Likewise.
16269 * tree-cfg.c (gimple_verify_flow_info): Likewise.
16270 (print_loop): Likewise.
16271 * tree-chkp-opt.c (chkp_print_addr): Likewise.
16272 (chkp_gather_checks_info): Likewise.
16273 (chkp_get_check_result): Likewise.
16274 (chkp_remove_check_if_pass): Likewise.
16275 (chkp_use_outer_bounds_if_possible): Likewise.
16276 (chkp_reduce_bounds_lifetime): Likewise.
16277 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
16278 (chkp_mark_completed_bounds): Likewise.
16279 (chkp_register_incomplete_bounds): Likewise.
16280 (chkp_mark_invalid_bounds): Likewise.
16281 (chkp_maybe_copy_and_register_bounds): Likewise.
16282 (chkp_build_returned_bound): Likewise.
16283 (chkp_get_bound_for_parm): Likewise.
16284 (chkp_build_bndldx): Likewise.
16285 (chkp_get_bounds_by_definition): Likewise.
16286 (chkp_generate_extern_var_bounds): Likewise.
16287 (chkp_get_bounds_for_decl_addr): Likewise.
16288 * tree-chrec.c (chrec_apply): Likewise.
16289 * tree-data-ref.c (dump_data_reference): Likewise.
16290 (dump_subscript): Likewise.
16291 (dump_data_dependence_relation): Likewise.
16292 (analyze_overlapping_iterations): Likewise.
16293 * tree-inline.c (expand_call_inline): Likewise.
16294 (tree_function_versioning): Likewise.
16295 * tree-into-ssa.c (dump_defs_stack): Likewise.
16296 (dump_currdefs): Likewise.
16297 (dump_names_replaced_by): Likewise.
16298 (dump_update_ssa): Likewise.
16299 (update_ssa): Likewise.
16300 * tree-object-size.c (pass_object_sizes::execute): Likewise.
16301 * tree-parloops.c (build_new_reduction): Likewise.
16302 (try_create_reduction_list): Likewise.
16303 (ref_conflicts_with_region): Likewise.
16304 (oacc_entry_exit_ok_1): Likewise.
16305 (oacc_entry_exit_single_gang): Likewise.
16306 * tree-pretty-print.h: Likewise.
16307 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
16308 (get_scalar_evolution): Likewise.
16309 (add_to_evolution): Likewise.
16310 (get_loop_exit_condition): Likewise.
16311 (analyze_evolution_in_loop): Likewise.
16312 (analyze_initial_condition): Likewise.
16313 (analyze_scalar_evolution): Likewise.
16314 (instantiate_scev): Likewise.
16315 (number_of_latch_executions): Likewise.
16316 (gather_chrec_stats): Likewise.
16317 (final_value_replacement_loop): Likewise.
16318 (scev_const_prop): Likewise.
16319 * tree-sra.c (dump_access): Likewise.
16320 (disqualify_candidate): Likewise.
16321 (create_access): Likewise.
16322 (reject): Likewise.
16323 (maybe_add_sra_candidate): Likewise.
16324 (create_access_replacement): Likewise.
16325 (analyze_access_subtree): Likewise.
16326 (analyze_all_variable_accesses): Likewise.
16327 (sra_modify_assign): Likewise.
16328 (initialize_constant_pool_replacements): Likewise.
16329 (find_param_candidates): Likewise.
16330 (decide_one_param_reduction): Likewise.
16331 (replace_removed_params_ssa_names): Likewise.
16332 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
16333 * tree-ssa-copy.c (dump_copy_of): Likewise.
16334 (copy_prop_visit_cond_stmt): Likewise.
16335 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
16336 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
16337 (record_equivalences_from_stmt): Likewise.
16338 * tree-ssa-dse.c (compute_trims): Likewise.
16339 (delete_dead_call): Likewise.
16340 (delete_dead_assignment): Likewise.
16341 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
16342 (forward_propagate_into_cond): Likewise.
16343 (pass_forwprop::execute): Likewise.
16344 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
16345 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
16347 (move_computations_worker): Likewise.
16348 (execute_sm): Likewise.
16349 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
16350 (remove_exits_and_undefined_stmts): Likewise.
16351 (remove_redundant_iv_tests): Likewise.
16352 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
16353 (adjust_iv_update_pos): Likewise.
16354 * tree-ssa-math-opts.c (bswap_replace): Likewise.
16355 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
16356 (value_replacement): Likewise.
16357 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
16358 * tree-ssa-pre.c (print_pre_expr): Likewise.
16359 (get_representative_for): Likewise.
16360 (create_expression_by_pieces): Likewise.
16361 (insert_into_preds_of_block): Likewise.
16362 (eliminate_insert): Likewise.
16363 (eliminate_dom_walker::before_dom_children): Likewise.
16364 (eliminate): Likewise.
16365 (remove_dead_inserted_code): Likewise.
16366 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
16367 * tree-ssa-reassoc.c (get_rank): Likewise.
16368 (eliminate_duplicate_pair): Likewise.
16369 (eliminate_plus_minus_pair): Likewise.
16370 (eliminate_not_pairs): Likewise.
16371 (undistribute_ops_list): Likewise.
16372 (eliminate_redundant_comparison): Likewise.
16373 (update_range_test): Likewise.
16374 (optimize_range_tests_var_bound): Likewise.
16375 (optimize_vec_cond_expr): Likewise.
16376 (rewrite_expr_tree): Likewise.
16377 (rewrite_expr_tree_parallel): Likewise.
16378 (linearize_expr): Likewise.
16379 (break_up_subtract): Likewise.
16380 (linearize_expr_tree): Likewise.
16381 (attempt_builtin_powi): Likewise.
16382 (attempt_builtin_copysign): Likewise.
16383 (transform_stmt_to_copy): Likewise.
16384 (transform_stmt_to_multiply): Likewise.
16385 (dump_ops_vector): Likewise.
16386 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
16387 (print_scc): Likewise.
16388 (set_ssa_val_to): Likewise.
16389 (visit_reference_op_store): Likewise.
16390 (visit_use): Likewise.
16391 (sccvn_dom_walker::before_dom_children): Likewise.
16392 (run_scc_vn): Likewise.
16393 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
16395 (expr_hash_elt::print): Likewise.
16396 (const_and_copies::pop_to_marker): Likewise.
16397 (const_and_copies::record_const_or_copy_raw): Likewise.
16398 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
16399 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
16400 (dump_predicates): Likewise.
16401 (find_uninit_use): Likewise.
16402 (warn_uninitialized_phi): Likewise.
16403 (pass_late_warn_uninitialized::execute): Likewise.
16404 * tree-ssa.c (verify_vssa): Likewise.
16405 (verify_ssa): Likewise.
16406 (maybe_optimize_var): Likewise.
16407 * tree-vrp.c (dump_value_range): Likewise.
16408 (dump_all_value_ranges): Likewise.
16409 (dump_asserts_for): Likewise.
16410 (register_edge_assert_for_2): Likewise.
16411 (vrp_visit_cond_stmt): Likewise.
16412 (vrp_visit_switch_stmt): Likewise.
16413 (vrp_visit_stmt): Likewise.
16414 (vrp_visit_phi_node): Likewise.
16415 (simplify_cond_using_ranges_1): Likewise.
16416 (fold_predicate_in): Likewise.
16417 (evrp_dom_walker::before_dom_children): Likewise.
16418 (evrp_dom_walker::push_value_range): Likewise.
16419 (evrp_dom_walker::pop_value_range): Likewise.
16420 (execute_early_vrp): Likewise.
16422 2017-05-16 Richard Biener <rguenther@suse.de>
16424 * dwarf2out.c (loc_list_from_tree_1): Do not create
16425 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
16427 2017-05-16 Richard Biener <rguenther@suse.de>
16429 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
16431 (note_variable_value_in_expr): If we resolved the decl ref
16432 do not push to the stack.
16434 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
16436 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
16437 operations in fast-math mode.
16438 (vaddq_f16): Likewise.
16439 (vmul_f16): Likewise.
16440 (vmulq_f16): Likewise.
16441 (vsub_f16): Likewise.
16442 (vsubq_f16): Likewise.
16443 * config/arm/neon.md (add<mode>3): New.
16445 (fma:<VH:mode>3): New. Also remove outdated comment.
16448 2017-05-16 Martin Liska <mliska@suse.cz>
16452 * ipa-devirt.c (warn_types_mismatch): Fix typo.
16453 (odr_types_equivalent_p): Likewise.
16455 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
16457 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
16459 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
16462 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
16463 non-interunit SSE move alternatives with '?'.
16464 (zero-extendsidi peephole2): New peephole to skip intermediate
16465 general register in SSE zero-extend sequence.
16467 2017-05-15 Jeff Law <law@redhat.com>
16469 * reorg.c (relax_delay_slots): Create a new variable to hold
16470 the temporary target rather than clobbering TARGET_LABEL.
16472 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
16473 missing argument to extract_bit_field call.
16474 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
16476 2017-05-15 Martin Liska <mliska@suse.cz>
16479 * gcc.c (process_command): Do not allow empty argument of -o option.
16481 2017-05-15 Renlin Li <renlin.li@arm.com>
16483 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
16484 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
16485 * config/aarch64/constraints.md (Usf): Add long call check.
16486 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
16487 (call_value): Likewise.
16488 (sibcall): Likewise.
16489 (sibcall_value): Likewise.
16491 (call_value_insn): New.
16492 (sibcall_insn): Update rtx pattern.
16493 (sibcall_value_insn): Likewise.
16494 (call_internal): Remove.
16495 (call_value_internal): Likewise.
16496 (sibcall_internal): Likewise.
16497 (sibcall_value_internal): Likewise.
16498 (call_reg): Likewise.
16499 (call_symbol): Likewise.
16500 (call_value_reg): Likewise.
16501 (call_value_symbol): Likewise.
16503 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
16506 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
16508 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
16510 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
16511 compatible with CCGOCmode and with CCZmode.
16513 2017-05-14 Martin Sebor <msebor@redhat.com>
16515 PR middle-end/77671
16516 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
16517 (gimple_fold_builtin_snprintf): Same.
16518 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
16519 (gimple_fold_builtin_snprintf): Same.
16520 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
16521 of character types.
16522 (is_call_safe): New function.
16523 (try_substitute_return_value): Call it.
16524 (try_simplify_call): New function.
16525 (pass_sprintf_length::handle_gimple_call): Call it.
16527 2017-05-14 Martin Sebor <msebor@redhat.com>
16529 PR middle-end/80669
16530 * builtins.c (expand_builtin_stpncpy): Simplify.
16532 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
16534 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
16535 * config/i386/i386.h
16536 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
16537 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
16538 (struct machine_function): Add new members call_ms2sysv,
16539 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
16540 (struct machine_frame_state): New fields sp_realigned and
16541 sp_realigned_offset.
16542 * config/i386/i386.c
16543 (enum xlogue_stub): New enum.
16544 (enum xlogue_stub_sets): New enum.
16545 (class xlogue_layout): New class.
16546 (struct ix86_frame): New fields stack_realign_allocate_offset,
16547 stack_realign_offset and outlined_save_offset. Modify comments to
16548 detail stack layout when using out-of-line stubs.
16549 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
16550 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
16551 -mcall-ms2sysv-xlogues.
16552 (stub_managed_regs): New static variable.
16553 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
16554 registers managed by out-of-line stub.
16555 (disable_call_ms2sysv_xlogues): New function.
16556 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
16557 m->call_ms2sysv when appropriate and compute frame layout for
16559 (sp_valid_at, fp_valid_at): New inline functions.
16560 (choose_basereg): New function.
16561 (choose_baseaddr): Add align parameter, use choose_basereg and modify
16563 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
16564 Use align parameter of choose_baseaddr to generated aligned SSE movs
16566 (pro_epilogue_adjust_stack): Modify to track
16567 machine_frame_state::sp_realigned.
16568 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
16569 (ix86_nsaved_sseregs): Likewise.
16570 (ix86_emit_save_regs): Likewise.
16571 (ix86_emit_save_regs_using_mov): Likewise.
16572 (ix86_emit_save_sse_regs_using_mov): Likewise.
16573 (get_scratch_register_on_entry): Likewise.
16574 (gen_frame_set): New function.
16575 (gen_frame_load): Likewise.
16576 (gen_frame_store): Likewise.
16577 (emit_outlined_ms2sysv_save): Likewise.
16578 (emit_outlined_ms2sysv_restore): Likewise.
16579 (ix86_expand_prologue): Modify stack re-alignment code and call
16580 emit_outlined_ms2sysv_save when appropriate.
16581 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
16582 parameter rtx_insn *insn, which allows the function to be used to only
16583 generate the notes.
16584 (ix86_expand_epilogue): Modify validity checks of frame and stack
16585 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
16586 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
16587 * config/i386/predicates.md
16588 (save_multiple): New predicate.
16589 (restore_multiple): Likewise.
16590 * config/i386/sse.md
16591 (save_multiple<mode>): New pattern.
16592 (save_multiple_realign<mode>): Likewise.
16593 (restore_multiple<mode>): Likewise.
16594 (restore_multiple_and_return<mode>): Likewise.
16595 (restore_multiple_leave_return<mode>): Likewise.
16596 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
16598 2017-05-14 Julia Koval <julia.koval@intel.com>
16600 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
16601 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
16602 (__builtin_ia32_xsetbv): New builtins.
16603 * config/i386/i386.c (ix86_expand_special_args_builtin):
16605 (ix86_expand_builtin): Special expand for new intrinsics.
16606 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
16607 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
16608 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
16610 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16612 * cfganal.c (inverted_post_order_compute): Change argument type
16614 * cfganal.h (inverted_post_order_compute): Adjust prototype.
16615 * df-core.c (rest_of_handle_df_initialize): Adjust.
16616 (rest_of_handle_df_finish): Likewise.
16617 (df_analyze_1): Likewise.
16618 (df_analyze): Likewise.
16619 (loop_inverted_post_order_compute): Change argument to be a vec *.
16620 (df_analyze_loop): Adjust.
16621 (df_get_n_blocks): Likewise.
16622 (df_get_postorder): Likewise.
16623 * df.h (struct df_d): Change field to be a vec.
16624 * lcm.c (compute_laterin): Adjust.
16625 (compute_available): Likewise.
16626 * lra-lives.c (lra_create_live_ranges_1): Likewise.
16627 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
16628 * tree-ssa-pre.c (compute_antic): Likewise.
16630 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16632 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
16633 (depth_first_search::depth_first_search): Change structure init
16634 function to this constructor.
16635 (depth_first_search::add_bb): Rename function to this member.
16636 (depth_first_search::execute): Likewise.
16637 (flow_dfs_compute_reverse_finish): Adjust.
16639 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16641 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
16642 (longest_simple_path): Likewise.
16643 * shrink-wrap.c (spread_components): Likewise.
16644 (disqualify_problematic_components): Likewise.
16645 (emit_common_heads_for_components): Likewise.
16646 (emit_common_tails_for_components): Likewise.
16647 (insert_prologue_epilogue_for_components): Likewise.
16649 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16651 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
16654 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16656 * df-core.c (df_set_blocks): Start using auto_bitmap.
16657 (df_compact_blocks): Likewise.
16658 * df-problems.c (df_rd_confluence_n): Likewise.
16659 * df-scan.c (df_insn_rescan_all): Likewise.
16660 (df_process_deferred_rescans): Likewise.
16661 (df_update_entry_block_defs): Likewise.
16662 (df_update_exit_block_uses): Likewise.
16663 (df_entry_block_bitmap_verify): Likewise.
16664 (df_exit_block_bitmap_verify): Likewise.
16665 (df_scan_verify): Likewise.
16666 * lra-constraints.c (lra_constraints): Likewise.
16667 (undo_optional_reloads): Likewise.
16668 (lra_undo_inheritance): Likewise.
16669 * lra-remat.c (calculate_gen_cands): Likewise.
16670 (do_remat): Likewise.
16671 * lra-spills.c (assign_spill_hard_regs): Likewise.
16672 (spill_pseudos): Likewise.
16673 * tree-ssa-pre.c (bitmap_set_and): Likewise.
16674 (bitmap_set_subtract_values): Likewise.
16676 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16678 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
16679 management with auto_bitmap.
16680 (fix_inter_tick): Likewise.
16681 (fix_recovery_deps): Likewise.
16682 * ira.c (add_store_equivs): Likewise.
16683 (find_moveable_pseudos): Likewise.
16684 (split_live_ranges_for_shrink_wrap): Likewise.
16685 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
16686 (rtx_reuse_manager::seen_def_p): Likewise.
16687 (rtx_reuse_manager::set_seen_def): Likewise.
16688 * print-rtl.h (class rtx_reuse_manager): Likewise.
16690 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16692 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
16694 (migrate_btr_def): Likewise.
16695 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
16696 * df-core.c (loop_post_order_compute): Likewise.
16697 (loop_inverted_post_order_compute): Likewise.
16698 * hsa-common.h: Likewise.
16699 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
16700 * init-regs.c (initialize_uninitialized_regs): Likewise.
16701 * ipa-inline.c (resolve_noninline_speculation): Likewise.
16702 (inline_small_functions): Likewise.
16703 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
16704 * ira.c (combine_and_move_insns): Likewise.
16705 (build_insn_chain): Likewise.
16706 * loop-invariant.c (find_invariants): Likewise.
16707 * lower-subreg.c (propagate_pseudo_copies): Likewise.
16708 * predict.c (tree_predict_by_opcode): Likewise.
16709 (predict_paths_leading_to): Likewise.
16710 (predict_paths_leading_to_edge): Likewise.
16711 (estimate_loops_at_level): Likewise.
16712 (estimate_loops): Likewise.
16713 * shrink-wrap.c (try_shrink_wrapping): Likewise.
16714 (spread_components): Likewise.
16715 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
16716 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
16717 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
16718 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
16719 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
16720 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
16721 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
16722 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
16723 (mark_threaded_blocks): Likewise.
16724 (thread_through_all_blocks): Likewise.
16725 * tree-ssa.c (verify_ssa): Likewise.
16726 (execute_update_addresses_taken): Likewise.
16727 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
16729 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16731 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
16733 (post_order_compute): Likewise.
16734 (inverted_post_order_compute): Likewise.
16735 (pre_and_rev_post_order_compute_fn): Likewise.
16737 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16739 * genrecog.c (int_set::int_set): Explicitly construct our
16740 auto_vec base class.
16741 * vec.h (auto_vec::auto_vec): New constructor.
16743 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16745 * bitmap.h (class auto_bitmap): New constructor taking
16746 bitmap_obstack * argument.
16748 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16750 * bitmap.h (class auto_bitmap): Change type of m_bits to
16751 bitmap_head, and adjust ctor / dtor and member operators.
16753 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
16755 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
16756 when returned register mode doesn't match original mode.
16758 2017-05-12 Jeff Law <law@redhat.com>
16759 Jakub Jelinek <jakub@redhat.com>
16761 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
16762 we look for cc setter after the compare-elim changes.
16763 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
16764 within the vector to match what compare-elim now expects.
16765 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
16766 (xorsi3_flags, one_cmplsi2_flags): Likewise.
16768 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
16769 after the compare-elim changes.
16770 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
16771 the vector to match what compare-elim now expects.
16772 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
16773 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
16774 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
16775 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
16776 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
16778 * config/visium/visium.c (single_set_and_flags): Fix where
16779 we look for cc setter after the compare-elim changes.
16780 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
16781 with the vector to match what compare-elim now expects.
16782 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
16783 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
16784 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
16785 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
16786 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
16787 (neg<mode>2_insn_set_overflow): Likewise.
16789 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
16791 PR middle-end/79794
16792 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
16793 maybe_expand_insn call, set ops[0].target. If still set after call,
16794 set alt_rtl. Add extra arg to recursive calls.
16795 (extract_bit_field): Add alt_rtl argument. Pass to
16797 * expmed.h (extract_bit_field): Fix prototype.
16798 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
16799 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
16800 to extract_bit_field_calls.
16801 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
16802 Pass alt_rtl to extract_bit_field calls.
16803 * calls.c (store_unaligned_arguments_into_psuedos)
16804 load_register_parameters): Pass extra NULL to extract_bit_field calls.
16805 * optabs.c (maybe_legitimize_operand): Clear op->target when call
16807 * optabs.h (struct expand_operand): Add target bitfield.
16809 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
16811 * compare-elim.c (try_eliminate_compare): Canonicalize
16812 operation with embedded compare to
16813 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
16814 (set (reg) (operation)].
16816 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
16818 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
16821 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
16822 cost of adding a carry flag for ADC instruction.
16823 [case MINUS]: Ignore the cost of subtracting a carry flag
16824 for SBB instruction.
16826 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
16828 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
16830 * config/rs6000/bmiintrin.h: New file.
16831 * config/rs6000/bmi2intrin.h: New file.
16832 * config/rs6000/x86intrin.h: New file.
16834 2017-05-12 Jeff Law <law@redhat.com>
16836 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
16839 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
16841 PR middle-end/80707
16842 * tree-cfg.c: Remove cfg edges of unreachable case statements.
16844 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
16846 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
16847 early expansion of vector divide builtins.
16848 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
16849 builtins identified as having unsigned arguments.
16851 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
16853 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
16854 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
16855 expansion of vector logical operations (and, andc, or, xor,
16858 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
16860 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
16861 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
16863 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
16865 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
16866 early GIMPLE expansion of vector multiplies.
16868 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
16870 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
16871 TARGET_HAVE_MOVT conditional.
16872 (movt splitter): Likewise.
16874 2017-05-12 Richard Biener <rguenther@suse.de>
16876 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
16877 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
16878 Fold all stmts not inplace.
16880 2017-05-12 Richard Biener <rguenther@suse.de>
16882 PR tree-optimization/80713
16883 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
16884 inserted_exprs bit for not removed stmts.
16886 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
16888 PR middle-end/69921
16889 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
16890 parallelized" attribute for parallelized OpenACC kernels.
16891 * omp-offload.c (execute_oacc_device_lower): Use it.
16893 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
16894 Set "oacc kernels" attribute.
16895 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
16896 parameter. Adjust all users.
16897 (oacc_fn_attrib_kernels_p): Remove function.
16898 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
16899 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
16900 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
16901 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
16902 assert "oacc kernels" attribute is set.
16904 2017-05-11 Carl Love <cel@us.ibm.com>
16906 * config/rs6000/rs6000-c: Add support for built-in functions
16907 vector unsigned char vec_popcnt (vector signed char)
16908 vector unsigned char vec_popcnt (vector unsigned char)
16909 vector unsigned short vec_popcnt (vector signed short)
16910 vector unsigned short vec_popcnt (vector unsigned short)
16911 vector unsigned int vec_popcnt (vector signed int)
16912 vector unsigned int vec_popcnt (vector unsigned int)
16913 vector unsigned long long vec_popcnt (vector signed long long)
16914 vector unsigned long long vec_popcnt (vector unsigned long long)
16915 vector signed long long vec_slo (vector signed long long,
16916 vector signed char)
16917 vector signed long long vec_slo (vector signed long long,
16918 vector unsigned char)
16919 vector unsigned long long vec_slo (vector unsigned long long,
16920 vector signed char)
16921 vector unsigned long long vec_slo (vector unsigned long long,
16922 vector unsigned char)
16923 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
16924 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
16925 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
16926 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
16927 * doc/extend.texi: Update the built-in documentation file for the
16928 new built-in functions.
16930 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
16932 * attribs.h (sorted_attr_string): Move machine independent
16933 functions for target clone support from the i386 port to common
16934 code. Rename ix86_function_versions to common_function_versions.
16935 Rename make_name to make_unique_name.
16936 (common_function_versions): Likewise.
16937 (make_unique_name): Likewise.
16938 (make_dispatcher_decl): Likewise.
16939 (is_function_default_version): Likewise.
16940 * attribs.c (attr_strcmp): Likewise.
16941 (sorted_attr_string): Likewise.
16942 (common_function_versions): Likewise.
16943 (make_unique_name): Likewise.
16944 (make_dispatcher_decl): Likewise.
16945 (is_function_default_version): Likewise.
16946 * config/i386/i386.c (attr_strcmp): Likewise.
16947 (sorted_attr_string): Likewise.
16948 (ix86_function_versions): Likewise.
16949 (make_name): Likewise.
16950 (make_dispatcher_decl): Likewise.
16951 (is_function_default_version): Likewise.
16952 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
16954 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16957 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
16958 Account for direct move costs for vec_construct of integer
16961 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
16964 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
16965 (UNSPEC_STX_ATOMIC): Ditto.
16966 (loaddi_via_sse): New insn.
16967 (storedi_via_sse): Ditto.
16968 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
16969 Update corresponding peephole2 patterns.
16970 (atomic_storedi_fpu): Ditto.
16972 2017-05-11 Julia Koval <julia.koval@intel.com>
16974 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
16975 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
16977 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
16978 (__builtin_ia32_rsqrt14ss_mask): New builtins.
16979 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
16981 2017-05-11 Nathan Sidwell <nathan@acm.org>
16983 * graphite-poly.c: Include dumpfile.h.
16985 * dumpfle.h (dump_function): Declare here ...
16986 * tree-dump.h (dump_function): ... not here.
16987 * dumpfile.c: #include tree-cfg.h.
16988 (dump_function): Move here from ...
16989 * tree-dump.c (dump_function): ... here.
16990 * gimplify.c: #include splay-tree.h, not tree-dump.h.
16991 * graphite-poly.c: Don't include tree-dump.h.
16992 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
16993 * print-tree.c: Likewise.
16994 * stor-layout.c: Likewise.
16995 * tree-nested.c: Likewise.
16997 * dumpfile.c (dump_start): Use TDF_FLAGS.
16998 (dump_enable_all): Fix TDF_KIND check thinko.
17000 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
17002 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17003 array entries to represent two legal parameterizations of the
17004 overloaded __builtin_cmpb function, as represented by the
17005 P6_OV_BUILTIN_CMPB constant.
17006 (altivec_resolve_overloaded_builtin): Add special case handling
17007 for the __builtin_cmpb function, as represented by the
17008 P6_OV_BUILTIN_CMPB constant.
17009 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
17010 (BU_P6_64BIT_2): New macro.
17011 (BU_P6_OVERLOAD_2): New macro
17012 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
17013 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
17014 (CMPB): Add overload support to represent both 32-bit and 64-bit
17015 compare-bytes function.
17016 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
17017 support for TARGET_CMPB.
17018 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
17019 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
17020 documentation of the __builtin_cmpb overloaded built-in function.
17022 2017-05-11 Richard Biener <rguenther@suse.de>
17024 PR tree-optimization/80705
17025 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
17026 bases are not vectorizable.
17028 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17030 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
17031 when counting register pressure.
17033 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17035 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
17036 (iv_ca_more_deps): Renamed to ...
17037 (iv_ca_compare_deps): ... this.
17038 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
17040 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17042 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
17044 (determine_group_iv_costs): ... here.
17045 (find_inv_vars_cb): Record inv var if it's not recorded before.
17047 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17049 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
17050 (get_shiftadd_cost): Ditto.
17052 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17054 * tree-ssa-address.c: Include header file.
17055 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
17057 (add_to_parts): Refactor.
17058 (addr_to_parts): New parameter. Update use of move_hint_to_base.
17059 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
17062 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17064 PR tree-optimization/53090
17065 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
17067 (extract_cond_operands): Detect condition with IV on both sides
17068 and return COMP_IV_EXPR_2.
17069 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
17070 (rewrite_use_compare): Simplify by removing call to function
17071 extract_cond_operands.
17073 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17075 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
17076 (extract_cond_operands): Detect condition comparing against non-
17077 invariant bound and return appropriate enum value.
17078 (find_interesting_uses_cond): Update use of extract_cond_operands.
17079 Handle its return value accordingly.
17080 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
17082 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17084 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
17085 nonlinear iv_use computation in loop invariant sensitive way.
17087 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17089 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
17090 (find_iv_candidates): Call relate_compare_use_with_all_cands.
17092 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17094 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
17095 (dump_cand): Support iv_cand.inv_exprs.
17096 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
17098 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
17101 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17103 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
17105 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
17106 as local function. Include necessary header files.
17107 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
17109 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17111 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
17113 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17115 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
17116 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
17117 RSHIFT_EXPR and BIT_NOT_EXPR.
17119 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17121 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
17122 (adjust_setup_cost): New parameter supporting round up adjustment.
17123 (struct address_cost_data): Delete.
17124 (force_expr_to_var_cost): Don't bound cost with spill_cost.
17125 (split_address_cost, ptr_difference_cost): Delete.
17126 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
17127 (struct ainc_cost_data): New struct.
17128 (get_address_cost_ainc): New function.
17129 (get_address_cost, get_computation_cost): Reimplement.
17130 (determine_group_iv_cost_address): Record inv_expr for all uses of
17132 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
17133 (iv_ca_has_deps): Reimplemented to ...
17134 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
17136 (iv_ca_extend): Call iv_ca_more_deps.
17138 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17140 * tree-ssa-address.c (struct mem_address): Move to header file.
17141 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
17142 * tree-ssa-address.h (struct mem_address): Move from C file.
17143 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
17145 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17147 * tree-affine.h (aff_combination_type): New interface.
17148 (aff_combination_zero_p): Remove static.
17149 (aff_combination_const_p): New interface.
17150 (aff_combination_singleton_var_p): New interfaces.
17152 2017-05-11 Richard Biener <rguenther@suse.de>
17154 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
17155 Skip unreachable blocks and destinations.
17156 (eliminate): Move stmt removal and fixup ...
17157 (fini_eliminate): ... here. Skip inserted exprs.
17158 (pass_pre::execute): Move fini_pre after fini_eliminate.
17159 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
17160 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
17161 PRE to get rid of dead code that has invalid SSA form and
17162 split critical edges again.
17164 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17166 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
17168 2017-05-11 Richard Biener <rguenther@suse.de>
17170 * passes.c (execute_function_todo): Verify loops if they are
17171 said to be up-to-date.
17172 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
17173 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
17175 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
17178 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
17179 handle calling assemble_external ourself.
17182 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
17183 modes with zero size. Enhance comment.
17185 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17187 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
17188 built-ins for vec_xl and vec_xst with short and char pointer
17191 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
17193 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
17194 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
17195 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
17196 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
17197 (_mm_maskz_min_round_ss): New intrinsics.
17198 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
17199 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
17200 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
17201 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
17202 (__builtin_ia32_minss_mask_round): New builtins.
17203 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
17204 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
17205 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
17207 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
17208 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
17210 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
17213 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
17215 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
17216 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
17217 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
17218 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
17219 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
17220 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
17221 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
17222 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
17223 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
17224 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
17225 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
17226 (__builtin_ia32_mulss_mask_round): New builtins.
17227 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
17228 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
17229 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
17231 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
17232 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
17234 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
17237 2017-05-10 Julia Koval <julia.koval@intel.com>
17239 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
17240 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
17241 (_mm256_setr_m128i): New intrinsics.
17243 2017-05-10 Julia Koval <julia.koval@intel.com>
17245 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
17246 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
17247 (_mm_maskz_rcp14_ss): New intrinsics.
17248 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
17249 (__builtin_ia32_rcp14ss_mask): New builtins.
17250 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
17252 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
17254 PR tree-optimization/51513
17255 * tree-cfg.c (gimple_seq_unreachable_p): New function.
17256 (assert_unreachable_fallthru_edge_p): Use it.
17257 (group_case_labels_stmt): Likewise.
17258 * tree-cfg.h: Prototype it.
17259 * stmt.c: Include cfghooks.h and tree-cfg.h.
17260 (emit_case_dispatch_table) <gap_label>: New local variable.
17261 Use it to fill dispatch table gaps.
17262 Test for default_label before updating probabilities.
17263 (expand_case) <default_label>: Remove unneeded initialization.
17264 Test for unreachable default case statement and remove its edge.
17265 Set default_label accordingly.
17266 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
17268 2017-05-10 Carl Love <cel@us.ibm.com>
17270 * config/rs6000/rs6000-c: Add support for built-in functions
17271 vector signed char vec_neg (vector signed char)
17272 vector signed short int vec_neg (vector short int)
17273 vector signed int vec_neg (vector signed int)
17274 vector signed long long vec_neg (vector signed long long)
17275 vector float vec_neg (vector float)
17276 vector double vec_neg (vector double)
17277 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
17279 * config/rs6000/altivec.h: Add define for vec_neg
17280 * doc/extend.texi: Update the built-in documentation for the
17281 new built-in functions.
17283 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17285 PR tree-optimization/77644
17286 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
17288 2017-05-10 Nathan Sidwell <nathan@acm.org>
17290 * dumpfile.h (TDI_lang_all): New.
17291 (TDF_KIND): New. Renumber others
17292 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
17294 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
17296 (get_dump_file_name): Adjust suffix generation.
17297 (dump_enable_all): Use TDF_KIND.
17298 * doc/invoke.texi (-fdump-lang-all): Document.
17300 * dumpfile.h: Tabify.
17302 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
17305 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
17306 Move member access before delete.
17308 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
17310 * tree-inline.c (expand_call_inline): Split block at stmt
17313 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
17316 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
17317 are now unused after splitting mov{sf,sd}_hardfloat.
17318 (f32_lr2): Likewise.
17319 (f32_lm): Likewise.
17320 (f32_lm2): Likewise.
17321 (f32_li): Likewise.
17322 (f32_li2): Likewise.
17323 (f32_lv): Likewise.
17324 (f32_sr): Likewise.
17325 (f32_sr2): Likewise.
17326 (f32_sm): Likewise.
17327 (f32_sm2): Likewise.
17328 (f32_si): Likewise.
17329 (f32_si2): Likewise.
17330 (f32_sv): Likewise.
17331 (f32_dm): Likewise.
17332 (f32_vsx): Likewise.
17333 (f32_av): Likewise.
17334 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
17335 For movsf, order stores so the VSX stores occur before the GPR
17336 store which encourages the register allocator to use a traditional
17337 FPR instead of a GPR. For movsd, order the stores so that the GPR
17338 store comes before the VSX stores to allow the power6 to work.
17339 This is due to the power6 not having a 32-bit integer store
17340 instruction from a FPR.
17341 (movsf_hardfloat): Likewise.
17342 (movsd_hardfloat): Likewise.
17344 2017-05-09 Martin Sebor <msebor@redhat.com>
17346 PR translation/80280
17347 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
17350 PR translation/80280
17351 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
17352 data member added in r247778.
17353 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
17355 2017-05-09 Nathan Sidwell <nathan@acm.org>
17357 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
17359 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
17362 2017-05-09 Marek Polacek <polacek@redhat.com>
17364 * doc/invoke.texi: Fix typo.
17366 2017-05-09 Richard Biener <rguenther@suse.de>
17368 * tree-vrp.c (vrp_val_is_max): Adjust comment.
17369 (vrp_val_is_min): Likewise.
17370 (set_value_range_to_value): Likewise.
17371 (set_value_range_to_nonnegative): Likewise.
17372 (gimple_assign_nonzero_p): Likewise.
17373 (gimple_stmt_nonzero_p): Likewise.
17374 (vrp_int_const_binop): Likewise. Remove unreachable case.
17375 (adjust_range_with_scev): Adjust comments.
17376 (compare_range_with_value): Likewise.
17377 (extract_range_from_phi_node): Likewise.
17378 (test_for_singularity): Likewise.
17380 2017-05-09 Richard Biener <rguenther@suse.de>
17382 * tree-vrp.c (get_single_symbol): Add assert that we don't
17383 get overflowed constants as invariant part.
17384 (compare_values_warnv): Add comment before the TREE_NO_WARNING
17385 checks. Use wi::cmp instead of recursing for integer constants.
17386 (compare_values): Just ignore whether we assumed undefined
17387 overflow instead of failing the compare.
17388 (extract_range_for_var_from_comparison_expr): Add comment before the
17389 TREE_NO_WARNING sets.
17390 (test_for_singularity): Likewise.
17391 (extract_range_from_comparison): Do not disable optimization
17392 when we assumed undefined overflow.
17393 (extract_range_basic): Remove init of unused var.
17395 2017-05-09 Richard Biener <rguenther@suse.de>
17397 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
17398 (extract_range_from_multiplicative_op_1): Adjust.
17399 (extract_range_from_binary_expr_1): Use int_const_binop.
17401 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
17404 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
17405 rs6000_store_data_bypass_p in seven define_bypass directives and
17406 in several comments.
17407 * config/rs6000/rs6000-protos.h: Add prototype for
17408 rs6000_store_data_bypass_p function.
17409 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
17410 function implements slightly different (rs6000-specific) semantics
17411 than store_data_bypass_p, returning false rather than aborting
17412 with assertion error when arguments do not satisfy the
17413 requirements of store data bypass.
17414 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
17415 rs6000_store_data_bypass_p.
17417 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
17419 * config/xtensa/xtensa-protos.h
17420 (xtensa_initial_elimination_offset): New declaration.
17421 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
17422 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
17423 macro definition, add case for FRAME_POINTER_REGNUM when
17424 FRAME_GROWS_DOWNWARD.
17425 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
17426 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
17427 xtensa_initial_elimination_offset.
17429 2017-05-08 Nathan Sidwell <nathan@acm.org>
17431 * doc/invoke.texi: Alphabetize -fdump options.
17433 2017-05-08 Martin Sebor <msebor@redhat.com>
17435 PR translation/80280
17436 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
17438 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
17440 * target.def (compute_frame_layout): New optional target hook.
17441 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
17442 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
17443 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
17445 * reload1.c (verify_initial_elim_offsets): Likewise.
17446 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
17447 (use_simple_return_p): Call arm_compute_frame_layout if needed.
17448 (arm_get_frame_offsets): Split up into this ...
17449 (arm_compute_frame_layout): ... and this function.
17451 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
17453 * config/aarch64/constraints.md (Usa): New constraint.
17454 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
17456 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
17458 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
17459 with_multilib_list after it has been checked.
17461 2017-05-08 Richard Biener <rguenther@suse.de>
17463 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
17464 (bitmap_set_subtract_values): Likewise.
17466 2017-05-08 Richard Biener <rguenther@suse.de>
17468 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
17469 (gimple_assign_nonzero): ... this and remove strict_overflow_p
17471 (gimple_stmt_nonzero_warnv_p): Rename to ...
17472 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
17474 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
17475 (extract_range_basic): Adjust, do not disable propagation on
17476 strict overflow sensitive simplification.
17477 (vrp_visit_cond_stmt): Likewise.
17479 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
17481 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
17482 body size unconditionally.
17484 2017-05-07 Jeff Law <law@redhat.com>
17487 2017-05-06 Jeff Law <law@redhat.com>
17488 PR tree-optimization/78496
17489 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
17492 PR tree-optimization/78496
17493 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
17494 (simplify_stmt_using_ranges): Call it.
17495 (vrp_dom_walker::before_dom_children): Extract equivalences
17496 from an ASSERT_EXPR with an equality comparison against a
17499 2017-05-06 Jeff Law <law@redhat.com>
17501 PR tree-optimization/78496
17502 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
17505 PR tree-optimization/78496
17506 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
17507 (simplify_stmt_using_ranges): Call it.
17508 (vrp_dom_walker::before_dom_children): Extract equivalences
17509 from an ASSERT_EXPR with an equality comparison against a
17512 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
17514 * lra-constraints.c (lra_copy_reg_equiv): New function.
17515 (split_reg): Use it to copy equivalence information from the
17516 original register to the spill register.
17518 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
17520 PR rtl-optimization/75964
17521 * simplify-rtx.c (simplify_const_relational_operation): Remove
17522 invalid handling of comparisons of integer ABS.
17524 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
17526 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
17527 initialize to zero.
17528 (init_regs): Remove declaration.
17529 (function_arg_advance_32): Initialize error_p as boolean variable.
17531 2017-05-05 Nathan Sidwell <nathan@acm.org>
17533 * store-motion.c (remove_reachable_equiv_notes): Reformat long
17534 lines. Use for (;;).
17536 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17538 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
17539 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
17540 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
17541 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
17542 VF=2 that require versioning.
17544 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17546 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
17549 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17551 * diagnostic.h (diagnostic_override_option_index): Convert from
17552 macro to inline function.
17554 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17556 * diagnostic.c (last_module_changed_p): New function.
17557 (set_last_module): New function.
17558 (diagnostic_report_current_module): Convert macro usage to
17559 the above functions.
17560 * diagnostic.h (diagnostic_context::last_module): Strengthen
17561 from const line_map * to const line_map_ordinary *.
17562 (diagnostic_last_module_changed): Delete macro.
17563 (diagnostic_set_last_module): Delete macro.
17565 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17567 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
17568 with diagnostic_report_diagnostic.
17569 (diagnostic_n_impl_richloc): Likewise.
17570 * diagnostic.h (report_diagnostic): Delete macro.
17571 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
17572 with diagnostic_report_diagnostic.
17573 * substring-locations.c (format_warning_va): Likewise.
17575 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17577 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
17578 save/restor of format_spec. Move option-printing code to...
17579 (print_option_information): ...this new function, and
17580 reimplement by simply printing to the pretty_printer,
17581 rather than appending to the format string.
17583 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17585 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
17586 handling logic into...
17587 (update_effective_level_from_pragmas): ...this new function.
17589 2017-05-04 Andrew Waterman <andrew@sifive.com>
17591 * config/riscv/riscv.opt (mstrict-align): New option.
17592 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
17593 (SLOW_UNALIGNED_ACCESS): Define.
17594 (riscv_slow_unaligned_access): Declare.
17595 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
17597 (riscv_slow_unaligned_access): New variable.
17598 (rocket_tune_info): Set slow_unaligned_access to true.
17599 (optimize_size_tune_info): Set slow_unaligned_access to false.
17600 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
17601 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
17602 (riscv_option_override): Set riscv_slow_unaligned_access.
17603 * doc/invoke.texi: Add -mstrict-align to RISC-V.
17605 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
17607 * config/riscv/riscv.md: Unify indentation.
17609 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
17614 * config/rs6000/rs6000.md (u code attribute): Add FIX and
17616 (extendsi<mode>2): Add support for doing sign extension via
17617 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
17618 don't have ISA 3.0 instructions.
17619 (extendsi<mode>2 splitter): Likewise.
17620 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
17621 generate the normal insns since SImode can now go in vector
17622 registers. Disallow the special UNSPECs needed for previous
17623 machines to hide SImode being used. Add new insns
17624 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
17625 (fix_trunc<mode>si2_stfiwx): Likewise.
17626 (fix_trunc<mode>si2_internal): Likewise.
17627 (fixuns_trunc<mode>si2): Likewise.
17628 (fixuns_trunc<mode>si2_stfiwx): Likewise.
17629 (fctiw<u>z_<mode>_smallint): Likewise.
17630 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
17631 of floating point to 32-bit integer from doing a direct move to
17632 the GPR registers to do a store.
17633 (fctiwz_<mode>): Break long line.
17635 2017-05-05 Bin Cheng <bin.cheng@arm.com>
17637 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
17638 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
17639 (addr_list, addr_offset_valid_p): New.
17640 (split_address_groups): Check offset validity with above function.
17641 (gt-tree-ssa-loop-ivopts.h): Include header file.
17643 2017-05-05 Nathan Sidwell <nathan@acm.org>
17645 * config.gcc (arm*-*-*): Add missing 'fi'.
17647 2017-05-05 Steve Ellcey <sellcey@cavium.com>
17649 * doc/invoke.texi (-fopt-info): Explicitly say order of options
17650 included in -fopt-info does not matter.
17651 * doc/optinfo.texi (-fopt-info): Fix description of default
17652 behavour. Explicitly say order of options included in -fopt-info
17655 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
17657 * config.gcc: Allow combinations of aprofile and rmprofile values for
17658 --with-multilib-list.
17659 * config/arm/t-multilib: New file.
17660 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
17661 variables. Remove setting of ISA and floating-point ABI in
17662 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
17663 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
17664 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
17665 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
17667 * config/arm/t-rmprofile: Likewise except for the matches changes.
17668 * doc/install.texi (--with-multilib-list): Document the combination of
17669 aprofile and rmprofile values and warn about pitfalls in doing that.
17671 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
17673 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
17674 (movdi_aarch64): Likewise.
17676 2017-05-05 Jakub Jelinek <jakub@redhat.com>
17678 PR tree-optimization/80632
17679 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
17681 (build_arrays): Initialize it for virtual phis.
17682 (fix_phi_nodes): Use it for virtual phis.
17684 PR tree-optimization/80558
17685 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
17686 [x, y] op z into [x op, y op z] for op & or | if conditions
17689 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
17690 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
17693 * config/arm/arm.md (use_literal_pool): Remove.
17694 (64-bit immediate split): No longer takes cost into consideration
17695 if arm_disable_literal_pool is enabled.
17696 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
17697 used when arm_disable_literal_pool is enabled.
17698 (arm_max_const_double_inline_cost): Remove use of
17699 arm_disable_literal_pool.
17700 (push_minipool_fix): Add assert.
17701 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
17702 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
17703 (no_literal_pool_sf_immediate): New.
17705 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17707 PR tree-optimization/80613
17708 * tree-ssa-dce.c (propagate_necessity): Remove cases for
17709 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
17711 2017-05-05 Richard Biener <rguenther@suse.de>
17713 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
17715 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
17717 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
17718 of this flag from insn conditions due to removal from r247495.
17720 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
17722 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
17724 (arm_early_store_addr_dep_ptr): Likewise.
17725 * config/arm/aarch-common-protos.h
17726 (arm_early_load_addr_dep_ptr): Add prototype.
17727 (arm_early_store_addr_dep_ptr): Likewise.
17728 * config/arm/cortex-a53.md: Add new bypasses.
17730 2017-05-05 Jakub Jelinek <jakub@redhat.com>
17732 * tree.c (next_type_uid): Change type to unsigned.
17733 (type_hash_canon): Decrement back next_type_uid if
17734 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
17735 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
17738 2017-05-04 Martin Sebor <msebor@redhat.com>
17740 * builtins.c: Fix a trivial typo in a comment.
17742 PR middle-end/79234
17743 * builtins.c (check_sizes): Adjust to handle reading past the end.
17744 Avoid printing excessive upper bound of ranges. Use %E to print
17745 tree nodes instead of converting them to %wu.
17746 (expand_builtin_memchr): New function.
17747 (compute_dest_size): Rename...
17748 (compute_objsize): ...to this.
17749 (expand_builtin_memcpy): Adjust.
17750 (expand_builtin_mempcpy): Adjust.
17751 (expand_builtin_strcat): Adjust.
17752 (expand_builtin_strcpy): Adjust.
17753 (check_strncat_sizes): Adjust.
17754 (expand_builtin_strncat): Adjust.
17755 (expand_builtin_strncpy): Adjust and simplify.
17756 (expand_builtin_memset): Adjust.
17757 (expand_builtin_bzero): Adjust.
17758 (expand_builtin_memcmp): Adjust.
17759 (expand_builtin): Handle memcmp.
17760 (maybe_emit_chk_warning): Check strncat just once.
17762 2017-05-04 Martin Sebor <msebor@redhat.com>
17764 PR preprocessor/79214
17765 PR middle-end/79222
17766 PR middle-end/79223
17767 * builtins.c (check_sizes): Add inlining context and issue
17768 warnings even when -Wno-system-headers is set.
17769 (check_strncat_sizes): Same.
17770 (expand_builtin_strncat): Same.
17771 (expand_builtin_memmove): New function.
17772 (expand_builtin_stpncpy): Same.
17773 (expand_builtin): Handle memmove and stpncpy.
17775 2017-05-04 Bin Cheng <bin.cheng@arm.com>
17777 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
17778 which is not used any more.
17780 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
17782 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
17784 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
17786 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
17787 (cortexa53_tunings): Likewise.
17788 (cortexa57_tunings): Likewise.
17789 (cortexa72_tunings): Likewise.
17790 (cortexa73_tunings): Likewise.
17792 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
17794 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
17795 Set loop alignment to 8.
17797 2017-05-04 Martin Sebor <msebor@redhat.com>
17799 PR translation/80280
17800 * builtins.c (expand_builtin_object_size): Add missing quoting to
17801 %D and like directives.
17802 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
17803 (hsa_type_for_tree_type): Same.
17804 (verify_function_arguments): Same.
17805 * symtab.c (symbol_table::change_decl_assembler_name): Same.
17806 * varasm.c (get_section): Same.
17809 2017-05-04 Martin Sebor <msebor@redhat.com>
17811 PR translation/80280
17812 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
17814 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
17816 * config/aarch64/aarch64.c (generic_addrcost_table):
17817 Change HI/TI mode setting.
17819 2017-05-04 Martin Jambor <mjambor@suse.cz>
17821 PR tree-optimization/80622
17822 * tree-sra.c (comes_initialized_p): New function.
17823 (build_accesses_from_assign): Only set write lazily when
17824 comes_initialized_p is false.
17825 (analyze_access_subtree): Use comes_initialized_p.
17826 (propagate_subaccesses_across_link): Assert !comes_initialized_p
17827 instead of testing for PARM_DECL.
17829 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17831 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
17832 constraint on operand 0 to allow more general addressing modes.
17833 Adjust output template.
17834 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
17836 * config/aarch64/aarch64-protos.h
17837 (aarch64_address_valid_for_prefetch_p): Declare prototype.
17838 * config/aarch64/constraints.md (Dp): New address constraint.
17839 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
17842 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
17844 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
17845 update use of estimate_ipcp_clone_size_and_time.
17846 (estimate_local_effects): Update use of
17847 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
17848 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
17849 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
17850 Return nonspecialized time.
17852 2017-05-04 Richard Biener <rguenther@suse.de>
17854 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
17855 for the last VUSE which def dominates the PHI. Directly call
17857 (get_continuation_for_phi_1): Remove.
17859 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
17861 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
17862 to explain the use of truncating division. Cap the number of
17863 iterations to the maximum given by nb_iterations_upper_bound,
17866 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
17868 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
17869 * configure: Regenerate.
17870 * config.in: Regenerate.
17871 * config/i386/driver-mingw32.c: new file.
17872 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
17873 * config.host: Link driver-mingw32.o on MinGW host.
17874 * doc/install.texi: Document new --enable-mingw-wildcard configure
17877 2017-05-04 Marek Polacek <polacek@redhat.com>
17879 PR tree-optimization/80612
17880 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
17882 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
17883 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
17885 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
17886 (movt splitter): Likewise.
17887 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
17888 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
17889 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
17890 block for Thumb-1 with MOVT.
17891 (thumb2_legitimate_address_p): Move code block ...
17892 (can_avoid_literal_pool_for_label_p): ... into this new function.
17893 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
17895 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
17896 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
17897 "M-profile targets with the MOVT instruction".
17899 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
17901 * config/arm/arm-builtins.c (arm_init_builtins): Rename
17902 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
17903 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
17905 2017-05-04 Martin Liska <mliska@suse.cz>
17907 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
17908 variable cond_code.
17910 2017-05-04 Richard Biener <rguenther@suse.de>
17912 * tree.c (array_at_struct_end_p): Handle arrays at struct
17913 end with flexarrays more conservatively. Refactor and treat
17914 arrays of arrays or aggregates more strict. Fix
17915 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
17916 * tree.c (array_at_struct_end_p): Adjust prototype.
17917 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
17918 * gimple-fold.c (get_range_strlen): Likewise.
17919 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
17921 2017-05-04 Richard Biener <rguenther@suse.de>
17923 PR tree-optimization/31130
17924 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
17926 (supports_overflow_infinity): Likewise.
17927 (is_negative_overflow_infinity): Likewise.
17928 (is_positive_overflow_infinity): Likewise.
17929 (is_overflow_infinity): Likewise.
17930 (stmt_overflow_infinity): Likewise.
17931 (overflow_infinity_range_p): Likewise.
17932 (usable_range_p): Remove as always returning true.
17933 (make_overflow_infinity): Remove.
17934 (negative_overflow_infinity): Likewise.
17935 (positive_overflow_infinity): Likewise.
17936 (avoid_overflow_infinity): Likewise.
17937 (set_value_range): Adjust accordingly.
17938 (set_value_range_to_nonnegative): Likewise, remove now unused
17939 overflow_infinity arg.
17940 (vrp_operand_equal_p): Adjust.
17941 (update_value_range): Likewise.
17942 (range_int_cst_singleton_p): Likewise.
17943 (operand_less_p): Likewise.
17944 (compare_values_warnv): Likewise.
17945 (extract_range_for_var_from_comparison_expr): Likewise.
17946 (vrp_int_const_binop): Likewise.
17947 (zero_nonzero_bits_from_vr): Likewise.
17948 (extract_range_from_multiplicative_op_1): Likewise.
17949 (extract_range_from_binary_expr_1): Likewise.
17950 (extract_range_from_unary_expr): Likewise.
17951 (extract_range_from_comparison): Likewise.
17952 (extract_range_basic): Likewise.
17953 (adjust_range_with_scev): Likewise.
17954 (compare_ranges): Likewise.
17955 (compare_range_with_value): Likewise.
17956 (dump_value_range): Likewise.
17957 (test_for_singularity): Likewise, remove strict_overflow_p parameter
17959 (simplify_cond_using_ranges): Adjust.
17961 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
17963 * brig-builtins.def: Added a builtin for class_f64.
17964 * builtin-types.def: Added a builtin type needed by class_f64.
17966 2017-05-03 Jason Merrill <jason@redhat.com>
17968 * timevar.def: Add TV_CONSTEXPR.
17970 2017-05-03 David Malcolm <dmalcolm@redhat.com>
17972 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
17974 2017-05-03 Martin Jambor <mjambor@suse.cz>
17976 * ipa-prop.c (ipa_update_after_lto_read): Removed.
17977 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
17978 * ipa-cp.c (ipcp_propagate_stage): Do not call
17979 ipa_update_after_lto_read.
17980 * ipa-inline.c (ipa_inline): Likewise.
17982 2017-05-03 Martin Jambor <mjambor@suse.cz>
17984 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
17985 tag. Added a default constructor and a destructor.
17986 (ipa_edge_args_sum_t): New class;
17987 (ipa_edge_args_sum): Declare.
17988 (ipa_edge_args_vector): Remove declaration.
17989 (IPA_EDGE_REF): Use ipa_edge_args_sum.
17990 (ipa_free_edge_args_substructures): Remove declaration.
17991 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
17992 (ipa_edge_args_info_available_for_edge_p): Likewise.
17993 * ipa-prop.c (ipa_edge_args_vector): Removed.
17994 (edge_removal_hook_holder): Likewise.
17995 (edge_duplication_hook_holder): Likewise.
17996 (ipa_edge_args_sum): New variable.
17997 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
17998 ipa_edge_args_vector.
17999 (ipa_free_edge_args_substructures): Likewise.
18000 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
18001 ipa_edge_args_vector.
18002 (ipa_edge_removal_hook): Turned into method
18003 ipa_edge_args_sum_t::remove.
18004 (ipa_edge_duplication_hook): Turned into method
18005 ipa_edge_args_sum_t::duplicate.
18006 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
18007 registering edge hooks.
18008 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
18009 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
18010 ipa_edge_args_sum instead of ipa_edge_args_vector.
18011 * ipa-profile.c (ipa_profile): Likewise.
18013 2017-05-03 Martin Jambor <mjambor@suse.cz>
18015 * symbol-summary.h (function_summary): New method exists.
18016 (function_summary::symtab_removal): Deallocate through release.
18017 (call_summary): New class.
18018 (gt_ggc_mx): New overload.
18019 (gt_pch_nx): Likewise.
18020 (gt_pch_nx): Likewise.
18022 2017-05-03 Jeff Law <law@redhat.com>
18024 PR tree-optimization/78496
18025 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
18026 from simplify_cond_using_ranges. Split off code to walk
18027 backwards through casts into ...
18028 (simplify_cond_using_ranges_2): New function.
18029 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
18030 (execute_vrp): After identifying jump threads, call
18031 simplify_cond_using_ranges_2.
18033 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
18036 * ipa-inline.h (inline_summary): Add ctor.
18037 (create_ggc): Do not use ggc_cleared_alloc.
18039 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
18040 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18042 * gcc.c (handle_braces): Support escaping in switch matching
18044 * doc/invoke.texi (Spec Files): Document it.
18045 Remove superfluous @code markup in items.
18047 2017-05-03 David Malcolm <dmalcolm@redhat.com>
18049 * diagnostic-show-locus.c (struct column_range): New struct.
18050 (get_affected_columns): New function.
18051 (get_printed_columns): New function.
18052 (struct correction): New struct.
18053 (correction::ensure_capacity): New function.
18054 (correction::ensure_terminated): New function.
18055 (struct line_corrections): New struct.
18056 (line_corrections::~line_corrections): New dtor.
18057 (line_corrections::add_hint): New function.
18058 (layout::print_trailing_fixits): Reimplement in terms of the new
18060 (selftest::test_overlapped_fixit_printing): New function.
18061 (selftest::diagnostic_show_locus_c_tests): Call it.
18063 2017-05-03 Nathan Sidwell <nathan@acm.org>
18065 Canonicalize canonical type hashing
18066 * tree.h (type_hash_canon_hash): Declare.
18067 * tree.c (type_hash_list, attribute_hash_list): Move into
18068 type_hash_canon_hash.
18069 (build_type_attribute_qual_variant): Break out hash code calc into
18070 type_hash_canon_hash.
18071 (type_hash_canon_hash): New. Generic type hash computation.
18072 (build_range_type_1, build_array_type_1, build_function_type,
18073 build_method_type_directly, build_offset_type, build_complex_type,
18074 make_vector_type): Call it.
18076 2017-05-03 Richard Biener <rguenther@suse.de>
18078 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
18079 When all DRs have unknown misaligned do not always peel
18080 when there is a store but apply the same costing model as if
18081 there were only loads.
18083 2017-05-03 Richard Biener <rguenther@suse.de>
18086 PR tree-optimization/80492
18087 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
18088 compare_base_decls returning dont-know properly.
18090 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
18092 * config/arm/iterators.md (CCSI): New mode iterator.
18093 (arch): New mode attribute.
18094 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
18095 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
18096 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
18097 code iterator for success result mode.
18098 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
18099 the corresponding new insn generators.
18101 2017-05-03 Bin Cheng <bin.cheng@arm.com>
18104 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18105 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
18107 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
18109 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
18110 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
18111 (DDR_A): Wrap DDR argument in brackets.
18112 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
18113 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
18114 (DDR_REVERSED_P): Likewise.
18116 2017-05-03 Jakub Jelinek <jakub@redhat.com>
18118 PR tree-optimization/79472
18119 * tree-switch-conversion.c (struct switch_conv_info): Add
18120 contiguous_range and default_case_nonstandard fields.
18121 (collect_switch_conv_info): Compute contiguous_range and
18122 default_case_nonstandard fields, don't clear final_bb if
18123 contiguous_range and only the default case doesn't have the required
18125 (check_all_empty_except_final): Set default_case_nonstandard instead
18126 of failing if contiguous_range and the default case doesn't have empty
18128 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
18129 and only the default case doesn't have the required constants. Skip
18131 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
18132 if default_case_nonstandard.
18133 (build_constructors): Build constant 1 just once. Assert that default
18134 values aren't inserted in between cases if contiguous_range. Skip
18136 (build_arrays): Skip virtual phis.
18137 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
18138 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
18139 Handle virtual phis.
18140 (gen_inbound_check): Handle default_case_nonstandard case.
18141 (process_switch): Adjust check_final_bb caller. Call
18142 gather_default_values with the first non-default case instead of
18143 default case if default_case_nonstandard.
18145 2017-05-02 Nathan Sidwell <nathan@acm.org>
18147 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
18148 check. Fix formatting.
18150 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
18152 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
18153 errors when comparing specialized and unspecialized times.
18155 2017-05-02 David Malcolm <dmalcolm@redhat.com>
18157 * diagnostic-show-locus.c
18158 (layout::should_print_annotation_line_p): Make private.
18159 (layout::print_annotation_line): Make private.
18160 (layout::annotation_line_showed_range_p): Make private.
18161 (layout::show_ruler): Make private.
18162 (layout::print_source_line): Make private. Pass in line and
18163 line_width, rather than calling location_get_source_line. Drop
18165 (layout::print_leading_fixits): New method.
18166 (layout::print_any_fixits): Rename to...
18167 (layout::print_trailing_fixits): ...this, and make private.
18168 Don't print newline fixits.
18169 (diagnostic_show_locus): Move logic for printing one row into...
18170 (layout::print_line): ...this new function. Move the
18171 location_get_source_line call and error-handling from
18172 print_source_line to here. Call print_leading_fixits, and rename
18173 print_any_fixits to print_trailing_fixits.
18174 (selftest::test_fixit_insert_containing_newline): Update now that
18175 newlines are partially supported.
18176 (selftest::test_fixit_insert_containing_newline_2): New test.
18177 (selftest::test_fixit_replace_containing_newline): Update comments.
18178 (selftest::diagnostic_show_locus_c_tests): Call the new test.
18179 * edit-context.c (class added_line): New class.
18180 (class edited_line): Describe newline handling in comment.
18181 (edited_line::actually_edited_p): New method.
18182 (edited_line::print_content): Delete redundant decl.
18183 (edited_line::m_predecessors): New field.
18184 (edited_file::print_content): Call edited_line::print_content.
18185 (edited_file::print_diff): Update to support newlines.
18186 (edited_file::print_diff_hunk): Likewise.
18187 (edited_file::print_run_of_changed_lines): New function.
18188 (edited_file::print_diff_line): Convert to...
18189 (print_diff_line): ...this.
18190 (edited_file::get_effective_line_count): New function.
18191 (edited_line::edited_line): Initialize new field m_predecessors.
18192 (edited_line::~edited_line): Clean up m_predecessors.
18193 (edited_line::apply_fixit): Handle newlines.
18194 (edited_line::get_effective_line_count): New function.
18195 (edited_line::print_content): New function.
18196 (edited_line::print_diff_lines): New function.
18197 (selftest::test_applying_fixits_insert_containing_newline): New
18199 (selftest::test_applying_fixits_replace_containing_newline): New
18201 (selftest::insert_line): New function.
18202 (selftest::test_applying_fixits_multiple_lines): Add example of
18204 (selftest::edit_context_c_tests): Call the new tests.
18206 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18208 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
18209 parameter cand. Update dump information.
18210 (get_computation_cost): Update uses.
18212 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18214 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
18215 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
18216 (get_computation_at, rewrite_use_address): Update use of
18217 get_computation_aff.
18219 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18221 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
18222 (get_computation): Delete.
18223 (get_computation_cost): Implement like get_computation_cost_at.
18224 Use get_computation_at.
18225 (get_computation_cost_at): Delete.
18226 (rewrite_use_nonlinear_expr): Use get_computation_at.
18227 (rewrite_use_compare, remove_unused_ivs): Ditto.
18229 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18231 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
18233 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18235 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
18236 (ivopts_global_cost_for_size): Rename parameter and update uses.
18237 (iv_ca_recount_cost): Update uses.
18238 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
18239 candidates seperately in n_invs and n_cands.
18240 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
18242 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18244 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
18245 (find_inv_vars_cb): New.
18246 (find_depends): Renamed to ...
18247 (find_inv_vars): ... this.
18248 (add_candidate_1, force_var_cost): Call find_inv_vars.
18249 (split_address_cost, determine_group_iv_cost_cond): Ditto.
18251 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18253 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
18254 inv_vars. Add inv_exprs.
18255 (struct iv_cand): Rename depends_on to inv_vars.
18256 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
18257 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
18258 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
18259 (dump_cand): Dump inv_vars.
18260 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
18261 (record_invariant, find_depends, add_candidate_1): Ditto.
18262 (set_group_iv_cost, force_var_cost): Ditto.
18263 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
18264 (get_computation_cost_at, get_computation_cost): Ditto.
18265 (determine_group_iv_cost_generic): Ditto.
18266 (determine_group_iv_cost_address): Ditto.
18267 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
18268 (determine_group_iv_costs): Ditto.
18269 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
18270 (iv_ca_set_remove_invariants): Renamed to ...
18271 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
18272 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
18273 (iv_ca_set_add_invariants): Renamed to ...
18274 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
18275 (iv_ca_set_cp): Use iv_ca_set_add_invs.
18276 (iv_ca_has_deps): Support inv_vars and inv_exprs.
18277 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
18278 (create_new_ivs): Remove useless dump.
18280 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18282 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
18284 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
18285 (iv_ca_set_no_cp, create_new_iv): Ditto.
18287 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18289 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
18291 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
18293 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
18294 function tree_check2.
18296 2017-05-02 Martin Liska <mliska@suse.cz>
18298 * doc/gcov.texi: Add missing preposition.
18299 * gcov.c (function_info::function_info): Properly fill up
18300 all member variables.
18302 2017-05-02 Tamar Christina <tamar.christina@arm.com>
18304 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
18306 2017-05-02 Tamar Christina <tamar.christina@arm.com>
18308 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
18310 2017-05-02 Martin Liska <mliska@suse.cz>
18313 * lto-streamer-in.c (lto_read_tree_1): Remove
18314 LTO_STREAMER_DEBUG.
18315 * lto-streamer.c (struct tree_hash_entry): Likewise.
18316 (struct tree_entry_hasher): Likewise.
18317 (tree_entry_hasher::hash): Likewise.
18318 (tree_entry_hasher::equal): Likewise.
18319 (lto_streamer_init): Likewise.
18320 (lto_orig_address_map): Likewise.
18321 (lto_orig_address_get): Likewise.
18322 (lto_orig_address_remove): Likewise.
18323 * lto-streamer.h: Likewise.
18324 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
18325 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
18327 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
18329 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
18330 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
18331 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
18332 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
18333 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
18334 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
18335 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
18336 (mm_maskz_sub_ss): New intrinsics.
18337 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
18338 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
18339 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
18340 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
18341 (__builtin_ia32_subss_mask_round): New builtins.
18342 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
18343 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
18344 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
18346 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
18347 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
18349 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
18352 2017-05-02 Martin Jambor <mjambor@suse.cz>
18354 PR tree-optimization/78687
18355 * tree-sra.c (access): New field parent.
18356 (process_subtree_disqualification): New function.
18357 (disqualify_candidate): Call it.
18358 (build_accesses_from_assign): Reset write flag if creating an
18360 (build_access_subtree): Fill in parent field and also prpagate
18361 down grp_write flag.
18362 (create_artificial_child_access): New parameter set_grp_write, set
18363 grp_write to its value.
18364 (propagate_subaccesses_across_link): Also propagate grp_write flag
18366 (propagate_all_subaccesses): Push the closest parent back to work
18367 queue if add_access_to_work_queue returned true.
18369 2017-05-02 Richard Biener <rguenther@suse.de>
18371 * common.opt (fstrict-overflow): Alias negative to fwrapv.
18372 * doc/invoke.texi (fstrict-overflow): Remove all traces of
18373 -fstrict-overflow documentation.
18374 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
18375 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
18376 flag_strict_overflow.
18377 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
18378 * lto-opts.c (lto_write_options): Do not stream it.
18379 * lto-wrapper.c (merge_and_complain): Do not handle it.
18380 * opts.c (default_options_table): Do not set -fstrict-overflow.
18381 (finish_options): Likewise do not clear it when sanitizing.
18382 * simplify-rtx.c (simplify_const_relational_operation): Do not
18383 test flag_strict_overflow.
18385 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
18387 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
18388 using enabled attribute.
18389 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
18390 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
18391 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
18392 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
18393 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
18394 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
18395 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
18396 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
18397 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
18398 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
18400 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
18402 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
18404 2017-05-02 Richard Biener <rguenther@suse.de>
18406 PR tree-optimization/80591
18408 2017-04-10 Richard Biener <rguenther@suse.de>
18410 * tree-ssa-structalias.c (find_func_aliases): Properly handle
18413 2017-05-02 Richard Biener <rguenther@suse.de>
18415 PR tree-optimization/80549
18416 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
18417 (cleanup_tree_cfg_noloop): Create forwarders to known loop
18418 headers if they do not have a preheader.
18420 2017-05-02 Martin Liska <mliska@suse.cz>
18423 * common.opt: Fix typo.
18424 * doc/invoke.texi: Likewise.
18426 2017-05-01 Jan Beulich <jbeulich@suse.com>
18428 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
18429 swapping, add (x,x,m,x,n) alternative.
18431 2017-05-01 Nathan Sidwell <nathan@acm.org>
18433 * calls.c (combine_pending_stack_adjustment_and_call): Remove
18434 unnecessary unadjusted_alignment check.
18436 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
18439 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
18441 * gimplify.c (gimplify_cilk_detach): New function.
18442 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
18443 * tree-core.h: Document EXPR_CILK_SPAWN.
18444 * tree.h (EXPR_CILK_SPAWN): Define.
18446 2017-05-01 David Malcolm <dmalcolm@redhat.com>
18448 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
18449 to use new fixit_hint representation, using the "replace" logic.
18450 (get_line_span_for_fixit_hint): Likewise.
18451 (layout::print_any_fixits): Likewise.
18452 (selftest::test_one_liner_many_fixits): Rename to...
18453 (selftest::test_one_liner_many_fixits_1): ...this, and update
18454 comment and expected output to reflect that the multiple fix-it
18455 hints are now consolidated into one insertion.
18456 (selftest::test_one_liner_many_fixits_2): New test.
18457 (selftest::test_diagnostic_show_locus_one_liner): Update for
18459 (selftest::test_fixit_consolidation): Update for fix-it API
18461 * diagnostic.c (print_parseable_fixits): Likewise.
18462 * edit-context.c (edited_line::m_line_events): Convert from
18463 auto_vec <line_event *> to auto_vec <line_event>.
18464 (class line_event): Convert from abstract base class to a concrete
18465 class, taking over the role of replace_event.
18466 (class insert_event): Delete.
18467 (class replace_event): Rename to class line_event. Convert to
18469 (edit_context::add_fixits): Reimplement.
18470 (edit_context::apply_insert): Delete.
18471 (edit_context::apply_replace): Rename to...
18472 (edit_context::apply_fixit): ...this. Convert to half-open range.
18473 (edited_file::apply_insert): Delete.
18474 (edited_file::apply_replace): Rename to...
18475 (edited_file::apply_fixit): ...this.
18476 (edited_line::~edited_line): Drop deletion of events.
18477 (edited_line::apply_insert): Delete.
18478 (edited_line::apply_replace): Rename to...
18479 (edited_line::apply_fixit): ...this. Convert to half-open range.
18480 Update for change to type of m_line_events.
18481 * edit-context.h (edit_context::apply_insert): Delete.
18482 (edit_context::apply_replace): Rename to...
18483 (edit_context::apply_fixit): ...this.
18485 2017-05-01 Martin Sebor <msebor@redhat.com>
18487 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
18490 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
18493 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
18494 __get_cpuid_max returns 0.
18495 (__get_cpuid_count): Ditto.
18497 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
18499 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
18500 replacement expression is another instance of one of its arguments.
18502 2017-05-01 Jakub Jelinek <jakub@redhat.com>
18505 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
18506 check for stack push/pop autoinc.
18507 * config/i386/i386.c (ix86_agi_dependent): Return false
18508 if the only reason why modified_in_p returned true is that
18509 addr is SP based and set_insn is a push or pop.
18511 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
18513 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
18516 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
18519 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
18520 (account_size_time): Use two predicates - exec_pred and
18522 (evaluate_conditions_for_known_args): Compute both clause and
18524 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
18525 (inline_summary_t::duplicate): Update.
18526 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
18528 (compute_inline_parameters): Likewise.
18529 (estimate_edge_size_and_time): Update caluclation of time.
18530 (estimate_node_size_and_time): Compute both time and nonspecialized
18532 (estimate_ipcp_clone_size_and_time): Update.
18533 (inline_merge_summary): Update.
18534 (do_estimate_edge_time): Update.
18535 (do_estimate_edge_size): Update.
18536 (do_estimate_edge_hints): Update.
18537 (inline_read_section, inline_write_summary): Stream both new predicates.
18538 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
18540 (compute_inlined_call_time): Cleanup.
18541 (big_speedup_p): Update.
18542 (edge_badness): Update.
18543 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
18544 (size_time_entry): Replace predicate by exec_predicate and
18545 nonconst_predicate.
18546 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
18547 (estimate_edge_time): Return also nonspec_time.
18548 (reset_edge_growth_cache): Update.
18550 2017-04-29 Jakub Jelinek <jakub@redhat.com>
18552 PR rtl-optimization/80491
18553 * ifcvt.c (noce_process_if_block): When looking for x setter
18554 with missing else_bb, don't check only the insn right before
18555 cond_earliest, but look for the last insn that x is modified in
18556 within the same bb.
18558 PR rtl-optimization/80491
18559 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
18561 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
18563 PR tree-optimization/80487
18564 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
18566 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18568 PR tree-optimization/79697
18569 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
18570 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
18571 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
18573 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
18574 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
18576 2017-04-28 Martin Sebor <msebor@redhat.com>
18578 PR tree-optimization/80523
18579 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
18580 (init_target_to_host_charmap, target_to_host, target_strtol10): New
18582 (maybe_warn, format_directive, parse_directive): Use new functions.
18583 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
18585 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
18587 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
18589 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
18591 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
18592 target_header_dir): Set correctly.
18593 * configure: Regenerated.
18594 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
18595 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
18596 instead of SYSTEM_HEADER_DIR.
18598 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
18600 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
18601 (estimate_local_effects): Likewise.
18602 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
18603 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
18604 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
18605 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
18606 do_estimate_edge_time, estimate_edge_time): Likewise.
18607 * ipa-inline-analysis.c (estimate_node_size_and_time,
18608 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
18609 (estimate_time_after_inlining): Remove.
18611 2017-04-28 Martin Liska <mliska@suse.cz>
18613 * doc/gcov.texi: Enhance documentation of gcov.
18615 2017-04-28 Martin Liska <mliska@suse.cz>
18617 * doc/gcov.texi: Sort options in alphabetic order.
18618 * doc/gcov-dump.texi: Likewise.
18619 * doc/gcov-tool.texi: Likewise.
18620 * gcov.c (print_usage): Likewise.
18621 * gcov-dump.c (print_usage): Likewise.
18622 * gcov-tool.c (print_merge_usage_message): Likewise.
18623 (print_rewrite_usage_message): Likewise.
18624 (print_overlap_usage_message): Likewise.
18626 2017-04-28 Martin Liska <mliska@suse.cz>
18628 PR gcov-profile/53915
18629 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
18631 2017-04-28 Martin Liska <mliska@suse.cz>
18633 PR gcov-profile/79891
18634 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
18635 is marked by compiler as living on a line.
18636 (get_cycles_count): Remove usage of the union.
18637 (output_intermediate_file): Likewise.
18638 (find_source): Fix GNU coding style.
18639 (accumulate_line_counts): Remove old non-all block mode.
18640 (output_lines): Remove usage of the union.
18641 * profile.c (output_location): Include all BBs, even if
18642 belonging to a same line (and file) as a previous BB.
18644 2017-04-28 Martin Liska <mliska@suse.cz>
18646 * gcov.c (process_args): Handle new argument 'w'.
18647 (read_graph_file): Assign ID to BBs.
18648 (output_branch_count): Display BB # if verbose flag is set.
18649 (output_lines): Likewise for arcs.
18650 (print_usage): Add '--verbose' option help.
18651 * doc/gcov.texi: Document --verbose (-w) option.
18653 2017-04-28 Martin Liska <mliska@suse.cz>
18655 * gcov.c (struct block_location_info): New struct.
18656 (process_file): Fill up the new structure.
18657 (read_graph_file): Replace usage of encoding by the newly added
18659 (add_line_counts): Likewise.
18660 (accumulate_line_counts): Remove usage of the union.
18661 (function_info::function_info): New function.
18662 (function_info::~function_info): Likewise.
18663 (process_file): Call delete instead of release_function.
18664 (release_function): Release the function.
18665 (release_structures): Call delete instead of release_function.
18666 (solve_flow_graph): Replace usage of num_blocks.
18667 (find_exception_blocks): Likewise.
18668 (output_lines): Fix GNU coding style.
18670 2017-04-28 Martin Liska <mliska@suse.cz>
18673 * coverage.c (coverage_remove_note_file): New function.
18674 * coverage.h: Declare the function.
18675 * toplev.c (finalize): Clean if an error has been seen.
18677 2017-04-28 Martin Liska <mliska@suse.cz>
18679 PR gcov-profile/80031
18680 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
18681 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
18682 * gcov.c (read_graph_file): Read just number of blocks.
18683 * profile.c (branch_prob): Do not stream 0 flags per a basic
18686 2017-04-28 Martin Liska <mliska@suse.cz>
18688 * gcov-dump.c (tag_*): Add new argument to declarations.
18689 (dump_gcov_file): Likewise.
18690 (tag_blocks): Add and use new argument depth.
18691 (tag_arcs): Likewise.
18692 (tag_lines): Likewise.
18693 (tag_counters): Likewise.
18694 (tag_summary): Likewise.
18695 (dump_working_sets): Use depth to do a proper indentation.
18697 2017-04-28 Jakub Jelinek <jakub@redhat.com>
18700 * cgraph.h (symtab_node::debug_symtab): No longer inline.
18701 * symtab.c (symtab_node::debug_symtab): Move definition here.
18703 2017-04-28 Richard Biener <rguenther@suse.de>
18705 * lto-streamer.h (LTO_major_version): Bump to 7.
18707 2017-04-28 Richard Biener <rguenther@suse.de>
18709 * tree-vrp.c (assert_info): New struct.
18710 (add_assert_info): New helper.
18711 (register_edge_assert_for_2): Refactor to add asserts to a vector
18713 (register_edge_assert_for_1): Likewise.
18714 (register_edge_assert_for): Likewise.
18715 (finish_register_edge_assert_for): New helper actually registering
18716 asserts where live on edge.
18717 (find_conditional_asserts): Adjust.
18718 (find_switch_asserts): Likewise.
18719 (evrp_dom_walker::try_find_new_range): Generalize.
18720 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
18722 2017-04-27 Marek Polacek <polacek@redhat.com>
18725 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
18726 arg10 and arg11 to itype.
18728 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
18730 * doc/extend.texi (Object Size Checking): Improve grammar.
18732 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
18735 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
18736 that the logic for permitting reciprocal estimates matches that
18739 2017-04-27 Jakub Jelinek <jakub@redhat.com>
18742 * tree.c (type_cache_hasher::equal): Only compare
18743 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
18744 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
18745 non-aggregate element types.
18746 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
18747 about the flag on ARRAY_TYPEs in the comment, formatting fix.
18749 2017-04-27 Richard Biener <rguenther@suse.de>
18751 PR middle-end/80533
18752 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
18753 stripping ARRAY_REFs from MEM_EXPR make sure we're not
18754 keeping a reference to a trailing array.
18756 2017-04-27 Richard Biener <rguenther@suse.de>
18758 PR middle-end/80539
18759 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
18760 being in loop-closed SSA form conservatively.
18761 (chrec_fold_multiply_poly_poly): Likewise.
18763 2017-04-27 Tamar Christina <tamar.christina@arm.com>
18765 PR middle-end/79665
18766 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
18767 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
18769 2017-04-27 Jakub Jelinek <jakub@redhat.com>
18772 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
18773 (aarch64_function_arg_alignment): Return unsigned int again, but still
18774 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
18775 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
18776 Don't emit -Wpsabi note.
18777 (aarch64_function_arg_boundary): Likewise.
18778 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
18781 2017-04-26 Nathan Sidwell <nathan@acm.org>
18783 * tree.h (crc32_unsigned_n): Declare.
18784 (crc32_unsigned, crc32_unsigned): Make inline.
18785 * tree.c (crc32_unsigned_bits): Replace with ...
18786 (crc32_unsigned_n): ... this.
18787 (crc32_unsigned, crc32_byte): Remove.
18788 (crc32_string): Remove unnecessary braces.
18790 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
18792 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
18793 * ipa-inline-analysis.c (MAX_TIME): Remove.
18794 (account_size_time): Use sreal for time.
18795 (dump_inline_summary): Update.
18796 (estimate_function_body_sizes): Update.
18797 (estimate_edge_size_and_time): Update.
18798 (estimate_calls_size_and_time): Update.
18799 (estimate_node_size_and_time): Update.
18800 (inline_merge_summary): Update.
18801 (inline_update_overall_summary): Update.
18802 (estimate_time_after_inlining): Update.
18803 (inline_read_section): Update.
18804 (inline_write_summary): Update.
18805 * ipa-inline.c (compute_uninlined_call_time): Update.
18806 (compute_inlined_call_time): Update.
18807 (recursive_inlining): Update.
18808 (inline_small_functions): Update.
18809 (dump_overall_stats): Update.
18810 * ipa-inline.h: Include sreal.h.
18811 (size_time_entry): Turn time to sreal.
18812 (inline_summary): Turn self_time nad time to sreal.
18814 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
18816 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
18818 (sreal::stream_out, sreal::stream_in): New.
18819 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
18821 2017-04-25 Jakub Jelinek <jakub@redhat.com>
18823 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
18826 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
18829 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
18831 (dimode_scalar_chain::compute_convert_gain): Ditto.
18832 (dimode_scalar_chain::make_vector_copies): Ditto.
18833 (dimode_scalar_chain::convert_reg): Ditto.
18834 (dimode_scalar_chain::convert_insn): Ditto.
18835 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
18836 (VI248_AVX512BW_1): New mode iterator.
18837 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
18838 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
18841 2017-04-25 Martin Sebor <msebor@redhat.com>
18843 PR tree-optimization/80497
18844 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
18845 constants are representable in HOST_WIDE_INT.
18846 (parse_directive): Ditto.
18848 2017-04-25 Martin Sebor <msebor@redhat.com>
18851 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
18852 (new_zero_array): Adjust signature.
18853 (dom_info::dom_init): Used unsigned rather that size_t.
18854 (dom_info::dom_info): Same.
18856 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18857 Jakub Jelinek <jakub@redhat.com>
18860 * config/arm/arm.c: Include gimple.h.
18861 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
18862 returns negative, increment ncrn only if it returned positive.
18863 (arm_needs_doubleword_align): Return int instead of bool,
18864 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
18865 members, but if there is any such non-FIELD_DECL
18866 > PARM_BOUNDARY aligned decl, return -1 instead of false.
18867 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
18868 returns negative, increment nregs only if it returned positive.
18869 (arm_setup_incoming_varargs): Likewise.
18870 (arm_function_arg_boundary): Emit -Wpsabi note if
18871 arm_needs_doubleword_align returns negative, return
18872 DOUBLEWORD_ALIGNMENT only if it returned positive.
18874 2017-04-25 Marek Polacek <polacek@redhat.com>
18877 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
18878 first argument to type.
18880 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
18883 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
18884 type checks to test for compatibility instead of equality.
18886 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18887 Jakub Jelinek <jakub@redhat.com>
18890 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
18892 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
18893 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
18894 the alignment computation, but return their maximum in warn_alignment.
18895 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
18896 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
18898 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
18899 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
18902 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18904 * config/arc/simdext.md (dmpyh): Fix typo.
18906 2017-04-25 Richard Biener <rguenther@suse.de>
18908 PR tree-optimization/80492
18909 * alias.c (compare_base_decls): Handle registers with asm
18910 specification conservatively.
18911 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
18912 compare_base_decls returning dont-know properly.
18914 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18916 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
18917 (legitimate_offset_address_p): New function.
18918 (arc_legitimate_address_p): Use above function.
18920 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18922 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
18924 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18926 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
18927 ACCH registers whenever they are available.
18929 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18931 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
18932 double regs fix when not used.
18934 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18936 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
18938 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
18939 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
18941 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18943 * config/arc/arc.c (arc_output_addsi): Check for h-register class
18944 when emitting short ADD instructions.
18946 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18948 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
18950 (cmpsi_cc_c_insn): Likewise.
18951 (cbranchsi4_scratch): Compute proper instruction length using
18952 compact_hreg_operand.
18953 * config/arc/predicates.md (compact_hreg_operand): New predicate.
18955 2017-04-25 Richard Biener <rguenther@suse.de>
18957 PR middle-end/80509
18958 * passes.c (pass_manager::pass_manager): Initialize
18959 m_name_to_pass_map.
18961 2017-04-25 Richard Biener <rguenther@suse.de>
18963 PR tree-optimization/79201
18964 * tree-ssa-sink.c (statement_sink_location): Handle calls.
18966 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18969 * config/s390/vector.md: Split MEM->GPR vector moves for
18970 non-s_operand addresses.
18972 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18975 * config/s390/predicates.md (reload_const_wide_int_operand): New
18977 * config/s390/s390.md ("movti"): Remove d/P alternative.
18978 ("movti_bigconst"): New pattern definition.
18980 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
18983 * s390-protos.h (s390_expand_cs_hqi): Removed.
18984 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
18985 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
18986 modes as well as CCZ1mode and CCZmode.
18987 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
18988 signature of s390_emit_compare_and_swap.
18989 (s390_expand_cs_hqi): Likewise, make static.
18990 (s390_expand_cs_tdsi): Generate an explicit compare before trying
18991 compare-and-swap, in some cases.
18992 (s390_expand_cs): Wrapper function.
18993 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
18995 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
18996 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
18997 patterns for small and large integers. Forbid symref memory operands.
18998 Move expander to s390.c. Require cc register.
18999 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
19000 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
19001 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
19002 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
19003 symref memory operands. Remove CC mode and call s390_match_ccmode
19005 ("atomic_exchange<mode>"): Allow and implement all integer modes.
19007 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
19009 * config/s390/s390.md (define_peephole2): New peephole to help
19010 combining the load-and-test pattern with volatile memory.
19012 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
19014 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
19015 with CCZmode for TARGET_Z196.
19017 2017-04-25 Jakub Jelinek <jakub@redhat.com>
19019 PR rtl-optimization/80501
19020 * combine.c (make_compound_operation_int): Set subreg_code to SET
19021 even for AND with mask of the sign bit of mode.
19023 PR rtl-optimization/80500
19024 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
19025 sum's initial value.
19027 2017-04-25 Julian Brown <julian@codesourcery.com>
19028 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19030 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
19032 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
19034 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
19036 2017-04-25 Julian Brown <julian@codesourcery.com>
19037 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19039 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
19040 (thunderx2t99_sha): New Reservation.
19042 2017-04-25 Julian Brown <julian@codesourcery.com>
19043 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19045 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
19046 type for 1-element load.
19048 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
19050 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
19052 2017-04-24 Martin Jambor <mjambor@suse.cz>
19054 PR tree-optimization/80293
19055 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
19056 char arrays not totally scalarizable if it is false.
19057 (analyze_all_variable_accesses): Pass correct value in the new
19058 parameter. Add a statistics counter.
19060 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
19062 PR middle-end/79931
19063 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
19065 2017-04-24 Richard Biener <rguenther@suse.de>
19067 PR tree-optimization/80494
19068 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
19069 out for complex types.
19071 2017-04-24 Richard Biener <rguenther@suse.de>
19073 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
19074 * tree-ssa-sccvn.c (print_scc): Print SCC size.
19075 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
19076 (DFS): Adjust and never fail.
19077 (sccvn_dom_walker::fail): Remove.
19078 (sccvn_dom_walker::before_dom_children): Adjust.
19079 (run_scc_vn): Likewise and never fail.
19080 * tree-ssa-pre.c (pass_pre::execute): Adjust.
19081 (pass_fre::execute): Likewise.
19083 2017-04-24 Richard Biener <rguenther@suse.de>
19085 PR tree-optimization/79725
19086 * tree-ssa-sink.c (statement_sink_location): Return whether
19087 failure reason was zero uses. Move that check later.
19088 (sink_code_in_bb): Deal with zero uses by removing the stmt
19091 2017-04-24 Richard Biener <rguenther@suse.de>
19094 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
19095 pointer-based references.
19097 2017-04-24 Richard Biener <rguenther@suse.de>
19100 * pass_manager.h (pass_manager::operator new): Remove.
19101 (pass_manager::operator delete): Likewise.
19102 * passes.c (pass_manager::operator new): Remove.
19103 (pass_manager::operator delete): Likewise.
19104 (pass_manager::pass_manager): Zero individual pass members.
19106 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
19109 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
19110 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
19111 Check "XEXP (src, 1)" operand here.
19112 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
19113 Check "XEXP (src, 1)" operand here.
19114 (dimode_scalar_chain::make_vector_copies): Detect count register
19115 of a shift instruction. Zero extend count register from QImode
19116 to DImode to satisfy vector shift pattern count operand predicate.
19117 Substitute vector shift count operand with a DImode copy.
19118 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
19121 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
19123 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
19124 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
19125 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
19126 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
19127 (UNSPEC_NOREX_MEM): Remove definition.
19129 2017-04-21 Richard Biener <rguenther@suse.de>
19131 PR tree-optimization/79547
19132 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
19133 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
19134 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
19135 without any constraints.
19137 2017-04-21 Richard Biener <rguenther@suse.de>
19139 PR tree-optimization/78847
19140 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
19142 2017-04-21 Richard Biener <rguenther@suse.de>
19144 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
19145 (build_distinct_type_copy): Likewise.
19146 (build_variant_type_copy): Likewise.
19147 * tree.c (build_qualified_type): Pass down mem-stat info.
19148 (build_distinct_type_copy): Likewise.
19149 (build_variant_type_copy): Likewise.
19151 2017-04-21 Richard Biener <rguenther@suse.de>
19153 PR tree-optimization/80237
19154 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
19156 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
19157 for a simplified result.
19159 2016-04-21 Richard Biener <rguenther@suse.de>
19161 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
19162 sth as strict as a simple_iv but a chrec without symbols and an
19163 operand defined in the loop we are peeling (and not some subloop).
19164 (propagate_constants_for_unrolling): Propagate all constants.
19166 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
19169 * config/i386/i386.c (print_reg): Remove assert for disalowed
19170 regno values, call output_operand_lossage instead.
19172 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
19175 * config/i386/constraints.md (Yc): New register constraint.
19176 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
19177 Use Yc constraint for alternative 2 of operand 0. Remove
19178 preferred_for_speed attribute.
19180 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
19182 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
19183 lastprivate clauses in SIMT case.
19185 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
19187 * doc/invoke.texi (-Wextra-semi): Document new warning option.
19189 2017-04-20 Richard Biener <rguenther@suse.de>
19191 PR tree-optimization/57796
19192 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
19193 as N scalar stores.
19194 (vect_model_load_cost): Cost gathers as N scalar loads.
19196 2017-04-20 Richard Biener <rguenther@suse.de>
19198 * ggc-page.c (ggc_allocated_p): Rename to ...
19199 (safe_lookup_page_table_entry): ... this and return the lookup
19201 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
19203 2017-04-20 Richard Biener <rguenther@suse.de>
19205 PR tree-optimization/80453
19206 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
19207 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
19208 from the conditions.
19209 (vn_phi_eq): Pass them down.
19210 (vn_phi_lookup): Record them.
19211 (vn_phi_insert): Likewise.
19213 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
19215 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
19216 uninitialized variable warning to avoid buffer overrun.
19218 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
19221 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
19222 is suppressed for '{ 0 }' in C.
19224 2017-04-20 Jakub Jelinek <jakub@redhat.com>
19226 * BASE-VER: Set to 8.0.0.
19228 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
19230 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
19231 priority .init_array and .fini_array section with SECTION_NOTYPE
19234 2017-04-20 Jakub Jelinek <jakub@redhat.com>
19236 PR middle-end/80423
19237 * tree.h (build_array_type): Add typeless_storage default argument.
19238 * tree.c (type_cache_hasher::equal): Also compare
19239 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
19240 (build_array_type): Add typeless_storage argument, set
19241 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
19243 (build_nonshared_array_type): Adjust build_array_type_1 caller.
19244 (build_array_type): Likewise. Add typeless_storage argument.
19246 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
19247 Jakub Jelinek <jakub@redhat.com>
19249 PR tree-optimization/80426
19250 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
19251 operation on symbolic operands, also compute the overflow for the
19252 invariant part when the operation degenerates into a negation.
19254 2017-04-19 Jakub Jelinek <jakub@redhat.com>
19257 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
19258 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
19261 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
19263 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
19266 * config/avr/avr.c (tree.h): Include it.
19267 (cgraph.h): Include it.
19268 (avr_encode_section_info): Don't warn for uninitialized progmem
19269 variable if it's just an alias.
19271 2017-04-19 Richard Biener <rguenther@suse.de>
19274 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
19275 when needed by AutoPGO.
19277 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
19280 * doc/lto.texi: Remove an extra 'that'.
19282 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
19284 PR rtl-optimization/80429
19285 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
19286 are only used in debug insns.
19288 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
19289 Vladimir Makarov <vmakarov@redhat.com>
19291 * config/sparc/predicates.md (input_operand): Add comment. Return
19292 true for any memory operand when LRA is in progress.
19293 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
19295 2017-04-18 Jeff Law <law@redhat.com>
19298 * mips.md ({return,simple_return}_internal): Do not overwrite
19301 2017-04-18 Jakub Jelinek <jakub@redhat.com>
19303 PR tree-optimization/80443
19304 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
19305 instead of adding 1, subtract -1 and similarly instead of subtracting
19308 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
19310 PR rtl-optimization/80357
19311 * haifa-sched.c (tmp_bitmap): New variable.
19312 (model_recompute): Handle duplicate use records.
19313 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
19314 (free_global_sched_pressure_data): Free it.
19316 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
19319 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
19320 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
19321 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
19322 instead of SYSTEM_HEADER_DIR.
19324 2017-04-18 Jeff Law <law@redhat.com>
19326 PR middle-end/80422
19327 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
19328 predecessors after walking up the insn chain.
19330 2017-04-18 Jakub Jelinek <jakub@redhat.com>
19333 * dwarf2out.c (modified_type_die): Try harder not to emit internal
19334 sizetype type into debug info.
19336 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
19339 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
19340 unneeded test for TARGET_UPPER_REGS_SF.
19341 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
19343 2017-04-18 Jakub Jelinek <jakub@redhat.com>
19346 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
19347 instead of gsi_after_labels.
19349 2017-04-18 Jeff Law <law@redhat.com>
19351 * regcprop.c (maybe_mode_change): Avoid creating copies of the
19355 2017-04-13 Jeff Law <law@redhat.com>
19356 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
19357 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
19359 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
19362 * config/avr/avr.c (intl.h): Include it.
19363 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
19365 2017-04-18 Martin Liska <mliska@suse.cz>
19367 PR gcov-profile/78783
19368 * gcov-tool.c (gcov_output_files): Validate that destination
19369 file is either removed by the tool or by a user.
19371 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
19372 Guy Benyei <guybe@mellanox.com>
19374 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
19375 block, and do not negate it, the stored id is already negative.
19377 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
19379 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
19381 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
19384 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
19385 masks of options that should be turned off if the VSX vector
19386 options are turned off.
19387 (OTHER_P8_VECTOR_MASKS): Likewise.
19388 (OTHER_VSX_VECTOR_MASKS): Likewise.
19389 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
19390 rs6000_disable_incompatible_switches to validate no type switches
19392 (rs6000_incompatible_switch): New function to disallow turning on
19393 other vector options if -mno-vsx, -mno-power8-vector, or
19394 -mno-power9-vector are specified.
19396 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
19398 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
19400 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
19402 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
19403 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
19404 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
19405 (ARG_POINTER_CFA_OFFSET): Likewise.
19407 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
19409 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
19410 conditions to take advantage of various optimizations.
19412 2017-04-13 Jeff Law <law@redhat.com>
19414 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
19415 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
19416 (zero_extendsidi2_dext): Likewise.
19418 2017-04-13 Jakub Jelinek <jakub@redhat.com>
19421 * fold-const.c (fold_ternary_loc): Revert
19422 use op0 instead of fold_convert_loc (loc, type, arg0) part of
19425 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
19427 PR rtl-optimization/80343
19428 * lra-remat.c (update_scratch_ops): Assign original hard reg to
19429 new scratch pseudo.
19431 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
19434 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
19435 to ubsan_encode_value.
19437 2017-04-13 Jeff Law <law@redhat.com>
19439 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
19440 appearing in DEBUG_INSNs.
19442 2017-04-13 Martin Liska <mliska@suse.cz>
19444 PR gcov-profile/80413
19445 * gcov-io.c (gcov_write_string): Copy to buffer just when
19446 allocated size is greater than zero.
19448 2017-04-13 Jakub Jelinek <jakub@redhat.com>
19451 * dwarf2out.c (decls_for_scope): Ignore declarations of
19452 current_function_decl in BLOCK_NONLOCALIZED_VARS.
19454 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
19457 * ipa-visibility.c (non_local_p): Fix typos.
19458 (localize_node): When localizing symbol in same comdat group,
19459 dissolve the group only when we know external symbols are going
19461 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
19463 2017-04-12 Jakub Jelinek <jakub@redhat.com>
19465 PR tree-optimization/79390
19466 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
19467 order does not result in usable sequence, retry with reversed operand
19473 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
19474 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
19475 op0 instead of fold_convert_loc (loc, type, arg0).
19477 2017-04-12 Jeff Law <law@redhat.com>
19479 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
19480 has a delay slot in the generated code.
19482 * config/cris/cris.md (cris_preferred_reload_class): Return
19483 GENNONACR_REGS rather than GENERAL_REGS.
19485 2017-04-12 Jakub Jelinek <jakub@redhat.com>
19488 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
19489 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
19490 signedness of the result type.
19492 2017-04-12 Richard Biener <rguenther@suse.de>
19493 Jeff Law <law@redhat.com>
19495 PR tree-optimization/80359
19496 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
19497 trim stores to TARGET_MEM_REFs.
19499 2017-04-12 Richard Biener <rguenther@suse.de>
19501 PR tree-optimization/79390
19502 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
19503 threading case even more.
19505 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
19508 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
19509 for quad_address_p for TImode, instead of just not indexed_address.
19511 2017-04-12 Richard Biener <rguenther@suse.de>
19512 Bernd Edlinger <bernd.edlinger@hotmail.de>
19514 PR middle-end/79671
19515 * alias.c (component_uses_parent_alias_set_from): Handle
19516 TYPE_TYPELESS_STORAGE.
19517 (get_alias_set): Likewise.
19518 * tree-core.h (tree_type_common): Add typeless_storage flag.
19519 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
19520 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
19521 for types containing members with TYPE_TYPELESS_STORAGE.
19522 (place_field): Likewise.
19523 (layout_type): Likewise for ARRAY_TYPE.
19524 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
19525 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
19526 TYPE_TYPELESS_STORAGE.
19527 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
19529 2017-04-12 Jakub Jelinek <jakub@redhat.com>
19532 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
19533 first argument to type.
19535 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19539 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
19540 CONST0_RTX (mode) rather than const0_rtx where appropriate.
19541 (rs6000_expand_binop_builtin): Likewise.
19542 (rs6000_expand_ternop_builtin): Likewise; also add missing
19543 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
19544 vshasigma built-ins.
19545 * doc/extend.texi: Document that vec_xxpermdi's third argument
19546 must be a constant.
19548 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
19550 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
19551 Use shift_const cost parameter when calculating gain of STV shifts.
19553 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
19555 PR rtl-optimization/70478
19556 * lra-constraints.c (process_alt_operands): Check memory for
19557 disfavoring memory insn operand.
19559 2017-04-11 Jakub Jelinek <jakub@redhat.com>
19561 PR middle-end/80100
19562 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
19563 left shift in unsigned HOST_WIDE_INT type.
19565 PR rtl-optimization/80385
19566 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
19567 (not (neg X)) into (plus X -1) for complex or non-integral modes.
19570 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
19571 if they have any depend clauses.
19573 2017-04-11 Martin Liska <mliska@suse.cz>
19576 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
19577 * ipa-split.c (split_function): Create a local comdat symbol
19578 if caller is in a comdat group.
19580 2017-04-11 Martin Liska <mliska@suse.cz>
19583 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
19586 2017-04-11 Martin Sebor <msebor@redhat.com>
19588 PR middle-end/80364
19589 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
19590 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
19592 (directive::set_width, directive::set_precision, format_character):
19594 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
19597 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
19600 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
19601 conflict, set target->arch_name instead of target->cpu_name.
19603 2017-04-11 Richard Biener <rguenther@suse.de>
19605 PR tree-optimization/80374
19606 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
19607 build_zero_cst, remove fold_convertible_p check again.
19609 2017-04-11 Martin Liska <mliska@suse.cz>
19612 * ubsan.c (instrument_object_size): Do not instrument register
19615 2017-04-11 Jakub Jelinek <jakub@redhat.com>
19618 * config/i386/i386-builtin-types.def
19619 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
19620 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
19621 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
19622 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
19623 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
19624 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
19625 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
19626 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
19627 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
19628 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
19629 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
19630 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
19631 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
19632 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
19633 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
19634 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
19635 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
19636 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
19637 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
19638 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
19639 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
19640 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
19641 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
19642 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
19643 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
19644 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
19645 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
19646 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
19647 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
19648 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
19649 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
19650 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
19651 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
19652 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
19653 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
19654 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
19655 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
19656 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
19657 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
19658 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
19659 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
19660 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
19661 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
19662 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
19663 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
19664 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
19666 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
19667 flag to second_arg_count, handle 4 argument function type _COUNT
19668 aliases, handle second_arg_count on second argument rather than last.
19670 2017-04-10 Jeff Law <law@redhat.com>
19672 PR tree-optimization/80374
19673 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
19674 record anything if we can not convert integer_zero_node to the
19677 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
19680 * config/rs6000/rs6000.c (rs6000_option_override_internal):
19681 Enhance special handling given to the TARGET_P9_MINMAX option in
19682 relation to certain other options.
19684 2017-04-10 Bin Cheng <bin.cheng@arm.com>
19686 PR tree-optimization/80153
19687 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
19688 remove POINTER_PLUS_EXPR's base part directly, rather than through
19691 2017-04-10 Richard Biener <rguenther@suse.de>
19692 Bin Cheng <bin.cheng@arm.com>
19694 PR tree-optimization/80153
19695 * tree-affine.c (aff_combination_to_tree): Get base pointer from
19696 the first element of pointer type aff_tree. Build result expr in
19698 (add_elt_to_tree): Convert to type unconditionally. Remove other
19699 fold_convert calls.
19700 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
19701 (rewrite_use_nonlinear_expr): Check invariant using iv information.
19703 2017-04-10 Richard Biener <rguenther@suse.de>
19705 * tree-ssa-structalias.c (find_func_aliases): Properly handle
19708 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
19710 PR rtl-optimization/70478
19711 * lra-constraints.c (curr_small_class_check): New.
19712 (update_and_check_small_class_inputs): New.
19713 (process_alt_operands): Update curr_small_class_check. Disfavor
19714 alternative insn memory operands. Check available regs for small
19717 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
19720 * config/mips/mips.opt (-mvirt): Update description.
19721 * doc/invoke.texi (-mvirt): Likewise.
19723 2017-04-10 Richard Biener <rguenther@suse.de>
19725 PR middle-end/80362
19726 * fold-const.c (fold_binary_loc): Look at unstripped ops when
19727 looking for NEGATE_EXPR in -A / -B to A / B folding.
19729 2017-04-10 Martin Liska <mliska@suse.cz>
19731 PR gcov-profile/80224
19732 * gcov.c (print_usage): Fix usage string.
19733 (get_gcov_intermediate_filename): Remove.
19734 (output_gcov_file): Use both for normal and intermediate format.
19735 (generate_results): Do not initialize special file for
19736 intermediate format.
19738 2017-04-10 Richard Biener <rguenther@suse.de>
19740 PR tree-optimization/80304
19741 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
19744 2017-04-10 Nathan Sidwell <nathan@acm.org>
19747 * config/rs6000/rs6000.c (rs6000_vector_type): New.
19748 (rs6000_init_builtins): Use it.
19750 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19752 * config/arm/arm.md (<mrc>): Add mode to SET source.
19753 (<mrrc>): Likewise.
19755 2017-04-10 Richard Biener <rguenther@suse.de>
19757 PR middle-end/80344
19758 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
19760 2017-04-10 Jakub Jelinek <jakub@redhat.com>
19763 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
19764 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
19765 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
19766 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
19767 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
19768 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
19769 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
19770 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
19771 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
19772 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
19773 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
19774 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
19775 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
19776 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
19777 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
19778 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
19779 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
19780 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
19781 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
19782 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
19783 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
19784 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
19785 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
19787 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
19789 PR rtl-optimization/70478
19790 * lra-constraints.c: Reverse the last patch.
19792 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
19794 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
19795 Add comment for WCHAR_T.
19797 2017-04-08 Martin Liska <mliska@suse.cz>
19800 2017-04-07 Martin Liska <mliska@suse.cz>
19803 * ipa-split.c (split_function): Add function part to a same comdat
19806 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19809 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
19811 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
19813 * rs6000/rs6000.c (vec_load_pendulum): Rename...
19814 (vec_pairing): ...to this.
19815 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
19816 (rs6000_sched_init): Adjust for name change.
19817 (struct rs6000_sched_context): Likewise.
19818 (rs6000_init_sched_context): Likewise.
19819 (rs6000_set_sched_context): Likewise.
19821 2017-04-07 Jakub Jelinek <jakub@redhat.com>
19827 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
19829 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
19830 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
19831 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
19833 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
19835 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
19837 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
19839 PR rtl-optimization/70703
19840 * ira-color.c (update_conflict_hard_regno_costs): Use
19841 int64_t instead of HOST_WIDE_INT.
19843 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
19845 PR rtl-optimization/70478
19846 * lra-constraints.c (process_alt_operands): Disfavor alternative
19847 insn memory operands.
19849 2017-04-07 Jeff Law <law@redhat.com>
19851 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
19852 CALL and NOTE_INSN_CALL_ARG_LOCATION.
19854 2017-04-07 Martin Liska <mliska@suse.cz>
19857 * config/aarch64/aarch64.c (aarch64_process_target_attr):
19858 Show error message instead of an ICE.
19860 2017-04-07 Martin Liska <mliska@suse.cz>
19863 * ipa-split.c (split_function): Add function part to a same comdat
19866 2017-04-07 Richard Biener <rguenther@suse.de>
19868 PR middle-end/80341
19869 * tree.c (get_unwidened): Also handle ! for_type case for
19871 * convert.c (do_narrow): Split out from ...
19872 (convert_to_integer_1): ... here. Do not pass final truncation
19873 type to get_unwidened for TRUNC_DIV_EXPR.
19875 2017-04-07 Richard Biener <rguenther@suse.de>
19877 * tree-affine.c (wide_int_ext_for_comb): Take type rather
19879 (aff_combination_const): Adjust.
19880 (aff_combination_scale): Likewise.
19881 (aff_combination_add_elt): Likewise.
19882 (aff_combination_add_cst): Likewise.
19883 (aff_combination_convert): Likewise.
19884 (add_elt_to_tree): Likewise. Remove unused argument.
19885 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
19887 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
19889 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
19891 * config/arm/arm.c (arm_default_short_enums): Use
19892 ARM_DEFAULT_SHORT_ENUMS.
19893 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
19895 2017-04-06 Jakub Jelinek <jakub@redhat.com>
19898 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
19899 members with redundant out-of-class redeclaration.
19901 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
19904 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
19905 * config/i386/i386.md (*zero_extendsidi2):
19906 Add (?*x,*x) and (?*v,*v) alternatives.
19908 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
19911 * config/i386/i386.c (ix86_expand_builtin)
19912 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
19913 mode from insn data. Convert operands to insn operand mode.
19914 Copy operands that don't satisfy insn predicate to a register.
19916 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
19918 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
19921 2017-04-06 Richard Biener <rguenther@suse.de>
19923 PR tree-optimization/80334
19924 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
19925 preserve alignment of accesses.
19927 2017-04-06 Richard Biener <rguenther@suse.de>
19929 PR tree-optimization/80262
19930 * tree-sra.c (build_ref_for_offset): Preserve address-space
19932 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
19933 Drop useless address-space information on MEM_REF offsets.
19935 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
19937 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
19939 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
19941 PR rtl-optimization/70703
19942 * ira-color.c (update_conflict_hard_regno_costs): Use
19943 HOST_WIDE_INT instead of long.
19945 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
19948 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
19949 not defined for x86_64 target. Add -mmmx target option when __SSE2__
19951 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
19952 for x86_64 target. Handle -m3dnowa option.
19954 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
19956 PR rtl-optimization/70703
19957 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
19958 (update_conflict_hard_regno_costs): Use long instead of unsigned
19959 arithmetic for cost calculation.
19961 2017-04-05 Jakub Jelinek <jakub@redhat.com>
19962 Bernd Edlinger <bernd.edlinger@hotmail.de>
19965 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
19968 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
19971 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
19972 ptr_mode with Pmode throughout.
19973 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
19974 into probe_stack_range and use DImode.
19976 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
19979 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
19980 call_eh_return is true.
19982 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19984 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
19985 Initialize last_match_fntype_index.
19987 2017-04-05 Jakub Jelinek <jakub@redhat.com>
19990 * tree-nvr.c: Include internal-fn.h.
19991 (pass_return_slot::execute): Ignore internal calls without
19994 2017-04-04 Jakub Jelinek <jakub@redhat.com>
19995 Richard Biener <rguenther@suse.de>
19998 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
19999 captures used multiple times, except for the last use.
20000 * generic-match-head.c: Include gimplify.h.
20002 2017-04-04 Jakub Jelinek <jakub@redhat.com>
20004 PR tree-optimization/79390
20005 * target.h (struct noce_if_info): Declare.
20006 * targhooks.h (default_noce_conversion_profitable_p): Declare.
20007 * target.def (noce_conversion_profitable_p): New target hook.
20008 * ifcvt.h (struct noce_if_info): New type, moved from ...
20009 * ifcvt.c (struct noce_if_info): ... here.
20010 (noce_conversion_profitable_p): Renamed to ...
20011 (default_noce_conversion_profitable_p): ... this. No longer
20013 (noce_try_store_flag_constants, noce_try_addcc,
20014 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
20015 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
20016 instead of noce_conversion_profitable_p.
20017 * config/i386/i386.c: Include ifcvt.h.
20018 (ix86_option_override_internal): Don't override
20019 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
20020 (ix86_noce_conversion_profitable_p): New function.
20021 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
20022 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
20023 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
20024 * doc/tm.texi: Regenerated.
20026 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20028 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
20031 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
20034 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
20035 instructions for small multiply cores.
20037 2017-04-04 Jeff Law <law@redhat.com>
20039 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
20041 (mips_expand_vec_perm_const): Initialize elements in orig_perm
20042 that are not set by the loop over the elements.
20044 2017-04-04 Jakub Jelinek <jakub@redhat.com>
20047 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
20048 int mode, convert_modes it to mode as unsigned, otherwise use
20049 lowpart_subreg to mode rather than SImode.
20050 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
20051 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
20052 Use DImode instead of SImode for the shift count operand.
20053 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
20056 2017-04-04 Richard Biener <rguenther@suse.de>
20058 PR middle-end/80281
20059 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
20060 arithmetic done for the negate or the plus. Simplify.
20061 (A - (-B) -> A + B): Likewise.
20062 * fold-const.c (split_tree): Make sure to not negate pointers.
20064 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
20066 PR rtl-optimization/60818
20067 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
20068 a compare of comparisons with the thing compared if this results
20069 in a different machine mode.
20071 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
20073 * alias.c (base_alias_check): Fix typo in comment.
20074 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
20075 * cgraphunit.c (symbol_table::compile): Likewise.
20076 * collect2.c (maybe_run_lto_and_relink): Likewise.
20077 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
20078 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
20079 * config/avr/avr.c (avr_map_op_t): Likewise.
20080 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
20081 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
20082 * config/epiphany/epiphany.md (movcc): Likewise.
20083 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
20084 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
20086 * config/mips/mips.c (mips_save_restore_reg): Likewise.
20087 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
20088 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
20089 * config/sh/sh.c (sh_rtx_costs): Likewise.
20090 * fold-const.c (fold_truth_andor): Likewise.
20091 * genautomata.c (collapse_flag): Likewise.
20092 * gengtype.h (struct type::u::s): Likewise.
20093 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
20094 * input.c (FORMAT_AMOUNT): Likewise.
20095 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
20096 (known_aggs_to_agg_replacement_list): Likewise.
20097 * ipa-inline-analysis.c: Likewise.
20098 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
20099 * ipa-polymorphic-call.c
20100 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
20101 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
20102 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
20104 * modulo-sched.c (apply_reg_moves): Likewise.
20105 * omp-expand.c (build_omp_regions_1): Likewise.
20106 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
20107 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
20108 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
20109 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
20110 * value-prof.c: Likewise.
20111 * var-tracking.c (val_reset): Likewise.
20113 2017-04-03 Richard Biener <rguenther@suse.de>
20115 PR tree-optimization/80275
20116 * fold-const.c (split_address_to_core_and_offset): Handle
20119 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
20121 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
20122 descriptors is at least equal to that of functions.
20124 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
20126 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
20128 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
20131 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
20132 (mov<IMOD4:mode>): New expander.
20133 (*mov<IMOD4:mode>_internal): New insn and split pattern.
20135 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
20137 PR rtl-optimization/79405
20138 * fwprop.c (propagations_left): New variable.
20139 (forward_propagate_into): Decrement it.
20140 (fwprop_init): Initialize it.
20141 (fw_prop): If the variable has reached zero, stop propagating.
20142 (fwprop_addr): Ditto.
20144 2017-03-31 Jakub Jelinek <jakub@redhat.com>
20147 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
20148 a FUNCTION_DECL, pass it as decl instead of origin to
20151 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
20153 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
20156 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
20159 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
20160 TARGET_VSX_SMALL_INTEGER.
20162 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20164 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
20165 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
20167 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
20169 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
20170 extraction from odd-numbered MSA register.
20172 2017-03-31 Jakub Jelinek <jakub@redhat.com>
20174 PR middle-end/80173
20175 * expmed.c (store_bit_field_1): Don't attempt to create
20176 a word subreg out of hard registers wider than word if they
20177 have HARD_REGNO_NREGS of 1 for their mode.
20179 PR middle-end/80163
20180 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
20181 conversions to integer types wider than word and pointer.
20184 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
20185 (rtx_equal_for_cselib_p): Pass 0 to it.
20186 * cselib.c (cselib_hasher::equal): Likewise.
20187 (rtx_equal_for_cselib_1): Add depth argument. If depth
20188 is 128, don't look up VALUE locs and punt. Increment
20189 depth in recursive calls when walking VALUE locs.
20191 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
20193 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
20194 (make_gcov_file_name): Use the canonical path name for generating
20196 (read_line): Fix handling of files with ascii null bytes.
20198 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
20200 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
20201 to initialise a vector register instead
20202 of using a const_int.
20204 2017-03-30 Jakub Jelinek <jakub@redhat.com>
20206 PR translation/80189
20207 * gimplify.c (omp_default_clause): Use %qs instead of %s in
20208 diagnostic messages.
20210 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
20213 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
20214 (dfp_diex_<mode>): Update mode of operand 1.
20215 * doc/extend.texi (dxex, dxexq): Document change to return type.
20216 (diex, diexq): Document change to argument type.
20218 2017-03-30 Martin Jambor <mjambor@suse.cz>
20221 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
20222 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
20223 it reflects the signature changes performed at the callee side.
20224 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
20225 to cgraph_build_function_type_skip_args.
20226 (build_function_decl_skip_args): Adjust call to the above function.
20228 2017-03-30 Jakub Jelinek <jakub@redhat.com>
20231 * config/i386/sse.md
20232 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
20233 register as dest whenever it is a MEM not rtx_equal_p to the
20234 corresponding dup operand, and when forcing into reg move the
20235 reg into the memory afterwards.
20236 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
20237 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
20238 for the force_reg mode.
20239 (avx512vl_vextractf128<mode>): Use register as dest either
20240 always when a MEM, or when it is a MEM not rtx_equal_p to the
20241 corresponding dup operand, or even not when it is a CONST_VECTOR
20242 depending on the mode and lo vs. hi.
20243 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
20245 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
20246 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
20247 Likewise. Require that operands[2] is even.
20248 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
20249 Remove extraneous parens. Require that operands[2] is a multiple
20251 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
20252 operands[0] is a MEM if <mask_applied>, the predicates/constraints
20253 disallow memory then.
20255 2017-03-30 Richard Biener <rguenther@suse.de>
20257 PR tree-optimization/77498
20258 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
20259 to non-constants over backedges.
20261 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
20263 PR rtl-optimization/80233
20264 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
20265 as last_combined_insn. Do not test for BARRIER_P separately.
20267 2017-03-29 Andreas Schwab <schwab@suse.de>
20270 * calls.c (prepare_call_address): Convert funexp to Pmode before
20271 copying to temp reg.
20273 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20275 PR tree-optimization/80158
20276 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
20277 Handle possible future case of more than one alternate
20279 (replace_rhs_if_not_dup): Likewise.
20280 (replace_one_candidate): Likewise.
20282 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
20284 PR rtl-optimization/80193
20285 * ira.c (ira): Do not check allocation for LRA.
20287 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
20289 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
20290 (nvptx_output_simt_exit): Declare.
20291 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
20292 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
20293 (init_softstack_frame): Move initialization of crtl->is_leaf to...
20294 (nvptx_declare_function_name): ...here. Emit declaration of local
20295 memory space buffer for omp_simt_enter insn.
20296 (nvptx_output_unisimt_switch): New.
20297 (nvptx_output_softstack_switch): New.
20298 (nvptx_output_simt_enter): New.
20299 (nvptx_output_simt_exit): New.
20300 * config/nvptx/nvptx.h (struct machine_function): New fields
20301 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
20302 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
20303 (UNSPECV_SIMT_EXIT): Ditto.
20304 (omp_simt_enter_insn): New insn.
20305 (omp_simt_enter): New expansion.
20306 (omp_simt_exit): New insn.
20307 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
20309 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
20310 (expand_GOMP_SIMT_ENTER_ALLOC): New.
20311 (expand_GOMP_SIMT_EXIT): New.
20312 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
20313 (GOMP_SIMT_ENTER_ALLOC): Ditto.
20314 (GOMP_SIMT_EXIT): Ditto.
20315 * target-insns.def (omp_simt_enter): New insn.
20316 (omp_simt_exit): Ditto.
20317 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
20319 (lower_rec_simd_input_clauses): Implement SIMT privatization.
20320 (lower_rec_input_clauses): Likewise.
20321 (lower_lastprivate_clauses): Handle SIMT privatization.
20323 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
20324 (ompdevlow_adjust_simt_enter): New.
20325 (find_simtpriv_var_op): New.
20326 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
20327 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
20329 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
20330 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
20331 (copy_decl_for_dup_finish): Ditto.
20333 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
20335 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
20338 * config/i386/i386.c (ix86_option_override_internal): Always
20339 allow -mpreferred-stack-boundary=3 for 64-bit targets.
20341 2017-03-28 Bin Cheng <bin.cheng@arm.com>
20343 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
20345 2017-03-28 Bin Cheng <bin.cheng@arm.com>
20347 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
20348 mark new edge's irreducible flag accordign to it.
20349 (vect_do_peeling): Check loop preheader edge's irreducible flag
20350 and pass it to function slpeel_add_loop_guard.
20352 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
20354 PR tree-optimization/80218
20355 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
20356 Update block frequencies and counts.
20358 2017-03-28 Richard Biener <rguenther@suse.de>
20360 PR tree-optimization/78644
20361 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
20362 of a simplification result we may not use it at all.
20364 2017-03-28 Richard Biener <rguenther@suse.de>
20367 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
20368 without arguments, generate default definition of a SSA name.
20370 2017-03-28 Richard Biener <rguenther@suse.de>
20372 PR middle-end/80222
20373 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
20374 TYPE_REF_CAN_ALIAS_ALL references.
20375 * fold-const.c (fold_indirect_ref_1): Likewise.
20377 2017-03-28 Martin Liska <mliska@suse.cz>
20380 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
20381 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
20383 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
20384 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
20386 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
20387 (EXTRA_SPECS): Define.
20388 (SUBTARGET_EXTRA_SPECS): Likewise.
20389 (SUBTARGET_CPP_SPEC): Likewise.
20390 * config/arc/elf.h (EXTRA_SPECS): Renamed to
20391 SUBTARGET_EXTRA_SPECS.
20392 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
20394 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
20396 * config/arc/simdext.md (vst64_insn): Update pattern.
20397 (vld32wh_insn): Likewise.
20398 (vld32wl_insn): Likewise.
20399 (vld64_insn): Likewise.
20400 (vld32_insn): Likewise.
20402 2017-03-28 Marek Polacek <polacek@redhat.com>
20405 * fold-const.c (fold_comparison): Use protected_set_expr_location
20406 instead of SET_EXPR_LOCATION.
20408 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
20410 * tree.c (add_expr): Avoid name lookup warning.
20412 2017-03-27 Jeff Law <law@redhat.com>
20414 PR tree-optimization/80216
20415 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
20416 function name. Limit recursion depth.
20417 (record_temporary_equivalences): Corresponding changes.
20419 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
20421 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
20424 2017-03-27 Jakub Jelinek <jakub@redhat.com>
20427 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
20429 * cfgcleanup.c (reg_note_cfa_p): New array.
20430 (insns_have_identical_cfa_notes): New function.
20431 (old_insns_match_p): Don't cross-jump in between /f
20432 and non-/f instructions. If both i1 and i2 are frame related,
20433 verify all CFA notes, their order and content.
20435 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
20438 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
20439 HImode and SImode with zero extend to DImode to one insn.
20440 (bswap<mode>2_extenddi): Likewise.
20441 (bswapsi2_extenddi): Likewise.
20442 (bswaphi2_extendsi): Likewise.
20443 (bswaphi2): Combine bswap HImode and SImode into one insn.
20444 Separate memory insns from swapping register.
20445 (bswapsi2): Likewise.
20446 (bswap<mode>2): Likewise.
20447 (bswaphi2_internal): Delete, no longer used.
20448 (bswapsi2_internal): Likewise.
20449 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
20450 store, and gpr<-gpr swap insns.
20451 (bswap<mode>2_store): Likewise.
20452 (bswaphi2_reg): Register only splitter, combine with the splitter.
20453 (bswaphi2 splitter): Likewise.
20454 (bswapsi2_reg): Likewise.
20455 (bswapsi2 splitter): Likewise.
20456 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
20457 the insns into load, store, and register/register insns.
20458 (bswapdi2_ldbrx): Likewise.
20459 (bswapdi2_load): Likewise.
20460 (bswapdi2_store): Likewise.
20461 (bswapdi2_reg): Likewise.
20463 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
20465 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
20466 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
20468 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
20471 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
20473 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
20474 special handling for target option conflicts between dform
20475 options (-mpower9-dform, -mpower9-dform-vector,
20476 -mpower9-dform-scalar) and -mno-direct-move.
20478 2017-03-27 Richard Biener <rguenther@suse.de>
20480 PR tree-optimization/80181
20481 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
20483 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
20485 * config/arc/predicates.md (move_double_src_operand): Replace the
20486 call to move_double_src_operand with a call to address_operand.
20488 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
20490 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
20491 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
20492 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
20494 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
20496 * config/arc/predicates.md (long_immediate_loadstore_operand):
20497 Consider scaled addresses cases.
20499 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
20501 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
20502 restored when in interrupt.
20503 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
20504 doesn't have delay slot.
20506 2017-03-27 Richard Biener <rguenther@suse.de>
20509 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
20510 inlined thunk clones.
20512 2017-03-27 Jakub Jelinek <jakub@redhat.com>
20515 * asan.c (instrument_derefs): Copy over last operand from
20516 original COMPONENT_REF to the new COMPONENT_REF with
20517 DECL_BIT_FIELD_REPRESENTATIVE.
20518 * ubsan.c (instrument_object_size): Likewise.
20520 2017-03-27 Richard Biener <rguenther@suse.de>
20522 PR tree-optimization/80170
20523 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
20524 sure DR/SCEV didnt fold in constants we do not see when looking
20525 at the reference base alignment.
20527 2017-03-27 Richard Biener <rguenther@suse.de>
20529 PR middle-end/80171
20530 * gimple-fold.c (fold_ctor_reference): Properly guard against
20531 NULL return value from canonicalize_constructor_val.
20533 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
20536 * config/i386/i386.c (ix86_expand_builtin)
20537 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
20538 flags reg setting and flags reg using instructions.
20539 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
20540 clobbering instructions to zero extend op2.
20542 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
20544 * doc/install.texi (Configuration) <--with-aix-soname>:
20545 Update link to AIX ld.
20547 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
20549 PR rtl-optimization/80160
20550 PR rtl-optimization/80159
20551 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
20552 reg_alternate_class into account.
20554 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
20557 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
20558 to consider in curr_insn_transform.
20560 2017-03-24 Jakub Jelinek <jakub@redhat.com>
20562 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
20563 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
20564 and emit_mode_inner.
20566 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20568 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
20569 argument to the overloaded builtin variants. Use the new flag to
20570 deprecate certain builtin variants.
20571 * config/s390/s390-builtin-types.def: Add new builtin types.
20572 * config/s390/s390-builtins.h: Support new flags field for
20573 overloaded builtins.
20574 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
20575 (s390_macro_to_expand): Enable vector float data type.
20576 (s390_cpu_cpp_builtins_internal): Indicate support of the new
20577 builtins by incrementing the __VEC__ version number.
20578 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
20580 (s390_resolve_overloaded_builtin): Emit error messages depending
20581 on the builtin flags.
20582 * config/s390/s390.c (s390_expand_builtin): Support additional
20583 flags argument. Change error message to match the messages
20584 emitted in s390-c.c.
20585 * config/s390/s390.md: New UNSPEC_* constants.
20586 (op_type): Add new instruction types.
20587 * config/s390/vecintrin.h: Add new builtins and test data class
20589 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
20590 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
20591 (VEC_INEXACT, VEC_NOINEXACT): New constants.
20592 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
20593 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
20594 ("vec_mergel<mode>"): V_HW -> VEC_HW.
20596 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
20597 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
20598 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
20599 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
20601 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
20602 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
20603 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
20604 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
20606 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
20607 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
20608 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
20609 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
20610 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
20611 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
20612 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
20614 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
20615 ("vec_scatter_element<V_HW_4:mode>_DI")
20616 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
20617 ("vec_fpint<mode>", "vflls")
20618 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
20619 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
20620 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
20621 ("*vec_cmphe<mode>_cc"): ... these.
20623 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
20624 mode constant instead of magic value.
20626 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20628 * config/s390/s390.c (s390_expand_vec_compare): Support other
20629 vector floating point modes than just V2DF.
20630 (s390_expand_vcond): Likewise.
20631 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
20632 (s390_cannot_change_mode_class): Prevent mode changes between TF
20633 and V1TF in vector registers.
20634 * config/s390/s390.md (DF, SF): New mode attributes.
20635 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
20636 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
20637 SFmode support for VRs.
20638 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
20640 (VFT, VF_HW): New mode iterators.
20641 (vw, sdx): New mode attributes.
20642 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
20643 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
20644 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
20645 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
20646 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
20647 also the new vector floating point modes. Renaming to ...
20649 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
20650 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
20651 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
20652 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
20653 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
20654 ("vec_unordered<mode>"): ... these.
20656 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
20657 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
20658 ("*vec_extendv2df"): New insn definitions.
20660 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20662 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
20663 ("mulditi3_2", "*muldi3_sign"): New patterns.
20664 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
20665 rename the pattern definition.
20667 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20669 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
20671 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
20673 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20675 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
20676 instruction if possible.
20677 * config/s390/vector.md (vec_halfnumelts): New mode
20679 ("*vec_vllezlf<mode>"): New pattern.
20681 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20683 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
20684 ("popcountv4si2", "popcountv2di2"): Rename to ...
20685 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
20686 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
20688 ("popcount<mode>2_vxe"): New pattern.
20690 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20692 * common/config/s390/s390-common.c (processor_flags_table): Add
20694 * config.gcc: Add arch12.
20695 * config/s390/driver-native.c (s390_host_detect_local_cpu):
20696 Default to arch12 for unknown CPU model numbers.
20697 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
20698 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
20699 PROCESSOR_max sanity check.
20700 * config/s390/s390-opts.h (enum processor_type): Add
20702 * config/s390/s390.c (processor_table): Add arch12.
20703 (s390_expand_builtin): Add check for B_VXE flag.
20704 (s390_issue_rate): Add PROCESSOR_ARCH12.
20705 (s390_get_sched_attrmask): Likewise.
20706 (s390_get_unit_mask): Likewise.
20707 (s390_sched_score): Enable z13 scheduling for arch12.
20708 (s390_sched_reorder): Likewise.
20709 (s390_sched_variable_issue): Likewise.
20710 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
20712 (s390_tune_attr): Use z13 scheduling also for arch12.
20713 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
20714 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
20715 (TARGET_VXE_P): New macros.
20716 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
20717 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
20718 * config/s390/s390.opt: Add arch12 as processor_type.
20720 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20722 * config/s390/s390.md
20723 ("fixuns_truncdddi2", "fixuns_trunctddi2")
20724 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
20725 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
20727 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
20728 Rename expanders to ...
20730 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
20731 ("fixuns_truncdddi2_emu"): ... these.
20733 ("fixuns_trunc<mode>si2_emu"): New expander.
20735 ("*fixuns_truncdfdi2_z13"): Rename to ...
20736 ("*fixuns_truncdfdi2_vx"): ... this.
20738 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20740 * config/s390/2964.md: Remove the single element vector compare
20741 instructions which are no longer used.
20742 * config/s390/s390.c (s390_select_ccmode): Remove handling of
20744 (s390_canonicalize_comparison): Remove handling of DFmode
20746 (s390_expand_vec_compare_scalar): Remove function.
20747 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
20748 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
20750 ("*cmp<mode>_ccs"): Add wfcdb instruction.
20752 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20754 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
20756 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
20757 will anyway by matched by mov<mode>_64dfp.
20759 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20761 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
20762 vlef/vstef. Add missing operand to vleif.
20764 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20766 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
20767 pair for all vector types with 64 bit elements.
20768 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
20769 * config/s390/vector.md (V_HW_64): ... here.
20770 (V_128_NOSINGLE): New mode iterator.
20771 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
20772 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
20773 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
20774 ("*vec_load_pairv2di"): Change to ...
20775 ("*vec_load_pair<mode>"): ... this one.
20777 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20779 * config/s390/constraints.md: Add comments.
20780 (jKK): Reject element sizes > 8 bytes.
20781 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
20783 * config/s390/s390.md: Add the s_operand checks formerly in
20784 s390_split_ok_p to various splitters where they are still
20786 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
20787 for 128 bit vectors. Plus two splitters.
20789 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20791 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
20794 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20797 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
20798 error if the boundary argument is not constant.
20800 2017-03-24 Jakub Jelinek <jakub@redhat.com>
20802 PR rtl-optimization/80112
20803 * loop-doloop.c (doloop_condition_get): Don't check condition
20804 if cmp isn't SET with IF_THEN_ELSE src.
20806 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20808 PR tree-optimization/80158
20809 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
20810 replacing a candidate statement, also replace it for the
20811 candidate's alternate interpretation.
20812 (replace_rhs_if_not_dup): Likewise.
20813 (replace_one_candidate): Likewise.
20815 2017-03-24 Richard Biener <rguenther@suse.de>
20817 PR tree-optimization/80167
20818 * graphite-isl-ast-to-gimple.c
20819 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
20821 (translate_isl_ast_to_gimple::get_rename): Likewise.
20823 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
20825 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
20826 handling of certain combinations of target options, including the
20827 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
20828 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
20830 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20833 * config/arm/arm.md (*load_multiple): Add reload_completed to
20834 matching condition.
20836 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20837 Richard Biener <rguenth@suse.de>
20839 PR tree-optimization/79908
20840 PR tree-optimization/80136
20841 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
20842 been cast away, gimplify_and_add suffices.
20844 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
20846 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
20848 2017-03-23 Richard Biener <rguenther@suse.de>
20850 PR tree-optimization/80032
20851 * gimplify.c (gimple_push_cleanup): Forced unconditional
20852 cleanups still have to go to the conditional_cleanups
20855 2017-03-22 Jakub Jelinek <jakub@redhat.com>
20857 PR tree-optimization/80072
20858 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
20860 (next_operand_entry_id): Change type to unsigned int.
20861 (sort_by_operand_rank): Make sure to return the right return value
20862 even if unsigned fields are bigger than INT_MAX.
20863 (struct oecount): Change cnt and id type to unsigned int.
20864 (oecount_hasher::equal): Formatting fix.
20865 (oecount_cmp): Make sure to return the right return value
20866 even if unsigned fields are bigger than INT_MAX.
20867 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
20870 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
20871 TREE_READONLY on result if writing it more than once.
20874 * doc/invoke.texi (-fsanitize=thread): Document that with
20875 -fnon-call-exceptions atomics are not able to throw
20879 * tsan.c: Include tree-eh.h.
20880 (instrument_builtin_call): Call maybe_clean_eh_stmt or
20881 maybe_clean_or_replace_eh_stmt where needed.
20882 (instrument_memory_accesses): Add cfg_changed argument.
20883 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
20884 if it returned true.
20885 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
20887 PR rtl-optimization/63191
20888 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
20889 wrapper function, moved the whole old content into ...
20890 (ix86_delegitimize_address_1): ... this. New inline function.
20891 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
20892 true as last argument instead of ix86_delegitimize_address.
20894 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
20896 * config/aarch64/aarch64.c (generic_branch_cost): Copy
20897 cortexa57_branch_cost.
20899 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
20901 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
20903 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20906 * doc/md.texi (Constraints): Document wA constraint.
20907 * config/rs6000/constraints.md (wA): New.
20908 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
20909 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
20910 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
20911 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
20913 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
20916 * gimplify.c (is_oacc_declared): New function.
20917 (oacc_default_clause): Use it to set default flags for acc declared
20918 variables inside parallel regions.
20919 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
20920 declared variables.
20921 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
20922 declare attribute to any decl as necessary.
20924 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
20927 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
20928 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
20929 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
20930 (arm_arch_lpae): This.
20931 * config/arm/arm.c (arm_arch7ve): Rename into ...
20932 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
20933 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
20936 2017-03-22 Martin Liska <mliska@suse.cz>
20939 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
20940 error message instead of an ICE.
20942 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20944 * doc/extend.texi (6.11 Additional Floating Types): Revise.
20946 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
20948 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
20950 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
20953 2017-03-21 Martin Sebor <msebor@redhat.com>
20955 * doc/extend.texi: Use "cannot" instead of "can't."
20956 * doc/hostconfig.texi: Same.
20957 * doc/install.texi: Same.
20958 * doc/invoke.texi: Same.
20959 * doc/loop.texi: Same.
20960 * doc/md.texi: Same.
20961 * doc/objc.texi: Same.
20962 * doc/rtl.texi: Same.
20963 * doc/tm.texi: Same.
20964 * doc/tm.texi.in: Same.
20965 * doc/trouble.texi: Same.
20967 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
20970 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
20971 (collect_checksum_attributes): Set it.
20972 (die_checksum_ordered): Use it.
20974 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20976 PR tree-optimization/79908
20977 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
20978 change: For a VA_ARG whose LHS has been cast away, use
20979 force_gimple_operand to construct the side effects.
20981 2017-03-21 David Malcolm <dmalcolm@redhat.com>
20983 PR translation/80001
20984 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
20985 more amenable to translation.
20986 (oacc_loop_auto_partitions): Likewise.
20988 2017-03-21 Marek Polacek <polacek@redhat.com>
20989 Martin Sebor <msebor@redhat.com>
20991 PR tree-optimization/80109
20992 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
20993 on INTEGRAL_TYPE_P.
20995 2017-03-21 Jakub Jelinek <jakub@redhat.com>
20996 Segher Boessenkool <segher@kernel.crashing.org>
20999 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
21000 check reg_used_between_p between insn and one of succ or succ2
21001 depending on if succ is artificial insn not inserted into insn
21004 2017-03-21 Martin Liska <mliska@suse.cz>
21006 PR gcov-profile/80081
21007 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
21008 * doc/gcc.texi: Include gcov-dump stuff.
21009 * doc/gcov-dump.texi: New file.
21011 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
21013 PR rtl-optimization/79150
21014 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
21015 conditional jump, if the jump is the last insn of the loop.
21017 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21018 Richard Biener <rguenth@suse.de>
21020 PR tree-optimization/79908
21021 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
21022 been cast away, use force_gimple_operand to construct the side
21025 2017-03-21 Martin Liska <mliska@suse.cz>
21027 PR libfortran/79956
21028 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
21031 2017-03-21 Brad Spengler <spender@grsecurity.net>
21034 * plugin.c (htab_hash_plugin): New function.
21035 (add_new_plugin): Use it and adjust.
21036 (parse_plugin_arg_opt): Adjust.
21037 (init_one_plugin): Likewise.
21039 2017-03-21 Richard Biener <rguenther@suse.de>
21041 PR tree-optimization/80032
21042 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
21043 if set force the cleanup to happen unconditionally.
21044 (gimplify_target_expr): Push inserted clobbers with force_uncond
21045 to avoid them being removed by control-dependent DCE.
21047 2017-03-21 Richard Biener <rguenther@suse.de>
21049 PR tree-optimization/80122
21050 * tree-inline.c (copy_bb): Do not expans va-arg packs or
21051 va_arg_pack_len when the inlined call stmt requires pack
21053 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
21055 2017-03-21 Jakub Jelinek <jakub@redhat.com>
21058 * tsan.c (instrument_builtin_call): If the memory model argument
21059 is not a constant, assume it is valid.
21062 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
21065 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
21067 PR rtl-optimization/79910
21068 * combine.c (can_combine_p): Do not allow combining an I0 or I1
21069 if its dest is used by an insn before I2 (other than the combined
21070 insns themselves, which are properly handled already).
21072 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
21075 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
21077 * combine.c (record_used_regs): New static function.
21078 (try_combine): Handle situations where there is an additional
21079 instruction between I2 and I3 which needs to have a LOG_LINK
21083 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
21085 * combine.c (try_combine): Delete redundant i1 test. Call
21086 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
21088 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21091 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
21092 alternatives 13/14.
21094 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21096 PR tree-optimization/80054
21097 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
21098 the optimization if a PHI or any of its arguments is not dominated
21099 by the candidate's basis. Use gphi* rather than gimple* as
21101 (replace_profitable_candidates): Clean up a gimple* variable that
21102 should be a gphi* variable.
21104 2017-03-20 Martin Sebor <msebor@redhat.com>
21107 * doc/extend.texi (attribute constructor): Document present limitation.
21109 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
21112 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
21113 __POWER9_VECTOR__ #ifdef control, change template definition to
21114 use Power9-specific built-in function.
21115 (vec_any_eq): Likewise.
21116 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
21117 to control outcomes from this test.
21118 (vector_ae_<mode>p): For VEC_F modes, likewise.
21120 2017-03-20 Ian Lance Taylor <iant@google.com>
21122 * config/i386/i386.c (ix86_function_regparm): Save an extra
21123 register for -fsplit-stack with DECL_STATIC_CHAIN.
21125 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
21128 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
21129 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
21131 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
21133 * config/riscv/riscv.c (riscv_print_operand): Use "fence
21135 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
21138 2017-03-20 Marek Polacek <polacek@redhat.com>
21141 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
21143 2017-03-20 Richard Biener <rguenther@suse.de>
21145 PR tree-optimization/80113
21146 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
21147 allocate extra SSA name for PHI def.
21148 (add_close_phis_to_outer_loops): Likewise.
21149 (add_close_phis_to_merge_points): Likewise.
21150 (copy_loop_close_phi_args): Likewise.
21151 (copy_cond_phi_nodes): Likewise.
21153 2017-03-20 Martin Liska <mliska@suse.cz>
21155 PR middle-end/79753
21156 * tree-chkp.c (chkp_build_returned_bound): Do not build
21157 returned bounds for a LHS that's not a BOUNDED_P type.
21159 2017-03-20 Martin Liska <mliska@suse.cz>
21163 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
21164 COMPLEX_CST and VECTOR_CST.
21166 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21169 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
21170 target operand. A new splitter adds the clobber statement in case
21171 the target operand is dead anyway.
21173 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
21175 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
21176 to age-old versions of binutils and glibc.
21178 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
21180 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
21182 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
21184 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
21186 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
21188 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
21189 requirement for binutils 2.13.
21191 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
21193 * combine.c (try_combine): Delete redundant i1 test. Call
21194 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
21196 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
21198 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
21199 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
21201 <riscv64-*-elf>: Re-arrange section
21202 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
21203 <riscv32-*-linux>: Likewise.
21204 <riscv64-*-elf>: Likewise
21205 <riscv64-*-linux>: Likewise.
21207 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
21210 * aarch64.opt(verbose-cost-dump): Fix typo.
21212 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
21215 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
21216 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
21218 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
21220 * reload.c (find_reloads): When reloading a nonoffsettable address,
21221 use RELOAD_OTHER for it and its address reloads.
21223 PR rtl-optimization/79910
21224 * combine.c (record_used_regs): New static function.
21225 (try_combine): Handle situations where there is an additional
21226 instruction between I2 and I3 which needs to have a LOG_LINK
21229 2017-03-17 Jeff Law <law@redhat.com>
21231 PR tree-optimization/71437
21232 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
21233 conditional in the hash table first.
21234 (vrp_dom_walker::before_dom_children): Extract condition from
21235 ASSERT_EXPR. Record condition, its inverion and any implied
21236 conditions as well.
21238 2017-03-17 Marek Polacek <polacek@redhat.com>
21239 Markus Trippelsdorf <markus@trippelsdorf.de>
21241 PR tree-optimization/80079
21242 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
21245 2017-03-17 Richard Biener <rguenther@suse.de>
21247 PR middle-end/80075
21248 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
21249 Properly verify the LHS before the RHS possibly claims to be
21251 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
21254 2017-03-17 Martin Jambor <mjambor@suse.cz>
21256 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
21257 (List of -O2 options): Likewise.
21258 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
21261 2017-03-17 Tom de Vries <tom@codesourcery.com>
21263 * gcov-dump.c (print_usage): Print bug_report_url.
21265 2017-03-17 Richard Biener <rguenther@suse.de>
21267 PR middle-end/80050
21268 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
21269 (parser::peek): Likewise.
21271 2017-03-17 Richard Biener <rguenther@suse.de>
21273 PR tree-optimization/80048
21274 * sese.c (free_sese_info): Properly release rename_map and
21275 copied_bb_map elements.
21277 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
21279 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
21280 Add linked-list forward and backlinks. Insert on
21281 construction, remove on destruction.
21282 (class pass_store_merging): Add m_stores_head field.
21283 (pass_store_merging::terminate_and_process_all_chains):
21284 Iterate over m_stores_head list.
21285 (pass_store_merging::terminate_all_aliasing_chains):
21287 (pass_store_merging::execute): Check for debug stmts first.
21288 Push new chains onto the m_stores_head stack.
21290 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
21293 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
21294 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
21295 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
21297 2017-03-16 Jeff Law <law@redhat.com>
21299 PR tree-optimization/71437
21300 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
21301 member function. Implementation moved into after_dom_children
21302 member function and into the threader's thread_outgoing_edges
21304 (dom_opt_dom_walker::after_dom_children): Simplify by moving
21305 some code into new thread_outgoing_edges.
21306 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
21307 definition. Simplify marker handling (do it here). Assume we always
21308 have the available expression and the const/copies tables.
21309 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
21311 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
21312 * tree-vrp.c (equiv_stack): No longer file scoped.
21313 (vrp_dom_walker): New class.
21314 (vrp_dom_walker::before_dom_children): New member function.
21315 (vrp_dom_walker::after_dom_children): Likewise.
21316 (identify_jump_threads): Setup domwalker. Use it rather than
21317 walking edges in a random order by hand. Simplify setup/finalization.
21318 (finalize_jump_threads): Remove.
21319 (vrp_finalize): Do not call identify_jump_threads here.
21320 (execute_vrp): Do it here instead and call thread_through_all_blocks
21323 PR tree-optimization/71437
21324 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
21326 (simplify_stmt_for_jump_threading): Add basic_block argument. All
21328 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
21329 (dom_opt_dom_walker::thread_across_edge): Remove
21330 handle_dominating_asserts argument. All callers changed.
21331 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
21332 changes. Remove calls to lhs_of_dominating_assert. Other
21333 uses of handle_dominating_asserts turn into unconditional code
21334 (simplify_control_stmt_condition_1): Likewise.
21335 (simplify_control_stmt_condition): Likewise.
21336 (thread_through_normal_block, thread_across_edge): Likewise.
21337 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
21338 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
21339 object if it is not an SSA_NAME.
21340 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
21341 before calling into the VRP specific simplifiers.
21342 (identify_jump_threads): Remove handle_dominating_asserts
21345 2017-03-16 Jakub Jelinek <jakub@redhat.com>
21348 * tree-diagnostic.c (default_tree_printer): No longer static.
21349 * tree-diagnostic.h (default_tree_printer): New prototype.
21351 2017-03-16 Tamar Christina <tamar.christina@arm.com>
21353 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
21354 Change ins into fmov.
21356 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21358 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
21359 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
21360 Use h_con constraint for operand 1.
21361 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
21362 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
21364 2017-03-15 Jeff Law <law@redhat.com>
21366 PR tree-optimization/71437
21367 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
21368 (record_temporary_equivalences): Use it.
21370 PR tree-optimization/71437
21371 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
21372 tree-ssa-scopedtables.
21373 (lookup_avail_expr, build_and_record_new_cond): Likewise.
21374 (record_conditions, record_cond, vuse_eq): Likewise.
21375 (record_edge_info): Adjust to API tweak of record_conditions.
21376 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
21377 (record_temporary_equivalences, optimize_stmt): Likewise.
21378 (eliminate_redundant_computations): Likewise.
21379 (record_equivalences_from_stmt): Likewise.
21380 * tree-ssa-scopedtables.c: Include options.h and params.h.
21381 (vuse_eq): New function, moved from tree-ssa-dom.c
21382 (build_and_record_new_cond): Likewise.
21383 (record_conditions): Likewise. Accept vector of conditions rather
21384 than edge_equivalence structure for first argument.
21385 for the first argument.
21386 (avail_exprs_stack::lookup_avail_expr): New member function, moved
21387 from tree-ssa-dom.c.
21388 (avail_exprs_stack::record_cond): Likewise.
21389 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
21390 from tree-ssa-dom.c.
21391 (avail_exprs_stack): Add new member functions lookup_avail_expr
21393 (record_conditions): Declare.
21395 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
21398 * lra-constraints.c (process_alt_operands): Increase reject for
21399 reloading an input/output operand.
21401 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
21404 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
21405 insns to convert from signed/unsigned char/short to IEEE 128-bit
21407 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
21409 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
21412 * config/i386/i386.c (ix86_vector_duplicate_value): Create
21413 subreg of inner mode for values already in registers.
21415 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
21417 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
21418 iteration reg is used after the loop.
21420 2017-03-14 Martin Sebor <msebor@redhat.com>
21422 PR tree-optimization/79800
21423 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
21424 precision in negative-positive range.
21425 (format_floating): Call non-const overload with adjusted precision.
21427 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
21430 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
21433 2017-03-14 Martin Sebor <msebor@redhat.com>
21435 PR middle-end/80020
21436 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
21437 * builtins.def (aligned_alloc): Use it.
21440 * Makefile.in (GTFILES): Add calls.c.
21441 * calls.c: Include "gt-calls.h".
21443 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
21445 PR rtl-optimization/79728
21446 * regs.h (struct target_regs): New field
21447 x_contains_allocatable_regs_of_mode.
21448 (contains_allocatable_regs_of_mode): New macro.
21449 * reginfo.c (init_reg_sets_1): Initialize it, and change
21450 contains_reg_of_mode so it includes global regs as well.
21451 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
21452 rather than contains_regs_of_mode.
21454 2017-03-14 Martin Liska <mliska@suse.cz>
21456 * doc/invoke.texi: Document options that can't be combined with
21457 -fcheck-pointer-bounds.
21459 2017-03-14 Martin Liska <mliska@suse.cz>
21461 PR middle-end/79831
21462 * doc/invoke.texi (-Wchkp): Document the option.
21464 2017-03-14 Martin Liska <mliska@suse.cz>
21466 * Makefile.in: Install gcov-dump.
21468 2017-03-14 Martin Liska <mliska@suse.cz>
21470 * multiple_target.c (expand_target_clones): Bail out for
21471 an invalid attribute.
21473 2017-03-14 Richard Biener <rguenther@suse.de>
21475 * alias.c (struct alias_set_entry): Pack properly.
21476 * cfgloop.h (struct loop): Likewise.
21477 * cse.c (struct set): Likewise.
21478 * ipa-utils.c (struct searchc_env): Likewise.
21479 * loop-invariant.c (struct invariant): Likewise.
21480 * lra-remat.c (struct cand): Likewise.
21481 * recog.c (struct change_t): Likewise.
21482 * rtl.h (struct address_info): Likewise.
21483 * symbol-summary.h (function_summary): Likewise.
21484 * tree-loop-distribution.c (struct partition): Likewise.
21485 * tree-object-size.c (struct object_size_info): Likewise.
21486 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
21487 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
21488 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
21489 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
21490 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
21491 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
21492 (struct _stmt_vec_info): Likewise.
21494 2017-03-14 Martin Liska <mliska@suse.cz>
21497 * multiple_target.c (create_dispatcher_calls): Check that
21498 a target can create a function dispatcher.
21500 2017-03-14 Martin Liska <mliska@suse.cz>
21503 * multiple_target.c (expand_target_clones): Drop local.local
21504 flag for default implementation.
21506 2017-03-14 Richard Biener <rguenther@suse.de>
21508 PR tree-optimization/80030
21509 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
21511 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
21513 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
21514 gcc_fallthrough() instead of __attribute__((fallthrough));
21516 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
21518 * doc/gcc.texi: Remove "up" link to (DIR).
21519 * doc/gccint.texi: Ditto.
21521 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
21523 * doc/install.texi (Specific) <avr>: Remove reference to
21526 2017-03-13 Jeff Law <law@redhat.com>
21528 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
21529 attribute rather than comments.
21531 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
21532 match_scratch operand is highest.
21534 2017-03-13 Martin Liska <mliska@suse.cz>
21536 PR middle-end/78339
21537 * ipa-pure-const.c (warn_function_noreturn): If the declarations
21538 is a CHKP clone, use original declaration.
21540 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
21542 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
21543 (arc_conditional_register_usage): Use a different allocation order
21544 when optimizing for size.
21545 * common/config/arc/arc-common.c (arc_option_optimization_table):
21546 Section anchors default on when optimizing for size.
21548 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
21550 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
21552 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
21554 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
21555 * config/arc/arc.md (cpu_facility): Add cd variant.
21556 (*movqi_insn): Add code density variant.
21557 (*movhi_insn): Likewise.
21558 (*movqi_insn): Likewise.
21559 (*addsi3_mixed): Likewise.
21560 (subsi3_insn): Likewise.
21562 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
21564 * config/arc/arc.md (movsi_cond_exec): Update constraint.
21566 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
21568 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
21569 expressions with MINUS and UNARY ops.
21571 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21574 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
21576 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
21577 between vec_select and vector argument.
21578 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
21579 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
21580 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
21581 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
21582 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
21583 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
21585 2017-03-13 Richard Biener <rguenther@suse.de>
21588 * params.def (vect-max-peeling-for-alignment): Fix typo.
21590 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
21592 * doc/install.texi (Specific) <mips-*-*>: Remove description of
21593 issue that only occurred with binutils below 2.18.
21595 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
21597 * doc/install.texi (Specific) <cris-axis-elf>: No longer
21598 refer to binutils 2.11/2.12 minimum.
21600 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
21602 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
21603 ftp.kernel.org and simplify binutils requirement.
21605 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
21607 * doc/invoke.texi (Warning Options): Fix spelling of link-time
21609 (Optimize Options): Ditto. Also remove redundancy.
21611 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21613 PR translation/79848
21614 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
21616 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
21617 to G_ to avoid double translation.
21619 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21621 PR translation/79923
21622 * auto-profile.c (get_combined_location): Convert leading
21623 character of diagnostics to lower case and remove trailing period.
21624 (read_profile): Likewise for various diagnostics.
21625 * config/arm/arm.c (arm_option_override): Remove trailing period
21626 from various diagnostics.
21627 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
21628 (msp430_expand_delay_cycles): Likewise.
21630 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21633 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
21634 full command-line argument, rather than just "str".
21635 (aarch64_validate_march): Likewise.
21636 (aarch64_validate_mtune): Likewise.
21638 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
21640 PR rtl-optimization/78911
21641 * lra-assigns.c (must_not_spill_p): New function.
21642 (spill_for): Use it.
21644 2017-03-10 Jakub Jelinek <jakub@redhat.com>
21646 PR tree-optimization/79981
21647 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
21648 ATOMIC_COMPARE_EXCHANGE ifn result.
21649 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
21650 IFN_ATOMIC_COMPARE_EXCHANGE.
21652 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21655 * opts.c (parse_sanitizer_options): Add missing question mark to
21656 "did you mean" message.
21658 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21660 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
21662 (VMULEUH_UNS): Likewise.
21663 (VMULOUB_UNS): Likewise.
21664 (VMULOUH_UNS): Likewise.
21665 * config/rs6000/rs6000.c (builtin_function_type): Remove
21666 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
21668 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21671 * read-rtl-function.c (function_reader::read_rtx_operand): Update
21672 x with result of extra_parsing_for_operand_code_0.
21673 (function_reader::extra_parsing_for_operand_code_0): Convert
21674 return type from void to rtx, returning x. When reading
21675 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
21676 larger size containing struct block_symbol.
21678 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
21680 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
21681 -mfloat128-hardware without -m64.
21683 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
21686 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
21687 entries to the case statement that marks unsigned arguments to
21688 overloaded functions.
21690 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
21692 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
21693 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
21695 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
21698 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
21699 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
21701 2017-03-10 Martin Liska <mliska@suse.cz>
21705 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
21706 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
21709 2017-03-10 Olivier Hainque <hainque@adacore.com>
21711 * tree-switch-conversion (array_value_type): Start by resetting
21712 candidate type to it's main variant.
21714 2017-03-10 Jakub Jelinek <jakub@redhat.com>
21716 PR rtl-optimization/79909
21717 * combine.c (try_combine): Use simplify_replace_rtx on individual
21718 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
21719 of the whole CALL_INSN_FUNCTION_USAGE.
21721 PR tree-optimization/79972
21722 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
21723 get_range_info on SSA_NAMEs. Formatting fixes.
21725 2017-03-10 Richard Biener <rguenther@suse.de>
21726 Jakub Jelinek <jakub@redhat.com>
21728 PR tree-optimization/77975
21729 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
21730 edge to be constant.
21731 (get_val_for): For constant x return it. Formatting fix.
21732 (loop_niter_by_eval): Avoid pointless looping if the next iteration
21733 would use the same bases as the current one.
21735 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21737 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
21738 instead of vec_select for V1TImode.
21739 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
21741 (VSX_LE_128): Add V1TI to this mode iterator.
21742 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
21743 (*vsx_le_perm_store_<mode>): Likewise.
21744 (pre-reload splitter for VSX stores): Likewise.
21745 (post-reload splitter for VSX stores): Likewise.
21746 (*vsx_xxpermdi2_le_<mode>): Likewise.
21747 (*vsx_lxvd2x2_le_<mode>): Likewise.
21748 (*vsx_stxvd2x2_le_<mode>): Likewise.
21750 2017-03-09 Michael Eager <eager@eagercon.com>
21752 Correct failures with --enable-checking=yes,rtl.
21754 * config/microblaze/microblaze.c (microblaze_expand_shift):
21755 Replace GET_CODE test with CONST_INT_P and INTVAL test with
21756 test for const0_rtx.
21757 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
21758 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
21760 2017-03-09 Richard Biener <rguenther@suse.de>
21762 PR tree-optimization/79977
21763 * graphite-scop-detection.c (scop_detection::merge_sese):
21764 Handle the case of extra exits to blocks dominating the entry.
21766 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
21768 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
21771 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
21773 PR rtl-optimization/79949
21774 * lra-constraints.c (process_alt_operands): Check memory when
21775 trying to predict a cycle. Print about the overall increase.
21777 2017-03-09 Richard Biener <rguenther@suse.de>
21779 PR middle-end/79971
21780 * gimple-expr.c (useless_type_conversion_p): Preserve
21781 TYPE_SATURATING for fixed-point types.
21783 2017-03-09 Richard Biener <rguenther@suse.de>
21786 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
21787 alignment of BLKmode params.
21789 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21792 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
21793 (VALL_NO_V2Q): Likewise.
21795 * config/aarch64/aarch64-simd.md
21796 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
21798 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
21799 VALL_NO_V2Q mode iterator.
21800 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
21802 2017-03-09 Martin Liska <mliska@suse.cz>
21804 PR tree-optimization/79631
21805 * tree-chkp-opt.c (chkp_is_constant_addr): Call
21806 tree_int_cst_sign_bit just for INTEGER constants.
21808 2017-03-09 Martin Liska <mliska@suse.cz>
21812 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
21815 2017-03-09 Marek Polacek <polacek@redhat.com>
21818 * tree.c (inchash::add_expr): Handle TREE_VEC.
21820 2017-03-09 Martin Liska <mliska@suse.cz>
21823 (chkp_narrow_size_and_offset): New function.
21824 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
21825 (void chkp_parse_bit_field_ref): New function.
21826 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
21827 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
21829 2017-03-09 Martin Liska <mliska@suse.cz>
21832 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
21833 (chkp_find_bounds_1): Remove gcc_unreachable.
21835 2017-03-09 Jakub Jelinek <jakub@redhat.com>
21838 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
21839 BUILT_IN_SYNC*, determine the access type from the size suffix and
21840 always build a MEM_REF with that type. Handle forgotten
21841 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
21844 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
21845 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
21846 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
21847 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
21848 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
21849 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
21850 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
21851 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
21852 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
21853 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
21854 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
21855 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
21856 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
21857 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
21858 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
21859 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
21860 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
21861 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
21862 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
21863 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
21864 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
21865 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
21866 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
21867 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
21868 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
21869 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
21870 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
21871 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
21872 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
21873 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
21874 definitions outside of __OPTIMIZE__ guarded section.
21877 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
21878 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
21879 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
21880 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
21883 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21885 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
21886 ("vfenez<mode>"): Add missing constraints.
21888 2017-03-08 Martin Sebor <msebor@redhat.com>
21891 * config/nds32/nds32.c (nds32_option_override):
21892 Fix misspelled diagnostic.
21894 2017-03-08 Jakub Jelinek <jakub@redhat.com>
21897 * gimplify.c (gimplify_omp_for): Replace index var in outer
21898 taskloop statement with an artificial variable and add
21899 OMP_CLAUSE_PRIVATE clause for it.
21901 2017-03-08 Richard Biener <rguenther@suse.de>
21903 PR tree-optimization/79955
21904 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
21905 for accesses that are completely outside of the variable.
21907 2017-03-08 Andrew Haley <aph@redhat.com>
21909 PR tree-optimization/79943
21910 * tree-ssa-loop-split.c (compute_new_first_bound): When
21911 calculating the new upper bound, (END-BEG) should be added, not
21914 2017-03-08 Jakub Jelinek <jakub@redhat.com>
21916 * config/avr/avr.md (setmemhi): Make sure match_dup
21917 operand number comes before match_scratch.
21919 2017-03-08 Richard Biener <rguenther@suse.de>
21921 PR tree-optimization/79920
21922 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
21923 with ncopies == 1 to ...
21924 (vect_transform_slp_perm_load): ... here. Properly compute
21925 all element loads by iterating VF times over the group. Do
21926 not handle ncopies (computed in a broken way) in
21927 vect_create_mask_and_perm.
21929 2017-03-08 Jakub Jelinek <jakub@redhat.com>
21932 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
21933 is a uniform vector, use uniform_vector_p return value instead of
21934 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
21936 2017-03-07 Marek Polacek <polacek@redhat.com>
21938 PR middle-end/79809
21939 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
21940 (alloca_call_type): Likewise.
21942 2017-03-07 Martin Liska <mliska@suse.cz>
21944 * gcov.c (process_args): Put comment to correct location.
21946 2017-03-07 Martin Liska <mliska@suse.cz>
21948 PR middle-end/68270
21949 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
21950 Use array_at_struct_end_p instead of DECL_CHAIN (field).
21951 (chkp_narrow_bounds_for_field): Likewise.
21952 (chkp_parse_array_and_component_ref): Pass one more argument to
21955 2017-03-07 Richard Biener <rguenther@suse.de>
21957 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
21960 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
21962 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
21963 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
21965 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21968 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
21969 to end of description.
21970 (PARAM_MAX_STORES_TO_MERGE): Likewise.
21972 2017-03-07 Jakub Jelinek <jakub@redhat.com>
21974 PR rtl-optimization/79901
21975 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
21977 (*avx512f_<code><mode>3<mask_name>): ... this.
21978 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
21979 iterator instead of VI8_AVX2_AVX512BW.
21981 PR rtl-optimization/79901
21982 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
21983 min/max expander, expand it using expand_vec_cond_expr.
21986 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
21989 2017-03-06 Jakub Jelinek <jakub@redhat.com>
21992 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
21993 to void * for PCH reasons.
21994 * dwarf2out.c (output_loc_operands, output_die): Cast
21995 v.val_vec.array to unsigned char *.
21997 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
22000 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
22003 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
22005 PR rtl-optimization/79571
22006 * lra-constraints.c (process_alt_operands): Calculate static
22007 reject and subtract it from overall when only addresses will be
22010 2017-03-06 Julia Koval <julia.koval@intel.com>
22013 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
22014 incoming stack boundary to 128 for 64-bit targets.
22016 2017-03-06 Richard Biener <rguenther@suse.de>
22018 PR tree-optimization/79894
22019 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
22020 to NULL after folding it.
22022 2017-03-06 Richard Biener <rguenther@suse.de>
22024 PR tree-optimization/79824
22025 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
22026 check disabling peeling for gaps.
22028 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
22030 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
22031 attributes): Document gettimeofday.
22033 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
22035 * config/s390/s390.c (s390_option_override_internal): Set
22036 PARAM_MIN_VECT_LOOP_BOUND
22038 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
22040 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
22041 * config/s390/s390.md: Likewise.
22043 2017-03-06 Jakub Jelinek <jakub@redhat.com>
22046 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
22047 (<avx2_avx512>_perm<mode>): Rename to ...
22048 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
22050 (<avx512>_perm<mode>_mask): Rename to ...
22051 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
22053 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
22054 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
22055 instead of VI8F_256_512.
22056 (avx512f_perm<mode>): New define_expand.
22057 (avx512f_perm<mode>_mask): Likewise.
22058 (avx512f_perm<mode>_1<mask_name>): New define_insn.
22059 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
22061 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
22063 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
22064 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
22066 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
22068 2017-03-06 Martin Liska <mliska@suse.cz>
22071 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
22072 when having a SSA NAME w/o VAR_DECL assigned to it.
22074 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
22076 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
22077 msa_dpsub_<su>_d): Fix MODE for vec_select.
22079 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
22081 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
22083 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
22085 2017-03-06 Richard Biener <rguenther@suse.de>
22087 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
22088 * plugin.c (register_plugin_info): Likewise.
22089 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
22091 2017-03-05 Jakub Jelinek <jakub@redhat.com>
22093 * config/i386/sse.md (sse_storehps, sse_storelps,
22094 avx_<castmode><avxsizesuffix>_<castmode>,
22095 avx512f_<castmode><avxsizesuffix>_<castmode>,
22096 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
22097 in condition that at least one operand is not a MEM.
22099 2017-03-03 Jakub Jelinek <jakub@redhat.com>
22101 PR middle-end/79805
22102 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
22103 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
22105 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
22106 gimple_call_nothrow_p flag based on whether original builtin can throw.
22107 If it can, emit following stmts on the fallthrough edge.
22108 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
22109 don't create new bb if inserting just debug stmts on the edge, try to
22110 insert them on the fallthru bb or just reset debug stmts.
22112 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
22115 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
22116 restore recog_data (including the operand rtxes inside it) around
22117 the call to get_insn_template.
22119 2017-03-03 Martin Sebor <msebor@redhat.com>
22121 PR tree-optimization/79699
22122 * context.c (context::~context): Free MPFR caches to avoid
22123 a memory leak on program exit.
22125 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22127 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
22128 Use wide_int::ulow () instead of .elt (0).
22130 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
22132 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
22133 (*pushxf): Limit oF constraint to 32bit targets and add oC
22134 constraint for 64bit targets.
22135 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
22136 (*pushdf): Change rmF constraint to rmC.
22138 2017-03-03 Martin Liska <mliska@suse.cz>
22140 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
22141 Remove unused variable.
22143 2017-03-03 Jakub Jelinek <jakub@redhat.com>
22146 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
22147 is a memory operand, increase num_memory.
22148 (ix86_expand_args_builtin): Likewise.
22150 2017-03-03 Jan Hubicka <jh@suse.cz>
22153 * ipa-devirt.c (maybe_record_node): Properly handle
22154 __cxa_pure_virtual visibility.
22156 2017-03-03 Martin Liska <mliska@suse.cz>
22158 PR tree-optimization/79803
22159 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
22161 (pass_loop_prefetch::execute): Disabled optimization if an
22162 assumption about L1 cache size is not met.
22164 2017-03-03 Martin Liska <mliska@suse.cz>
22166 PR rtl-optimization/79574
22167 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
22168 (hash_scan_set): Likewise.
22169 (dump_hash_table): Likewise.
22170 (hoist_code): Likewise.
22172 2017-03-03 Richard Biener <rguenther@suse.de>
22174 * fixed-value.c (fixed_from_string): Restore use of elt (1)
22175 in place of uhigh ().
22176 (fixed_convert_from_real): Likewise.
22178 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
22181 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
22183 2017-03-03 Richard Biener <rguenther@suse.de>
22185 PR middle-end/79818
22186 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
22187 TYPE_OVERFLOW_UNDEFINED check.
22189 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22191 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
22193 (vector_ae_<mode>_p): Likewise.
22194 (vector_nez_<mode>_p): Likewise.
22195 (vector_ne_v2di_p): Likewise.
22196 (vector_ae_v2di_p): Likewise.
22197 (vector_ne_<mode>_p): Likewise.
22198 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
22200 (vsx_tsqrt<mode>2_fe): Likewise.
22202 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
22205 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
22207 2017-03-02 Jakub Jelinek <jakub@redhat.com>
22209 PR rtl-optimization/79780
22210 * cprop.c (one_cprop_pass): When second and further conditional trap
22211 in a single basic block is turned into an unconditional trap, turn it
22212 into a deleted note to avoid RTL verification failures.
22214 2017-03-02 Richard Biener <rguenther@suse.de>
22216 * fold-const.c (const_binop): Use ulow () instead of elt (0).
22218 2017-03-02 Richard Biener <rguenther@suse.de>
22220 PR tree-optimization/79345
22222 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
22223 param and abort the walk, returning -1 if it is hit.
22224 (walk_aliased_vdefs): Take a limit param and pass it on.
22225 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
22226 defaulting to 0 and return a signed int.
22227 * tree-ssa-uninit.c (struct check_defs_data): New struct.
22228 (check_defs): New helper.
22229 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
22230 about uninitialized memory.
22231 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
22232 bogus uninitialized warning.
22233 (fixed_convert_from_real): Likewise.
22235 2017-03-02 Bin Cheng <bin.cheng@arm.com>
22237 PR tree-optimization/66768
22238 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
22239 iv_use if base object can't be determined.
22241 2017-03-02 Jakub Jelinek <jakub@redhat.com>
22243 PR tree-optimization/79345
22244 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
22245 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
22246 (get_pattern_stats): Initialize it.
22247 * genemit.c (gen_expand): Verify match_scratch numbers come after
22248 match_operand/match_dup numbers.
22249 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
22250 match_scratch numbers.
22251 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
22253 * config/s390/s390.md (trunctdsd2): Likewise.
22255 2017-03-02 Richard Biener <rguenther@suse.de>
22257 * wide-int.h (wide_int_storage::operator=): Implement in terms
22260 2017-03-02 Richard Biener <rguenther@suse.de>
22262 PR tree-optimization/79777
22263 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
22264 the to insert expression to sth existing.
22266 2017-03-01 Martin Sebor <msebor@redhat.com>
22268 PR middle-end/79692
22269 * gimple-ssa-sprintf.c
22270 (directive::known_width_and_precision): New function.
22271 (format_integer): Use it.
22272 (get_mpfr_format_length): Consider the full range of precision
22273 when computing %g output with the # flag. Set the likely byte
22274 count to 3 rather than 1 when precision is indeterminate.
22275 (format_floating): Correct the lower bound of precision.
22277 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22279 * doc/invoke.texi: Document default code model for 64-bit Linux.
22281 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22284 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
22285 udiv rather than div since input pattern is unsigned.
22287 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
22289 * config/i386/i386.c (print_reg): Warn for values of
22290 unsupported size in integer register.
22292 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
22295 * config/rs6000/predicates.md (current_file_function_operand): Do
22296 not allow self calls to be local if the function is replaceable.
22298 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
22301 * config/rs6000/altivec.h (vec_ctz and others): Change the
22302 preprocessor macro that controls conditional compilation from
22303 _ARCH_PWR9 to __POWER9_VECTOR__.
22304 (vec_all_ne): Change parameterization of __altivec_scalar_pred
22305 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
22306 control (instead of _ARCH_PWR9 control) so that template
22307 definition uses power9-specific function.
22308 (vec_any_eq): Likewise.
22309 (vec_all_ne): Change macro definition to use a power9-specific
22310 expansion under #ifdef __POWER9_VECTOR__ control (instead of
22311 _ARCH_PWR9 control).
22312 (vec_any_eq) Likewise.
22313 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
22314 expansion for CMPNEF to remove support for xvcmpnesp instruction.
22315 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
22316 support for xvcmpnedp instruction.
22317 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
22318 macro expansion so that Power9 implementation of vec_all_ne does
22319 not use the AltiVec predicate framework.
22320 (VCMPNEH_P): Likewise.
22321 (VCMPNEW_P): Likewise.
22322 (VCMPNED_P): Likewise.
22323 (VCMPNEFP_P): Likewise.
22324 (VCMPNEDP_P): Likewise.
22325 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
22326 implementation of vec_any_eq to not use AltiVec predicate
22328 (VCMPAEH_P): Likewise.
22329 (VCMPAEW_P): Likewise.
22330 (VCMPAED_P): Likewise.
22331 (VCMPAEFP_P): Likewise.
22332 (VCMPAEDP_P): Likewise.
22333 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
22334 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
22335 not use the AltiVec predicate framework.
22336 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
22337 of vec_any_eq to not use AltiVec predicate framework.
22338 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
22339 support for predefined __POWER9_VECTOR__ macro to indicate that
22340 Power9 instruction selection is enabled.
22341 (altivec_overloaded_builtins): Remove extraneous
22342 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
22343 function argument types RS6000_BTI_bool_V16QI and
22344 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
22345 entry for overloaded function argument types RS6000_BTI_bool_V4SI
22346 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
22347 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
22348 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
22349 Power9 for implementations of vec_cmpne. Change the signature for
22350 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
22351 (representing vec_all_ne) to remove the previously described first
22352 argument of type RS6000_BTI_INTSI, as this was an artifact of
22353 reliance on the AltiVec predicate framework, which is no longer
22354 used in the implementation of these functions. Add
22355 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
22356 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
22357 since, unlike the AltiVec predicate framework implementation, we
22358 do not share function descriptors between vec_alle and vec_anyeq.
22359 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
22360 set of modes that receive special treatment even when
22361 TARGET_P9_VECTOR is true. The special treatment emits code that
22362 does not depend on Power9 instructions.
22363 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
22364 define_expand to not rely on AltiVec predicate framework.
22365 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
22367 (vector_ne_v2di_p): Change this define_expand to not rely on
22368 AltiVec predicate framework.
22369 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
22371 (vector_ne_<mode>_p): Change this define_expand to not rely on
22372 AltiVec predicate framework.
22373 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
22375 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
22376 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
22377 define_insn pattern.
22378 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
22379 define_insn pattern because the xvcmpne<VSs>. instruction is not
22381 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
22382 instruction is not supported.
22384 2017-03-01 Jakub Jelinek <jakub@redhat.com>
22386 * config/nvptx/nvptx.c: Include intl.h.
22388 2017-03-01 Martin Jambor <mjambor@suse.cz>
22391 * ipa-prop.h (ipa_bits): Removed field known.
22392 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
22393 to pointers. Adjusted their comments to warn about their sharing.
22394 (ipcp_transformation_summary): Change bits to a vector of pointers.
22395 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
22396 (ipa_get_ipa_bits_for_value): Declare.
22397 * tree-vrp.h (value_range): Mark as GTY((for_user)).
22398 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
22399 (ipa_bits_hash_table): Likewise.
22400 (ipa_vr_ggc_hash_traits): Likewise.
22401 (ipa_vr_hash_table): Likewise.
22402 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
22403 being pointers and vr_known being removed.
22404 (ipa_set_jf_unknown): Likewise.
22405 (ipa_get_ipa_bits_for_value): New function.
22406 (ipa_set_jfunc_bits): Likewise.
22407 (ipa_get_value_range): New overloaded functions.
22408 (ipa_set_jfunc_vr): Likewise.
22409 (ipa_compute_jump_functions_for_edge): Use the above functions to
22410 construct bits and vr parts of jump functions.
22411 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
22412 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
22414 (ipcp_grow_transformations_if_necessary): Also allocate
22415 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
22417 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
22418 them. Fix too long lines.
22419 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
22420 vr_known being removed.
22421 (ipa_read_jump_function): Use new setter functions to construct bits
22422 and vr parts of jump functions or set them to NULL.
22423 (write_ipcp_transformation_info): Adjust for bits being pointers.
22424 (read_ipcp_transformation_info): Likewise.
22425 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
22427 Include gt-ipa-prop.h.
22428 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
22430 (ipcp_store_bits_results): Likewise.
22431 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
22432 Do not write to existing jump functions but use a temporary instead.
22434 2017-03-01 Jakub Jelinek <jakub@redhat.com>
22437 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
22438 attempt to use its first operand as BIT_FIELD_REF base.
22440 2017-03-01 Richard Biener <rguenther@suse.de>
22442 PR middle-end/79721
22443 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
22444 interpolating formula in wrapping arithmetic.
22445 (chrec_apply): Convert chrec_evaluate return value to wanted type.
22447 2017-03-01 Jakub Jelinek <jakub@redhat.com>
22449 PR tree-optimization/79734
22450 * tree-vect-generic.c (expand_vector_condition): Optimize
22451 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
22452 Handle VEC_COND_EXPR where comparison has different inner width from
22453 type's inner width.
22455 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
22457 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
22458 markup, and similar issues. Remove @opindex entries for things
22459 that aren't options. Add missing -mmpy-option entries.
22461 2017-02-28 Jakub Jelinek <jakub@redhat.com>
22463 PR tree-optimization/79737
22464 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
22465 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
22466 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
22467 instead of byte_size. Formatting fix.
22468 (shift_bytes_in_array_right): Formatting fix.
22470 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
22473 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
22474 condition on optimize for the leaf function test.
22476 2017-02-28 Martin Liska <mliska@suse.cz>
22479 * read-rtl-function.c (function_reader::handle_unknown_directive):
22480 Bail out when one uses -flto.
22482 2017-02-28 Martin Liska <mliska@suse.cz>
22484 * common.opt: Replace space with tabular for options of <number>
22486 * config/i386/i386.opt: Show <number> value for
22487 -mlarge-data-threshold.
22488 * opts.c (print_filtered_help): Do not display number in hexadecimal
22491 2017-02-28 Martin Liska <mliska@suse.cz>
22493 * common.opt: Fix --help=option -Q for options which are of
22496 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
22498 * config/i386/i386.c (print_reg): Error out for values
22499 of 8-bit size in invalid integer register.
22501 2017-02-28 Martin Sebor <msebor@redhat.com>
22503 PR tree-optimization/79691
22504 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
22506 2017-02-28 Jakub Jelinek <jakub@redhat.com>
22509 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
22510 gcc_unreachable with output_operand_lossage.
22512 2017-02-28 Richard Biener <rguenther@suse.de>
22514 PR tree-optimization/79740
22515 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
22517 (visit_nary_op): Insert the nary into the hashtable if we
22518 pattern-matched sth.
22519 * tree-ssa-pre.c (eliminate_insert): Robustify.
22521 2017-02-28 Richard Biener <rguenther@suse.de>
22523 PR middle-end/79731
22524 * fold-const.c (decode_field_reference): Reject out-of-bound
22527 2017-02-28 Jakub Jelinek <jakub@redhat.com>
22529 * config/i386/i386.c: Include intl.h.
22530 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
22531 instead of just cond ? "..." : "...".
22532 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
22533 * coverage.c (read_counts_file): Likewise.
22534 * omp-offload.c: Include intl.h.
22535 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
22536 of just cond ? "..." : "...".
22537 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
22538 of just cond ? "..." : "...".
22540 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
22543 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
22545 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
22546 'tune for' CPU name.
22547 * config/arm/arm-cpu-data.h: Regenerated.
22549 2017-02-28 Richard Biener <rguenther@suse.de>
22551 PR tree-optimization/79732
22552 * tree-inline.c (expand_call_inline): Do not shadow var.
22554 2017-02-28 Richard Biener <rguenther@suse.de>
22556 PR tree-optimization/79723
22557 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
22558 address-space properly.
22560 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
22562 * doc/optinfo.texi (Optimization groups): Fix option used for
22564 * doc/invoke.texi (-fopt-info): Document "omp".
22565 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
22566 (OPTGROUP_ALL): Add OPTGROUP_OMP.
22567 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
22568 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
22569 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
22571 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
22573 * dumpfile.c (optgroup_options): Instead of "openmp", associate
22574 OPTGROUP_OMP with "omp".
22576 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
22579 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
22580 for arithmetic shift of unsigned V2DI.
22582 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
22584 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
22585 arc/linux.h headers.
22586 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
22587 (LINK_SPEC): Likewise.
22588 (ARC_TLS_EXTRA_START_SPEC): Likewise.
22589 (EXTRA_SPECS): Likewise.
22590 (STARTFILE_SPEC): Likewise.
22591 (ENDFILE_SPEC): Likewise.
22592 (LIB_SPEC): Likewise.
22593 (TARGET_SDATA_DEFAULT): Likewise.
22594 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
22595 (MULTILIB_DEFAULTS): Likewise.
22596 (DWARF2_UNWIND_INFO): Likewise.
22597 * config/arc/big.h: New file.
22598 * config/arc/elf.h: Likewise.
22599 * config/arc/linux.h: Likewise.
22600 * config/arc/t-uClibc: Remove.
22602 2017-02-27 Bin Cheng <bin.cheng@arm.com>
22604 PR tree-optimization/77536
22605 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
22606 (tree_transform_and_unroll_loop): Use above function to compute the
22607 estimated niter of unrolled loop and use it when scaling profile.
22608 Also use count info rather than frequency if it's non-zero.
22609 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
22610 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
22611 (vect_transform_loop): Call above function.
22613 2017-02-27 Richard Biener <rguenther@suse.de>
22615 PR tree-optimization/45397
22616 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
22617 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
22618 (visit_nary_op): Add pattern matching for CSEing sign-changed
22619 or truncated operations with wider ones.
22621 2017-02-27 Richard Biener <rguenther@suse.de>
22623 PR tree-optimization/79690
22624 * tree-vect-stmts.c (vectorizable_store): Use vector type
22625 built from the DR with address-space.
22627 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
22629 * doc/invoke.texi (Optimize Options): Refine the description
22630 of asan-use-after-return.
22632 2017-02-25 Alan Modra <amodra@gmail.com>
22634 PR rtl-optimization/79584
22635 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
22636 base, not ad->base_term, the reg within base. Remove assertion
22637 that ad->base == ad->base_term. Replace gen_int_mode using
22638 bogus mode with const0_rtx.
22640 2017-02-25 Jakub Jelinek <jakub@redhat.com>
22642 PR middle-end/79396
22643 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
22644 FMA_EXPR like tcc_binary or tcc_unary.
22646 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
22649 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
22651 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
22652 (output_loc_operands): Handle DW_OP_call_ref and
22653 DW_OP_GNU_variable_value.
22654 (struct variable_value_struct): New type.
22655 (struct variable_value_hasher): Likewise.
22656 (variable_value_hash): New variable.
22657 (string_types): Remove.
22658 (copy_loc_descr): New function.
22659 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
22660 (prepend_loc_descr_to_each): New function.
22661 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
22662 instead of add_loc_descr_to_each if the first argument is single
22663 location list and the second has multiple.
22664 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
22665 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
22666 when looking for variable value which doesn't have other location info.
22667 (loc_list_from_tree): Formatting fix.
22668 (gen_array_type_die): Simplify DW_AT_string_length handling.
22669 (adjust_string_types): Remove.
22670 (gen_subprogram_die): Don't call adjust_string_types nor test/set
22671 string_types. Call resolve_variable_values.
22672 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
22673 (resolve_addr_in_expr): Likewise. Add A argument.
22674 (copy_deref_exprloc): Remove deref argument. Adjust for the
22675 original expression being DW_OP_GNU_variable_value with optionally
22676 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
22677 optionally after it.
22678 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
22679 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
22680 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
22681 (variable_value_hasher::hash, variable_value_hasher::equal): New
22683 (resolve_variable_value_in_expr, resolve_variable_value,
22684 resolve_variable_values, note_variable_value_in_expr,
22685 note_variable_value): New functions.
22686 (dwarf2out_early_finish): Call note_variable_value on all toplevel
22689 2017-02-24 Jakub Jelinek <jakub@redhat.com>
22692 * opts.h (handle_generated_option): Add GENERATED_P argument.
22693 * opts-common.c (handle_option): Adjust function comment.
22694 (handle_generated_option): Add GENERATED_P argument, pass it to
22696 (control_warning_option): Pass false to handle_generated_option
22698 * opts.c (maybe_default_option): Pass true to handle_generated_option
22700 * optc-gen.awk: Likewise.
22702 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22704 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
22705 a REG, look at the REG it is a SUBREG of.
22706 (splitter for cmpeqsi_t): Ditto.
22708 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22710 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
22711 the special USEs with the pattern of the insn, not the insn itself.
22713 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
22716 * doc/invoke.texi: Document -mload-store-pairs.
22718 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22719 Sandra Loosemore <sandra@codesourcery.com>
22721 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
22722 argument isn't a CONST_INT.
22723 (nios2_alternate_compare_const): Assert op is a CONST_INT.
22724 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
22725 (nios2_validate_compare): Bypass alternate compare logic if *op2
22726 is not a CONST_INT.
22727 (ldstwm_operation_p): Return false if first_base is not a REG or
22728 if first_offset is not a CONST_INT.
22730 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22732 * config/cris/cris.md: Use correct operand in a define_peephole2.
22734 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22736 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
22738 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22740 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
22741 this_insn if it is an INSN or JUMP_INSN.
22742 (force_offsettable): Look at base, not at addr.
22743 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
22744 on things that aren't necessarily CONST_INTs.
22746 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
22748 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
22749 -mfpmath=sse is the default also for x86-32 targets with SSE2
22750 instruction set when @option{-ffast-math} is enabled
22752 2017-02-24 Jeff Law <law@redhat.com>
22754 PR rtl-optimizatoin/79286
22755 * ira.c (update_equiv_regs): Drop may_trap_p exception to
22758 2017-02-24 Richard Biener <rguenther@suse.de>
22760 PR tree-optimization/79389
22761 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
22764 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
22766 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
22767 function comment to reflect reality.
22768 (loop_exits_before_overflow): Fix typo in function description.
22770 2017-02-24 Richard Biener <rguenther@suse.de>
22772 PR tree-optimization/79389
22773 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
22774 properly that a threading opportunity exists. Detect conditional
22775 copy/constant propagation opportunities.
22777 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
22779 * config/visium/visium.md (type): Add trap.
22780 (b): New mode attribute.
22781 (*btst): Rename into...
22782 (*btst<mode>): ...this and adjust.
22783 (*cbranchsi4_btst_insn): Rename into...
22784 (*cbranch<mode>4_btst_insn): ...this and adjust.
22785 (trap): New define_insn.
22787 2017-02-23 Jakub Jelinek <jakub@redhat.com>
22789 PR tree-optimization/79389
22790 * ifcvt.c (struct noce_if_info): Add rev_cond field.
22791 (noce_reversed_cond_code): New function.
22792 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
22793 reversed_comparison_code. Formatting fix.
22794 (noce_try_store_flag): Test rev_cond != NULL in addition to
22795 reversed_comparison_code.
22796 (noce_try_store_flag_constants): Likewise.
22797 (noce_try_store_flag_mask): Likewise.
22798 (noce_try_addcc): Use rev_cond if non-NULL instead of
22799 reversed_comparison_code.
22800 (noce_try_cmove_arith): Likewise. Formatting fixes.
22801 (noce_try_minmax, noce_try_abs): Clear rev_cond.
22802 (noce_find_if_block): Initialize rev_cond.
22803 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
22804 instead of false as last argument never attempt to reverse it
22807 2017-02-23 Bin Cheng <bin.cheng@arm.com>
22809 PR tree-optimization/79663
22810 * tree-predcom.c (combine_chains): Process refs in reverse order
22811 only for ZERO length chains, and add explaining comment.
22813 2017-02-23 Jeff Law <law@redhat.com>
22815 PR tree-optimization/79578
22816 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
22817 in call to operand_equal_p.
22819 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
22822 * config/i386/cpuid.h: Fix another undefined behavior.
22824 2017-02-23 Richard Biener <rguenther@suse.de>
22826 PR tree-optimization/79683
22827 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
22828 vector types for data-refs.
22830 2017-02-23 Martin Liska <mliska@suse.cz>
22832 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
22834 2017-02-23 Jakub Jelinek <jakub@redhat.com>
22836 PR middle-end/79665
22837 * internal-fn.c (get_range_pos_neg): Moved to ...
22838 * tree.c (get_range_pos_neg): ... here. No longer static.
22839 * tree.h (get_range_pos_neg): New prototype.
22840 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
22841 are known to be in between 0 and signed maximum inclusive, try to
22842 expand both unsigned and signed divmod and use the cheaper one from
22845 2017-02-22 Jeff Law <law@redhat.com>
22847 PR tree-optimization/79578
22848 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
22849 to compare base operands.
22851 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
22854 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
22855 gpc_reg_operand instead of fpr_reg_operand.
22857 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
22859 * config/mips/mips.c (mips_return_in_memory): Force FP
22860 vector types to be returned in memory for o32 ABI.
22862 2017-02-22 Jakub Jelinek <jakub@redhat.com>
22864 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
22865 instead of DW_TAG_member for static data member declarations and don't
22866 set no_linkage_name for static inline data members.
22867 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
22870 2017-02-22 Martin Liska <mliska@suse.cz>
22872 * doc/invoke.texi: Replace inequality signs with square brackets
22875 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22877 PR tree-optimization/68644
22878 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
22880 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
22883 * lra-constraints.c (simplify_operand_subreg): Handle
22884 WORD_REGISTER_OPERATIONS targets.
22886 2017-02-22 Jakub Jelinek <jakub@redhat.com>
22889 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
22890 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
22891 elimination by swapping fld*.
22893 2017-02-22 Richard Biener <rguenther@suse.de>
22895 PR tree-optimization/79673
22896 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
22897 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
22898 irrelevant address-space qualifiers and avoiding a
22899 ADDR_SPACE_CONVERT_EXPR from fold_convert.
22901 2017-02-22 Richard Biener <rguenther@suse.de>
22903 PR tree-optimization/79666
22904 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
22905 to not symbolically negate if that may introduce undefined
22908 2017-02-22 Martin Liska <mliska@suse.cz>
22911 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
22912 * data-streamer-out.c (streamer_write_gcov_count_stream):
22914 * value-prof.c (stream_out_histogram_value): Make assert more
22915 precise based on type of counter.
22917 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
22920 * config/i386/i386.md (standard_x87sse_constant_load splitter):
22921 Use nonimmediate_operand instead of memory_operand for operand 1.
22922 (float-extend standard_x87sse_constant_load splitter): Ditto.
22924 2017-02-21 Jeff Law <law@redhat.com>
22926 PR tree-optimization/79621
22927 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
22928 blocks with edges to themselves.
22930 2017-02-21 Jakub Jelinek <jakub@redhat.com>
22933 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
22934 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
22935 Use gimple_call_builtin_p.
22938 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
22939 on temporarily removed DEBUG_INSNs.
22941 PR tree-optimization/79649
22942 * tree-loop-distribution.c (classify_partition): Give up on
22943 non-generic address space loads/stores.
22945 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
22947 * doc/loop.texi (Loop manipulation): Remove nonexistent
22948 tree_ssa_loop_version from the documentation.
22949 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
22951 2017-02-21 Jakub Jelinek <jakub@redhat.com>
22954 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
22955 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
22956 * config/rs6000/rs6000.c: Include except.h.
22957 (rs6000_expand_split_stack_prologue): Call
22958 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
22960 2017-02-21 Martin Jambor <mjambor@suse.cz>
22963 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
22964 have been analyzed.
22966 2017-02-21 Martin Jambor <mjambor@suse.cz>
22968 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
22969 for backward compatibility only.
22970 * doc/invoke.texi (Option Summary): Remove all references to
22971 -fipa-cp-alignment.
22973 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
22977 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
22979 * lra-constraints.c (curr_insn_transform): Handle
22980 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
22982 2017-02-21 Martin Liska <mliska@suse.cz>
22984 * config/i386/i386.opt: Replace -masm-dialect with -masm.
22986 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
22988 PR translation/79638
22989 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
22991 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
22994 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
22995 (arm_function_ok_for_sibcall): Return false for an indirect call by
22996 descriptor if all the argument registers are used.
22997 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
22998 alignment of the function.
23000 2017-02-21 Jakub Jelinek <jakub@redhat.com>
23002 PR tree-optimization/61441
23003 * simplify-rtx.c (simplify_const_unary_operation): For
23004 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
23005 the sNaN unmodified.
23007 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
23009 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
23010 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
23011 instead of SYSTEM_HEADER_DIR.
23013 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
23014 Martin Liška <mliska@suse.cz>
23016 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
23017 Fix typos and grammar, use active voice, and clarify.
23019 2017-02-20 Marek Polacek <polacek@redhat.com>
23021 PR middle-end/79537
23022 * gimplify.c (gimplify_expr): Handle unused *&&L;.
23025 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
23027 2017-02-20 Jakub Jelinek <jakub@redhat.com>
23030 * config/i386/i386.c (ix86_expand_builtin): Handle
23031 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
23032 ix86_builtins_isa[fcode].isa as a requirement of those
23033 flags and any other flag in the bitmask.
23034 (ix86_init_mmx_sse_builtins): Use 0 instead of
23035 ~OPTION_MASK_ISA_64BIT as mask.
23036 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
23037 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
23038 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
23039 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
23041 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
23044 * lra-constraints.c (split_reg): Check requested split mode
23045 is supported by the register.
23047 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
23049 * lra-constraints.c (simplify_operand_subreg): Remove early
23052 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
23055 * lra-constraints.c (curr_insn_transform): Tighten condition
23056 for converting SUBREG reloads from OP_OUT to OP_INOUT.
23058 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
23061 * lra-constraints.c (curr_insn_transform): Handle
23062 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
23064 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
23067 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
23069 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
23071 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
23074 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
23077 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23079 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
23082 2017-02-18 Jakub Jelinek <jakub@redhat.com>
23084 * final.c (last_columnnum, override_columnnum): New variables.
23085 (final_start_function): Set last_columnnum, pass it to begin_prologue
23086 hook and pass 0 to dwarf2out_begin_prologue.
23087 (final_scan_insn): Update override_columnnum. Pass last_columnnum
23088 to source_line debug hook.
23089 (notice_source_line): Compute last_columnnum and for debug_column_info
23090 return true on column changes.
23091 * debug.h (struct gcc_debug_hooks): Add column argument to
23092 source_line and begin_prologue hooks.
23093 (debug_nothing_int_charstar_int_bool): Remove prototype.
23094 (debug_nothing_int_int_charstar,
23095 debug_nothing_int_int_charstar_int_bool): New prototypes.
23096 (dwarf2out_begin_prologue): Add column argument.
23097 * debug.c (do_nothing_debug_hooks): Adjust source_line and
23098 begin_prologue hooks.
23099 (debug_nothing_int_charstar_int_bool): Remove.
23100 (debug_nothing_int_int_charstar,
23101 debug_nothing_int_int_charstar_int_bool): New functions.
23102 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
23103 through to dwarf2out_source_line.
23104 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
23105 (dwarf2out_source_line): Add column argument, emit it if requested.
23106 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
23108 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
23109 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
23110 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
23111 through to dwarf2out_begin_prologue.
23112 (vmsdbgout_source_line): Add column argument, pass it through to
23113 dwarf2out_source_line.
23114 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
23115 dbxout_source_line caller.
23116 (dbxout_source_line): Add column argument.
23118 * common.opt (gno-column-info, gcolumn-info): New options.
23119 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
23120 (check_die): Also test for multiple DW_AT_decl_column attributes.
23121 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
23122 DW_AT_decl_column if requested.
23123 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
23125 (gen_variable_die): Likewise.
23126 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
23127 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
23130 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
23131 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
23132 (ix86_handle_option): Handle OPT_m3dnowa.
23133 * doc/invoke.texi (-m3dnowa): Document.
23134 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
23135 -m3dnowa instead of -m3dnow -march=athlon.
23138 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
23139 instead of gcc_assert for K, r and R code checks. Formatting fixes.
23141 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23144 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
23145 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
23146 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
23147 generator for vsx_xxpermdi_<mode>_be.
23148 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
23149 force big-endian semantics.
23150 (vsx_xxpermdi_<mode>_be): New define_expand with same
23151 implementation as previous version of vsx_xxpermdi_<mode>.
23153 2017-02-17 Jakub Jelinek <jakub@redhat.com>
23155 PR tree-optimization/79327
23156 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
23157 variable, its initialization and use.
23159 2017-02-17 Julia Koval <julia.koval@intel.com>
23161 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
23162 (OPTION_MASK_ISA_PKU_UNSET): New.
23163 (ix86_handle_option): Handle -mrdpid.
23164 * config/i386/cpuid.h (bit_RDPID): New.
23165 * config/i386/driver-i386.c (host_detect_local_cpu):
23166 Detect RDPID feature.
23167 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
23168 * config/i386/i386-c.c (ix86_target_macros_internal):
23170 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
23171 (ix86_valid_target_attribute_inner_p): Add "rdpid".
23172 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
23173 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
23174 * config/i386/i386.md (define_insn "rdpid"): New.
23175 * config/i386/i386.opt Add -mrdpid.
23176 * config/i386/immintrin.h (_rdpid_u32): New.
23178 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
23180 PR rtl-optimization/79541
23181 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
23182 instead of transforming it into USE.
23184 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
23186 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
23187 If HONOR_SNANS (SFmode) force the input to a register.
23188 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
23189 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
23190 an frsp or similar insn.
23192 2017-02-17 Martin Liska <mliska@suse.cz>
23194 PR rtl-optimization/79577
23195 * params.def (selsched-max-sched-times): Increase minimum to 1.
23197 2017-02-17 Martin Liska <mliska@suse.cz>
23199 PR rtl-optimization/79574
23200 * gcse.c (want_to_gcse_p): Prevent integer overflow.
23202 2017-02-17 Martin Liska <mliska@suse.cz>
23204 PR tree-optimization/79529
23205 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
23206 ssa_defined_default_def_p to handle cases which are implicitly
23208 * tree-ssa.c (ssa_defined_default_def_p): New function.
23209 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
23210 which are implicitly defined.
23211 * tree-ssa.h (ssa_defined_default_def_p): Declare.
23213 2017-02-17 Richard Biener <rguenther@suse.de>
23215 PR middle-end/79576
23216 * params.def (max-ssa-name-query-depth): Limit to 10.
23218 2017-02-17 Richard Biener <rguenther@suse.de>
23220 PR tree-optimization/79552
23221 * tree-ssa-structalias.c (visit_loadstore): Properly verify
23224 2017-02-17 Richard Biener <rguenther@suse.de>
23227 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
23229 2017-02-17 Marek Polacek <polacek@redhat.com>
23231 PR middle-end/79536
23232 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
23233 (fold_negate_expr): New wrapper.
23235 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
23237 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
23238 Correct terminology and de-emphasize pre-standard behavior.
23240 2017-02-16 Alan Modra <amodra@gmail.com>
23242 PR rtl-optimization/79286
23243 * ira.c (def_dominates_uses): New function.
23244 (update_equiv_regs): Don't create an equivalence for insns that
23245 may trap where the register def does not dominate the use.
23247 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
23249 PR rtl-optimization/78127
23250 * lra.c (lra): Call lra_eliminate before finish the loop after
23253 2017-02-16 Richard Biener <rguenther@suse.de>
23255 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
23257 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
23258 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
23259 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
23260 (isl_val_int_from_wi): New function.
23261 (extract_affine_gmp): Rename to ...
23262 (extract_affine_wi): ... this, take a widest_int.
23263 (extract_affine_int): Just wrap extract_affine_wi.
23264 (add_param_constraints): Use isl_val_int_from_wi.
23265 (add_loop_constraints): Likewise, and extract_affine_wi.
23267 2017-02-15 Jeff Law <law@redhat.com>
23269 PR middle-end/79521
23270 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
23271 ira_init_register_move_cost_if_necessary.
23273 2017-02-15 Martin Sebor <msebor@redhat.com>
23275 PR middle-end/32003
23276 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
23277 removed in a prior commit.
23279 2017-02-15 Bin Cheng <bin.cheng@arm.com>
23281 PR tree-optimization/79347
23282 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
23283 counters during peeling.
23285 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
23287 * Makefile.in (site.exp): Remove "set ISLVER".
23289 2017-02-15 Jakub Jelinek <jakub@redhat.com>
23292 * real.c (real_from_integer): Call real_convert even for decimal.
23294 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
23297 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
23299 2017-02-14 Andrew Pinski <apinski@cavium.com>
23301 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
23302 cores and change the partno/implementer to be correct.
23303 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
23304 the 'B" as the implementer.
23305 * config/aarch64/aarch64-tune.md: Regenerate.
23307 2017-02-14 Carl Love <cel@us.ibm.com>
23309 * config/rs6000/rs6000.c: Add case statement entry to make the
23310 xvcvuxdsp built-in argument unsigned.
23311 * config/rs6000/vsx.md: Fix the source and return operand types so they
23312 match the instruction definitions from the ISA document. Fix typo
23313 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
23316 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
23319 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
23320 member early_clobber_alts.
23321 * lra-lives.c (reg_early_clobber_p): New.
23322 (process_bb_lives): Use it.
23323 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
23324 (debug_operand_data): Initialize early_clobber_alts.
23325 (setup_operand_alternative): Set up early_clobber_alts.
23326 (collect_non_operand_hard_regs): Ditto. Pass early clobber
23327 alternatives to new_insn_reg.
23328 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
23330 (lra_update_insn_regno_info): Pass the new arg.
23332 2017-02-14 Jakub Jelinek <jakub@redhat.com>
23334 PR middle-end/79505
23335 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
23336 (new_oacc_loop_raw): Don't clear already cleared fields.
23339 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
23340 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
23341 _mm512_prefetch_i64gather_ps): New inline functions and macros.
23343 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
23346 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
23348 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
23351 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
23352 the extra instruction to the right place to store 128-bit constant
23355 2017-02-14 Martin Sebor <msebor@redhat.com>
23357 PR middle-end/79448
23358 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
23359 warning for strings of unknown length.
23361 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
23363 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
23365 2017-02-14 Jeff Law <law@redhat.com>
23368 * ira-costs.c (scan_one_insn): Initialize register move costs
23369 for pseudos seen in USE/CLOBBER insns.
23371 PR tree-optimization/79095
23372 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
23373 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
23374 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
23375 if the operands are known to be not equal, then the resulting range
23377 (intersect_ranges): If the new range is ~[0,0] and the old range is
23378 wide, then prefer ~[0,0].
23379 * tree-vrp.c (overflow_comparison_p_1): New function.
23380 (overflow_comparison_p): New function.
23381 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
23382 if NAME is used in an overflow test.
23383 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
23384 overflow check that can be expressed as an equality test, then adjust
23385 ops to be that equality test.
23387 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23389 * config/s390/s390-builtin-types.def: Remove flags argument.
23390 * config/s390/s390.c (s390_init_builtins): Likewise.
23392 2017-02-14 Martin Liska <mliska@suse.cz>
23394 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
23395 vector. Fix trailing white spaces.
23397 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
23399 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
23402 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23404 PR rtl-optimization/68664
23405 * config/arm/arm.c (arm_sched_can_speculate_insn):
23406 New function. Declare prototype.
23407 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
23409 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23411 PR rtl-optimization/68664
23412 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
23414 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
23416 2017-02-14 Amit Pawar <amit.pawar@amd.com>
23418 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
23419 max skip bytes for function, loop and jump.
23421 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23423 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
23424 ABS_EXPR for gimple dump.
23426 2017-02-14 Jakub Jelinek <jakub@redhat.com>
23429 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
23431 PR tree-optimization/79408
23432 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
23433 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
23434 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
23435 also if rhs1 is INTEGER_CST.
23437 2017-02-14 Richard Biener <rguenther@suse.de>
23439 PR middle-end/79432
23440 * tree-into-ssa.c (insert_phi_nodes): When the function can
23441 have abnormal edges rewrite SSA names with broken use-def
23442 dominance out of SSA and register them for PHI insertion.
23444 2017-02-13 Martin Sebor <msebor@redhat.com>
23446 PR middle-end/79496
23447 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
23448 clearing info.nowrite flag when snprintf size argument is a range.
23450 2017-02-13 Jakub Jelinek <jakub@redhat.com>
23452 * cprop.c (cprop_jump): Add missing space in string literal.
23453 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
23454 (get_constraint_for_component_ref): Likewise.
23455 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
23456 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
23457 * lra-constraints.c (process_alt_operands): Likewise.
23458 * ipa-inline.c (inline_small_functions): Likewise.
23459 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
23460 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
23461 * trans-mem.c (diagnose_tm_1_op): Likewise.
23462 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
23463 (grid_parallel_clauses_gridifiable): Likewise.
23465 * config/nvptx/mkoffload.c (process): Add space in between
23468 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
23469 "MOD4_SSE_REGS" and "ALL_REGS".
23471 * spellcheck.c (test_data): Add , in between "foo" and "food".
23473 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23476 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
23477 boundary crossing check and subsequent code generation agree.
23479 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23481 * config/aarch64/aarch64.c (has_memory_op): Delete.
23482 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
23485 2017-02-13 Jakub Jelinek <jakub@redhat.com>
23487 PR rtl-optimization/79388
23488 PR rtl-optimization/79450
23489 * combine.c (distribute_notes): When removing TEM_INSN for which
23490 corresponding dest has last value recorded, invalidate that last
23493 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23495 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
23496 of explicit '@'. Add missing assembly comment marker on branch costs
23499 2017-02-13 Nathan Sidwell <nathan@acm.org>
23501 * gengtype-lex.l (<in_struct>): Add '/'.
23503 2017-02-13 Martin Liska <mliska@suse.cz>
23506 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
23508 2017-02-13 Richard Biener <rguenther@suse.de>
23510 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
23512 * configure: Re-generate.
23513 * config.in: Likewise.
23514 * graphite-dependences.c: Simplify as if
23515 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
23516 * graphite-isl-ast-to-gimple.c: Likewise.
23517 * graphite-optimize-isl.c: Likewise.
23518 * graphite-poly.c: Likewise.
23519 * graphite-sese-to-poly.c: Likewise.
23520 * graphite.h: Likewise.
23521 * toplev.c: Include isl/version.h and use isl_version () for
23522 printing the ISL version.
23523 * doc/install.texi: Update ISL requirement.
23525 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
23527 * doc/standards.texi (Standards): Update reference to
23530 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
23532 * doc/extend.texi (Named Address Spaces): sourceware.org now
23534 * doc/install.texi (Binaries): Ditto.
23537 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
23539 * doc/cpp.texi: Replace "stringify"/"stringification" with C
23540 standard terminology "stringize"/"stringizing" throughout.
23541 * doc/cppinternals.texi: Likewise.
23543 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
23545 * doc/extend.texi: Fix some spelling mistakes and typos.
23546 * doc/invoke.texi: Likewise.
23548 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
23551 * params.def (inline-min-speedup) Change from 10 to 8.
23553 2017-02-11 Jakub Jelinek <jakub@redhat.com>
23555 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
23558 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
23561 * ipa-inline-analysis.c (get_minimal_bb): New function.
23562 (record_modified): Use it.
23563 (remap_edge_change_prob): Handle also ancestor functions.
23565 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
23567 * doc/contrib.texi (Contributors): Remove broken link into
23568 the Mauve CVS repository.
23570 2017-02-11 Jakub Jelinek <jakub@redhat.com>
23572 PR middle-end/79454
23573 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
23574 result computation whenever lhs doesn't have vector mode, not
23575 just when it has BLKmode.
23577 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
23579 * doc/makefile.texi (profiledbootstrap): Refer to the
23580 installation instructions only in textual form.
23582 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23585 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
23587 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
23589 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
23590 (Specific): Update mingw-w64 reference.
23592 (Specific): Remove broken link to Renesas RX processor.
23594 2017-02-10 Richard Biener <rguenther@suse.de>
23596 * toplev.c (process_options): Do not mention obsolete graphite
23597 options when printing sorry message about missing graphite support.
23598 Mention -floop-nest-optimize.
23600 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
23602 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
23603 (vtst_p16): Likewise.
23604 (vtstq_p8): Likewise.
23605 (vtstq_p16): Likewise.
23607 (vtstq_p64): Likewise.
23608 * config/arm/arm_neon.h (vgetq_lane_p64): New.
23609 (vset_lane_p64): New.
23610 (vsetq_lane_p64): New.
23612 2017-02-10 Jakub Jelinek <jakub@redhat.com>
23614 PR tree-optimization/79411
23615 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
23616 stmt operands are SSA_NAMEs used in abnormal phis.
23617 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
23620 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
23623 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
23626 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
23628 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
23630 2017-02-09 Jakub Jelinek <jakub@redhat.com>
23632 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
23636 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
23637 not arbitrary TREE_CONSTANT.
23640 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
23641 "omp declare target link" attribute unless is_global_var.
23642 * omp-offload.c (find_link_var_op): Likewise.
23644 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
23645 Chung-Lin Tang <cltang@codesourcery.com>
23647 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
23649 (gimplify_adjust_omp_clauses): Don't delete TILE.
23650 (gimplify_omp_for): Deal with TILE.
23651 * internal-fn.c (expand_GOACC_TILE): New function.
23652 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
23654 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
23655 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
23657 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
23658 avoid DIV for outermost collapse var.
23659 (expand_oacc_for): Insert tile element loop as needed. Adjust.
23660 Remove out of date comments, fix whitespace.
23661 * omp-general.c (omp_extract_for_data): Deal with tiling.
23662 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
23663 adjust OLF_DIM_BASE value.
23664 (struct omp_for_data): Add tiling field.
23665 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
23666 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
23667 for auto loops. Remove default auto determining, moved to
23668 oacc_loop_fixed_partitions.
23669 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
23670 stmts, add e_mask field.
23671 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
23672 (oacc_thread_numbers): Use oacc_dim_call.
23673 (oacc_xform_tile): New.
23674 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
23675 (finish_oacc_loop): Adjust for ifns vector.
23676 (oacc_loop_discover_walk): Append loop abstraction sites to list,
23677 add case for GOACC_TILE fns.
23678 (oacc_loop_xform_loop): Delete.
23679 (oacc_loop_process): Iterate over call list directly, and add
23680 handling for GOACC_TILE fns.
23681 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
23683 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
23684 vector partitioning to outer loops. Assign 2 partitions to loops
23685 when available. Add TILE handling.
23686 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
23687 (execite_oacc_device_lower): Process GOACC_TILE fns,
23688 ignore unknown specs.
23689 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
23690 * tree.c (omp_clause_num_ops): Adjust TILE ops.
23691 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
23693 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
23695 * configure.ac (ACX_BUGURL): Update.
23696 * configure: Regenerate.
23698 2017-02-09 Richard Biener <rguenther@suse.de>
23700 PR tree-optimization/69823
23701 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
23702 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
23704 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
23706 * config/arc/arc-c.def: Add __NPS400__ definition.
23707 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
23708 (TARGET_NPS400): Define.
23710 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
23712 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
23714 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
23715 pointer, arch_info.
23716 (arc_cpu_types): Fill the arch_info field with a pointer into the
23717 arc_arch_types table.
23718 (arc_selected_cpu): Declare.
23719 * config/arc/arc.c (arc_selected_cpu): Make global.
23720 (arc_selected_arch): Delete.
23721 (arc_base_cpu): Delete.
23722 (arc_override_options): Remove references to deleted variables,
23723 update access to arch information.
23724 (ARC_OPT): Update access to arch information.
23725 (ARC_OPTX): Likewise.
23726 * config/arc/arc.h (arc_base_cpu): Remove declaration.
23727 (TARGET_ARC600): Update access to arch information.
23728 (TARGET_ARC601): Likewise.
23729 (TARGET_ARC700): Likewise.
23730 (TARGET_EM): Likewise.
23731 (TARGET_HS): Likewise.
23732 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
23735 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
23738 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
23739 condition/operands for integer GE/LE/GEU/LEU operations.
23741 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
23743 PR translation/79397
23744 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
23747 2017-02-08 Martin Jambor <mjambor@suse.cz>
23750 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
23751 whether allocation happened.
23752 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
23753 nothing was allocated.
23755 2017-02-08 Jakub Jelinek <jakub@redhat.com>
23757 PR tree-optimization/79408
23758 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
23759 constant, but SSA_NAME with a known integer range, use the minimum
23760 of that range instead of op1 to determine if modulo can be replaced
23761 with its first operand.
23763 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23765 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
23767 2017-02-08 Richard Biener <rguenther@suse.de>
23769 PR tree-optimization/71824
23770 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
23771 Check all loops contained in the merged region.
23773 2017-02-07 Andrew Pinski <apinski@cavium.com>
23775 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
23777 2017-02-07 Andrew Pinski <apinski@cavium.com>
23779 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
23780 (thunderxt88): Likewise.
23781 (thunderxt81): Disable LSE and change v8.1 to v8.
23782 (thunderxt83): Likewise.
23784 2017-02-07 Jakub Jelinek <jakub@redhat.com>
23785 Richard Biener <rguenther@suse.de>
23787 PR middle-end/79399
23788 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
23789 type from int to size_t.
23790 * ira-costs.c (struct_costs_size): Change type from int to size_t.
23792 2017-02-07 Jakub Jelinek <jakub@redhat.com>
23794 PR rtl-optimization/79386
23795 * cprop.c (bypass_conditional_jumps): Initialize
23796 bypass_last_basic_block already before splitting bbs after
23797 unconditional traps...
23798 (bypass_conditional_jumps): ... rather than here.
23801 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
23802 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
23803 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
23804 fix -masm=intel patterns.
23806 2017-02-07 Richard Biener <rguenther@suse.de>
23808 PR tree-optimization/79256
23809 PR middle-end/79278
23810 * builtins.c (get_object_alignment_2): Use min_align_of_type
23811 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
23812 and ADJUST_FIELD_ALIGN.
23814 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
23816 * doc/tm.texi: Regenerate.
23817 * stor-layout.c (layout_decl): Adjust.
23818 (update_alignment_for_field): Likewise.
23819 (place_field): Likewise.
23820 (min_align_of_type): Likewise.
23821 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
23822 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
23823 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
23824 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
23825 * config/frv/frv.c (frv_adjust_field_align): Likewise.
23826 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
23827 * config/i386/i386.c (x86_field_alignment): Likewise.
23828 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
23829 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
23830 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
23831 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
23832 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
23833 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
23837 2017-01-30 Richard Biener <rguenther@suse.de>
23839 PR tree-optimization/79256
23840 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
23841 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
23844 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
23846 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
23847 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
23848 builtins to SImode and emit a zero-extend, if necessary.
23850 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
23852 * docs/invoke.texi (RISC-V Options): Alphabetize.
23854 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
23856 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
23859 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
23861 * config/riscv/riscv.c: New file.
23862 * common/config/riscv/riscv-common.c: Likewise.
23863 * config.gcc: Likewise.
23864 * config/riscv/constraints.md: Likewise.
23865 * config/riscv/elf.h: Likewise.
23866 * config/riscv/generic.md: Likewise.
23867 * config/riscv/linux.h: Likewise.
23868 * config/riscv/multilib-generator: Likewise.
23869 * config/riscv/peephole.md: Likewise.
23870 * config/riscv/pic.md: Likewise.
23871 * config/riscv/predicates.md: Likewise.
23872 * config/riscv/riscv-builtins.c: Likewise.
23873 * config/riscv/riscv-c.c: Likewise.
23874 * config/riscv/riscv-ftypes.def: Likewise.
23875 * config/riscv/riscv-modes.def: Likewise.
23876 * config/riscv/riscv-opts.h: Likewise.
23877 * config/riscv/riscv-protos.h: Likewise.
23878 * config/riscv/riscv.h: Likewise.
23879 * config/riscv/riscv.md: Likewise.
23880 * config/riscv/riscv.opt: Likewise.
23881 * config/riscv/sync.md: Likewise.
23882 * config/riscv/t-elf-multilib: Likewise.
23883 * config/riscv/t-linux: Likewise.
23884 * config/riscv/t-linux-multilib: Likewise.
23885 * config/riscv/t-riscv: Likewise.
23886 * configure.ac: Likewise.
23887 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
23888 Waterman as RISC-V maintainers.
23889 * doc/install.texi: Add RISC-V entries.
23890 * doc/invoke.texi: Add RISC-V options section.
23891 * doc/md.texi: Add RISC-V constraints section.
23892 * configure: Regenerated.
23894 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
23897 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
23898 false values to be constant vectors with all 0 or all 1 bits set.
23899 (vcondu<mode><mode>): Likewise.
23900 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
23902 (fpmask_comparison_operator): Update comment.
23903 (vecint_comparison_operator): New predicate.
23904 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
23905 vector conditionals when the true and false values are constant
23906 vectors with all 0 bits or all 1 bits set.
23908 2017-02-06 Martin Sebor <msebor@redhat.com>
23910 PR tree-optimization/79376
23911 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
23913 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
23915 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
23916 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
23917 to simplify split condition.
23919 2017-02-06 Jakub Jelinek <jakub@redhat.com>
23921 * omp-expand.c (oxpand_omp_atomic_fetch_op,
23922 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
23925 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
23927 PR rtl-optimization/68664
23928 * target.def (can_speculate_insn): New hook.
23929 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
23930 * doc/tm.texi: Regenerate.
23931 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
23932 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
23933 (rs6000_sched_can_speculate_insn): New function.
23935 2017-02-06 Jakub Jelinek <jakub@redhat.com>
23937 PR tree-optimization/79284
23938 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
23939 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
23940 vectorizable_mask_load_store, vectorizable_operation,
23941 vect_is_simple_cond, get_same_sized_vectype): Use it instead
23942 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
23943 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
23944 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
23945 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
23946 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
23947 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
23948 is_gimple_assign (stmt). Replace another such test with
23949 is_gimple_assign (stmt).
23951 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
23954 * config/avr/avr.c (rtl-iter.h): Include it.
23955 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
23956 (avr_legitimate_combined_insn): ...and implementation.
23958 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
23960 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
23961 * config/s390/s390.c (s390_const_operand_ok)
23962 (s390_canonicalize_comparison, s390_extract_part)
23963 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
23964 (s390_contiguous_bitmask_p, s390_rtx_costs)
23965 (legitimize_pic_address): Likewise.
23966 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
23967 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
23968 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
23969 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
23970 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
23972 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
23974 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
23975 REGNO($0) == REGNO($1).
23977 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23979 * config/s390/linux.h(SIZE_TYPE): Add comment.
23981 2017-02-06 Julian Brown <julian@codesourcery.com>
23982 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23983 Virendra Pathak <virendra.pathak@broadcom.com>
23985 * config/aarch64/aarch64-cores.def: Change the scheduler
23987 * config/aarch64/aarch64.md: Include thunderx2t99.md.
23988 * config/aarch64/thunderx2t99.md: New file.
23990 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
23992 * doc/standards.texi (Go Language): Update link to language
23995 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
23997 * tree-eh.c (lower_resx): Sanitize profile.
23998 (cleanup_empty_eh_move_lp): Likewise.
24000 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
24003 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
24005 * cfgloopmanip.h (loop_version): Update prototype.
24006 * modulo-sched.c (sms_schedule): Update call of loop_version.
24007 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
24008 * tree-parloops.c (gen_parallel_loop): Likewise.
24009 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
24010 * tree-ssa-loop-split.c (split_loop): Likewise.
24011 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
24012 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
24014 2017-02-05 Martin Liska <mliska@suse.cz>
24017 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
24019 (print_operand_address): Initialize a struct to zero.
24021 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
24023 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
24024 garbage collector only in textual form.
24026 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
24028 * doc/extend.texi (x86 specific memory model extensions for
24029 transactional memory): Simplify a phrase.
24031 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
24034 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
24035 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
24036 (atomic_storedi_1): Likewise.
24038 2017-02-04 Jakub Jelinek <jakub@redhat.com>
24040 PR tree-optimization/79338
24041 * tree-parloops.c (gather_scalar_reductions): Don't call
24042 vect_analyze_loop_form for loop->inner before destroying loop's
24045 2017-02-03 Martin Sebor <msebor@redhat.com>
24047 PR tree-optimization/79327
24048 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
24049 when precision has resulted in leading zeros.
24050 (format_integer): Adjust the likely counter to assume an unknown
24051 argument that may be zero is non-zero.
24053 2017-02-03 Jason Merrill <jason@redhat.com>
24056 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
24057 avoid copying non-taken branch.
24059 2017-02-03 Jakub Jelinek <jakub@redhat.com>
24061 PR tree-optimization/79340
24062 * tree-vect-loop.c (vectorizable_reduction): Release
24063 vec_defs elements after safe_splicing them into other vectors.
24066 PR tree-optimization/79327
24067 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
24068 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
24070 (format_integer): Use wide_int_to_tree instead of build_int_cst
24071 + to_?hwi. If argmin is NULL, just set argmin and argmax to
24072 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
24073 of shortest and longest sequence.
24075 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
24077 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
24078 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
24080 2017-02-03 Walter Lee <walt@tilera.com>
24083 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
24084 after initial stackframe link reg save.
24085 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
24087 2017-02-03 Jakub Jelinek <jakub@redhat.com>
24090 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
24091 wu for stxssp alternative.
24093 2017-02-03 Martin Sebor <msebor@redhat.com>
24095 PR tree-optimization/79352
24096 * gimple-fold.c (get_range_strlen): Add argument.
24097 (get_range_strlen): Change return type to bool.
24098 (get_maxval_strlen): Pass in a dummy argument.
24099 * gimple-fold.h (get_range_strlen): Change return type to bool.
24100 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
24101 * tree.h (array_at_struct_end_p): Add argument.
24102 * tree.c (array_at_struct_end_p): Handle it.
24104 2017-02-03 Martin Liska <mliska@suse.cz>
24107 * multiple_target.c (create_dispatcher_calls): Redirect edge
24108 from a caller of a dispatcher.
24109 (expand_target_clones): Make the clones local.
24110 (ipa_target_clone): Do both target clones and resolvers.
24111 (ipa_dispatcher_calls): Remove the pass.
24112 (pass_dispatcher_calls::gate): Likewise.
24113 (make_pass_dispatcher_calls): Likewise.
24114 * passes.def (pass_target_clone): Put as very first IPA early
24117 2017-02-03 Martin Liska <mliska@suse.cz>
24119 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
24120 in case of a function with ifunc attribute.
24122 2017-02-03 Martin Liska <mliska@suse.cz>
24124 * cgraph.c (cgraph_node::dump): Dump function version info.
24125 * symtab.c (symtab_node::dump_base): Add missing new line.
24127 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
24129 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
24130 (ifcombine_ifandif): Use it.
24132 2017-02-03 Martin Liska <mliska@suse.cz>
24134 * doc/invoke.texi: Document default value for
24135 use-after-scope-direct-emission-threshold.
24137 2017-02-03 Martin Liska <mliska@suse.cz>
24139 PR tree-optimization/79339
24140 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
24141 (format_floating): Likewise.
24143 2017-02-03 Martin Liska <mliska@suse.cz>
24146 * ipa-prop.c (ipa_node_params_t::insert): Remove current
24148 (ipa_node_params_t::remove): Likewise.
24149 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
24150 initialization from removed ipa_node_params_t::insert.
24151 (ipa_node_params::~ipa_node_params): Move from removed
24152 ipa_node_params_t::release.
24153 * symbol-summary.h (symbol_summary::m_released): New member.
24154 Do not release a summary twice. Do not allow to call finalizer
24155 for types of a summary that live in GGC memory.
24157 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24159 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
24162 2017-02-02 Martin Sebor <msebor@redhat.com>
24164 PR middle-end/79275
24165 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
24166 (format_string): Tighten up the range of output for non-constant
24167 strings and correct the expected range for wide non-constant strings.
24169 2017-02-02 Martin Sebor <msebor@redhat.com>
24171 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
24173 PR middle-end/32003
24174 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
24176 (-fdump-tree-@var): Add to index and document how to come up
24177 with pass-specific option and dump file names.
24178 (-fdump-passes): Clarify where to look for output.
24180 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
24182 PR middle-end/77445
24183 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
24184 statistics of the analyzed path; allow threading for speed when
24185 any of BBs along the path are optimized for speed.
24187 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
24189 PR middle-end/78468
24190 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
24191 settings of the virtual registers.
24194 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
24196 * explow.c (get_dynamic_stack_size): Take known alignment of stack
24197 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
24200 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24202 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
24203 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
24205 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24207 * config/s390/s390.md: Add missing comments with the expanded
24209 * config/s390/vector.md: Likewise.
24210 * config/s390/vx-builtins.md: Likewise.
24212 2017-02-02 Jakub Jelinek <jakub@redhat.com>
24215 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
24216 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
24217 conditions on a single line.
24219 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24221 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
24222 __S390_VX__ to __VX__.
24224 2017-02-01 Andrew Pinski <apinski@cavium.com>
24226 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
24227 stmt_info to record_stmt_cost.
24228 (vect_get_known_peeling_cost): Pass stmt_info if known to
24230 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
24231 cpu_vector_cost field into
24232 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
24233 field into vec_int_stmt_cost and vec_fp_stmt_cost.
24234 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
24235 splitting of scalar_stmt_cost and vec_stmt_cost.
24236 (thunderx_vector_cost): Likewise.
24237 (cortexa57_vector_cost): LIkewise.
24238 (exynosm1_vector_cost): Likewise.
24239 (xgene1_vector_cost): Likewise.
24240 (thunderx2t99_vector_cost): Improve after the splitting of the two
24242 (aarch64_builtin_vectorization_cost): Update for the splitting of
24243 scalar_stmt_cost and vec_stmt_cost.
24245 2017-02-01 Torvald Riegel <triegel@redhat.com>
24246 Richard Henderson <rth@redhat.com>
24248 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
24249 conditional on existance of a fast atomic load.
24250 * optabs-query.c (can_atomic_load_p): New function.
24251 * optabs-query.h (can_atomic_load_p): Declare it.
24252 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
24253 no fast atomic load is available for the particular size of access.
24254 (expand_atomic_compare_and_swap): Likewise.
24255 (expand_atomic_load): Likewise.
24256 (expand_atomic_store): Likewise.
24257 (expand_atomic_fetch_op): Likewise.
24258 * testsuite/lib/target-supports.exp
24259 (check_effective_target_sync_int_128): Remove x86 because it provides
24260 no fast atomic load.
24261 (check_effective_target_sync_int_128_runtime): Likewise.
24263 2017-02-01 Richard Biener <rguenther@suse.de>
24265 * graphite.c: Include tree-vectorizer.h for find_loop_location.
24266 (graphite_transform_loops): Provide opt-info for optimized nests.
24267 * tree-parloop.c (parallelize_loops): Provide opt-info for
24268 parallelized loops.
24270 2017-02-01 Richard Biener <rguenther@suse.de>
24272 PR middle-end/79315
24273 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
24274 was not set before.
24276 2017-02-01 Richard Biener <rguenther@suse.de>
24278 PR tree-optimization/71824
24279 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
24280 Verify the loops are valid in the merged SESE region.
24281 (scop_detection::can_represent_loop_1): Check analyzing the
24282 evolution of the number of iterations in the region succeeds.
24284 2017-01-31 Ian Lance Taylor <iant@golang.org>
24286 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
24287 REG_ARGS_SIZE note to 32-bit push insns and call insn.
24289 2017-01-31 David Malcolm <dmalcolm@redhat.com>
24291 PR preprocessor/79210
24292 * input.c (get_substring_ranges_for_loc): Replace line_width
24293 assertion with error-handling.
24295 2017-01-31 Richard Biener <rguenther@suse.de>
24297 PR tree-optimization/77318
24298 * graphite-sese-to-poly.c (extract_affine): Fix assert.
24299 (create_pw_aff_from_tree): Take loop parameter.
24300 (add_condition_to_pbb): Pass loop of the condition to
24301 create_pw_aff_from_tree.
24303 2017-01-31 Jakub Jelinek <jakub@redhat.com>
24305 * config/s390/s390.c (s390_asan_shadow_offset): New function.
24306 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
24308 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
24312 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
24314 (convert_int_to_float128): Likewise.
24315 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
24316 (convert_int_to_float128): Likewise.
24317 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
24318 (UNSPEC_IEEE128_CONVERT): Likewise.
24319 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
24320 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
24321 Use local variables for IBM extended format.
24322 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
24323 (fix_trunc<mode>si2_fprs): Likewise.
24324 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
24325 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
24326 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
24327 to know that we can now have integers of all sizes in vector
24329 (fix<uns>_<mode>di2_hw): Likewise.
24330 (float<uns>_<mode>si2_hw): Likewise.
24331 (fix_<mode>si2_hw): Likewise.
24332 (fixuns_<mode>si2_hw): Likewise.
24333 (float<uns>_<mode>di2_hw): Likewise.
24334 (float_<mode>di2_hw): Likewise.
24335 (float_<mode>si2_hw): Likewise.
24336 (floatuns_<mode>di2_hw): Likewise.
24337 (floatuns_<mode>si2_hw): Likewise.
24338 (xscvqp<su>wz_<mode>): Delete, no longer used.
24339 (xscvqp<su>dz_<mode>): Likewise.
24340 (xscv<su>dqp_<mode>): Likewise.
24341 (ieee128_mfvsrd_64bit): Likewise.
24342 (ieee128_mfvsrd_32bit): Likewise.
24343 (ieee128_mfvsrwz): Likewise.
24344 (ieee128_mtvsrw): Likewise.
24345 (ieee128_mtvsrd_64bit): Likewise.
24346 (ieee128_mtvsrd_32bit): Likewise.
24348 2017-01-31 Martin Liska <mliska@suse.cz>
24351 * ipa-prop.c (ipa_free_all_node_params): Call release method
24352 instead of ~sumbol_summary to not to trigger double times
24355 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
24357 PR tree-optimization/71691
24358 * bitmap.h (class auto_bitmap): New.
24359 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
24360 is_maybe_undefined instead of ssa_undefined_value_p.
24362 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24364 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
24365 __S390_ARCH_LEVEL__ to __ARCH__.
24367 2017-01-31 Jakub Jelinek <jakub@redhat.com>
24369 PR tree-optimization/79267
24370 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
24371 if should_remove_lhs_p is true.
24373 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
24376 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
24377 (add_alignment_attribute): New.
24378 (base_type_die): Add alignment attribute.
24379 (subrange_type_die): Likewise.
24380 (modified_type_die): Likewise.
24381 (gen_array_type_die): Likewise.
24382 (gen_descr_array_type_die: Likewise.
24383 (gen_enumeration_type_die): Likewise.
24384 (gen_subprogram_die): Likewise.
24385 (gen_variable_die): Likewise.
24386 (gen_field_die): Likewise.
24387 (gen_ptr_to_mbr_type_die): Likewise.
24388 (gen_struct_or_union_type_die): Likewise.
24389 (gen_subroutine_type_die): Likewise.
24390 (gen_typedef_die): Likewise.
24391 (base_type_cmp): Compare alignment attribute.
24393 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24396 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
24397 (setb_unsigned) New pattern for setb with CCUNS.
24398 * config/rs6000/rs6000.c (expand_block_compare): Use a different
24399 subfc./subfe sequence to avoid overflow problems. Generate a
24400 shorter sequence with cmpld/setb for power9.
24401 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
24402 for generating subfc. instruction.
24403 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
24404 now uses this instruction.
24406 2017-01-30 Ian Lance Taylor <iant@google.com>
24409 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
24410 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
24412 2017-01-30 Martin Sebor <msebor@redhat.com>
24414 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
24415 Move constant to the right of a relational operator.
24416 (get_mpfr_format_length, format_character, format_string): Ditto.
24417 (should_warn_p, maybe_warn): Same.
24419 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
24421 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
24424 * asan.c (get_translation_unit_decl): Remove function.
24425 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
24427 2017-01-30 Martin Liska <mliska@suse.cz>
24429 PR gcov-profile/79259
24430 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
24431 -fprofile-generate.
24433 2017-01-30 Martin Liska <mliska@suse.cz>
24436 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
24437 Initialize variables with NULL value.
24439 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
24442 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
24444 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
24446 2017-01-30 Richard Biener <rguenther@suse.de>
24448 PR tree-optimization/79276
24449 * tree-vrp.c (process_assert_insertions): Properly adjust common
24450 when removing a duplicate.
24452 * gcc.dg/torture/pr79276.c: New testcase.
24454 2017-01-30 Richard Biener <rguenther@suse.de>
24456 PR tree-optimization/79256
24457 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
24458 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
24460 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
24462 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
24465 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
24466 ("*r<noxa>sbg_<mode>_sll_bitmask")
24467 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
24468 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
24469 Use contiguous_bitmask_nowrap_operand.
24471 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24474 * config/rs6000/altivec.h (vec_xl): Revise #define.
24475 (vec_xst): Likewise.
24477 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
24479 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
24481 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
24483 PR rtl-optimization/79194
24484 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
24485 traps before call to bypass_conditional_jumps.
24487 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
24489 PR tree-optimization/71374
24490 * lra-constraints.c (check_conflict_input_operands): New.
24491 (match_reload): Use it.
24493 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
24496 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
24497 account to calculate conflict_set.
24499 2017-01-27 Bin Cheng <bin.cheng@arm.com>
24501 PR rtl-optimization/78559
24502 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
24503 other_insn in combine.
24505 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
24507 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
24508 uint16_type_node for BT_UINT16.
24510 2017-01-27 David Malcolm <dmalcolm@redhat.com>
24512 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
24513 "RTL Tests" to menu.
24514 (GIMPLE Tests): New node.
24515 (RTL Tests): New node.
24517 2017-01-27 Richard Biener <rguenther@suse.de>
24519 PR tree-optimization/79245
24520 * tree-loop-distribution.c (distribute_loop): Apply cost
24521 modeling also to detected patterns.
24523 2017-01-27 Richard Biener <rguenther@suse.de>
24525 PR tree-optimization/71433
24526 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
24527 (compare_assert_loc): New function.
24528 (process_assert_insertions): Sort and optimize assert locations
24529 to remove duplicates and push down identical assertions on
24530 edges to their destination block.
24532 2017-01-27 Richard Biener <rguenther@suse.de>
24534 PR tree-optimization/79244
24535 * tree-vrp.c (remove_range_assertions): Forcefully propagate
24536 out SSA names even if abnormal.
24538 2017-01-27 Jakub Jelinek <jakub@redhat.com>
24540 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
24541 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
24542 instead of MPFR_RNDN.
24544 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
24547 * arm.c (arm_option_override): Don't call build_target_option_node
24548 until after doing all option overrides.
24549 (arm_valid_target_attribute_tree): Likewise.
24551 2017-01-27 Martin Liska <mliska@suse.cz>
24553 * doc/invoke.texi (-fprofile-arcs): Document profiling support
24554 for {cd}tors and C++ {cd}tors.
24556 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
24558 * config/s390/s390.md ("*setmem_long_and")
24559 ("*setmem_long_and_31z"): Use zero_extend instead of and.
24561 2017-01-26 Martin Sebor <msebor@redhat.com>
24563 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
24566 2017-01-26 Martin Sebor <msebor@redhat.com>
24568 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
24569 HAVE_DFmode before using XFmode or DFmode.
24570 (parse_directive): Avoid using the z length modifier to avoid
24571 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
24573 PR middle-end/78703
24574 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
24575 to accept adjustment as an array.
24576 (get_int_range): New function.
24577 (struct directive): Make width and prec arrays.
24578 (directive::set_width, directive::set_precision): Call get_int_range.
24579 (format_integer, format_floating): Handle width and precision ranges.
24580 (format_string, parse_directive): Same.
24582 2017-01-26 Jakub Jelinek <jakub@redhat.com>
24585 * dwarf2out.c (generate_skeleton_bottom_up): For children with
24586 comdat_type_p set, just clone them, but keep the children in the
24590 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
24591 which have direct callers with -fvar-tracking-assignments enabled
24593 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
24594 inside of type units.
24596 2017-01-26 Martin Sebor <msebor@redhat.com>
24598 PR middle-end/78703
24599 * gimple-ssa-sprintf.c (struct result_range): Add likely and
24601 (struct format_result): Replace number_chars, number_chars_min,
24602 and number_chars_max with a single member of struct result_range.
24604 (format_result::operator+=): Adjust.
24605 (struct fmtresult): Remove bounded. Handle likely and unlikely
24607 (fmtresult::adjust_for_width_or_precision): New function.
24608 (fmtresult:type_max_digits): New function.
24609 (bytes_remaining): Handle likely and unlikely counters.
24610 (min_bytes_remaining): Remove.
24611 (format_percent): Simplify.
24612 (format_integer, format_floating): Set likely and unlikely counters.
24613 (get_string_length, format_character, format_string): Same.
24614 (format_plain, should_warn_p): New function.
24615 (maybe_warn): Call should_warn_p. Update diagnostic messages
24616 and handle those for all directives, including plain strings.
24617 (format_directive): Handle likely and unlikely counters.
24618 Remove unnecessary quoting from diagnostics. Add an informational
24620 (add_bytes): Remove.
24621 (pass_sprintf_length::compute_format_length): Simplify.
24622 (try_substitute_return_value): Handle likely and unlikely counters.
24624 2017-01-26 Carl Love <cel@us.ibm.com>
24626 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
24627 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
24629 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
24632 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
24633 endianess for subregs into account.
24634 * lra-constraints.c (lra_constraints): Do risky transformations
24635 always on the first iteration.
24636 * lra-lives.c (check_pseudos_live_through_calls): Add arg
24637 last_call_used_reg_set.
24638 (process_bb_lives): Define and use last_call_used_reg_set.
24639 * lra.c (lra): Always continue after lra_constraints on the first
24642 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
24644 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
24646 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
24648 2017-01-26 Jakub Jelinek <jakub@redhat.com>
24650 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
24651 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
24652 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
24653 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
24654 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
24655 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
24656 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
24657 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
24658 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
24660 2017-01-26 Marek Polacek <polacek@redhat.com>
24663 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
24664 for the third operand.
24666 2017-01-26 Jakub Jelinek <jakub@redhat.com>
24668 PR middle-end/79236
24669 * omp-low.c (struct omp_context): Add simt_stmt field.
24670 (scan_omp_for): Return omp_context *.
24671 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
24672 context to the _simt_ SIMD stmt.
24673 (lower_omp_for): For combined SIMD with sibling _simt_
24674 SIMD, make sure to use the same decls in _looptemp_
24675 clauses as in the sibling.
24677 2017-01-26 David Sherwood <david.sherwood@arm.com>
24679 PR middle-end/79212
24680 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
24683 2017-01-26 Jakub Jelinek <jakub@redhat.com>
24686 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
24687 emit fld b; fld a; if possible.
24689 * brig-builtins.def: Update copyright years.
24690 * config/arm/arm_acle_builtins.def: Update copyright years.
24692 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
24695 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
24696 constraint instead of o for the stxsd instruction.
24698 2017-01-25 Carl Love <cel@us.ibm.com>
24700 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
24701 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
24703 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
24705 * doc/invoke.texi (C++ Dialect Options): Fix typo.
24707 2017-01-25 Richard Biener <rguenther@suse.de>
24709 PR tree-optimization/69264
24710 * target.def (vector_alignment_reachable): Improve documentation.
24711 * doc/tm.texi: Regenerate.
24712 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
24714 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
24715 earlier changes with respect to TYPE_USER_ALIGN.
24716 (vector_alignment_reachable_p): Likewise. Improve dumping.
24718 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24721 * config/arm/arm.md (xordi3): Force constant operand into a register
24724 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24726 * doc/invoke.texi (-fstore-merging): Correct default optimization
24727 levels at which it is enabled.
24728 (-O): Move -fstore-merging from list to...
24731 2017-01-25 Richard Biener <rguenther@suse.de>
24734 * omp-expand.c: Include debug.h.
24735 (expand_omp_taskreg): Make sure to generate early debug before
24736 outlining anything from a function.
24737 (expand_omp_target): Likewise.
24738 (grid_expand_target_grid_body): Likewise.
24740 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
24743 * asan.c (get_translation_unit_decl): New function.
24744 (asan_add_global): Extract modules file name from globals
24745 TRANSLATION_UNIT_DECL name.
24747 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
24750 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
24751 for long calls with APCS frame and VFP.
24753 2017-01-24 David Malcolm <dmalcolm@redhat.com>
24755 * cfg.c (original_copy_tables_initialized_p): New function.
24756 * cfg.h (original_copy_tables_initialized_p): New decl.
24757 * cfgrtl.c (relink_block_chain): Guard the call to
24758 free_original_copy_tables with a call to
24759 original_copy_tables_initialized_p.
24760 * cgraph.h (symtab_node::native_rtl_p): New decl.
24761 * cgraphunit.c (symtab_node::native_rtl_p): New function.
24762 (symtab_node::needed_p): Don't assert for early assembly output
24763 for __RTL functions.
24764 (cgraph_node::finalize_function): Set "force_output" for __RTL
24766 (cgraph_node::analyze): Bail out early for __RTL functions.
24767 (analyze_functions): Update assertion to support __RTL functions.
24768 (cgraph_node::expand): Bail out early for __RTL functions.
24769 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
24771 * function.h (struct function): Update comment for field
24773 * gimple-expr.c: Include "tree-pass.h".
24774 (gimple_has_body_p): Return false for __RTL functions.
24775 * Makefile.in (OBJS): Add run-rtl-passes.o.
24776 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
24778 (gcc::pass_manager::get_clean_slate): New accessor.
24779 * passes.c: Include "insn-addr.h".
24780 (should_skip_pass_p): Add logging. Update logic for running
24781 "expand" to be compatible with both __GIMPLE and __RTL. Guard
24782 property-provider override so it is only done for gimple passes.
24784 (skip_pass): New function.
24785 (execute_one_pass): Call skip_pass when skipping passes.
24786 * read-md.c (md_reader::read_char): Support filtering
24787 the input to a subset of line numbers.
24788 (md_reader::md_reader): Initialize fields
24789 m_first_line and m_last_line.
24790 (md_reader::read_file_fragment): New function.
24791 * read-md.h (md_reader::read_file_fragment): New decl.
24792 (md_reader::m_first_line): New field.
24793 (md_reader::m_last_line): New field.
24794 * read-rtl-function.c (function_reader::create_function): Only
24795 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
24796 curr_properties. Set DECL_INITIAL to a dummy block.
24797 (read_rtl_function_body_from_file_range): New function.
24798 * read-rtl-function.h (read_rtl_function_body_from_file_range):
24800 * run-rtl-passes.c: New file.
24801 * run-rtl-passes.h: New file.
24803 2017-01-24 Jeff Law <law@redhat.com>
24805 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
24808 2017-01-24 Bin Cheng <bin.cheng@arm.com>
24810 PR tree-optimization/79159
24811 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
24812 (record_nonwrapping_iv): Improve boundary using above function if no
24813 value range information.
24815 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
24816 Martin Jambor <mjambor@suse.cz>
24818 * brig-builtins.def: New file.
24819 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
24820 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
24821 (DEF_HSAIL_SAT_BUILTIN): Likewise.
24822 (DEF_HSAIL_INTR_BUILTIN): Likewise.
24823 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
24824 * builtin-types.def (BT_INT8): New.
24825 (BT_INT16): Likewise.
24826 (BT_UINT8): Likewise.
24827 (BT_UINT16): Likewise.
24828 (BT_FN_ULONG): Likewise.
24829 (BT_FN_UINT_INT): Likewise.
24830 (BT_FN_UINT_ULONG): Likewise.
24831 (BT_FN_UINT_LONG): Likewise.
24832 (BT_FN_UINT_PTR): Likewise.
24833 (BT_FN_ULONG_PTR): Likewise.
24834 (BT_FN_INT8_FLOAT): Likewise.
24835 (BT_FN_INT16_FLOAT): Likewise.
24836 (BT_FN_UINT32_FLOAT): Likewise.
24837 (BT_FN_UINT16_FLOAT): Likewise.
24838 (BT_FN_UINT8_FLOAT): Likewise.
24839 (BT_FN_UINT64_FLOAT): Likewise.
24840 (BT_FN_UINT16_UINT32): Likewise.
24841 (BT_FN_UINT32_UINT16): Likewise.
24842 (BT_FN_UINT16_UINT16_UINT16): Likewise.
24843 (BT_FN_INT_PTR_INT): Likewise.
24844 (BT_FN_UINT_PTR_UINT): Likewise.
24845 (BT_FN_LONG_PTR_LONG): Likewise.
24846 (BT_FN_ULONG_PTR_ULONG): Likewise.
24847 (BT_FN_VOID_UINT64_UINT64): Likewise.
24848 (BT_FN_UINT8_UINT8_UINT8): Likewise.
24849 (BT_FN_INT8_INT8_INT8): Likewise.
24850 (BT_FN_INT16_INT16_INT16): Likewise.
24851 (BT_FN_INT_INT_INT): Likewise.
24852 (BT_FN_UINT_FLOAT_UINT): Likewise.
24853 (BT_FN_FLOAT_UINT_UINT): Likewise.
24854 (BT_FN_ULONG_UINT_UINT): Likewise.
24855 (BT_FN_ULONG_UINT_PTR): Likewise.
24856 (BT_FN_ULONG_ULONG_ULONG): Likewise.
24857 (BT_FN_UINT_UINT_UINT): Likewise.
24858 (BT_FN_VOID_UINT_PTR): Likewise.
24859 (BT_FN_UINT_UINT_PTR: Likewise.
24860 (BT_FN_UINT32_UINT64_PTR): Likewise.
24861 (BT_FN_INT_INT_UINT_UINT): Likewise.
24862 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
24863 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
24864 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
24865 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
24866 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
24867 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
24868 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
24869 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
24870 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
24871 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
24872 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
24873 * doc/frontends.texi: List BRIG FE.
24874 * doc/install.texi (Testing): Add BRIG tesring requirements.
24875 * doc/invoke.texi (Overall Options): Mention BRIG.
24876 * doc/standards.texi (Standards): Doucment BRIG HSA version.
24878 2017-01-24 Richard Biener <rguenther@suse.de>
24880 PR translation/79208
24881 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
24883 2017-01-24 Martin Jambor <mjambor@suse.cz>
24886 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
24887 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
24888 and known_contexts.
24890 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
24892 PR middle-end/79123
24893 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
24894 casts from signed to unsigned really don't have a range.
24896 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
24898 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
24899 GMP_RNDx for compatiblity.
24901 2017-01-24 Martin Liska <mliska@suse.cz>
24904 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
24905 that would prevent us to call alloca with -1 as argument.
24907 2017-01-24 Jakub Jelinek <jakub@redhat.com>
24909 * dwarf2out.c (output_compilation_unit_header, output_file_names):
24910 Avoid -Wformat-security warning.
24912 2017-01-23 Andrew Pinski <apinski@cavium.com>
24914 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
24917 2017-01-23 Martin Sebor <msebor@redhat.com>
24919 PR middle-end/78703
24920 * gimple-ssa-sprintf.c (warn_level): New global.
24921 (format_integer): Use it here and throughout the rest of the file.
24922 Use the same switch to compute sign as base.
24923 (maybe_warn): New function.
24924 (format_directive): Factor out warnings into maybe_warn.
24925 Add debugging output. Use warn_level.
24926 (add_bytes): Use warn_level.
24927 (pass_sprintf_length::compute_format_length): Add debugging output.
24928 (try_substitute_return_value): Same.
24929 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
24931 PR middle-end/78703
24932 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
24933 (struct fmtresult, format_integer, format_floating): Adjust.
24934 (fmtresult::fmtresult): Set max correctly in two argument ctor.
24935 (get_string_length, format_string,format_directive): Same.
24936 (pass_sprintf_length::compute_format_length): Same.
24937 (try_substitute_return_value): Simplify slightly.
24939 PR middle-end/78703
24940 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
24941 (fmtresult::operator+=): Outlined.
24942 (struct fmtresult): Add ctors.
24943 (struct conversion_spec): Rename...
24944 (struct directive): ...to this. Add and remove data members.
24945 (directive::set_width, directive::set_precision): New functions.
24946 (format_percent): Use fmtresult ctor.
24947 (get_width_and_precision): Remove.
24948 (format_integer): Make naming changes. Avoid computing width and
24950 (format_floating): Same. Adjust indentation.
24951 (format_character, format_none): New functions.
24952 (format_string): Moved character handling to format_character.
24953 (format_directive): Remove arguments, change return type.
24954 (parse_directive): New function.
24955 (pass_sprintf_length::compute_format_length): Move directive
24956 parsing to parse_directive.
24958 2017-01-23 Jakub Jelinek <jakub@redhat.com>
24960 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
24961 (assign_assembler_name_if_needed): ... this.
24962 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
24963 (assign_assembler_name_if_needed): ... this.
24964 (free_lang_data_in_cgraph): Adjust callers.
24965 * cgraphunit.c (cgraph_node::analyze): Likewise.
24966 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
24969 2017-01-23 Richard Biener <rguenther@suse.de>
24971 PR tree-optimization/79088
24972 PR tree-optimization/79188
24973 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
24974 resetting loop bounds after last path deletion. Reset loop
24975 bounds of the target loop, make code match the comments.
24976 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
24977 Make sure loops need no fixups.
24979 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
24981 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
24982 exponent support with double type for first argument.
24983 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
24984 type returned by __builtin_vec_extract_sig,
24985 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
24986 functions from "vector int" to "vector unsigned int" or from
24987 "vector long long int" to "vector unsigned long long int".
24988 Changed type returned by __builtin_vec_extract_exp,
24989 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
24990 functions from "vector int" to "vector unsigned int" or from
24991 "vector long long int" to "vector unsigned long long int".
24992 Changed return type of __builtin_vec_test_data_class,
24993 __builtin_vec_test_data_class_sp, and
24994 __builtin_vec_test_data_class_dp from "vector int" to
24995 "vector bool int" or from "vector long long int" to "vector bool
24996 long long int" and changed second argument type from "unsigned
24997 int" to "int". Added new overloaded function forms "vector float
24998 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
24999 "vector float __builtin_vec_insert_exp_sp (vector float, vector
25000 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
25001 double, vector unsigned long long int)" and "vector double
25002 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
25003 long int)". Changed return type of
25004 __builtin_scalar_test_data_class and
25005 __builtin_scalar_test_data_class_sp and
25006 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
25007 int" and changed second argument from "unsigned int" to "int".
25008 Changed type returned by __builtin_scalar_test_neg,
25009 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
25010 from "int" to "bool int". Added new overloaded function form
25011 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
25012 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
25013 exponent double-precision with floating point first argument.
25014 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
25015 documentation of scalar_test_data_class, scalar_test_neg,
25016 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
25017 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
25018 vec_test_data_class built-in functions to reflect refinements in
25019 their type signatures.
25021 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
25023 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
25025 (aarch64_elf_asm_destructor): Likewise.
25027 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
25029 PR rtl-optimization/78634
25030 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
25031 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
25032 * ifcvt.c (noce_try_cmove): Add missing cost check.
25034 PR rtl-optimization/71724
25035 * combine.c (if_then_else_cond): Look for situations where it is
25036 beneficial to undo the work of one of the recursive calls.
25038 2017-01-23 Bin Cheng <bin.cheng@arm.com>
25040 PR tree-optimization/70754
25041 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
25042 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
25043 combined stmt before it if not NULL.
25044 (combine_chains): Process refs reversely and compute dominance point
25047 2017-01-23 Martin Liska <mliska@suse.cz>
25049 PR tree-optimization/79196
25050 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
25051 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
25053 (strlen_optimize_stmt): Call the renamed function.
25055 2017-01-23 Michael Matz <matz@suse.de>
25057 PR tree-optimization/78384
25058 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
25060 2017-01-23 Richard Biener <rguenther@suse.de>
25062 PR tree-optimization/79186
25063 * tree-vrp.c (register_new_assert_for): Make sure we've seen
25064 both incoming edges before moving an assert.
25066 2017-01-23 Martin Jambor <mjambor@suse.cz>
25068 * ipa-prop.c (load_from_param_1): Removed.
25069 (load_from_unmodified_param): Bits from load_from_param_1 put back
25071 (load_from_param): Removed.
25072 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
25073 with stmt. Reverted back to use of load_from_unmodified_param.
25075 2017-01-23 Martin Jambor <mjambor@suse.cz>
25078 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
25079 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
25080 field a pointer to garbage collected vector, mark lattices and
25081 ipcp_orig_node with GTY((skip)).
25082 (ipa_get_param_count): Adjust to descriptors being a pointer.
25083 (ipa_get_param): Likewise.
25084 (ipa_get_type): Likewise.
25085 (ipa_get_param_move_cost): Likewise.
25086 (ipa_set_param_used): Likewise.
25087 (ipa_get_controlled_uses): Likewise.
25088 (ipa_set_controlled_uses): Likewise.
25089 (ipa_is_param_used): Likewise.
25090 (ipa_node_params_t): Move into garbage collector. New methods insert
25092 (ipa_node_params_sum): Annotate wth GTY(()).
25093 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
25095 (ipa_load_from_parm_agg): Adjust declaration.
25096 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
25097 * ipa-profile.c (ipa_profile): Likewise.
25098 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
25099 (ipa_populate_param_decls): Make descriptors parameter garbage
25101 (ipa_dump_param): Adjust to descriptors being a pointer.
25102 (ipa_alloc_node_params): Likewise.
25103 (ipa_initialize_node_params): Likewise.
25104 (load_from_param_1): Make descriptors parameter garbage collected.
25105 (load_from_unmodified_param): Likewise.
25106 (load_from_param): Likewise.
25107 (ipa_load_from_parm_agg): Likewise.
25108 (ipa_node_params::~ipa_node_params): Removed.
25109 (ipa_free_all_node_params): Remove call to delete operator.
25110 (ipa_node_params_t::insert): New.
25111 (ipa_node_params_t::remove): Likewise.
25112 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
25113 copy known_csts and known_contexts vectors.
25114 (ipa_read_node_info): Adjust to descriptors being a pointer.
25115 (ipcp_modif_dom_walker): Make m_descriptors field garbage
25117 (ipcp_transform_function): Make descriptors variable garbage
25120 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
25122 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
25123 * config/i386/avx512dqintrin.h: Ditto.
25124 * config/i386/avx512fintrin.h: Ditto.
25125 * config/i386/i386.c: Handle new builtins.
25126 * config/i386/i386-builtin.def: Add new builtins.
25127 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
25128 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
25130 2017-01-23 Jakub Jelinek <jakub@redhat.com>
25131 Martin Liska <mliska@suse.cz>
25133 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
25134 * asan.c (asan_expand_poison_ifn): Support stores and use
25135 appropriate ASAN report function.
25136 * internal-fn.c (expand_ASAN_POISON_USE): New function.
25137 * internal-fn.def (ASAN_POISON_USE): Declare.
25138 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
25139 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
25140 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
25141 ASAN_POISON calls w/o LHS.
25142 * tree-ssa.c (execute_update_addresses_taken): Create clobber
25143 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
25144 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
25145 * gimplify.c (asan_poison_variables): Add attribute
25146 use_after_scope_memory to variables that really needs to live
25148 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
25149 having the attribute.
25151 2017-01-23 Martin Liska <mliska@suse.cz>
25153 * asan.c (create_asan_shadow_var): New function.
25154 (asan_expand_poison_ifn): Likewise.
25155 * asan.h (asan_expand_poison_ifn): New declaration.
25156 * internal-fn.c (expand_ASAN_POISON): Likewise.
25157 * internal-fn.def (ASAN_POISON): New builtin.
25158 * sanopt.c (pass_sanopt::execute): Expand
25159 asan_expand_poison_ifn.
25160 * tree-inline.c (copy_decl_for_dup_finish): Make function
25162 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
25163 * tree-ssa.c (is_asan_mark_p): New function.
25164 (execute_update_addresses_taken): Rewrite local variables
25165 (identified just by use-after-scope as addressable) into SSA.
25167 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
25169 * doc/install.texi (Specific): opensource.apple.com uses https
25170 now. Remove trailing slash.
25172 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
25174 * README.Portability: Remove note on an Irix compatibility issue.
25176 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
25178 * gcov.c (INCLUDE_ALGORITHM): Define.
25179 (INCLUDE_VECTOR): Define.
25180 No longer include <vector> and <algorithm> directly.
25182 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
25184 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
25186 * doc/invoke.texi (Code Gen Options): Ditto.
25188 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
25191 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
25193 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
25195 rtl-optimization/79125
25196 * cprop.c (local_cprop_pass): Handle cases where we make an
25197 unconditional trap.
25199 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
25203 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
25204 read from, for big endian.
25206 2017-01-20 Jiong Wang <jiong.wang@arm.com>
25208 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
25209 register pauth builtins for LP64 only.
25211 2017-01-20 Marek Polacek <polacek@redhat.com>
25214 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
25217 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
25219 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
25221 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
25222 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
25223 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
25225 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25228 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
25229 in big-endian mode when they are not a single duplicated value.
25231 2017-01-20 Richard Biener <rguenther@suse.de>
25233 * BASE-VER: Bump to 7.0.1.
25235 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
25237 * omp-low.c (omplow_simd_context): New struct. Use it...
25238 (lower_rec_simd_input_clauses): ...here and...
25239 (lower_rec_input_clauses): ...here to hold common data. Adjust all
25240 references to idx, lane, max_vf, is_simt.
25242 2017-01-20 Graham Markall <graham.markall@embecosm.com>
25244 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
25247 2017-01-20 Martin Jambor <mjambor@suse.cz>
25249 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
25250 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
25252 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
25253 (GTFILES): Rename hsa.c to hsa-common.c.
25254 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
25255 * hsa-dump.c: Likewise.
25256 * hsa-gen.c: Likewise.
25257 * hsa-regalloc.c: Likewise.
25258 * ipa-hsa.c: Likewise.
25259 * omp-expand.c: Likewise.
25260 * omp-low.c: Likewise.
25261 * toplev.c: Likewise.
25263 2017-01-20 Marek Polacek <polacek@redhat.com>
25266 * doc/invoke.texi: Document -Wduplicated-branches.
25267 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
25268 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
25269 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
25270 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
25271 return 0 only when not OEP_LEXICOGRAPHIC.
25272 (fold_build_cleanup_point_expr): Use the expression
25273 location when building CLEANUP_POINT_EXPR.
25274 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
25275 * tree.c (add_expr): Handle error_mark_node.
25277 2017-01-20 Martin Liska <mliska@suse.cz>
25280 * tree-profile.c (init_ic_make_global_vars): Do not call
25282 (gimple_init_gcov_profiler): Likewise.
25284 2017-01-20 Martin Liska <mliska@suse.cz>
25287 * cgraph.h (maybe_create_reference): Remove argument and
25289 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
25291 * ipa-cp.c (create_specialized_node): Likewise.
25292 * symtab.c (symtab_node::maybe_create_reference): Handle
25293 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
25295 2017-01-20 Martin Liska <mliska@suse.cz>
25297 * read-rtl-function.c (function_reader::create_function): Use
25298 build_decl instread of build_decl_stat.
25300 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
25302 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
25303 * config/i386/avx512dqintrin.h: Ditto.
25304 * config/i386/avx512fintrin.h: Ditto.
25305 * config/i386/i386-builtin-types.def: Add new types.
25306 * config/i386/i386.c: Handle new types.
25307 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
25308 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
25309 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
25310 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
25311 (__builtin_ia32_kshiftridi): New.
25312 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
25314 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
25318 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
25319 define to rs6000_init_stack_protect_guard.
25320 (rs6000_init_stack_protect_guard): New function.
25322 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
25323 Yunqiang Su <yunqiang.su@imgtec.com>
25325 * config.gcc (supported_defaults): Add madd4.
25326 (with_madd4): Add validation.
25327 (all_defaults): Add madd4.
25328 * config/mips/mips.opt (mmadd4): New option.
25329 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
25331 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
25333 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
25334 (ISA_HAS_FUSED_MADD4): Likewise.
25335 * doc/invoke.texi (-mmadd4): Document the new option.
25336 * doc/install.texi (--with-madd4): Document the new option.
25338 2017-01-19 Jiong Wang <jiong.wang@arm.com>
25340 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
25341 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
25342 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
25343 (aarch64_init_pauth_hint_builtins): New.
25344 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
25345 (aarch64_expand_builtin): Expand new builtins.
25347 2017-01-19 Jiong Wang <jiong.wang@arm.com>
25349 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
25350 * combine-stack-adj.c (no_unhandled_cfa): Handle
25351 REG_CFA_TOGGLE_RA_MANGLE.
25352 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
25353 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
25354 info for return address signing.
25355 (aarch64_expand_epilogue): Likewise.
25357 2017-01-19 Jiong Wang <jiong.wang@arm.com>
25359 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
25360 * config/aarch64/aarch64-protos.h
25361 (aarch64_return_address_signing_enabled): New declaration.
25362 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
25364 (aarch64_expand_prologue): Sign return address before it's pushed onto
25366 (aarch64_expand_epilogue): Authenticate return address fetched from
25368 (aarch64_override_options): Sanity check for ILP32 and ISA level.
25369 (aarch64_attributes): New function attributes for "sign-return-address".
25370 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
25371 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
25372 ("*do_return"): Generate combined instructions according to key index.
25373 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
25374 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
25376 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
25377 * config/aarch64/aarch64.opt (msign-return-address=): New.
25378 * doc/extend.texi (AArch64 Function Attributes): Documents
25379 "sign-return-address=".
25380 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
25382 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
25384 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
25385 overall option summary.
25387 2017-01-19 Jiong Wang <jiong.wang@arm.com>
25389 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
25390 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
25391 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
25392 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
25394 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
25396 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
25397 -mpower9-minmax by default for -mcpu=power9.
25398 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
25399 128-bit floating point.
25401 2017-01-20 Alan Modra <amodra@gmail.com>
25403 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
25404 optimizing for size.
25406 2017-01-20 Alan Modra <amodra@gmail.com>
25409 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
25410 for strcmp and strncmp from corresponding builtin decl.
25412 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
25414 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
25415 instead of i386/rtems-64.h.
25416 * config/i386/rtems-64.h: Remove.
25418 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
25422 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
25424 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
25426 2017-01-19 Tamar Christina <tamar.christina@arm.com>
25428 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
25429 Change int to HOST_WIDE_INT.
25430 * config/aarch64/aarch64-protos.h
25431 (aarch64_simd_gen_const_vector_dup): Likewise.
25432 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
25434 2017-01-19 David Malcolm <dmalcolm@redhat.com>
25436 * langhooks-def.h (lhd_type_for_size): New decl.
25437 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
25438 * langhooks.c (lhd_type_for_size): New function, taken from
25441 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
25443 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
25444 define_bypass for CR latency.
25445 (power9-cracked-alu): Update bypass latency and remove power9-branch.
25446 (power9-alu2): Add define_bypass for CR latency.
25448 (power9-mul): Update insn latency.
25449 (power9-mul-compare): Update insn latency, bypass latency and remove
25452 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25454 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
25456 * config/aarch64/aarch64.md
25457 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
25458 aarch64_nopcrelative_literal_loads.
25459 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
25461 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
25463 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
25464 TARGET_LOONGSON_3A.
25465 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
25467 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
25470 * config.gcc (supported_defaults): Add lxc1-sxc1.
25471 (with_lxc1_sxc1): Add validation.
25472 (all_defaults): Add lxc1-sxc1.
25473 * config/mips/mips.opt (mlxc1-sxc1): New option.
25474 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
25476 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
25477 __mips_no_lxc1_sxc1.
25478 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
25479 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
25480 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
25482 2017-01-19 Richard Biener <rguenther@suse.de>
25484 PR tree-optimization/72488
25485 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
25486 sure to restore SSA info.
25487 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
25489 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
25491 PR rtl-optimization/79121
25492 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
25493 of the inner type when shifting an extended value.
25495 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
25498 * symtab.c (symtab_node::equal_address_to): Fix comparing of
25499 interposable aliases.
25501 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
25504 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
25505 Use the evmergelohi instruction.
25506 (mov_si<mode>_e500_subreg4_2_le): Likewise.
25507 (mov_sitf_e500_subreg8_2_be): Likewise.
25508 (mov_sitf_e500_subreg12_2_le): Likewise.
25509 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
25510 (mov_si<mode>_e500_subreg4_2_be): Likewise.
25511 (mov_sitf_e500_subreg8_2_le): Likewise.
25512 (mov_sitf_e500_subreg12_2_be): Likewise.
25514 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25516 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
25517 attribute from vecsimple to vecperm.
25518 (altivec_vbpermq2): Likewise.
25520 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25523 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
25525 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25526 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
25527 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
25528 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
25529 case where N arg is SIZE_MAX.
25530 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
25531 (cmpstrsi): Add pattern.
25533 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
25535 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
25536 __builtin_vec_revb builtins.
25537 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
25538 built-in functions to support generation of the ISA 3.0 XXBR<x>
25539 vector byte reverse instructions.
25540 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
25541 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
25542 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
25543 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
25544 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
25545 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
25546 (P9V_BUILTIN_VEC_REVB): Likewise.
25547 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
25548 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
25549 (p9_xxbrq_v16qi): Likewise.
25550 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
25551 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
25552 (p9_xxbrh_v8hi): Likewise.
25553 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
25554 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
25555 vec_revb built-in functions.
25557 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
25559 PR rtl-optimization/78952
25560 * config/i386/i386.md (any_extract): New code iterator.
25561 (*insvqi_2): Use any_extract for source operand.
25562 (*insvqi_3): Use any_shiftrt for source operand.
25564 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
25566 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
25568 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
25570 2017-01-18 Matthias Klose <doko@ubuntu.com>
25572 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
25574 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25576 * config/rs6000/altivec.h (vec_bperm): Change #define.
25577 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
25578 (altivec_vbpermq2): New define_insn.
25579 (altivec_vbpermd): Likewise.
25580 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
25581 function interface.
25582 (VBPERMD): Likewise.
25583 (VBPERM): New polymorphic function interface.
25584 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
25585 Add entries for P9V_BUILTIN_VEC_VBPERM.
25586 * doc/extend.texi: Add interfaces for vec_bperm.
25588 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25590 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
25591 first letter of error messages.
25592 (s390_resolve_overloaded_builtin): Likewise.
25593 * config/s390/s390.c (s390_expand_builtin): Likewise.
25594 (s390_invalid_arg_for_unprototyped_fn): Likewise.
25595 (s390_valid_target_attribute_inner_p): Likewise.
25596 * config/s390/s390.md ("tabort"): Likewise.
25598 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
25600 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
25601 (ISA_AVOID_DIV_HILO): New macro.
25602 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
25603 (ISA_HAS_DDIV): Likewise.
25605 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
25607 * doc/invoke.texi (fabi-version): Correct number of occurrences.
25609 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
25611 * doc/invoke.texi (fabi-version): Spelling fix.
25613 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
25616 * doc/invoke.texi (fabi-version): Mention mangling fix for
25619 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
25622 * doc/invoke.texi (fabi-version): Document discriminator mangling.
25624 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
25627 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
25628 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
25630 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
25632 (stack_protect_test): Ditto.
25633 * config/rs6000/rs6000.opt (mstack-protector-guard=,
25634 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
25636 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
25637 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
25638 -mstack-protector-guard-offset=.
25639 (RS/6000 and PowerPC Options): Ditto.
25641 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
25643 * config/i386/i386.h (MASK_CLASS_P): New define.
25644 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
25645 there are no registers from different register sets also when
25646 mask registers are used. Update function comment.
25647 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
25648 to (*k/*r) and (*k/*km) alternatives.
25650 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
25652 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
25653 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
25654 (EH_RETURN_HANDLER_RTX): New define.
25655 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
25656 Force frame pointer in EH return functions.
25657 (aarch64_expand_epilogue): Add barrier for eh_return.
25658 (aarch64_final_eh_return_addr): Remove.
25659 (aarch64_eh_return_handler_rtx): New function.
25660 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
25662 (aarch64_eh_return_handler_rtx): New prototype.
25664 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25666 * config/rs6000/altivec.h (vec_rlmi): New #define.
25667 (vec_vrlnm): Likewise.
25668 (vec_rlnm): Likewise.
25669 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
25670 (UNSPEC_VRLNM): Likewise.
25671 (VIlong): New mode iterator.
25672 (altivec_vrl<VI_char>mi): New define_insn.
25673 (altivec_vrl<VI_char>nm): Likewise.
25674 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
25676 (VRLDNM): Likewise.
25677 (RLNM): New polymorphic function entry.
25678 (VRLWMI): New monomorphic function entry.
25679 (VRLDMI): Likewise.
25680 (RLMI): New polymorphic function entry.
25681 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
25682 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
25683 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
25686 2017-01-17 Jakub Jelinek <jakub@redhat.com>
25689 * dwarf2out.c (field_byte_offset): Restore the
25690 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
25691 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
25692 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
25695 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
25698 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
25700 2017-01-17 Jakub Jelinek <jakub@redhat.com>
25703 * dwarf2out.c (add_data_member_location_attribute): For constant
25704 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
25705 instead of DW_AT_data_member_location, DW_AT_bit_offset and
25706 DW_AT_byte_size attributes.
25708 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
25710 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
25711 after forcing to constant memory when the code model is medium.
25713 2017-01-17 Julia Koval <julia.koval@intel.com>
25716 * config/i386/avx512fintrin.h
25717 (_mm512_i32gather_ps): Change __addr type to void const*.
25718 (_mm512_mask_i32gather_ps): Ditto.
25719 (_mm512_i32gather_pd): Ditto.
25720 (_mm512_mask_i32gather_pd): Ditto.
25721 (_mm512_i64gather_ps): Ditto.
25722 (_mm512_mask_i64gather_ps): Ditto.
25723 (_mm512_i64gather_pd): Ditto.
25724 (_mm512_mask_i64gather_pd): Ditto.
25725 (_mm512_i32gather_epi32): Ditto.
25726 (_mm512_mask_i32gather_epi32): Ditto.
25727 (_mm512_i32gather_epi64): Ditto.
25728 (_mm512_mask_i32gather_epi64): Ditto.
25729 (_mm512_i64gather_epi32): Ditto.
25730 (_mm512_mask_i64gather_epi32): Ditto.
25731 (_mm512_i64gather_epi64): Ditto.
25732 (_mm512_mask_i64gather_epi64): Ditto.
25733 (_mm512_i32scatter_ps): Change __addr type to void*.
25734 (_mm512_mask_i32scatter_ps): Ditto.
25735 (_mm512_i32scatter_pd): Ditto.
25736 (_mm512_mask_i32scatter_pd): Ditto.
25737 (_mm512_i64scatter_ps): Ditto.
25738 (_mm512_mask_i64scatter_ps): Ditto.
25739 (_mm512_i64scatter_pd): Ditto.
25740 (_mm512_mask_i64scatter_pd): Ditto.
25741 (_mm512_i32scatter_epi32): Ditto.
25742 (_mm512_mask_i32scatter_epi32): Ditto.
25743 (_mm512_i32scatter_epi64): Ditto.
25744 (_mm512_mask_i32scatter_epi64): Ditto.
25745 (_mm512_i64scatter_epi32): Ditto.
25746 (_mm512_mask_i64scatter_epi32): Ditto.
25747 (_mm512_i64scatter_epi64): Ditto.
25748 (_mm512_mask_i64scatter_epi64): Ditto.
25749 * config/i386/avx512pfintrin.h
25750 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
25751 (_mm512_mask_prefetch_i32gather_ps): Ditto.
25752 (_mm512_mask_prefetch_i64gather_pd): Ditto.
25753 (_mm512_mask_prefetch_i64gather_ps): Ditto.
25754 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
25755 (_mm512_prefetch_i32scatter_ps): Ditto.
25756 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
25757 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
25758 (_mm512_prefetch_i64scatter_pd): Ditto.
25759 (_mm512_prefetch_i64scatter_ps): Ditto.
25760 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
25761 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
25762 * config/i386/avx512vlintrin.h
25763 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
25764 (_mm_mmask_i32gather_ps): Ditto.
25765 (_mm256_mmask_i32gather_pd): Ditto.
25766 (_mm_mmask_i32gather_pd): Ditto.
25767 (_mm256_mmask_i64gather_ps): Ditto.
25768 (_mm_mmask_i64gather_ps): Ditto.
25769 (_mm256_mmask_i64gather_pd): Ditto.
25770 (_mm_mmask_i64gather_pd): Ditto.
25771 (_mm256_mmask_i32gather_epi32): Ditto.
25772 (_mm_mmask_i32gather_epi32): Ditto.
25773 (_mm256_mmask_i32gather_epi64): Ditto.
25774 (_mm_mmask_i32gather_epi64): Ditto.
25775 (_mm256_mmask_i64gather_epi32): Ditto.
25776 (_mm_mmask_i64gather_epi32): Ditto.
25777 (_mm256_mmask_i64gather_epi64): Ditto.
25778 (_mm_mmask_i64gather_epi64): Ditto.
25779 (_mm256_i32scatter_ps): Change __addr type to void*.
25780 (_mm256_mask_i32scatter_ps): Ditto.
25781 (_mm_i32scatter_ps): Ditto.
25782 (_mm_mask_i32scatter_ps): Ditto.
25783 (_mm256_i32scatter_pd): Ditto.
25784 (_mm256_mask_i32scatter_pd): Ditto.
25785 (_mm_i32scatter_pd): Ditto.
25786 (_mm_mask_i32scatter_pd): Ditto.
25787 (_mm256_i64scatter_ps): Ditto.
25788 (_mm256_mask_i64scatter_ps): Ditto.
25789 (_mm_i64scatter_ps): Ditto.
25790 (_mm_mask_i64scatter_ps): Ditto.
25791 (_mm256_i64scatter_pd): Ditto.
25792 (_mm256_mask_i64scatter_pd): Ditto.
25793 (_mm_i64scatter_pd): Ditto.
25794 (_mm_mask_i64scatter_pd): Ditto.
25795 (_mm256_i32scatter_epi32): Ditto.
25796 (_mm256_mask_i32scatter_epi32): Ditto.
25797 (_mm_i32scatter_epi32): Ditto.
25798 (_mm_mask_i32scatter_epi32): Ditto.
25799 (_mm256_i32scatter_epi64): Ditto.
25800 (_mm256_mask_i32scatter_epi64): Ditto.
25801 (_mm_i32scatter_epi64): Ditto.
25802 (_mm_mask_i32scatter_epi64): Ditto.
25803 (_mm256_i64scatter_epi32): Ditto.
25804 (_mm256_mask_i64scatter_epi32): Ditto.
25805 (_mm_i64scatter_epi32): Ditto.
25806 (_mm_mask_i64scatter_epi32): Ditto.
25807 (_mm256_i64scatter_epi64): Ditto.
25808 (_mm256_mask_i64scatter_epi64): Ditto.
25809 (_mm_i64scatter_epi64): Ditto.
25810 (_mm_mask_i64scatter_epi64): Ditto.
25811 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
25812 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
25813 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
25814 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
25815 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
25816 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
25817 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
25818 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
25819 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
25820 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
25821 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
25822 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
25823 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
25824 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
25825 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
25826 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
25827 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
25828 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
25829 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
25830 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
25831 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
25832 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
25833 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
25834 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
25835 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
25836 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
25837 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
25838 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
25839 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
25840 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
25841 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
25842 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
25843 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
25844 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
25845 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
25846 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
25847 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
25848 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
25849 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
25850 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
25851 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
25852 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
25853 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
25854 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
25855 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
25856 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
25857 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
25858 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
25859 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
25860 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
25861 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
25862 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
25863 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
25864 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
25865 definitions accordingly.
25867 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
25868 Kuan-Lin Chen <kuanlinchentw@gmail.com>
25871 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
25874 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
25877 * ira-conflicts.c (ira_build_conflicts): Update total conflict
25878 hard regs for inner regno.
25880 2017-01-17 Martin Liska <mliska@suse.cz>
25883 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
25884 assumption and add comment.
25886 2017-01-17 Nathan Sidwell <nathan@acm.org>
25888 * ipa-visibility.c (localize_node): New function, broken out of ...
25889 (function_and_variable_visibility): ... here. Call it.
25891 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
25893 PR middle-end/77445
25894 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
25895 correctly set frequency of oudgoing edge.
25896 (duplicate_thread_path): Fix profile updating.
25898 2017-01-17 Jakub Jelinek <jakub@redhat.com>
25901 * configure.ac: Add GCC_BASE_VER.
25902 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
25903 version from BASE-VER file.
25904 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
25905 (gcc.o): Depend on $(BASEVER).
25906 * common.opt (dumpfullversion): New option.
25907 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
25908 * doc/invoke.texi: Document -dumpfullversion.
25909 * doc/install.texi: Document --with-gcc-major-version-only.
25910 * configure: Regenerated.
25912 2017-01-17 Richard Biener <rguenther@suse.de>
25914 PR tree-optimization/71433
25915 * tree-vrp.c (register_new_assert_for): Merge same asserts
25916 on all incoming edges.
25917 (process_assert_insertions_for): Handle insertions at the
25920 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
25922 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
25923 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
25925 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
25928 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
25931 2017-01-17 Alan Modra <amodra@gmail.com>
25934 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
25935 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
25936 symbolic stack limit when pic.
25938 2017-01-16 Martin Sebor <msebor@redhat.com>
25940 PR tree-optimization/78608
25941 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
25943 2017-01-16 Jeff Law <law@redhat.com>
25946 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
25947 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
25948 for several include directories that may be relative to sysroot.
25949 * config/i386/x-mingw32 (gplus_includedir): Define.
25950 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
25951 (native_system_includedir): Likewise.
25952 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
25953 override if TARGET_SYSTEM_ROOT is defined.
25954 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
25956 PR tree-optimization/79090
25957 PR tree-optimization/33562
25958 PR tree-optimization/61912
25959 PR tree-optimization/77485
25960 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
25961 and computed trims into the dump file.
25963 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
25965 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
25967 2017-01-16 Jakub Jelinek <jakub@redhat.com>
25970 * gimplify.c (gimplify_init_constructor): If want_value and
25971 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
25975 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
25976 sequence. Formatting fixes.
25977 (doloop_optimize): Formatting fixes.
25980 * gcc.c (debug_level_greater_than_spec_func): New function.
25981 (static_spec_functions): Add debug-level-gt spec function.
25982 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
25984 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
25985 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
25986 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
25987 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
25988 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
25989 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
25991 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
25993 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
25994 QImode fixups to general and mask registers only.
25996 2017-01-16 Carl Love <cel@us.ibm.com>
25998 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
25999 for built-in functions
26000 vector signed char vec_nabs (vector signed char)
26001 vector signed short vec_nabs (vector signed short)
26002 vector signed int vec_nabs (vector signed int)
26003 vector signed long long vec_nabs (vector signed long long)
26004 vector float vec_nabs (vector float)
26005 vector double vec_nabs (vector double)
26006 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
26008 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
26009 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
26010 * doc/extend.texi: Update the documentation file for the new built-in
26013 2017-01-16 Martin Sebor <msebor@redhat.com>
26015 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
26018 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26020 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
26021 UNSPEC_VSX__XXSPLTD to require special splat handling.
26023 2017-01-16 David Malcolm <dmalcolm@redhat.com>
26026 * system.h: Poison strndup.
26028 2017-01-16 Alan Modra <amodra@gmail.com>
26031 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
26034 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
26036 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
26038 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
26040 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
26041 call recog here. Assert that INSN_CODE (insn) is non-negative.
26043 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
26046 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
26048 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
26049 in the currently scheduled RTL fragment.
26051 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
26053 PR rtl-optimization/78751
26054 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
26057 2017-01-14 Jeff Law <law@redhat.com>
26059 PR tree-optimization/79090
26060 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
26061 variable length stores.
26062 (compute_trims): Delete dead assignment to *trim_tail.
26063 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
26066 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
26068 PR rtl-optimization/78626
26069 PR rtl-optimization/78727
26070 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
26071 of a block, and split such blocks after everything else is finished.
26073 2017-01-14 Alan Modra <amodra@gmail.com>
26076 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
26077 target legitimate_combined_insn.
26078 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
26079 (rs6000_legitimate_combined_insn): New function.
26080 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
26082 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
26083 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
26084 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
26086 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
26088 * doc/frontends.texi (G++ and GCC): Remove references to Java.
26090 2017-01-13 Jeff Law <law@redhat.com>
26092 PR tree-optimization/33562
26093 PR tree-optimization/61912
26094 PR tree-optimization/77485
26095 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
26097 (delete_dead_assignment): Likewise.
26098 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
26099 statement to delete_dead_call and delete_dead_assignment.
26101 2017-01-13 David Malcolm <dmalcolm@redhat.com>
26104 * substring-locations.c (format_warning_va): Strengthen case 1 so
26105 that both endpoints of the substring must be within the format
26106 range for just the substring to be printed.
26108 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
26110 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
26111 * config/i386/i386.c (ix86_target_string): Add missing options
26112 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
26113 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
26114 flags_other and ix86_target_other to flags2_other. Display unknown
26116 (ix86_valid_target_attribute_inner_p): Add missing options and
26117 reorder options by implied ISAs, as in ix86_target_string.
26119 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
26121 * hash-table.h (hash_table::too_empty_p): New function.
26122 (hash_table::expand): Use it.
26123 (hash_table::traverse): Likewise.
26124 (hash_table::empty_slot): Use sizeof (value_type) instead of
26125 sizeof (PTR) to convert bytes to elements. Shrink the table
26126 if the current size is excessive for the current number of
26129 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
26131 * ira-costs.c (record_reg_classes): Break from the inner loop
26132 early once alt_fail is known to be true. Update outer loop
26133 handling accordingly.
26135 2017-01-13 Jeff Law <law@redhat.com>
26137 * tree-ssa-dse.c (decrement_count): New function.
26138 (increment_start_addr, maybe_trim_memstar_call): Likewise.
26139 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
26140 when we know the partially dead statement is a mem* function.
26142 PR tree-optimization/61912
26143 PR tree-optimization/77485
26144 * tree-ssa-dse.c: Include expr.h.
26145 (maybe_trim_constructor_store): New function.
26146 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
26148 PR tree-optimization/33562
26149 PR tree-optimization/61912
26150 PR tree-optimization/77485
26151 * doc/invoke.texi: Document new dse-max-object-size param.
26152 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
26153 * tree-ssa-dse.c: Include params.h.
26154 (dse_store_status): New enum.
26155 (initialize_ao_ref_for_dse): New, partially extracted from
26157 (valid_ao_ref_for_dse, normalize_ref): New.
26158 (setup_live_bytes_from_ref, compute_trims): Likewise.
26159 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
26160 (maybe_trim_partially_dead_store): Likewise.
26161 (maybe_trim_complex_store): Likewise.
26162 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
26163 Track what bytes live from the original store. Return tri-state
26164 for dead, partially dead or live.
26165 (dse_dom_walker): Add constructor, destructor and new private members.
26166 (delete_dead_call, delete_dead_assignment): New extracted from
26168 (dse_optimize_stmt): Make a member of dse_dom_walker.
26169 Use initialize_ao_ref_for_dse.
26171 PR tree-optimization/33562
26172 PR tree-optimization/61912
26173 PR tree-optimization/77485
26174 * sbitmap.h (bitmap_count_bits): Prototype.
26175 (bitmap_clear_range, bitmap_set_range): Likewise.
26176 * sbitmap.c (bitmap_clear_range): New function.
26177 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
26179 2017-01-13 Martin Liska <mliska@suse.cz>
26182 * function.c (set_cfun): Add new argument force.
26183 * function.h (set_cfun): Likewise.
26184 * ipa-inline-transform.c (inline_call): Use the function when
26185 strict alising from is dropped for function we inline to.
26187 2017-01-13 Richard Biener <rguenther@suse.de>
26189 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
26190 for dumping GIMPLE INTEGER_CSTs.
26192 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26194 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
26195 to 201112L since C++17.
26197 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
26200 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
26201 if -fsanitize=kernel-address is present.
26203 2017-01-13 Richard Biener <rguenther@suse.de>
26205 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
26206 as _Literal ( type ) number in case usual suffixes do not
26207 preserve all information.
26209 2017-01-13 Richard Biener <rguenther@suse.de>
26211 PR tree-optimization/77283
26212 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
26213 and ssa-iterators.h.
26214 (is_feasible_trace): Implement a cost model based on joiner
26217 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
26220 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
26221 char or short to __float128/_Float128 directly.
26223 2017-01-12 Martin Sebor <msebor@redhat.com>
26225 to -Wformat-overflow.
26226 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
26227 (min_bytes_remaining): Same.
26228 (get_string_length): Same.
26229 (format_string): Same.
26230 (format_directive): Same.
26232 (pass_sprintf_length::handle_gimple_call): Same.
26234 2017-01-12 Jakub Jelinek <jakub@redhat.com>
26236 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
26237 info.nowrite calls with no lhs that can't throw. Return bool
26238 whether gsi_remove has been called or not.
26239 (pass_sprintf_length::handle_gimple_call): Return bool whether
26240 try_substitute_return_value called gsi_remove. Formatting fix.
26241 (pass_sprintf_length::execute): Don't use gsi_remove if
26242 handle_gimple_call returned true.
26245 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
26246 be removed due to side-effects, don't remove following barrier nor
26247 turn the successor edge into fallthru edge.
26249 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26252 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
26253 element-reversing loads and stores as not swappable.
26255 2017-01-12 Nathan Sidwell <nathan@acm.org>
26256 Nicolai Stange <nicstange@gmail.com>
26258 * combine.c (try_combine): Don't ignore result of overlap checking
26259 loop. Combine overlap & asm check into single loop.
26261 2017-01-12 Richard Biener <rguenther@suse.de>
26263 * tree-pretty-print.c (dump_generic_node): Provide -gimple
26264 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
26266 2017-01-12 Richard Biener <rguenther@suse.de>
26268 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
26269 and TS_TARGET_OPTION directly derive from TS_BASE.
26270 * tree-core.h (tree_optimization_option): Derive from tree_base.
26271 (tree_target_option): Likewise.
26273 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
26275 * config/i386/i386.c (memory_address_length): Increase len
26276 only when rip_relative_addr_p returns false.
26278 2017-01-11 Julia Koval <julia.koval@intel.com>
26280 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
26281 (OPTION_MASK_ISA_SGX_SET): New.
26282 (ix86_handle_option): Handle OPT_msgx.
26283 * config.gcc: Added sgxintrin.h.
26284 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
26285 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
26286 * config/i386/i386.c (ix86_target_string): Add -msgx.
26288 (ix86_option_override_internal): Handle new options.
26289 (ix86_valid_target_attribute_inner_p): Add sgx.
26290 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
26291 * config/i386/i386.opt: Add msgx.
26292 * config/i386/sgxintrin.h: New file.
26293 * config/i386/x86intrin.h: Add sgxintrin.h.
26295 2017-01-11 Jakub Jelinek <jakub@redhat.com>
26298 * fold-const.c (maybe_nonzero_address): Return 1 for function
26300 (tree_single_nonzero_warnv_p): Don't handle function local objects
26304 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
26307 2017-01-11 David Malcolm <dmalcolm@redhat.com>
26310 * opts.c: Include "spellcheck.h"
26311 (struct string_fragment): New struct.
26312 (struct edit_distance_traits<const string_fragment &>): New
26314 (get_closest_sanitizer_option): New function.
26315 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
26317 2017-01-11 Jakub Jelinek <jakub@redhat.com>
26319 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
26321 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
26322 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
26323 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
26324 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
26325 for initial die_offset if dwarf_split_debug_info.
26326 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
26327 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
26329 (output_skeleton_debug_sections): Formatting fix. Use
26330 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
26331 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
26333 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
26335 * config/arm/cortex-a53.md: Add bypasses for
26336 cortex_a53_r2f_cvt.
26337 (cortex_a53_r2f): Only use for transfers.
26338 (cortex_a53_f2r): Likewise.
26339 (cortex_a53_r2f_cvt): Add reservation for conversions.
26340 (cortex_a53_f2r_cvt): Likewise.
26342 2017-01-11 Tamar Christina <tamar.christina@arm.com>
26344 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
26345 to all inlined functions, change static to extern.
26347 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
26350 * config/arm/arm.c (legitimize_pic_address): Handle reference to
26352 (arm_assemble_integer): Likewise.
26354 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
26356 * config.gcc: Use new awk script to check CPU, FPU and architecture
26357 parameters for --with-... options.
26358 * config/arm/parsecpu.awk: New file
26359 * config/arm/arm-cpus.in: New file.
26360 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
26362 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
26364 * config/arm/t-arm: Update dependency rules.
26365 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
26366 of processing .def files.
26367 * config/arm/genopt.sh: Deleted.
26368 * config/arm/gentune.sh: Deleted.
26369 * config/arm/arm-cores.def: Deleted.
26370 * config/arm/arm-arches.def: Deleted.
26371 * config/arm/arm-fpus.def: Deleted.
26372 * config/arm/arm-tune.md: Regenerated.
26373 * config/arm/arm-tables.opt: Regenerated.
26374 * config/arm/arm-cpu.h: New generated file.
26375 * config/arm/arm-cpu-data.h: New generated file.
26376 * config/arm/arm-cpu-cdata.h: New generated file.
26378 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
26381 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
26383 (input_varpool_node): Unpack dynamically_initialized bit.
26385 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
26387 PR rtl-optimization/79032
26388 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
26389 the alignment of the adjusted memory reference against that of MODE,
26390 instead of the alignment of the original memory reference.
26392 2017-01-11 Martin Jambor <mjambor@suse.cz>
26394 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
26396 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
26397 decorated functions.
26399 2017-01-11 Richard Biener <rguenther@suse.de>
26401 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
26402 set range/nonnull info for PHI results. Do not set it on
26403 stmts marked for removal.
26405 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
26407 * expr.c (store_field): In the bitfield case, fetch the return value
26408 from the registers before applying a single big-endian adjustment.
26409 Always do a final load for a BLKmode value not larger than a word.
26411 2017-01-10 David Malcolm <dmalcolm@redhat.com>
26414 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
26415 that we correctly handle column numbers greater than
26416 LINE_MAP_MAX_COLUMN_NUMBER.
26418 2017-01-10 Martin Sebor <msebor@redhat.com>
26420 PR middle-end/78245
26421 * gimple-ssa-sprintf.c (get_destination_size): Call
26422 {init,fini}object_sizes.
26423 * tree-object-size.c (addr_object_size): Adjust.
26424 (pass_through_call): Adjust.
26425 (pass_object_sizes::execute): Adjust.
26426 * tree-object-size.h (fini_object_sizes): Declare.
26428 2017-01-10 Martin Sebor <msebor@redhat.com>
26430 PR tree-optimization/78775
26431 * builtins.c (get_size_range): Move...
26432 * calls.c: ...to here.
26433 (alloc_max_size): Accept zero argument.
26434 (operand_signed_p): Remove.
26435 (maybe_warn_alloc_args_overflow): Call get_size_range.
26436 * calls.h (get_size_range): Declare.
26438 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
26440 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
26441 from TI's devices.csv file as of September 2016.
26442 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
26444 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
26446 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
26447 * doc/invoke.texi: Likewise.
26448 * doc/md.texi: Likewise.
26449 * doc/objc.texi: Likewise.
26451 2017-01-10 Joshua Conner <joshconner@google.com>
26453 * config/arm/fuchsia-elf.h: New file.
26454 * config/fuchsia.h: New file.
26455 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
26456 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
26458 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
26460 2016-01-10 Richard Biener <rguenther@suse.de>
26462 PR tree-optimization/79034
26463 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
26464 Propagate out degenerate PHIs in the joiner.
26466 2017-01-10 Martin Liska <mliska@suse.cz>
26468 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
26469 (sort_congruence_classes_by_decl_uid): Likewise.
26470 (sort_congruence_class_groups_by_decl_uid): Likewise.
26471 (sem_item_optimizer::merge_classes): Sort class, groups in these
26472 classes and members in the groups by DECL_UID of declarations.
26473 This would make merge operations stable.
26475 2017-01-10 Martin Liska <mliska@suse.cz>
26477 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
26478 usage of m_classes_vec.
26479 (sem_item_optimizer::~sem_item_optimizer): Likewise.
26480 (sem_item_optimizer::get_group_by_hash): Likewise.
26481 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
26482 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
26483 (sem_item_optimizer::verify_classes): Likewise.
26484 (sem_item_optimizer::process_cong_reduction): Likewise.
26485 (sem_item_optimizer::dump_cong_classes): Likewise.
26486 (sem_item_optimizer::merge_classes): Likewise.
26487 * ipa-icf.h (congruence_class_hash): Rename from
26488 congruence_class_group_hash. Remove declaration of m_classes_vec.
26490 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
26492 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
26493 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
26494 * config.gcc: Add avx512vpopcntdqintrin.h.
26495 * config/i386/avx512vpopcntdqintrin.h: New.
26496 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
26497 * config/i386/i386-builtin-types.def: Add new types.
26498 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
26499 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
26500 __builtin_ia32_vpopcountq_v8di_mask): New.
26501 * config/i386/i386-c.c (ix86_target_macros_internal): Define
26502 __AVX512VPOPCNTDQ__.
26503 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
26504 (PTA_AVX512VPOPCNTDQ): Define.
26505 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
26506 TARGET_AVX512VPOPCNTDQ_P): Define.
26507 * config/i386/i386.opt: Add mavx512vpopcntdq.
26508 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
26509 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
26511 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
26513 PR middle-end/77484
26514 * predict.def (PRED_CALL): Set to 67.
26516 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
26518 * expr.c (store_field): In the bitfield case, if the value comes from
26519 a function call and is of an aggregate type returned in registers, do
26520 not modify the field mode; extract the value in all cases if the mode
26521 is BLKmode and the size is not larger than a word.
26523 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
26526 * config/i386/cpuid.h: Fix undefined behavior.
26528 2017-01-04 Jeff Law <law@redhat.com>
26530 PR tree-optimization/79007
26531 PR tree-optimization/67955
26532 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
26533 conservative for pt.null when flag_non_call_exceptions is on.
26535 2017-01-09 Jakub Jelinek <jakub@redhat.com>
26537 PR translation/79019
26538 PR translation/79020
26539 * params.def (PARAM_INLINE_MIN_SPEEDUP,
26540 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
26541 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
26543 * config/avr/avr.opt (maccumulate-args): Likewise.
26544 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
26545 * common.opt (freport-bug): Likewise.
26546 * cif-code.def (CIF_FINAL_ERROR): Likewise.
26547 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
26548 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
26549 translatable string.
26550 * config/i386/i386.c (function_value_32): Likewise.
26551 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
26552 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
26554 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
26555 * common/config/msp430/msp430-common.c (msp430_handle_option):
26557 * symtab.c (symtab_node::verify_base): Likewise.
26558 * opts.c (set_debug_level): Likewise.
26559 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
26560 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
26561 missing whitespace to translatable strings.
26562 * config/avr/avr.md (bswapsi2): Fix typo in comment.
26563 * config/sh/superh.h: Likewise.
26564 * config/i386/xopintrin.h: Likewise.
26565 * config/i386/znver1.md: Likewise.
26566 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
26567 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
26568 * double-int.h (struct double_int): Likewise.
26569 * double-int.c (div_and_round_double): Likewise.
26570 * wide-int.cc: Likewise.
26571 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
26572 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
26573 * cfgcleanup.c (crossjumps_occured): Renamed to ...
26574 (crossjumps_occurred): ... this.
26575 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
26578 PR tree-optimization/78899
26579 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
26580 returning bool return struct loop *, NULL for failure and the new
26582 (versionable_outer_loop_p): Don't version outer loop if it has
26583 dont_vectorized bit set.
26584 (tree_if_conversion): When versioning outer loop, ensure
26585 tree_if_conversion is performed also on the inner loop of the
26586 non-vectorizable outer loop copy.
26587 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
26588 LOOP_VECTORIZED in inner loop of the scalar outer loop and
26589 prevent vectorization of it.
26590 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
26591 the outer loop vectorization of the non-scalar version is attempted
26592 before vectorization of the inner loop in scalar version. If
26593 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
26594 vectorization of its inner loop.
26595 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
26596 has 2 inner loops, rename also on edges from bb whose single pred
26597 is outer_loop->header. Fix typo in function comment.
26599 2017-01-09 Martin Sebor <msebor@redhat.com>
26602 * asan.c (asan_emit_stack_protection): Increase local buffer size
26603 to avoid snprintf truncation warning.
26605 2017-01-09 Andrew Pinski <apinski@cavium.com>
26607 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
26608 to reference thunderx2t99 for the tuning structure
26609 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
26611 (thunderx2t99_extra_costs): This.
26612 * config/aarch64/aarch64-tune.md: Regenerate.
26613 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
26614 (vulcan_addrcost_table): This.
26615 (vulcan_regmove_cost): Rename to ...
26616 (thunderx2t99_regmove_cost): This.
26617 (vulcan_vector_cost): Rename to ...
26618 (thunderx2t99_vector_cost): this.
26619 (vulcan_branch_cost): Rename to ...
26620 (thunderx2t99_branch_cost): This.
26621 (vulcan_tunings): Rename to ...
26622 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
26623 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
26625 2017-01-09 Martin Jambor <mjambor@suse.cz>
26629 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
26630 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
26631 (propagate_vr_accross_jump_function): Use the above function for all
26632 value range computations for pass-through jump functions and type
26633 converasion from explicit value range values.
26634 (ipcp_propagate_stage): Do not attempt to deduce types of formal
26635 parameters from TYPE_ARG_TYPES.
26636 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
26637 (ipa_write_node_info): Stream type of the actual argument.
26638 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
26640 2017-01-09 Martin Liska <mliska@suse.cz>
26643 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
26644 (lookup_compiler): Do not show error message with have_E.
26646 2017-01-09 Jakub Jelinek <jakub@redhat.com>
26648 PR tree-optimization/78938
26649 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
26650 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
26651 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
26652 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
26655 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26657 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
26660 2017-01-09 Richard Biener <rguenther@suse.de>
26662 PR tree-optimization/78997
26663 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
26664 name condition properly.
26666 2017-01-09 Richard Biener <rguenther@suse.de>
26669 * dwarf2out.c (is_cxx): New overload with context.
26670 (is_naming_typedef_decl): Use it.
26672 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
26674 * invoke.texi (Option Summary): Correct spacing in option lists
26675 and add line breaks to fix over-long lines.
26677 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
26679 PR middle-end/17660
26681 * extend.texi (Common Variable Attributes): Add xref to GCC
26682 Internals manual to explain mode attribute keywords.
26684 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
26687 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
26688 and Preprocessor Options.
26689 (Options for Linking): Document -pthread here....
26690 (RS/6000 and PowerPC Options): ...not here.
26691 (Solaris 2 Options): ...or here.
26692 * doc/cppopts.texi: Document -pthread.
26694 2017-01-08 Martin Sebor <msebor@redhat.com>
26696 PR middle-end/77708
26697 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
26698 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
26699 New member functions.
26700 (format_directive): Used them.
26702 (pass_sprintf_length::handle_gimple_call): Same.
26703 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
26704 to avoid truncation for any argument.
26705 (extract_affine_mul): Same.
26706 * tree.c (get_file_function_name): Same.
26708 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
26710 PR middle-end/77484
26711 * predict.def (PRED_INDIR_CALL): Set to 86.
26713 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
26715 PR preprocessor/54124
26716 * doc/cppopts.texi: Reformat -d subtable to list the full name
26717 of the options. Add cross-reference to the docs for the general
26718 compiler -d options.
26719 * doc/invoke.texi (Developer Options): Add cross-reference to the
26720 preprocessor-specific -d option documentation.
26722 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
26724 PR preprocessor/13498
26725 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
26726 redudant material, and reflect new command-line options.
26727 (System Headers): Likewise.
26729 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
26731 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
26732 -isystem, and -idirafter. Copy-edit.
26733 * doc/cppopts.texi: Copy-edit. Remove contradiction about
26734 default for -ftrack-macro-expansion. Delete obsolete and
26735 badly-formatted implementation details about -fdebug-cpp output.
26736 * doc/cppwarnopts.texi: Copy-edit.
26738 2017-01-07 David Malcolm <dmalcolm@redhat.com>
26741 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
26742 that the transition from a max line width >= 1<<10 to narrower
26743 lines works correctly.
26745 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
26747 * doc/options.texi (PerFunction): New.
26748 * opt-functions.awk (switch_flags): Map both Optimization and
26749 PerFunction to CL_OPTIMIZATION.
26750 * opth-gen.awk: Test for PerFunction flag along with
26752 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
26753 it only when the latter is present. Skip those that don't in
26754 the hash function generator.
26755 * common.opt (fvar-tracking): Mark as PerFunction instead of
26757 (fvar-tracking-assignments): Likewise.
26758 (fvar-tracking-assignments-toggle): Likewise.
26759 (fvar-tracking-uninit): Likewise.
26761 2017-01-07 Jakub Jelinek <jakub@redhat.com>
26763 PR translation/79018
26764 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
26767 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
26770 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
26771 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
26772 TARGET_LONG_JUMP_TABLE_OFFSETS.
26773 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
26774 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
26775 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
26776 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
26777 * config/m68k/m68k.md (tablejump expander): Likewise.
26778 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
26779 TARGET_LONG_JUMP_TABLE_OFFSETS.
26780 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
26781 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
26783 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
26784 David Holsgrove <david.holsgrove@xilinx.com>
26786 * common/config/microblaze/microblaze-common.c
26787 (TARGET_EXCEPT_UNWIND_INFO): Remove.
26788 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
26790 * config/microblaze/microblaze.c (microblaze_must_save_register)
26791 (microblaze_expand_epilogue, microblaze_return_addr): Handle
26793 (microblaze_eh_return): New function.
26794 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
26795 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
26796 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
26797 * config/microblaze/microblaze.md (eh_return): New pattern.
26799 2017-01-06 Jakub Jelinek <jakub@redhat.com>
26801 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
26802 GCC_DIAGNOSTIC_STRINGIFY): Define.
26804 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
26806 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26808 * config/arm/arm.md (<mcrr>): New.
26810 * config/arm/arm.c (arm_arch5te): New.
26811 (arm_option_override): Set arm_arch5te.
26812 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
26814 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
26815 (arm_mcrr_qualifiers): ... this. New.
26816 (MRRC_QUALIFIERS): Define to...
26817 (arm_mrrc_qualifiers): ... this. New.
26818 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
26820 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
26821 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
26822 (MRRCI, mrrc, MRRC): New.
26823 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
26824 VUNSPEC_MRRC2): New.
26826 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26828 * config/arm/arm.md (<mcr>): New.
26830 * config/arm/arm.c (arm_coproc_builtin_available): Add
26831 support for mcr, mrc, mcr2 and mrc2.
26832 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
26833 (arm_mcr_qualifiers): ... this. New.
26834 (MRC_QUALIFIERS): Define to ...
26835 (arm_mrc_qualifiers): ... this. New.
26836 (MCR_QUALIFIERS): Define to ...
26837 (arm_mcr_qualifiers): ... this. New.
26838 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
26840 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
26841 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
26842 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
26843 VUNSPEC_MRC2): New.
26845 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26847 * config/arm/arm.md (*ldc): New.
26851 * config/arm/arm.c (arm_coproc_builtin_available): Add
26852 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
26853 (arm_coproc_ldc_stc_legitimate_address): New.
26854 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
26855 'qualifier_const_pointer'.
26856 (LDC_QUALIFIERS): Define to...
26857 (arm_ldc_qualifiers): ... this. New.
26858 (STC_QUALIFIERS): Define to...
26859 (arm_stc_qualifiers): ... this. New.
26860 * config/arm/arm-protos.h
26861 (arm_coproc_ldc_stc_legitimate_address): New.
26862 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
26863 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
26864 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
26865 stc2, stcl, stc2l): New.
26866 * config/arm/constraints.md (Uz): New.
26867 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
26868 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
26869 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
26870 VUNSPEC_STC2L): New.
26872 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26874 * config/arm/arm.md (<cdp>): New.
26875 * config/arm/arm.c (neon_const_bounds): Rename this ...
26876 (arm_const_bounds): ... this.
26877 (arm_coproc_builtin_available): New.
26878 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
26879 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
26880 (CDP_QUALIFIERS): Define to...
26881 (arm_cdp_qualifiers): ... this. New.
26883 (arm_expand_builtin_args): Add case for 6 arguments.
26884 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
26885 (arm_const_bounds): ... this.
26886 (arm_coproc_builtin_available): New.
26887 * config/arm/arm_acle.h (__arm_cdp): New.
26889 * config/arm/arm_acle_builtins.def (cdp): New.
26891 * config/arm/iterators.md (CDPI,CDP,cdp): New.
26892 * config/arm/neon.md: Rename all 'neon_const_bounds' to
26893 'arm_const_bounds'.
26894 * config/arm/types.md (coproc): New.
26895 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
26896 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
26897 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
26898 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
26900 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26902 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
26903 (UBINOP_QUALIFIERS): New.
26905 (acle_builtin_data): New. Change comment.
26906 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
26907 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
26908 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
26909 arm_acle_builtins.def.
26910 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
26911 (arm_init_acle_builtins): New.
26912 (CRC32_BUILTIN): Remove.
26913 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
26914 crc32cb, crc32ch and crc32cw.
26915 (arm_init_crc32_builtins): Remove.
26916 (arm_init_builtins): Use arm_init_acle_builtins rather
26917 than arm_init_crc32_builtins.
26918 (arm_expand_acle_builtin): New.
26919 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
26920 * config/arm/arm_acle_builtins.def: New.
26922 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26924 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
26925 (arm_builtin_datum): ... this.
26926 (arm_init_neon_builtin): Rename to ...
26927 (arm_init_builtin): ... this. Add a new parameters PREFIX
26928 and USE_SIG_IN_NAME.
26929 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
26930 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
26931 'arm_builtin_datum'.
26932 (arm_init_vfp_builtins): Likewise.
26933 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
26934 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
26935 (arm_expand_neon_args): Rename to ...
26936 (arm_expand_builtin_args): ... this. Rename builtin_arg
26937 enum values and differentiate between ARG_BUILTIN_MEMORY
26938 and ARG_BUILTIN_NEON_MEMORY.
26939 (arm_expand_neon_builtin_1): Rename to ...
26940 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
26941 values, arm_expand_builtin_args and add bool parameter NEON.
26942 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
26943 (arm_expand_vfp_builtin): Likewise.
26944 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
26946 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
26948 PR middle-end/77484
26949 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
26950 * predict.c (tree_estimate_probability_bb): Reverse direction of
26951 polymorphic call predictor.
26953 2017-01-06 David Malcolm <dmalcolm@redhat.com>
26955 * passes.c (execute_one_pass): Split out pass-skipping logic into...
26956 (determine_pass_name_match): ...this new function and...
26957 (should_skip_pass_p): ...this new function.
26959 2017-01-06 Nathan Sidwell <nathan@acm.org>
26961 * ipa-visibility.c (function_and_variable_visibility): Reformat
26962 comments and long lines. Remove extrneous if.
26963 * symtab.c (symtab_node::make_decl_local): Fix code format.
26964 (symtab_node::set_section_for_node): Fix comment typo.
26966 2017-01-06 Martin Liska <mliska@suse.cz>
26969 * lra-constraints.c: Rename invariant to lra_invariant.
26970 * predict.c (set_even_probabilities): Initialize e to NULL.
26972 2017-01-05 Martin Sebor <msebor@redhat.com>
26974 PR tree-optimization/78910
26975 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
26976 (format_integer): Correct off-by-one error in the handling
26977 of precision with negative numbers in signed conversions..
26979 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
26981 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
26983 2017-01-05 Jakub Jelinek <jakub@redhat.com>
26985 PR tree-optimization/71016
26986 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
26987 factor_out_conditional_conversion. Formatting fix.
26988 (factor_out_conditional_conversion): Add cond_stmt argument.
26989 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
26990 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
26993 2017-01-05 David Malcolm <dmalcolm@redhat.com>
26995 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
26996 read-rtl-function.o, and selftest-rtl.o.
26997 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
26998 (selftest::aarch64_test_loading_full_dump): New function.
26999 (selftest::aarch64_run_selftests): New function.
27000 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
27001 selftest::aarch64_run_selftests.
27002 * config/i386/i386.c
27003 (selftest::ix86_test_loading_dump_fragment_1): New function.
27004 (selftest::ix86_test_loading_call_insn): New function.
27005 (selftest::ix86_test_loading_full_dump): New function.
27006 (selftest::ix86_test_loading_unspec): New function.
27007 (selftest::ix86_run_selftests): Call the new functions.
27008 * emit-rtl.c (maybe_set_max_label_num): New function.
27009 * emit-rtl.h (maybe_set_max_label_num): New decl.
27010 * function.c (instantiate_decls): Guard call to
27011 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
27012 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
27014 * gensupport.c (gen_reader::gen_reader): Pass "false"
27015 for new "compact" param of rtx_reader.
27016 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
27017 rather than an empty string for NULL strings.
27018 * read-md.c: Potentially include config.h rather than bconfig.h.
27019 Wrap include of errors.h with #ifdef GENERATOR_FILE.
27020 (have_error): New global, copied from errors.c.
27021 (md_reader::read_name): Rename to...
27022 (md_reader::read_name_1): ...this, adding "out_loc" param,
27023 and converting "missing name or number" to returning false, rather
27025 (md_reader::read_name): Reimplement in terms of read_name_1.
27026 (md_reader::read_name_or_nil): New function.
27027 (md_reader::read_string): Handle "(nil)" by returning NULL.
27028 (md_reader::md_reader): Add new param "compact".
27029 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
27030 (md_reader::read_file): New method.
27031 * read-md.h (md_reader::md_reader): Add new param "compact".
27032 (md_reader::read_file): New method.
27033 (md_reader::is_compact): New accessor.
27034 (md_reader::read_name): Convert return type from void to file_location.
27035 (md_reader::read_name_or_nil): New decl.
27036 (md_reader::read_name_1): New decl.
27037 (md_reader::m_compact): New field.
27038 (noop_reader::noop_reader): Pass "false" for new "compact" param
27040 (rtx_reader::rtx_reader): Add new "compact" param.
27041 (rtx_reader::read_rtx_operand): Make virtual and convert return
27042 type from void to rtx.
27043 (rtx_reader::read_until): New decl.
27044 (rtx_reader::handle_any_trailing_information): New virtual function.
27045 (rtx_reader::postprocess): New virtual function.
27046 (rtx_reader::finalize_string): New virtual function.
27047 (rtx_reader::m_in_call_function_usage): New field.
27048 (rtx_reader::m_reuse_rtx_by_id): New field.
27049 * read-rtl-function.c: New file.
27050 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
27051 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
27052 (selftest::verify_three_block_rtl_cfg): New decl.
27053 * read-rtl-function.h: New file.
27054 * read-rtl.c: Potentially include config.h rather than bconfig.h.
27055 For host, include function.h, memmodel.h, and emit-rtl.h.
27056 (one_time_initialization): New function.
27057 (struct compact_insn_name): New struct.
27058 (compact_insn_names): New array.
27059 (find_code): Handle insn codes in compact dumps.
27060 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
27061 (bind_subst_iter_and_attr): Likewise.
27062 (add_condition_to_string): Likewise.
27063 (add_condition_to_rtx): Likewise.
27064 (apply_attribute_uses): Likewise.
27065 (add_current_iterators): Likewise.
27066 (apply_iterators): Likewise.
27067 (initialize_iterators): Guard usage of apply_subst_iterator with
27068 #ifdef GENERATOR_FILE.
27069 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
27070 (md_reader::read_mapping): Likewise.
27071 (add_define_attr_for_define_subst): Likewise.
27072 (add_define_subst_attr): Likewise.
27073 (read_subst_mapping): Likewise.
27074 (check_code_iterator): Likewise.
27075 (rtx_reader::read_rtx): Likewise. Move one-time initialization
27077 (one_time_initialization): New function.
27078 (rtx_reader::read_until): New method.
27079 (read_flags): New function.
27080 (parse_reg_note_name): New function.
27081 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
27082 Handle reuse_rtx ids.
27083 Wrap iterator lookup within #ifdef GENERATOR_FILE.
27084 Add parsing support for RTL dumps, mirroring the special-cases in
27085 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
27086 values, and calling handle_any_trailing_information.
27087 (rtx_reader::read_rtx_operand): Convert return type from void
27088 to rtx, returning return_rtx. Handle case 'e'. Call
27089 finalize_string on XSTR and XTMPL fields.
27090 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
27091 "(nil)" values were omitted. Call the postprocess vfunc on the
27093 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
27094 class ctor. Initialize m_in_call_function_usage. Call
27095 one_time_initialization.
27096 * rtl-tests.c (selftest::test_uncond_jump): Call
27097 set_new_first_and_last_insn.
27098 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
27099 * selftest-rtl.c: New file.
27100 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
27101 (selftest::get_insn_by_uid): New decl.
27102 * selftest-run-tests.c (selftest::run_tests): Call
27103 read_rtl_function_c_tests.
27104 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
27105 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
27108 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
27110 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
27111 operands in a special way. Assert that pos+len <= mode precision.
27113 2017-01-05 Jakub Jelinek <jakub@redhat.com>
27115 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
27116 3 argument Alias with unlimited for the negative form.
27117 (fno-vect-cost-model): Removed.
27119 2017-01-05 Martin Liska <mliska@suse.cz>
27121 * hsa-gen.c (gen_hsa_divmod): New function.
27122 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
27124 2017-01-05 Martin Liska <mliska@suse.cz>
27127 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
27130 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27132 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
27133 small constant length operands.
27135 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27137 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
27138 between loop iterations.
27140 2017-01-05 Martin Liska <mliska@suse.cz>
27143 * gimplify.c (gimplify_decl_expr): Compare to
27144 asan_poisoned_variables instread of checking flags.
27145 (gimplify_target_expr): Likewise.
27146 (gimplify_expr): Likewise.
27147 (gimplify_function_tree): Conditionally initialize
27148 asan_poisoned_variables.
27150 2017-01-04 Jeff Law <law@redhat.com>
27152 PR tree-optimizatin/78812
27153 * rtl.h (contains_mem_rtx_p): Prototype.
27154 * ifcvt.c (containts_mem_rtx_p): Move from here to...
27155 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
27156 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
27157 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
27158 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
27160 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
27162 * input.c (assert_char_at_range): Default-initialize actual_range.
27164 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
27166 * df-scan.c (df_ref_create_structure): Make regno unsigned,
27167 to match the caller.
27169 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
27171 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
27172 insns after final jump in test to emit dummy move.
27174 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
27176 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
27177 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
27179 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
27181 * multiple_target.c (create_dispatcher_calls): Init e_next.
27182 * tree-ssa-loop-split.c (split_loop): Init border.
27183 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
27186 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
27191 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
27192 (altivec_register_operand): Do not return true if the operand
27193 contains a SUBREG mixing SImode and SFmode.
27194 (vsx_register_operand): Likewise.
27195 (vsx_reg_sfsubreg_ok): New predicate.
27196 (vfloat_operand): Do not return true if the operand contains a
27197 SUBREG mixing SImode and SFmode.
27198 (vint_operand): Likewise.
27199 (vlogical_operand): Likewise.
27200 (gpc_reg_operand): Likewise.
27201 (int_reg_operand): Likewise.
27202 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
27203 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
27204 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
27206 (rs6000_emit_move_si_sf_subreg): New helper function.
27207 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
27208 fixup SUBREGs involving SImode and SFmode.
27209 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
27210 numbers for the new peephole2 optimization.
27211 (peephole2 for SFmode unions): New peephole2 to optimize cases in
27212 the GLIBC math library that do AND/IOR/XOR operations on single
27213 precision floating point.
27214 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
27215 target macros to say whether we need to avoid SUBREGs mixing
27217 (TARGET_ALLOW_SF_SUBREG): Likewise.
27218 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
27219 (UNSPEC_SI_FROM_SF): Likewise.
27220 (iorxor): Change spacing.
27221 (and_ior_xor): New iterator for AND, IOR, and XOR.
27222 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
27223 (movdi_from_sf_zero_ext): Likewise.
27224 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
27225 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
27226 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
27227 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
27228 (fms<mode>4): Likewise.
27229 (fnma<mode>4): Likewise.
27230 (fnms<mode>4): Likewise.
27231 (nfma<mode>4): Likewise.
27232 (nfms<mode>4): Likewise.
27234 2017-01-04 Marek Polacek <polacek@redhat.com>
27237 * doc/invoke.texi: Document -Wpointer-compare.
27239 2017-01-04 Jakub Jelinek <jakub@redhat.com>
27241 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
27244 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
27245 descriptions for -gdwarf-5 and emit them as uleb128 instead of
27248 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
27251 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
27252 documentation of the powerpc_popcntb_ok attribute.
27253 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
27254 code to issue warning messages if a requested CPU configuration is
27255 not supported by the binary (assembler and loader) toolchain.
27256 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
27257 made to define a built-in function that has been disabled.
27258 (paired_init_builtins): Add assertion to prevent ICE if attempt is
27259 made to define a built-in function that has been disabled.
27260 (altivec_init_builtins): Add comment explaining why definition
27261 of the DST built-in functions is not preceded by an assertion
27262 check. Add assertions to prevent ICE if attempts are made to
27263 define an altivec predicate or an abs* built-in function that has
27265 (htm_init_builtins): Add comment explaining why definition of the
27266 htm built-in functions is not preceded by an assertion check.
27268 2017-01-04 Jeff Law <law@redhat.com>
27270 PR tree-optimizatin/67955
27271 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
27272 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
27273 the points-to solution does not include pt_null. Use DECL_PT_UID
27276 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
27278 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
27279 Use gen_int_mode instead of gen_lopwart for const_int operands.
27281 2017-01-04 Jakub Jelinek <jakub@redhat.com>
27283 PR tree-optimization/71563
27284 * match.pd: Simplify X << Y into X if Y is known to be 0 or
27285 out of range value - has low bits known to be zero.
27287 2017-01-04 Alan Modra <amodra@gmail.com>
27289 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
27290 * configure: Regenerate.
27291 * config.in: Regenerate.
27293 2017-01-04 Jakub Jelinek <jakub@redhat.com>
27296 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
27297 a substring of the message, but strcmp with the whole message. Ifdef
27298 ENABLE_NLS, translate the message first using dgettext.
27300 2017-01-03 Jeff Law <law@redhat.com>
27302 PR tree-optimizatin/78856
27303 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
27304 (mark_threaded_blocks): Remove code to truncate thread paths that
27305 cross multiple loop headers. Instead invalidate the cached loop
27306 iteration information and handle case of a thread path walking
27307 into an irreducible region.
27309 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
27312 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
27313 assertions. Add support for doing the signbit if the IEEE 128-bit
27314 floating point value is in a GPR.
27315 * config/rs6000/rs6000.md (Fsignbit): Delete.
27316 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
27317 Update the length attribute if the value is in a GPR.
27318 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
27319 the sign or zero extension instruction, since the value is always 0/1.
27320 (signbit<mode>2_dm2): Delete using <Fsignbit>.
27323 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
27324 extracting SImode to a GPR register so that we can generate a
27325 store, limit the vector to be in a traditional Altivec register
27326 for the vextuwrx instruction.
27328 2017-01-03 Ian Lance Taylor <iant@google.com>
27330 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
27332 2017-01-03 Martin Sebor <msebor@redhat.com>
27334 PR tree-optimization/78696
27335 * gimple-ssa-sprintf.c (format_floating): Correct handling of
27336 precision. Use MPFR for %f for greater fidelity. Correct handling
27338 (pass_sprintf_length::compute_format_length): Set width and precision
27339 specified by asrerisk to void_node for vararg functions.
27340 (try_substitute_return_value): Adjust dump output.
27342 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
27344 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
27346 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
27348 * doc/invoke.texi (SPARC options): Document -mlra as the default.
27349 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
27350 -mlra/-mno-lra was passed to the compiler.
27352 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
27354 PR rtl-optimization/65618
27355 * emit-rtl.c (try_split): Move initialization of "before" and
27356 "after" to just before the call to emit_insn_after_setloc.
27358 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
27360 * doc/md.texi (Standard Names): Remove reference to Java frontend.
27362 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
27364 * dwarf2out.c (gen_enumeration_type_die): When
27365 -gno-strict-dwarf, add a DW_AT_encoding attribute.
27367 2017-01-03 Jakub Jelinek <jakub@redhat.com>
27369 PR tree-optimization/78965
27370 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
27371 Change first argument from const call_info & to call_info &. For %n
27372 set info.nowrite to false.
27374 PR middle-end/78901
27375 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
27376 possibly throwing calls.
27378 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
27379 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
27380 and fns handling, rather than in a separate case SSA_NAME.
27382 2017-01-02 Jeff Law <law@redhat.com>
27384 * config/darwin-driver.c (darwin_driver_init): Const-correctness
27385 fixes for first_period and second_period variables.
27387 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
27390 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
27391 (*insvqi_1): New insn pattern.
27392 (*insvqi_1_mem_rex64): Ditto.
27393 (*insvqi_2): Ditto.
27394 (*insvqi_3): Rename from *insvqi.
27396 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
27398 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
27400 * doc/cfg.texi (Edges): Remove reference to Java.
27401 (Maintaining the CFG): Ditto.
27403 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
27405 PR middle-end/77674
27406 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
27407 transparent aliases.
27409 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
27411 PR middle-end/77484
27412 * predict.def (PRED_CALL): Update hitrate.
27413 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
27414 * predict.c (tree_estimate_probability_bb): Split CALL predictor
27415 into direct/indirect/polymorphic variants.
27417 2017-01-01 Jakub Jelinek <jakub@redhat.com>
27419 Update copyright years.
27421 * gcc.c (process_command): Update copyright notice dates.
27422 * gcov-dump.c (print_version): Ditto.
27423 * gcov.c (print_version): Ditto.
27424 * gcov-tool.c (print_version): Ditto.
27425 * gengtype.c (create_file): Ditto.
27426 * doc/cpp.texi: Bump @copying's copyright year.
27427 * doc/cppinternals.texi: Ditto.
27428 * doc/gcc.texi: Ditto.
27429 * doc/gccint.texi: Ditto.
27430 * doc/gcov.texi: Ditto.
27431 * doc/install.texi: Ditto.
27432 * doc/invoke.texi: Ditto.
27434 Copyright (C) 2017 Free Software Foundation, Inc.
27436 Copying and distribution of this file, with or without modification,
27437 are permitted in any medium without royalty provided the copyright
27438 notice and this notice are preserved.