1 2017-09-20 Jakub Jelinek <jakub@redhat.com>
3 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
4 (DEBUG_LTO_ABBREV_SECTION): Likewise.
5 (DEBUG_LTO_MACINFO_SECTION): Likewise.
6 (DEBUG_MACRO_SECTION): Likewise.
7 (DEBUG_LTO_MACRO_SECTION): Likewise.
8 (DEBUG_STR_DWO_SECTION): Likewise.
9 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
10 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
11 (DEBUG_LTO_DWO_LINE_SECTION): Define.
12 (DEBUG_LTO_LINE_STR_SECTION): Define.
13 (init_sections_and_labels): Initialize debug_line_str_section
14 variable. Initialize debug_loc_section for -gdwarf-5 to
15 DEBUG_LOCLISTS_SECTION. Formatting fixes.
17 2017-09-20 Richard Biener <rguenther@suse.de>
19 * graphite-sese-to-poly.c (extract_affine): Properly handle
20 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
22 2017-09-20 Richard Biener <rguenther@suse.de>
24 PR tree-optimization/81373
25 * graphite-scop-detection.c (build_cross_bb_scalars_def):
26 Force SESE live-out defs to be handled even if they are
29 2017-09-19 Jeff Law <law@redhat.com>
31 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
32 nothing for stack adjustments with REG_STACK_CHECK.
33 * sched-deps.c (parse_add_or_inc): Reject insns with
34 REG_STACK_CHECK from dependency breaking.
35 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
36 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
37 * reg-notes.def (STACK_CHECK): New note.
39 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
40 (ix86_expand_prologue): Dump stack clash info as needed.
41 Call ix86_adjust_stack_and_probe_stack_clash as needed.
43 * function.c (dump_stack_clash_frame_info): New function.
44 * function.h (dump_stack_clash_frame_info): Prototype.
45 (enum stack_clash_probes): New enum.
47 * config/alpha/alpha.c (alpha_expand_prologue): Also check
48 flag_stack_clash_protection.
49 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
50 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
51 (arm_frame_pointer_required): Likewise.
52 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
53 (ia64_expand_prologue): Likewise.
54 * config/mips/mips.c (mips_expand_prologue): Likewise.
55 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
56 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
57 (sparc_flat_expand_prologue): Likewise.
58 * config/spu/spu.c (spu_expand_prologue): Likewise.
60 * explow.c: Include "params.h".
61 (anti_adjust_stack_and_probe_stack_clash): New function.
62 (get_stack_check_protect): Likewise.
63 (compute_stack_clash_protection_loop_data): Likewise.
64 (emit_stack_clash_protection_loop_start): Likewise.
65 (emit_stack_clash_protection_loop_end): Likewise.
66 (allocate_dynamic_stack_space): Use get_stack_check_protect.
67 Use anti_adjust_stack_and_probe_stack_clash.
68 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
69 (emit_stack_clash_protection_loop_start): Likewise.
70 (emit_stack_clash_protection_loop_end): Likewise.
71 * rtl.h (get_stack_check_protect): Prototype.
72 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
73 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
74 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
76 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
78 * doc/tm.texi: Rebuilt.
79 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
80 get_stack_check_protect.
81 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
82 * config/arm/arm.c (arm_expand_prologue): Likewise.
83 (arm_frame_pointer_required): Likewise.
84 * config/i386/i386.c (ix86_expand_prologue): Likewise.
85 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
86 * config/mips/mips.c (mips_expand_prologue): Likewise.
87 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
88 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
89 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
90 (sparc_flat_expand_prologue): Likewise.
92 * common.opt (-fstack-clash-protection): New option.
93 * flag-types.h (enum stack_check_type): Note difference between
94 -fstack-check= and -fstack-clash-protection.
95 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
96 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
97 * toplev.c (process_options): Issue warnings/errors for cases
98 not handled with -fstack-clash-protection.
99 * doc/invoke.texi (-fstack-clash-protection): Document new option.
100 (-fstack-check): Note additional problem with -fstack-check=generic.
101 Note that -fstack-check is primarily for Ada and refer users
102 to -fstack-clash-protection for stack-clash-protection.
103 Document new params for stack clash protection.
105 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
107 * config/i386/i386.md (*scc_bt<mode>): New insn_and_split pattern.
108 (*scc_bt<mode>_1): Ditto.
109 (*scc_bt<mode>_mask): Ditto.
111 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
113 * config/i386/i386.c (ix86_split_long_move): Do not handle
114 address used for LEA in a special way.
116 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
118 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
120 2017-09-19 Martin Sebor <msebor@redhat.com>
123 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
124 of incompatible types.
126 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
128 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
129 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
130 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
131 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
133 2017-09-19 Richard Biener <rguenther@suse.de>
135 PR tree-optimization/82244
136 * tree-vrp.c (remove_range_assertions): Do not propagate
137 a constant to abnormals but replace the assert with a copy.
139 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
141 PR rtl-optimization/57878
142 PR rtl-optimization/68988
143 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
144 avoidance test involving non_reload_pseudos. Move frequency test
145 below the general fragmentation avoidance test.
147 2017-09-19 Richard Biener <rguenther@suse.de>
149 PR tree-optimization/69728
150 * graphite-sese-to-poly.c (schedule_error): New global.
151 (add_loop_schedule): Handle empty domain by failing the
153 (build_original_schedule): Handle schedule_error.
155 2017-09-19 Richard Biener <rguenther@suse.de>
157 * graphite-scop-detection.c (scop_detection::can_represent_loop):
158 Do not iterate to sibling loops but only to siblings of inner
161 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
164 * config/m68k/m68k.md (moveq feeding equality comparison): Check
165 that the registers are different.
167 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
169 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
170 to processor_model and "amdfam17h" to arch_names_table.
171 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
173 2017-09-18 Jakub Jelinek <jakub@redhat.com>
176 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
178 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
179 Alan Hayward <alan.hayward@arm.com>
180 David Sherwood <david.sherwood@arm.com>
182 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
184 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
186 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
187 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
188 vect_schedule_slp_instance.
189 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
190 Update call to vect_slp_analyze_node_operations. Simplify return
192 (vect_slp_analyze_bb_1): Update call accordingly.
193 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
194 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
195 (vect_schedule_slp): Update call accordingly.
197 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
198 Alan Hayward <alan.hayward@arm.com>
199 David Sherwood <david.sherwood@arm.com>
201 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
202 with types that aren't in fact scalar.
204 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
206 * tree-vect-slp.c (vect_record_max_nunits): New function,
208 (vect_build_slp_tree_1): ...here.
209 (vect_build_slp_tree_2): Call it for phis too.
211 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
213 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
214 to vect_get_vec_def_for_operand when getting the mask operand.
216 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
217 Alan Hayward <alan.hayward@arm.com>
218 David Sherwood <david.sherwood@arm.com>
220 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
223 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
224 Alan Hayward <alan.hayward@arm.com>
225 David Sherwood <david.sherwood@arm.com>
227 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
228 calculation for vector booleans.
230 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
231 Alan Hayward <alan.hayward@arm.com>
232 David Sherwood <david.sherwood@arm.com>
234 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
236 (vect_transform_stmt): ...here.
237 (vect_analyze_stmt): Use it instead of calling
238 vectorizable_live_operation directly.
240 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
242 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
243 non-SIMT targets in acc vector loops.
245 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
247 * configure.ac: Add arc and check if assembler supports gdwarf2.
248 * configure: Regenerate.
250 2017-09-18 Richard Biener <rguenther@suse.de>
252 PR tree-optimization/82220
253 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
254 epilogue niters from the min_profitable_iters compute.
256 2017-09-18 Jakub Jelinek <jakub@redhat.com>
259 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
260 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
261 (ix86_init_pic_reg): Revert 2017-09-01 changes.
263 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
266 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
267 switching to a new text section.
269 2017-09-18 Richard Biener <rguenther@suse.de>
271 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
273 (build_alias_set): Reject aliases with no access function.
275 2017-09-18 Richard Biener <rguenther@suse.de>
277 PR tree-optimization/79622
278 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
280 (build_cross_bb_scalars_use): Likewise.
282 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
284 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
286 2017-09-18 Alan Modra <amodra@gmail.com>
289 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
290 stack_pointer_rtx for count 0. Update comments. Break up
291 large rtl expression.
293 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
295 config/i386/i386.c: (xlogue_layout::STUB_NAME_MAX_LEN): Increase to 20
297 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
298 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
299 or avx version of the stub.
301 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
304 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
305 compute the minimum stack alignment. Also update preferred stack
306 boundary for leaf functions.
308 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
310 PR tree-optimization/82228
311 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
314 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
316 * common/config/nds32/nds32-common.c
317 (nds32_option_optimization_table): Refine formatting.
318 (nds32_option_optimization_table): Use -fsched-pressure and
319 -fomit-frame-pointer for specific optimization level.
321 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
323 * config/nds32/nds32.c: Refine formatting and comments.
324 * config/nds32/nds32.h: Likewise.
325 * config/nds32/nds32.md: Likewise.
326 * config/nds32/nds32-cost.c: Likewise.
327 * config/nds32/nds32-isr.c: Likewise.
328 * config/nds32/nds32-md-auxiliary.c: Likewise.
329 * config/nds32/nds32-multiple.md: Likewise.
330 * config/nds32/nds32-predicates.c: Likewise.
332 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
333 Jakub Jelinek <jakub@redhat.com>
335 Add support for -std=c++2a.
336 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
338 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
340 2017-09-15 Steve Ellcey <sellcey@cavium.com>
343 * doc/extend.texi (Common Function Attributes): Add
344 references to ARM, AArch64, and S/390 specific attributes.
345 (Function Specific Option Pragmas): Add AArch64 and S/390
346 to list of back ends that support the target pragma.
348 2017-09-15 Nathan Sidwell <nathan@acm.org>
350 * doc/standards.texi: Fix C++17 description. Update URLs for
353 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
355 * common.opt (Wcast-align=strict): New warning option.
356 * doc/invoke.texi: Document -Wcast-align=strict.
358 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
360 * cgraph.h (cgraph_thunk_info): Add comments.
361 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
362 assert for VIRTUAL_* arguments stricter.
364 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
366 PR tree-optimization/71026
367 * match.pd: Move RDIV patterns from fold-const.c
368 * fold-const.c (distribute_real_division): Removed.
369 (fold_binary_loc): Remove calls to distribute_real_divison.
371 2017-09-15 Jakub Jelinek <jakub@redhat.com>
373 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
374 c++1z and gnu++1z as deprecated. Change other references to
375 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
376 Change -Wc++1z-compat to -Wc++17-compat.
377 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
378 * dwarf2out.c (highest_c_language): Handle C++17.
379 (gen_compile_unit_die): Likewise.
381 2017-09-15 Jakub Jelinek <jakub@redhat.com>
383 PR rtl-optimization/82192
384 * combine.c (make_extraction): Don't look through non-paradoxical
385 SUBREGs or TRUNCATE if pos + len is or might be bigger than
388 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
389 Alan Hayward <alan.hayward@arm.com>
390 David Sherwood <david.sherwood@arm.com>
392 * target.def (function_arg_offset): New hook.
393 * targhooks.h (default_function_arg_offset): Declare.
394 * targhooks.c (default_function_arg_offset): New function.
395 * function.c (locate_and_pad_parm): Use
396 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
397 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
398 (TARGET_FUNCTION_ARG_OFFSET): ...this.
399 * doc/tm.texi: Regenerate.
400 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
401 * config/spu/spu.c (spu_function_arg_offset): New function.
402 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
403 * system.h (FUNCTION_ARG_OFFSET): Poison.
405 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
406 Alan Hayard <alan.hayward@arm.com>
407 David Sherwood <david.sherwood@arm.com>
409 * target.def (truly_noop_truncation): New hook.
410 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
411 than TRULY_NOOP_TRUNCATION.
412 * hooks.h (hook_bool_uint_uint_true): Declare.
413 * hooks.c (hook_bool_uint_uint_true): New function.
414 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
415 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
416 * doc/tm.texi: Regenerate.
417 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
418 rather than TRULY_NOOP_TRUNCATION in comments.
419 (simplify_comparison): Likewise.
420 (record_truncated_value): Likewise.
421 * expmed.c (extract_bit_field_1): Likewise.
422 (extract_split_bit_field): Likewise.
423 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
424 instead of TRULY_NOOP_TRUNCATION.
425 * function.c (assign_parm_setup_block): Likewise.
426 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
427 * rtlhooks.c: Include target.h.
428 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
429 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
430 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
431 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
432 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
433 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
434 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
435 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
436 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
437 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
438 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
439 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
440 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
441 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
442 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
443 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
444 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
445 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
446 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
447 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
448 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
449 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
450 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
451 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
452 * config/mips/mips.c (mips_truly_noop_truncation): New function.
453 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
454 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
455 rather than TRULY_NOOP_TRUNCATION in comments.
456 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
457 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
458 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
459 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
460 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
461 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
462 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
463 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
464 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
465 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
466 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
467 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
468 rather than TRULY_NOOP_TRUNCATION in comments.
469 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
470 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
471 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
472 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
473 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
474 TRULY_NOOP_TRUNCATION condition.
475 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
476 (TRULY_NOOP_TRUNCATION): Delete.
477 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
478 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
479 * config/spu/spu.c (spu_truly_noop_truncation): New function.
480 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
481 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
482 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
483 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
484 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
485 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
486 rather than TRULY_NOOP_TRUNCATION in comments.
487 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
488 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
489 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
490 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
491 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
492 * system.h (TRULY_NOOP_TRUNCATION): Poison.
494 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
497 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
498 (*cmp_ior): Likewise.
499 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
500 (*ior_scc_scc_cmp): Likewise.
501 (*and_scc_scc): Likewise.
502 (*and_scc_scc_cmp): Likewise.
504 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
505 Alan Hayard <alan.hayward@arm.com>
506 David Sherwood <david.sherwood@arm.com>
508 * target.def (can_change_mode_class): New hook.
509 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
510 (hard_regno_nregs): Likewise.
511 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
512 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
513 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
514 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
515 (LOAD_EXTEND_OP): Update accordingly.
516 * doc/tm.texi: Regenerate.
517 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
518 CANNOT_CHANGE_MODE_CLASS.
519 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
520 (REG_CAN_CHANGE_MODE_P): ...this new macro.
521 * combine.c (simplify_set): Update accordingly.
522 * emit-rtl.c (validate_subreg): Likewise.
523 * recog.c (general_operand): Likewise.
524 * regcprop.c (mode_change_ok): Likewise.
525 * reload1.c (choose_reload_regs): Likewise.
526 (inherit_piecemeal_p): Likewise.
527 * rtlanal.c (simplify_subreg_regno): Likewise.
528 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
529 instead of CANNOT_CHANGE_MODE_CLASS.
530 (reload_cse_simplify_operands): Likewise.
531 * reload.c (push_reload): Use targetm.can_change_mode_class
532 instead of CANNOT_CHANGE_MODE_CLASS.
533 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
534 REG_CANNOT_CHANGE_MODE_P.
535 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
536 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
537 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
538 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
539 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
540 (arm_can_change_mode_class): New function.
541 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
542 than CANNOT_CHANGE_MODE_CLASS in comments.
543 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
544 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
545 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
546 (ix86_can_change_mode_class): ...this new function, inverting the
547 sense of the return value.
548 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
549 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
550 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
551 (ia64_can_change_mode_class): New function.
552 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
553 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
554 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
555 (m32c_can_change_mode_class): ...this new function, inverting the
556 sense of the return value.
557 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
558 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
559 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
560 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
561 (mips_can_change_mode_class): ...this new function, inverting the
562 sense of the return value.
563 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
564 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
565 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
566 (msp430_can_change_mode_class): New function.
567 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
568 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
569 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
570 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
571 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
572 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
573 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
574 (pa_cannot_change_mode_class): Replace with...
575 (pa_can_change_mode_class): ...this new function, inverting the
576 sense of the return value.
577 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
578 than CANNOT_CHANGE_MODE_CLASS in comments.
579 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
580 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
581 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
582 (pdp11_cannot_change_mode_class): Replace with...
583 (pdp11_can_change_mode_class): ...this new function, inverting the
584 sense of the return value.
585 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
586 * config/powerpcspe/powerpcspe-protos.h
587 (rs6000_cannot_change_mode_class_ptr): Delete.
588 * config/powerpcspe/powerpcspe.c
589 (rs6000_cannot_change_mode_class_ptr): Delete.
590 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
591 (rs6000_option_override_internal): Assign to
592 targetm.can_change_mode_class instead of
593 rs6000_cannot_change_mode_class_ptr.
594 (rs6000_cannot_change_mode_class): Replace with...
595 (rs6000_can_change_mode_class): ...this new function, inverting the
596 sense of the return value.
597 (rs6000_debug_cannot_change_mode_class): Replace with...
598 (rs6000_debug_can_change_mode_class): ...this new function.
599 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
600 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
601 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
602 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
603 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
605 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
606 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
607 (rs6000_option_override_internal): Assign to
608 targetm.can_change_mode_class instead of
609 rs6000_cannot_change_mode_class_ptr.
610 (rs6000_cannot_change_mode_class): Replace with...
611 (rs6000_can_change_mode_class): ...this new function, inverting the
612 sense of the return value.
613 (rs6000_debug_cannot_change_mode_class): Replace with...
614 (rs6000_debug_can_change_mode_class): ...this new function.
615 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
616 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
617 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
618 (s390_can_change_mode_class): ...this new function, inverting the
619 sense of the return value.
620 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
621 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
622 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
623 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
624 (sh_cannot_change_mode_class): Replace with...
625 (sh_can_change_mode_class): ...this new function, inverting the
626 sense of the return value.
627 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
628 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
629 (sparc_can_change_mode_class): New function.
630 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
631 * config/spu/spu.c (spu_can_change_mode_class): New function.
632 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
633 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
634 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
635 (visium_can_change_mode_class): New function.
636 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
638 2017-09-15 Richard Biener <rguenther@suse.de>
640 PR tree-optimization/82217
641 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
642 but not undefined case.
644 2017-09-15 Jakub Jelinek <jakub@redhat.com>
647 * postreload.c (reload_cse_simplify_operands): Skip
648 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
650 2017-09-15 Richard Biener <rguenther@suse.de>
652 PR tree-optimization/68823
653 * graphite-scop-detection.c (build_alias_set): If we have a
654 possible dependence check whether we can handle them by just
655 looking at the DRs DR_ACCESS_FNs.
656 (build_scops): If build_alias_set fails, fail the SCOP.
658 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
660 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
661 to support float128 built-in functions that require the ISA 3.0
663 (BU_FLOAT128_3_HW): Likewise.
664 (SQRTF128): Add support for the IEEE 128-bit square root and fma
668 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
669 support for built-in functions that need the ISA 3.0 IEEE 128-bit
670 floating point instructions.
671 (rs6000_invalid_builtin): Likewise.
672 (rs6000_builtin_mask_names): Likewise.
673 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
674 (RS6000_BTM_FLOAT128_HW): Likewise.
675 (RS6000_BTM_COMMON): Likewise.
676 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
678 * doc/extend.texi (RS/6000 built-in functions): Document the
679 IEEE 128-bit floating point square root and fused multiply-add
682 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
684 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
685 reg (r2) isn't in the set of registers defined in the prologue.
687 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
688 Alan Hayward <alan.hayward@arm.com>
689 David Sherwood <david.sherwood@arm.com>
691 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
692 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
693 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
694 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
695 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
697 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
698 max_vectorization_factor.
699 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
701 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
702 Alan Hayward <alan.hayward@arm.com>
703 David Sherwood <david.sherwood@arm.com>
705 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
706 (vect_worthwhile_without_simd_p): Declare.
707 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
708 (vectorizable_reduction): Use it.
709 * tree-vect-stmts.c (vectorizable_shift): Likewise.
710 (vectorizable_operation): Likewise.
712 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
713 Alan Hayward <alan.hayward@arm.com>
714 David Sherwood <david.sherwood@arm.com>
716 * tree-vectorizer.h (vect_get_num_copies): New function.
717 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
718 * tree-vect-loop.c (vectorizable_reduction): Likewise.
719 (vectorizable_induction): Likewise.
720 (vectorizable_live_operation): Likewise.
721 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
722 (vectorizable_bswap): Likewise.
723 (vectorizable_call): Likewise.
724 (vectorizable_conversion): Likewise.
725 (vectorizable_assignment): Likewise.
726 (vectorizable_shift): Likewise.
727 (vectorizable_operation): Likewise.
728 (vectorizable_store): Likewise.
729 (vectorizable_load): Likewise.
730 (vectorizable_condition): Likewise.
731 (vectorizable_comparison): Likewise.
732 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
734 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
735 Alan Hayward <alan.hayward@arm.com>
736 David Sherwood <david.sherwood@arm.com>
738 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
741 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
742 Alan Hayward <alan.hayward@arm.com>
743 David Sherwood <david.sherwood@arm.com>
745 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
746 an inline wrapper that provides a location.
747 (gimple_build_vector): Likewise.
748 * gimple-fold.c (gimple_build_vector_from_val): New function.
749 (gimple_build_vector): Likewise.
750 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
751 functions to build the initial value. Always return a gimple value.
752 (get_initial_defs_for_reduction): Likewise. Only compute
754 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
755 vect_init_vector on the results from get_initial_def(s)_for_reduction.
756 (vectorizable_induction): Use gimple_build_vector rather than
759 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
760 Alan Hayward <alan.hayward@arm.com>
761 David Sherwood <david.sherwood@arm.com>
763 * target.h (vec_perm_indices): New typedef.
764 (auto_vec_perm_indices): Likewise.
765 * optabs-query.h: Include target.h
766 (can_vec_perm_p): Take a vec_perm_indices *.
767 * optabs-query.c (can_vec_perm_p): Likewise.
768 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
769 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
770 * tree-vect-generic.c (lower_vec_perm): Likewise.
771 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
772 (vect_grouped_load_supported): Likewise.
773 (vect_shift_permute_load_chain): Likewise.
774 (vect_permute_store_chain): Use auto_vec_perm_indices.
775 (vect_permute_load_chain): Likewise.
776 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
777 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
778 Update uses of can_vec_perm_p.
779 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
780 mode with a number of elements. Take a vec_perm_indices *.
781 (vect_create_epilog_for_reduction): Update accordingly.
782 Use auto_vec_perm_indices.
783 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
784 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
785 (vect_transform_slp_perm_load): Likewise.
786 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
787 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
788 (vect_gen_perm_mask_checked): Likewise.
789 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
790 (vect_gen_perm_mask_checked): Likewise.
791 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
792 (vectorizable_store): Likewise.
793 (vectorizable_load): Likewise.
794 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
795 (vectorizable_bswap): Likewise.
797 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
798 Alan Hayward <alan.hayward@arm.com>
799 David Sherwood <david.sherwood@arm.com>
801 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
802 * tree.c (build_vector): Likewise.
803 (build_vector_from_ctor): Update accordingly.
804 (build_vector_from_val): Likewise.
805 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
806 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
807 * tree-vect-generic.c (add_rshift): Likewise.
808 (expand_vector_divmod): Likewise.
809 (optimize_vector_constructor): Likewise.
810 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
811 (vect_transform_slp_perm_load): Likewise.
812 (vect_schedule_slp_instance): Likewise.
813 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
814 (vectorizable_call): Likewise.
815 (vect_gen_perm_mask_any): Likewise. Add elements in order.
816 * expmed.c (make_tree): Likewise.
817 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
818 a vector passed to build_vector.
819 (fold_convert_const): Likewise.
820 (exact_inverse): Likewise.
821 (fold_ternary_loc): Likewise.
822 (fold_relational_const): Likewise.
823 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
824 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
825 (const_unop): Likewise. Store the reduction accumulator in a
826 variable rather than an array.
827 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
828 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
829 the new vector, rather than constructing it after the input arrays.
830 (native_interpret_vector): Use auto_vec<tree> when building
831 a vector passed to build_vector. Add elements in order.
832 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
833 auto_vec<tree> when building a vector passed to build_vector.
834 (vect_create_epilog_for_reduction): Likewise.
835 (vectorizable_induction): Likewise.
836 (get_initial_def_for_reduction): Likewise. Fix indentation of
838 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
840 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
841 passed to build_vector.
843 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
844 Alan Hayward <alan.hayward@arm.com>
845 David Sherwood <david.sherwood@arm.com>
847 * tree-core.h (tree_base::u): Add an "nelts" field.
848 (tree_vector): Use VECTOR_CST_NELTS as the length.
849 * tree.c (tree_size): Likewise.
850 (make_vector): Initialize VECTOR_CST_NELTS.
851 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
852 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
853 TYPE_VECTOR_SUBPARTS.
854 * expr.c (const_vector_mask_from_tree): Consistently use "units"
855 as the number of units, setting it from VECTOR_CST_NELTS.
856 (const_vector_from_tree): Likewise.
857 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
858 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
859 (fold_negate_expr_1): Likewise.
860 (fold_convert_const): Likewise.
861 (const_binop): Likewise. Differentiate the number of output and
863 (const_unop): Likewise.
864 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
865 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
866 in cases that did the opposite.
868 2017-09-14 Richard Biener <rguenther@suse.de>
870 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
873 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
875 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
877 2017-09-14 Jakub Jelinek <jakub@redhat.com>
880 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
881 if and where to split a bb, except for splitting before debug insn
882 sequences followed by non-label real insn. Delete debug insns
883 in between basic blocks.
885 * combine.c (make_compound_operation_int): Formatting fixes.
887 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
888 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
889 * config/netbsd.h (LINK_EH_SPEC): Likewise.
890 * config/sol2.h (LINK_EH_SPEC): Likewise.
891 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
892 * config/s390/linux.h (LINK_SPEC): Likewise.
893 * config/freebsd.h (LINK_EH_SPEC): Likewise.
894 * config/openbsd.h (LINK_EH_SPEC): Likewise.
895 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
896 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
897 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
898 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
899 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
900 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
902 2017-09-13 Jakub Jelinek <jakub@redhat.com>
904 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
906 (ENDFILE_LINUX_SPEC): Likewise.
907 (LINK_EH_SPEC): Likewise.
908 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
909 (LINK_OS_LINUX_SPEC32): Likewise.
910 (LINK_OS_LINUX_SPEC64): Likewise.
911 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
912 (LINK_OS_LINUX_SPEC): Likewise.
914 2017-09-13 Martin Liska <mliska@suse.cz>
917 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
918 CASE_HIGH is NULL_TREE.
920 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
921 Alan Hayward <alan.hayward@arm.com>
922 David Sherwood <david.sherwood@arm.com>
924 * target.def (secondary_memory_needed): New hook.
925 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
926 instead of SECONDARY_MEMORY_NEEDED.
927 (secondary_memory_needed_mode): Likewise.
928 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
929 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
930 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
931 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
932 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
933 * doc/tm.texi: Regenerate.
934 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
935 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
936 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
937 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
938 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
939 * config/i386/i386.c (inline_secondary_memory_needed): Put the
940 mode argument first and change the reg_class arguments to reg_class_t.
941 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
942 Make static. Update the call to inline_secondary_memory_needed.
943 (ix86_register_move_cost): Update the call to
944 inline_secondary_memory_needed.
945 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
946 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
948 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
949 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
951 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
952 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
953 * config/mips/mips.c (mips_secondary_memory_needed): Make static
954 and match hook interface. Add comment from mips.h.
955 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
956 * config/mmix/mmix.md (truncdfsf2): Refer to
957 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
959 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
960 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
961 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
962 (pa_secondary_memory_needed): New function.
963 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
964 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
965 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
966 (pdp11_secondary_memory_needed): Make static and match hook interface.
967 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
968 * config/powerpcspe/powerpcspe-protos.h
969 (rs6000_secondary_memory_needed_ptr): Delete.
970 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
972 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
973 (rs6000_option_override_internal): Assign to
974 targetm.secondary_memory_needed rather than
975 rs6000_secondary_memory_needed_ptr.
976 (rs6000_secondary_memory_needed): Match hook interface.
977 (rs6000_debug_secondary_memory_needed): Likewise.
978 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
979 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
980 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
981 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
982 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
983 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
985 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
986 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
987 (rs6000_option_override_internal): Assign to
988 targetm.secondary_memory_needed rather than
989 rs6000_secondary_memory_needed_ptr.
990 (rs6000_secondary_memory_needed): Match hook interface.
991 (rs6000_debug_secondary_memory_needed): Likewise.
992 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
993 * config/s390/s390.c (s390_secondary_memory_needed): New function.
994 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
995 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
996 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
997 (sparc_secondary_memory_needed): New function.
998 * lra-constraints.c (check_and_process_move): Refer to
999 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
1001 (curr_insn_transform): Likewise.
1002 (process_alt_operands): Use targetm.secondary_memory_needed
1003 instead of TARGET_SECONDARY_MEMORY_NEEDED.
1004 (check_secondary_memory_needed_p): Likewise.
1005 (choose_split_class): Likewise.
1006 * reload.c: Unconditionally include code that was previously
1007 conditional on SECONDARY_MEMORY_NEEDED.
1008 (push_secondary_reload): Use targetm.secondary_memory_needed
1009 instead of TARGET_SECONDARY_MEMORY_NEEDED.
1010 (push_reload): Likewise.
1011 * reload1.c: Unconditionally include code that was previously
1012 conditional on SECONDARY_MEMORY_NEEDED.
1013 (choose_reload_regs): Use targetm.secondary_memory_needed
1014 instead of TARGET_SECONDARY_MEMORY_NEEDED.
1015 (gen_reload): Likewise.
1016 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
1018 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
1019 Alan Hayward <alan.hayward@arm.com>
1020 David Sherwood <david.sherwood@arm.com>
1022 * target.def (secondary_memory_needed_mode): New hook:
1023 * targhooks.c (default_secondary_memory_needed_mode): Declare.
1024 * targhooks.h (default_secondary_memory_needed_mode): New function.
1025 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
1026 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
1027 * doc/tm.texi: Regenerate.
1028 * lra-constraints.c (check_and_process_move): Use
1029 targetm.secondary_memory_needed_mode instead of
1030 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
1031 (curr_insn_transform): Likewise.
1032 * reload.c (get_secondary_mem): Likewise.
1033 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
1034 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
1036 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
1037 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
1038 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
1039 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
1040 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
1042 * config/powerpcspe/powerpcspe-protos.h
1043 (rs6000_secondary_memory_needed_mode): Delete.
1044 * config/powerpcspe/powerpcspe.c
1045 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
1046 (rs6000_secondary_memory_needed_mode): Make static.
1047 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
1048 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
1050 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
1052 (rs6000_secondary_memory_needed_mode): Make static.
1053 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
1054 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
1055 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
1056 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
1057 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
1059 (sparc_secondary_memory_needed_mode): New function.
1060 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
1062 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
1064 * config/aarch64/constraints.md (Umq): New constraint.
1065 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
1067 (mov<mode>): Update condition.
1069 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1071 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
1072 when bitposition is the same.
1074 2017-09-13 Richard Biener <rguenther@suse.de>
1076 * dwarf2out.c (output_die_symbol): Remove.
1077 (output_die): Do not output a DIEs symbol.
1079 2017-09-13 Richard Biener <rguenther@suse.de>
1082 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
1083 default-def to avoid breaking iterator update with the weird
1084 interaction with cgraph_update_edges_for_call_stmt_node.
1086 2017-09-13 Richard Biener <rguenther@suse.de>
1088 * tree-cfg.c (verify_gimple_assign_binary): Add verification
1089 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
1090 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
1091 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
1093 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
1095 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
1096 Disable pc relative literal load irrespective of
1097 TARGET_FIX_ERR_A53_84341 for default.
1099 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
1101 * config/sparc/sparc.c (output_return): Output the source location of
1102 the insn in the delay slot, if any.
1103 (output_sibcall): Likewise.
1105 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
1108 * common.opt (-static-pie): New alias.
1109 (shared): Negate static-pie.
1110 (-no-pie): Update help text.
1112 (static-pie): New option.
1113 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
1114 -static-pie support.
1115 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
1116 (LINK_EH_SPEC): Likewise.
1117 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
1118 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
1119 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
1120 * gcc.c (LINK_COMMAND_SPEC): Likewise.
1121 (init_gcc_specs): Likewise.
1122 (init_spec): Likewise.
1123 (display_help): Update help message for -pie.
1124 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
1127 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
1129 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
1130 (movdi_aarch64): Likewise.
1131 (movti_aarch64): Likewise.
1133 2017-09-12 Simon Wright <simon@pushface.org>
1136 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
1137 calculation of the minor version, always output as 0.
1139 2017-09-12 Jakub Jelinek <jakub@redhat.com>
1142 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
1143 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
1144 on it early, rather than manual conversion late. For
1145 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
1146 instead of performing manual conversion.
1148 2017-09-12 Carl Love <cel@us.ibm.com>
1150 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
1151 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
1153 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
1154 VMULOSW): Add definitions.
1155 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
1156 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
1157 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
1158 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
1159 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
1161 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
1163 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
1165 (movti_aarch64): Likewise.
1166 (movdf_aarch64): Likewise.
1167 (movtf_aarch64): Likewise.
1168 (load_pairdi): Likewise.
1169 (store_pairdi): Likewise.
1170 (load_pairdf): Likewise.
1171 (store_pairdf): Likewise.
1172 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
1173 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
1174 (ldr_got_small_<mode>): Likewise.
1175 (ldr_got_small_28k_<mode>): Likewise.
1176 (ldr_got_tiny): Likewise.
1177 * config/aarch64/iterators.md (ldst_sz): New.
1178 (ldpstp_sz): Likewise.
1179 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
1181 (thunderx_load): Split load_8 to load_16.
1182 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
1184 (thunderx2t99_storepair_basic): Split store_8 to store_16.
1185 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
1186 (xgene1_store_pair): Split store_8 to store_16.
1187 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
1188 (falkor_st_0_st_sd): Split store_8 to store_16.
1190 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
1192 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
1193 and store1/2/3/4 to store_4/8/12/16.
1194 * config/aarch64/aarch64.md: Update for rename.
1195 * config/arm/arm.md: Likewise.: Likewise.
1196 * config/arm/arm.c: Likewise.
1197 * config/arm/thumb1.md: Likewise.
1198 * config/arm/thumb2.md: Likewise.
1199 * config/arm/vfp.md: Likewise.
1200 * config/arm/arm-generic.md: Likewise.
1201 * config/arm/arm1020e.md: Likewise.
1202 * config/arm/arm1026ejs.md: Likewise.
1203 * config/arm/arm1136jfs.md: Likewise.
1204 * config/arm/arm926ejs.md: Likewise.
1205 * config/arm/cortex-a15.md: Likewise.
1206 * config/arm/cortex-a17.md: Likewise.
1207 * config/arm/cortex-a5.md: Likewise.
1208 * config/arm/cortex-a53.md: Likewise.
1209 * config/arm/cortex-a57.md: Likewise.
1210 * config/arm/cortex-a7.md: Likewise.
1211 * config/arm/cortex-a8.md: Likewise.
1212 * config/arm/cortex-a9.md: Likewise.
1213 * config/arm/cortex-m4.md: Likewise.
1214 * config/arm/cortex-m7.md: Likewise.
1215 * config/arm/cortex-r4.md: Likewise.
1216 * config/arm/exynos-m1.md: Likewise.
1217 * config/arm/fa526.md: Likewise.
1218 * config/arm/fa606te.md: Likewise.
1219 * config/arm/fa626te.md: Likewise.
1220 * config/arm/fa726te.md: Likewise.
1221 * config/arm/fmp626.md: Likewise.
1222 * config/arm/iwmmxt.md: Likewise.
1223 * config/arm/ldmstm.md: Likewise.
1224 * config/arm/marvell-pj4.md: Likewise.
1225 * config/arm/xgene1.md: Likewise.
1226 * config/aarch64/thunderx.md: Likewise.
1227 * config/aarch64/thunderx2t99.md: Likewise.
1228 * config/aarch64/falkor.md: Likewise.
1230 2017-09-12 Martin Liska <mliska@suse.cz>
1232 * attribs.c (private_lookup_attribute): New function.
1233 * attribs.h (private_lookup_attribute): Declared here.
1234 (lookup_attribute): Called from this place.
1236 2017-09-12 Richard Biener <rguenther@suse.de>
1238 PR tree-optimization/82157
1239 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
1240 stmts with side-effects.
1242 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1243 Alan Hayward <alan.hayward@arm.com>
1244 David Sherwood <david.sherwood@arm.com>
1246 * target.def (hard_regno_nregs): New hook.
1247 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
1248 * targhooks.h (default_hard_regno_nregs): Declare.
1249 * targhooks.c (default_hard_regno_nregs): New function.
1250 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
1251 (TARGET_HARD_REGNO_NREGS): ...this hook.
1252 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
1253 (CLASS_MAX_NREGS): Likewise.
1254 * doc/tm.texi: Regenerate.
1255 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
1256 instead of HARD_REGNO_NREGS.
1257 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
1258 HARD_REGNO_NREGS in the comment.
1259 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
1260 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
1261 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
1262 Return an unsigned int.
1263 (TARGET_HARD_REGNO_NREGS): Redefine.
1264 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
1265 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
1266 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
1267 (arc_hard_regno_nregs): New function.
1268 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
1269 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
1270 (arm_hard_regno_nregs): New function.
1271 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
1272 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
1273 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
1274 (TARGET_HARD_REGNO_NREGS): Redefine.
1275 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
1276 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
1277 (HARD_REGNO_NREGS): Delete.
1278 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
1279 (cr16_hard_regno_nregs): New function.
1280 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
1281 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
1282 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
1283 (cris_hard_regno_nregs): New function.
1284 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
1285 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
1286 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
1287 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
1288 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
1289 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
1290 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
1291 (frv_hard_regno_nregs): Make static. Take and return an
1293 (frv_class_max_nregs): Remove outdated copy of documentation.
1294 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
1295 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
1296 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
1297 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
1298 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
1299 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
1300 (TARGET_HARD_REGNO_NREGS): Redefine.
1301 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
1302 (CLASS_MAX_NREGS): Update comment.
1303 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
1304 (ia64_hard_regno_nregs): New function.
1305 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
1306 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
1307 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
1308 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
1309 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
1311 (m32c_hard_regno_nregs): Likewise. Make static.
1312 (TARGET_HARD_REGNO_NREGS): Redefine.
1313 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
1314 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
1315 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
1316 (m68k_hard_regno_nregs): New function.
1317 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
1318 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
1319 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
1320 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
1321 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
1322 Take and return an unsigned int.
1323 (TARGET_HARD_REGNO_NREGS): Redefine.
1324 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
1325 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
1326 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
1327 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
1328 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
1329 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
1330 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
1331 (msp430_hard_regno_nregs): Make static. Take and return an
1333 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
1334 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
1335 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
1336 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
1337 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
1338 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
1339 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
1340 (TARGET_HARD_REGNO_NREGS): Redefine.
1341 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
1342 (PA_HARD_REGNO_NREGS): ...this.
1343 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
1344 (PA_HARD_REGNO_NREGS): ...this.
1345 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
1346 (pa_hard_regno_nregs): New function.
1347 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
1348 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
1349 (pdp11_hard_regno_nregs): New function.
1350 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
1351 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
1352 (rs6000_hard_regno_nregs_hook): New function.
1353 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
1354 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
1355 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
1356 Take and return an unsigned int. Move earlier in file.
1357 (TARGET_HARD_REGNO_NREGS): Redefine.
1358 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
1359 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
1360 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
1361 (rl78_hard_regno_nregs): Make static. Take and return an
1363 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
1364 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
1365 (rs6000_hard_regno_nregs_hook): New function.
1366 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
1367 * config/rx/rx.c (rx_hard_regno_nregs): New function.
1368 (TARGET_HARD_REGNO_NREGS): Redefine.
1369 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
1370 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
1371 instead of HARD_REGNO_NREGS.
1372 (s390_hard_regno_nregs): New function.
1373 (s390_hard_regno_mode_ok): Add comment from s390.h.
1374 (TARGET_HARD_REGNO_NREGS): Redefine.
1375 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
1376 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
1377 (sh_hard_regno_nregs): New function.
1378 (sh_pass_in_reg_p): Use it.
1379 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
1380 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
1381 (sparc_hard_regno_nregs): New function.
1382 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
1383 * config/spu/spu.c (spu_hard_regno_nregs): New function.
1384 (spu_function_arg_advance): Use it, supplying a valid register number.
1385 (TARGET_HARD_REGNO_NREGS): Redefine.
1386 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
1387 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
1388 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
1389 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
1390 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
1391 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
1392 (CLASS_MAX_NREGS): Remove copy of old documentation.
1393 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
1394 (visium_hard_regno_nregs): New function.
1395 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
1396 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
1397 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
1398 xtensa_hard_regno_nregs): New function.
1399 * system.h (HARD_REGNO_NREGS): Poison.
1401 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1403 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
1404 hard_regno_nregs instead of HARD_REGNO_NREGS.
1405 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
1406 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
1407 (c6x_expand_epilogue): Likewise.
1408 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
1409 (frv_read_iacc_argument): Likewise.
1410 * config/sh/sh.c: Include regs.h.
1411 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
1412 (regs_used): Likewise.
1413 (output_stack_adjust): Likewise.
1414 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
1415 * expmed.c: Include regs.h.
1416 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
1417 * ree.c: Include regs.h.
1418 (combine_reaching_defs): Use hard_regno_nregs instead of
1420 (add_removable_extension): Likewise.
1422 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1424 * regs.h (hard_regno_nregs): Turn into a function.
1425 (end_hard_regno): Update accordingly.
1426 * caller-save.c (setup_save_areas): Likewise.
1427 (save_call_clobbered_regs): Likewise.
1428 (replace_reg_with_saved_mem): Likewise.
1429 (insert_restore): Likewise.
1430 (insert_save): Likewise.
1431 * combine.c (can_change_dest_mode): Likewise.
1432 (move_deaths): Likewise.
1433 (distribute_notes): Likewise.
1434 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
1435 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
1436 (rs6000_split_multireg_move): Likewise.
1437 (rs6000_register_move_cost): Likewise.
1438 (rs6000_memory_move_cost): Likewise.
1439 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
1440 (rs6000_split_multireg_move): Likewise.
1441 (rs6000_register_move_cost): Likewise.
1442 (rs6000_memory_move_cost): Likewise.
1443 * cselib.c (cselib_reset_table): Likewise.
1444 (cselib_lookup_1): Likewise.
1445 * emit-rtl.c (set_mode_and_regno): Likewise.
1446 * function.c (aggregate_value_p): Likewise.
1447 * ira-color.c (setup_profitable_hard_regs): Likewise.
1448 (check_hard_reg_p): Likewise.
1449 (calculate_saved_nregs): Likewise.
1450 (assign_hard_reg): Likewise.
1451 (improve_allocation): Likewise.
1452 (calculate_spill_cost): Likewise.
1453 * ira-emit.c (modify_move_list): Likewise.
1454 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
1455 (ira_hard_reg_in_set_p): Likewise.
1456 * ira.c (setup_reg_mode_hard_regset): Likewise.
1457 (clarify_prohibited_class_mode_regs): Likewise.
1458 (check_allocation): Likewise.
1459 * lra-assigns.c (find_hard_regno_for_1): Likewise.
1460 (lra_setup_reg_renumber): Likewise.
1461 (setup_try_hard_regno_pseudos): Likewise.
1462 (spill_for): Likewise.
1463 (assign_hard_regno): Likewise.
1464 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
1465 * lra-constraints.c (in_class_p): Likewise.
1466 (lra_constraint_offset): Likewise.
1467 (simplify_operand_subreg): Likewise.
1468 (lra_constraints): Likewise.
1469 (split_reg): Likewise.
1470 (split_if_necessary): Likewise.
1471 (invariant_p): Likewise.
1472 (inherit_in_ebb): Likewise.
1473 * lra-lives.c (process_bb_lives): Likewise.
1474 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
1475 (get_hard_regs): Likewise.
1476 (do_remat): Likewise.
1477 * lra-spills.c (assign_spill_hard_regs): Likewise.
1478 * mode-switching.c (create_pre_exit): Likewise.
1479 * postreload.c (reload_combine_recognize_pattern): Likewise.
1480 * recog.c (peep2_find_free_register): Likewise.
1481 * regcprop.c (kill_value_regno): Likewise.
1482 (set_value_regno): Likewise.
1483 (copy_value): Likewise.
1484 (maybe_mode_change): Likewise.
1485 (find_oldest_value_reg): Likewise.
1486 (copyprop_hardreg_forward_1): Likewise.
1487 * regrename.c (check_new_reg_p): Likewise.
1488 (regrename_do_replace): Likewise.
1489 * reload.c (push_reload): Likewise.
1490 (combine_reloads): Likewise.
1491 (find_dummy_reload): Likewise.
1492 (operands_match_p): Likewise.
1493 (find_reloads): Likewise.
1494 (find_equiv_reg): Likewise.
1495 (reload_adjust_reg_for_mode): Likewise.
1496 * reload1.c (count_pseudo): Likewise.
1497 (count_spilled_pseudo): Likewise.
1498 (find_reg): Likewise.
1499 (clear_reload_reg_in_use): Likewise.
1500 (free_for_value_p): Likewise.
1501 (allocate_reload_reg): Likewise.
1502 (choose_reload_regs): Likewise.
1503 (reload_adjust_reg_for_temp): Likewise.
1504 (emit_reload_insns): Likewise.
1505 (delete_output_reload): Likewise.
1506 * rtlanal.c (subreg_get_info): Likewise.
1507 * sched-deps.c (sched_analyze_reg): Likewise.
1508 * sel-sched.c (init_regs_for_mode): Likewise.
1509 (mark_unavailable_hard_regs): Likewise.
1510 (choose_best_reg_1): Likewise.
1511 (verify_target_availability): Likewise.
1512 * valtrack.c (dead_debug_insert_temp): Likewise.
1513 * var-tracking.c (track_loc_p): Likewise.
1514 (emit_note_insn_var_location): Likewise.
1515 * varasm.c (make_decl_rtl): Likewise.
1516 * reginfo.c (choose_hard_reg_mode): Likewise.
1517 (init_reg_modes_target): Refer directly to
1518 this_target_regs->x_hard_regno_nregs.
1520 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1522 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
1523 instead of hard_regno_nregs.
1525 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1527 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
1528 end_hard_regno instead of hard_regno_nregs.
1529 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
1530 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
1531 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
1532 * ira-color.c (improve_allocation): Likewise.
1533 * lra-assigns.c (find_hard_regno_for_1): Likewise.
1534 * lra-lives.c (mark_regno_live): Likewise.
1535 (mark_regno_dead): Likewise.
1536 * lra-remat.c (operand_to_remat): Likewise.
1537 * lra.c (collect_non_operand_hard_regs): Likewise.
1538 * postreload.c (reload_combine_note_store): Likewise.
1539 (move2add_valid_value_p): Likewise.
1540 * reload.c (regno_clobbered_p): Likewise.
1542 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1544 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
1546 * config/v850/v850.c (v850_reorg): Likewise.
1547 * reload.c (refers_to_regno_for_reload_p): Likewise.
1548 (find_equiv_reg): Likewise.
1549 * reload1.c (reload_reg_reaches_end_p): Likewise.
1551 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1553 * caller-save.c (add_used_regs): Use REG_NREGS instead of
1555 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
1556 * config/arm/arm.c (output_move_neon): Likewise.
1557 (arm_attr_length_move_neon): Likewise.
1558 (neon_split_vcombine): Likewise.
1559 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
1560 (c6x_mark_reg_written): Likewise.
1561 (c6x_dwarf_register_span): Likewise.
1562 * config/i386/i386.c (ix86_save_reg): Likewise.
1563 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
1564 (rws_access_reg): Likewise.
1565 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
1566 * mode-switching.c (create_pre_exit): Likewise.
1567 * ree.c (combine_reaching_defs): Likewise.
1568 (add_removable_extension): Likewise.
1569 * regcprop.c (find_oldest_value_reg): Likewise.
1570 (copyprop_hardreg_forward_1): Likewise.
1571 * reload.c (reload_inner_reg_of_subreg): Likewise.
1572 (push_reload): Likewise.
1573 (combine_reloads): Likewise.
1574 (find_dummy_reload): Likewise.
1575 (reload_adjust_reg_for_mode): Likewise.
1576 * reload1.c (find_reload_regs): Likewise.
1577 (forget_old_reloads_1): Likewise.
1578 (reload_reg_free_for_value_p): Likewise.
1579 (reload_adjust_reg_for_temp): Likewise.
1580 (emit_reload_insns): Likewise.
1581 (delete_output_reload): Likewise.
1582 * sel-sched.c (choose_best_reg_1): Likewise.
1583 (choose_best_pseudo_reg): Likewise.
1585 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1586 Alan Hayward <alan.hayward@arm.com>
1587 David Sherwood <david.sherwood@arm.com>
1589 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
1590 * target.def (slow_unaligned_access): New hook.
1591 * targhooks.h (default_slow_unaligned_access): Declare.
1592 * targhooks.c (default_slow_unaligned_access): New function.
1593 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
1594 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
1595 * doc/tm.texi: Regenerate.
1596 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
1597 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
1598 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
1600 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
1601 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
1603 (rs6000_slow_unaligned_access): New function.
1604 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
1605 (expand_block_compare): Likewise.
1606 (expand_strn_compare): Likewise.
1607 (rs6000_rtx_costs): Likewise.
1608 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
1609 (riscv_slow_unaligned_access): Likewise.
1610 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
1611 (riscv_slow_unaligned_access_p): ...this and make static.
1612 (riscv_option_override): Update accordingly.
1613 (riscv_slow_unaligned_access): New function.
1614 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
1615 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
1616 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
1617 (rs6000_slow_unaligned_access): New function.
1618 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
1619 (rs6000_rtx_costs): Likewise.
1620 * config/rs6000/rs6000-string.c (expand_block_compare)
1621 (expand_strn_compare): Use targetm.slow_unaligned_access instead
1622 of SLOW_UNALIGNED_ACCESS.
1623 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
1624 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
1625 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
1626 of SLOW_UNALIGNED_ACCESS.
1627 * expmed.c (simple_mem_bitfield_p): Likewise.
1628 * expr.c (alignment_for_piecewise_move): Likewise.
1629 (emit_group_load_1): Likewise.
1630 (emit_group_store): Likewise.
1631 (copy_blkmode_from_reg): Likewise.
1632 (emit_push_insn): Likewise.
1633 (expand_assignment): Likewise.
1634 (store_field): Likewise.
1635 (expand_expr_real_1): Likewise.
1636 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
1637 * lra-constraints.c (simplify_operand_subreg): Likewise.
1638 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
1639 * gimple-ssa-store-merging.c: Likewise in block comment at start
1641 * tree-ssa-strlen.c: Include target.h.
1642 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
1643 of SLOW_UNALIGNED_ACCESS.
1644 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
1646 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1648 PR rtl-optimization/82185
1649 * expmed.c (emit_store_flag_int): Only test tem if it has been
1652 2017-09-12 Richard Biener <rguenther@suse.de>
1655 * match.pd ((FTYPE) N CMP CST): Fix typo.
1657 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
1659 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
1661 (mips_near_type_p): Add 'short_call' attribute as a synonym
1663 * doc/extend.texi (short_call): Document new function attribute.
1665 2017-09-12 Jakub Jelinek <jakub@redhat.com>
1668 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
1669 assertion check that in the condition.
1670 (get_atomic_generic_size): Likewise. Before testing if parameter
1671 has pointer type, if it has array type, call for C++
1672 default_conversion to perform array-to-pointer conversion.
1674 2017-09-12 Richard Biener <rguenther@suse.de>
1676 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
1677 for operations we cannot scalarize.
1679 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
1681 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
1682 vectors heap vectors. Clean up comments.
1683 Make visited_bbs a reference.
1684 (profitable_jump_thread_path): Make GC
1685 vectors heap vectors. Clean up comments.
1687 (convert_and_register_jump_thread_path): Make GC vectors heap
1689 (check_subpath_and_update_thread_path): Same. Clean up comments.
1690 Make visited_bbs a reference.
1691 (handle_phi): Abstract common code to to
1692 register_jump_thread_path_if_profitable.
1693 Rename VAR_BB to DEF_BB.
1695 Make GC vectors heap vectors.
1696 Make visited_bbs a reference.
1697 (handle_assignment): Same.
1698 (register_jump_thread_path_if_profitable): New.
1699 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
1701 Make GC vectors heap vectors. Clean up comments.
1702 Make visited_bbs a reference.
1703 (find_jump_threads_backwards): Make visited_bbs live in the stack.
1704 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
1707 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
1710 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
1711 words of E_DImode object are reachable by xtensa_uimm8x4 access.
1713 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
1715 Revert r251800 and r251799.
1717 2017-09-11 Martin Jambor <mjambor@suse.cz>
1720 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
1721 arguments in advance.
1722 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
1723 use it to find predecessor edges.
1724 (naive_outof_ssa): Collect vector of predecessors.
1726 2017-09-08 Jason Merrill <jason@redhat.com>
1728 PR c++/70029 - ICE with ref-qualifier and -flto
1729 * langhooks.h (struct lang_hooks_for_types): Add
1730 copy_lang_qualifiers.
1731 * attribs.c (build_type_attribute_qual_variant): Use it.
1732 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
1734 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
1735 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
1737 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
1740 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
1741 (*mulsi3_sp64): New instruction.
1742 (mulsi3): New expander.
1744 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
1746 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
1748 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1750 * sancov.c: Include memmodel.h.
1752 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
1755 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
1758 2017-09-07 Carl Love <cel@us.ibm.com>
1760 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
1761 the sldi instruction.
1763 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
1765 * sancov.c: Include tm_p.h.
1767 2017-09-07 Jakub Jelinek <jakub@redhat.com>
1770 * output.h (switch_to_other_text_partition): New declaration.
1771 * varasm.c (switch_to_other_text_partition): New function.
1772 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
1773 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
1774 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
1775 to the other text partition before emitting LCL label and switch back
1776 after emitting the word after it.
1778 2017-09-07 Richard Biener <rguenther@suse.de>
1780 * passes.def (pass_split_crit_edges): Remove instance before PRE.
1781 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
1782 critical edges here, after loop init.
1783 (pass_data_pre): Remove PROP_no_crit_edges flags.
1784 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
1785 for valueization of call args to avoid leaking VN_TOP.
1786 (visit_use): Assert we do not visit default defs.
1787 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
1788 Use error_mark_node to more easily detect leaking VN_TOP.
1789 All default-defs are varying, not VN_TOP. Mark them visited.
1790 (run_scc_vn): Make code match comment.
1792 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
1794 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
1795 OPTION_MASK_FLOAT128_KEYWORD.
1796 (POWERPC_MASKS): Likewise.
1797 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
1798 support for the -mfloat128-type option, and make -mfloat128
1799 default on PowerPC Linux systems. Define or undefine
1800 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
1801 Define __float128 to be __ieee128 if IEEE 128-bit support is
1802 enabled, or undefine it.
1803 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
1804 Delete defining __FLOAT128_TYPE__.
1805 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
1806 -mfloat128-type option and make -mfloat128 default on PowerPC
1808 (TARGET_FLOAT128_TYPE): Likewise.
1809 (-mfloat128-type): Likewise.
1810 * config/rs6000/rs6000.c (rs6000_option_override_internal):
1811 Delete the -mfloat128-type option and make -mfloat128 default on
1812 PowerPC Linux systems. Always use __ieee128 to be the keyword for
1813 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
1814 128-bit floating point is enabled. Change tests from using
1815 -mfloat128-type to -mfloat128.
1816 (rs6000_mangle_type): Use the correct mangling for the __float128
1817 type even if normal long double is restricted to 64-bits.
1818 (floatn_mode): Enable the _Float128 type by default on VSX Linux
1820 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
1821 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
1822 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
1824 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
1825 documentation for -mfloat128.
1827 2017-09-06 Olivier Hainque <hainque@adacore.com>
1829 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
1831 2017-09-06 Wish Wu <wishwu007@gmail.com>
1832 Jakub Jelinek <jakub@redhat.com>
1834 * asan.c (initialize_sanitizer_builtins): Add
1835 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
1836 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
1837 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
1838 BT_FN_VOID_UINT64_PTR variables.
1839 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
1840 (BT_FN_VOID_UINT16_UINT16): Likewise.
1841 (BT_FN_VOID_UINT32_UINT32): Likewise.
1842 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
1843 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
1844 (BT_FN_VOID_UINT64_PTR): Likewise.
1845 * common.opt (flag_sanitize_coverage): New variable.
1846 (fsanitize-coverage=trace-pc): Remove.
1847 (fsanitize-coverage=): Add.
1848 * flag-types.h (enum sanitize_coverage_code): New enum.
1849 * fold-const.c (fold_range_test): Disable non-short-circuit
1850 optimization if flag_sanitize_coverage.
1851 (fold_truth_andor): Likewise.
1852 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
1853 * opts.c (COVERAGE_SANITIZER_OPT): Define.
1854 (coverage_sanitizer_opts): New array.
1855 (get_closest_sanitizer_option): Add OPTS argument, handle also
1856 OPT_fsanitize_coverage_.
1857 (parse_sanitizer_options): Adjusted to also handle
1858 OPT_fsanitize_coverage_.
1859 (common_handle_option): Add OPT_fsanitize_coverage_.
1860 * sancov.c (instrument_comparison, instrument_switch): New function.
1861 (sancov_pass): Add trace-cmp support.
1862 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
1863 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
1864 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
1865 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
1866 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
1867 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
1868 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
1869 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
1870 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
1871 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
1873 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
1875 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
1876 error. Only quit immediately if parsing is complete.
1877 (BEGIN): Initialize fatal_err and parse_done.
1878 (begin fpu, end fpu): Check number of arguments.
1879 (begin arch, end arch): Likewise.
1880 (begin cpu, end cpu): Likewise.
1881 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
1882 (optalias): Likewise.
1884 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
1886 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
1887 * config/arm/arm-isa.h: Delete. Move definitions to ...
1888 * arm-cpus.in: ... here. Use new feature and fgroup values.
1889 * config/arm/arm.c (arm_option_override): Use lower case for feature
1891 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
1892 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
1893 * config/arm/parsecpu.awk (END): Add new command 'isa'.
1895 (print_isa_bits_for): New function.
1896 (gen_isa): New function.
1897 (gen_comm_data): Use print_isa_bits_for.
1898 (define feature): New keyword.
1899 (define fgroup): New keyword.
1900 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
1901 (arm-isa.h): Add rule to generate file.
1902 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
1903 case for feature bit names.
1905 2017-09-06 Richard Biener <rguenther@suse.de>
1907 * tree-ssa-pre.c (NECESSARY): Remove.
1908 (create_expression_by_pieces): Do not touch pass-local flags.
1909 (insert_into_preds_of_block): Likewise.
1910 (do_pre_regular_insertion): Likewise.
1911 (eliminate_insert): Likewise.
1912 (eliminate_dom_walker::before_dom_children): Likewise.
1913 (fini_eliminate): Do not look at inserted_exprs.
1914 (mark_operand_necessary): Remove.
1915 (remove_dead_inserted_code): Replace with simple work-list
1916 algorithm based on inserted_exprs and SSA uses.
1917 (pass_pre::execute): Re-order fini_eliminate and
1918 remove_dead_inserted_code.
1920 2017-09-06 Olivier Hainque <hainque@adacore.com>
1922 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
1923 for VxWorks 7. Adjust surrounding comments.
1925 2017-09-06 Richard Biener <rguenther@suse.de>
1927 * gimple-ssa-strength-reduction.c
1928 (find_candidates_dom_walker::before_dom_children): Also allow
1931 2017-09-06 Richard Biener <rguenther@suse.de>
1933 PR tree-optimization/82108
1934 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
1935 for gap in the non-permutation SLP case.
1937 2017-09-06 Martin Jambor <mjambor@suse.cz>
1939 PR tree-optimization/82078
1940 * tree-sra.c (sort_and_splice_var_accesses): Move call to
1941 add_access_to_work_queue...
1942 (build_accesses_from_assign): ...here.
1943 (propagate_all_subaccesses): Make sure racc is the group
1944 representative, if there is one.
1946 2017-09-06 Jakub Jelinek <jakub@redhat.com>
1949 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
1952 2017-09-06 Richard Biener <rguenther@suse.de>
1954 * gimple-ssa-strength-reduction.c
1955 (find_candidates_dom_walker::before_doom_children): Use a
1956 type and not a mode check.
1958 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
1961 * config/arm/predicates.md (arm_general_adddi_operand): Create new
1963 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
1965 2017-09-05 Jeff Law <law@redhat.com>
1967 PR tree-optimization/64910
1968 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
1969 cases where we have 3 or more operands.
1971 2017-09-05 Jakub Jelinek <jakub@redhat.com>
1974 * omp-low.c (lower_omp_for): Recompute tree invariant if
1975 gimple_omp_for_initial/final is ADDR_EXPR.
1978 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
1979 into gimple val before gimplification fo the COND_EXPR.
1981 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
1983 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
1984 REGION_COPY argument.
1985 (thread_through_all_blocks): Remove unused argument to
1986 duplicate_thread_path.
1988 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1989 Alan Hayward <alan.hayward@arm.com>
1990 David Sherwood <david.sherwood@arm.com>
1992 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
1993 Take a scalar_mode rather than a machine_mode.
1994 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
1995 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
1996 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
1997 (aarch64_gen_adjusted_ldpstp): Likewise.
1998 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
2000 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2001 Alan Hayward <alan.hayward@arm.com>
2002 David Sherwood <david.sherwood@arm.com>
2004 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
2005 Take a scalar_int_mode instead of a machine_mode.
2006 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
2007 (aarch64_output_scalar_simd_mov_immediate): Likewise.
2008 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
2009 (aarch64_simd_attr_length_rglist): Delete.
2010 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
2011 a scalar_int_mode instead of a machine_mode.
2012 (aarch64_add_offset): Likewise.
2013 (aarch64_internal_mov_immediate): Likewise
2014 (aarch64_add_constant_internal): Likewise.
2015 (aarch64_add_constant): Likewise.
2016 (aarch64_movw_imm): Likewise.
2017 (aarch64_rtx_arith_op_extract_p): Likewise.
2018 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
2019 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
2020 Remove assert that the mode isn't a vector.
2021 (aarch64_output_scalar_simd_mov_immediate): Likewise.
2022 (aarch64_expand_mov_immediate): Update calls after above changes.
2023 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
2024 (aarch64_and_bitmask_imm): Check for scalar integer modes.
2025 (aarch64_move_imm): Likewise.
2026 (aarch64_can_const_movi_rtx_p): Likewise.
2027 (aarch64_strip_extend): Likewise.
2028 (aarch64_extr_rtx_p): Likewise.
2029 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
2030 a CONST_INT when the mode parameter is VOIDmode.
2031 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
2033 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2035 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
2036 * stor-layout.c (bitwise_mode_for_mode): Likewise.
2037 (bitwise_type_for_mode): Update accordingly.
2039 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2041 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
2042 * stor-layout.c (mode_for_size_tree): Likewise.
2043 (mode_for_array): Update accordingly.
2044 (layout_decl): Likewise.
2045 (compute_record_mode): Likewise. Only set the mode once.
2047 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2049 * target.def (get_mask_mode): Change return type to opt_mode.
2051 * doc/tm.texi: Regenerate.
2052 * targhooks.h (default_get_mask_mode): Return an opt_mode.
2053 * targhooks.c (default_get_mask_mode): Likewise.
2054 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
2055 * optabs-query.c (can_vec_mask_load_store_p): Update use of
2056 targetm.get_mask_mode.
2057 * tree.c (build_truth_vector_type): Likewise.
2059 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2061 * machmode.h (mode_for_vector): Return an opt_mode.
2062 * stor-layout.c (mode_for_vector): Likewise.
2063 (mode_for_int_vector): Update accordingly.
2064 (layout_type): Likewise.
2065 * config/i386/i386.c (emit_memmov): Likewise.
2066 (ix86_expand_set_or_movmem): Likewise.
2067 (ix86_expand_vector_init): Likewise.
2068 (ix86_get_mask_mode): Likewise.
2069 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
2071 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
2072 * expmed.c (extract_bit_field_1): Likewise.
2073 * expr.c (expand_expr_real_2): Likewise.
2074 * optabs-query.c (can_vec_perm_p): Likewise.
2075 (can_vec_mask_load_store_p): Likewise.
2076 * optabs.c (expand_vec_perm): Likewise.
2077 * targhooks.c (default_get_mask_mode): Likewise.
2078 * tree-vect-stmts.c (vectorizable_store): Likewise.
2079 (vectorizable_load): Likewise.
2080 (get_vectype_for_scalar_type_and_size): Likewise.
2082 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2084 * machmode.h (mode_for_int_vector): New function.
2085 * stor-layout.c (mode_for_int_vector): Likewise.
2086 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
2087 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
2088 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
2089 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
2090 (s390_expand_vcond): Likewise.
2092 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2094 * machmode.h (opt_machine_mode): New type.
2095 (opt_mode<T>): Allow construction from anything that can be
2097 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
2098 (mode_for_size): Return an opt_machine_mode.
2099 * stor-layout.c (mode_for_size): Likewise.
2100 (mode_for_size_tree): Update call accordingly.
2101 (bitwise_mode_for_mode): Likewise.
2102 (make_fract_type): Likewise.
2103 (make_accum_type): Likewise.
2104 * caller-save.c (replace_reg_with_saved_mem): Update call
2106 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
2107 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
2108 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
2109 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
2110 * expmed.c (extract_bit_field_1): Likewise.
2111 * reload.c (get_secondary_mem): Likewise.
2112 * varasm.c (assemble_integer): Likewise.
2113 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
2116 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2118 * machmode.h (decimal_float_mode_for_size): New function.
2119 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
2120 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
2121 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
2122 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
2123 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
2124 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
2126 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2128 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
2129 (get_builtin_sync_mode): Likewise.
2130 (expand_ifn_atomic_compare_exchange): Likewise.
2131 (expand_builtin_atomic_clear): Likewise.
2132 (expand_builtin_atomic_test_and_set): Likewise.
2133 (fold_builtin_atomic_always_lock_free): Likewise.
2134 * calls.c (compute_argument_addresses): Likewise.
2135 (emit_library_call_value_1): Likewise.
2136 (store_one_arg): Likewise.
2137 * combine.c (combine_instructions): Likewise.
2138 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
2139 * config/arm/arm.c (arm_function_value): Likewise.
2140 (aapcs_allocate_return_reg): Likewise.
2141 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
2142 * config/i386/i386.c (construct_container): Likewise.
2143 (ix86_gimplify_va_arg): Likewise.
2144 (ix86_expand_sse_cmp): Likewise.
2145 (emit_memmov): Likewise.
2146 (emit_memset): Likewise.
2147 (expand_small_movmem_or_setmem): Likewise.
2148 (ix86_expand_pextr): Likewise.
2149 (ix86_expand_pinsr): Likewise.
2150 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
2151 * config/microblaze/microblaze.c (microblaze_block_move_straight):
2153 * config/mips/mips.c (mips_function_value_1) Likewise.
2154 (mips_block_move_straight): Likewise.
2155 (mips_expand_ins_as_unaligned_store): Likewise.
2156 * config/powerpcspe/powerpcspe.c
2157 (rs6000_darwin64_record_arg_advance_flush): Likewise.
2158 (rs6000_darwin64_record_arg_flush): Likewise.
2159 * config/rs6000/rs6000.c
2160 (rs6000_darwin64_record_arg_advance_flush): Likewise.
2161 (rs6000_darwin64_record_arg_flush): Likewise.
2162 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
2163 (sparc_function_value_1): Likewise.
2164 * config/spu/spu.c (adjust_operand): Likewise.
2165 (spu_emit_branch_or_set): Likewise.
2166 (arith_immediate_p): Likewise.
2167 * emit-rtl.c (gen_lowpart_common): Likewise.
2168 * expr.c (expand_expr_real_1): Likewise.
2169 * function.c (assign_parm_setup_block): Likewise.
2170 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
2171 * reload1.c (alter_reg): Likewise.
2172 * stor-layout.c (mode_for_vector): Likewise.
2173 (layout_type): Likewise.
2175 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2177 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
2178 (spu_convert_move): Likewise.
2179 * lower-subreg.c (resolve_simple_move): Likewise.
2181 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2184 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
2185 define_insn to a define_expand.
2186 (altivec_vsum2sws_direct): New define_insn.
2187 (altivec_vsumsws): Convert from a define_insn to a define_expand.
2189 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
2191 * config/arm/arm.c (arm_option_params_internal): Improve setting of
2194 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
2199 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
2200 realign stack if stack alignment needed is less than incoming
2203 2017-09-05 Marek Polacek <polacek@redhat.com>
2206 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
2209 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
2211 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
2213 2017-09-05 Richard Biener <rguenther@suse.de>
2215 PR tree-optimization/82084
2216 * fold-const.c (can_native_encode_string_p): Handle wide characters.
2218 2017-09-05 Richard Biener <rguenther@suse.de>
2220 PR tree-optimization/82102
2221 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
2223 2017-09-05 Martin Liska <mliska@suse.cz>
2225 PR tree-optimization/82032
2226 * tree-cfg.c (generate_range_test): New function.
2227 * tree-cfg.h (generate_range_test): Declared here.
2228 * tree-cfgcleanup.c (convert_single_case_switch): New function.
2229 (cleanup_control_expr_graph): Use it.
2230 * tree-switch-conversion.c (try_switch_expansion): Remove
2232 (emit_case_nodes): Use generate_range_test.
2234 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
2237 * config/i386/i386.md (*<btsc><mode>_mask): Add
2238 TARGET_USE_BT to insn constraint.
2239 (*btr<mode>_mask): Ditto.
2241 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
2243 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
2244 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
2246 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2249 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
2250 TARGET_NEON and TARGET_IWMMXT.
2251 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
2252 TARGET_NEON and TARGET_IWMMXT.
2253 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
2255 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
2257 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
2258 (ix86_rewrite_tls_address): Ditto.
2259 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
2260 (ix86_rewrite_tls_address_1): Ditto.
2261 (ix86_rewrite_tls_address): Ditto.
2262 * config/i386/predicates.md (tls_address_pattern): New predicate.
2263 * config/i386/i386.md (TLS address splitter): New splitter.
2265 2017-09-04 Richard Biener <rguenther@suse.de>
2267 PR tree-optimization/82084
2268 * fold-const.h (can_native_encode_string_p): Declare.
2269 * fold-const.c (can_native_encode_string_p): Factor out from ...
2270 (native_encode_string): ... here.
2271 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
2272 vectorizing stores from constants we later cannot handle.
2274 2017-09-04 Marek Polacek <polacek@redhat.com>
2277 * doc/invoke.texi: Update -Wtautological-compare documentation.
2279 2017-09-04 Jeff Law <law@redhat.com>
2281 PR tree-optimization/64910
2282 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
2283 swap the first and last operand if the last is a constant.
2285 2017-09-04 Marek Polacek <polacek@redhat.com>
2288 * convert.c (do_narrow): When sanitizing signed integer overflows,
2289 bail out for signed types.
2290 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
2292 2017-09-04 Richard Biener <rguenther@suse.de>
2294 PR tree-optimization/82060
2295 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
2296 Move devirtualization after stmt folding and before EH/AB/noreturn
2297 cleanup to get the stmt refs canonicalized. Use a bool instead
2298 of gimple_modified_p since that doesn't work for NOPs. Schedule
2299 NOPs generated by folding for removal.
2301 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2302 Alan Hayward <alan.hayward@arm.com>
2303 David Sherwood <david.sherwood@arm.com>
2305 * coretypes.h (pad_direction): New enum.
2306 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
2307 (FUNCTION_ARG_PADDING): Likewise.
2308 * target.def (function_arg_padding): New hook.
2309 * targhooks.h (default_function_arg_padding): Declare.
2310 * targhooks.c (default_function_arg_padding): New function.
2311 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
2312 (TARGET_FUNCTION_ARG_PADDING): ...this.
2313 * doc/tm.texi: Regenerate.
2314 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
2315 instead of direction.
2316 (compute_argument_addresses): Likewise.
2317 (load_register_parameters): Likewise.
2318 (emit_library_call_value_1): Likewise.
2319 (store_one_arg): Use targetm.calls.function_arg_padding instead
2320 of FUNCTION_ARG_PADDING.
2321 (must_pass_in_stack_var_size_or_pad): Likewise.
2322 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
2323 (emit_group_store): Likewise.
2324 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
2325 instead of FUNCTION_ARG_PADDING.
2326 (emit_push_insn): Likewise, and propagate enum change throughout
2328 * function.h (direction): Delete.
2329 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
2331 * function.c (assign_parm_find_stack_rtl): Likewise.
2332 (assign_parm_setup_block_p): Likewise.
2333 (assign_parm_setup_block): Likewise.
2334 (gimplify_parameters): Likewise.
2335 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
2336 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
2338 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
2339 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
2340 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
2341 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
2342 (aarch64_function_arg_padding): ...this new function.
2343 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
2344 (TARGET_FUNCTION_ARG_PADDING): Redefine.
2345 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
2346 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
2347 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
2348 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2349 (arm_pad_arg_upward): Replace with...
2350 (arm_function_arg_padding): ...this new function.
2351 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
2353 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
2354 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
2355 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2356 (ia64_hpux_function_arg_padding): Replace with...
2357 (ia64_function_arg_padding): ...this new function. Use pad_direction
2358 instead of direction. Check for TARGET_HPUX.
2359 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
2360 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2361 (iq2000_function_arg_padding): New function.
2362 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
2363 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
2364 (mips_function_arg_padding): ...this new function.
2365 (mips_pad_reg_upward): Update accordingly.
2366 (TARGET_FUNCTION_ARG_PADDING): Redefine.
2367 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
2368 targetm.calls.function_arg_padding.
2369 (FUNCTION_ARG_PADDING): Delete.
2370 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
2371 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
2372 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
2373 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
2374 (nios2_block_reg_padding): Return pad_direction instead of direction.
2375 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
2376 instead of direction.
2377 (nios2_function_arg_padding): Likewise. Make static.
2378 (TARGET_FUNCTION_ARG_PADDING): Redefine.
2379 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
2380 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
2381 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
2382 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2383 (pa_function_arg_padding): Make static. Return pad_direction instead
2385 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
2386 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
2387 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
2388 instead of direction. Use targetm.calls.function_arg_padding.
2389 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
2390 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
2391 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
2392 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
2393 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
2395 (function_arg_padding): Rename to...
2396 (rs6000_function_arg_padding): ...this. Make static. Return
2397 pad_direction instead of direction.
2398 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
2399 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
2400 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
2401 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
2402 instead of direction. Use targetm.calls.function_arg_padding.
2403 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
2404 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
2405 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
2406 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
2407 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2408 (function_arg_padding): Rename to...
2409 (rs6000_function_arg_padding): ...this. Make static. Return
2410 pad_direction instead of direction.
2411 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
2412 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
2413 * config/s390/s390.c (s390_function_arg_padding): New function.
2414 (TARGET_FUNCTION_ARG_PADDING): Redefine.
2415 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
2416 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
2417 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2418 (function_arg_padding): Rename to...
2419 (sparc_function_arg_padding): ...this. Make static. Return
2420 pad_direction instead of direction.
2421 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
2422 * config/spu/spu.c (spu_function_arg_padding): New function.
2423 (TARGET_FUNCTION_ARG_PADDING): Redefine.
2424 * system.h (FUNCTION_ARG_PADDING): Poison.
2426 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2427 Alan Hayward <alan.hayward@arm.com>
2428 David Sherwood <david.sherwood@arm.com>
2430 * target.def (modes_tieable_p): New hook.
2431 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
2432 (TARGET_MODES_TIEABLE_P): ...this.
2433 * doc/tm.texi.in: Regenerate.
2434 * hooks.h (hook_bool_mode_mode_true): Declare.
2435 * hooks.c (hook_bool_mode_mode_true): New function.
2436 * combine.c (subst): Use targetm.modes_tieable_p instead of
2438 * dse.c (find_shift_sequence): Likewise.
2439 * expmed.c (extract_low_bits): Likewise.
2440 * lower-subreg.c: Include target.h.
2441 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
2443 * rtlanal.c (rtx_cost): Likewise.
2444 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
2445 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
2446 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
2447 (TARGET_MODES_TIEABLE_P): Redefine.
2448 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
2449 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
2450 (TARGET_MODES_TIEABLE_P): Redefine.
2451 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
2452 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
2453 (arc_modes_tieable_p): New function.
2454 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
2455 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
2456 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
2457 (arm_modes_tieable_p): Make static.
2458 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
2459 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
2460 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
2461 (TARGET_MODES_TIEABLE_P): Redefine.
2462 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
2463 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
2464 (TARGET_MODES_TIEABLE_P): Redefine.
2465 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
2466 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
2467 (cr16_modes_tieable_p): New function.
2468 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
2469 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
2470 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
2471 (TRULY_NOOP_TRUNCATION): Update comment.
2472 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
2473 (TRULY_NOOP_TRUNCATION): Update comment.
2474 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
2475 (frv_modes_tieable_p): New function.
2476 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
2477 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
2478 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
2479 (TARGET_MODES_TIEABLE_P): Redefine.
2480 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
2481 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
2482 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
2483 (TARGET_MODES_TIEABLE_P): Redefine.
2484 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
2485 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
2486 (ia64_modes_tieable_p): New function.
2487 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
2488 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
2489 (iq2000_modes_tieable_p): New function.
2490 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
2491 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
2492 (lm32_modes_tieable_p): New function.
2493 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
2494 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
2495 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
2496 (TARGET_MODES_TIEABLE_P): Redefine.
2497 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
2498 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
2499 (m32r_modes_tieable_p): New function.
2500 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
2501 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
2502 (m68k_modes_tieable_p): New function.
2503 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
2504 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
2505 (mcore_modes_tieable_p): New function.
2506 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
2507 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
2509 (TARGET_MODES_TIEABLE_P): Redefine.
2510 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
2511 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
2512 * config/mips/mips.c (mips_modes_tieable_p): Make static.
2513 (TARGET_MODES_TIEABLE_P): Redefine.
2514 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
2515 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
2516 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
2517 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
2518 (mn10300_modes_tieable_p): ...this and make static.
2519 (TARGET_MODES_TIEABLE_P): Redefine.
2520 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
2521 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
2522 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
2523 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
2524 (msp430_modes_tieable_p): Make static.
2525 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
2526 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
2527 (TARGET_MODES_TIEABLE_P): Redefine.
2528 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
2529 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
2530 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
2531 (TARGET_MODES_TIEABLE_P): Redefine.
2532 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
2533 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
2534 * config/pa/pa.c (pa_modes_tieable_p): Make static.
2535 (TARGET_MODES_TIEABLE_P): Redefine.
2536 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
2537 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
2538 (pdp11_modes_tieable_p): New function.
2539 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
2540 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
2541 (rs6000_modes_tieable_p): New function.
2542 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
2543 * config/powerpcspe/powerpcspe.md: Update comment.
2544 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
2545 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
2546 (TARGET_MODES_TIEABLE_P): Redefine.
2547 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
2548 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
2549 (rl78_modes_tieable_p): New function.
2550 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
2551 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
2552 (rs6000_modes_tieable_p): New function.
2553 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
2554 * config/rs6000/rs6000.md: Update comment.
2555 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
2556 * config/rx/rx.c (rx_modes_tieable_p): New function.
2557 (TARGET_MODES_TIEABLE_P): Redefine.
2558 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
2559 * config/s390/s390.c (s390_modes_tieable_p): New function.
2560 (TARGET_MODES_TIEABLE_P): Redefine.
2561 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
2562 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
2563 (sh_modes_tieable_p): New function.
2564 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
2565 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
2566 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
2567 (sparc_modes_tieable_p): Make static.
2568 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
2569 * config/spu/spu.c (spu_modes_tieable_p): New function.
2570 (TARGET_MODES_TIEABLE_P): Redefine.
2571 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
2572 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
2573 (TARGET_MODES_TIEABLE_P): Redefine.
2574 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
2575 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
2576 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
2577 * config/v850/v850.c (v850_modes_tieable_p): New function.
2578 (TARGET_MODES_TIEABLE_P): Redefine.
2579 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
2580 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
2581 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
2582 (visium_modes_tieable_p): New function.
2583 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
2584 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
2585 (xtensa_modes_tieable_p): New function.
2586 * system.h (MODES_TIEABLE_P): Poison.
2588 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2589 Alan Hayward <alan.hayward@arm.com>
2590 David Sherwood <david.sherwood@arm.com>
2592 * target.def (hard_regno_mode_ok): New hook.
2593 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
2594 (TARGET_HARD_REGNO_MODE_OK): ...this.
2595 * doc/tm.texi.in: Regenerate.
2596 * hooks.h (hook_bool_uint_mode_true): Declare.
2597 * hooks.c (hook_bool_uint_mode_true): New function.
2598 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
2600 * genpreds.c (write_insn_preds_c): Add an include of target.h.
2601 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
2602 instead of HARD_REGNO_MODE_OK.
2603 * caller-save.c: Include target.h.
2604 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
2606 * combine.c (can_combine_p): Likewise.
2607 (combinable_i3pat): Likewise.
2608 (can_change_dest_mode): Likewise.
2609 * expr.c (init_expr_target): Likewise.
2610 (convert_move): Likewise.
2611 (convert_modes): Likewise.
2612 * ira.c (setup_prohibited_class_mode_regs): Likewise.
2613 (setup_prohibited_mode_move_regs): Likewise.
2614 * ira.h (target_ira): Likewise.
2615 * lra-assigns.c (find_hard_regno_for_1): Likewise.
2616 * lra-constraints.c (process_alt_operands): Likewise.
2617 (split_reg): Likewise.
2618 * recog.c (peep2_find_free_register): Likewise.
2619 * ree.c (combine_reaching_defs): Likewise.
2620 * regcprop.c (maybe_mode_change): Likewise.
2621 * reginfo.c (init_reg_sets_1): Likewise.
2622 (choose_hard_reg_mode): Likewise.
2623 (simplifiable_subregs): Likewise.
2624 * regrename.c (check_new_reg_p): Likewise.
2625 * reload.c (find_valid_class): Likewise.
2626 (find_valid_class_1): Likewise.
2627 (reload_inner_reg_of_subreg): Likewise.
2628 (push_reload): Likewise.
2629 (combine_reloads): Likewise.
2630 (find_dummy_reload): Likewise.
2631 (find_reloads): Likewise.
2632 * reload1.c (find_reg): Likewise.
2633 (set_reload_reg): Likewise.
2634 (allocate_reload_reg): Likewise.
2635 (choose_reload_regs): Likewise.
2636 (reload_adjust_reg_for_temp): Likewise.
2637 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
2638 (simplify_subreg_regno): Likewise.
2639 * sel-sched.c (init_regs_for_mode): Likewise.
2640 * varasm.c (make_decl_rtl): Likewise.
2641 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
2642 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
2644 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
2645 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
2646 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2647 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
2648 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
2649 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2650 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
2651 (arc_mode_class): Delete.
2652 (HARD_REGNO_MODE_OK): Delete.
2653 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2654 (arc_hard_regno_mode_ok): Rename old array to...
2655 (arc_hard_regno_mode_ok_modes): ...this.
2656 (arc_conditional_register_usage): Update accordingly.
2657 (arc_mode_class): Make static.
2658 (arc_hard_regno_mode_ok): New function.
2659 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
2660 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
2661 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2662 (arm_hard_regno_mode_ok): Make static.
2663 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
2665 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
2666 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
2667 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
2669 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2670 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
2671 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
2672 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
2673 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
2674 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2675 * config/bfin/predicates.md (valid_reg_operand): Use
2676 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
2677 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
2678 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
2679 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2680 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
2681 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
2682 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2683 (cr16_hard_regno_mode_ok): Make static and return a bool.
2684 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
2685 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2686 (cris_hard_regno_mode_ok): New function.
2687 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
2688 (epiphany_mode_class): Delete.
2689 (HARD_REGNO_MODE_OK): Delete.
2690 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
2691 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2692 (hard_regno_mode_ok): Rename to...
2693 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
2694 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
2695 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
2697 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
2698 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
2699 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2700 (frv_hard_regno_mode_ok): Make static and return a bool.
2701 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
2703 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
2704 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
2705 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
2706 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
2708 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2709 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
2710 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
2711 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
2713 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2714 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
2715 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2716 (ia64_hard_regno_mode_ok): New function.
2717 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
2718 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2719 (iq2000_hard_regno_mode_ok): New function.
2720 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
2721 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2722 (lm32_hard_regno_mode_ok): New function.
2723 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
2724 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
2725 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
2726 instead of HARD_REGNO_MODE_OK.
2727 (m32c_hard_regno_ok): Rename to...
2728 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
2729 (m32c_cannot_change_mode_class): Update accordingly.
2730 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2731 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
2732 (m32r_mode_class): Delete.
2733 (HARD_REGNO_MODE_OK): Delete.
2734 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2735 (m32r_hard_regno_mode_ok): Rename to...
2736 (m32r_hard_regno_modes): ...this.
2737 (m32r_mode_class): Make static.
2738 (m32r_hard_regno_mode_ok): New function.
2739 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
2740 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
2741 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2742 (m68k_hard_regno_mode_ok): Make static.
2743 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
2744 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2745 (mcore_hard_regno_mode_ok): New function.
2746 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
2747 (HARD_REGNO_MODE_OK): Delete.
2748 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
2750 (microblaze_hard_regno_mode_ok_p): ...this and make static.
2751 (microblaze_hard_regno_mode_ok): New function.
2752 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2753 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
2754 (mips_hard_regno_mode_ok): Delete.
2755 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
2756 (mips_hard_regno_mode_ok_p): ...this and make static.
2757 (mips_hard_regno_mode_ok_p): Rename to...
2758 (mips_hard_regno_mode_ok_uncached): ...this.
2759 (mips_hard_regno_mode_ok): New function.
2760 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
2761 of HARD_REGNO_MODE_OK.
2762 (mips_option_override): Update after above name changes.
2763 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2764 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
2765 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
2766 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
2767 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
2768 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2769 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
2770 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
2771 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
2772 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2773 (msp430_hard_regno_mode_ok): Make static and return a bool.
2774 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
2775 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
2776 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
2778 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2779 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
2780 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
2781 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
2782 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
2783 (PA_HARD_REGNO_MODE_OK): ...this
2784 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
2785 (PA_HARD_REGNO_MODE_OK): ...this.
2786 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2787 (pa_hard_regno_mode_ok): New function.
2788 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
2789 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2790 (pdp11_hard_regno_mode_ok): New function.
2791 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
2792 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
2794 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
2796 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2797 (rs6000_hard_regno_mode_ok): Rename to...
2798 (rs6000_hard_regno_mode_ok_uncached): ...this.
2799 (rs6000_init_hard_regno_mode_ok): Update accordingly.
2800 (rs6000_hard_regno_mode_ok): New function.
2801 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
2802 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
2803 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
2804 (riscv_hard_regno_mode_ok): ...this and make static.
2805 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2806 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
2807 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
2808 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2809 (rl78_hard_regno_mode_ok): Make static and return bool.
2810 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
2811 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
2813 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
2814 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2815 (rs6000_hard_regno_mode_ok): Rename to...
2816 (rs6000_hard_regno_mode_ok_uncached): ...this.
2817 (rs6000_init_hard_regno_mode_ok): Update accordingly.
2818 (rs6000_hard_regno_mode_ok): New function.
2819 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
2820 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
2821 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2822 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
2823 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
2824 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
2825 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2826 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
2827 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
2828 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2829 (sh_hard_regno_mode_ok): Make static.
2830 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
2831 instead of HARD_REGNO_MODE_OK.
2832 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
2833 (sparc_mode_class): Delete.
2834 (HARD_REGNO_MODE_OK): Delete.
2835 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2836 (hard_regno_mode_classes): Make static.
2837 (sparc_mode_class): Likewise.
2838 (sparc_hard_regno_mode_ok): New function.
2839 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
2840 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
2841 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
2843 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2844 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
2845 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
2846 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
2847 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
2848 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2849 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
2850 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
2851 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2852 (visium_hard_regno_mode_ok): New function.
2853 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
2854 instead of HARD_REGNO_MODE_OK.
2855 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
2856 (HARD_REGNO_MODE_OK): Delete.
2857 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
2858 (xtensa_hard_regno_mode_ok_p): ...this and make static.
2859 (xtensa_option_override): Update accordingly.
2860 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2861 (xtensa_hard_regno_mode_ok): New function.
2862 * system.h (HARD_REGNO_MODE_OK): Poison.
2864 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2865 Alan Hayward <alan.hayward@arm.com>
2866 David Sherwood <david.sherwood@arm.com>
2868 * target.def (hard_regno_call_part_clobbered): New hook.
2869 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
2870 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
2871 * doc/tm.texi: Regenerate.
2872 * hooks.h (hook_bool_uint_mode_false): Declare.
2873 * hooks.c (hook_bool_uint_mode_false): New function.
2874 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2875 * cselib.c (cselib_process_insn): Use
2876 targetm.hard_regno_call_part_clobbered instead of
2877 HARD_REGNO_CALL_PART_CLOBBERED.
2878 * ira-conflicts.c (ira_build_conflicts): Likewise.
2879 * ira-costs.c (ira_tune_allocno_costs): Likewise.
2880 * lra-constraints.c (need_for_call_save_p): Likewise.
2881 * lra-lives.c: Include target.h.
2882 (check_pseudos_live_through_calls): Use
2883 targetm.hard_regno_call_part_clobbered instead of
2884 HARD_REGNO_CALL_PART_CLOBBERED.
2885 * regcprop.c: Include target.h.
2886 (copyprop_hardreg_forward_1): Use
2887 targetm.hard_regno_call_part_clobbered instead of
2888 HARD_REGNO_CALL_PART_CLOBBERED.
2889 * reginfo.c (choose_hard_reg_mode): Likewise.
2890 * regrename.c (check_new_reg_p): Likewise.
2891 * reload.c (find_equiv_reg): Likewise.
2892 * reload1.c (emit_reload_insns): Likewise.
2893 * sched-deps.c (deps_analyze_insn): Likewise.
2894 * sel-sched.c (init_regs_for_mode): Likewise.
2895 (mark_unavailable_hard_regs): Likewise.
2896 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
2897 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2898 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
2900 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2901 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2902 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
2904 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
2906 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2907 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2908 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
2910 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2911 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2912 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
2914 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2915 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
2917 * config/powerpcspe/powerpcspe.c
2918 (rs6000_hard_regno_call_part_clobbered): New function.
2919 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2920 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2921 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
2923 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2924 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2925 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
2927 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2928 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2929 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
2931 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2932 Alan Hayward <alan.hayward@arm.com>
2933 David Sherwood <david.sherwood@arm.com>
2935 * rtl.h (subreg_memory_offset): Declare.
2936 * emit-rtl.c (subreg_memory_offset): New function.
2937 * expmed.c (store_bit_field_1): Use it.
2938 * expr.c (undefined_operand_subword_p): Likewise.
2939 * simplify-rtx.c (simplify_subreg): Likewise.
2941 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
2943 PR rtl-optimization/57448
2946 * optabs.c (expand_atomic_load): Place compiler memory barriers if
2947 using atomic_load pattern.
2948 (expand_atomic_store): Likewise.
2950 2017-09-04 Jakub Jelinek <jakub@redhat.com>
2953 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
2954 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
2955 handling. Use replace_call_with_value with NULL instead of
2956 gsi_replace, unlink_stmt_vdef and release_defs.
2958 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
2961 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
2963 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2966 * rtl.h (emit_library_call_value_1): Declare.
2967 (emit_library_call): Replace declaration with a series of overloads.
2968 Remove the parameter count argument.
2969 (emit_library_call_value): Likewise.
2970 * calls.c (emit_library_call_value_1): Make global. Replace varargs
2971 with an "rtx_mode_t *".
2972 (emit_library_call_value): Delete.
2973 (emit_library_call): Likewise.
2974 * asan.c (asan_emit_stack_protection): Update calls accordingly.
2975 (asan_emit_allocas_unpoison): Likewise.
2976 * builtins.c (expand_builtin_powi): Likewise.
2977 (expand_asan_emit_allocas_unpoison): Likewise.
2978 * cfgexpand.c (expand_main_function): Likewise.
2979 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
2980 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
2981 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
2982 * config/arm/arm.c (arm_trampoline_init): Likewise.
2983 (arm_call_tls_get_addr): Likewise.
2984 (arm_expand_divmod_libfunc): Likewise.
2985 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
2986 (smulsi3_highpart): Likewise.
2987 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
2988 (c6x_expand_compare): Likewise.
2989 (c6x_expand_movmem): Likewise.
2990 * config/frv/frv.c (frv_trampoline_init): Likewise.
2991 * config/i386/i386.c (ix86_trampoline_init): Likewise.
2992 (ix86_expand_divmod_libfunc): Likewise.
2993 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
2994 (ia64_expand_compare): Likewise.
2995 (ia64_profile_hook): Likewise.
2996 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
2997 (nonlocal_goto): Likewise.
2998 (restore_stack_nonlocal): Likewise.
2999 * config/m32r/m32r.c (block_move_call): Likewise.
3000 (m32r_trampoline_init): Likewise.
3001 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
3002 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
3003 (m68k_call_m68k_read_tp): Likewise.
3004 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
3005 (microblaze_expand_divide): Likewise.
3006 * config/mips/mips.h (mips_args): Likewise.
3007 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
3008 (MIPS_ICACHE_SYNC): Likewise.
3009 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
3010 (nios2_trampoline_init): Likewise.
3011 * config/pa/pa.c (hppa_tls_call): Likewise.
3012 (pa_trampoline_init): Likewise.
3013 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
3014 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
3015 (expand_strn_compare): Likewise.
3016 (rs6000_generate_compare): Likewise.
3017 (rs6000_expand_float128_convert): Likewise.
3018 (output_profile_hook): Likewise.
3019 (rs6000_trampoline_init): Likewise.
3020 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
3021 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
3022 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
3023 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
3024 (rs6000_generate_compare): Likewise.
3025 (rs6000_expand_float128_convert): Likewise.
3026 (output_profile_hook): Likewise.
3027 (rs6000_trampoline_init): Likewise.
3028 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
3029 * config/sh/sh.c (sh_trampoline_init): Likewise.
3030 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
3031 (sparc_emit_float_lib_cmp): Likewise.
3032 (sparc32_initialize_trampoline): Likewise.
3033 (sparc64_initialize_trampoline): Likewise.
3034 (sparc_profile_hook): Likewise.
3035 * config/spu/spu.c (ea_load_store): Likewise.
3036 * config/spu/spu.md (floatunssidf2): Likewise.
3037 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
3038 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
3039 * config/visium/visium.c (expand_block_move_4): Likewise.
3040 (expand_block_move_2): Likewise.
3041 (expand_block_move_1): Likewise.
3042 (expand_block_set_4): Likewise.
3043 (expand_block_set_2): Likewise.
3044 (expand_block_set_1): Likewise.
3045 (visium_trampoline_init): Likewise.
3046 (visium_profile_hook): Likewise.
3047 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
3048 (xtensa_setup_frame_addresses): Likewise.
3049 (xtensa_trampoline_init): Likewise.
3050 * except.c (sjlj_emit_function_enter): Likewise.
3051 (sjlj_emit_function_exit): Likewise.
3052 * explow.c (allocate_dynamic_stack_space): Likewise.
3053 (probe_stack_range): Likewise.
3054 * expr.c (convert_mode_scalar): Likewise.
3055 * optabs.c (expand_binop): Likewise.
3056 (expand_twoval_binop_libfunc): Likewise.
3057 (expand_unop): Likewise.
3058 (prepare_cmp_insn): Likewise.
3059 (prepare_float_lib_cmp): Likewise.
3060 (expand_float): Likewise.
3061 (expand_fix): Likewise.
3062 (expand_fixed_convert): Likewise.
3063 (maybe_emit_sync_lock_test_and_set): Likewise.
3064 (expand_atomic_compare_and_swap): Likewise.
3065 (expand_mem_thread_fence): Likewise.
3066 (expand_atomic_fetch_op): Likewise.
3068 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
3070 * doc/generic.texi (OpenACC): Adjust URL.
3071 * doc/invoke.texi (C Dialect Options): Ditto.
3073 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
3075 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
3076 predicate for operand 1. Add (m,<S>) constraint.
3077 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
3078 Prevent memory operand 1 with register operand 2.
3080 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
3082 PR rtl-optimization/82024
3083 * combine.c (try_combine): If the combination result is a PARALLEL,
3084 and we only need to retain the SET in there that would be placed
3085 at I2, check that we can place that at I3 instead, before doing so.
3087 2017-09-01 Jakub Jelinek <jakub@redhat.com>
3090 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
3092 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
3093 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
3096 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
3097 Jeff Law <law@redhat.com>
3099 * varasm.c (bss_initializer_p): Do not put constants into .bss
3100 (categorize_decl_for_section): Handle bss_initializer_p returning
3101 false when DECL_INITIAL is NULL.
3103 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3106 * config/s390/s390.c (s390_can_inline_p): New function.
3108 2017-09-01 Jeff Law <law@redhat.com>
3110 PR tree-optimization/82052
3111 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
3112 Always initialize the returned slot after a hash table miss
3113 when INSERT is true.
3115 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
3117 * config/s390/s390.md (mem_signal_fence): Remove.
3118 * doc/md.texi (mem_signal_fence): Remove.
3119 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
3121 * target-insns.def (mem_signal_fence): Remove.
3123 2017-09-01 Jakub Jelinek <jakub@redhat.com>
3126 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
3129 * asan.c (create_odr_indicator): Strip name encoding from assembler
3130 name before appending it after __odr_asan_.
3132 2017-09-01 Martin Liska <mliska@suse.cz>
3134 PR tree-optimization/82059
3135 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
3136 frequency only when an edge is redirected.
3138 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
3140 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
3141 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
3142 (arc_conditional_register_usage): Remove ARC600 lp_count
3144 (arc_file_start): Emit Tag_ARC_CPU_variation.
3145 (arc_can_use_doloop_p): New conditions to use ZOLs.
3146 (hwloop_fail): New function.
3147 (hwloop_optimize): Likewise.
3148 (hwloop_pattern_reg): Likewise.
3149 (arc_doloop_hooks): New struct, to be used with reorg_loops.
3150 (arc_reorg_loops): New function, calls reorg_loops.
3151 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
3152 (arc600_corereg_hazard): Remove ZOL checking, case handled by
3154 (arc_loop_hazard): Remove function, functionality moved into
3156 (arc_hazard): Remove arc_loop_hazard call.
3157 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
3158 into hwloop_optimize.
3159 (arc_label_align): Remove ZOL handling.
3160 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
3161 * config/arc/arc.md (doloop_begin): Remove pattern.
3162 (doloop_begin_i): Likewise.
3163 (doloop_end_i): Likewise.
3164 (doloop_fallback): Likewise.
3165 (doloop_fallback_m): Likewise.
3166 (doloop_end): Reimplement expand.
3167 (arc_lp): New pattern for LP instruction.
3168 (loop_end): New pattern.
3169 (loop_fail): Likewise.
3170 (decrement_and_branch_until_zero): Likewise.
3171 * config/arc/arc.opt (mlpc-width): New option.
3172 * doc/invoke.texi (mlpc-width): Document option.
3174 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
3176 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
3177 (arc_ccfsm_advance): Fix checking for delay slots.
3178 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
3180 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
3182 * config/arc/arc.md (movqi_insn): Add stores to save constant long
3184 (movhi_insn): Update store instruction constraint which are saving
3185 6-bit short immediates.
3186 (movsi_insn): Consider also short scaled load operations.
3187 (zero_extendhisi2_i): Use Usd constraint instead of T.
3188 (extendhisi2_i): Add q constraint.
3189 (arc_clzsi2): Add type and length attributes.
3190 (arc_ctzsi2): Likewise.
3191 * config/arc/constraints.md (Usc): Update constraint, the
3192 assembler can parse two relocations for a single instruction.
3194 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
3196 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
3197 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
3199 2017-08-31 Olivier Hainque <hainque@adacore.com>
3201 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
3202 match as powerpc-wrs-vxworks*.
3204 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
3206 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
3207 register constraint for by-element operand.
3208 (aarch64_mls_elt_merge<mode>): Likewise.
3210 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
3212 * config/arc/arc.c (arc_can_follow_jump): Check for short
3215 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
3217 * config.gcc: Use g.opt for arc.
3218 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
3219 functionality moved to ...
3220 (legitimate_scaled_address_p): New function, ...here.
3221 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
3222 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
3223 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
3225 (arc_override_options): Handle G option.
3226 (arc_output_pic_addr_const): Correct function definition.
3227 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
3228 (arc_decl_anon_ns_mem_p): Delete.
3229 (arc_in_small_data_p): Overhaul this function to take into
3230 consideration the value given via G option.
3231 (arc_rewrite_small_data_1): Renamed and corrected old
3232 arc_rewrite_small_data function.
3233 (arc_rewrite_small_data): New function.
3234 (small_data_pattern): Don't use pic_offset_table_rtx.
3235 * config/arc/arc.h (CC1_SPEC): Recognize G option.
3236 * config/arc/simdext.md (movmisalignv2hi): Use
3237 prepare_move_operands function.
3239 (movmisalign*): Likewise.
3240 * doc/invoke.texi (ARC options): Document -G option.
3242 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
3244 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
3246 * config/arc/arc.c (arc_print_operand): Output scalled address for
3247 sdata whenever is possible.
3248 (arc_in_small_data_p): Allow sdata for 64bit datum when double
3249 load/stores are available.
3250 (compact_sda_memory_operand): Check for the alignment required by
3251 code density instructions.
3252 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
3254 * config/arc/constraints.md (Usd): Update constraint.
3255 (Us0): New constraint.
3256 (Usc): Update constraint.
3258 2017-08-31 Richard Biener <rguenther@suse.de>
3261 * dwarf2out.c (dwarf2out_early_global_decl): Process each
3264 2017-08-31 Tamar Christina <tamar.christina@arm.com>
3266 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
3267 Resize type_signature.
3269 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
3270 Alan Hayward <alan.hayward@arm.com>
3271 David Sherwood <david.sherwood@arm.com>
3273 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
3274 subregs whose inner modes can be stored in GPRs.
3275 (aarch64_classify_index): Likewise.
3277 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
3278 Alan Hayward <alan.hayward@arm.com>
3279 David Sherwood <david.sherwood@arm.com>
3281 * config/aarch64/iterators.md (V_cmp_result): Rename to...
3282 (V_INT_EQUIV): ...this.
3283 (v_cmp_result): Rename to...
3284 (v_int_equiv): ...this.
3285 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
3286 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
3287 (copysign<mode>3): Likewise.
3288 (aarch64_simd_bsl<mode>_internal): Likewise.
3289 (aarch64_simd_bsl<mode>): Likewise.
3290 (vec_cmp<mode><mode>): Likewise.
3291 (vcond<mode><mode>): Likewise.
3292 (vcond<v_cmp_mixed><mode>): Likewise.
3293 (vcondu<mode><v_cmp_mixed>): Likewise.
3294 (aarch64_cm<optab><mode>): Likewise.
3295 (aarch64_cmtst<mode>): Likewise.
3296 (aarch64_fac<optab><mode>): Likewise.
3297 (vec_perm_const<mode>): Likewise.
3298 (vcond_mask_<mode><v_cmp_result>): Rename to...
3299 (vcond_mask_<mode><v_int_equiv>): ...this.
3300 (vec_cmp<mode><v_cmp_result>): Rename to...
3301 (vec_cmp<mode><v_int_equiv>): ...this.
3303 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
3304 Alan Hayward <alan.hayward@arm.com>
3305 David Sherwood <david.sherwood@arm.com>
3307 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
3309 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
3310 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
3311 (UNSPEC_LD4_DREG): New unspecs.
3312 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
3313 (aarch64_ld2<mode>_dreg_be): Replace with...
3314 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
3316 (aarch64_ld3<mode>_dreg_le)
3317 (aarch64_ld3<mode>_dreg_be): Replace with...
3318 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
3320 (aarch64_ld4<mode>_dreg_le)
3321 (aarch64_ld4<mode>_dreg_be): Replace with...
3322 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
3325 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3327 PR tree-optimization/81987
3328 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
3329 insert an initializer in a location not dominated by the stride
3332 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
3334 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
3335 on the entire header of the finally block in the fallthru case.
3337 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
3339 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
3341 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
3343 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
3344 rs6000_emit_move_from_cr and call renamed function.
3345 (rs6000_emit_prologue): Call renamed functions.
3346 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
3347 movesi_from_cr, remove volatile CRs.
3349 2017-08-30 Jon Beniston <jon@beniston.com>
3350 Richard Biener <rguenther@suse.de>
3352 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
3353 of VECTOR_MODE_P check.
3354 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
3355 element vector types.
3357 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3359 * df.h (df_read_modify_subreg_p): Remove in favor of...
3360 * rtl.h (read_modify_subreg_p): ...this new function. Take a
3361 const_rtx instead of an rtx.
3362 * cprop.c (local_cprop_find_used_regs): Update accordingly.
3363 * df-problems.c (df_word_lr_mark_ref): Likewise.
3364 * ira-lives.c (mark_pseudo_reg_live): Likewise.
3365 (mark_pseudo_reg_dead): Likewise.
3366 (mark_ref_dead): Likewise.
3367 * reginfo.c (init_subregs_of_mode): Likewise.
3368 * sched-deps.c (sched_analyze_1): Likewise.
3369 * df-scan.c (df_def_record_1): Likewise.
3370 (df_uses_record): Likewise.
3371 (df_read_modify_subreg_p): Remove in favor of...
3372 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
3373 const_rtx instead of an rtx.
3375 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3376 Alan Hayward <alan.hayward@arm.com>
3377 David Sherwood <david.sherwood@arm.com>
3379 * rtl.h (partial_subreg_p): New function.
3380 * caller-save.c (save_call_clobbered_regs): Use it.
3381 * calls.c (expand_call): Likewise.
3382 * combine.c (combinable_i3pat): Likewise.
3383 (simplify_set): Likewise.
3384 (make_extraction): Likewise.
3385 (make_compound_operation_int): Likewise.
3386 (gen_lowpart_or_truncate): Likewise.
3387 (force_to_mode): Likewise.
3388 (make_field_assignment): Likewise.
3389 (reg_truncated_to_mode): Likewise.
3390 (record_truncated_value): Likewise.
3391 (move_deaths): Likewise.
3392 * cse.c (record_jump_cond): Likewise.
3393 (cse_insn): Likewise.
3394 * cselib.c (cselib_lookup_1): Likewise.
3395 * expmed.c (extract_bit_field_using_extv): Likewise.
3396 * function.c (assign_parm_setup_reg): Likewise.
3397 * ifcvt.c (noce_convert_multiple_sets): Likewise.
3398 * ira-build.c (create_insn_allocnos): Likewise.
3399 * lra-coalesce.c (merge_pseudos): Likewise.
3400 * lra-constraints.c (match_reload): Likewise.
3401 (simplify_operand_subreg): Likewise.
3402 (curr_insn_transform): Likewise.
3403 * lra-lives.c (process_bb_lives): Likewise.
3404 * lra.c (new_insn_reg): Likewise.
3405 (lra_substitute_pseudo): Likewise.
3406 * regcprop.c (mode_change_ok): Likewise.
3407 (maybe_mode_change): Likewise.
3408 (copyprop_hardreg_forward_1): Likewise.
3409 * reload.c (push_reload): Likewise.
3410 (find_reloads): Likewise.
3411 (find_reloads_subreg_address): Likewise.
3412 * reload1.c (alter_reg): Likewise.
3413 (eliminate_regs_1): Likewise.
3414 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3416 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
3418 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
3419 back to if statements, including unpack.
3421 2017-08-30 Martin Liska <mliska@suse.cz>
3424 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
3426 (func_checker::compare_asm_inputs_outputs): ... this function.
3427 (func_checker::compare_gimple_asm): Use the function to compare
3428 also ASM constrains.
3429 * ipa-icf-gimple.h: Rename the function.
3431 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3432 Alan Hayward <alan.hayward@arm.com>
3433 David Sherwood <david.sherwood@arm.com>
3435 * coretypes.h (complex_mode): New type.
3436 * gdbhooks.py (build_pretty_printer): Handle it.
3437 * machmode.h (complex_mode): New class.
3438 (complex_mode::includes_p): New function.
3439 (is_complex_int_mode): Likewise.
3440 (is_complex_float_mode): Likewise.
3441 * genmodes.c (get_mode_class): Handle complex mode classes.
3442 * function.c (expand_function_end): Use is_complex_int_mode.
3444 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3445 Alan Hayward <alan.hayward@arm.com>
3446 David Sherwood <david.sherwood@arm.com>
3448 * coretypes.h (scalar_mode_pod): New typedef.
3449 * gdbhooks.py (build_pretty_printer): Handle it.
3450 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
3451 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
3452 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
3453 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
3456 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3457 Alan Hayward <alan.hayward@arm.com>
3458 David Sherwood <david.sherwood@arm.com>
3460 * machmode.h (mode_for_vector): Take a scalar_mode instead
3462 * stor-layout.c (mode_for_vector): Likewise.
3463 * explow.c (promote_mode): Use as_a <scalar_mode>.
3464 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
3466 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3467 Alan Hayward <alan.hayward@arm.com>
3468 David Sherwood <david.sherwood@arm.com>
3470 * target.def (preferred_simd_mode): Take a scalar_mode
3471 instead of a machine_mode.
3472 * targhooks.h (default_preferred_simd_mode): Likewise.
3473 * targhooks.c (default_preferred_simd_mode): Likewise.
3474 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
3475 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
3476 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
3477 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
3478 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
3479 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
3480 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
3481 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
3483 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
3484 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
3485 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
3486 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
3487 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
3488 * doc/tm.texi: Regenerate.
3489 * optabs-query.c (can_vec_mask_load_store_p): Return false for
3492 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3493 Alan Hayward <alan.hayward@arm.com>
3494 David Sherwood <david.sherwood@arm.com>
3496 * target.def (scalar_mode_supported_p): Take a scalar_mode
3497 instead of a machine_mode.
3498 * targhooks.h (default_scalar_mode_supported_p): Likewise.
3499 * targhooks.c (default_scalar_mode_supported_p): Likewise.
3500 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
3501 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
3502 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
3503 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
3504 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
3505 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
3506 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
3507 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
3508 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
3509 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
3510 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
3511 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
3513 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
3514 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
3515 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
3516 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
3517 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
3519 * doc/tm.texi: Regenerate.
3521 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3522 Alan Hayward <alan.hayward@arm.com>
3523 David Sherwood <david.sherwood@arm.com>
3525 * coretypes.h (opt_scalar_mode): New typedef.
3526 * gdbhooks.py (build_pretty_printers): Handle it.
3527 * machmode.h (mode_iterator::get_2xwider): Add overload for
3529 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
3531 * expr.c (convert_mode_scalar): Likewise.
3532 * omp-low.c (omp_clause_aligned_alignment): Likewise.
3533 * optabs.c (expand_float): Likewise.
3534 (expand_fix): Likewise.
3535 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
3537 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3538 Alan Hayward <alan.hayward@arm.com>
3539 David Sherwood <david.sherwood@arm.com>
3541 * optabs.c (expand_float): Explicitly check for scalars before
3542 using a branching expansion.
3543 (expand_fix): Likewise.
3545 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3546 Alan Hayward <alan.hayward@arm.com>
3547 David Sherwood <david.sherwood@arm.com>
3549 * expr.c (convert_mode): Split scalar handling out into...
3550 (convert_mode_scalar): ...this new function. Treat the modes
3553 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3554 Alan Hayward <alan.hayward@arm.com>
3555 David Sherwood <david.sherwood@arm.com>
3557 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
3559 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
3561 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3562 Alan Hayward <alan.hayward@arm.com>
3563 David Sherwood <david.sherwood@arm.com>
3565 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
3566 rather than a machine_mode.
3567 (fixed_from_string): Likewise.
3568 (fixed_convert): Likewise.
3569 (fixed_convert_from_int): Likewise.
3570 (fixed_convert_from_real): Likewise.
3571 (real_convert_from_fixed): Likewise.
3572 * fixed-value.c (fixed_from_double_int): Likewise.
3573 (fixed_from_string): Likewise.
3574 (fixed_convert): Likewise.
3575 (fixed_convert_from_int): Likewise.
3576 (fixed_convert_from_real): Likewise.
3577 (real_convert_from_fixed): Likewise.
3578 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
3580 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3581 Alan Hayward <alan.hayward@arm.com>
3582 David Sherwood <david.sherwood@arm.com>
3584 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
3585 of separate mode class checks. Do not allow vector modes here.
3586 (immed_wide_int_const): Use as_a <scalar_mode>.
3587 * explow.c (trunc_int_for_mode): Likewise.
3588 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
3589 (wi::shwi): Likewise.
3590 (wi::min_value): Likewise.
3591 (wi::max_value): Likewise.
3592 * dwarf2out.c (loc_descriptor): Likewise.
3593 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
3596 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3597 Alan Hayward <alan.hayward@arm.com>
3598 David Sherwood <david.sherwood@arm.com>
3600 * tree.h (SCALAR_TYPE_MODE): New macro.
3601 * expr.c (expand_expr_addr_expr_1): Use it.
3602 (expand_expr_real_2): Likewise.
3603 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
3604 (fold_convert_const_fixed_from_int): Likewise.
3605 (fold_convert_const_fixed_from_real): Likewise.
3606 (native_encode_fixed): Likewise
3607 (native_encode_complex): Likewise
3608 (native_encode_vector): Likewise.
3609 (native_interpret_fixed): Likewise.
3610 (native_interpret_real): Likewise.
3611 (native_interpret_complex): Likewise.
3612 (native_interpret_vector): Likewise.
3613 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
3614 (simd_clone_adjust_argument_types): Likewise.
3615 (simd_clone_init_simd_arrays): Likewise.
3616 (simd_clone_adjust): Likewise.
3617 * stor-layout.c (layout_type): Likewise.
3618 * tree.c (build_minus_one_cst): Likewise.
3619 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
3620 * tree-inline.c (estimate_move_cost): Likewise.
3621 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
3622 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
3623 (vectorizable_reduction): Likewise.
3624 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
3625 (vect_recog_mixed_size_cond_pattern): Likewise.
3626 (check_bool_pattern): Likewise.
3627 (adjust_bool_pattern): Likewise.
3628 (search_type_for_mask_1): Likewise.
3629 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
3630 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
3631 (vectorizable_load): Likewise.
3632 (vectorizable_store): Likewise.
3633 * ubsan.c (ubsan_encode_value): Likewise.
3634 * varasm.c (output_constant): Likewise.
3636 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3637 Alan Hayward <alan.hayward@arm.com>
3638 David Sherwood <david.sherwood@arm.com>
3640 * coretypes.h (scalar_mode): New class.
3641 * machmode.h (scalar_mode): Likewise.
3642 (scalar_mode::includes_p): New function.
3643 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
3644 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
3645 * genmodes.c (get_mode_class): Handle remaining scalar modes.
3646 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
3647 * expmed.c (store_bit_field_1): Likewise.
3648 (extract_bit_field_1): Likewise.
3649 * expr.c (write_complex_part): Likewise.
3650 (read_complex_part): Likewise.
3651 (emit_move_complex_push): Likewise.
3652 (expand_expr_real_2): Likewise.
3653 * function.c (assign_parm_setup_reg): Likewise.
3654 (assign_parms_unsplit_complex): Likewise.
3655 * optabs.c (expand_binop): Likewise.
3656 * rtlanal.c (subreg_get_info): Likewise.
3657 * simplify-rtx.c (simplify_immed_subreg): Likewise.
3658 * varasm.c (output_constant_pool_2): Likewise.
3660 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3661 Alan Hayward <alan.hayward@arm.com>
3662 David Sherwood <david.sherwood@arm.com>
3664 * expmed.c (extract_high_half): Use scalar_int_mode and remove
3666 (expmed_mult_highpart_optab): Likewise.
3667 (expmed_mult_highpart): Likewise.
3669 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3670 Alan Hayward <alan.hayward@arm.com>
3671 David Sherwood <david.sherwood@arm.com>
3673 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
3674 instead of a machine_mode.
3675 (builtin_memset_read_str): Likewise.
3676 * builtins.c (c_readstr): Likewise.
3677 (builtin_memcpy_read_str): Likewise.
3678 (builtin_strncpy_read_str): Likewise.
3679 (builtin_memset_read_str): Likewise.
3680 (builtin_memset_gen_str): Likewise.
3681 (expand_builtin_signbit): Use scalar_int_mode for local variables.
3682 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
3683 instead of a machine_mode.
3684 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
3686 (make_extraction): Likewise.
3687 (try_widen_shift_mode): Take and return scalar_int_modes instead
3689 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
3690 a scalar_int_mode instead of a machine_mode.
3691 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
3692 (avr_addr_space_pointer_mode): Likewise.
3693 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
3694 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
3695 (msp430_unwind_word_mode): Likewise.
3696 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
3697 (spu_addr_space_pointer_mode): Likewise.
3698 (spu_addr_space_address_mode): Likewise.
3699 (spu_libgcc_cmp_return_mode): Likewise.
3700 (spu_libgcc_shift_count_mode): Likewise.
3701 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
3702 (rl78_addr_space_pointer_mode): Likewise.
3703 (fl78_unwind_word_mode): Likewise.
3704 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
3706 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
3707 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
3708 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
3709 (mips_valid_pointer_mode): Likewise.
3710 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
3711 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
3712 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
3714 (ft32_addr_space_address_mode): Likewise.
3715 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
3716 scalar_int_mode instead of a machine_mode.
3717 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
3719 (m32c_addr_space_address_mode): Likewise.
3720 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
3721 (rs6000_eh_return_filter_mode): Likewise.
3722 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
3723 (rs6000_eh_return_filter_mode): Likewise.
3724 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
3725 (s390_libgcc_shift_count_mode): Likewise.
3726 (s390_unwind_word_mode): Likewise.
3727 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
3729 * target.def (mode_rep_extended): Likewise.
3730 (valid_pointer_mode): Likewise.
3731 (addr_space.valid_pointer_mode): Likewise.
3732 (eh_return_filter_mode): Return a scalar_int_mode rather than
3734 (libgcc_cmp_return_mode): Likewise.
3735 (libgcc_shift_count_mode): Likewise.
3736 (unwind_word_mode): Likewise.
3737 (addr_space.pointer_mode): Likewise.
3738 (addr_space.address_mode): Likewise.
3739 * doc/tm.texi: Regenerate.
3740 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
3742 (do_jump): Use scalar_int_mode for local variables.
3743 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
3744 rather than a machine_mode.
3745 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
3746 (scompare_loc_descriptor_wide): Likewise.
3747 (scompare_loc_descriptor_narrow): Likewise.
3748 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
3750 * except.c (sjlj_emit_dispatch_table): Likewise.
3751 (expand_builtin_eh_copy_values): Likewise.
3752 * explow.c (convert_memory_address_addr_space_1): Likewise.
3753 Take a scalar_int_mode rather than a machine_mode.
3754 (convert_memory_address_addr_space): Take a scalar_int_mode rather
3755 than a machine_mode.
3756 (memory_address_addr_space): Use scalar_int_mode for local variables.
3757 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
3758 rather than a machine_mode.
3759 * expmed.c (mask_rtx): Likewise.
3760 (init_expmed_one_conv): Likewise.
3761 (expand_mult_highpart_adjust): Likewise.
3762 (extract_high_half): Likewise.
3763 (expmed_mult_highpart_optab): Likewise.
3764 (expmed_mult_highpart): Likewise.
3765 (expand_smod_pow2): Likewise.
3766 (expand_sdiv_pow2): Likewise.
3767 (emit_store_flag_int): Likewise.
3768 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
3770 (extract_low_bits): Likewise.
3771 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
3773 * expr.c (pieces_addr::adjust): Likewise.
3774 (can_store_by_pieces): Likewise.
3775 (store_by_pieces): Likewise.
3776 (clear_by_pieces_1): Likewise.
3777 (expand_expr_addr_expr_1): Likewise.
3778 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
3779 (expand_expr_real_1): Likewise.
3780 (try_casesi): Likewise.
3781 * final.c (shorten_branches): Likewise.
3782 * fold-const.c (fold_convert_const_int_from_fixed): Change the
3783 type of "mode" to machine_mode.
3784 * internal-fn.c (expand_arith_overflow_result_store): Take a
3785 scalar_int_mode rather than a machine_mode.
3786 (expand_mul_overflow): Use scalar_int_mode for local variables.
3787 * loop-doloop.c (doloop_modify): Likewise.
3788 (doloop_optimize): Likewise.
3789 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
3790 than a machine_mode.
3791 (expand_doubleword_shift_condmove): Likewise.
3792 (expand_doubleword_shift): Likewise.
3793 (expand_doubleword_clz): Likewise.
3794 (expand_doubleword_popcount): Likewise.
3795 (expand_doubleword_parity): Likewise.
3796 (expand_absneg_bit): Use scalar_int_mode for local variables.
3797 (prepare_float_lib_cmp): Likewise.
3798 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
3799 rather than a machine_mode.
3800 (convert_memory_address_addr_space): Likewise.
3801 (get_mode_bounds): Likewise.
3802 (get_address_mode): Return a scalar_int_mode rather than a
3804 * rtlanal.c (get_address_mode): Likewise.
3805 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
3806 than a machine_mode.
3807 * targhooks.c (default_mode_rep_extended): Likewise.
3808 (default_valid_pointer_mode): Likewise.
3809 (default_addr_space_valid_pointer_mode): Likewise.
3810 (default_eh_return_filter_mode): Return a scalar_int_mode rather
3811 than a machine_mode.
3812 (default_libgcc_cmp_return_mode): Likewise.
3813 (default_libgcc_shift_count_mode): Likewise.
3814 (default_unwind_word_mode): Likewise.
3815 (default_addr_space_pointer_mode): Likewise.
3816 (default_addr_space_address_mode): Likewise.
3817 * targhooks.h (default_eh_return_filter_mode): Likewise.
3818 (default_libgcc_cmp_return_mode): Likewise.
3819 (default_libgcc_shift_count_mode): Likewise.
3820 (default_unwind_word_mode): Likewise.
3821 (default_addr_space_pointer_mode): Likewise.
3822 (default_addr_space_address_mode): Likewise.
3823 (default_mode_rep_extended): Take a scalar_int_mode rather than
3825 (default_valid_pointer_mode): Likewise.
3826 (default_addr_space_valid_pointer_mode): Likewise.
3827 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
3829 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
3830 rather than a machine_mode.
3831 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
3832 for local variables.
3833 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
3834 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
3835 than a machine_mode.
3837 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3838 Alan Hayward <alan.hayward@arm.com>
3839 David Sherwood <david.sherwood@arm.com>
3841 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
3842 the mode argument to scalar_int_mode.
3843 (do_jump_by_parts_zero_rtx): Likewise.
3844 (do_jump_by_parts_equality_rtx): Likewise.
3845 (do_jump_by_parts_greater): Take a mode argument.
3846 (do_jump_by_parts_equality): Likewise.
3847 (do_jump_1): Update calls accordingly.
3849 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3850 Alan Hayward <alan.hayward@arm.com>
3851 David Sherwood <david.sherwood@arm.com>
3853 * is-a.h (safe_dyn_cast): New function.
3854 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
3855 (jump_table_for_label): Likewise.
3856 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
3857 instead of an rtx_insn *.
3858 (shorten_branches): Use dyn_cast instead of LABEL_P and
3859 JUMP_TABLE_DATA_P. Use jump_table_for_label and
3860 rtx_jump_table_data::get_data_mode.
3861 (final_scan_insn): Likewise.
3863 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3864 Alan Hayward <alan.hayward@arm.com>
3865 David Sherwood <david.sherwood@arm.com>
3867 * combine.c (try_combine): Use is_a <scalar_int_mode> when
3868 trying to combine a full-register integer set with a subreg
3871 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3872 Alan Hayward <alan.hayward@arm.com>
3873 David Sherwood <david.sherwood@arm.com>
3875 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
3876 that is always either address_mode or pointer_mode.
3878 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3879 Alan Hayward <alan.hayward@arm.com>
3880 David Sherwood <david.sherwood@arm.com>
3882 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
3883 when the two are known to be equal.
3885 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3886 Alan Hayward <alan.hayward@arm.com>
3887 David Sherwood <david.sherwood@arm.com>
3889 * simplify-rtx.c (simplify_const_unary_operation): Use
3890 is_a <scalar_int_mode> instead of checking for a nonzero
3891 precision. Forcibly convert op_mode to a scalar_int_mode
3892 in that case. More clearly differentiate the operand and
3893 result modes and use the former when deciding what the value
3894 of a count-bits operation should be. Use is_int_mode instead
3895 of checking for a MODE_INT. Remove redundant check for whether
3896 this mode has a zero precision.
3898 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3899 Alan Hayward <alan.hayward@arm.com>
3900 David Sherwood <david.sherwood@arm.com>
3902 * optabs.c (widen_leading): Change the type of the mode argument
3903 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
3904 (widen_bswap): Likewise.
3905 (expand_parity): Likewise.
3906 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
3907 (expand_ffs): Likewise.
3908 (epand_unop): Check for scalar integer modes before calling the
3911 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3912 Alan Hayward <alan.hayward@arm.com>
3913 David Sherwood <david.sherwood@arm.com>
3915 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
3917 (expand_expr_real_1): Update call accordingly.
3919 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3920 Alan Hayward <alan.hayward@arm.com>
3921 David Sherwood <david.sherwood@arm.com>
3923 * expmed.c (store_bit_field_using_insv): Add op0_mode and
3924 value_mode arguments. Use scalar_int_mode internally.
3925 (store_bit_field_1): Rename the new integer mode from imode
3926 to op0_mode and use it instead of GET_MODE (op0). Update calls
3927 to store_split_bit_field, store_bit_field_using_insv and
3928 store_fixed_bit_field.
3929 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
3930 Use scalar_int_mode internally. Use a bit count rather than a mode
3931 when calculating the largest bit size for get_best_mode.
3932 Update calls to store_split_bit_field and store_fixed_bit_field_1.
3933 (store_fixed_bit_field_1): Add mode and value_mode arguments.
3934 Remove assertion that OP0 has a scalar integer mode.
3935 (store_split_bit_field): Add op0_mode and value_mode arguments.
3936 Update calls to extract_fixed_bit_field.
3937 (extract_bit_field_using_extv): Add an op0_mode argument.
3938 Use scalar_int_mode internally.
3939 (extract_bit_field_1): Rename the new integer mode from imode to
3940 op0_mode and use it instead of GET_MODE (op0). Update calls to
3941 extract_split_bit_field, extract_bit_field_using_extv and
3942 extract_fixed_bit_field.
3943 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
3944 to extract_split_bit_field and extract_fixed_bit_field_1.
3945 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
3946 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
3948 (extract_split_bit_field): Add an op0_mode argument. Update call
3949 to extract_fixed_bit_field.
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 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
3956 * explow.c (hard_function_value): Likewise.
3957 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
3958 convert_to_mode call outside the loop.
3959 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
3960 for the mode iterator. Require the mode specified by max_pieces
3962 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
3964 (copy_blkmode_to_reg): Likewise.
3965 (set_storage_via_setmem): Likewise.
3966 * optabs.c (prepare_cmp_insn): Likewise.
3967 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
3968 * stor-layout.c (finish_bitfield_representative): Likewise.
3970 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3971 Alan Hayward <alan.hayward@arm.com>
3972 David Sherwood <david.sherwood@arm.com>
3974 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
3975 * expr.c (convert_move): Use them.
3976 (convert_modes): Likewise.
3977 (store_expr_with_bounds): Likewise.
3979 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3980 Alan Hayward <alan.hayward@arm.com>
3981 David Sherwood <david.sherwood@arm.com>
3983 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
3984 parameter for the mode of "x". Remove the "known_x", "known_mode"
3985 and "known_ret" arguments. Change the type of the mode argument
3987 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
3988 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
3989 (reg_num_sign_bit_copies_for_combine): Likewise.
3990 * rtlanal.c (nonzero_bits1): Likewise.
3991 (num_sign_bit_copies1): Likewise.
3992 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
3993 (reg_num_sign_bit_copies_general): Likewise.
3994 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
3995 (reg_nonzero_bits_general): Likewise.
3997 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3998 Alan Hayward <alan.hayward@arm.com>
3999 David Sherwood <david.sherwood@arm.com>
4001 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
4002 than in subroutines. Return 1 for non-integer modes.
4003 (cached_num_sign_bit_copies): Change the type of the mode parameter
4005 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
4006 classes. Handle CONST_INT_P first and then check whether X also
4007 has a scalar integer mode. Check the same thing for inner registers
4008 of a SUBREG and for values that are being extended or truncated.
4010 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4011 Alan Hayward <alan.hayward@arm.com>
4012 David Sherwood <david.sherwood@arm.com>
4014 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
4015 in subroutines. Return the mode mask for non-integer modes.
4016 (cached_nonzero_bits): Change the type of the mode parameter
4018 (nonzero_bits1): Likewise. Remove early exit for other mode
4019 classes. Handle CONST_INT_P first and then check whether X
4020 also has a scalar integer mode.
4022 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4023 Alan Hayward <alan.hayward@arm.com>
4024 David Sherwood <david.sherwood@arm.com>
4026 * expr.c (widest_int_mode_for_size): Make the comment match the code.
4027 Return a scalar_int_mode and assert that the size is greater than
4029 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
4030 (op_by_pieces_d::op_by_pieces_d): Likewise.
4031 (op_by_pieces_d::run): Likewise.
4032 (can_store_by_pieces): Likewise.
4034 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4035 Alan Hayward <alan.hayward@arm.com>
4036 David Sherwood <david.sherwood@arm.com>
4038 * combine.c (extract_left_shift): Add a mode argument and update
4040 (make_compound_operation_int): Change the type of the mode parameter
4041 to scalar_int_mode and update the call to extract_left_shift.
4043 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4044 Alan Hayward <alan.hayward@arm.com>
4045 David Sherwood <david.sherwood@arm.com>
4047 * combine.c (simplify_and_const_int): Change the type of the mode
4048 parameter to scalar_int_mode.
4049 (simplify_and_const_int_1): Likewise. Update recursive call.
4051 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4052 Alan Hayward <alan.hayward@arm.com>
4053 David Sherwood <david.sherwood@arm.com>
4055 * combine.c (simplify_compare_const): Check that the mode is a
4056 scalar_int_mode (rather than VOIDmode) before testing its
4058 (simplify_comparison): Move COMPARISON_P handling out of the
4059 loop and restrict the latter part of the loop to scalar_int_modes.
4060 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
4061 and when considering SUBREG_REGs. Use is_int_mode instead of
4062 checking GET_MODE_CLASS against MODE_INT.
4064 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4065 Alan Hayward <alan.hayward@arm.com>
4066 David Sherwood <david.sherwood@arm.com>
4068 * combine.c (try_widen_shift_mode): Move check for equal modes to...
4069 (simplify_shift_const_1): ...here. Use scalar_int_mode for
4070 shift_unit_mode and for modes involved in scalar shifts.
4072 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4073 Alan Hayward <alan.hayward@arm.com>
4074 David Sherwood <david.sherwood@arm.com>
4076 * combine.c (force_int_to_mode): New function, split out from...
4077 (force_to_mode): ...here. Keep xmode up-to-date and use it
4078 instead of GET_MODE (x).
4080 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4081 Alan Hayward <alan.hayward@arm.com>
4082 David Sherwood <david.sherwood@arm.com>
4084 * optabs-query.h (extraction_insn::struct_mode): Change type to
4085 opt_scalar_int_mode and update comment.
4086 (extraction_insn::field_mode): Change type to scalar_int_mode.
4087 (extraction_insn::pos_mode): Likewise.
4088 * combine.c (make_extraction): Update accordingly.
4089 * optabs-query.c (get_traditional_extraction_insn): Likewise.
4090 (get_optab_extraction_insn): Likewise.
4091 * recog.c (simplify_while_replacing): Likewise.
4092 * expmed.c (narrow_bit_field_mem): Change the type of the mode
4093 parameter to opt_scalar_int_mode.
4095 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4096 Alan Hayward <alan.hayward@arm.com>
4097 David Sherwood <david.sherwood@arm.com>
4099 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
4100 to a scalar_int_mode instead of a machine_mode.
4101 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
4102 (get_best_mode): Return a boolean and use a pointer argument to store
4103 the selected mode. Replace the limit mode parameter with a bit limit.
4104 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
4105 for the values returned by bit_field_mode_iterator::next_mode.
4106 (store_bit_field): Update call to get_best_mode.
4107 (store_fixed_bit_field): Likewise.
4108 (extract_fixed_bit_field): Likewise.
4109 * expr.c (optimize_bitfield_assignment_op): Likewise.
4110 * fold-const.c (optimize_bit_field_compare): Likewise.
4111 (fold_truth_andor_1): Likewise.
4112 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
4113 Update for new type of m_mode.
4114 (get_best_mode): As above.
4116 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4117 Alan Hayward <alan.hayward@arm.com>
4118 David Sherwood <david.sherwood@arm.com>
4120 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
4121 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
4122 (store_bit_field): Check is_a <scalar_int_mode> before calling
4123 strict_volatile_bitfield_p.
4124 (extract_bit_field): Likewise.
4126 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4127 Alan Hayward <alan.hayward@arm.com>
4128 David Sherwood <david.sherwood@arm.com>
4130 * target.def (cstore_mode): Return a scalar_int_mode.
4131 * doc/tm.texi: Regenerate.
4132 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
4133 * targhooks.h (default_cstore_mode): Likewise.
4134 * targhooks.c (default_cstore_mode): Likewise, using a forced
4136 * expmed.c (emit_cstore): Expect the target of the cstore to be
4139 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4140 Alan Hayward <alan.hayward@arm.com>
4141 David Sherwood <david.sherwood@arm.com>
4143 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
4145 (niter_desc): Likewise mode.
4146 (iv_analyze): Add a mode parameter.
4148 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
4149 and change its type to scalar_int_mode.
4150 * loop-iv.c: Update commentary at head of file.
4151 (iv_constant): Pass the mode paraeter before the rtx it describes
4152 and change its type to scalar_int_mode. Remove VOIDmode handling.
4153 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
4154 (iv_extend): Likewise.
4155 (shorten_into_mode): Likewise.
4156 (iv_add): Use scalar_int_mode.
4157 (iv_mult): Likewise.
4158 (iv_shift): Likewise.
4159 (canonicalize_iv_subregs): Likewise.
4160 (get_biv_step_1): Pass the outer_mode parameter before the rtx
4161 it describes and change its mode to scalar_int_mode. Also change
4162 the type of the returned inner_mode to scalar_int_mode.
4163 (get_biv_step): Likewise, turning outer_mode from a pointer
4164 into a direct parameter. Update call to get_biv_step_1.
4165 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
4166 iv_constant and get_biv_step.
4167 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
4168 and change its type to scalar_int_mode. Don't initialise iv->mode
4169 to VOIDmode and remove later checks for its still being VOIDmode.
4170 Update calls to iv_analyze_op and iv_analyze_expr. Check
4171 is_a <scalar_int_mode> when changing the mode under consideration.
4172 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
4173 Update call to iv_analyze_expr.
4174 (iv_analyze_op): Add a mode parameter. Reject subregs whose
4175 inner register is not also a scalar_int_mode. Update call to
4177 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
4178 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
4179 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
4180 separate mode class checks. Update calls to iv_analyze. Remove
4181 fix-up of VOIDmodes after iv_analyze_biv.
4182 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
4183 don't have a scalar_int_mode. Update call to biv_p.
4185 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4186 Alan Hayward <alan.hayward@arm.com>
4187 David Sherwood <david.sherwood@arm.com>
4189 * cfgexpand.c (convert_debug_memory_address): Use
4190 as_a <scalar_int_mode>.
4191 * combine.c (expand_compound_operation): Likewise.
4192 (make_extraction): Likewise.
4193 (change_zero_ext): Likewise.
4194 (simplify_comparison): Likewise.
4195 * cse.c (cse_insn): Likewise.
4196 * dwarf2out.c (minmax_loc_descriptor): Likewise.
4197 (mem_loc_descriptor): Likewise.
4198 (loc_descriptor): Likewise.
4199 * expmed.c (init_expmed_one_mode): Likewise.
4200 (synth_mult): Likewise.
4201 (emit_store_flag_1): Likewise.
4202 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
4203 of a comparison with size.
4204 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
4205 (reduce_to_bit_field_precision): Likewise.
4206 * function.c (expand_function_end): Likewise.
4207 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
4208 * loop-doloop.c (doloop_modify): Likewise.
4209 * optabs.c (expand_binop): Likewise.
4210 (expand_unop): Likewise.
4211 (expand_copysign_absneg): Likewise.
4212 (prepare_cmp_insn): Likewise.
4213 (maybe_legitimize_operand): Likewise.
4214 * recog.c (const_scalar_int_operand): Likewise.
4215 * rtlanal.c (get_address_mode): Likewise.
4216 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4217 (simplify_cond_clz_ctz): Likewise.
4218 * tree-nested.c (get_nl_goto_field): Likewise.
4219 * tree.c (build_vector_type_for_mode): Likewise.
4220 * var-tracking.c (use_narrower_mode): Likewise.
4222 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4223 Alan Hayward <alan.hayward@arm.com>
4224 David Sherwood <david.sherwood@arm.com>
4226 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
4227 * builtins.c (expand_builtin_signbit): Use it.
4228 * cfgexpand.c (expand_debug_expr): Likewise.
4229 * dojump.c (do_jump): Likewise.
4230 (do_compare_and_jump): Likewise.
4231 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
4232 * expmed.c (make_tree): Likewise.
4233 * expr.c (expand_expr_real_2): Likewise.
4234 (expand_expr_real_1): Likewise.
4235 (try_casesi): Likewise.
4236 * fold-const-call.c (fold_const_call_ss): Likewise.
4237 * fold-const.c (unextend): Likewise.
4238 (extract_muldiv_1): Likewise.
4239 (fold_single_bit_test): Likewise.
4240 (native_encode_int): Likewise.
4241 (native_encode_string): Likewise.
4242 (native_interpret_int): Likewise.
4243 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
4244 * internal-fn.c (expand_addsub_overflow): Likewise.
4245 (expand_neg_overflow): Likewise.
4246 (expand_mul_overflow): Likewise.
4247 (expand_arith_overflow): Likewise.
4248 * match.pd: Likewise.
4249 * stor-layout.c (layout_type): Likewise.
4250 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
4251 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
4252 * tree-ssanames.c (get_range_info): Likewise.
4253 * tree-switch-conversion.c (array_value_type) Likewise.
4254 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
4255 (vect_recog_divmod_pattern): Likewise.
4256 (vect_recog_mixed_size_cond_pattern): Likewise.
4257 * tree-vrp.c (extract_range_basic): Likewise.
4258 (simplify_float_conversion_using_ranges): Likewise.
4259 * tree.c (int_fits_type_p): Likewise.
4260 * ubsan.c (instrument_bool_enum_load): Likewise.
4261 * varasm.c (mergeable_string_section): Likewise.
4262 (narrowing_initializer_constant_valid_p): Likewise.
4263 (output_constant): Likewise.
4265 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4266 Alan Hayward <alan.hayward@arm.com>
4267 David Sherwood <david.sherwood@arm.com>
4269 * machmode.h (NARROWEST_INT_MODE): New macro.
4270 * expr.c (alignment_for_piecewise_move): Use it instead of
4271 GET_CLASS_NARROWEST_MODE (MODE_INT).
4272 (push_block): Likewise.
4273 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
4275 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
4277 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4278 Alan Hayward <alan.hayward@arm.com>
4279 David Sherwood <david.sherwood@arm.com>
4281 * postreload.c (move2add_valid_value_p): Change the type of the
4282 mode parameter to scalar_int_mode.
4283 (move2add_use_add2_insn): Add a mode parameter and use it instead
4285 (move2add_use_add3_insn): Likewise.
4286 (reload_cse_move2add): Update accordingly.
4288 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4289 Alan Hayward <alan.hayward@arm.com>
4290 David Sherwood <david.sherwood@arm.com>
4292 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
4294 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
4295 * optabs.c (expand_unop): Likewise.
4297 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4298 Alan Hayward <alan.hayward@arm.com>
4299 David Sherwood <david.sherwood@arm.com>
4301 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
4302 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
4303 (popcount_loc_descriptor): Likewise.
4304 (bswap_loc_descriptor): Likewise.
4305 (rotate_loc_descriptor): Likewise.
4306 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
4307 calling the functions above.
4309 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4310 Alan Hayward <alan.hayward@arm.com>
4311 David Sherwood <david.sherwood@arm.com>
4313 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
4315 (try_combine): Likewise.
4316 (simplify_if_then_else): Likewise.
4317 * cse.c (cse_insn): Likewise.
4318 * dwarf2out.c (mem_loc_descriptor): Likewise.
4319 * emit-rtl.c (gen_lowpart_common): Likewise.
4320 * simplify-rtx.c (simplify_truncation): Likewise.
4321 (simplify_binary_operation_1): Likewise.
4322 (simplify_const_relational_operation): Likewise.
4323 (simplify_ternary_operation): Likewise.
4324 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
4326 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4327 Alan Hayward <alan.hayward@arm.com>
4328 David Sherwood <david.sherwood@arm.com>
4330 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
4331 * reload.c (push_reload): Likewise.
4332 (find_reloads): Likewise.
4334 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4335 Alan Hayward <alan.hayward@arm.com>
4336 David Sherwood <david.sherwood@arm.com>
4338 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
4339 (make_compound_operation_int): Likewise.
4340 (change_zero_ext): Likewise.
4341 * expr.c (convert_move): Likewise.
4342 (convert_modes): Likewise.
4343 * fwprop.c (forward_propagate_subreg): Likewise.
4344 * loop-iv.c (get_biv_step_1): Likewise.
4345 * optabs.c (widen_operand): Likewise.
4346 * postreload.c (move2add_valid_value_p): Likewise.
4347 * recog.c (simplify_while_replacing): Likewise.
4348 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4349 (simplify_binary_operation_1): Likewise. Remove redundant
4350 mode equality check.
4352 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4353 Alan Hayward <alan.hayward@arm.com>
4354 David Sherwood <david.sherwood@arm.com>
4356 * combine.c (combine_simplify_rtx): Add checks for
4357 is_a <scalar_int_mode>.
4358 (simplify_if_then_else): Likewise.
4359 (make_field_assignment): Likewise.
4360 (simplify_comparison): Likewise.
4361 * ifcvt.c (noce_try_bitop): Likewise.
4362 * loop-invariant.c (canonicalize_address_mult): Likewise.
4363 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4365 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4366 Alan Hayward <alan.hayward@arm.com>
4367 David Sherwood <david.sherwood@arm.com>
4369 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
4370 is_a <scalar_int_mode> instead of != BLKmode.
4372 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4373 Alan Hayward <alan.hayward@arm.com>
4374 David Sherwood <david.sherwood@arm.com>
4376 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
4377 instead of != VOIDmode.
4378 * combine.c (if_then_else_cond): Likewise.
4379 (change_zero_ext): Likewise.
4380 * dwarf2out.c (mem_loc_descriptor): Likewise.
4381 (loc_descriptor): Likewise.
4382 * rtlanal.c (canonicalize_condition): Likewise.
4383 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
4385 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4386 Alan Hayward <alan.hayward@arm.com>
4387 David Sherwood <david.sherwood@arm.com>
4389 * simplify-rtx.c (simplify_binary_operation_1): Use
4390 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
4392 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4393 Alan Hayward <alan.hayward@arm.com>
4394 David Sherwood <david.sherwood@arm.com>
4396 * wide-int.h (int_traits<unsigned char>) New class.
4397 (int_traits<unsigned short>) Likewise.
4398 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
4399 Use GET_MODE_UNIT_PRECISION and remove redundant test for
4401 * combine.c (set_nonzero_bits_and_sign_copies): Use
4402 is_a <scalar_int_mode>.
4403 (find_split_point): Likewise.
4404 (combine_simplify_rtx): Likewise.
4405 (simplify_logical): Likewise.
4406 (expand_compound_operation): Likewise.
4407 (expand_field_assignment): Likewise.
4408 (make_compound_operation): Likewise.
4409 (extended_count): Likewise.
4410 (change_zero_ext): Likewise.
4411 (simplify_comparison): Likewise.
4412 * dwarf2out.c (scompare_loc_descriptor): Likewise.
4413 (ucompare_loc_descriptor): Likewise.
4414 (minmax_loc_descriptor): Likewise.
4415 (mem_loc_descriptor): Likewise.
4416 (loc_descriptor): Likewise.
4417 * expmed.c (init_expmed_one_mode): Likewise.
4418 * lra-constraints.c (lra_constraint_offset): Likewise.
4419 * optabs.c (prepare_libcall_arg): Likewise.
4420 * postreload.c (move2add_note_store): Likewise.
4421 * reload.c (operands_match_p): Likewise.
4422 * rtl.h (load_extend_op): Likewise.
4423 * rtlhooks.c (gen_lowpart_general): Likewise.
4424 * simplify-rtx.c (simplify_truncation): Likewise.
4425 (simplify_unary_operation_1): Likewise.
4426 (simplify_binary_operation_1): Likewise.
4427 (simplify_const_binary_operation): Likewise.
4428 (simplify_const_relational_operation): Likewise.
4429 (simplify_subreg): Likewise.
4430 * stor-layout.c (bitwise_mode_for_mode): Likewise.
4431 * var-tracking.c (adjust_mems): Likewise.
4432 (prepare_call_arguments): Likewise.
4434 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4435 Alan Hayward <alan.hayward@arm.com>
4436 David Sherwood <david.sherwood@arm.com>
4438 * machmode.h (is_int_mode): New fuction.
4439 * combine.c (find_split_point): Use it.
4440 (combine_simplify_rtx): Likewise.
4441 (simplify_if_then_else): Likewise.
4442 (simplify_set): Likewise.
4443 (simplify_shift_const_1): Likewise.
4444 (simplify_comparison): Likewise.
4445 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
4446 * cse.c (notreg_cost): Likewise.
4447 (cse_insn): Likewise.
4448 * cselib.c (cselib_lookup_1): Likewise.
4449 * dojump.c (do_jump_1): Likewise.
4450 (do_compare_rtx_and_jump): Likewise.
4451 * dse.c (get_call_args): Likewise.
4452 * dwarf2out.c (rtl_for_decl_init): Likewise.
4453 (native_encode_initializer): Likewise.
4454 * expmed.c (emit_store_flag_1): Likewise.
4455 (emit_store_flag): Likewise.
4456 * expr.c (convert_modes): Likewise.
4457 (store_field): Likewise.
4458 (expand_expr_real_1): Likewise.
4459 * fold-const.c (fold_read_from_constant_string): Likewise.
4460 * gimple-ssa-sprintf.c (get_format_string): Likewise.
4461 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
4462 * optabs.c (expand_binop): Likewise.
4463 (expand_unop): Likewise.
4464 (expand_abs_nojump): Likewise.
4465 (expand_one_cmpl_abs_nojump): Likewise.
4466 * simplify-rtx.c (mode_signbit_p): Likewise.
4467 (val_signbit_p): Likewise.
4468 (val_signbit_known_set_p): Likewise.
4469 (val_signbit_known_clear_p): Likewise.
4470 (simplify_relational_operation_1): Likewise.
4471 * tree.c (vector_type_mode): Likewise.
4473 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4474 Alan Hayward <alan.hayward@arm.com>
4475 David Sherwood <david.sherwood@arm.com>
4477 * machmode.h (smallest_mode_for_size): Fix formatting.
4478 (smallest_int_mode_for_size): New function.
4479 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
4480 instead of smallest_mode_for_size.
4481 * combine.c (make_extraction): Likewise.
4482 * config/arc/arc.c (arc_expand_movmem): Likewise.
4483 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
4484 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
4485 * config/s390/s390.c (s390_expand_insv): Likewise.
4486 * config/sparc/sparc.c (assign_int_registers): Likewise.
4487 * config/spu/spu.c (spu_function_value): Likewise.
4488 (spu_function_arg): Likewise.
4489 * coverage.c (get_gcov_type): Likewise.
4490 (get_gcov_unsigned_t): Likewise.
4491 * dse.c (find_shift_sequence): Likewise.
4492 * expmed.c (store_bit_field_1): Likewise.
4493 * expr.c (convert_move): Likewise.
4494 (store_field): Likewise.
4495 * internal-fn.c (expand_arith_overflow): Likewise.
4496 * optabs-query.c (get_best_extraction_insn): Likewise.
4497 * optabs.c (expand_twoval_binop_libfunc): Likewise.
4498 * stor-layout.c (layout_type): Likewise.
4499 (initialize_sizetypes): Likewise.
4500 * targhooks.c (default_get_mask_mode): Likewise.
4501 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
4503 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4504 Alan Hayward <alan.hayward@arm.com>
4505 David Sherwood <david.sherwood@arm.com>
4507 * machmode.h (opt_mode::else_blk): New function.
4508 (int_mode_for_mode): Declare.
4509 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
4510 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
4512 * cfgexpand.c (expand_debug_expr): Likewise.
4513 * combine.c (gen_lowpart_or_truncate): Likewise.
4514 (gen_lowpart_for_combine): Likewise.
4515 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
4516 * config/avr/avr.c (avr_to_int_mode): Likewise.
4517 (avr_out_plus_1): Likewise.
4518 (avr_out_plus): Likewise.
4519 (avr_out_round): Likewise.
4520 * config/i386/i386.c (ix86_split_to_parts): Likewise.
4521 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
4522 (s390_expand_vcond): Likewise.
4523 * config/spu/spu.c (spu_split_immediate): Likewise.
4524 (spu_expand_mov): Likewise.
4525 * dse.c (get_stored_val): Likewise.
4526 * expmed.c (store_bit_field_1): Likewise.
4527 (convert_extracted_bit_field): Use int_mode_for_mode instead of
4529 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
4530 (extract_low_bits): Likewise.
4531 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
4532 handling rather than repeating the check.
4533 (emit_group_store): Likewise.
4534 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
4535 * optabs.c (expand_absneg_bit): Likewise.
4536 (expand_copysign_absneg): Likewise.
4537 (expand_copysign_bit): Likewise.
4538 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
4539 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
4540 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
4541 * var-tracking.c (prepare_call_arguments): Likewise.
4542 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
4543 int_mode_for_mode instead of mode_for_size.
4544 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
4546 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4547 Alan Hayward <alan.hayward@arm.com>
4548 David Sherwood <david.sherwood@arm.com>
4550 * machmode.h (int_mode_for_size): New function.
4551 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
4552 instead of mode_for_size.
4553 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
4555 * combine.c (expand_field_assignment): Use int_mode_for_size
4556 instead of mode_for_size.
4557 (make_extraction): Likewise.
4558 (simplify_shift_const_1): Likewise.
4559 (simplify_comparison): Likewise.
4560 * dojump.c (do_jump): Likewise.
4561 * dwarf2out.c (mem_loc_descriptor): Likewise.
4562 * emit-rtl.c (init_derived_machine_modes): Likewise.
4563 * expmed.c (flip_storage_order): Likewise.
4564 (convert_extracted_bit_field): Likewise.
4565 * expr.c (copy_blkmode_from_reg): Likewise.
4566 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
4567 * internal-fn.c (expand_mul_overflow): Likewise.
4568 * lower-subreg.c (simple_move): Likewise.
4569 * optabs-libfuncs.c (init_optabs): Likewise.
4570 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4571 * tree.c (vector_type_mode): Likewise.
4572 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
4573 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
4574 * tree-vect-generic.c (expand_vector_parallel): Likewise.
4575 * tree-vect-stmts.c (vectorizable_load): Likewise.
4576 (vectorizable_store): Likewise.
4578 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4579 Alan Hayward <alan.hayward@arm.com>
4580 David Sherwood <david.sherwood@arm.com>
4582 * coretypes.h (pod_mode): New type.
4583 (scalar_int_mode_pod): New typedef.
4584 * machmode.h (pod_mode): New class.
4585 (int_n_data_t::m): Change type to scalar_int_mode_pod.
4586 * genmodes.c (emit_mode_int_n): Update accordingly.
4587 * lower-subreg.h (target_lower_subreg): Change type to
4588 scalar_int_mode_pod.
4589 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
4590 scalar_int_mode_pod.
4592 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4593 Alan Hayward <alan.hayward@arm.com>
4594 David Sherwood <david.sherwood@arm.com>
4596 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
4597 machine_mode to scalar_int_mode.
4598 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
4599 (rs6000_option_override_internal): Remove cast to int.
4600 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
4601 machine_mode to scalar_int_mode.
4602 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
4603 (rs6000_option_override_internal): Remove cast to int.
4604 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
4605 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
4607 * config/s390/s390.c (s390_expand_builtin): Likewise.
4608 * coretypes.h (scalar_int_mode): New type.
4609 (opt_scalar_int_mode): New typedef.
4610 * machmode.h (scalar_int_mode): New class.
4611 (scalar_int_mode::includes_p): New function.
4612 (byte_mode): Change type to scalar_int_mode.
4613 (word_mode): Likewise.
4614 (ptr_mode): Likewise.
4615 * emit-rtl.c (byte_mode): Likewise.
4616 (word_mode): Likewise.
4617 (ptr_mode): Likewise.
4618 (init_derived_machine_modes): Update accordingly.
4619 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
4620 and MODE_PARTIAL_INT.
4621 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
4622 opt_scalar_int_mode.
4624 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4625 Alan Hayward <alan.hayward@arm.com>
4626 David Sherwood <david.sherwood@arm.com>
4628 * target.def (libgcc_floating_mode_supported_p): Take a
4630 * doc/tm.texi: Regenerate.
4631 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
4633 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
4634 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
4637 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4638 Alan Hayward <alan.hayward@arm.com>
4639 David Sherwood <david.sherwood@arm.com>
4641 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
4642 * doc/tm.texi: Regenerate.
4643 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
4644 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
4645 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
4646 * targhooks.h (default_floatn_mode): Likewise.
4647 * targhooks.c (default_floatn_mode): Likewise.
4648 * tree.c (build_common_tree_nodes): Update accordingly.
4650 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4651 Alan Hayward <alan.hayward@arm.com>
4652 David Sherwood <david.sherwood@arm.com>
4654 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
4655 (mode_iterator::iterate_p): Likewise.
4656 (mode_iterator::get_wider): Likewise.
4657 * expr.c (init_expr_target): Use opt_scalar_float_mode.
4659 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4660 Alan Hayward <alan.hayward@arm.com>
4661 David Sherwood <david.sherwood@arm.com>
4663 * coretypes.h (opt_scalar_float_mode): New typedef.
4664 * machmode.h (float_mode_for_size): New function.
4665 * emit-rtl.c (double_mode): Delete.
4666 (init_emit_once): Use float_mode_for_size.
4667 * stor-layout.c (layout_type): Likewise.
4668 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
4670 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4671 Alan Hayward <alan.hayward@arm.com>
4672 David Sherwood <david.sherwood@arm.com>
4674 * output.h (assemble_real): Take a scalar_float_mode.
4675 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
4676 * config/arm/arm.md (consttable_4): Likewise.
4677 (consttable_8): Likewise.
4678 (consttable_16): Likewise.
4679 * config/mips/mips.md (consttable_float): Likewise.
4680 * config/s390/s390.c (s390_output_pool_entry): Likewise.
4681 * varasm.c (assemble_real): Take a scalar_float_mode.
4682 (output_constant_pool_2): Update accordingly.
4683 (output_constant): Likewise.
4685 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4686 Alan Hayward <alan.hayward@arm.com>
4687 David Sherwood <david.sherwood@arm.com>
4689 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
4690 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
4691 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
4692 (native_encode_real): Likewise.
4693 (native_interpret_real): Likewise.
4694 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
4695 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
4697 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4698 Alan Hayward <alan.hayward@arm.com>
4699 David Sherwood <david.sherwood@arm.com>
4701 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
4702 <scalar_float_mode>. Simplify.
4703 (gen_extend_conv_libfunc): Likewise.
4705 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4706 Alan Hayward <alan.hayward@arm.com>
4707 David Sherwood <david.sherwood@arm.com>
4709 * coretypes.h (scalar_float_mode): New type.
4710 * machmode.h (mode_traits::from_int): Use machine_mode if
4711 USE_ENUM_MODES is defined.
4712 (is_a): New function.
4714 (dyn_cast): Likewise.
4715 (scalar_float_mode): New class.
4716 (scalar_float_mode::includes_p): New function.
4717 (is_float_mode): Likewise.
4718 * gdbhooks.py (MachineModePrinter): New class.
4719 (build_pretty_printer): Use it for scalar_float_mode.
4720 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
4721 (format_helper::format_helper): Turn into a template.
4722 * genmodes.c (get_mode_class): New function.
4723 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
4724 or machine_mode if none.
4725 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
4726 as_a <scalar_float_mode>.
4727 * dwarf2out.c (mem_loc_descriptor): Likewise.
4728 (insert_float): Likewise.
4729 (add_const_value_attribute): Likewise.
4730 * simplify-rtx.c (simplify_immed_subreg): Likewise.
4731 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
4732 (expand_unop): Update accordingly.
4733 (expand_abs_nojump): Likewise.
4734 (expand_copysign_absneg): Take a scalar_float_mode.
4735 (expand_copysign_bit): Likewise.
4736 (expand_copysign): Update accordingly.
4738 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4739 Alan Hayward <alan.hayward@arm.com>
4740 David Sherwood <david.sherwood@arm.com>
4742 * coretypes.h (opt_mode): New class.
4743 * machmode.h (opt_mode): Likewise.
4744 (opt_mode::else_void): New function.
4745 (opt_mode::require): Likewise.
4746 (opt_mode::exists): Likewise.
4747 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
4748 (GET_MODE_2XWIDER_MODE): Likewise.
4749 (mode_iterator::get_wider): Update accordingly.
4750 (mode_iterator::get_2xwider): Likewise.
4751 (mode_iterator::get_known_wider): Likewise, turning into a template.
4752 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
4753 forcing a wider mode to exist.
4754 * config/cr16/cr16.h (LONG_REG_P): Likewise.
4755 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
4756 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
4757 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
4758 * lower-subreg.c (init_lower_subreg): Likewise.
4759 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
4760 on the final iteration.
4761 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
4762 a wider mode exists before asking for a move pattern.
4763 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
4764 forcing a wider mode to exist.
4765 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
4766 returning false if no such mode exists.
4767 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
4768 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
4769 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
4770 Avoid checking for a MODE_INT if we already know the mode is not a
4772 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
4773 forcing a wider mode to exist.
4774 (expmed_mult_highpart_optab): Likewise.
4775 (expmed_mult_highpart): Likewise.
4776 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
4778 * lto-streamer-in.c (lto_input_mode_table): Likewise.
4779 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
4780 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
4781 * internal-fn.c (expand_mul_overflow): Update use of
4782 GET_MODE_2XWIDER_MODE.
4783 * omp-low.c (omp_clause_aligned_alignment): Likewise.
4784 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
4785 GET_MODE_WIDER_MODE.
4786 (convert_plusminus_to_widen): Likewise.
4787 * tree-switch-conversion.c (array_value_type): Likewise.
4788 * var-tracking.c (emit_note_insn_var_location): Likewise.
4789 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
4790 Return false inside rather than outside the loop if no wider mode
4792 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
4793 and GET_MODE_2XWIDER_MODE
4794 (can_compare_p): Use else_void.
4795 * gdbhooks.py (OptMachineModePrinter): New class.
4796 (build_pretty_printer): Use it for opt_mode.
4798 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4799 Alan Hayward <alan.hayward@arm.com>
4800 David Sherwood <david.sherwood@arm.com>
4802 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
4803 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
4805 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4806 Alan Hayward <alan.hayward@arm.com>
4807 David Sherwood <david.sherwood@arm.com>
4809 * machmode.h (mode_traits): New structure.
4810 (get_narrowest_mode): New function.
4811 (mode_iterator::start): Likewise.
4812 (mode_iterator::iterate_p): Likewise.
4813 (mode_iterator::get_wider): Likewise.
4814 (mode_iterator::get_known_wider): Likewise.
4815 (mode_iterator::get_2xwider): Likewise.
4816 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
4817 (FOR_EACH_MODE): Likewise.
4818 (FOR_EACH_MODE_FROM): Likewise.
4819 (FOR_EACH_MODE_UNTIL): Likewise.
4820 (FOR_EACH_WIDER_MODE): Likewise.
4821 (FOR_EACH_2XWIDER_MODE): Likewise.
4822 * builtins.c (expand_builtin_strlen): Use new mode iterators.
4823 * combine.c (simplify_comparison): Likewise
4824 * config/i386/i386.c (type_natural_mode): Likewise.
4825 * cse.c (cse_insn): Likewise.
4826 * dse.c (find_shift_sequence): Likewise.
4827 * emit-rtl.c (init_derived_machine_modes): Likewise.
4828 (init_emit_once): Likewise.
4829 * explow.c (hard_function_value): Likewise.
4830 * expmed.c (extract_fixed_bit_field_1): Likewise.
4831 (extract_bit_field_1): Likewise.
4832 (expand_divmod): Likewise.
4833 (emit_store_flag_1): Likewise.
4834 * expr.c (init_expr_target): Likewise.
4835 (convert_move): Likewise.
4836 (alignment_for_piecewise_move): Likewise.
4837 (widest_int_mode_for_size): Likewise.
4838 (emit_block_move_via_movmem): Likewise.
4839 (copy_blkmode_to_reg): Likewise.
4840 (set_storage_via_setmem): Likewise.
4841 (compress_float_constant): Likewise.
4842 * omp-low.c (omp_clause_aligned_alignment): Likewise.
4843 * optabs-query.c (get_best_extraction_insn): Likewise.
4844 * optabs.c (expand_binop): Likewise.
4845 (expand_twoval_unop): Likewise.
4846 (expand_twoval_binop): Likewise.
4847 (widen_leading): Likewise.
4848 (widen_bswap): Likewise.
4849 (expand_parity): Likewise.
4850 (expand_unop): Likewise.
4851 (prepare_cmp_insn): Likewise.
4852 (prepare_float_lib_cmp): Likewise.
4853 (expand_float): Likewise.
4854 (expand_fix): Likewise.
4855 (expand_sfix_optab): Likewise.
4856 * postreload.c (move2add_use_add2_insn): Likewise.
4857 * reg-stack.c (reg_to_stack): Likewise.
4858 * reginfo.c (choose_hard_reg_mode): Likewise.
4859 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
4860 * stor-layout.c (mode_for_size): Likewise.
4861 (smallest_mode_for_size): Likewise.
4862 (mode_for_vector): Likewise.
4863 (finish_bitfield_representative): Likewise.
4864 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
4865 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
4866 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
4867 * var-tracking.c (prepare_call_arguments): Likewise.
4869 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4870 Alan Hayward <alan.hayward@arm.com>
4871 David Sherwood <david.sherwood@arm.com>
4873 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
4874 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
4875 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
4876 * machmode.h (mode_size): Move earlier in file.
4877 (mode_precision): Likewise.
4878 (mode_inner): Likewise.
4879 (mode_nunits): Likewise.
4880 (mode_unit_size): Likewise.
4881 (unit_unit_precision): Likewise.
4882 (mode_wider): Likewise.
4883 (mode_2xwider): Likewise.
4884 (machine_mode): New class.
4885 (mode_to_bytes): New function.
4886 (mode_to_bits): Likewise.
4887 (mode_to_precision): Likewise.
4888 (mode_to_inner): Likewise.
4889 (mode_to_unit_size): Likewise.
4890 (mode_to_unit_precision): Likewise.
4891 (mode_to_nunits): Likewise.
4892 (GET_MODE_SIZE): Use mode_to_bytes.
4893 (GET_MODE_BITSIZE): Use mode_to_bits.
4894 (GET_MODE_PRECISION): Use mode_to_precision.
4895 (GET_MODE_INNER): Use mode_to_inner.
4896 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
4897 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
4898 (GET_MODE_NUNITS): Use mode_to_nunits.
4899 * system.h (ALWAYS_INLINE): New macro.
4900 * config/powerpcspe/powerpcspe-c.c
4901 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
4902 int for arg1_mode and arg2_mode.
4904 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4905 Alan Hayward <alan.hayward@arm.com>
4906 David Sherwood <david.sherwood@arm.com>
4908 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
4909 Prefix mode names with E_ in case statements.
4910 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4911 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
4912 (aarch64_split_simd_move): Likewise.
4913 (aarch64_gen_storewb_pair): Likewise.
4914 (aarch64_gen_loadwb_pair): Likewise.
4915 (aarch64_gen_store_pair): Likewise.
4916 (aarch64_gen_load_pair): Likewise.
4917 (aarch64_get_condition_code_1): Likewise.
4918 (aarch64_constant_pool_reload_icode): Likewise.
4919 (get_rsqrte_type): Likewise.
4920 (get_rsqrts_type): Likewise.
4921 (get_recpe_type): Likewise.
4922 (get_recps_type): Likewise.
4923 (aarch64_gimplify_va_arg_expr): Likewise.
4924 (aarch64_simd_container_mode): Likewise.
4925 (aarch64_emit_load_exclusive): Likewise.
4926 (aarch64_emit_store_exclusive): Likewise.
4927 (aarch64_expand_compare_and_swap): Likewise.
4928 (aarch64_gen_atomic_cas): Likewise.
4929 (aarch64_emit_bic): Likewise.
4930 (aarch64_emit_atomic_swap): Likewise.
4931 (aarch64_emit_atomic_load_op): Likewise.
4932 (aarch64_evpc_trn): Likewise.
4933 (aarch64_evpc_uzp): Likewise.
4934 (aarch64_evpc_zip): Likewise.
4935 (aarch64_evpc_ext): Likewise.
4936 (aarch64_evpc_rev): Likewise.
4937 (aarch64_evpc_dup): Likewise.
4938 (aarch64_gen_ccmp_first): Likewise.
4939 (aarch64_gen_ccmp_next): Likewise.
4940 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
4941 (alpha_emit_xfloating_libcall): Likewise.
4942 (emit_insxl): Likewise.
4943 (alpha_arg_type): Likewise.
4944 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
4945 (arc_preferred_simd_mode): Likewise.
4946 (arc_secondary_reload): Likewise.
4947 (get_arc_condition_code): Likewise.
4948 (arc_print_operand): Likewise.
4949 (arc_legitimate_constant_p): Likewise.
4950 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4951 * config/arc/arc.md (casesi_load): Likewise.
4952 (casesi_compact_jump): Likewise.
4953 * config/arc/predicates.md (proper_comparison_operator): Likewise.
4954 (cc_use_register): Likewise.
4955 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4956 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
4957 (arm_init_iwmmxt_builtins): Likewise.
4958 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
4959 (neon_expand_vector_init): Likewise.
4960 (arm_attr_length_move_neon): Likewise.
4961 (maybe_get_arm_condition_code): Likewise.
4962 (arm_emit_vector_const): Likewise.
4963 (arm_preferred_simd_mode): Likewise.
4964 (arm_output_iwmmxt_tinsr): Likewise.
4965 (thumb1_output_casesi): Likewise.
4966 (thumb2_output_casesi): Likewise.
4967 (arm_emit_load_exclusive): Likewise.
4968 (arm_emit_store_exclusive): Likewise.
4969 (arm_expand_compare_and_swap): Likewise.
4970 (arm_evpc_neon_vuzp): Likewise.
4971 (arm_evpc_neon_vzip): Likewise.
4972 (arm_evpc_neon_vrev): Likewise.
4973 (arm_evpc_neon_vtrn): Likewise.
4974 (arm_evpc_neon_vext): Likewise.
4975 (arm_validize_comparison): Likewise.
4976 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
4977 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
4978 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
4979 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
4980 (c6x_preferred_simd_mode): Likewise.
4981 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
4982 (epiphany_rtx_costs): Likewise.
4983 * config/epiphany/predicates.md (proper_comparison_operator):
4985 * config/frv/frv.c (condexec_memory_operand): Likewise.
4986 (frv_emit_move): Likewise.
4987 (output_move_single): Likewise.
4988 (output_condmove_single): Likewise.
4989 (frv_hard_regno_mode_ok): Likewise.
4990 (frv_matching_accg_mode): Likewise.
4991 * config/h8300/h8300.c (split_adds_subs): Likewise.
4992 (h8300_rtx_costs): Likewise.
4993 (h8300_print_operand): Likewise.
4994 (compute_mov_length): Likewise.
4995 (output_logical_op): Likewise.
4996 (compute_logical_op_length): Likewise.
4997 (compute_logical_op_cc): Likewise.
4998 (h8300_shift_needs_scratch_p): Likewise.
4999 (output_a_shift): Likewise.
5000 (compute_a_shift_length): Likewise.
5001 (compute_a_shift_cc): Likewise.
5002 (expand_a_rotate): Likewise.
5003 (output_a_rotate): Likewise.
5004 * config/i386/i386.c (classify_argument): Likewise.
5005 (function_arg_advance_32): Likewise.
5006 (function_arg_32): Likewise.
5007 (function_arg_64): Likewise.
5008 (function_value_64): Likewise.
5009 (ix86_gimplify_va_arg): Likewise.
5010 (ix86_legitimate_constant_p): Likewise.
5011 (put_condition_code): Likewise.
5012 (split_double_mode): Likewise.
5013 (ix86_avx256_split_vector_move_misalign): Likewise.
5014 (ix86_expand_vector_logical_operator): Likewise.
5015 (ix86_split_idivmod): Likewise.
5016 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
5017 (ix86_build_const_vector): Likewise.
5018 (ix86_build_signbit_mask): Likewise.
5019 (ix86_match_ccmode): Likewise.
5020 (ix86_cc_modes_compatible): Likewise.
5021 (ix86_expand_branch): Likewise.
5022 (ix86_expand_sse_cmp): Likewise.
5023 (ix86_expand_sse_movcc): Likewise.
5024 (ix86_expand_int_sse_cmp): Likewise.
5025 (ix86_expand_vec_perm_vpermi2): Likewise.
5026 (ix86_expand_vec_perm): Likewise.
5027 (ix86_expand_sse_unpack): Likewise.
5028 (ix86_expand_int_addcc): Likewise.
5029 (ix86_split_to_parts): Likewise.
5030 (ix86_vectorize_builtin_gather): Likewise.
5031 (ix86_vectorize_builtin_scatter): Likewise.
5032 (avx_vpermilp_parallel): Likewise.
5033 (inline_memory_move_cost): Likewise.
5034 (ix86_tieable_integer_mode_p): Likewise.
5035 (x86_maybe_negate_const_int): Likewise.
5036 (ix86_expand_vector_init_duplicate): Likewise.
5037 (ix86_expand_vector_init_one_nonzero): Likewise.
5038 (ix86_expand_vector_init_one_var): Likewise.
5039 (ix86_expand_vector_init_concat): Likewise.
5040 (ix86_expand_vector_init_interleave): Likewise.
5041 (ix86_expand_vector_init_general): Likewise.
5042 (ix86_expand_vector_set): Likewise.
5043 (ix86_expand_vector_extract): Likewise.
5044 (emit_reduc_half): Likewise.
5045 (ix86_emit_i387_round): Likewise.
5046 (ix86_mangle_type): Likewise.
5047 (ix86_expand_round_sse4): Likewise.
5048 (expand_vec_perm_blend): Likewise.
5049 (canonicalize_vector_int_perm): Likewise.
5050 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
5051 (expand_vec_perm_1): Likewise.
5052 (expand_vec_perm_interleave3): Likewise.
5053 (expand_vec_perm_even_odd_pack): Likewise.
5054 (expand_vec_perm_even_odd_1): Likewise.
5055 (expand_vec_perm_broadcast_1): Likewise.
5056 (ix86_vectorize_vec_perm_const_ok): Likewise.
5057 (ix86_expand_vecop_qihi): Likewise.
5058 (ix86_expand_mul_widen_hilo): Likewise.
5059 (ix86_expand_sse2_abs): Likewise.
5060 (ix86_expand_pextr): Likewise.
5061 (ix86_expand_pinsr): Likewise.
5062 (ix86_preferred_simd_mode): Likewise.
5063 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
5064 * config/i386/sse.md (*andnot<mode>3): Likewise.
5065 (<mask_codefor><code><mode>3<mask_name>): Likewise.
5066 (*<code><mode>3): Likewise.
5067 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
5068 (ia64_expand_atomic_op): Likewise.
5069 (ia64_arg_type): Likewise.
5070 (ia64_mode_to_int): Likewise.
5071 (ia64_scalar_mode_supported_p): Likewise.
5072 (ia64_vector_mode_supported_p): Likewise.
5073 (expand_vec_perm_broadcast): Likewise.
5074 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
5075 (iq2000_function_arg_advance): Likewise.
5076 (iq2000_function_arg): Likewise.
5077 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
5078 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
5079 (m68k_libcall_value): Likewise.
5080 (m68k_function_value): Likewise.
5081 (sched_attr_op_type): Likewise.
5082 * config/mcore/mcore.c (mcore_output_move): Likewise.
5083 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
5085 (microblaze_function_arg): Likewise.
5086 * config/mips/mips.c (mips16_build_call_stub): Likewise.
5087 (mips_print_operand): Likewise.
5088 (mips_mode_ok_for_mov_fmt_p): Likewise.
5089 (mips_vector_mode_supported_p): Likewise.
5090 (mips_preferred_simd_mode): Likewise.
5091 (mips_expand_vpc_loongson_even_odd): Likewise.
5092 (mips_expand_vec_unpack): Likewise.
5093 (mips_expand_vi_broadcast): Likewise.
5094 (mips_expand_vector_init): Likewise.
5095 (mips_expand_vec_reduc): Likewise.
5096 (mips_expand_msa_cmp): Likewise.
5097 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
5098 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
5099 (cc_flags_for_mode): Likewise.
5100 * config/msp430/msp430.c (msp430_print_operand): Likewise.
5101 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
5102 (nds32_output_casesi_pc_relative): Likewise.
5103 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
5104 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
5105 (nvptx_gen_unpack): Likewise.
5106 (nvptx_gen_pack): Likewise.
5107 (nvptx_gen_shuffle): Likewise.
5108 (nvptx_gen_wcast): Likewise.
5109 (nvptx_preferred_simd_mode): Likewise.
5110 * config/pa/pa.c (pa_secondary_reload): Likewise.
5111 * config/pa/predicates.md (base14_operand): Likewise.
5112 * config/powerpcspe/powerpcspe-c.c
5113 (altivec_resolve_overloaded_builtin): Likewise.
5114 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
5116 (rs6000_preferred_simd_mode): Likewise.
5117 (output_vec_const_move): Likewise.
5118 (rs6000_expand_vector_extract): Likewise.
5119 (rs6000_split_vec_extract_var): Likewise.
5120 (reg_offset_addressing_ok_p): Likewise.
5121 (rs6000_legitimate_offset_address_p): Likewise.
5122 (rs6000_legitimize_address): Likewise.
5123 (rs6000_emit_set_const): Likewise.
5124 (rs6000_const_vec): Likewise.
5125 (rs6000_emit_move): Likewise.
5126 (spe_build_register_parallel): Likewise.
5127 (rs6000_darwin64_record_arg_recurse): Likewise.
5128 (swap_selector_for_mode): Likewise.
5129 (spe_init_builtins): Likewise.
5130 (paired_init_builtins): Likewise.
5131 (altivec_init_builtins): Likewise.
5132 (do_load_for_compare): Likewise.
5133 (rs6000_generate_compare): Likewise.
5134 (rs6000_expand_float128_convert): Likewise.
5135 (emit_load_locked): Likewise.
5136 (emit_store_conditional): Likewise.
5137 (rs6000_output_function_epilogue): Likewise.
5138 (rs6000_handle_altivec_attribute): Likewise.
5139 (rs6000_function_value): Likewise.
5140 (emit_fusion_gpr_load): Likewise.
5141 (emit_fusion_p9_load): Likewise.
5142 (emit_fusion_p9_store): Likewise.
5143 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
5144 (fusion_gpr_mem_load): Likewise.
5145 (fusion_addis_mem_combo_load): Likewise.
5146 (fusion_addis_mem_combo_store): Likewise.
5147 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
5148 (fusion_gpr_mem_load): Likewise.
5149 (fusion_addis_mem_combo_load): Likewise.
5150 (fusion_addis_mem_combo_store): Likewise.
5151 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5153 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
5154 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
5155 (rs6000_preferred_simd_mode): Likewise.
5156 (output_vec_const_move): Likewise.
5157 (rs6000_expand_vector_extract): Likewise.
5158 (rs6000_split_vec_extract_var): Likewise.
5159 (reg_offset_addressing_ok_p): Likewise.
5160 (rs6000_legitimate_offset_address_p): Likewise.
5161 (rs6000_legitimize_address): Likewise.
5162 (rs6000_emit_set_const): Likewise.
5163 (rs6000_const_vec): Likewise.
5164 (rs6000_emit_move): Likewise.
5165 (rs6000_darwin64_record_arg_recurse): Likewise.
5166 (swap_selector_for_mode): Likewise.
5167 (paired_init_builtins): Likewise.
5168 (altivec_init_builtins): Likewise.
5169 (rs6000_expand_float128_convert): Likewise.
5170 (emit_load_locked): Likewise.
5171 (emit_store_conditional): Likewise.
5172 (rs6000_output_function_epilogue): Likewise.
5173 (rs6000_handle_altivec_attribute): Likewise.
5174 (rs6000_function_value): Likewise.
5175 (emit_fusion_gpr_load): Likewise.
5176 (emit_fusion_p9_load): Likewise.
5177 (emit_fusion_p9_store): Likewise.
5178 * config/rx/rx.c (rx_gen_move_template): Likewise.
5179 (flags_from_mode): Likewise.
5180 * config/s390/predicates.md (s390_alc_comparison): Likewise.
5181 (s390_slb_comparison): Likewise.
5182 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
5183 (s390_vector_mode_supported_p): Likewise.
5184 (s390_cc_modes_compatible): Likewise.
5185 (s390_match_ccmode_set): Likewise.
5186 (s390_canonicalize_comparison): Likewise.
5187 (s390_emit_compare_and_swap): Likewise.
5188 (s390_branch_condition_mask): Likewise.
5189 (s390_rtx_costs): Likewise.
5190 (s390_secondary_reload): Likewise.
5191 (__SECONDARY_RELOAD_CASE): Likewise.
5192 (s390_expand_cs): Likewise.
5193 (s390_preferred_simd_mode): Likewise.
5194 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
5195 * config/sh/sh.c (sh_print_operand): Likewise.
5196 (dump_table): Likewise.
5197 (sh_secondary_reload): Likewise.
5198 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
5199 * config/sh/sh.md (casesi_worker_1): Likewise.
5200 (casesi_worker_2): Likewise.
5201 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
5202 (fcc_comparison_operator): Likewise.
5203 * config/sparc/sparc.c (sparc_expand_move): Likewise.
5204 (emit_soft_tfmode_cvt): Likewise.
5205 (sparc_preferred_simd_mode): Likewise.
5206 (output_cbranch): Likewise.
5207 (sparc_print_operand): Likewise.
5208 (sparc_expand_vec_perm_bmask): Likewise.
5209 (vector_init_bshuffle): Likewise.
5210 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
5211 (spu_vector_mode_supported_p): Likewise.
5212 (spu_expand_insv): Likewise.
5213 (spu_emit_branch_or_set): Likewise.
5214 (spu_handle_vector_attribute): Likewise.
5215 (spu_builtin_splats): Likewise.
5216 (spu_builtin_extract): Likewise.
5217 (spu_builtin_promote): Likewise.
5218 (spu_expand_sign_extend): Likewise.
5219 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
5220 (tilegx_simd_int): Likewise.
5221 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
5222 (tilepro_simd_int): Likewise.
5223 * config/v850/v850.c (const_double_split): Likewise.
5224 (v850_print_operand): Likewise.
5225 (ep_memory_offset): Likewise.
5226 * config/vax/vax.c (vax_rtx_costs): Likewise.
5227 (vax_output_int_move): Likewise.
5228 (vax_output_int_add): Likewise.
5229 (vax_output_int_subtract): Likewise.
5230 * config/visium/predicates.md (visium_branch_operator): Likewise.
5231 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
5232 (visium_print_operand_address): Likewise.
5233 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
5234 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
5235 (xtensa_expand_conditional_branch): Likewise.
5236 (xtensa_copy_incoming_a7): Likewise.
5237 (xtensa_output_literal): Likewise.
5238 * dfp.c (decimal_real_maxval): Likewise.
5239 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
5241 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5242 Alan Hayward <alan.hayward@arm.com>
5243 David Sherwood <david.sherwood@arm.com>
5245 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
5246 (mode_nunits_inline): Likewise.
5247 (mode_inner_inline): Likewise.
5248 (mode_unit_size_inline): Likewise.
5249 (mode_unit_precision_inline): Likewise.
5250 (emit_insn_modes_h): Likewise. Also emit a #define of the
5252 (emit_mode_wider): Add an E_ prefix to mode names.
5253 (emit_mode_complex): Likewise.
5254 (emit_mode_inner): Likewise.
5255 (emit_mode_adjustments): Likewise.
5256 (emit_mode_int_n): Likewise.
5257 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
5258 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
5259 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
5260 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
5261 (CRC32_BUILTIN, ENTRY): Likewise.
5262 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
5263 (aarch64_pop_regs): Likewise.
5264 (aarch64_process_components): Likewise.
5265 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
5266 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
5267 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
5268 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
5269 * config/arm/arm.c (arm_init_libfuncs): Likewise.
5270 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
5272 * config/i386/i386-builtin.def (pcmpestr): Likewise.
5273 (pcmpistr): Likewise.
5274 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
5275 * config/mmix/mmix.c (mmix_output_condition): Likewise.
5276 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
5278 * config/rl78/rl78.c (mduc_regs): Likewise.
5279 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
5280 (htm_expand_builtin): Likewise.
5281 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
5282 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
5283 * config/xtensa/xtensa.c (print_operand): Likewise.
5284 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
5285 (NUM_MODE_VECTOR_INT): Likewise.
5286 * genoutput.c (null_operand): Likewise.
5287 (output_operand_data): Likewise.
5288 * genrecog.c (print_parameter_value): Likewise.
5289 * lra.c (debug_operand_data): Likewise.
5291 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5293 * dwarf2out.c (scompare_loc_descriptor_wide)
5294 (scompare_loc_descriptor_narrow): New functions, split out from...
5295 (scompare_loc_descriptor): ...here.
5296 * expmed.c (emit_store_flag_int): New function, split out from...
5297 (emit_store_flag): ...here.
5299 2017-08-30 Richard Biener <rguenther@suse.de>
5301 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
5302 (dwarf2out_early_finish): Move setting of AT_pubnames from
5303 early debug output to early finish.
5305 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
5307 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
5308 and -mdata-region to the assembler.
5310 2017-08-30 Richard Biener <rguenther@suse.de>
5312 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
5314 (gen_subprogram_die): Add DW_AT_object_pointer only early.
5315 (dwarf2out_early_global_decl): Only generate a DIE for the
5316 abstract origin if it doesn't already exist or is a declaration DIE.
5317 (resolve_addr): Do not add the linkage name twice when
5318 generating a stub DIE for the DW_TAG_GNU_call_site target.
5320 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5322 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5323 Use machine_mode rather than int for arg1_mode.
5325 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
5328 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
5329 that the second argument of the built-in functions to unpack
5330 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
5331 switch statement instead a lot of if statements.
5332 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
5333 Allow 64-bit values to be in Altivec registers as well as
5334 traditional floating point registers.
5335 (pack<mode>, FMOVE128_VSX iterator): Likewise.
5337 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
5339 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
5340 MAX_REGS_PER_ADDRESS == 1.
5342 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
5344 * config/i386/i386.opt (flag_fentry): Do not init to -1.
5345 * config/i386/i386.c (ix86_option_override_internal): Simplify
5346 setting of opts->x_flag_entry.
5348 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5349 Jakub Jelinek <jakub@redhat.com>
5350 Richard Biener <rguenther@suse.de>
5352 PR tree-optimization/81503
5353 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
5354 folded constant fits in the target type; reorder tests for clarity.
5356 2017-08-29 Martin Liska <mliska@suse.cz>
5358 * passes.def: Include pass_lower_switch.
5359 * stmt.c (dump_case_nodes): Remove and move to
5360 tree-switch-conversion.
5361 (case_values_threshold): Likewise.
5362 (expand_switch_as_decision_tree_p): Likewise.
5363 (emit_case_decision_tree): Likewise.
5364 (expand_case): Likewise.
5365 (balance_case_nodes): Likewise.
5366 (node_has_low_bound): Likewise.
5367 (node_has_high_bound): Likewise.
5368 (node_is_bounded): Likewise.
5369 (emit_case_nodes): Likewise.
5370 (struct simple_case_node): New struct.
5371 (add_case_node): Remove.
5372 (emit_case_dispatch_table): Use vector instead of case_list.
5373 (reset_out_edges_aux): Remove.
5374 (compute_cases_per_edge): Likewise.
5375 (expand_case): Build list of simple_case_node.
5376 (expand_sjlj_dispatch_table): Use it.
5377 * tree-switch-conversion.c (struct case_node): Moved from
5378 stmt.c and adjusted.
5379 (emit_case_nodes): Likewise.
5380 (node_has_low_bound): Likewise.
5381 (node_has_high_bound): Likewise.
5382 (node_is_bounded): Likewise.
5383 (case_values_threshold): Likewise.
5384 (reset_out_edges_aux): Likewise.
5385 (compute_cases_per_edge): Likewise.
5386 (add_case_node): Likewise.
5387 (dump_case_nodes): Likewise.
5388 (balance_case_nodes): Likewise.
5389 (expand_switch_as_decision_tree_p): Likewise.
5390 (emit_jump): Likewise.
5391 (emit_case_decision_tree): Likewise.
5392 (try_switch_expansion): Likewise.
5393 (do_jump_if_equal): Likewise.
5394 (emit_cmp_and_jump_insns): Likewise.
5395 (fix_phi_operands_for_edge): New function.
5396 (record_phi_operand_mapping): Likewise.
5397 (class pass_lower_switch): New pass.
5398 (pass_lower_switch::execute): New function.
5399 (make_pass_lower_switch): Likewise.
5400 (conditional_probability):
5401 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
5402 * tree-pass.h: Add make_pass_lower_switch.
5404 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
5407 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
5410 2017-08-29 Richard Biener <rguenther@suse.de>
5412 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
5413 we do not add a DW_AT_inline attribute twice.
5414 (gen_subprogram_die): Remove code setting DW_AT_inline on
5415 DECL_ABSTRACT_P nodes.
5417 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
5419 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
5420 calls to internal functions.
5421 (gimplify_modify_expr): Likewise.
5422 * tree-call-cdce.c (use_internal_fn): Likewise.
5423 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
5424 (convert_to_divmod): Set the nothrow flag.
5425 * tree-if-conv.c (predicate_mem_writes): Likewise.
5426 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
5427 (vectorizable_call): Likewise.
5428 (vectorizable_store): Likewise.
5429 (vectorizable_load): Likewise.
5430 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
5431 (vect_recog_mask_conversion_pattern): Likewise.
5433 2017-08-29 Martin Liska <mliska@suse.cz>
5436 * gcc.c (driver_handle_option): Add new argument.
5437 * opts-common.c (handle_option): Pass
5438 target_option_override_hook.
5439 * opts-global.c (lang_handle_option): Add new option.
5440 (set_default_handlers): Add new argument.
5441 (decode_options): Likewise.
5442 * opts.c (target_handle_option): Likewise.
5443 (common_handle_option): Call target_option_override_hook.
5444 * opts.h (struct cl_option_handler_func): Add hook for
5445 target option override.
5446 (struct cl_option_handlers): Likewise.
5447 (set_default_handlers): Add new argument.
5448 (decode_options): Likewise.
5449 (common_handle_option): Likewise.
5450 (target_handle_option): Likewise.
5451 * toplev.c (toplev::main): Pass targetm.target_option.override
5454 2017-08-29 Richard Biener <rguenther@suse.de>
5455 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
5457 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
5458 life to the active subtree.
5460 2017-08-28 Jeff Law <law@redhat.com>
5462 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
5463 derive_equivalences.
5464 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
5466 (edge_info::derive_equivalences): New private member function
5468 * tree-ssa-dom.c (class edge_info): Changed from a struct
5469 to a class. Add ctor/dtor, methods and data members.
5470 (edge_info::edge_info): Renamed from allocate_edge_info.
5471 Initialize additional members.
5472 (edge_info::~edge_info): New.
5473 (free_dom_edge_info): Delete the edge info.
5474 (record_edge_info): Use new class & associated member functions.
5475 Tighten forms for testing for edge equivalences.
5476 (record_temporary_equivalences): Iterate over the simple
5477 equivalences rather than assuming there's only one per edge.
5478 (cprop_into_successor_phis): Iterate over the simple
5479 equivalences rather than assuming there's only one per edge.
5480 (optimize_stmt): Use operand_equal_p rather than pointer
5481 equality for mini-DSE code.
5483 2017-08-28 Nathan Sidwell <nathan@acm.org>
5485 * gcc.c (execute): Fold SIGPIPE handling into switch
5486 statement. Adjust internal error message.
5488 2017-08-28 Richard Biener <rguenther@suse.de>
5491 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
5492 Do nothing for removed DIEs.
5494 2017-08-28 Richard Biener <rguenther@suse.de>
5496 PR tree-optimization/81977
5497 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
5500 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
5503 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
5504 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
5505 using targetm.gen_mem_thread_fence.
5507 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
5510 * config/i386/i386.md (*<btsc><mode>): Change operand 2
5511 predicate to register_operand. Reorder operands.
5512 (*btr<mode>): Ditto.
5513 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
5514 (*btr<mode>_mask): Ditto.
5516 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
5518 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
5519 * config/rs6000/xmmintrin.h: New file.
5520 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
5522 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5525 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
5526 parameter and_insn and return it.
5527 (recombine_lvx_pattern): Insert a copy to ensure availability of
5528 the base register of the copied masking operation at the point of
5529 the instruction replacement.
5530 (recombine_stvx_pattern): Likewise.
5532 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
5534 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
5535 undocumented switches.
5536 (-mpower9-dform-vector): Likewise.
5537 (-mpower9-dform): Likewise.
5538 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
5539 comments to delete references to -mpower9-dform* switches.
5540 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
5541 Delete reference to -mpower9-dform* switches, test for
5542 -mpower9-vector instead.
5543 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
5544 (OTHER_P9_VECTOR_MASKS): Likewise.
5545 (POWERPC_MASKS): Likewise.
5546 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
5547 tests against -mpower9-dform* to -mpower9-vector. Delete code
5548 that checked for -mpower9-dform* consistancy with other options.
5549 Add test for -mpower9-misc to enable other power9 switches.
5550 (rs6000_init_hard_regno_mode_ok): Likewise.
5551 (rs6000_option_override_internal): Likewise.
5552 (rs6000_emit_prologue): Likewise.
5553 (rs6000_emit_epilogue): Likewise.
5554 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
5555 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
5556 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
5558 (emit_fusion_p9_store): Likewise.
5559 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
5560 resetting these macros if the assembler does not support ISA 3.0
5562 (TARGET_P9_DFORM_VECTOR): Likewise.
5563 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
5564 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
5566 2017-08-25 Alan Modra <amodra@gmail.com>
5568 PR rtl-optimization/81747
5569 * cse.c (cse_extended_basic_block): Don't attempt to record
5570 equivalences for degenerate conditional branches that branch
5571 to their fall-through.
5573 2017-08-24 Martin Sebor <msebor@redhat.com>
5576 * gimple-fold.c (size_must_be_zero_p): New function.
5577 (gimple_fold_builtin_memory_op): Call it.
5579 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
5581 * config/rs6000/mm_malloc.h: New file.
5583 2017-08-24 Bin Cheng <bin.cheng@arm.com>
5585 PR tree-optimization/81913
5586 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
5587 analysis when either IVs in condition can wrap.
5589 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
5591 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
5592 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
5594 2017-08-24 Richard Biener <rguenther@suse.de>
5597 * targhooks.c (default_target_can_inline_p): Properly
5598 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
5599 is present and always compare.
5600 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
5601 infer -mfpmath=sse from TARGET_SSE_P.
5602 (ix86_can_inline_p): Properly use target_option_default_node when
5603 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
5605 2017-08-24 Richard Biener <rguenther@suse.de>
5608 * dwarf2out.c (output_die): Handle flag_generate_offload like
5610 (output_comp_unit): Likewise.
5611 (gen_array_type_die): Likewise.
5612 (dwarf2out_early_finish): Likewise.
5613 (note_variable_value_in_expr): Likewise.
5614 (dwarf2out_finish): Likewise. Adjust assert.
5615 * cgraphunit.c (symbol_table::compile): Move setting of
5616 flag_generate_offload earlier ...
5617 (symbol_table::finalize_compilation_unit): ... here, before
5618 early debug finalization.
5620 2017-08-24 Richard Biener <rguenther@suse.de>
5622 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
5623 and ipa-fnsummary.h.
5624 (ix86_can_inline_p): When ix86_fpmath flags do not match
5625 check whether the callee uses FP math at all.
5627 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
5630 * tree-ssanames.c (get_nonzero_bits): Use element_precision
5631 instead of TYPE_PRECISION.
5633 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
5634 Alan Hayward <alan.hayward@arm.com>
5635 David Sherwood <david.sherwood@arm.com>
5637 * combine.c (make_extraction): Use subreg_offset_from_lsb.
5639 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
5641 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
5643 (ix86_frame::stack_realign_allocate): New field.
5644 (struct machine_frame_state): Modify comments.
5645 (machine_frame_state::sp_realigned_fp_end): New field.
5646 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
5648 (sp_valid_at): Add assertion to assure no attempt to access invalid
5649 offset of a realigned stack.
5650 (fp_valid_at): Likewise.
5651 (choose_baseaddr): Modify comments.
5652 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
5653 ix86_expand_prologue.
5654 (ix86_expand_prologue): Modify stack realignment and allocation.
5655 (ix86_expand_epilogue): Modify comments.
5656 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
5657 avx2_runtime, avx512f, and avx512f_runtime.
5659 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
5661 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
5662 (mstackrealign): Do not init to -1.
5663 * config/i386/i386.c (ix86_option_override_internal):
5664 Check opts_set, not opts when setting default value of
5665 opts->x_ix86_force_align_arg_pointer.
5667 2017-08-23 Richard Biener <rguenther@suse.de>
5669 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
5670 lang_hooks.decl_printable_name.
5671 * print-rtl-function.c (print_rtx_function): Likewise.
5672 * tree-pretty-print.c (dump_function_header): Likewise.
5674 2017-08-23 Richard Biener <rguenther@suse.de>
5677 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
5678 -g0 at compile-time.
5680 2017-08-23 Tamar Christina <tamar.christina@arm.com>
5683 * doc/sourcebuild.texi (Other hardware attributes):
5686 2017-08-23 Tamar Christina <tamar.christina@arm.com>
5689 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
5690 Fix single-use check.
5692 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
5694 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
5696 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
5698 * doc/install.texi: Modify to add more details on running selected
5701 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
5703 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
5704 is combined with -mabi=ms.
5705 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
5708 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5710 PR tree-optimization/81488
5711 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
5712 and cached_basis fields.
5713 (MAX_SPREAD): New constant.
5714 (alloc_cand_and_find_basis): Initialize new fields.
5715 (clear_visited): New function.
5716 (create_phi_basis_1): Rename from create_phi_basis, set visited
5717 and cached_basis fields.
5718 (create_phi_basis): New wrapper function.
5719 (phi_add_costs_1): Rename from phi_add_costs, add spread
5720 parameter, set visited field, short-circuit when limits reached.
5721 (phi_add_costs): New wrapper function.
5722 (record_phi_increments_1): Rename from record_phi_increments, set
5724 (record_phi_increments): New wrapper function.
5725 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
5726 (phi_incr_cost): New wrapper function.
5727 (all_phi_incrs_profitable_1): Rename from
5728 all_phi_incrs_profitable, set visited field.
5729 (all_phi_incrs_profitable): New wrapper function.
5731 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
5732 Alan Hayward <alan.hayward@arm.com>
5733 David Sherwood <david.sherwood@arm.com>
5735 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
5736 that takes the outer and inner modes.
5737 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
5738 comparison as the canonical test for a paradoxical subreg.
5739 * combine.c (simplify_set): Use paradoxical_subreg_p.
5740 (make_extraction): Likewise.
5741 (force_to_mode): Likewise.
5742 (rtx_equal_for_field_assignment_p): Likewise.
5743 (gen_lowpart_for_combine): Likewise.
5744 (simplify_comparison): Likewise.
5745 * cse.c (equiv_constant): Likewise.
5746 * expmed.c (store_bit_field_1): Likewise.
5747 * final.c (alter_subreg): Likewise.
5748 * fwprop.c (propagate_rtx): Likewise.
5749 (forward_propagate_subreg): Likewise.
5750 * ira-conflicts.c (ira_build_conflicts): Likewise.
5751 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
5752 * lra-constraints.c (curr_insn_transform): Likewise.
5753 (split_reg): Likewise.
5754 * lra-eliminations.c (move_plus_up): Likewise.
5755 (lra_eliminate_regs_1): Likewise.
5756 * recog.c (general_operand): Likewise.
5757 * ree.c (combine_reaching_defs): Likewise.
5758 * reload.c (push_reload): Likewise.
5759 (find_reloads): Likewise.
5760 * reload1.c (elimination_effects): Likewise.
5761 (compute_reload_subreg_offset): Likewise.
5762 (choose_reload_regs): Likewise.
5763 * rtlanal.c (subreg_lsb_1): Likewise.
5764 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5765 (simplify_subreg): Likewise.
5766 * var-tracking.c (track_loc_p): Likewise.
5767 * emit-rtl.c (byte_lowpart_offset): Likewise.
5768 (paradoxical_subreg_p): Delete out-of-line definition.
5770 2017-08-22 Jeff Law <law@redhat.com>
5772 PR tree-optimization/81741
5773 PR tree-optimization/71947
5774 * tree-ssa-dom.c: Include tree-inline.h.
5775 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
5776 equivalences if one is more expensive to compute than the other.
5777 * tree-ssa-scopedtables.h (class const_or_copies): Make
5778 record_const_or_copy_raw method private.
5779 (class avail_exprs_stack): New method simplify_binary_operation.
5780 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
5781 avail_exprs_stack::simplify_binary_operation as needed.
5782 (avail_exprs_stack::simplify_binary_operation): New function.
5784 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
5786 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
5787 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
5788 (DOT_SYMBOLS): Likewise.
5789 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
5790 (RELOCATABLE_NEEDS_FIXUP): Likewise.
5791 (RS6000_ABI_NAME): Likewise.
5792 (TARGET_CMODEL): Likewise.
5793 (TOC_SECTION_ASM_OP): Likewise.
5794 (SET_CMODEL): New macro.
5795 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
5797 2017-08-22 Richard Biener <rguenther@suse.de>
5799 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
5800 to allow for free-lang-data replacements similar to verify_type_variant.
5802 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
5803 Alan Hayward <alan.hayward@arm.com>
5804 David Sherwood <david.sherwood@arm.com>
5806 * config/aarch64/aarch64.md (casesi): Use DImode rather than
5807 VOIDmode for the LABEL_REF.
5809 2017-08-22 Richard Biener <rguenther@suse.de>
5811 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
5813 2017-08-22 Richard Biener <rguenther@suse.de>
5815 * common.opt (feliminate-dwarf2-dups): Ignore.
5816 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
5817 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
5818 same_die_p_wrap, compute_section_prefix,
5819 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
5820 (comdat_symbol_id, comdat_symbol_number): Likewise.
5821 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
5823 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
5824 (output_die): Mark unreachable path unreachable.
5825 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
5826 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
5827 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
5828 (dwarf2out_early_finish): Likewise.
5830 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
5832 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
5834 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
5837 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
5838 not VAR_P. Filter attribute warnings with OPT_Wattributes.
5839 (avr_attribute_table) <io, io_low, address>: Initialize
5840 .decl_required with true.
5842 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
5844 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
5845 undocumented debugging options.
5846 (-mvsx-scalar-double): Likewise.
5847 (-mallow-df-permute): Likewise.
5848 (-mvectorize-builtins): Likewise.
5849 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
5850 (rs6000_builtin_vectorized_function): Likewise.
5851 (rs6000_builtin_md_vectorized_function): Likewise.
5852 (rs6000_opt_vars): Likewise.
5854 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
5857 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
5858 (*btrq_imm): Rename from *btrq.
5859 (*btcq_imm): Rename from *btcq.
5860 (btsc): New code attribute.
5861 (*<btsc><mode>): New insn pattern.
5862 (*btr<mode>): Ditto.
5863 (*<btsc><mode>_mask): New insn_and_split pattern.
5864 (*btr<mode>_mask): Ditto.
5866 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5867 Alan Hayward <alan.hayward@arm.com>
5868 David Sherwood <david.sherwood@arm.com>
5870 * function.c (pad_below): Simplify padding calculation.
5872 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5873 Alan Hayward <alan.hayward@arm.com>
5874 David Sherwood <david.sherwood@arm.com>
5876 * target.def (function_prologue): Remove frame size argument.
5877 (function_epilogue): Likewise.
5878 * doc/tm.texi: Regenerate.
5879 * final.c (final_start_function): Update call to function_prologue.
5880 (final_end_function): Update call to function_epilogue.
5881 (default_function_pro_epilogue): Remove frame size argument.
5882 * output.h (default_function_pro_epilogue): Likewise.
5883 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
5884 (arm_output_function_prologue): Likewise.
5885 * config/frv/frv.c (frv_function_prologue): Likewise.
5886 (frv_function_epilogue): Likewise.
5887 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
5888 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
5889 (ia64_output_function_epilogue): Likewise.
5890 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
5891 (m32r_output_function_epilogue): Likewise.
5892 * config/microblaze/microblaze.c (microblaze_function_prologue)
5893 (microblaze_function_epilogue): Likewise.
5894 * config/mips/mips.c (mips_output_function_prologue): Likewise.
5895 (mips_output_function_epilogue): Likewise.
5896 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
5897 (mmix_target_asm_function_epilogue): Likewise.
5898 * config/msp430/msp430.c (msp430_start_function): Likewise.
5899 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
5900 (nds32_asm_function_epilogue): Likewise.
5901 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
5902 * config/pa/pa.c (pa_output_function_prologue): Likewise.
5903 (pa_output_function_epilogue): Likewise.
5904 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
5905 (rs6000_output_function_epilogue): Likewise.
5906 * config/rl78/rl78.c (rl78_start_function): Likewise.
5907 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
5908 (rs6000_output_function_epilogue): Likewise.
5909 * config/rx/rx.c (rx_output_function_prologue): Likewise.
5910 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
5911 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
5912 (sparc_asm_function_epilogue): Likewise.
5914 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5916 * tree.h (type_has_mode_precision_p): New function.
5917 * convert.c (convert_to_integer_1): Use it.
5918 * expr.c (expand_expr_real_2): Likewise.
5919 (expand_expr_real_1): Likewise.
5920 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
5921 * match.pd: Likewise.
5922 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
5923 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
5924 * tree-tailcall.c (process_assignment): Likewise.
5925 * tree-vect-loop.c (vectorizable_reduction): Likewise.
5926 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
5927 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
5928 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
5929 (vectorizable_assignment): Likewise.
5930 (vectorizable_shift): Likewise.
5931 (vectorizable_operation): Likewise.
5932 * tree-vrp.c (register_edge_assert_for_2): Likewise.
5934 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
5936 * match.pd: Add pow (C, x) simplification.
5938 2017-08-21 Richard Biener <rguenther@suse.de>
5940 PR tree-optimization/81900
5941 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
5942 for blocks with abnormal predecessors.
5943 (compute_antic): Do not set visited flag prematurely.
5945 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
5948 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
5950 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5952 * stor-layout.h (vector_type_mode): Move to...
5953 * tree.h (vector_type_mode): ...here.
5954 * stor-layout.c (vector_type_mode): Move to...
5955 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
5957 2017-08-21 Richard Biener <rguenther@suse.de>
5959 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
5960 register_external_die hooks.
5961 (debug_false_tree_charstarstar_uhwistar): Declare.
5962 (debug_nothing_tree_charstar_uhwi): Likewise.
5963 * debug.c (do_nothing_debug_hooks): Adjust.
5964 (debug_false_tree_charstarstar_uhwistar): New do nothing.
5965 (debug_nothing_tree_charstar_uhwi): Likewise.
5966 * dbxout.c (dbx_debug_hooks): Adjust.
5967 (xcoff_debug_hooks): Likewise.
5968 * sdbout.c (sdb_debug_hooks): Likewise.
5969 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
5970 * dwarf2out.c (macinfo_label_base): New global.
5971 (dwarf2out_register_external_die): New function for the
5972 register_external_die hook.
5973 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
5974 (dwarf2_debug_hooks): Use them.
5975 (dwarf2_lineno_debug_hooks): Adjust.
5976 (struct die_struct): Add with_offset flag.
5977 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
5978 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
5979 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
5980 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
5981 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
5982 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
5983 defining section names for the early LTO debug variants.
5984 (reset_indirect_string): New helper.
5985 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
5986 (print_dw_val): Add support for offsetted symbol references.
5987 (get_ultimate_context): Split out from is_cxx.
5988 (is_cxx): Use get_ultimate_context.
5989 (is_fortran): Add decl overload.
5990 (compute_comp_unit_symbol): Split out worker from
5991 compute_section_prefix.
5992 (compute_section_prefix): Call compute_comp_unit_symbol and
5993 set comdat_type_p here.
5994 (output_die): Skip DIE symbol output for the LTO added one.
5995 Handle DIE symbol references with offset.
5996 (output_comp_unit): Guard section name mangling properly.
5997 For LTO debug sections emit a symbol at the section beginning
5998 which we use to refer to its DIEs.
5999 (add_abstract_origin_attribute): For DIEs registered via
6000 dwarf2out_register_external_die directly refer to the early
6001 DIE rather than indirectly through the shadow one we created.
6002 Remove obsolete call to dwarf2out_abstract_function for
6003 non-function/block origins.
6004 (gen_array_type_die): When generating early LTO debug do
6005 not emit DW_AT_string_length.
6006 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
6007 late when in LTO. As suggested place a gcc_unreachable for
6008 the DECL_ABSTRACT_P case.
6009 (gen_subprogram_die): Avoid another specification DIE
6010 for early built declarations/definitions for the late LTO case.
6011 (gen_variable_die): Add type references for late duplicated VLA dies
6013 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
6014 we have the abstract instance already.
6015 (process_scope_var): Adjust decl DIE contexts in LTO which
6016 first puts them in limbo.
6017 (gen_decl_die): Do not generate type DIEs late apart from
6018 types for VLAs or for decls we do not yet have a DIE. Do not
6019 call dwarf2out_abstract_function late.
6020 (dwarf2out_early_global_decl): Make sure to create DIEs
6021 for abstract instances of a decl first.
6022 (dwarf2out_late_global_decl): Adjust comment.
6023 (output_macinfo_op): With multiple macro sections use
6024 macinfo_label_base to distinguish labels.
6025 (output_macinfo): Likewise. Update macinfo_label_base.
6026 Pass in the line info label.
6027 (note_variable_value_in_expr): When generating LTO resolve
6028 all variable values here by generating DIEs as needed.
6029 (init_sections_and_labels): Add early LTO debug flag parameter
6030 and generate different sections and names if set. Add generation
6031 counter for the labels so we can have multiple of them.
6032 (reset_dies): Helper to allow DIEs to be output multiple times.
6033 (dwarf2out_finish): When outputting DIEs to the fat part of an
6034 LTO object first reset DIEs.
6035 (dwarf2out_early_finish): Output early DIEs when generating LTO.
6036 (modified_type_die): Check for decl_ultimate_origin being self
6038 (gen_type_die_with_usage): Likewise.
6039 (gen_typedef_die): Allow decl_ultimate_origin being self.
6040 (set_decl_abstract_flags): Remove.
6041 (set_block_abstract_flags): Likewise.
6042 (dwarf2out_abstract_function): Treat the early generated DIEs
6043 as the abstract copy and only add DW_AT_inline and
6044 DW_AT_artificial here and call set_decl_origin_self.
6045 If the DIE has an abstract origin don't do anything.
6046 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
6047 if we have none yet (Go fails to build one, PR78628).
6048 (variably_modified_type_p): Prevent endless recursion for Ada
6049 cyclic pointer types.
6050 * lto-streamer-in.c: Include debug.h.
6051 (dref_queue): New global.
6052 (lto_read_tree_1): Stream in DIE references.
6053 (lto_input_tree): Register DIE references.
6054 (input_function): Stream DECL_DEBUG_ARGS.
6055 * lto-streamer-out.c: Include debug.h.
6056 (lto_write_tree_1): Output DIE references.
6057 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
6058 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
6059 (output_function): Stream DECL_DEBUG_ARGS.
6060 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
6061 Stream DECL_ABSTRACT_ORIGIN.
6062 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
6063 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
6064 DECL_CONTEXT for file-scope decls.
6065 * lto-streamer.h (struct dref_entry): Declare.
6066 (dref_queue): Likewise.
6067 * cfgexpand.c (pass_expand::execute): Do not call the
6068 outlining_inline_function hook here.
6069 * lto-wrapper.c (debug_obj): New global.
6070 (tool_cleanup): Unlink it if required.
6071 (debug_objcopy): New function.
6072 (run_gcc): Handle early debug sections in the IL files by
6073 extracting them to separate files, partially linkin them and
6074 feeding the result back as result to the linker.
6075 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
6076 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
6077 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
6078 sections into a separate segment.
6079 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
6081 (darwin_asm_dwarf_section): Likewise.
6082 (darwin_asm_output_dwarf_offset): Likewise.
6083 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
6085 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
6086 Alan Hayward <alan.hayward@arm.com>
6087 David Sherwood <david.sherwood@arm.com>
6089 * read-md.h (md_reader::record_potential_iterator_use): Replace
6090 pointer argument with an rtx and an index.
6091 * read-rtl.c (iterator_group::apply_iterator): Likewise.
6092 (apply_mode_iterator): Likewise.
6093 (apply_code_iterator): Likewise.
6094 (apply_int_iterator): Likewise.
6095 (apply_subst_iterator): Likewise.
6096 (record_iterator_use): Likewise.
6097 (record_attribute_use): Likewise.
6098 (md_reader::record_potential_iterator_use): Likewise. Update calls
6099 to record_iterator_use and apply_iterator.
6100 (iterator_use): Replace ptr with x and index.
6101 (attribute_use): Likewise.
6102 (apply_attribute_uses): Update calls to apply_iterator.
6103 (apply_iterators): Likewise. Update initialization of iterator_use.
6104 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
6105 and record_potential_iterator_use.
6106 (rtx_reader::read_rtx_operand): Likewise.
6108 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
6109 Alan Hayward <alan.hayward@arm.com>
6110 David Sherwood <david.sherwood@arm.com>
6112 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
6115 2017-08-21 Richard Biener <rguenther@suse.de>
6118 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
6119 at struct end conservatively when comparing common bases.
6121 2017-08-21 Richard Biener <rguenther@suse.de>
6123 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
6124 (mem_ref_in_stmt): Remove.
6125 (determine_max_movement): Use ref index to get at the reference.
6126 (invariantness_dom_walker::before_dom_children): Deal with
6127 lim data already initialized.
6128 (gather_mem_refs_stmt): Initialize lim data and record ref index.
6130 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
6132 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
6133 (TARGET_ISA_ROUND): Ditto.
6134 (TARGET_ROUND): Ditto.
6135 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
6136 * config/i386/i386.md: Ditto.
6137 * config/i386/sse.md: Ditto.
6138 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
6139 with OPTION_MASK_ISA_SSE4_1.
6141 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
6144 * doc/extend.texi (x86 Built-in Functions): Correct the name of
6145 __builtin_ia32_lzcnt_u16.
6147 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
6150 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
6151 (rs6000_set_current_function): Rewrite function to use it.
6153 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
6156 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
6157 and TYPE_WARN_IF_NOT_ALIGN.
6158 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
6159 (handle_warn_if_not_align): New.
6160 (place_union_field): Call handle_warn_if_not_align.
6161 (place_field): Call handle_warn_if_not_align.
6162 Copy TYPE_WARN_IF_NOT_ALIGN.
6163 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
6164 (layout_type): Likewise.
6165 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
6167 (tree_decl_common): Add warn_if_not_align.
6168 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
6169 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
6170 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
6171 (DECL_WARN_IF_NOT_ALIGN): Likewise.
6172 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
6173 * doc/extend.texi: Document warn_if_not_aligned attribute.
6174 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
6176 2017-08-17 Martin Liska <mliska@suse.cz>
6179 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
6180 (get_data_dependence): Use it as pointer type.
6181 (distribute_loop): Likewise.
6183 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6185 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
6186 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
6187 (p8_vmrgow_<mode>_direct): New define_insn.
6188 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
6189 handle endianness for vmrgew and vmrgow permute patterns.
6191 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
6193 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
6194 * config/rs6000/rs6000-cpus.def: Remove comment.
6195 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
6196 (POWERPC_MASKS): Likewise.
6197 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
6198 use of TARGET_VSX_TIMODE.
6199 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
6200 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
6201 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
6202 (rs6000_option_override_internal): Remove dead code.
6203 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
6204 (rs6000_legitimize_reload_address): Likewise.
6205 (rs6000_legitimate_address_p): Likewise.
6206 (rs6000_opt_masks): Delete "vsx-timode".
6207 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
6208 from function comment.
6209 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
6210 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
6211 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
6213 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
6214 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
6215 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
6218 (define_peephole2): Likewise.
6220 2017-08-17 Martin Sebor <msebor@redhat.com>
6223 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
6224 past the end of an array.
6225 (test_pp_format): Add test cases.
6227 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
6229 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
6230 missing ECF_NOTHROW flags.
6232 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
6235 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
6236 operands residing in integer registers.
6237 (*vsx_le_perm_load_<mode>): Likewise.
6238 (*vsx_le_perm_store_<mode>): Likewise.
6239 (define_peephole2): Add peepholes to optimize the above.
6241 2017-08-17 Marek Polacek <polacek@redhat.com>
6244 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
6245 to mimic what shorten_compare did. Change the return type to bool.
6246 (fold_cond_expr_with_comparison): Update call to
6247 operand_equal_for_comparison_p.
6248 (fold_ternary_loc): Likewise.
6250 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
6252 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
6254 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
6256 2017-08-17 Richard Biener <rguenther@suse.de>
6258 * tree-ssa-structalias.c (solve_graph): When propagating
6259 to successors update the graphs succ edges and avoid duplicate work.
6261 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
6264 * config/i386/i386.c (ix86_option_override_internal): Save target
6265 specific options after ix86_stack_protector_guard_reg was changed.
6267 2017-08-17 Richard Biener <rguenther@suse.de>
6269 PR tree-optimization/81827
6270 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
6271 (new_var_info): Initialize it conservatively.
6272 (get_call_vi): Mark register vars.
6273 (new_scalar_tmp_constraint_exp): Likewise.
6274 (handle_rhs_call): Likewise.
6275 (handle_const_call): Likewise.
6276 (create_function_info_for): Likewise.
6277 (solve_constraints): Sort varinfos to separate register from
6278 non-register vars to pack points-to solution bitmaps during
6281 2017-08-17 Marek Polacek <polacek@redhat.com>
6283 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
6285 2017-08-17 Richard Biener <rguenther@suse.de>
6287 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
6288 to true when overflow is undefined and we saturated the result.
6290 2017-08-17 Alan Modra <amodra@gmail.com>
6293 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
6294 Don't use store multiple if only one reg needs saving.
6295 (interesting_frame_related_regno): New function.
6296 (rs6000_frame_related): Don't emit frame info for regs that
6298 (rs6000_emit_epilogue): Likewise.
6300 2017-08-16 Nathan Sidwell <nathan@acm.org>
6302 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
6303 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
6304 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
6305 * tree.c (free_lang_data_in_type): Use else-if chain. Always
6306 clear TYPE_LANG_1. Remove obsolete member-function stripping.
6307 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
6308 (verify_type): Adjust for TYPE_BINFO move.
6309 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
6310 process TYPE_BINFO directly.
6311 (hash_tree): Likewise.
6312 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
6314 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
6317 2017-08-16 David Malcolm <dmalcolm@redhat.com>
6319 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
6321 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
6324 * config/i386/i386.md (*anddi_1_btr): Change predicates of
6325 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
6326 Add ix86_binary_operator_ok to insn constraint.
6327 (*iordi_1_bts): Ditto.
6328 (*xordi_1_btc): Ditto.
6329 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
6330 Update corresponding peephole2 pattern.
6334 2017-08-16 Bin Cheng <bin.cheng@arm.com>
6336 PR tree-optimization/81832
6337 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
6338 copy loop header which has IFN_LOOP_DIST_ALIAS call.
6340 2017-08-16 Marek Polacek <polacek@redhat.com>
6343 * fold-const.c (fold_indirect_ref_1): Restore original behavior
6344 regarding size_zero_node.
6346 2017-08-16 Martin Liska <mliska@suse.cz>
6349 * config.gcc: Respect previously set extra_objs in case
6352 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
6354 PR tree-optimization/81835
6355 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
6356 the phi SSA_NAME. Check that the condition in a COND_EXPR does
6357 not depend on the phi.
6359 2017-08-16 Alan Modra <amodra@gmail.com>
6361 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
6364 2017-08-16 Alan Modra <amodra@gmail.com>
6366 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
6367 (save_reg_p): ..into this. Update all callers.
6368 (first_reg_to_save): Simplify.
6370 2017-08-16 Alan Modra <amodra@gmail.com>
6372 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
6375 2017-08-15 Joseph Myers <joseph@codesourcery.com>
6379 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
6380 constant count if that count is less than 32.
6382 2017-08-15 Nathan Sidwell <nathan@acm.org>
6384 * gcc.c (execute): Emit friendlier message if inferior is killed
6385 by an external cause.
6387 2017-08-15 Richard Biener <rguenther@suse.de>
6389 PR tree-optimization/81790
6390 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
6391 CONSTRUCTORs from simplifying and VN.
6393 2017-08-14 Martin Sebor <msebor@redhat.com>
6395 * builtin-attrs.def: Add comments.
6397 2017-08-14 Martin Sebor <msebor@redhat.com>
6400 * doc/extend.texi (attribute nonstring): Document new attribute.
6402 2017-08-14 Martin Sebor <msebor@redhat.com>
6405 * tree-diagnostic.c (default_tree_printer): Handle %G.
6406 * gimple-pretty-print.h (percent_G_format): Declare new function.
6407 * gimple-pretty-print.c (percent_G_format): Define.
6408 * tree-pretty-print.c (percent_K_format): Add argument.
6410 2017-08-14 Martin Sebor <msebor@redhat.com>
6412 PR translation/79998
6413 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
6414 Remove a stray space.
6416 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
6419 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
6420 (*iordi_1_bts): Ditto.
6421 (*xordi_1_btc): Ditto.
6423 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6426 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
6427 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6429 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
6430 quoted strings, and make more translator-friendly.
6431 (darwin_rs6000_override_options): Likewise.
6432 (rs6000_option_override_internal): Likewise.
6433 (rs6000_return_in_memory): Fix overlong line.
6434 (init_cmulative_args): Use quoted strings, and make more
6435 translator-friendly.
6436 (rs6000_pass_by_reference): Fix overlong line.
6437 (def_builtin): Use quoted strings.
6438 (altivec_expand_predicate_builtin): Use quoted strings, and make
6439 more translator-friendly.
6440 (htm_expand_builtin): Use quoted strings.
6441 (cpu_expand_builtin): Use quoted strings, and make more
6442 translator-friendly.
6443 (altivec_expand_builtin): Likewise.
6444 (paired_expand_predicate_builtin): Likewise.
6445 (rs6000_invalid_builtin): Likewise.
6446 (builtin_function_type): Use quoted strings.
6447 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
6448 more translator-friendly.
6449 (rs6000_trampoline_init): Likewise.
6450 (rs6000_handle_altivec_attribute): Likewise.
6451 (rs6000_inner_target_options): Use quoted strings.
6452 (rs6000_disable_incompatible_switches): Likewise.
6453 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
6454 strings, and make more translator-friendly.
6455 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
6457 2017-08-14 Bin Cheng <bin.cheng@arm.com>
6459 PR tree-optimization/81799
6460 * tree-loop-distribution.c (version_loop_by_alias_check): Force
6461 cond_expr to simple gimple operand.
6463 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
6466 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
6468 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
6472 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
6473 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
6474 TARGET_GASISR_PROLOGUES.
6475 * config/avr/avr.c (avr_option_override): Same.
6476 (avr_pass_pre_proep::execute): Same.
6478 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
6481 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
6482 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
6484 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
6486 * config/i386/i386.md (*load_tp_<mode>): Redefine as
6487 define_insn_and_split. Split to a memory load from 0 in
6488 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
6489 using PTR mode iterator.
6490 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
6491 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
6492 (*add_tp_<mode>): Redefine as define_insn_and_split.
6493 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
6494 address space. Merge with *add_tp_x32 using PTR mode iterator.
6495 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
6496 Split to an add with a memory load from 0 in
6497 DEFAULT_TLS_SEG_REG address space.
6499 2017-08-12 Andrew Pinski <apinski@cavium.com>
6501 * config/aarch64/aarch64-option-extensions.def (rdma):
6502 Fix feature string to what Linux prints out in /proc/cpuinfo.
6504 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
6507 * dwarf2out.c (modified_type_die): For C typedef types that have
6508 an ultimate origin, process the ultimate origin instead of the
6510 (gen_typedef_die): Assert that input DECLs have no ultimate
6512 (gen_type_die_with_usage): For typedef variants that have an
6513 ultimate origin, just call gen_decl_die on the original DECL.
6514 (process_scope_var): Avoid creating DIEs for local typedefs and
6515 concrete static variables.
6517 2017-08-12 Alan Modra <amodra@gmail.com>
6521 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
6522 match gnu-user.h startfile.
6523 (ENDFILE_LINUX_SPEC): Similarly.
6525 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
6528 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
6530 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
6532 2017-08-11 Tamar Christina <tamar.christina@arm.com>
6533 * config/aarch64/aarch64.md (mov<mode>): Change.
6534 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
6535 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
6536 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
6538 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
6540 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
6541 for storage order barriers.
6543 2017-08-11 Martin Liska <mliska@suse.cz>
6546 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
6547 variables of void type.
6549 2017-08-11 Martin Liska <mliska@suse.cz>
6551 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
6552 TARGET_SUPPORTS_ALIASES.
6553 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
6554 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
6555 (optimize_weakref): Likewise.
6556 * symtab.c (symtab_node::noninterposable_alias): Likewise.
6557 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
6558 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
6560 2017-08-11 Martin Liska <mliska@suse.cz>
6563 * config/i386/i386.c (make_resolver_func): Do complete
6564 refactoring of the function.
6566 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
6569 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
6570 * config/i386/i386.c (ix86_stack_protect_guard): Use
6571 ix86_stack_protect_guard_symbol_str to generate varible declaration.
6572 * doc/invoke.texi (x86 Options): Document
6573 -mstack-protector-guard-symbol= option.
6575 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
6577 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
6578 * config/i386/i386.c (ix86_split_stack_guard): New function.
6579 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
6580 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
6581 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
6582 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
6583 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
6584 (split_stack_space_check): Call ix86_split_stack_guard.
6586 2017-08-10 Martin Sebor <msebor@redhat.com>
6588 * print-tree.c (print_node): Print location using the established
6590 Replace spaces with colons.
6591 (debug_raw, debug): Ditto.
6593 2017-08-10 Martin Sebor <msebor@redhat.com>
6596 * pretty-print.c (pp_format): Correct the handling of %s precision.
6598 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
6601 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
6603 (ix86_finalize_stack_frame_flags): This. Also clear
6604 frame_pointer_needed if -fno-omit-frame-pointer is used without
6606 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
6607 with ix86_finalize_stack_frame_flags.
6608 (ix86_expand_epilogue): Likewise.
6609 (ix86_expand_split_stack_prologue): Likewise.
6610 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
6612 2017-08-10 Martin Liska <mliska@suse.cz>
6615 * c-attribs.c (handle_target_attribute):
6616 Report warning for an empty string argument of target attribute.
6618 2017-08-09 Jakub Jelinek <jakub@redhat.com>
6621 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
6623 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
6624 or DECL_NONLOCAL labels.
6625 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
6626 or DECL_NONLOCAL labels here.
6628 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
6630 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
6631 to indicate when early gimple folding has been disabled.
6632 (rs6000_gimple_fold_builtin): Add debug content.
6633 (rs6000_invalid_builtin): Fix whitespace.
6634 (rs6000_expand_builtin): Fix whitespace.
6635 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
6637 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
6640 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
6641 SAVE_MULTIPLE if not all the registers that saves, should be saved.
6643 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
6645 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
6646 (qdf24xx): Likewise.
6647 * config/aarch64/aarch64.md: Include falkor.md.
6648 * config/aarch64/falkor.md: New.
6650 2017-08-09 Marek Polacek <polacek@redhat.com>
6653 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
6654 * diagnostic.c (emit_diagnostic): Add a comment.
6655 (emit_diagnostic_valist): New function.
6657 2017-08-09 Marek Polacek <polacek@redhat.com>
6660 * input.c (make_location): New overload.
6661 * input.h (make_location): Declare.
6663 2017-08-08 Alan Modra <amodra@gmail.com>
6664 H.J. Lu <hongjiu.lu@intel.com>
6667 * gcc.c (NO_PIE_SPEC): Delete.
6668 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
6670 (LINK_PIE_SPEC): ..to here.
6671 (LINK_COMMAND_SPEC): Support -no-pie.
6672 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
6673 chain of crtbegin*.o selection, update for PIE_SPEC changes and
6675 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
6676 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
6677 (ENDFILE_CRTEND_SPEC): Similarly.
6679 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
6682 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
6683 (mstack-protector-guard-offset=): Ditto.
6684 * config/i386/i386.c (ix86_option_override): Handle
6685 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
6687 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
6688 ix86_stack_protect_guard_offset variables.
6689 (TARGET_STACK_PROTECT_GUARD): Always define.
6690 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
6691 and -mstack-protector-guard-offset= options.
6693 2017-08-08 Bin Cheng <bin.cheng@arm.com>
6695 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
6696 boundary case for the last candidate.
6698 2017-08-08 Bin Cheng <bin.cheng@arm.com>
6700 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
6701 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
6703 2017-08-08 Tamar Christina <tamar.christina@arm.com>
6706 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
6707 * config/aarch64/aarch64-builtins.c
6708 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
6709 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
6710 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
6712 2017-08-08 Tamar Christina <tamar.christina@arm.com>
6713 Andrew Pinski <pinskia@gmail.com>
6716 * internal-fn.def (XORSIGN): New.
6717 * optabs.def (xorsign_optab): New.
6718 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
6719 (convert_expand_mult_copysign): New.
6720 (pass_optimize_widening_mul::execute): Call
6721 convert_expand_mult_copysign.
6723 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6725 PR tree-optimization/81354
6726 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
6727 Insert on edges rather than explicitly creating landing pads.
6728 (analyze_candidates_and_replace): Commit edge inserts.
6730 2017-08-08 Richard Biener <rguenther@suse.de>
6733 * tree-ssa-loop-niter.c: Include tree-dfa.h.
6734 (expand_simple_operations): Also look through ADDR_EXPRs with
6735 MEM_REF bases treating them as POINTER_PLUS_EXPR.
6737 2017-08-08 Richard Biener <rguenther@suse.de>
6739 PR tree-optimization/81723
6740 * tree-vect-slp.c (struct bst_traits): New hash traits.
6741 (bst_fail): New global.
6742 (vect_build_slp_tree_2): New worker, split out from ...
6743 (vect_build_slp_tree): ... this now wrapping it with using
6744 bst_fail set to cache SLP tree build fails. Properly handle
6746 (vect_analyze_slp_instance): Allocate and free bst_fail.
6748 2017-08-08 Martin Liska <mliska@suse.cz>
6751 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
6752 LABEL_DECLs that can be from a different function.
6754 2017-08-08 Bin Cheng <bin.cheng@arm.com>
6756 PR tree-optimization/81744
6757 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
6758 loop's number of iterations.
6760 2017-08-08 Martin Liska <mliska@suse.cz>
6762 * asan.c: Include header files.
6763 * attribs.c (build_decl_attribute_variant): New function moved
6765 (build_type_attribute_qual_variant): Likewise.
6766 (cmp_attrib_identifiers): Likewise.
6767 (simple_cst_list_equal): Likewise.
6768 (omp_declare_simd_clauses_equal): Likewise.
6769 (attribute_value_equal): Likewise.
6770 (comp_type_attributes): Likewise.
6771 (build_type_attribute_variant): Likewise.
6772 (lookup_ident_attribute): Likewise.
6773 (remove_attribute): Likewise.
6774 (merge_attributes): Likewise.
6775 (merge_type_attributes): Likewise.
6776 (merge_decl_attributes): Likewise.
6777 (merge_dllimport_decl_attributes): Likewise.
6778 (handle_dll_attribute): Likewise.
6779 (attribute_list_equal): Likewise.
6780 (attribute_list_contained): Likewise.
6781 * attribs.h (lookup_attribute): New function moved from tree.[ch].
6782 (lookup_attribute_by_prefix): Likewise.
6783 * bb-reorder.c: Include header files.
6784 * builtins.c: Likewise.
6785 * calls.c: Likewise.
6786 * cfgexpand.c: Likewise.
6787 * cgraph.c: Likewise.
6788 * cgraphunit.c: Likewise.
6789 * convert.c: Likewise.
6790 * dwarf2out.c: Likewise.
6791 * final.c: Likewise.
6792 * fold-const.c: Likewise.
6793 * function.c: Likewise.
6794 * gimple-expr.c: Likewise.
6795 * gimple-fold.c: Likewise.
6796 * gimple-pretty-print.c: Likewise.
6797 * gimple.c: Likewise.
6798 * gimplify.c: Likewise.
6799 * hsa-common.c: Likewise.
6800 * hsa-gen.c: Likewise.
6801 * internal-fn.c: Likewise.
6802 * ipa-chkp.c: Likewise.
6803 * ipa-cp.c: Likewise.
6804 * ipa-devirt.c: Likewise.
6805 * ipa-fnsummary.c: Likewise.
6806 * ipa-inline.c: Likewise.
6807 * ipa-visibility.c: Likewise.
6809 * lto-cgraph.c: Likewise.
6810 * omp-expand.c: Likewise.
6811 * omp-general.c: Likewise.
6812 * omp-low.c: Likewise.
6813 * omp-offload.c: Likewise.
6814 * omp-simd-clone.c: Likewise.
6815 * opts-global.c: Likewise.
6816 * passes.c: Likewise.
6817 * predict.c: Likewise.
6818 * sancov.c: Likewise.
6819 * sanopt.c: Likewise.
6820 * symtab.c: Likewise.
6821 * toplev.c: Likewise.
6822 * trans-mem.c: Likewise.
6823 * tree-chkp.c: Likewise.
6824 * tree-eh.c: Likewise.
6825 * tree-into-ssa.c: Likewise.
6826 * tree-object-size.c: Likewise.
6827 * tree-parloops.c: Likewise.
6828 * tree-profile.c: Likewise.
6829 * tree-ssa-ccp.c: Likewise.
6830 * tree-ssa-live.c: Likewise.
6831 * tree-ssa-loop.c: Likewise.
6832 * tree-ssa-sccvn.c: Likewise.
6833 * tree-ssa-structalias.c: Likewise.
6834 * tree-ssa.c: Likewise.
6835 * tree-streamer-in.c: Likewise.
6836 * tree-vectorizer.c: Likewise.
6837 * tree-vrp.c: Likewise.
6839 * ubsan.c: Likewise.
6840 * varasm.c: Likewise.
6841 * varpool.c: Likewise.
6842 * tree.c: Remove functions moved to attribs.[ch].
6844 * config/aarch64/aarch64.c: Add attrs.h header file.
6845 * config/alpha/alpha.c: Likewise.
6846 * config/arc/arc.c: Likewise.
6847 * config/arm/arm.c: Likewise.
6848 * config/avr/avr.c: Likewise.
6849 * config/bfin/bfin.c: Likewise.
6850 * config/c6x/c6x.c: Likewise.
6851 * config/cr16/cr16.c: Likewise.
6852 * config/cris/cris.c: Likewise.
6853 * config/darwin.c: Likewise.
6854 * config/epiphany/epiphany.c: Likewise.
6855 * config/fr30/fr30.c: Likewise.
6856 * config/frv/frv.c: Likewise.
6857 * config/ft32/ft32.c: Likewise.
6858 * config/h8300/h8300.c: Likewise.
6859 * config/i386/winnt.c: Likewise.
6860 * config/ia64/ia64.c: Likewise.
6861 * config/iq2000/iq2000.c: Likewise.
6862 * config/lm32/lm32.c: Likewise.
6863 * config/m32c/m32c.c: Likewise.
6864 * config/m32r/m32r.c: Likewise.
6865 * config/m68k/m68k.c: Likewise.
6866 * config/mcore/mcore.c: Likewise.
6867 * config/microblaze/microblaze.c: Likewise.
6868 * config/mips/mips.c: Likewise.
6869 * config/mmix/mmix.c: Likewise.
6870 * config/mn10300/mn10300.c: Likewise.
6871 * config/moxie/moxie.c: Likewise.
6872 * config/msp430/msp430.c: Likewise.
6873 * config/nds32/nds32-isr.c: Likewise.
6874 * config/nds32/nds32.c: Likewise.
6875 * config/nios2/nios2.c: Likewise.
6876 * config/nvptx/nvptx.c: Likewise.
6877 * config/pa/pa.c: Likewise.
6878 * config/pdp11/pdp11.c: Likewise.
6879 * config/powerpcspe/powerpcspe.c: Likewise.
6880 * config/riscv/riscv.c: Likewise.
6881 * config/rl78/rl78.c: Likewise.
6882 * config/rx/rx.c: Likewise.
6883 * config/s390/s390.c: Likewise.
6884 * config/sh/sh.c: Likewise.
6885 * config/sol2.c: Likewise.
6886 * config/sparc/sparc.c: Likewise.
6887 * config/spu/spu.c: Likewise.
6888 * config/stormy16/stormy16.c: Likewise.
6889 * config/tilegx/tilegx.c: Likewise.
6890 * config/tilepro/tilepro.c: Likewise.
6891 * config/v850/v850.c: Likewise.
6892 * config/vax/vax.c: Likewise.
6893 * config/visium/visium.c: Likewise.
6894 * config/xtensa/xtensa.c: Likewise.
6896 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
6899 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
6900 constraints since the -mupper-regs-* switches have been
6902 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
6903 into a vector from a double word element that was extracted from
6904 another vector, and eliminate extra XXPERMDI instructions.
6905 (vsx_concat_<mode>_2): Likewise.
6906 (vsx_concat_<mode>_3): Likewise.
6907 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
6908 concat to allow optimizing inserts from previous extracts.
6910 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
6912 * config/i386/i386.c (ix86_stack_protect_guard): Generate
6913 memory reference to a SSP offset in TLS address space.
6914 (ix86_print_operand) <case '@'>: Remove.
6915 (ix86_print_operand_punct_valid_p): Remove '@' code.
6916 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
6918 (stack_tls_protect_set_<mode>): Remove.
6919 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
6920 (stack_tls_protect_test_<mode>): Remove.
6921 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
6923 2017-08-07 Olivier Hainque <hainque@adacore.com>
6926 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
6928 2017-08-07 Douglas Rupp <rupp@adacore.com>
6930 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
6931 variable was referenced as multidir in command.
6933 2017-08-07 Jakub Jelinek <jakub@redhat.com>
6936 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
6939 2017-08-07 Martin Liska <mliska@suse.cz>
6941 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
6942 * config/rl78/rl78.c: Add include of attribs.h.
6943 * config/sh/sh.c: Likewise.
6944 * config/v850/v850.c: Likewise.
6946 2017-08-07 Tom de Vries <tom@codesourcery.com>
6949 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
6951 2017-08-07 Martin Liska <mliska@suse.cz>
6953 * config/mips/mips.c: Include attribs.h.
6955 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
6958 * doc/invoke.texi: Document change in behvaior for -Ofast for
6961 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
6963 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
6965 (aarch64_pop_regs): Likewise.
6966 (aarch64_gen_load_pair): Likewise.
6967 (aarch64_save_callee_saves): Likewise.
6968 (aarch64_restore_callee_saves): Likewise.
6970 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
6972 * config/i386/i386.c: Revert the last change.
6974 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
6977 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
6979 (ix86_finalize_stack_frame_flags): This. Also clear
6980 frame_pointer_needed if -fno-omit-frame-pointer is used without
6982 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
6983 with ix86_finalize_stack_frame_flags.
6984 (ix86_expand_epilogue): Likewise.
6985 (ix86_expand_split_stack_prologue): Likewise.
6987 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
6990 * config/i386/i386.c (get_builtin_code_for_version): Set priority
6991 to P_AES for Westmere.
6993 2017-08-07 Jonathan Yong <10walls@gmail.com>
6995 * config/i386/mingw.opt (fset-stack-executable): Removed.
6996 * config/i386/cygming.opt (fset-stack-executable): Moved
6998 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
7000 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
7002 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
7004 2017-08-07 Marek Polacek <polacek@redhat.com>
7007 * fold-const.c (fold_indirect_ref_1): Check type_domain.
7009 2017-08-07 Martin Liska <mliska@suse.cz>
7011 * attribs.h (canonicalize_attr_name): New function.
7012 (cmp_attribs): Move from c-format.c and adjusted.
7013 (is_attribute_p): Moved from tree.h.
7014 * tree-inline.c: Add new includes.
7015 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
7016 (private_is_attribute_p): Remove.
7017 (private_lookup_attribute): Likewise.
7018 (private_lookup_attribute_by_prefix): Simplify.
7019 (remove_attribute): Use is_attribute_p.
7020 * tree.h: Remove removed declarations.
7022 2017-08-07 Jakub Jelinek <jakub@redhat.com>
7025 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
7026 instead of computing it in the function. Formatting fix.
7027 (expand_case): Don't rely on default_edge being the first edge,
7028 clear it if removing it, pass default_edge to
7029 emit_case_dispatch_table.
7030 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
7033 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
7035 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
7036 insn in the function, emit NOP after the insn.
7038 2017-08-06 Tom de Vries <tom@codesourcery.com>
7040 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
7043 2017-08-06 Tom de Vries <tom@codesourcery.com>
7045 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
7048 2017-08-04 Yury Gribov <tetra2005@gmail.com>
7050 PR tree-optimization/57371
7051 * match.pd: New pattern.
7053 2017-08-04 Marek Polacek <polacek@redhat.com>
7056 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
7057 perform the computation in offset_int.
7059 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
7061 PR tree-optimization/81136
7062 * tree-vectorizer.h: Include tree-hash-traits.h.
7063 (vec_base_alignments): New typedef.
7064 (vec_info): Add a base_alignments field.
7065 (vect_record_base_alignments): Declare.
7066 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
7068 (DR_IS_CONDITIONAL_IN_STMT): New macro.
7069 (create_data_ref): Add an is_conditional_in_stmt argument.
7070 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
7071 the is_conditional_in_stmt field.
7072 (data_ref_loc): Add an is_conditional_in_stmt field.
7073 (get_references_in_stmt): Set the is_conditional_in_stmt field.
7074 (find_data_references_in_stmt): Update call to create_data_ref.
7075 (graphite_find_data_references_in_stmt): Likewise.
7076 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
7077 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
7078 (vect_record_base_alignment): New function.
7079 (vect_record_base_alignments): Likewise.
7080 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
7081 for nested statements even if we fail to compute a misalignment.
7082 Use pooled base alignments for unconditional references.
7083 (vect_find_same_alignment_drs): Compare base addresses instead
7085 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
7086 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
7088 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
7090 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
7091 Add an explicit name for the enum. Use auto_vec for slp_instances
7093 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
7095 (_bb_vec_info): Add a constructor and destructor.
7096 (vinfo_for_stmt): Return NULL for uids of -1 as well.
7097 (destroy_loop_vec_info): Delete.
7098 (vect_destroy_datarefs): Likewise.
7099 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
7100 (vec_info::vec_info): New function.
7101 (vec_info::~vec_info): Likewise.
7102 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
7103 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
7104 destroy_loop_vec_info.
7105 * tree-vect-loop.c (new_loop_vec_info): Replace with...
7106 (_loop_vec_info::_loop_vec_info): ...this.
7107 (destroy_loop_vec_info): Replace with...
7108 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
7109 the stmt_vec_infos. Leave handling of vec_info information to its
7110 destructor. Remove explicit vector releases.
7111 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
7112 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
7113 * tree-vect-slp.c (new_bb_vec_info): Replace with...
7114 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
7115 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
7116 (destroy_bb_vec_info): Replace with...
7117 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
7118 information to its destructor.
7119 (vect_slp_analyze_bb_1): Use new and delete instead of
7120 new_bb_vec_info and destroy_bb_vec_info.
7121 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
7124 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
7126 * tree-data-ref.h (subscript): Add access_fn field.
7127 (data_dependence_relation): Add could_be_independent_p.
7128 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
7129 (same_access_functions): Move to tree-data-ref.c.
7130 * tree-data-ref.c (ref_contains_union_access_p): New function.
7131 (access_fn_component_p): Likewise.
7132 (access_fn_components_comparable_p): Likewise.
7133 (dr_analyze_indices): Add a reference to access_fn_component_p.
7134 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
7136 (constant_access_functions): Likewise.
7137 (add_other_self_distances): Likewise.
7138 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
7139 (initialize_data_dependence_relation): Use XCNEW and remove
7140 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
7141 of access functions that have the same type. Allow the
7142 subsequence to end with different bases in some circumstances.
7143 Record the chosen access functions in SUB_ACCESS_FN.
7144 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
7145 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
7146 (subscript_dependence_tester_1): Likewise dra and drb.
7147 (build_classic_dist_vector): Update calls accordingly.
7148 (subscript_dependence_tester): Likewise.
7149 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
7150 DDR_COULD_BE_INDEPENDENT_P.
7151 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
7152 comp_alias_ddrs instead of may_alias_ddrs.
7153 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
7155 (vect_analyze_data_ref_dependence): Use it if
7156 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
7157 distance vectors if that fails.
7158 (dependence_distance_ge_vf): New function.
7159 (vect_prune_runtime_alias_test_list): Use it. Don't clear
7160 LOOP_VINFO_MAY_ALIAS_DDRS.
7162 2017-08-04 Richard Biener <rguenther@suse.de>
7165 * fold-const.c (fold_binary_loc): Properly restrict
7166 minus_var0 && minus_var1 case when associating undefined overflow
7169 2017-08-04 Tom de Vries <tom@codesourcery.com>
7171 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
7173 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7175 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7176 Don't start diagnostic messages with a capital letter.
7177 * config/rs6000/rs6000.c (rs6000_option_override_internal):
7179 (rs6000_invalid_builtin): Likewise.
7180 (rs6000_trampoline_init): Likewise.
7182 2017-08-03 Jakub Jelinek <jakub@redhat.com>
7185 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
7186 after setting changeable df flags.
7188 2017-08-03 Richard Biener <rguenther@suse.de>
7190 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
7191 up if the use is in USE - X.
7193 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
7195 * toplev.c (dumpfile.h): New include.
7196 (internal_error_reentered): New static function. Use it...
7197 (internal_error_function): ...here to handle reentered internal_error.
7199 2017-08-03 Richard Biener <rguenther@suse.de>
7202 * fold-const.c (split_tree): Add minus_var and minus_con
7203 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
7204 here but always use minus_*.
7205 (associate_trees): Assert we never associate with MINUS_EXPR
7206 and NULL first operand. Do not recurse for PLUS_EXPR operands
7207 when associating as MINUS_EXPR either.
7208 (fold_binary_loc): Track minus_var and minus_con.
7210 2017-08-03 Tom de Vries <tom@codesourcery.com>
7213 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
7214 ACCEL_COMPILER, apply finish_options on
7215 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
7217 2017-08-03 Tom de Vries <tom@codesourcery.com>
7220 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
7221 function_entry_patch_area_size > 0.
7223 2017-08-03 Jakub Jelinek <jakub@redhat.com>
7226 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
7227 instead of 10??LU, perform unit multiplication in wide_int,
7228 don't change alloc_object_size_limit if the limit is larger
7231 PR tree-optimization/81655
7232 PR tree-optimization/81588
7233 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
7234 the case when ranges[i].low and high are 1 for unsigned type with
7238 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
7239 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
7241 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
7243 * tree-vrp.h: Add include guard.
7245 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
7248 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
7249 (ud2): New insn pattern.
7250 * config/i386/i386.c (ix86_expand_epilogue):
7251 For naked functions, generate ud2 instead of trap insn.
7253 2017-08-02 Marek Polacek <polacek@redhat.com>
7256 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
7258 2017-08-02 Tom de Vries <tom@codesourcery.com>
7259 Cesar Philippidis <cesar@codesourcery.com>
7261 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
7262 Add missing edge probabilities.
7264 2017-08-02 Tamar Christina <tamar.christina@arm.com>
7266 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
7269 2017-08-02 Jakub Jelinek <jakub@redhat.com>
7272 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
7273 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
7274 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
7275 prologue_seq sequences - if any.
7277 2017-08-02 Richard Biener <rguenther@suse.de>
7279 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
7280 via vectors if supported, integer extracts via punning if supported
7281 or otherwise vector extracts.
7283 2017-08-02 Richard Biener <rguenther@suse.de>
7285 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
7287 (bitmap_insert_into_set): ... this.
7289 2017-08-02 Richard Biener <rguenther@suse.de>
7291 PR tree-optimization/81633
7293 2015-08-17 Alan Hayward <alan.hayward@arm.com>
7295 PR tree-optimization/71752
7296 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
7298 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
7300 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
7301 (machine_function::call_ms2sysv_pad_out): Remove field.
7302 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
7303 (ix86_compute_frame_layout): Likewise.
7305 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
7308 * config/i386/i386.c (ix86_set_func_type): Disallow naked
7309 attribute with interrupt attribute.
7311 2017-08-01 Andrew Pinski <apinski@cavium.com>
7313 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
7314 BIT_INSERT_EXPR's operand 1
7315 to see if the types precision matches.
7317 2017-08-01 Martin Liska <mliska@suse.cz>
7320 * builtins.c (expand_builtin_memcpy_args): Remove.
7321 (expand_builtin_memcpy): Call newly added function
7322 expand_builtin_memory_copy_args.
7323 (expand_builtin_memcpy_with_bounds): Likewise.
7324 (expand_builtin_mempcpy): Remove last argument.
7325 (expand_builtin_mempcpy_with_bounds): Likewise.
7326 (expand_builtin_memory_copy_args): New function created from
7327 expand_builtin_mempcpy_args with small modifications.
7328 (expand_builtin_mempcpy_args): Remove.
7329 (expand_builtin_stpcpy): Remove unused argument.
7330 (expand_builtin): Likewise.
7331 (expand_builtin_with_bounds): Likewise.
7333 2017-08-01 Martin Liska <mliska@suse.cz>
7336 Make mempcpy more optimal (PR middle-end/70140).
7338 2017-08-01 Jakub Jelinek <jakub@redhat.com>
7341 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
7342 __builtin_vec_cmpne verify both arguments are compatible vectors
7343 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
7344 verify arg1_type is a pointer or array type. For __builtin_vec_st,
7345 move computation of aligned to after checking the argument types.
7349 * config/rs6000/vsx.md (vextract_fp_from_shorth,
7350 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
7353 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
7354 Doug Rupp <rupp@adacore.com>
7355 Olivier Hainque <hainque@adacore.com>
7357 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
7358 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
7360 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
7361 for TARGET_OS_CPP_BUILTIN.
7362 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
7363 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
7365 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
7366 passing required abi options to the assembler for EABI configurations.
7367 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
7368 of .text.hot and .text.unlikely sections for kernel modules when
7369 using ARM style exceptions.
7370 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
7371 options. Add EXTRA_CC1_SPEC.
7372 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
7374 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
7376 (ARM_TARGET2_DWARF_FORMAT): Define.
7377 * config/arm/t-vxworks: Adjust multilib control to removal of the
7378 Diab command line options.
7380 2017-08-01 Martin Liska <mliska@suse.cz>
7382 PR gcov-profile/81561
7383 * gcov.c (unblock): Make unblocking safe as we need to preserve
7384 index correspondence of blocks and block_lists.
7386 2017-08-01 Richard Biener <rguenther@suse.de>
7388 PR tree-optimization/81181
7389 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
7390 (compute_antic): ... end of iteration here.
7392 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
7394 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
7395 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
7396 (ftree-slp-vectorize): Likewise.
7397 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
7398 can no longer be set independent of flag_tree_loop_vectorize.
7399 * omp-general.c (emp_max_vf): Likewise.
7400 * opts.c (enable_fdo_optimizations): Remove references to
7401 flag_tree_vectorize, these are now implicit.
7402 (common_handle_option): Remove handling for OPT_ftree_vectorize,
7403 and leave it for the options machinery.
7405 2017-08-01 Martin Liska <mliska@suse.cz>
7408 * builtins.c (expand_builtin_memcpy_args): Remove.
7409 (expand_builtin_memcpy): Call newly added function
7410 expand_builtin_memory_copy_args.
7411 (expand_builtin_memcpy_with_bounds): Likewise.
7412 (expand_builtin_mempcpy): Remove last argument.
7413 (expand_builtin_mempcpy_with_bounds): Likewise.
7414 (expand_builtin_memory_copy_args): New function created from
7415 expand_builtin_mempcpy_args with small modifications.
7416 (expand_builtin_mempcpy_args): Remove.
7417 (expand_builtin_stpcpy): Remove unused argument.
7418 (expand_builtin): Likewise.
7419 (expand_builtin_with_bounds): Likewise.
7421 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
7424 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
7425 print "ds:" only for immediates in generic address space.
7427 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
7430 * config/i386/i386.c (ix86_funciton_naked): New prototype.
7431 (ix86_function_ok_for_sibcall): Return false for naked functions.
7433 2017-08-01 Richard Biener <rguenther@suse.de>
7435 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
7436 (compute_antic): Seed worklist with exit block predecessors.
7437 * cfganal.c (dfs_find_deadend): For a cycle return the source
7438 of the edge closing it.
7440 2017-08-01 Tamar Christina <tamar.christina@arm.com>
7442 * config/aarch64/aarch64.c
7443 (aarch64_can_const_movi_rtx_p): Move 0 check.
7445 2017-08-01 Bin Cheng <bin.cheng@arm.com>
7447 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
7448 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
7450 * match.pd: Ditto in address comparison pattern.
7452 2017-08-01 Bin Cheng <bin.cheng@arm.com>
7454 PR tree-optimization/81627
7455 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
7456 closed ssa form for store-store chain.
7458 2017-08-01 Bin Cheng <bin.cheng@arm.com>
7460 PR tree-optimization/81620
7461 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
7462 for store-store chain.
7464 2017-08-01 Jakub Jelinek <jakub@redhat.com>
7466 PR tree-optimization/81588
7467 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
7468 ranges[i].in_p, invert comparison code ccode. For >/>=,
7469 swap rhs1 and rhs2 and comparison code unconditionally,
7470 for </<= don't do that. Don't swap rhs1/rhs2 again if
7471 ranges[i].in_p, instead invert comparison code ccode if
7472 opcode or oe->rank is BIT_IOR_EXPR.
7475 * optabs.def (vec_extract_optab, vec_init_optab): Change from
7476 a direct optab to conversion optab.
7477 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
7478 with GET_MODE_INNER as last argument instead of optab_handler.
7479 * expmed.c (extract_bit_field_1): Likewise. Use vector from
7480 vector extraction if possible and optab is available.
7481 * expr.c (store_constructor): Use convert_optab_handler instead
7482 of optab_handler. Use vector initialization from smaller
7483 vectors if possible and optab is available.
7484 * tree-vect-stmts.c (vectorizable_load): Likewise.
7485 * doc/md.texi (vec_extract, vec_init): Document that the optabs
7487 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
7488 of vec_init from half-sized vectors with the same element mode.
7489 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
7490 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
7491 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
7492 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
7493 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
7494 after mode in gen_vec_extract* calls.
7495 (vec_extract<mode>): Renamed to ...
7496 (vec_extract<mode><ssescalarmodelower>): ... this.
7497 (vec_extract<mode><ssehalfvecmodelower>): New expander.
7498 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
7499 element mode after mode in gen_vec_init* calls.
7500 (VEC_INIT_HALF_MODE): New mode iterator.
7501 (vec_init<mode>): Renamed to ...
7502 (vec_init<mode><ssescalarmodelower>): ... this.
7503 (vec_init<mode><ssehalfvecmodelower>): New expander.
7504 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
7505 (vec_extractv2sfsf): ... this.
7506 (vec_initv2sf): Renamed to ...
7507 (vec_initv2sfsf): ... this.
7508 (vec_extractv2si): Renamed to ...
7509 (vec_extractv2sisi): ... this.
7510 (vec_initv2si): Renamed to ...
7511 (vec_initv2sisi): ... this.
7512 (vec_extractv4hi): Renamed to ...
7513 (vec_extractv4hihi): ... this.
7514 (vec_initv4hi): Renamed to ...
7515 (vec_initv4hihi): ... this.
7516 (vec_extractv8qi): Renamed to ...
7517 (vec_extractv8qiqi): ... this.
7518 (vec_initv8qi): Renamed to ...
7519 (vec_initv8qiqi): ... this.
7520 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
7521 (vec_init<mode>): Renamed to ...
7522 (vec_init<mode><VEC_base_l>): ... this.
7523 (vec_extract<mode>): Renamed to ...
7524 (vec_extract<mode><VEC_base_l>): ... this.
7525 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
7526 (vec_initv2sfsf): ... this.
7527 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
7528 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
7529 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
7530 element mode after mode in gen_vec_init* calls.
7531 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
7532 (vec_init<mode><Vel>): ... this.
7533 (vec_extract<mode>): Renamed to ...
7534 (vec_extract<mode><Vel>): ... this.
7535 * config/aarch64/iterators.md (Vel): New mode attribute.
7536 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
7537 Add element mode after mode in gen_vec_extract* calls.
7538 * config/s390/vector.md (non_vec_l): New mode attribute.
7539 (vec_extract<mode>): Renamed to ...
7540 (vec_extract<mode><non_vec_l>): ... this.
7541 (vec_init<mode>): Renamed to ...
7542 (vec_init<mode><non_vec_l>): ... this.
7543 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
7544 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
7546 * config/arm/iterators.md (V_elem_l): New mode attribute.
7547 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
7548 (vec_extract<mode><V_elem_l>): ... this.
7549 (vec_extractv2di): Renamed to ...
7550 (vec_extractv2didi): ... this.
7551 (vec_init<mode>): Renamed to ...
7552 (vec_init<mode><V_elem_l>): ... this.
7553 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
7554 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
7555 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
7556 Add element mode after gen_vec_extract* calls.
7557 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
7558 (vec_init<mode><unitmode>): ... this.
7559 (vec_extract<mode>): Renamed to ...
7560 (vec_extract<mode><unitmode>): ... this.
7561 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
7562 (vec_init<mode><unitmode>): ... this.
7563 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
7564 (vec_initv2sfsf): ... this.
7565 (vec_extractv2sf): Renamed to ...
7566 (vec_extractv2sfsf): ... this.
7567 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
7568 Add element mode after gen_vec_extract* calls.
7569 * config/mips/mips.md (unitmode): New mode iterator.
7570 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
7571 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
7572 * config/spu/spu.md (inner_l): New mode attribute.
7573 (vec_init<mode>): Renamed to ...
7574 (vec_init<mode><inner_l>): ... this.
7575 (vec_extract<mode>): Renamed to ...
7576 (vec_extract<mode><inner_l>): ... this.
7577 * config/sparc/sparc.md (veltmode): New mode iterator.
7578 (vec_init<VMALL:mode>): Renamed to ...
7579 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
7580 * config/ia64/vect.md (vec_initv2si): Renamed to ...
7581 (vec_initv2sisi): ... this.
7582 (vec_initv2sf): Renamed to ...
7583 (vec_initv2sfsf): ... this.
7584 (vec_extractv2sf): Renamed to ...
7585 (vec_extractv2sfsf): ... this.
7586 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
7587 (vec_init<mode>): Renamed to ...
7588 (vec_init<mode><VEC_base_l>): ... this.
7589 (vec_extract<mode>): Renamed to ...
7590 (vec_extract<mode><VEC_base_l>): ... this.
7591 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
7592 (vec_initv2sfsf): ... this.
7593 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
7594 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
7595 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
7596 gen_vec_init* calls.
7598 2017-08-01 Richard Biener <rguenther@suse.de>
7600 PR tree-optimization/81297
7601 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
7602 TREE_OVERFLOW from INTEGER_CSTs.
7604 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
7606 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
7608 2017-07-31 Carl Love <cel@us.ibm.com>
7610 * config/rs6000/rs6000-c: Add support for built-in functions
7611 vector signed char vec_xl_be (signed long long, signed char *);
7612 vector unsigned char vec_xl_be (signed long long, unsigned char *);
7613 vector signed int vec_xl_be (signed long long, signed int *);
7614 vector unsigned int vec_xl_be (signed long long, unsigned int *);
7615 vector signed long long vec_xl_be (signed long long, signed long long *);
7616 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
7617 vector signed short vec_xl_be (signed long long, signed short *);
7618 vector unsigned short vec_xl_be (signed long long, unsigned short *);
7619 vector double vec_xl_be (signed long long, double *);
7620 vector float vec_xl_be (signed long long, float *);
7621 * config/rs6000/altivec.h (vec_xl_be): Add #define.
7622 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
7623 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
7625 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
7626 (altivec_expand_builtin): Add switch statement to call
7627 altivec_expand_xl_be for each builtin.
7628 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
7629 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
7630 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
7631 __builtin_vsx_le_be_v16qi.
7632 * doc/extend.texi: Update the built-in documentation file for the
7633 new built-in functions.
7635 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
7638 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
7640 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
7642 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7644 * config.gcc: Add z14.
7645 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
7646 CPU model numbers for z13s and z14.
7647 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
7649 * config/s390/s390-opts.h (enum processor_type): Rename
7650 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
7651 * config/s390/s390.c (processor_table): Add field for CPU name to
7652 be passed to Binutils.
7653 (s390_asm_output_machine_for_arch): Use the new field in
7654 processor_table for Binutils.
7655 (s390_expand_builtin): Replace arch12 with z14.
7656 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
7657 (s390_get_sched_attrmask): Likewise.
7658 (s390_get_unit_mask): Likewise.
7659 * config/s390/s390.opt: Add z14 to processor_type enum.
7661 2017-07-31 Martin Jambor <mjambor@suse.cz>
7664 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
7665 regardless of optimization level.
7667 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
7668 Martin Liska <mliska@suse.cz>
7670 * predict.def: Remove old comment and adjust probability.
7671 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
7674 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
7677 * config/i386/i386.c (ix86_function_naked): New function.
7678 (ix86_can_use_return_insn_p): Return false for naked functions.
7679 (ix86_expand_prologue): Skip prologue for naked functions.
7680 (ix86_expand_epilogue): Skip epilogue for naked functions
7681 and emit trap instruction.
7682 (ix86_warn_func_return): New function.
7683 (ix86_attribute_table): Add "naked" attribute specification.
7684 (TARGET_WARN_FUNC_RETURN): Define.
7685 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
7687 2017-07-31 Martin Liska <mliska@suse.cz>
7689 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
7690 (dump_gimple_bb_header): Always dump BB info.
7691 (pp_cfg_jump): Do not append info about BB when dumping a jump.
7693 2017-07-31 Martin Liska <mliska@suse.cz>
7696 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
7697 also with current_function_decl non-null equality.
7699 2017-07-31 Jakub Jelinek <jakub@redhat.com>
7702 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
7703 change type to the element type, instead add eltype variable and
7704 use it where we are interested in the element type.
7706 PR tree-optimization/81603
7707 * ipa-polymorphic-call.c
7708 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
7709 offset arithmetic in offset_int, bail out if the resulting bit offset
7710 doesn't fit into shwi.
7712 2017-07-31 Martin Liska <mliska@suse.cz>
7714 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
7715 (gimplify_save_expr): Fix comment.
7717 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
7720 * config/i386/i386.c (ix86_function_arg): Update arguments for
7722 (ix86_compute_frame_layout): Set the initial stack offset to
7723 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
7724 INCOMING_FRAME_SP_OFFSET.
7725 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
7726 stack before exception handler returns.
7727 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
7728 the 'ERROR_CODE' for exception handler.
7730 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
7732 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
7733 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
7734 (ASM_OUTPUT_REG_POP): Ditto.
7735 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
7736 instead of asm_fprintf to output pure string.
7738 2017-07-29 Jakub Jelinek <jakub@redhat.com>
7740 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
7741 to imported_module_or_decl hook.
7742 (debug_nothing_tree_tree_tree_bool): Remove.
7743 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
7744 * debug.c (do_nothing_debug_hooks): Use
7745 debug_nothing_tree_tree_tree_bool_bool instead of
7746 debug_nothing_tree_tree_tree_bool.
7747 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
7748 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
7749 * sdbout.c (sdb_debug_hooks): Likewise.
7750 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
7751 (gen_namespace_die): Add DW_AT_export_symbols attribute if
7753 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
7754 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
7755 attribute, don't add anything.
7757 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7759 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
7760 (fold_build2_stat_loc): Likewise.
7761 (fold_build3_stat_loc): Likewise.
7762 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
7763 (fold_build1_loc): Remove macro.
7764 (fold_build2_loc): Likewise.
7765 (fold_build3_loc): Likewise.
7767 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7769 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
7770 (gimple_build_debug_bind_source_stat): Likewise.
7771 * gimple.h (gimple_build_debug_bind): Remove macro.
7772 (gimple_build_debug_bind_source): Likewise.
7774 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7776 * bitmap.c (bitmap_alloc): Adjust.
7777 (bitmap_gc_alloc): Likewise.
7778 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
7780 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7782 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
7783 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
7784 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
7785 (bitmap_gc_alloc_stat): Likewise.
7786 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
7788 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7790 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
7791 * rtl.h (shallow_copy_rtx): Remove macro.
7793 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7795 * emit-rtl.c (gen_raw_REG): Adjust.
7796 * gengenrtl.c (gendef): Likewise.
7797 * rtl.c (rtx_alloc_stat): Remove _stat from name.
7798 * rtl.h (rtx_alloc): Remove macro.
7800 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7802 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
7803 (build_tree_list_stat): Likewise.
7804 * tree.h (build_tree_list): Remove macro.
7805 (build_tree_list_vec): Likewise.
7807 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7809 * tree.c (make_vector_stat): Remove _stat from name.
7810 (build_vector_stat): Likewise.
7811 * tree.h (make_vector_stat): Remove macro.
7812 (build_vector_stat): Likewise.
7814 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7816 * tree.h (build_var_debug_value): Remove prototype.
7818 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7820 * tree.c (tree_cons_stat): Remove _stat from name.
7821 * tree.h (tree_cons): Remove macro.
7823 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7825 * tree.c (build_vl_exp_stat): Remove _stat from name.
7826 * tree.h (build_vl_exp): Remove macro.
7828 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7830 * tree.c (build_decl_stat): Remove _stat from name.
7831 * tree.h (build_decl): Remove macro.
7833 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7835 * gimple.c (gimple_build_with_ops_stat): Adjust.
7836 (gimple_alloc_stat): Remove _stat from name.
7837 * gimple.h (gimple_alloc): Remove macro.
7839 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7841 * tree.c (make_tree_vec_stat): Remove _stat from name.
7842 (grow_tree_vec_stat): Likewise.
7843 * tree.h (make_tree_vec_stat): Adjust prototype.
7844 (grow_tree_vec_stat): Likewise.
7845 (make_tree_vec): Remove macro.
7846 (grow_tree_vec): Likewise.
7848 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7850 * fold-const.c (fold_build1_stat_loc): Adjust.
7851 (fold_build2_stat_loc): Likewise.
7852 (fold_build3_stat_loc): Likewise.
7853 * tree.c (build0_stat): Remove _stat from name.
7854 (build1_stat): Likewise.
7855 (build2_stat): Likewise.
7856 (build3_stat): Likewise.
7857 (build4_stat): Likewise.
7858 (build5_stat): Likewise.
7859 * tree.h (build1_loc): Remove macro, and rename _stat function
7861 (build2_loc): Likewise.
7862 (build3_loc): Likewise.
7863 (build4_loc): Likewise.
7864 (build5_loc): Likewise.
7866 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7868 * tree.c (make_int_cst_stat): Remove _stat from name.
7869 * tree.h (make_int_cst_stat): Adjust prototype.
7870 (make_int_cst): Remove macro.
7872 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7874 * tree.c (make_tre_binfo_stat): Remove _stat from name.
7875 * tree.h (make_tree_binfo_stat): Adjust prototype.
7876 (make_tree_binfo): Remove.
7878 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7880 * tree.c (copy_node_stat): Rename to copy_node.
7881 (build_distinct_type_copy): Adjust.
7882 * tree.h (copy_node_stat): Adjust prototype.
7883 (copy_node): Remove macro.
7885 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7887 * tree.c (make_node_stat): rename to make_node.
7888 (build_tree_list_stat): Adjust.
7889 (build0_stat): Likewise.
7890 (build2_stat): Likewise.
7891 (build3_stat): Likewise.
7892 (build4_stat): Likewise.
7893 (build5_stat): Likewise.
7894 (build_decl_stat): Likewise.
7895 * tree.h (make_node_stat): Adjust prototype.
7896 (make_node): remove macro.
7898 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
7900 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
7901 (PPC_FEATURE2_SCV): Likewise.
7902 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
7904 2017-07-28 Tamar Christina <tamar.christina@arm.com>
7906 * config/aarch64/aarch64.c
7907 (aarch64_internal_mov_immediate): Add new special pattern.
7908 * config/aarch64/aarch64.md (*movdi_aarch64):
7909 Add reg/32bit const mov case.
7911 2017-07-28 Tamar Christina <tamar.christina@arm.com>
7912 Richard Sandiford <richard.sandiford@linaro.org>
7914 * config/aarch64/aarch64.md (mov<mode>): Generalize.
7915 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
7916 Add integer and movi cases.
7917 (movi-split-hf-df-sf split, fp16): New.
7918 (enabled): Added TARGET_FP_F16INST.
7919 * config/aarch64/iterators.md (GPF_HF): New.
7920 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
7922 2017-07-28 Tamar Christina <tamar.christina@arm.com>
7924 * config/aarch64/aarch64.c
7925 (aarch64_simd_container_mode): Add prototype.
7926 (aarch64_expand_mov_immediate): Add HI support.
7927 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
7928 (aarch64_can_const_movi_rtx_p): New.
7929 (aarch64_preferred_reload_class):
7930 Remove restrictions of using FP registers for certain SIMD operations.
7931 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
7932 (aarch64_valid_floating_const): Add integer move validation.
7933 (aarch64_simd_imm_scalar_p): Remove.
7934 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
7935 (aarch64_legitimate_constant_p): Expand list of supported cases.
7936 * config/aarch64/aarch64-protos.h
7937 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
7938 (aarch64_reinterpret_float_as_int): New.
7939 (aarch64_simd_imm_scalar_p): Remove.
7940 * config/aarch64/constraints.md (Uvi): New.
7941 (Dd): Split into Ds and new Dd.
7942 * config/aarch64/aarch64.md (*movsi_aarch64):
7944 (*movdi_aarch64): Add SIMD mov case.
7946 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7948 * tree-predcom.c: (struct chain): Handle store-store chain in which
7949 stores for elimination only store loop invariant values.
7950 (execute_pred_commoning_chain): Ditto.
7951 (prepare_initializers_chain_store_elim): Ditto.
7952 (prepare_finalizers): Ditto.
7953 (is_inv_store_elimination_chain): New function.
7954 (initialize_root_vars_store_elim_1): New function.
7956 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7958 * tree-predcom.c: Revise general description of the pass.
7959 (enum chain_type): New enum type for store elimination.
7960 (struct chain): New field supporting store elimination.
7961 (struct component): Ditto.
7962 (dump_chain): Dump store-stores chain.
7963 (release_chain): Release resources.
7964 (split_data_refs_to_components): Compute and create component
7965 contains only stores for elimination.
7966 (get_chain_last_ref_at): New function.
7967 (make_invariant_chain): Initialization.
7968 (make_rooted_chain): Specify chain type in parameter and record it.
7969 (add_looparound_copies): Skip for store-stores chain.
7970 (determine_roots_comp): Compute type of chain and pass it to
7972 (initialize_root_vars_store_elim_2): New function.
7973 (finalize_eliminated_stores): New function.
7974 (remove_stmt): Handle store for elimination.
7975 (execute_pred_commoning_chain): Execute predictive commoning on
7977 (determine_unroll_factor): Skip unroll for store-stores chain.
7978 (prepare_initializers_chain_store_elim): New function.
7979 (prepare_initializers_chain): Hanlde store-store chain.
7980 (prepare_finalizers_chain, prepare_finalizers): New function.
7981 (tree_predictive_commoning_loop): Return integer value indicating
7982 if loop is unrolled or lcssa form is corrupted.
7983 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
7985 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7987 * tree-predcom.c (initialize_root): Delete.
7988 (execute_pred_commoning_chain): Initialize root vars and replace
7989 reference of non-combined chain directly, rather than call above
7992 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7994 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
7995 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
7997 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7999 * tree-predcom.c (struct chain): New field init_seq.
8000 (release_chain): Release init_seq.
8001 (prepare_initializers_chain): Record intialization stmts in above
8003 (insert_init_seqs): New function.
8004 (tree_predictive_commoning_loop): Call insert_init_seqs.
8006 2017-07-28 Bin Cheng <bin.cheng@arm.com>
8008 * tree-predcom.c (determine_roots_comp): Skip trivial components.
8010 2017-07-28 Richard Biener <rguenther@suse.de>
8012 * match.pd: Remove superfluous :c.
8013 * genmatch.c (simplify::id): Add member.
8014 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
8016 (current_id): New global.
8017 (dt_node::parent): Move from ...
8018 (dt_operand::parent): ... here. Add for_id member.
8019 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
8020 (decision_tree::find_node): Relax order requirement when
8021 merging DT_TRUE nodes to ones inbetween the current simplify
8022 and the one we try to merge with. Add diagnostic whenever
8023 we need to enforce pattern order by not merging.
8024 (decision_tree::insert): Set current_id.
8025 (decision_tree::print_node): Dump parent node and for_id.
8026 (parser::last_id): Add member.
8027 (parser::push_simplify): Assign unique id.
8028 (parser::parser): Initialize last_id.
8030 2017-07-28 Martin Liska <mliska@suse.cz>
8033 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
8034 gimple_build_debug_bind.
8036 2017-07-28 Richard Biener <rguenther@suse.de>
8038 PR tree-optimization/81502
8039 * match.pd: Add pattern combining BIT_INSERT_EXPR with
8041 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
8043 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
8044 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
8045 for BIT_FIELD_REF args.
8046 * fold-const.c (make_bit_field_ref): Likewise.
8047 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
8049 2017-07-28 Jakub Jelinek <jakub@redhat.com>
8052 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
8053 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
8054 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
8055 Or it into SANITIZER_UNDEFINED.
8056 * ubsan.c: Include gimple-fold.h and varasm.h.
8057 (ubsan_expand_ptr_ifn): New function.
8058 (instrument_pointer_overflow): New function.
8059 (maybe_instrument_pointer_overflow): New function.
8060 (instrument_object_size): Formatting fix.
8061 (pass_ubsan::execute): Call instrument_pointer_overflow
8062 and maybe_instrument_pointer_overflow.
8063 * internal-fn.c (expand_UBSAN_PTR): New function.
8064 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
8065 * sanitizer.def (__ubsan_handle_pointer_overflow,
8066 __ubsan_handle_pointer_overflow_abort): New builtins.
8067 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
8068 * internal-fn.def (UBSAN_PTR): New internal function.
8069 * opts.c (sanitizer_opts): Add pointer-overflow.
8070 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
8071 * fold-const.c (build_range_check): Compute pointer range check in
8072 integral type if pointer arithmetics would be needed. Formatting
8075 2017-07-28 Martin Liska <mliska@suse.cz>
8078 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
8079 parameters that are of a variable-length.
8081 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
8083 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
8085 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
8086 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
8087 (CRT_CALL_STATIC_FUNCTION): Likewise.
8088 (ASM_DEFAULT_SPEC): New define.
8089 (ASM_SPEC32): Likewise.
8090 (ASM_SPEC64): Likewise.
8091 (ASM_SPEC_COMMON): Likewise.
8092 (ASM_SPEC): Likewise.
8093 (INVALID_64BIT): Likewise.
8094 (LINK_OS_DEFAULT_SPEC): Likewise.
8095 (LINK_OS_SPEC32): Likewise.
8096 (LINK_OS_SPEC64): Likewise.
8097 (POWERPC_LINUX): Likewise.
8098 (PTRDIFF_TYPE): Likewise.
8099 (RESTORE_FP_PREFIX): Likewise.
8100 (RESTORE_FP_SUFFIX): Likewise.
8101 (SAVE_FP_PREFIX): Likewise.
8102 (SAVE_FP_SUFFIX): Likewise.
8103 (SIZE_TYPE): Likewise.
8104 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
8105 (TARGET_64BIT): Likewise.
8106 (TARGET_64BIT): Likewise.
8107 (TARGET_AIX): Likewise.
8108 (WCHAR_TYPE_SIZE): Likewise.
8109 (WCHAR_TYPE): Undefine.
8110 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
8111 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
8112 (CPP_OS_RTEMS_SPEC): Delete.
8113 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
8114 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
8116 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
8118 2017-07-28 Jakub Jelinek <jakub@redhat.com>
8120 PR tree-optimization/81578
8121 * tree-parloops.c (build_new_reduction): Bail out if
8122 reduction_code isn't one of the standard OpenMP reductions.
8123 Move the details printing after that decision.
8125 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
8127 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
8128 related to reload_in_progress.
8129 (splat_input_operand): Likewise.
8130 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
8132 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
8134 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
8135 (TARGET_INSTANTIATE_DECLS): Likewise.
8136 (legitimate_indexed_address_p): Delete reload_in_progress code.
8137 (rs6000_debug_legitimate_address_p): Likewise.
8138 (rs6000_eliminate_indexed_memrefs): Likewise.
8139 (rs6000_emit_le_vsx_store): Likewise.
8140 (rs6000_emit_move_si_sf_subreg): Likewise.
8141 (rs6000_emit_move): Likewise.
8142 (register_to_reg_type): Likewise.
8143 (rs6000_pre_atomic_barrier): Likewise.
8144 (rs6000_machopic_legitimize_pic_address): Likewise.
8145 (rs6000_allocate_stack_temp): Likewise.
8146 (rs6000_address_for_fpconvert): Likewise.
8147 (rs6000_address_for_altivec): Likewise.
8148 (rs6000_secondary_memory_needed_rtx): Delete function.
8149 (rs6000_check_sdmode): Likewise.
8150 (rs6000_alloc_sdmode_stack_slot): Likewise.
8151 (rs6000_instantiate_decls): Likewise.
8152 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
8153 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
8154 Delete reload_in_progress.
8155 (*vec_reload_and_plus_<mptrsize>): Likewise.
8156 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
8157 (vsx_div_v2di): Likewise.
8158 (vsx_udiv_v2di): Likewise.
8160 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
8162 * config/rs6000/rs6000.opt (mlra): Replace with stub.
8163 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
8164 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
8165 (rs6000_debug_reg_global): Delete print of LRA status.
8166 (rs6000_option_override_internal): Delete dead LRA related code.
8167 (rs6000_lra_p): Delete function.
8168 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
8170 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
8172 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
8173 * config/riscv/rtems.h: New file.
8175 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8176 Sudakshina Das <sudi.das@arm.com>
8178 * config/aarch64/aarch64.md
8179 (define_split for and<mode>3nr_compare): Move
8180 non aarch64_logical_operand to a register.
8181 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
8182 register immediate operand to a register.
8183 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
8185 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
8188 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
8190 2017-07-27 Richard Biener <rguenther@suse.de>
8192 PR tree-optimization/81573
8193 PR tree-optimization/81494
8194 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
8195 multi defuse cycle case.
8197 2017-07-27 Richard Biener <rguenther@suse.de>
8199 PR tree-optimization/81571
8200 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
8203 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
8205 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
8206 earlier and only if MASK_FPU is set. Adjust formatting.
8208 2017-07-27 Martin Liska <mliska@suse.cz>
8210 * opt-functions.awk: Add validation of value of Init.
8211 * optc-gen.awk: Pass new argument.
8213 2017-07-27 Martin Liska <mliska@suse.cz>
8215 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
8216 Fix wrong condition.
8218 2017-07-27 Martin Liska <mliska@suse.cz>
8220 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
8221 BBs and edges seen by autoFDO.
8223 2017-07-27 Richard Biener <rguenther@suse.de>
8225 PR tree-optimization/81502
8226 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
8227 with incompatible but same sized type.
8228 (execute_update_addresses_taken): Likewise.
8230 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
8232 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
8233 flag_tree_loop_vectorize rather than flag_tree_vectorize.
8235 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8238 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
8239 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
8240 Change s_operand to memory_operand.
8242 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
8244 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
8245 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
8246 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
8247 Emit instructions rather than returning an expression. Handle TFmode
8248 and KFmode by casting to TImode.
8249 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
8250 (rs6000_emit_le_vsx_store): Likewise.
8251 * config/rs6000/vsx.md (VSX_TI): New iterator.
8252 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
8253 (*vsx_le_undo_permute_<mode>): Likewise.
8254 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
8255 emit the split sequence.
8256 (*vsx_le_perm_store_<mode>): Likewise.
8258 2017-07-27 Jakub Jelinek <jakub@redhat.com>
8260 PR tree-optimization/81555
8261 PR tree-optimization/81556
8262 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
8263 if true, force CHANGED for the recursive invocation.
8264 (reassociate_bb): Remember original length of ops array, pass
8265 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
8267 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
8268 attributes for noipa attribute. For naked attribute use
8269 lookup_attribute first before lookup_attribute_spec.
8270 * final.c (rest_of_handle_final): Disable IPA RA for functions with
8272 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
8273 for functions with noipa attribute.
8274 (cgraph_externally_visible_p): Return true for functions with noipa
8276 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
8277 for functions with noipa attribute.
8278 * doc/extend.texi: Document noipa function attribute.
8279 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
8280 also for functions with noipa attribute.
8281 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
8283 2017-07-26 Andrew Pinski <apinski@cavium.com>
8285 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
8286 vec_unalign_load_cost and vec_unalign_store_cost.
8288 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
8290 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
8291 -mvsx-small-integer option.
8292 (ISA_3_0_MASKS_IEEE): Likewise.
8293 (OTHER_VSX_VECTOR_MASKS): Likewise.
8294 (POWERPC_MASKS): Likewise.
8295 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
8296 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
8297 code, only testing for DImode being allowed in non-VSX floating
8299 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
8300 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
8302 (rs6000_option_override_internal): Delete -mvsx-small-integer.
8303 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
8304 TARGET_P8_VECTOR test.
8305 (rs6000_secondary_reload_simple_move): Likewise.
8306 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
8307 since TARGET_P9_VECTOR was already tested.
8308 (rs6000_opt_masks): Remove -mvsx-small-integer.
8309 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
8310 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
8312 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
8313 test for TARGET_VEXTRACTUB was used, and that uses
8315 (p9 extract splitter): Likewise.
8316 (vsx_extract_<mode>_di_p9): Likewise.
8317 (vsx_extract_<mode>_store_p9): Likewise.
8318 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
8319 for TARGET_P9_VECTOR was used. Delete code that is now dead with
8320 the elimination of TARGET_VSX_SMALL_INTEGER.
8321 (vsx_extract_<mode>_p8): Likewise.
8322 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
8323 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
8324 (vsx_set_<mode>_p9): Likewise.
8325 (vsx_set_v4sf_p9): Likewise.
8326 (vsx_set_v4sf_p9_zero): Likewise.
8327 (vsx_insert_extract_v4sf_p9): Likewise.
8328 (vsx_insert_extract_v4sf_p9_2): Likewise.
8329 * config/rs6000/rs6000.md (sign extend splitter): Change
8330 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
8331 (floatsi<mode>2_lfiwax_mem): Likewise.
8332 (floatunssi<mode>2_lfiwzx_mem): Likewise.
8333 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
8334 since a test for TARGET_P9_VECTOR was used.
8335 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
8336 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
8337 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
8338 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
8339 TARGET_P8_VECTOR test.
8340 (fix_trunc<mode>si2_stfiwx): Likewise.
8341 (fix_trunc<mode>si2_internal): Likewise.
8342 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
8343 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
8345 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
8346 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
8347 TARGET_P8_VECTOR test.
8348 (fixuns_trunc<mode>si2_stfiwx): Likewise.
8349 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
8350 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
8352 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
8353 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
8354 since a test for TARGET_P9_VECTOR was used.
8355 (splitter for loading small constants): Likewise.
8357 2017-07-26 Andrew Pinski <apinski@cavium.com>
8359 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
8362 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
8365 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
8366 (fp_valid_at): Likewise.
8368 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
8370 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
8371 (qdf24xx_addrcost_table): Likewise.
8372 (cortexa57_tunings): Update to use generic_branch_cost.
8373 (cortexa72_tunings): Likewise.
8374 (cortexa73_tunings): Likewise.
8375 (qdf24xx_tunings): Likewise.
8377 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
8379 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
8380 (thunderx2t99_branch_cost): Likewise.
8381 (cortexa35_tunings): Update to use generic_branch_cost.
8382 (cortexa53_tunings): Likewise.
8383 (cortexa57_tunings): Likewise.
8384 (cortexa72_tunings): Likewise.
8385 (cortexa73_tunings): Likewise.
8386 (thunderx2t99_tunings): Likewise.
8388 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
8390 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
8391 (sparc_option_override): Honour MASK_FSMULD.
8392 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
8393 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
8394 * config/sparc/sparc.opt (mfsmuld): New option.
8395 * doc/invoke.texi (mfsmuld): Document option.
8397 2017-07-26 Marek Polacek <polacek@redhat.com>
8400 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
8402 2017-07-26 Richard Biener <rguenther@suse.de>
8404 * gimple-match-head.c (do_valueize): Return OP if valueize
8406 (get_def): New helper to get at the def stmt of a SSA name
8408 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
8409 do_valueize to get at the def stmt.
8410 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
8412 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
8415 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
8417 2017-07-26 Martin Liska <mliska@suse.cz>
8420 * function.c (expand_function_start): Make expansion of
8421 nonlocal_goto_save_area after parm_birth_insn.
8423 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
8425 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
8426 from all CPU target flags enable members.
8428 2017-07-26 Richard Biener <rguenther@suse.de>
8430 * genmatch.c (dt_simplify::gen): Make iterator vars const.
8431 (decision_tree::gen): Make 'type' const.
8432 (write_predicate): Likewise.
8434 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
8436 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
8437 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
8438 (rs6000_option_override_internal): Likewise.
8439 (rs6000_expand_vector_set): Likewise.
8440 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
8441 (TARGET_UPPER_REGS_SF): Likewise.
8442 (TARGET_UPPER_REGS_DI): Likewise.
8443 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
8444 (TARGET_DIRECT_MOVE_64BIT): Likewise.
8445 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
8446 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
8447 (Splitters for DI constants in Altivec registers): Likewise.
8448 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
8449 (vsx_set_v4sf_p9): Likewise.
8450 (vsx_set_v4sf_p9_zero): Likewise.
8451 (vsx_insert_extract_v4sf_p9): Likewise.
8452 (vsx_insert_extract_v4sf_p9_2): Likewise.
8454 2017-07-25 Carl Love <cel@us.ibm.com>
8456 * doc/extend.texi: Update the built-in documentation file for the
8457 existing built-in functions
8458 vector signed char vec_cnttz (vector signed char);
8459 vector unsigned char vec_cnttz (vector unsigned char);
8460 vector signed short vec_cnttz (vector signed short);
8461 vector unsigned short vec_cnttz (vector unsigned short);
8462 vector signed int vec_cnttz (vector signed int);
8463 vector unsigned int vec_cnttz (vector unsigned int);
8464 vector signed long long vec_cnttz (vector signed long long);
8465 vector unsigned long long vec_cnttz (vector unsigned long long);
8467 2017-07-25 Andrew Pinski <apinski@cavium.com>
8469 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
8470 accesses where the use is for the first operand of a BIT_INSERT.
8472 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
8475 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
8476 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
8478 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
8480 * config/i386/gstabs.h: Delete.
8481 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
8483 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
8485 * config/i386/i386.c (ix86_decompose_address): Do not check for
8486 register RTX when looking at index_reg or base_reg.
8487 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
8489 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
8491 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
8492 to update EH info here.
8494 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
8496 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
8498 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
8500 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
8502 2017-07-25 Torsten Duwe <duwe@suse.de>
8504 * common.opt: Introduce -fpatchable-function-entry
8505 command line option, and its variables function_entry_patch_area_size
8506 and function_entry_patch_area_start.
8507 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
8508 including a two-value parser.
8509 * target.def (print_patchable_function_entry): New target hook.
8510 * targhooks.h (default_print_patchable_function_entry): New function.
8511 * targhooks.c (default_print_patchable_function_entry): Likewise.
8512 * toplev.c (process_options): Switch off IPA-RA if
8513 patchable function entries are being generated.
8514 * varasm.c (assemble_start_function): Look at the
8515 patchable-function-entry command line switch and current
8516 function attributes and maybe generate NOP instructions by
8517 calling the print_patchable_function_entry hook.
8518 * doc/extend.texi: Document patchable_function_entry attribute.
8519 * doc/invoke.texi: Document -fpatchable_function_entry
8520 command line option.
8521 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
8523 * doc/tm.texi: Re-generate.
8525 2017-07-25 Jakub Jelinek <jakub@redhat.com>
8528 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
8529 TARGET_AVX512DQ rather than TARGET_AVX512BW.
8531 2017-07-25 Tamar Christina <tamar.christina@arm.com>
8533 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
8535 2017-07-25 Richard Biener <rguenther@suse.de>
8537 PR tree-optimization/81455
8538 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
8539 not walk in cycles when looking for guards.
8541 2017-07-25 Richard Biener <rguenther@suse.de>
8543 PR tree-optimization/81529
8544 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
8545 when optimizing backedge uses.
8547 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
8549 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
8551 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
8552 to dl_section_ref. On AIX, append an expression to subtract
8553 the size of the section length to dl_section_ref.
8555 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
8557 * configure.ac: If any of the config.* scripts fail, exit 1.
8558 * configure: Regenerate.
8560 2017-07-25 Richard Biener <rguenther@suse.de>
8563 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
8564 of immediate uses, be more verbose on errors.
8566 2017-07-25 Richard Biener <rguenther@suse.de>
8568 PR tree-optimization/81510
8569 * tree-vect-loop.c (vect_is_simple_reduction): When the
8570 reduction stmt is not inside the loop bail out.
8572 2017-07-25 Richard Biener <rguenther@suse.de>
8574 PR tree-optimization/81303
8575 * tree-vect-loop-manip.c (vect_loop_versioning): Build
8576 profitability check against LOOP_VINFO_NITERSM1.
8578 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
8580 * domwalk.c (cmp_bb_postorder): Simplify.
8581 (sort_bbs_postorder): New function. Use it...
8582 (dom_walker::walk): ...here to optimize common cases.
8584 2017-07-25 Martin Liska <mliska@suse.cz>
8587 * ipa-visibility.c (function_and_variable_visibility): Make the
8588 redirection just on target that supports aliasing.
8589 Fix GNU coding style.
8591 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
8594 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
8596 * config/arm/rtems.h: Likewise.
8597 * config/bfin/rtems.h: Likewise.
8598 * config/i386/rtemself.h: Likewise.
8599 * config/lm32/rtems.h: Likewise.
8600 * config/m32c/rtems.h: Likewise.
8601 * config/m68k/rtemself.h: Likewise.
8602 * config/microblaze/rtems.h: Likewise.
8603 * config/mips/rtems.h: Likewise.
8604 * config/moxie/rtems.h: Likewise.
8605 * config/nios2/rtems.h: Likewise.
8606 * config/powerpcspe/rtems.h: Likewise.
8607 * config/rs6000/rtems.h: Likewise.
8608 * config/rtems.h: Likewise.
8609 * config/sh/rtems.h: Likewise.
8610 * config/sh/rtemself.h: Likewise.
8611 * config/sparc/rtemself.h: Likewise.
8613 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
8616 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
8617 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
8618 * tree-ssa-structalias.c (alias_get_name): Same.
8620 2017-07-25 Bin Cheng <bin.cheng@arm.com>
8623 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
8624 instructions if no du chain is found.
8626 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
8628 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
8630 2017-07-25 Richard Biener <rguenther@suse.de>
8633 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
8636 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
8638 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
8640 (ISA_2_7_MASKS_SERVER): Likewise.
8641 (ISA_3_0_MASKS_IEEE): Likewise.
8642 (OTHER_P8_VECTOR_MASKS): Likewise.
8643 (OTHER_VSX_VECTOR_MASKS): Likewise.
8644 (POWERPC_MASKS): Likewise.
8645 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
8646 duplicate list of options.
8647 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
8648 explicit -mupper-regs options.
8649 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
8650 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
8651 alias for -mupper-regs-df.
8652 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
8653 (rs6000_init_hard_regno_mode_ok): Likewise.
8654 (rs6000_option_override_internal): Likewise.
8655 (rs6000_opt_masks): Likewise.
8656 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
8657 options in terms of whether -mvsx or -mpower8-vector was used.
8658 (TARGET_UPPER_REGS_DI): Likewise.
8659 (TARGET_UPPER_REGS_SF): Likewise.
8660 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
8661 -mupper-regs-* options.
8663 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
8665 * passes.c (emergency_dump_function): Print some empty lines and a
8666 header before the RTL dump.
8668 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
8670 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
8672 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
8675 * config/aarch64/aarch64.c (aarch64_classify_symbol):
8676 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
8678 2017-07-24 Carl Love <cel@us.ibm.com>
8680 * config/rs6000/rs6000-c.c: Add support for built-in functions
8681 vector float vec_extract_fp32_from_shorth (vector unsigned short);
8682 vector float vec_extract_fp32_from_shortl (vector unsigned short);
8683 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
8684 vec_extract_fp_from_shortl): Add defines for the two builtins.
8685 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
8686 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
8688 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
8689 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
8690 * doc/extend.texi: Update the built-in documentation file for the
8691 new built-in function.
8693 2017-07-24 Jakub Jelinek <jakub@redhat.com>
8696 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
8698 * doc/generic.texi: Likewise.
8699 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
8700 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
8702 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
8704 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
8705 (aarch64_mls_elt_merge<mode>): Likewise.
8707 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
8709 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
8710 having __cxa_atexit.
8712 2017-07-23 Michael Collison <michael.collison@arm.com>
8714 * config/arm/arm.c (arm_option_override): Deprecate
8715 use of -mstructure-size-boundary.
8716 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
8717 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
8719 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8722 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
8723 Reduce cost estimate for direct moves.
8725 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
8728 * config/i386/i386.c (ix86_option_override_internal): Disable
8729 BMI, BMI2 and TBM instructions for -m16.
8731 2017-07-21 Carl Love <cel@us.ibm.com>
8733 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8734 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
8735 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
8736 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
8737 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
8738 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
8739 VMULOSW): New enum "unspec" values.
8740 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
8741 altivec_vmulosw): New patterns.
8742 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
8743 VMULOSW): Add definitions.
8745 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
8747 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
8748 (qdf24xx): Likewise.
8749 * config/aarch64/aarch64-options-extensions.def (rdma); New.
8750 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
8751 (AARCH64_FL_V8_1): Renumber.
8752 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
8753 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
8754 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
8755 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
8756 rdma to feature modifiers list.
8758 2017-07-21 Yury Gribov <tetra2005@gmail.com>
8761 * ipa-visibility (function_and_variable_visibility): Convert
8762 recursive PLT call to direct call if appropriate.
8764 2017-07-21 Andrew Pinski <apinski@cavium.com>
8766 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
8767 operand 1 to see if the types precision matches.
8768 * fold-const.c (operand_equal_p): Likewise.
8770 2017-07-21 Richard Biener <rguenther@suse.de>
8772 PR tree-optimization/81303
8773 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
8774 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
8775 (vect_peeling_hash_get_lowest_cost): Adjust.
8776 (vect_enhance_data_refs_alignment): Likewise. Use
8777 vect_get_peeling_costs_all_drs to compute the penalty for no
8778 peeling to match up costs.
8780 2017-07-21 Richard Biener <rguenther@suse.de>
8782 PR tree-optimization/81500
8783 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
8784 we didn't identify a reduction path.
8786 2017-07-21 Tom de Vries <tom@codesourcery.com>
8787 Cesar Philippidis <cesar@codesourcery.com>
8789 PR gcov-profile/81442
8790 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
8793 2017-07-21 Tom de Vries <tom@codesourcery.com>
8796 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
8798 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
8799 nvptx_override_options_after_change.
8801 2017-07-21 Ulrich Drepper <drepper@redhat.com>
8803 * dwarf2out.c (output_file_names): Avoid double testing for
8806 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
8808 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
8810 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
8812 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
8814 (try_crossjump_to_edge): Do not punt on partitioned functions.
8816 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
8818 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
8819 Put all BBs reachable only via paths crossing cold region to cold
8821 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
8823 2016-07-21 Richard Biener <rguenther@suse.de>
8825 PR tree-optimization/81303
8826 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
8827 into account prologue and epilogue iterations when raising
8828 min_profitable_iters to sth at least covering one vector iteration.
8830 2017-07-21 Tamar Christina <tamar.christina@arm.com>
8832 * config/arm/arm.c (arm_test_cpu_arch_dat):
8835 2017-07-20 Nathan Sidwell <nathan@acm.org>
8837 Remove TYPE_METHODS.
8838 * tree.h (TYPE_METHODS): Delete.
8839 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
8840 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
8841 (dbxout_type_methods): Scan TYPE_FIELDS.
8842 (dbxout_type): Don't check TYPE_METHODS here.
8843 * function.c (use_register_for_decl): Always ignore register for
8844 class types when not optimizing.
8845 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
8846 * tree.c (free_lang_data_in_type): Stitch out member functions and
8847 templates from TYPE_FIELDS.
8848 (build_distinct_type_copy, verify_type_variant,
8849 verify_type): Member fns are on TYPE_FIELDS.
8850 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
8851 * tree-pretty-print.c (dump_generic_node): Likewise.
8853 2017-07-20 Jakub Jelinek <jakub@redhat.com>
8856 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
8857 V2TImode and V4TImode.
8858 (ix86_expand_vector_extract): Likewise.
8859 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
8860 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
8861 (ssescalarmode): Handle V4TImode and V2TImode.
8862 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
8863 (*vec_extractv2ti, *vec_extractv4ti): New insns.
8864 (VEXTRACTI128_MODE): New mode iterator.
8865 (splitter for *vec_extractv?ti first element): New.
8866 (VEC_INIT_MODE): New mode iterator.
8867 (vec_init<mode>): Consolidate 3 expanders into one using
8868 VEC_INIT_MODE mode iterator.
8870 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
8872 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
8873 non_spilled_static_chain_regno_p.
8875 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
8877 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
8879 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
8881 * bb-reorder.c (connect_traces): Allow copying of blocks within
8884 2017-07-20 Richard Biener <rguenther@suse.de>
8886 * gimple.h (gimple_phi_result): Add gphi * overload.
8887 (gimple_phi_result_ptr): Likewise.
8888 (gimple_phi_arg): Likewise. Adjust index assert to only
8889 allow actual argument accesses rather than all slots available
8891 (gimple_phi_arg_def): Add gphi * overload.
8892 * tree-phinodes.c (make_phi_node): Initialize only actual
8894 (resize_phi_node): Clear memory not covered by old node,
8895 do not initialize excess argument slots.
8896 (reserve_phi_args_for_new_edge): Initialize new argument slot
8899 2017-07-20 Bin Cheng <bin.cheng@arm.com>
8901 PR tree-optimization/81388
8903 2016-07-21 Bin Cheng <bin.cheng@arm.com>
8905 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
8906 by removing computation of may_be_zero.
8908 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
8909 Tom de Vries <tom@codesourcery.com>
8912 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
8913 when gimple level profile disagrees with what RTL expander did.
8915 2017-07-20 Richard Biener <rguenther@suse.de>
8917 PR tree-optimization/61171
8918 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
8919 (vect_analyze_stmt): Add slp instance parameter.
8920 (vectorizable_reduction): Likewise.
8921 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
8922 (vect_is_simple_reduction): Deal with chains not detected
8923 as SLP reduction chain, specifically not properly associated
8924 chains containing a mix of plus/minus.
8925 (get_reduction_op): Remove.
8926 (get_initial_defs_for_reduction): Simplify, pass in whether
8927 this is a reduction chain, pass in the SLP node for the PHIs.
8928 (vect_create_epilog_for_reduction): Get the SLP instance as
8930 (vectorizable_reduction): Get the SLP instance as arg.
8931 During analysis remember the SLP node with the PHIs in the
8932 instance. Simplify getting at the vectorized reduction PHIs.
8933 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
8934 through SLP instance.
8935 (vect_slp_analyze_operations): Likewise.
8936 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
8937 (vect_transform_stmt): Likewise.
8939 2017-07-20 Tom de Vries <tom@codesourcery.com>
8941 PR tree-optimization/81489
8942 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
8943 read of phi arg location to before loop that modifies phi.
8945 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
8947 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
8950 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
8953 * except.c (execute): Fix ordering issue.
8955 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
8957 PR rtl-optimization/81423
8958 * combine.c (make_compound_operation_int): Don't try to optimize
8959 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
8961 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
8963 PR rtl-optimization/81423
8964 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
8965 with a constant that is -1 in the truncated to mode.
8967 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
8969 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
8970 (determine_unlikely_bbs): ... here.
8971 * predict.h (propagate_unlikely_bbs_forward): Declare.
8972 * cfgexpand.c (pass_expand::execute): Use it.
8973 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
8975 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
8976 propagate_unlikely_bbs_forward.
8978 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
8981 * except.c (maybe_add_nop_after_section_switch): New function.
8984 2017-07-19 Tom de Vries <tom@codesourcery.com>
8986 * gimple.h (gimple_phi_set_arg): Make assert more strict.
8988 2017-07-19 Tom de Vries <tom@codesourcery.com>
8990 * gimple.h (gimple_phi_arg): Make assert more strict.
8992 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
8994 * config.gcc (powerpc*-*-*): Add mmintrin.h.
8995 * config/rs6000/mmintrin.h: New file.
8996 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
8998 2017-07-19 Jakub Jelinek <jakub@redhat.com>
9000 PR tree-optimization/81346
9001 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
9003 2017-07-19 Tom de Vries <tom@codesourcery.com>
9005 * config/nvptx/nvptx.md (VECIM): Add V2DI.
9007 2017-07-19 Tom de Vries <tom@codesourcery.com>
9009 * config/nvptx/nvptx-modes.def: Add V2DImode.
9010 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
9011 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
9012 (nvptx_output_mov_insn): Handle lack of mov.b128.
9013 (nvptx_print_operand): Handle 'H' and 'L' codes.
9014 (nvptx_vector_mode_supported): Allow V2DImode.
9015 (nvptx_preferred_simd_mode): New function.
9016 (nvptx_data_alignment): New function.
9017 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
9018 nvptx_preferred_simd_mode.
9019 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
9021 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
9023 2017-07-19 Tom de Vries <tom@codesourcery.com>
9025 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
9026 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
9027 (nvptx_vector_mode_supported): New function. Allow V2SImode.
9028 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
9029 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
9030 (mov<VECIM>_insn): New define_insn.
9031 (define_expand "mov<VECIM>): New define_expand.
9033 2017-07-19 Tom de Vries <tom@codesourcery.com>
9035 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
9037 2017-07-19 Jakub Jelinek <jakub@redhat.com>
9039 PR tree-optimization/81346
9040 * fold-const.h (fold_div_compare, range_check_type): Declare.
9041 * fold-const.c (range_check_type): New function.
9042 (build_range_check): Use range_check_type.
9043 (fold_div_compare): No longer static, rewritten into
9044 a match.pd helper function.
9045 (fold_comparison): Don't call fold_div_compare here.
9046 * match.pd (X / C1 op C2): New optimization using fold_div_compare
9049 2017-07-19 Nathan Sidwell <nathan@acm.org>
9051 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
9052 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
9053 * tree.c (find_decls_types_r, verify_type): Use
9054 TYPE_{MIN,MAX}_VALUE_RAW.
9055 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
9056 (hash_tree): Likewise.
9057 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
9059 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
9062 2017-07-18 Tom de Vries <tom@codesourcery.com>
9065 * omp-expand.c (expand_omp_for_static_chunk): Handle
9066 equal-argument loop exit phi.
9068 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
9071 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
9072 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
9073 operand 2 predicate.
9074 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
9075 operand 2 predicate.
9076 (ror,rol -> rorx splitters): Use const_int_operand as
9077 operand 2 predicate.
9079 2017-06-18 Richard Biener <rguenther@suse.de>
9081 PR tree-optimization/81410
9082 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
9083 the gap in the ! slp_perm SLP case after each group.
9085 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
9088 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
9091 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
9094 * predict.c (set_even_probabilities): Cleanup; do not affect
9095 probabilities that are already known.
9096 (combine_predictions_for_bb): Call even when count is set.
9098 2017-07-18 Nathan Sidwell <nathan@acm.org>
9100 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
9103 2017-07-18 Bin Cheng <bin.cheng@arm.com>
9106 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
9107 optimization for loop niter analysis.
9109 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
9112 * config/avr/avr.c (avr_optimize_casesi): Don't use
9113 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
9115 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
9117 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
9118 body_cost_vec from _vect_peel_extended_info.
9119 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
9120 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
9123 2017-07-18 Bin Cheng <bin.cheng@arm.com>
9125 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
9127 2017-07-18 Richard Biener <rguenther@suse.de>
9129 PR tree-optimization/80620
9130 PR tree-optimization/81403
9131 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
9132 info when re-using a VN table entry.
9134 2017-07-18 Richard Biener <rguenther@suse.de>
9136 PR tree-optimization/81418
9137 * tree-vect-loop.c (vectorizable_reduction): Properly compute
9138 vectype_in. Verify that with lane-reducing reduction operations
9139 we have a single def-use cycle.
9141 2017-07-17 Carl Love <cel@us.ibm.com>
9143 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
9145 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9146 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
9147 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
9148 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
9149 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
9150 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
9151 VMULOSW): New enum "unspec" values.
9152 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
9153 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
9154 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
9155 altivec_vmulosw): New patterns.
9156 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
9157 VMULOSW): Add definitions.
9158 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
9160 * config/alpha/alpha.c: Include predict.h.
9162 2017-07-17 Yury Gribov <tetra2005@gmail.com>
9164 * tree-vrp.c (compare_assert_loc): Fix comparison function
9165 to return predictable results.
9167 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
9169 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
9172 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
9173 * doc/invoke.texi (mexpand-adddi): Update text.
9175 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
9177 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
9178 that also clobbers the CC register. The old expand code is moved
9180 (*arc_clzsi2): ... here.
9181 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
9182 the CC register. The old expand code is moved to ...
9183 (arc_ctzsi2): ... here.
9185 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
9187 * config/arc/arc.opt (mindexed-loads): Use initial value
9188 TARGET_INDEXED_LOADS_DEFAULT.
9189 (mauto-modify-reg): Use initial value
9190 TARGET_AUTO_MODIFY_REG_DEFAULT.
9191 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
9192 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
9193 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
9194 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
9196 2017-07-17 Martin Liska <mliska@suse.cz>
9199 * opts.c (finish_options): Do not allow -fgnu-tm
9200 w/ -fsanitize={kernel-,}address. Say sorry.
9202 2017-07-17 Bin Cheng <bin.cheng@arm.com>
9205 * tree-loop-distribution.c (classify_partition): Only assert on
9206 numer of iterations.
9207 (merge_dep_scc_partitions): Delete prameter. Update function call.
9208 (distribute_loop): Remove code handling loop with unknown niters.
9209 (pass_loop_distribution::execute): Skip loop with unknown niters.
9211 2017-07-17 Bin Cheng <bin.cheng@arm.com>
9214 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
9215 function sort_partitions_by_post_order.
9217 2017-07-17 Bin Cheng <bin.cheng@arm.com>
9219 PR tree-optimization/81374
9220 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
9221 the max index of basic blocks, rather than number of basic blocks.
9223 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
9225 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
9227 (arc_legitimate_pic_operand_p): Likewise.
9228 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
9230 (arc_needs_pcl_p): Likewise.
9231 (arc_legitimate_pc_offset_p): Likewise.
9232 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
9233 function is also used in constrains.md.
9234 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
9235 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
9236 PLUS. Only return true/false in known cases, otherwise assert.
9237 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
9238 is already called in arc_legitimate_constant_p.
9239 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
9241 (LEGITIMATE_PIC_OPERAND_P): Use
9242 arc_raw_symbolic_reference_mentioned_p function.
9243 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
9248 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
9249 Andrew Burgess <andrew.burgess@embecosm.com>
9251 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
9252 (arc_return_address_register): New function.
9253 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
9254 (arc_handle_fndecl_attribute): Add naked attribute.
9255 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
9256 (TARGET_WARN_FUNC_RETURN): Likewise.
9257 (arc_allocate_stack_slots_for_args): New function.
9258 (arc_warn_func_return): Likewise.
9259 (machine_function): Change type fn_type.
9260 (arc_compute_function_type): Consider new naked function type,
9261 change function return type.
9262 (arc_must_save_register): Adapt to handle new
9263 arc_compute_function_type's return type.
9264 (arc_expand_prologue): Likewise.
9265 (arc_expand_epilogue): Likewise.
9266 (arc_return_address_regs): Delete.
9267 (arc_return_address_register): New function.
9268 (arc_epilogue_uses): Use above function.
9269 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
9270 (arc_function_type): Change encoding, add naked type.
9271 (ARC_INTERRUPT_P): Change to handle the new encoding.
9272 (ARC_FAST_INTERRUPT_P): Likewise.
9273 (ARC_NORMAL_P): Define.
9274 (ARC_NAKED_P): Likewise.
9275 (arc_compute_function_type): Delete prototype.
9276 * config/arc/arc.md (in_ret_delay_slot): Use
9277 arc_return_address_register function.
9278 (simple_return): Likewise.
9279 (p_return_i): Likewise.
9281 2017-07-17 Jakub Jelinek <jakub@redhat.com>
9283 PR tree-optimization/81428
9284 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
9285 can't be built for those types.
9287 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
9289 Remove stuff dead since r239246.
9291 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
9292 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
9293 (avr_inform_devices): Remove dead stuff.
9295 2017-07-17 Tamar Christina <tamar.christina@arm.com>
9297 * config/arm/arm_neon.h: Fix softp typo.
9299 2017-07-17 Jakub Jelinek <jakub@redhat.com>
9301 PR tree-optimization/81365
9302 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
9303 aggregate moves onto bb predecessor edges, make sure there are no
9304 loads that could alias the lhs in between the start of bb and the
9307 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
9310 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
9311 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
9312 [LSHIFTRT, outer_code = TRUNCATE]: Same.
9314 2017-07-17 Jakub Jelinek <jakub@redhat.com>
9316 PR tree-optimization/81396
9317 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
9318 (init_symbolic_number): Initialize it to 1.
9319 (perform_symbolic_merge): Add n_ops from both operands into the new
9321 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
9322 without base_addr as useless if they need more than one operation.
9323 (bswap_replace): Handle !bswap case for NULL base_addr.
9325 2017-07-17 Tom de Vries <tom@codesourcery.com>
9328 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
9331 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
9333 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
9334 conditional builtin define __FIX_LEON3FT_B2BST.
9336 2017-07-17 Daniel Cederman <cederman@gaisler.com>
9338 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
9339 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
9342 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
9344 PR rtl-optimization/81424
9345 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
9346 to remove potential trapping from operands if -fnon-call-exceptions.
9348 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9350 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
9351 profile_proability for scalling.
9352 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
9354 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9356 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
9358 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9360 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
9361 fixpoint arithmetics.
9363 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9365 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
9366 fixpoint arithmetics.
9368 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9370 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
9371 fixpoint arithmetics.
9373 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9375 * profile-count.h (profile_probability::from_reg_br_prob_note,
9376 profile_probability::to_reg_br_prob_note): New functions.
9377 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
9378 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
9379 * predict.c (probability_reliable_p): Update.
9380 (edge_probability_reliable_p): Update.
9381 (br_prob_note_reliable_p): Update.
9382 (invert_br_probabilities): Update.
9383 (add_reg_br_prob_note): New function.
9384 (combine_predictions_for_insn): Update.
9385 * asan.c (asan_clear_shadow): Update.
9386 * cfgbuild.c (compute_outgoing_frequencies): Update.
9387 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
9388 (update_br_prob_note): Update.
9389 (rtl_verify_edges): Update.
9390 (purge_dead_edges): Update.
9391 (fixup_reorder_chain): Update.
9392 * emit-rtl.c (try_split): Update.
9393 * ifcvt.c (cond_exec_process_insns): Update.
9394 (cond_exec_process_if_block): Update.
9395 (dead_or_predicable): Update.
9396 * internal-fn.c (expand_addsub_overflow): Update.
9397 (expand_neg_overflow): Update.
9398 (expand_mul_overflow): Update.
9399 * loop-doloop.c (doloop_modify): Update.
9400 * loop-unroll.c (compare_and_jump_seq): Update.
9401 * optabs.c (emit_cmp_and_jump_insn_1): Update.
9402 * predict.h: Update.
9403 * reorg.c (mostly_true_jump): Update.
9405 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
9406 * config/alpha/alpha.c (emit_unlikely_jump): Update.
9407 * config/arc/arc.c: (emit_unlikely_jump): Update.
9408 * config/arm/arm.c: (emit_unlikely_jump): Update.
9409 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
9410 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
9411 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
9412 (ix86_print_operand): Update.
9413 (ix86_split_fp_branch): Update.
9414 (predict_jump): Update.
9415 * config/ia64/ia64.c (ia64_print_operand): Update.
9416 * config/mmix/mmix.c (mmix_print_operand): Update.
9417 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
9418 (rs6000_expand_split_stack_prologue): Update.
9419 * config/rs6000/rs6000.c: Update.
9420 * config/s390/s390.c (s390_expand_vec_strlen): Update.
9421 (s390_expand_vec_movstr): Update.
9422 (s390_expand_cs_tdsi): Update.
9423 (s390_expand_split_stack_prologue): Update.
9424 * config/sh/sh.c (sh_print_operand): Update.
9425 (expand_cbranchsi4): Update.
9426 (expand_cbranchdi4): Update.
9427 * config/sparc/sparc.c (output_v9branch): Update.
9428 * config/spu/spu.c (get_branch_target): Update.
9429 (ea_load_store_inline): Update.
9430 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
9431 * config/tilepro/tilepro.c: Update.
9433 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
9435 * gimplify.c (mostly_copy_tree_r): Revert latest change.
9436 (gimplify_save_expr): Likewise.
9438 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
9440 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
9442 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
9444 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
9446 * timevar.def (TV_IPA_FNSUMMARY): Define.
9448 2017-07-16 Daniel Cederman <cederman@gaisler.com>
9450 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
9451 to back store errata sensitive sequence from being generated.
9452 (sqrtdf2_fix): Likewise.
9454 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
9456 * tree-ssa-threadupdate.c (compute_path_counts,
9457 update_joiner_offpath_counts): Use profile_probability.
9459 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
9462 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
9464 * config/arm/arm-c.c (arm_cpu_builtins): Define
9465 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
9467 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
9469 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9470 array entries to represent __ieee128 versions of the
9471 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
9472 scalar_extract_sig, and scalar_insert_exp built-in functions.
9473 (altivec_resolve_overloaded_builtin): Add special case handling
9474 for the __builtin_scalar_insert_exp function, as represented by
9475 the P9V_BUILTIN_VEC_VSIEDP constant.
9476 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
9477 exponent support for __ieee128 argument.
9478 (VSESQP): Add scalar extract signature support for __ieee128
9480 (VSTDCNQP): Add scalar test negative support for __ieee128
9482 (VSIEQP): Add scalar insert exponent support for __int128 argument
9483 with __ieee128 result.
9484 (VSIEQPF): Add scalar insert exponent support for __ieee128
9485 argument with __ieee128 result.
9486 (VSTDCQP): Add scalar test data class support for __ieee128
9488 (VSTDCNQP): Add overload support for scalar test negative with
9490 (VSTDCQP): Add overload support for scalar test data class
9492 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
9494 (UNSPEC_VSX_SIEXPQP): New constant.
9495 (xsxexpqp): New insn for VSX scalar extract exponent quad
9497 (xsxsigqp): New insn for VSX scalar extract significand quad
9499 (xsiexpqpf): New insn for VSX scalar insert exponent quad
9500 precision with floating point argument.
9501 (xststdcqp): New expand for VSX scalar test data class quad
9503 (xststdcnegqp): New expand for VSX scalar test negative quad
9505 (xststdcqp): New insn to match expansions for VSX scalar test data
9506 class quad precision and VSX scalar test negative quad precision.
9507 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
9508 special case operand checking to enforce that second operand of
9509 VSX scalar test data class with quad precision argument is a 7-bit
9511 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
9512 prototypes and descriptions of __ieee128 versions of
9513 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
9514 scalar_test_data_class, and scalar_test_neg built-in functions.
9516 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9518 PR tree-optimization/81162
9519 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
9520 replace a negate with an add.
9522 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
9524 * doc/invoke.texi (arm/-mcpu): Document +crypto.
9526 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
9528 * config/arm/arm-c.c (arm_cpu_builtins): Define
9529 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
9531 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
9533 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
9534 (armv8-r): Set ARM Cortex-R52 as default CPU.
9535 * config/arm/arm-tables.opt: Regenerate.
9536 * config/arm/arm-tune.md: Regenerate.
9537 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
9539 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
9540 extension for -mcpu=cortex-r52.
9542 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
9544 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
9545 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
9546 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
9547 (fp-armv8): Define it as FP_ARMv8 only.
9548 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
9549 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
9551 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
9552 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
9553 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
9554 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
9555 than TARGET_FPU_ARMV8.
9556 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
9557 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
9558 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
9560 * config/arm/neon.md (neon_vrint): Likewise.
9561 (neon_vcvt): Likewise.
9562 (neon_<fmaxmin_op><mode>): Likewise.
9563 (<fmaxmin><mode>3): Likewise.
9564 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
9565 * config/arm/predicates.md (arm_cond_move_operator): Check against
9566 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
9568 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
9570 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
9573 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9575 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
9576 loop in comment with memset.
9578 2017-07-14 Martin Liska <mliska@suse.cz>
9580 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
9581 * dwarf2out.c (is_java): Remove the function.
9582 (output_pubname): Remove usage of the function.
9583 (lower_bound_default): Remove usage of DW_LANG_Java.
9584 (gen_compile_unit_die): Likewise.
9585 * gcc.c: Remove compiler defaults for .java and .zip files.
9586 * gimple-expr.c (remove_suffix): Change as there's no longer
9587 extension than 4-letter one.
9588 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
9589 (gimplify_save_expr): Likewise.
9590 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
9591 as it's possible even for other languages than Java.
9592 * langhooks.h (struct lang_hooks): Remove Java from a comment.
9593 * lto-opts.c (lto_write_options): Remove reference to Java.
9594 * opts.c (strip_off_ending): Update file extension handling.
9595 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
9596 * tree-eh.c (lower_resx): Likewise.
9597 * tree.c (free_lang_data_in_type): Remove dead code.
9598 (find_decls_types_r): Likewise.
9599 (build_common_builtin_nodes): Remove Java from a comment.
9600 (verify_type): Remove dead code.
9601 * varasm.c (assemble_external): Remove Java from a comment.
9603 2017-07-14 Martin Liska <mliska@suse.cz>
9605 * opts.c (finish_options): Add quotes.
9606 (common_handle_option): Likewise.
9608 2017-07-14 Martin Liska <mliska@suse.cz>
9610 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
9611 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
9613 * dwarf2out.c (lower_bound_default): Do not handle
9615 (gen_compile_unit_die): Likewise.
9616 * gcc.c: Remove default extension binding for GNU Pascal.
9617 * stmt.c: Remove Pascal language from a comment.
9618 * xcoffout.c: Likewise.
9620 2017-07-13 David Malcolm <dmalcolm@redhat.com>
9623 * diagnostic-show-locus.c (fixit_cmp): New function.
9624 (layout::layout): Sort m_fixit_hints.
9625 (column_range::column_range): Assert that the values are valid.
9626 (struct char_span): New struct.
9627 (correction::overwrite): New method.
9628 (struct source_line): New struct.
9629 (line_corrections::add_hint): Add assertions. Reimplement memcpy
9630 calls in terms of classes source_line and char_span, and
9631 correction::overwrite.
9632 (selftest::test_overlapped_fixit_printing_2): New function.
9633 (selftest::diagnostic_show_locus_c_tests): Call it.
9635 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
9637 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
9638 early if there is no lhs.
9640 2017-07-13 Martin Liska <mliska@suse.cz>
9642 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
9643 (gen_reference_type_die): Likewise.
9644 * stor-layout.c: Remove Pascal-related comment.
9646 2017-07-13 Martin Liska <mliska@suse.cz>
9648 * opts.c (finish_options): Add quotes to error messages.
9649 (parse_sanitizer_options): Likewise.
9651 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
9653 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
9655 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
9657 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
9659 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
9661 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
9663 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
9665 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
9668 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
9669 provides the hardware capability bits, define the macro
9670 __BUILTIN_CPU_SUPPORTS__.
9671 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
9672 if GLIBC does not provide the hardware capability bits. Add a
9673 gcc_unreachable call if the built-in cpu function is neither
9674 __builtin_cpu_is nor __builtin_cpu_supports.
9675 (rs6000_get_function_versions_dispatcher): Change the warning
9676 that an old GLIBC is used which does not export the capability
9677 bits to be an error.
9678 * doc/extend.texi (target_clones attribute): Document the
9679 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
9680 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
9681 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
9682 the macros defined by GCC if the newer GLIBC is available.
9684 2017-07-12 Jeff Law <law@redhat.com>
9686 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
9687 remaining includes slightly.
9688 * config/riscv/riscv-builtins.c: Include profile-count.h.
9690 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
9693 * config/avr/avr.c (avr_set_current_function): In diagnostic
9694 messages: Quote keywords and (parts of) identifiers.
9695 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
9698 2017-07-12 Carl Love <cel@us.ibm.com>
9700 * config/rs6000/rs6000-c.c: Add support for built-in functions
9701 vector bool char vec_revb (vector bool char);
9702 vector bool short vec_revb (vector short char);
9703 vector bool int vec_revb (vector bool int);
9704 vector bool long long vec_revb (vector bool long long);
9705 * doc/extend.texi: Update the built-in documentation file for the
9706 new built-in functions.
9708 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9710 * config/s390/s390.md: Remove movcc splitter.
9712 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9714 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
9715 load/store on condition.
9717 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
9720 * config/avr/avr.c (avr_encode_section_info)
9721 [progmem && !TREE_READONLY]: Error if progmem object needs
9724 2017-07-11 Michael Collison <michael.collison@arm.com>
9726 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
9729 2017-07-11 Carl Love <cel@us.ibm.com>
9731 * config/rs6000/rs6000-c.c: Add support for builtins
9732 vector unsigned int vec_parity_lsbb (vector signed int);
9733 vector unsigned int vec_parity_lsbb (vector unsigned int);
9734 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
9735 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
9736 vector unsigned long long vec_parity_lsbb (vector signed long long);
9737 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
9738 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
9739 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
9740 * doc/extend.texi: Update the built-in documentation file for the
9741 new built-in functions.
9743 2017-07-11 David Malcolm <dmalcolm@redhat.com>
9745 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
9746 (layout::m_primary_loc): New field.
9747 (layout::layout): Initialize new field. Move location filtering
9748 logic from here to...
9749 (layout::maybe_add_location_range): ...this new method. Add
9750 support for filtering to just the lines already specified by other
9752 (layout::will_show_line_p): New method.
9753 (gcc_rich_location::add_location_if_nearby): New method.
9754 (selftest::test_add_location_if_nearby): New test function.
9755 (selftest::diagnostic_show_locus_c_tests): Call it.
9756 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
9759 2017-07-11 Tom de Vries <tom@codesourcery.com>
9761 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
9762 (bb_first_real_insn): New function.
9763 (nvptx_single): Add extra initialization of broadcasted condition
9766 2017-07-11 Nathan Sidwell <nathan@acm.org>
9768 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
9770 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
9772 * doc/extend.texi (AVR Function Attributes): Remove weblink to
9773 Binutils doc as TEXI will mess them up.
9774 * doc/invoke.texi (AVR Options): Same here.
9776 2017-07-11 Daniel Cederman <cederman@gaisler.com>
9778 * config/sparc/sparc.opt (mfix-ut700): New option.
9779 (mfix-gr712rc): Likewise.
9780 (sparc_fix_b2bst): New variable.
9781 * doc/invoke.texi (SPARC options): Document them.
9782 (ARM options): Fix warnings.
9783 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
9784 instructions to prevent sequences that can trigger the store-store
9785 errata for certain LEON3FT processors.
9786 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
9787 (sparc_option_override): Set sparc_fix_b2bst appropriately.
9788 * config/sparc/sparc.md (fix_b2bst): New attribute.
9789 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
9791 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
9794 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
9796 (*rsqrtsf2_sse): Ditto.
9798 (div<mode>3): Macroize insn from divdf3 and divsf3
9799 using MODEF mode iterator.
9801 2017-07-10 Martin Sebor <msebor@redhat.com>
9803 PR tree-optimization/80397
9804 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
9805 instead of testing for equality to INTEGER_TYPE.
9807 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
9809 * config.gcc: Remove uclibc from arc target spec.
9811 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
9813 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
9815 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
9818 * lto-wrapper.c (remove_option): New function.
9819 (merge_and_complain): Merge PIC/PIE options more realistically.
9821 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
9823 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
9827 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
9828 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
9829 * config.in: Regenerate.
9830 * configure: Regenerate.
9831 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
9832 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
9833 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
9834 (TARGET_GASISR_PROLOGUES): ...target mask.
9835 * common/config/avr/avr-common.c
9836 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
9837 Set -mgas-isr-prologues.
9838 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
9839 INSERT_PASS_BEFORE for it.
9840 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
9841 * config/avr/avr.c (avr_option_override)
9842 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
9843 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
9844 (avr_attribute_table) <no_gccisr>: Add new function attribute.
9845 (avr_set_current_function) <is_no_gccisr>: Init machine field.
9846 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
9848 (make_avr_pass_pre_proep): New function.
9849 (emit_push_sfr) <treg>: Add argument to function and use it
9851 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
9852 and set machine->gasisr.yes.
9853 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
9854 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
9855 __gcc_isr.n_pushed to .L__stack_usage.
9856 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
9857 (avr_asm_final_postscan_insn): ...this new static function.
9858 * config/avr/avr.h (machine_function)
9859 <is_no_gccisr, use_L__stack_usage>: New fields.
9860 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
9861 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
9862 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
9863 (gasisr, *gasisr): New expander and insn.
9864 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
9865 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
9866 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
9868 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
9870 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
9873 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
9875 Move jump-tables out of .text again.
9878 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
9879 (ASM_OUTPUT_ADDR_VEC): New function.
9880 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
9881 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
9882 INSN_ADDRESSes as asm comment.
9883 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
9884 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
9885 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
9886 * config/avr/avr.md (*tablejump): Adjust comment.
9887 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
9888 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
9890 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
9891 (avr_output_addr_vec): New proto.
9892 (avr_log_t) <insn_addresses>: New field.
9894 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
9897 * config/i386/i386.c (ix86_function_arg_advance): Set
9898 outgoing_args_on_stack to true if there are outgoing arguments
9900 (ix86_function_arg): Likewise.
9901 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
9902 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
9903 * config/i386/i386.h (machine_function): Add
9904 outgoing_args_on_stack.
9906 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
9908 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
9910 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
9912 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
9914 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
9915 (REAL_H): Remove $(MACHMODE_H).
9916 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
9918 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
9919 $(MACHMODE_H) and double-int.h.
9920 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
9922 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
9925 2017-07-07 Andrew Pinski <apinski@cavium.com>
9927 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
9928 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
9930 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
9932 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
9933 Add warning if GCC was not configured to link against a GLIBC that
9934 exports the hardware capability bits.
9935 (make_resolver_func): Make resolver function private and not a
9936 COMDAT function. Create the name with clone_function_name instead
9937 of make_unique_name.
9940 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
9941 correct operand in doing the split.
9943 2017-07-07 Carl Love <cel@us.ibm.com>
9945 * config/rs6000/rs6000-c: Add support for built-in function
9946 vector unsigned short vec_pack_to_short_fp32 (vector float,
9948 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
9949 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
9950 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
9951 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
9952 (convert_4f32_8i16): Add define_expand.
9953 * doc/extend.texi: Update the built-in documentation file for the
9954 new built-in function.
9956 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9958 * config/sparc/m8.md: New file.
9959 * config/sparc/sparc.md: Include m8.md.
9961 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9963 * config/sparc/sparc.opt: New option -mvis4b.
9964 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
9965 (sparc_option_override): Handle VIS4B.
9966 (enum sparc_builtins): Define
9967 SPARC_BUILTIN_DICTUNPACK{8,16,32},
9968 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
9969 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
9970 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
9971 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
9972 (check_constant_argument): New function.
9973 (sparc_vis_init_builtins): Define builtins
9974 __builtin_vis_dictunpack{8,16,32},
9975 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
9976 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
9977 __builtin_vis_fpcmpde{8,16,32}shl and
9978 __builtin_vis_fpcmpur{8,16,32}shl.
9979 (sparc_expand_builtin): Check that the constant operands to
9980 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
9981 constant and in range.
9982 * config/sparc/sparc-c.c (sparc_target_macros): Handle
9984 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
9985 (SPARC_IMM5_P): Likewise.
9986 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
9987 (enabled): Handle vis4b.
9988 (UNSPEC_DICTUNPACK): New unspec.
9989 (UNSPEC_FPCMPSHL): Likewise.
9990 (UNSPEC_FPUCMPSHL): Likewise.
9991 (UNSPEC_FPCMPDESHL): Likewise.
9992 (UNSPEC_FPCMPURSHL): Likewise.
9993 (cpu_feature): New CPU feature `vis4b'.
9994 (dictunpack{8,16,32}): New insns.
9995 (FPCSMODE): New mode iterator.
9996 (fpcscond): New code iterator.
9997 (fpcsucond): Likewise.
9998 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
9999 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
10000 (fpcmpde{8,16,32}{si,di}shl): Likewise.
10001 (fpcmpur{8,16,32}{si,di}shl): Likewise.
10002 * config/sparc/constraints.md: Define constraints `q' for unsigned
10003 2-bit integer constants and `t' for unsigned 5-bit integer
10005 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
10007 (imm5_operand_dictunpack16): Likewise.
10008 (imm5_operand_dictunpack32): Likewise.
10009 (imm2_operand): Likewise.
10010 * doc/invoke.texi (SPARC Options): Document -mvis4b.
10011 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
10012 ditunpack* and fpcmp*shl builtins.
10014 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10016 * config.gcc: Handle m8 in --with-{cpu,tune} options.
10017 * config.in: Add HAVE_AS_SPARC6 define.
10018 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
10020 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
10022 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
10023 (CPP_CPU_SPEC): Handle m8.
10024 (ASM_CPU_SPEC): Likewise.
10025 * config/sparc/sparc-opts.h (enum processor_type): Add
10027 * config/sparc/sparc.c (m8_costs): New struct.
10028 (sparc_option_override): Handle TARGET_CPU_m8.
10029 (sparc32_initialize_trampoline): Likewise.
10030 (sparc64_initialize_trampoline): Likewise.
10031 (sparc_issue_rate): Likewise.
10032 (sparc_register_move_cost): Likewise.
10033 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
10034 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
10035 (ASM_CPU64_DEFAULT_SPEC): Likewise.
10036 (CPP_CPU_SPEC): Handle M8.
10037 (ASM_CPU_SPEC): Likewise.
10038 (AS_M8_FLAG): Define.
10039 * config/sparc/sparc.md: Add m8 to the cpu attribute.
10040 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
10041 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
10043 * configure: Regenerate.
10044 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
10047 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10049 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
10051 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
10052 ("*movdi_insn_sp32"): Do not set v3pipe.
10053 ("*movsi_insn"): Likewise.
10054 ("*movdi_insn_sp64"): Likewise.
10055 ("*movsf_insn"): Likewise.
10056 ("*movdf_insn_sp32"): Likewise.
10057 ("*movdf_insn_sp64"): Likewise.
10058 ("*zero_extendsidi2_insn_sp64"): Likewise.
10059 ("*sign_extendsidi2_insn"): Likewise.
10060 ("*mov<VM32:mode>_insn"): Likewise.
10061 ("*mov<VM64:mode>_insn_sp64"): Likewise.
10062 ("*mov<VM64:mode>_insn_sp32"): Likewise.
10063 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
10064 ("<vlop:code><VL:mode>3"): Likewise.
10065 ("*not_<vlop:code><VL:mode>3"): Likewise.
10066 ("*nand<VL:mode>_vis"): Likewise.
10067 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
10068 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
10069 ("one_cmpl<VL:mode>2"): Likewise.
10070 ("faligndata<VM64:mode>_vis"): Likewise.
10071 ("alignaddrsi_vis"): Likewise.
10072 ("alignaddrdi_vis"): Likweise.
10073 ("alignaddrlsi_vis"): Likewise.
10074 ("alignaddrldi_vis"): Likewise.
10075 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
10076 ("bmaskdi_vis"): Likewise.
10077 ("bmasksi_vis"): Likewise.
10078 ("bshuffle<VM64:mode>_vis"): Likewise.
10079 ("cmask8<P:mode>_vis"): Likewise.
10080 ("cmask16<P:mode>_vis"): Likewise.
10081 ("cmask32<P:mode>_vis"): Likewise.
10082 ("pdistn<P:mode>_vis"): Likewise.
10083 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
10085 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10087 * config/sparc/sparc.md ("subtype"): New insn attribute.
10088 ("*wrgsr_sp64"): Set insn subtype.
10089 ("*rdgsr_sp64"): Likewise.
10090 ("alignaddrsi_vis"): Likewise.
10091 ("alignaddrdi_vis"): Likewise.
10092 ("alignaddrlsi_vis"): Likewise.
10093 ("alignaddrldi_vis"): Likewise.
10094 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
10095 ("fexpand_vis"): Likewise.
10096 ("fpmerge_vis"): Likewise.
10097 ("faligndata<VM64:mode>_vis"): Likewise.
10098 ("bshuffle<VM64:mode>_vis"): Likewise.
10099 ("cmask8<P:mode>_vis"): Likewise.
10100 ("cmask16<P:mode>_vis"): Likewise.
10101 ("cmask32<P:mode>_vis"): Likewise.
10102 ("fchksm16_vis"): Likewise.
10103 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
10104 ("fmean16_vis"): Likewise.
10105 ("fp<plusminus_insn>64_vis"): Likewise.
10106 ("<plusminus_insn>v8qi3"): Likewise.
10107 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
10108 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
10109 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
10110 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
10111 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
10112 ("*movqi_insn"): Likewise.
10113 ("*movhi_insn"): Likewise.
10114 ("*movsi_insn"): Likewise.
10115 ("movsi_pic_gotdata_op"): Likewise.
10116 ("*movdi_insn_sp32"): Likewise.
10117 ("*movdi_insn_sp64"): Likewise.
10118 ("movdi_pic_gotdata_op"): Likewise.
10119 ("*movsf_insn"): Likewise.
10120 ("*movdf_insn_sp32"): Likewise.
10121 ("*movdf_insn_sp64"): Likewise.
10122 ("*zero_extendhisi2_insn"): Likewise.
10123 ("*zero_extendqihi2_insn"): Likewise.
10124 ("*zero_extendqisi2_insn"): Likewise.
10125 ("*zero_extendqidi2_insn"): Likewise.
10126 ("*zero_extendhidi2_insn"): Likewise.
10127 ("*zero_extendsidi2_insn_sp64"): Likewise.
10128 ("ldfsr"): Likewise.
10129 ("prefetch_64"): Likewise.
10130 ("prefetch_32"): Likewise.
10131 ("tie_ld32"): Likewise.
10132 ("tie_ld64"): Likewise.
10133 ("*tldo_ldub_sp32"): Likewise.
10134 ("*tldo_ldub1_sp32"): Likewise.
10135 ("*tldo_ldub2_sp32"): Likewise.
10136 ("*tldo_ldub_sp64"): Likewise.
10137 ("*tldo_ldub1_sp64"): Likewise.
10138 ("*tldo_ldub2_sp64"): Likewise.
10139 ("*tldo_ldub3_sp64"): Likewise.
10140 ("*tldo_lduh_sp32"): Likewise.
10141 ("*tldo_lduh1_sp32"): Likewise.
10142 ("*tldo_lduh_sp64"): Likewise.
10143 ("*tldo_lduh1_sp64"): Likewise.
10144 ("*tldo_lduh2_sp64"): Likewise.
10145 ("*tldo_lduw_sp32"): Likewise.
10146 ("*tldo_lduw_sp64"): Likewise.
10147 ("*tldo_lduw1_sp64"): Likewise.
10148 ("*tldo_ldx_sp64"): Likewise.
10149 ("*mov<VM32:mode>_insn"): Likewise.
10150 ("*mov<VM64:mode>_insn_sp64"): Likewise.
10151 ("*mov<VM64:mode>_insn_sp32"): Likewise.
10153 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10155 * config/sparc/sparc.md ("type"): New insn type viscmp.
10156 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
10158 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
10159 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
10160 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
10161 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
10163 ("n7_vis_logical_11cycle"): Likewise.
10164 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
10165 * config/sparc/niagara2.md ("niag3_vis": Likewise.
10166 * config/sparc/niagara.md ("niag_vis"): Likewise.
10167 * config/sparc/ultra3.md ("us3_fga"): Likewise.
10168 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
10170 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10172 * config/sparc/sparc.md: New instruction type `bmask'.
10173 (bmaskdi_vis): Use the `bmask' type.
10174 (bmasksi_vis): Likewise.
10175 * config/sparc/ultra3.md (us3_array): Likewise.
10176 * config/sparc/niagara7.md (n7_array): Likewise.
10177 * config/sparc/niagara4.md (n4_array): Likewise.
10178 * config/sparc/niagara2.md (niag2_vis): Likewise.
10179 (niag3_vis): Likewise.
10180 * config/sparc/niagara.md (niag_vis): Likewise.
10182 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
10184 * ipa-comdats.c: Remove optimize check from gate.
10185 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
10186 for functions not optimized.
10187 (ipa_fn_summary_read): Skip optimize check.
10188 (ipa_fn_summary_write): Likewise.
10189 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
10191 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
10193 (can_inline_edge_p): Check flag_pcc_struct_return for match.
10194 (check_callers): Give up on caller which is not optimized.
10195 (inline_small_functions): Likewise.
10196 (ipa_inline): Do not give up when not optimizing.
10197 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
10198 away unoptimizes cdtors.
10199 (whole_program_function_and_variable_visibility): Do
10200 ipa_discover_readonly_nonaddressable_vars in LTO mode.
10201 * ipa.c (process_references): Do not check optimize.
10202 (symbol_table::remove_unreachable_nodes): Update optimize check.
10203 (set_writeonly_bit): Update optimize check.
10204 (pass_ipa_cdtor_merge::gate): Do not check optimize.
10205 (pass_ipa_single_use::gate): Remove.
10207 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
10209 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
10210 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
10211 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
10212 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
10213 permute_load, permute_store, adjust_extract, adjust_splat,
10214 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
10215 replace_swap_with_copy, dump_swap_insn_table,
10216 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
10217 recombine_lvx_pattern, recombine_stvx_pattern,
10218 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
10219 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
10220 to file rs6000-p8swap.c.
10221 * config/rs6000/rs6000-p8swap.c: New file.
10222 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
10223 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
10224 and rs6000*-*-* targets.
10226 2017-07-06 David Malcolm <dmalcolm@redhat.com>
10228 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
10230 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
10232 * lto-wrapper.c (merge_and_complain): Do not merge
10233 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
10234 fsigned_zeros, ftrapping_math, fwrapv.
10235 (append_compiler_options): Do not track these options.
10236 (append_linker_options): Likewie
10238 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
10240 * cgraphunit.c (cgraph_node::finalize_function): When
10241 !flag_toplevel_reorde set no_reorder flag.
10242 (varpool_node::finalize_decl): Likewise.
10243 (symbol_table::compile): Drop no toplevel reorder path.
10245 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
10247 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
10248 edges; zero probability is not better than uninitialized.
10250 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
10252 * asan.h (asan_sanitize_allocas_p): Declare.
10253 * asan.c (asan_sanitize_allocas_p): New function.
10254 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
10255 (handle_builtin_alloca): Likewise.
10256 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
10257 if !asan_sanitize_allocas_p.
10258 * params.def (asan-instrument-allocas): Add new option.
10259 * params.h (ASAN_PROTECT_ALLOCAS): Define.
10260 * opts.c (common_handle_option): Disable allocas sanitization for
10263 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
10265 * asan.c: Include gimple-fold.h.
10266 (get_last_alloca_addr): New function.
10267 (handle_builtin_stackrestore): Likewise.
10268 (handle_builtin_alloca): Likewise.
10269 (asan_emit_allocas_unpoison): Likewise.
10270 (get_mem_refs_of_builtin_call): Add new parameter, remove const
10271 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
10272 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
10273 (instrument_builtin_call): Pass gimple iterator to
10274 get_mem_refs_of_builtin_call.
10275 (last_alloca_addr): New global.
10276 * asan.h (asan_emit_allocas_unpoison): Declare.
10277 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
10278 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
10279 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
10280 if function calls alloca.
10281 * gimple-fold.c (replace_call_with_value): Remove static keyword.
10282 * gimple-fold.h (replace_call_with_value): Declare.
10283 * internal-fn.c: Include asan.h.
10284 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
10285 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
10287 2017-07-06 David Malcolm <dmalcolm@redhat.com>
10289 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
10290 (C_SELFTEST_FLAGS): New.
10291 (CPP_SELFTEST_FLAGS): New.
10292 (SELFTEST_DEPS): New, from deps of s-selftest.
10293 (C_SELFTEST_DEPS): New, from deps of s-selftest.
10294 (CPP_SELFTEST_DEPS): New.
10295 (selftest): Add dependency on s-selftest-c++.
10296 (s-selftest): Rename to...
10297 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
10298 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
10299 than SELFTEST_FLAGS.
10300 (selftest-gdb): Rename to...
10301 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
10303 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
10304 (selftest-valgrind): Rename to...
10305 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
10307 (selftest-valgrind): Reintroduce as an alias for
10308 selftest-c-valgrind.
10309 (s-selftest-c++): New.
10310 (selftest-c++-gdb): New.
10311 (selftest-c++-valgrind): New.
10313 2017-07-06 Olivier Hainque <hainque@adacore.com>
10315 * gcc.c (process_command): When deciding if undefined variables
10316 should be ignored when processing specs, accept "gcc -v" as well.
10318 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
10320 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
10321 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
10323 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
10325 * config/arm/arm-cpus.in (armv8-r): Add new entry.
10326 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
10327 * config/arm/arm-tables.opt: Regenerate.
10328 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
10330 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
10332 2017-07-06 Carl Love <cel@us.ibm.com>
10334 * ChangeLog: Clean up from mid air collision
10336 2017-07-06 Carl Love <cel@us.ibm.com>
10338 * config/rs6000/rs6000-c.c: Add support for built-in functions
10339 vector signed int vec_subc (vector signed int, vector signed int);
10340 vector signed __int128 vec_subc (vector signed __int128,
10341 vector signed __int128);
10342 vector unsigned __int128 vec_subc (vector unsigned __int128,
10343 vector unsigned __int128);
10344 vector signed int vec_sube (vector signed int, vector signed int,
10345 vector signed int);
10346 vector unsigned int vec_sube (vector unsigned int,
10347 vector unsigned int,
10348 vector unsigned int);
10349 vector signed __int128 vec_sube (vector signed __int128,
10350 vector signed __int128,
10351 vector signed__int128);
10352 vector unsigned __int128 vec_sube (vector unsigned __int128,
10353 vector unsigned __int128,
10354 vector unsigned __int128);
10355 vector signed int vec_subec (vector signed int, vector signed int,
10356 vector signed int);
10357 vector unsigned int vec_subec (vector unsigned int,
10358 vector unsigned int,
10359 vector unsigned int);
10360 vector signed __int128 vec_subec (vector signed __int128,
10361 vector signed __int128,
10362 vector signed__int128);
10363 vector unsigned __int128 vec_subec (vector unsigned __int128,
10364 vector unsigned __int128,
10365 vector unsigned __int128);
10366 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
10367 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
10368 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
10369 BU_ALTIVEC_OVERLOAD_X definitions.
10370 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
10371 * doc/extend.texi: Update the built-in documentation file for the new
10372 built-in functions.
10374 2017-07-06 David Malcolm <dmalcolm@redhat.com>
10377 * diagnostic-show-locus.c (layout::layout): Use start and finish
10378 spelling location for the start and finish of each range.
10379 * genmatch.c (linemap_client_expand_location_to_spelling_point):
10380 Add unused aspect param.
10381 * input.c (expand_location_1): Add "aspect" param, and use it
10382 to access the correct part of the location.
10383 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
10385 (expand_location_to_spelling_point): Likewise.
10386 (linemap_client_expand_location_to_spelling_point): Add "aspect"
10387 param, and pass it to expand_location_1.
10389 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
10391 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
10392 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
10393 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
10394 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
10395 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
10396 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
10397 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
10398 _mm_maskz_getmant_ss): New intrinsics.
10399 (__builtin_ia32_getexpss128_mask): Changed to ...
10400 __builtin_ia32_getexpss128_round ... this.
10401 (__builtin_ia32_getexpsd128_mask): Changed to ...
10402 __builtin_ia32_getexpsd128_round ... this.
10403 * config/i386/i386-builtin-types.def
10404 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
10405 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
10406 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
10407 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
10408 __builtin_ia32_getmantss_mask_round): New builtins.
10409 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
10410 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
10411 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
10412 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
10413 * config/i386/sse.md
10414 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
10415 avx512f_sgetexp<mode><mask_scalar_name>
10416 <round_saeonly_scalar_name> ... this.
10417 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
10418 %0, %1, %2<round_saeonly_op3>}): Changed to ...
10419 vgetexp<ssescalarmodesuffix>
10420 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
10421 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
10422 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
10423 avx512f_vgetmant<mode><mask_scalar_name>
10424 <round_saeonly_scalar_name> ... this.
10425 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
10426 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
10427 vgetmant<ssescalarmodesuffix>
10428 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
10429 %0<mask_scalar_operand4>, %1, %2
10430 <round_saeonly_scalar_mask_op4>, %3} ... this.
10431 * config/i386/subst.md (mask_scalar_operand4,
10432 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
10433 round_saeonly_scalar_nimm_predicate): New subst attributes.
10435 2017-07-06 Julia Koval <julia.koval@intel.com>
10437 * config/i386/i386.c (ix86_erase_embedded_rounding):
10438 Remove code for old rounding pattern.
10440 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
10442 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
10444 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
10446 * doc/sourcebuild.texi (Test Directives, Variants of
10447 dg-require-support): Add documentation for dg-require-stack-check.
10449 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
10451 * config/i386/subst.md (mask_scalar, round_scalar,
10452 round_saeonly_scalar): New meta-templates.
10453 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
10454 round_scalar_mask_operand3, round_scalar_mask_op3,
10455 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
10456 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
10457 round_saeonly_scalar_constraint,
10458 round_saeonly_scalar_prefix): New subst attribute.
10459 * config/i386/sse.md
10460 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
10461 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
10462 <round_scalar_name> ... this.
10463 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
10464 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
10465 <round_scalar_name> ... this.
10466 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
10467 <sse>_vm<code><mode>3<mask_scalar_name>
10468 <round_saeonly_scalar_name> ... this.
10469 (v<plusminus_mnemonic><ssescalarmodesuffix>
10470 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
10471 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
10472 v<plusminus_mnemonic><ssescalarmodesuffix>
10473 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
10474 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
10475 (v<multdiv_mnemonic><ssescalarmodesuffix>
10476 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
10477 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
10478 v<multdiv_mnemonic><ssescalarmodesuffix>
10479 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
10480 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
10481 (v<maxmin_float><ssescalarmodesuffix>
10482 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
10483 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
10484 v<maxmin_float><ssescalarmodesuffix>
10485 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
10486 %0<mask_scalar_operand3>, %1, %<iptr>2
10487 <round_saeonly_scalar_mask_op3>} ... this.
10489 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
10491 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
10492 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
10494 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
10495 Alan Hayward <alan.hayward@arm.com>
10496 David Sherwood <david.sherwood@arm.com>
10498 * combine.c (simplify_if_then_else): Remove "enum" before
10500 * compare-elim.c (can_eliminate_compare): Likewise.
10501 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
10503 (aarch64_lookup_simd_builtin_type): Likewise.
10504 (aarch64_simd_builtin_type): Likewise.
10505 (aarch64_init_simd_builtin_types): Likewise.
10506 (aarch64_simd_expand_args): Likewise.
10507 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
10509 (aarch64_reverse_mask): Likewise.
10510 (aarch64_simd_emit_reg_reg_move): Likewise.
10511 (aarch64_gen_adjusted_ldpstp): Likewise.
10512 (aarch64_ccmp_mode_to_code): Likewise.
10513 (aarch64_operands_ok_for_ldpstp): Likewise.
10514 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10515 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
10517 (aarch64_min_divisions_for_recip_mul): Likewise.
10518 (aarch64_reassociation_width): Likewise.
10519 (aarch64_get_condition_code_1): Likewise.
10520 (aarch64_simd_emit_reg_reg_move): Likewise.
10521 (aarch64_simd_attr_length_rglist): Likewise.
10522 (aarch64_reverse_mask): Likewise.
10523 (aarch64_operands_ok_for_ldpstp): Likewise.
10524 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10525 (aarch64_gen_adjusted_ldpstp): Likewise.
10526 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
10528 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
10529 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
10530 (arm_lookup_simd_builtin_type): Likewise.
10531 (arm_simd_builtin_type): Likewise.
10532 (arm_init_simd_builtin_types): Likewise.
10533 (arm_expand_builtin_args): Likewise.
10534 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
10535 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
10536 (ft32_setup_incoming_varargs): Likewise.
10537 (ft32_function_arg): Likewise.
10538 (ft32_function_arg_advance): Likewise.
10539 (ft32_pass_by_reference): Likewise.
10540 (ft32_arg_partial_bytes): Likewise.
10541 (ft32_valid_pointer_mode): Likewise.
10542 (ft32_addr_space_pointer_mode): Likewise.
10543 (ft32_addr_space_legitimate_address_p): Likewise.
10544 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
10546 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
10547 (ix86_emit_outlined_ms2sysv_restore): Likewise.
10548 (iamcu_alignment): Likewise.
10549 (canonicalize_vector_int_perm): Likewise.
10550 (ix86_noce_conversion_profitable_p): Likewise.
10551 (ix86_mpx_bound_mode): Likewise.
10552 (ix86_operands_ok_for_move_multiple): Likewise.
10553 * config/microblaze/microblaze-protos.h
10554 (microblaze_expand_conditional_branch_reg): Likewise.
10555 * config/microblaze/microblaze.c
10556 (microblaze_expand_conditional_branch_reg): Likewise.
10557 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
10559 (rs6000_reassociation_width): Likewise.
10560 (rs6000_invalid_binary_op): Likewise.
10561 (fusion_p9_p): Likewise.
10562 (emit_fusion_p9_load): Likewise.
10563 (emit_fusion_p9_store): Likewise.
10564 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
10566 (riscv_hard_regno_mode_ok_p): Likewise.
10567 (riscv_address_insns): Likewise.
10568 (riscv_split_symbol): Likewise.
10569 (riscv_legitimize_move): Likewise.
10570 (riscv_function_value): Likewise.
10571 (riscv_hard_regno_nregs): Likewise.
10572 (riscv_expand_builtin): Likewise.
10573 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
10574 (riscv_build_integer): Likewise.
10575 (riscv_split_integer): Likewise.
10576 (riscv_legitimate_constant_p): Likewise.
10577 (riscv_cannot_force_const_mem): Likewise.
10578 (riscv_regno_mode_ok_for_base_p): Likewise.
10579 (riscv_valid_base_register_p): Likewise.
10580 (riscv_valid_offset_p): Likewise.
10581 (riscv_valid_lo_sum_p): Likewise.
10582 (riscv_classify_address): Likewise.
10583 (riscv_legitimate_address_p): Likewise.
10584 (riscv_address_insns): Likewise.
10585 (riscv_load_store_insns): Likewise.
10586 (riscv_force_binary): Likewise.
10587 (riscv_split_symbol): Likewise.
10588 (riscv_force_address): Likewise.
10589 (riscv_legitimize_address): Likewise.
10590 (riscv_move_integer): Likewise.
10591 (riscv_legitimize_const_move): Likewise.
10592 (riscv_legitimize_move): Likewise.
10593 (riscv_address_cost): Likewise.
10594 (riscv_subword): Likewise.
10595 (riscv_output_move): Likewise.
10596 (riscv_canonicalize_int_order_test): Likewise.
10597 (riscv_emit_int_order_test): Likewise.
10598 (riscv_function_arg_boundary): Likewise.
10599 (riscv_pass_mode_in_fpr_p): Likewise.
10600 (riscv_pass_fpr_single): Likewise.
10601 (riscv_pass_fpr_pair): Likewise.
10602 (riscv_get_arg_info): Likewise.
10603 (riscv_function_arg): Likewise.
10604 (riscv_function_arg_advance): Likewise.
10605 (riscv_arg_partial_bytes): Likewise.
10606 (riscv_function_value): Likewise.
10607 (riscv_pass_by_reference): Likewise.
10608 (riscv_setup_incoming_varargs): Likewise.
10609 (riscv_print_operand): Likewise.
10610 (riscv_elf_select_rtx_section): Likewise.
10611 (riscv_save_restore_reg): Likewise.
10612 (riscv_for_each_saved_reg): Likewise.
10613 (riscv_register_move_cost): Likewise.
10614 (riscv_hard_regno_mode_ok_p): Likewise.
10615 (riscv_hard_regno_nregs): Likewise.
10616 (riscv_class_max_nregs): Likewise.
10617 (riscv_memory_move_cost): Likewise.
10618 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
10619 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
10620 (rl78_addr_space_address_mode): Likewise.
10621 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10623 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
10624 (rs6000_reassociation_width): Likewise.
10625 (rs6000_invalid_binary_op): Likewise.
10626 (fusion_p9_p): Likewise.
10627 (emit_fusion_p9_load): Likewise.
10628 (emit_fusion_p9_store): Likewise.
10629 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
10630 (ok_for_simple_move_operands): Likewise.
10631 (ok_for_simple_move_strict_operands): Likewise.
10632 (ok_for_simple_arith_logic_operands): Likewise.
10633 (visium_legitimize_reload_address): Likewise.
10634 (visium_select_cc_mode): Likewise.
10635 (output_cbranch): Likewise.
10636 (visium_split_double_move): Likewise.
10637 (visium_expand_copysign): Likewise.
10638 (visium_expand_int_cstore): Likewise.
10639 (visium_expand_fp_cstore): Likewise.
10640 * config/visium/visium.c (visium_pass_by_reference): Likewise.
10641 (visium_function_arg): Likewise.
10642 (visium_function_arg_advance): Likewise.
10643 (visium_libcall_value): Likewise.
10644 (visium_setup_incoming_varargs): Likewise.
10645 (visium_legitimate_constant_p): Likewise.
10646 (visium_legitimate_address_p): Likewise.
10647 (visium_legitimize_address): Likewise.
10648 (visium_secondary_reload): Likewise.
10649 (visium_register_move_cost): Likewise.
10650 (visium_memory_move_cost): Likewise.
10651 (prepare_move_operands): Likewise.
10652 (ok_for_simple_move_operands): Likewise.
10653 (ok_for_simple_move_strict_operands): Likewise.
10654 (ok_for_simple_arith_logic_operands): Likewise.
10655 (visium_function_value_1): Likewise.
10656 (rtx_ok_for_offset_p): Likewise.
10657 (visium_legitimize_reload_address): Likewise.
10658 (visium_split_double_move): Likewise.
10659 (visium_expand_copysign): Likewise.
10660 (visium_expand_int_cstore): Likewise.
10661 (visium_expand_fp_cstore): Likewise.
10662 (visium_split_cstore): Likewise.
10663 (visium_select_cc_mode): Likewise.
10664 (visium_split_cbranch): Likewise.
10665 (output_cbranch): Likewise.
10666 (visium_print_operand_address): Likewise.
10667 * expmed.c (flip_storage_order): Likewise.
10668 * expmed.h (emit_cstore): Likewise.
10669 (flip_storage_order): Likewise.
10670 * genrecog.c (validate_pattern): Likewise.
10671 * hsa-gen.c (gen_hsa_addr): Likewise.
10672 * internal-fn.c (expand_arith_overflow): Likewise.
10673 * ira-color.c (allocno_copy_cost_saving): Likewise.
10674 * lra-assigns.c (find_hard_regno_for_1): Likewise.
10675 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
10676 (process_invariant_for_inheritance): Likewise.
10677 * lra-eliminations.c (move_plus_up): Likewise.
10678 * omp-low.c (lower_oacc_reductions): Likewise.
10679 * simplify-rtx.c (simplify_subreg): Likewise.
10680 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
10681 (TARGET_CHKP_BOUND_MODE): Likewise..
10682 * targhooks.c (default_chkp_bound_mode): Likewise.
10683 (default_setup_incoming_vararg_bounds): Likewise.
10684 * targhooks.h (default_chkp_bound_mode): Likewise.
10685 (default_setup_incoming_vararg_bounds): Likewise.
10686 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
10687 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
10688 (have_whole_vector_shift): Likewise.
10689 * tree-vect-stmts.c (vectorizable_load): Likewise.
10690 * doc/tm.texi: Regenerate.
10692 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
10694 Graceful degrade if Binutils PR21472 is not available.
10697 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
10698 .rodata in flash test fails.
10699 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
10700 * confgure: Regenerate.
10701 * config.in: Regenerate.
10702 * config/avr/avr.c (avr_asm_named_section)
10703 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
10704 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
10705 (avr_asm_init_sections): Same.
10707 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
10709 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
10710 (fma<VH:mode>4_intrinsic): Likewise.
10711 (*fmsub<VCVTF:mode>4): Likewise.
10712 (*fmsub<VH:mode>4_intrinsic): Likewise.
10714 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
10717 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
10718 Don't depend on "optimize > 0".
10719 (out_movhi_r_mr, out_movqi_mr_r): Same.
10720 (out_movhi_mr_r, out_movqi_r_mr): Same.
10721 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
10722 io_address_operand on "optimize > 0".
10724 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10726 * tree-loop-distribution.c: Add general explanantion on the pass.
10727 (generate_loops_for_partition): Mark distributed loop.
10728 (pg_add_dependence_edges): New parameter. Handle alias data
10729 dependence specially and record it in the parameter if asked.
10730 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
10731 (init_partition_graph_vertices, add_partition_graph_edge): New.
10732 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
10733 (free_partition_graph_vdata, build_partition_graph): New.
10734 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
10735 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
10736 (data_ref_segment_size, latch_dominated_by_data_ref): New.
10737 (compute_alias_check_pairs, version_loop_by_alias_check): New.
10738 (version_for_distribution_p, finalize_partitions): New.
10739 (distribute_loop): Handle alias data dependence specially. Factor
10740 out loop fusion code as functions and call these functions.
10742 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10744 * tree-loop-distribution.c (classify_partition): New parameter and
10745 better handle reduction statement.
10746 (rdg_build_partitions): Revise comment.
10747 (distribute_loop): Compute statements in all partitions and pass it
10748 to classify_partition.
10750 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10752 * tree-loop-distribution.c (enum partition_type): New.
10753 (struct partition): New field type.
10754 (partition_merge_into): Add parameter. Update partition type.
10755 (data_dep_in_cycle_p, update_type_for_merge): New functions.
10756 (build_rdg_partition_for_vertex): Compute partition type.
10757 (rdg_build_partitions): Dump partition type.
10758 (distribute_loop): Update calls to partition_merge_into.
10760 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10762 * tree-loop-distribution.c (struct ddr_hasher): New.
10763 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
10765 (classify_partition): Call get_data_dependence.
10766 (pg_add_dependence_edges): Ditto.
10767 (distribute_loop): Release data dependence hash table.
10769 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10771 * tree-loop-distribution.c (ref_base_address): Delete.
10772 (similar_memory_accesses): Rename ...
10773 (share_memory_accesses): ... to this. Check if partitions access
10774 the same memory reference.
10775 (distribute_loop): Call share_memory_accesses.
10777 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10779 * tree-loop-distribution.c (struct partition): New field recording
10780 its data reference.
10781 (partition_alloc, partition_free): Init and release data refs.
10782 (partition_merge_into): Merge data refs.
10783 (build_rdg_partition_for_vertex): Collect data refs for partition.
10784 (pg_add_dependence_edges): Change parameters from vector to bitmap.
10786 (distribute_loop): Remve data refs from vertice data of partition
10789 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10791 * tree-loop-distribution.c (params.h): Include header file.
10792 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
10793 (datarefs_vec): New global var.
10794 (create_rdg_vertices): Use datarefs_vec directly.
10795 (free_rdg): Don't free data references.
10796 (build_rdg): Update use. Don't free data references.
10797 (distribute_loop): Compute global variable for data references.
10798 Bail out if there are too many data references.
10800 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10802 * tree-loop-distribution.c (loop_nest): New global var.
10803 (build_rdg): Use loop directly, rather than loop nest.
10804 (pg_add_dependence_edges): Remove loop nest parameter. Use global
10806 (distribute_loop): Compute global variable loop nest. Update use.
10808 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10810 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
10811 (partition_merge_into): New parameter. Dump reason for fusion.
10812 (distribute_loop): Update use of partition_merge_into.
10814 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10816 * tree-loop-distribution.c (bb_top_order_index): New.
10817 (bb_top_order_index_size, bb_top_order_cmp): New.
10818 (stmts_from_loop): Use topological order.
10819 (pass_loop_distribution::execute): Compute and release topological
10820 order for basic blocks.
10822 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10824 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
10827 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10829 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
10830 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
10831 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
10832 * internal-fn.def (LOOP_DIST_ALIAS): New.
10833 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
10834 (fold_loop_internal_call): ... this.
10835 (vect_loop_dist_alias_call): New function.
10836 (set_uid_loop_bbs): Call fold_loop_internal_call.
10837 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
10840 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
10843 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
10844 Require dead FLAGS_REG at the beginning of a peephole.
10846 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
10849 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
10850 arguments in the call to __builtin_ia32_sbb_u32.
10851 (_subborrow_u64): Swap _X and _Y arguments in the call to
10852 __builtin_ia32_sbb_u64.
10854 2017-07-04 Jakub Jelinek <jakub@redhat.com>
10857 * tree-vrp.c (compare_assert_loc): Turn into a function template
10858 with stable template parameter. Only test if a->e is NULL,
10859 !a->e == !b->e has been verified already. Use e == NULL or
10860 e != NULL instead of e or ! e tests. If stable is true, don't use
10861 iterative_hash_expr, on the other side allow a or b or both NULL
10862 and sort the NULLs last.
10863 (process_assert_insertions): Sort using compare_assert_loc<false>
10864 instead of compare_assert_loc, later sort using
10865 compare_assert_loc<true> before calling process_assert_insertions_for
10866 in a loop. Use break instead of continue once seen NULL pointer.
10868 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
10870 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
10871 Cortex-R7 and Cortex-R8 processors.
10873 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
10875 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
10876 uninitialized while src is not.
10878 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
10880 * common/config/arm/arm-common.c: Adjust include path for
10882 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
10883 (arm-cpu.h): Create in build directory. Adjust dependency rules.
10884 (arm-cpu-data.h): Likewise.
10885 (arm-cpu-cdata.h): Likewise.
10886 * config/arm/arm-cpu.h: Delete.
10887 * config/arm/arm-cpu-cdata.h: Delete.
10888 * config/arm/arm-cpu-data.h: Delete.
10890 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
10892 * config/arm/arm-cpus.in (cortex-a55): New.
10893 (cortex-a75): Likewise.
10894 (cortex-a75.cortex-a55): Likewise.
10895 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
10897 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
10898 * config/arm/arm-cpu-cdata.h: Regenerate.
10899 * config/arm/arm-cpu-data.h: Regenerate.
10900 * config/arm/arm-cpu.h: Regenerate.
10901 * config/arm/arm-tables.opt: Regenerate.
10902 * config/arm/arm-tune.md: Regenerate.
10904 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
10906 * haifa-sched.c (sched_create_recovery_edges): Update profile.
10908 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
10910 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
10913 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
10915 PR tree-optimization/81292
10916 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
10917 full_string_p, also call adjust_related_strinfos if the adjustment
10918 is simple, otherwise invalidate related strinfos.
10920 2017-07-04 Martin Liska <mliska@suse.cz>
10923 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
10924 newly created variable as DECL_IGNORED_P.
10926 2017-07-04 Martin Liska <mliska@suse.cz>
10929 * ipa-inline.c (inline_small_functions):
10930 Use xstrdup_for_dump.
10932 2017-07-04 Tom de Vries <tom@codesourcery.com>
10934 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
10936 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
10939 * config/darwin.c (darwin_function_switched_text_sections):
10942 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
10944 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
10946 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
10948 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
10950 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10952 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
10953 min_profitable_iters, and th as inclusive lower bounds.
10954 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
10955 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
10956 for min_profitable_iters and min_profitable_estimate.
10957 (vect_transform_loop): Treat th as an inclusive lower bound.
10958 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
10960 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
10963 * config/darwin.c (darwin_function_switched_text_sections):
10964 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
10965 in two pieces, and suppress the use of buf.
10967 2017-07-03 Nathan Sidwell <nathan@acm.org>
10969 * hash-table.h (hash_table_mod1): Fix indentation.
10971 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10973 PR middle-end/81290
10974 * predict.c (force_edge_cold): Be more careful about propagation
10976 * profile-count.h (profile_probability::guessed,
10977 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
10979 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
10981 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
10983 * doc/invoke.texi (rcpc architecture extension): Document it.
10985 2017-07-03 Richard Biener <rguenther@suse.de>
10987 PR tree-optimization/60510
10988 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
10989 the scalar reduction PHI and use it.
10990 (vectorizable_reduction): Properly guard the single_defuse_cycle
10991 path for non-SLP reduction chains where we cannot use it.
10992 Rework reduc_def/index and vector type deduction. Rework
10993 vector operand gathering during reduction op code-gen.
10994 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
10995 chains dissolve the chain and leave it to non-SLP reduction
10998 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11000 * tree-data-ref.h (dr_alignment): Declare.
11001 * tree-data-ref.c (dr_alignment): New function.
11002 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
11003 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
11005 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
11007 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11009 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
11010 and base_misalignment fields.
11011 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
11012 * tree-data-ref.c: Include builtins.h.
11013 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
11014 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
11015 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
11016 * tree-vect-data-refs.c: Include tree-cfg.h.
11017 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
11018 fields instead of calculating an alignment here.
11019 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
11020 innermost_loop_behavior fields.
11022 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11024 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
11026 (DR_STEP_ALIGNMENT): New macro.
11027 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
11028 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
11029 (create_data_ref): Print it.
11030 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
11031 to tell whether the step preserves vector (mis)alignment.
11032 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
11033 Move the check for an integer step and generalise to all INTEGER_CST.
11034 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
11035 Print the outer step alignment.
11037 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11039 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
11040 with offset_alignment.
11041 (DR_ALIGNED_TO): Delete.
11042 (DR_OFFSET_ALIGNMENT): New macro.
11043 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
11044 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
11045 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
11046 (create_data_ref): Likewise.
11047 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
11048 (vect_analyze_data_refs): Likewise.
11049 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
11050 creating dummy innermost behavior.
11052 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11054 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
11055 with a "innermost_loop_behavior *" and refeence tree.
11056 * tree-data-ref.c (dr_analyze_innermost): Likewise.
11057 (create_data_ref): Update call accordingly.
11058 * tree-predcom.c (find_looparound_phi): Likewise.
11060 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11062 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
11063 fields with dr_wrt_vec_loop.
11064 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
11065 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
11066 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
11067 (vect_dr_behavior): New function.
11068 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
11069 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
11070 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
11071 track whether the step preserves the misalignment.
11072 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
11073 Use vect_dr_behavior.
11074 (vect_setup_realignment): Update call accordingly.
11075 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
11076 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
11077 call to vect_create_addr_base_for_vector_ref.
11078 (vect_create_cond_for_align_checks): Likewise.
11079 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
11080 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
11081 (vect_recog_mask_conversion_pattern): Likewise.
11082 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
11083 (new_stmt_vec_info): Remove redundant zeroing.
11085 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
11087 * common/config/arm/arm-common.c (arm_be8_option): New function.
11088 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
11089 (ISA_ARMv6): Add isa_bit_be8.
11090 * config/arm/arm.h (arm_be8_option): Add prototype.
11091 (BE8_SPEC_FUNCTION): New define.
11092 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
11093 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
11094 (mlittle-endian): Similarly.
11095 (mbe8, mbe32): New options.
11096 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
11097 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
11099 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11101 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
11103 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11105 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
11106 (cleanup_tree_cfg_bb): Use it.
11107 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
11109 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
11111 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11114 * loop-doloop.c (add_test): Update profile.
11116 2017-07-03 Martin Liska <mliska@suse.cz>
11119 * sanopt.c (rewrite_usage_of_param): New function.
11120 (sanitize_rewrite_addressable_params): Likewise.
11121 (pass_sanopt::execute): Call rewrite_usage_of_param.
11123 2017-07-03 Richard Biener <rguenther@suse.de>
11125 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
11126 back to using VIEW_CONVERT_EXPR.
11128 2017-07-03 Martin Liska <mliska@suse.cz>
11131 * doc/extend.texi: Document when a resolver function is
11132 generated for target_clones.
11134 2017-07-03 Martin Liska <mliska@suse.cz>
11136 * asan.c (asan_emit_stack_protection): Unpoison just red zones
11137 and shadow memory of auto variables which are subject of
11138 use-after-scope sanitization.
11139 (asan_expand_mark_ifn): Add do set only when is_poison.
11141 2016-07-03 Richard Biener <rguenther@suse.de>
11143 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
11145 (vect_force_simple_reduction): Record reduction def -> phi mapping.
11146 (vectorizable_reduction): Perform reduction PHI creation when
11147 visiting a reduction PHI and adjust and simplify code generation
11148 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
11149 (vect_transform_loop): Visit reduction PHIs.
11150 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
11151 defs into the SLP tree.
11152 (vect_build_slp_tree): Reduction defs terminate the recursion.
11153 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
11155 (vect_get_vec_defs_for_stmt_copy): Export.
11156 (vect_get_vec_defs): Likewise.
11157 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
11159 (vect_get_vec_defs_for_stmt_copy): Declare.
11160 (vect_get_vec_defs): Likewise.
11162 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11164 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
11165 parameter with a "loop" parameter and use it instead of the
11166 loop containing DR_STMT. Don't check simple_iv when doing
11167 BB analysis. Describe the two analysis modes in the comment.
11169 2017-07-03 Tom de Vries <tom@codesourcery.com>
11171 PR tree-optimization/69468
11172 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
11173 (find_same_succ_bb): Handle ignore_edge_flags.
11175 2017-07-03 Tom de Vries <tom@codesourcery.com>
11177 PR tree-optimization/81192
11178 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
11180 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
11182 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
11184 2017-07-03 Tom de Vries <tom@codesourcery.com>
11186 PR tree-optimization/81192
11187 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
11188 BB_SAME_SUCC (bb) == NULL.
11190 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11192 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
11195 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11197 * dumpfile.c: Include profile-count.h
11198 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
11200 (insert_cond_bb): Update profile.
11201 * tree-cfg.h (insert_cond_bb): Update prototype.
11202 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
11203 * tree-dump.c: Do not include tree-cfg.
11205 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11207 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
11209 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11211 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
11214 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11216 * tree-complex.c (expand_complex_div_wide): update profile.
11218 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
11219 Alan Hayward <alan.hayward@arm.com>
11220 David Sherwood <david.sherwood@arm.com>
11222 * Makefile.in (MACHMODE_H): Remove insn-modes.h
11223 (CORETYPES_H): New define.
11224 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
11225 (insn-modes-inline.h, s-modes-inline-h): New rules.
11226 (generated_files): Add insn-modes-inline.h.
11227 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
11228 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
11229 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
11230 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
11231 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
11232 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
11233 (build/gencodes.o, build/genconditions.o): Likewise.
11234 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
11235 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
11236 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
11237 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
11238 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
11239 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
11240 * coretypes.h: Include everything up to real.h for generators.
11241 Include insn-modes.h first. Include wide-int-print.h after
11242 wide-int.h. Include insn-modes-inline.h and then machmode.h.
11243 * machmode.h: Don't include insn-modes.h here.
11244 * function-tests.c: Remove includes of signop.h, machmode.h,
11245 double-int.h and wide-int.h.
11247 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
11249 * optc-save-gen.awk: Likewise.
11250 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
11251 * godump.c: Remove include of wide-int-print.h.
11252 * pretty-print.h: Likewise.
11253 * wide-int-print.cc: Likewise.
11254 * wide-int.cc: Likewise.
11255 * hash-map-tests.c: Remove include of signop.h.
11256 * hash-set-tests.c: Likewise.
11257 * rtl-tests.c: Likewise.
11258 * mkconfig.sh: Remove include of machmode.h.
11259 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
11261 (emit_insn_modes_inline_h): ...this new function. Emit the code
11262 into an insn-modes-inline.h header file, adding appropriate
11263 include guards and end comments.
11264 (emit_insn_modes_c_header): Remove include of machmode.h.
11265 (emit_min_insn_modes_c_header): Include coretypes.h rather than
11267 (main): Handle -i flag and call emit_insn_modes_inline_h when
11270 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
11272 * tree-ssa-strlen.c (strinfo): Rename the length field to
11273 nonzero_chars. Add a full_string_p field.
11274 (compare_nonzero_chars, zero_length_string_p): New functions.
11275 (get_addr_stridx): Add an offset_out parameter.
11276 Use compare_nonzero_chars.
11277 (get_stridx): Update accordingly. Use compare_nonzero_chars.
11278 (new_strinfo): Update after above changes to strinfo.
11279 (set_endptr_and_length): Set full_string_p.
11280 (get_string_length): Update after above changes to strinfo.
11281 (unshare_strinfo): Update call to new_strinfo.
11282 (maybe_invalidate): Likewise.
11283 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
11284 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
11285 as a uhwi instead of an shwi. Update after above changes to
11286 strinfo and new_strinfo.
11287 (zero_length_string): Assert that chainsi contains full strings.
11288 Use zero_length_string_p. Update call to new_strinfo.
11289 (adjust_related_strinfos): Update after above changes to strinfo.
11290 Copy full_string_p from origsi.
11291 (adjust_last_stmt): Use zero_length_string_p.
11292 (handle_builtin_strlen): Update after above changes to strinfo and
11293 new_strinfo. Install the lhs as the string length if the previous
11294 entry didn't describe a full string.
11295 (handle_builtin_strchr): Update after above changes to strinfo
11297 (handle_builtin_strcpy): Likewise.
11298 (handle_builtin_strcat): Likewise.
11299 (handle_builtin_malloc): Likewise.
11300 (handle_pointer_plus): Likewise.
11301 (handle_builtin_memcpy): Likewise. Track nonzero characters
11302 that aren't necessarily followed by a nul terminator.
11303 (handle_char_store): Likewise.
11305 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
11307 PR tree-optimization/80769
11308 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
11309 for malloc and calloc. Document the new invariant that all related
11310 strinfos have delayed lengths or none do.
11311 (verify_related_strinfos): Move earlier in file.
11312 (set_endptr_and_length): New function, split out from...
11313 (get_string_length): ...here. Also set the lengths of related
11315 (zero_length_string): Assert that chainsi has known (rather than
11317 (adjust_related_strinfos): Likewise.
11319 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
11321 PR tree-optimization/81136
11322 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
11323 assert that two references with the same misalignment have the same
11324 compile-time misalignment if those compile-time misalignments
11327 2017-07-01 Andi Kleen <ak@linux.intel.com>
11329 * print-tree.c (print_node): Print all attributes.
11331 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11333 * cfg.c (scale_bbs_frequencies): New function.
11334 * cfg.h (scale_bbs_frequencies): Declare it.
11335 * cfgloopanal.c (single_likely_exit): Cleanup.
11336 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
11338 (scale_loop_profile): Likewise.
11339 (loop_version): Likewise.
11340 (create_empty_loop_on_edge): Update.
11341 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
11342 scale_loop_frequencies, scale_loop_profile, loopify,
11343 loop_version): Update prototypes.
11344 * modulo-sched.c (sms_schedule): Update.
11345 * predict.c (unlikely_executed_edge_p): Also check probability.
11346 (probably_never_executed_edge_p): Fix typo.
11347 * tree-if-conv.c (version_loop_for_if_conversion): Update.
11348 * tree-parloops.c (gen_parallel_loop): Update.
11349 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
11350 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
11351 * tree-ssa-loop-split.c (split_loop): Update.
11352 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
11353 * tree-vect-loop-manip.c (vect_do_peeling): Update.
11354 (vect_loop_versioning): Update.
11355 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
11357 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11359 * trans-mem.c (split_bb_make_tm_edge): Update profile.
11361 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11363 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
11364 to keep profile consistent.
11366 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11368 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
11369 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
11370 * profile-count.h (max_safe_multiplier): Make unsigned.
11371 (profile_count::guessed_zero): New.
11373 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11375 * bb-reorder.c (fix_up_crossing_landing_pad,
11376 fix_crossing_conditional_branches): Use make_single_succ_edge
11377 to keep profile consistent.
11379 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11381 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
11384 2017-07-01 Jakub Jelinek <jakub@redhat.com>
11387 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
11388 the right scopes, make sure cond_jump isn't preserved between multiple
11389 iterations. Search for fallthru edge whenever there are 3+ edges and
11390 use find_fallthru_edge for it.
11392 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
11394 Patch by Alexander Monakov <amonakov@ispras.ru>
11395 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
11396 probabilities consistently.
11398 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
11400 * pa.c (pa_expand_compare_and_swap_loop): Update call of
11401 emit_cmp_and_jump_insns.
11403 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
11406 * tree-inline.c (expand_call_inline): Combine profile statuses.
11408 2017-06-30 Andrew Pinski <apinski@cavium.com>
11410 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
11411 fold_stmt returned true.
11413 2017-06-30 Nathan Sidwell <nathan@acm.org>
11415 * ggc.h (empty_string): Delete.
11416 * cfgexpand.c (expand_asm_stmt): Use plain "".
11417 * optabs.c (expand_asm_memory_barrier): Likewise.
11418 * stringpool.c (empty_string): Delete.
11419 (digit_vector, digit_string): Delete.
11420 (ggc_alloc_string): Use plain "", don't optimize single digit
11421 strings. Use ggc_alloc_atomic.
11423 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
11425 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
11426 comparison set and one other set, use the cost of the non-comparison
11429 2017-06-30 Nathan Sidwell <nathan@acm.org>
11431 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
11434 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
11436 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
11437 loops. Remove now unneeded calls to gimple_switch_set_label() that
11438 just set removed labels to NULL_TREE.
11440 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
11442 * tree-ssanames.c (set_range_info_raw): Abstract from ...
11443 (set_range_info): ...here. Only call set_range_info_raw if domain
11445 (set_nonzero_bits): Call set_range_info_raw.
11446 * tree-ssanames.h (set_range_info_raw): New.
11448 2017-06-30 Jakub Jelinek <jakub@redhat.com>
11451 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
11452 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
11453 of nonimmediate_operand and <store_mask_constraint> instead of m
11454 for the input operand. For V8FI iterator, always split if input
11455 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
11456 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
11457 <store_mask_predicate> instead of register_operand and
11458 <store_mask_constraint> instead of v for the input operand. Make
11459 sure both operands aren't MEMs for if not <mask_applied>.
11461 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
11463 * lto-wrapper.c (copy_file) Close both file descriptors before
11466 2017-06-30 Martin Liska <mliska@suse.cz>
11469 * multiple_target.c (create_dispatcher_calls): Make ifunc
11470 also for function that don't have calls or are not referenced.
11472 2017-06-30 Richard Biener <rguenther@suse.de>
11474 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
11475 analyze the first scalar stmt. Move vector type computation
11476 for the BB case here from ...
11477 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
11478 live operation processing in the SLP case properly.
11480 2017-06-30 Richard Biener <rguenther@suse.de>
11482 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
11484 2017-06-30 Martin Liska <mliska@suse.cz>
11487 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
11488 before BUILT_IN_UNWIND_RESUME when ASAN is used.
11490 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
11492 * doc/invoke.texi (AArch64): Add missing options and remove redundant
11495 2017-06-30 Richard Biener <rguenther@suse.de>
11497 PR tree-optimization/81249
11498 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
11499 condition reduction result to original scalar type.
11501 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11503 * profile-count.h (enum profile_quality): Fix typos and whitespace
11506 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11508 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
11509 type for branch probabilities.
11511 2017-06-29 Julian Brown <julian@codesourcery.com>
11512 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11514 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
11515 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
11516 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
11517 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
11519 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11521 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
11522 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
11523 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
11524 CC usage from generic code to here.
11525 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
11526 CC usage into the target macros.
11528 2017-06-29 Maya Rashish <coypu@sdf.org>
11530 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
11533 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
11535 * arm/arm-builtins.c: Include profile-count.h
11536 * except.c (sjlj_emit_function_enter): Use
11537 profile_probability::unlikely.
11539 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
11541 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
11542 and tocrel_offset be pointer args rather than implicitly using
11544 (legitimate_constant_pool_address_p, rs6000_emit_move,
11545 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
11546 tocrel_offset and use in toc_relative_expr_p call.
11547 (print_operand, print_operand_address): Use static tocrel_base_oac
11548 and tocrel_offset_oac.
11549 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
11552 2017-06-29 Maya Rashish <coypu@sdf.org>
11554 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
11556 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
11558 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
11559 objects, take into account only the alignment of 'op0' and 'mode1' if
11562 2017-06-29 Steve Ellcey <sellcey@cavium.com>
11564 * ccmp.c (ccmp_tree_comparison_p): New function.
11565 (ccmp_candidate_p): Update to use above function.
11566 (get_compare_parts): New function.
11567 (expand_ccmp_next): Update to use new functions.
11568 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
11570 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
11571 take mode as argument.
11572 * ccmp.h (expand_ccmp_expr): Add mode as argument.
11573 * expr.c (expand_expr_real_1): Pass mode as argument.
11575 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
11577 * combine.c (combine_instructions): Print insns to dump_file, together
11580 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
11582 * asan.c (asan_emit_stack_protection): Update.
11583 (create_cond_insert_point): Update.
11584 * auto-profile.c (afdo_propagate_circuit): Update.
11585 * basic-block.h (struct edge_def): Turn probability to
11586 profile_probability.
11587 (EDGE_FREQUENCY): Update.
11588 * bb-reorder.c (find_traces_1_round): Update.
11589 (better_edge_p): Update.
11590 (sanitize_hot_paths): Update.
11591 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
11592 (make_single_succ_edge): Update.
11593 (check_bb_profile): Update.
11594 (dump_edge_info): Update.
11595 (update_bb_profile_for_threading): Update.
11596 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
11597 probabilitycount to 0.
11598 * cfgbuild.c (compute_outgoing_frequencies): Update.
11599 * cfgcleanup.c (try_forward_edges): Update.
11600 (outgoing_edges_match): Update.
11601 (try_crossjump_to_edge): Update.
11602 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
11603 (expand_gimple_tailcall): Update.
11604 (construct_init_block): Use make_single_succ_edge.
11605 (construct_exit_block): Use make_single_succ_edge.
11606 * cfghooks.c (verify_flow_info): Update.
11607 (redirect_edge_succ_nodup): Update.
11608 (split_edge): Update.
11609 (account_profile_record): Update.
11610 * cfgloopanal.c (single_likely_exit): Update.
11611 * cfgloopmanip.c (scale_loop_profile): Update.
11612 (set_zero_probability): Remove.
11613 (duplicate_loop_to_header_edge): Update.
11614 * cfgloopmanip.h (loop_version): Update prototype.
11615 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
11616 (force_nonfallthru_and_redirect): Update.
11617 (update_br_prob_note): Update.
11618 (rtl_verify_edges): Update.
11619 (purge_dead_edges): Update.
11620 (rtl_lv_add_condition_to_bb): Update.
11621 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
11622 * cgraphunit.c (init_lowered_empty_function): Update.
11623 (cgraph_node::expand_thunk): Update.
11624 * cilk-common.c: Include profile-count.h
11625 * dojump.c (inv): Remove.
11626 (jumpifnot): Update.
11627 (jumpifnot_1): Update.
11628 (do_jump_1): Update.
11630 (do_jump_by_parts_greater_rtx): Update.
11631 (do_compare_rtx_and_jump): Update.
11632 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
11633 do_jump_1. do_compare_rtx_and_jump): Update prototype.
11634 * dwarf2cfi.c: Include profile-count.h
11635 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
11636 (sjlj_emit_dispatch_table): Likewise.
11637 * explow.c: Include profile-count.h
11638 * expmed.c (emit_store_flag_force): Update.
11639 (do_cmp_and_jump): Update.
11640 * expr.c (compare_by_pieces_d::generate): Update.
11641 (compare_by_pieces_d::finish_mode): Update.
11642 (emit_block_move_via_loop): Update.
11643 (store_expr_with_bounds): Update.
11644 (store_constructor): Update.
11645 (expand_expr_real_2): Update.
11646 (expand_expr_real_1): Update.
11647 * expr.h (try_casesi, try_tablejump): Update prototypes.
11648 * gimple-pretty-print.c (dump_probability): Update.
11649 (dump_profile): New.
11650 (dump_gimple_label): Update.
11651 (dump_gimple_bb_header): Update.
11652 * graph.c (draw_cfg_node_succ_edges): Update.
11653 * hsa-gen.c (convert_switch_statements): Update.
11654 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
11655 (find_if_case_1): Update.
11656 (find_if_case_2): Update.
11657 * internal-fn.c (expand_arith_overflow_result_store): Update.
11658 (expand_addsub_overflow): Update.
11659 (expand_neg_overflow): Update.
11660 (expand_mul_overflow): Update.
11661 (expand_vector_ubsan_overflow): Update.
11662 * ipa-cp.c (good_cloning_opportunity_p): Update.
11663 * ipa-split.c (split_function): Use make_single_succ_edge.
11664 * ipa-utils.c (ipa_merge_profiles): Update.
11665 * loop-doloop.c (add_test): Update.
11666 (doloop_modify): Update.
11667 * loop-unroll.c (compare_and_jump_seq): Update.
11668 (unroll_loop_runtime_iterations): Update.
11669 * lra-constraints.c (lra_inheritance): Update.
11670 * lto-streamer-in.c (input_cfg): Update.
11671 * lto-streamer-out.c (output_cfg): Update.
11672 * mcf.c (adjust_cfg_counts): Update.
11673 * modulo-sched.c (sms_schedule): Update.
11674 * omp-expand.c (expand_omp_for_init_counts): Update.
11675 (extract_omp_for_update_vars): Update.
11676 (expand_omp_ordered_sink): Update.
11677 (expand_omp_for_ordered_loops): Update.
11678 (expand_omp_for_generic): Update.
11679 (expand_omp_for_static_nochunk): Update.
11680 (expand_omp_for_static_chunk): Update.
11681 (expand_cilk_for): Update.
11682 (expand_omp_simd): Update.
11683 (expand_omp_taskloop_for_outer): Update.
11684 (expand_omp_taskloop_for_inner): Update.
11685 * omp-simd-clone.c (simd_clone_adjust): Update.
11686 * optabs.c (expand_doubleword_shift): Update.
11687 (expand_abs): Update.
11688 (emit_cmp_and_jump_insn_1): Update.
11689 (expand_compare_and_swap_loop): Update.
11690 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
11691 * predict.c (predictable_edge_p): Update.
11692 (edge_probability_reliable_p): Update.
11693 (set_even_probabilities): Update.
11694 (combine_predictions_for_insn): Update.
11695 (combine_predictions_for_bb): Update.
11696 (propagate_freq): Update.
11697 (estimate_bb_frequencies): Update.
11698 (force_edge_cold): Update.
11699 * profile-count.c (profile_count::dump): Add missing space into dump.
11700 (profile_count::debug): Add newline.
11701 (profile_count::differs_from_p): Explicitly convert to unsigned.
11702 (profile_count::stream_in): Update.
11703 (profile_probability::dump): New member function.
11704 (profile_probability::debug): New member function.
11705 (profile_probability::differs_from_p): New member function.
11706 (profile_probability::differs_lot_from_p): New member function.
11707 (profile_probability::stream_in): New member function.
11708 (profile_probability::stream_out): New member function.
11709 * profile-count.h (profile_count_quality): Rename to ...
11710 (profile_quality): ... this one.
11711 (profile_probability): New.
11712 (profile_count): Update.
11713 * profile.c (compute_branch_probabilities): Update.
11714 * recog.c (peep2_attempt): Update.
11715 * sched-ebb.c (schedule_ebbs): Update.
11716 * sched-rgn.c (find_single_block_region): Update.
11717 (compute_dom_prob_ps): Update.
11718 (schedule_region): Update.
11719 * sel-sched-ir.c (compute_succs_info): Update.
11720 * stmt.c (struct case_node): Update.
11721 (do_jump_if_equal): Update.
11722 (get_outgoing_edge_probs): Update.
11723 (conditional_probability): Update.
11724 (emit_case_dispatch_table): Update.
11725 (expand_case): Update.
11726 (expand_sjlj_dispatch_table): Update.
11727 (emit_case_nodes): Update.
11728 * targhooks.c: Update.
11729 * tracer.c (better_p): Update.
11730 (find_best_successor): Update.
11731 * trans-mem.c (expand_transaction): Update.
11732 * tree-call-cdce.c: Update.
11733 * tree-cfg.c (gimple_split_edge): Upate.
11734 (move_sese_region_to_fn): Upate.
11735 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
11736 * tree-eh.c (lower_resx): Upate.
11737 (cleanup_empty_eh_move_lp): Upate.
11738 * tree-if-conv.c (version_loop_for_if_conversion): Update.
11739 * tree-inline.c (copy_edges_for_bb): Update.
11740 (copy_cfg_body): Update.
11741 * tree-parloops.c (gen_parallel_loop): Update.
11742 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
11743 (gimple_gen_time_profiler): Update.
11744 * tree-ssa-dce.c (remove_dead_stmt): Update.
11745 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
11746 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
11747 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
11748 (unloop_loops): Update.
11749 (try_peel_loop): Update.
11750 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
11751 * tree-ssa-loop-split.c (connect_loops): Update.
11752 (split_loop): Update.
11753 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
11754 (hoist_guard): Update.
11755 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
11756 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
11757 (value_replacement): Update.
11758 * tree-ssa-reassoc.c (branch_fixup): Update.
11759 * tree-ssa-tail-merge.c (replace_block_by): Update.
11760 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
11761 (create_edge_and_update_destination_phis): Update.
11762 (compute_path_counts): Update.
11763 (recompute_probabilities): Update.
11764 (update_joiner_offpath_counts): Update.
11765 (freqs_to_counts_path): Update.
11766 (duplicate_thread_path): Update.
11767 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
11768 (struct switch_conv_info): Update.
11769 (gen_inbound_check): Update.
11770 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
11771 (vect_do_peeling): Update.
11772 (vect_loop_versioning): Update.
11773 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
11774 (optimize_mask_stores): Update.
11775 * ubsan.c (ubsan_expand_null_ifn): Update.
11776 * value-prof.c (gimple_divmod_fixed_value): Update.
11777 (gimple_divmod_fixed_value_transform): Update.
11778 (gimple_mod_pow2): Update.
11779 (gimple_mod_pow2_value_transform): Update.
11780 (gimple_mod_subtract): Update.
11781 (gimple_mod_subtract_transform): Update.
11782 (gimple_ic): Update.
11783 (gimple_stringop_fixed_value): Update.
11784 (gimple_stringops_transform): Update.
11785 * value-prof.h: Update.
11787 2017-06-29 Carl Love <cel@us.ibm.com>
11789 * config/rs6000/rs6000-c.c: Add support for built-in functions
11790 vector signed int vec_signed (vector float);
11791 vector signed long long vec_signed (vector double);
11792 vector signed int vec_signed2 (vector double, vector double);
11793 vector signed int vec_signede (vector double);
11794 vector signed int vec_signedo (vector double);
11795 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
11796 instruction generator.
11797 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
11798 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
11799 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
11801 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
11802 vunsignede_v2df): Add define_expands.
11803 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
11804 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
11805 VEC_UNSIGNEDO): Add definitions.
11806 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
11807 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
11808 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
11809 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
11810 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
11811 * config/rs6000/altivec.h (vec_signed, vec_signed2,
11812 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
11813 vec_unsignede, vec_unsignedo): Add builtin defines.
11814 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
11816 * doc/extend.texi: Update the built-in documentation file for the
11817 new built-in functions.
11819 2017-06-29 Richard Biener <rguenther@suse.de>
11821 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
11822 reduction chains to LOOP_VINFO_REDUCTIONS.
11823 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
11824 SLP reductions after processing reduction chains.
11826 2017-06-29 Nathan Sidwell <nathan@acm.org>
11828 * builtins.c (fold_builtin_FUNCTION): Use
11829 lang_hooks.decl_printable_name.
11831 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
11833 PR middle-end/81194
11834 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
11835 with only one label.
11836 * stmt.c (expand_case): Assert NCASES is greater than one.
11838 2017-06-29 Richard Biener <rguenther@suse.de>
11840 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
11842 (group_case_labels): Likewise.
11843 (find_taken_edge): Push sanity checking on val to workers...
11844 (find_taken_edge_cond_expr): ... here
11845 (find_taken_edge_switch_expr): ... and here, handle cases
11846 with just a default label.
11847 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
11848 (group_case_labels): Likewise.
11849 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
11850 group_case_labels does anything cleanup the CFG again.
11852 2017-06-29 Bin Cheng <bin.cheng@arm.com>
11854 PR tree-optimization/81196
11855 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
11856 exit condition comparing two IVs.
11858 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
11860 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
11861 profile to the dummy entry at the end of the list of architectures.
11862 * config/arm/arm-cpu-cdata.h: Regenerated.
11864 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11865 Michael Collison <michael.collison@arm.com>
11868 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
11870 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
11871 (*aarch64_reg_<mode>3_minus_mask): New pattern.
11872 (*aarch64_<optab>_reg_di3_mask2): New pattern.
11873 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
11874 of shift when the shift amount is masked with constant equal to
11875 the size of the mode.
11876 * config/aarch64/predicates.md (subreg_lowpart_operator): New
11879 2017-06-29 Martin Liska <mliska@suse.cz>
11881 * config/i386/i386.opt: Change range from [1,5] to [0,5].
11883 2017-06-29 Yury Gribov <tetra2005@gmail.com>
11886 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
11889 (edge_growth_cache_entry::edge_growth_cache_entry): New
11891 (reset_edge_growth_cache): Update to use constructor.
11893 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
11895 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
11896 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
11897 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
11899 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
11901 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
11902 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
11904 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
11906 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
11907 (*-linux-uclibc*): Add t-uclibc tmake_file.
11908 * config/t-musl: New.
11909 * config/t-uclibc: New.
11911 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
11913 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
11915 (gen_comm_data): Emit architectural setting of arch_prof.
11916 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
11918 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
11919 (armv8-m.base, armv8-m.main): Likewise.
11920 * arm-protos.h (arm_build_target): Add profile field.
11921 (arch_option): Likewise.
11922 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
11924 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
11925 arm_active_target.profile.
11927 2017-06-28 Richard Biener <rguenther@suse.de>
11929 PR middle-end/81227
11930 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
11931 TYPE_OVERFLOW_WRAPS.
11932 * match.pd (negate_expr_p): Likewise.
11933 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
11934 fold_build2, not fold_binary.
11936 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
11938 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
11939 Convert memory address to Pmode.
11940 (aarch64_print_operand): Assert MEM operands are always Pmode.
11942 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
11945 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
11946 Remove redundant if.
11947 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
11948 * config/arm/aarch-common-protos.h
11949 (aarch_forward_to_shift_is_not_shifted_re): Remove.
11950 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
11952 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
11955 * multiple_target.c (create_dispatcher_calls): Set the default
11956 clone to be static, not public.
11958 2017-06-28 Richard Biener <rguenther@suse.de>
11960 * tree-vect-loop.c (vectorizable_reduction): Move special
11961 cond reduction IV var creation ...
11962 (vect_create_epilog_for_reduction): ... here. Remove induction_index
11963 parameter. Use STMT_VINFO_VECTYPE.
11964 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
11967 2017-06-28 Martin Liska <mliska@suse.cz>
11970 * ipa-visibility.c (non_local_p): Handle visibility.
11972 2017-06-28 Martin Liska <mliska@suse.cz>
11975 * common.opt: Add IntegerRange to various options.
11976 * opt-functions.awk (integer_range_info): New function.
11977 * optc-gen.awk: Add integer_range_info to cl_options struct.
11978 * opts-common.c (decode_cmdline_option): Handle
11979 CL_ERR_INT_RANGE_ARG.
11980 (cmdline_handle_error): Likewise.
11981 * opts.c (print_filtered_help): Show valid interval in
11982 when --help is provided.
11983 * opts.h (struct cl_option): Add range_min and range_max fields.
11984 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
11986 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
11988 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
11989 (x * C EQ/NE y * C): New transformation.
11991 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
11993 * genmultilib (combination_space): Accept '+' in option names.
11995 2017-06-28 Martin Liska <mliska@suse.cz>
11998 * asan.c (instrument_derefs): Bail out inner references
11999 that are hard register variables.
12001 2017-06-28 Jakub Jelinek <jakub@redhat.com>
12004 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
12005 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
12007 2017-06-28 Richard Biener <rguenther@suse.de>
12009 * tree-vectorizer.h (vect_get_vec_defs): Remove.
12010 (vect_get_slp_defs): Adjust.
12011 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
12013 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
12015 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
12016 get_initial_defs_for_reduction instead of vect_get_vec_defs.
12017 (vectorizable_reduction): Adjust.
12018 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
12020 (vect_get_slp_defs): Likewise.
12021 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
12022 (vectorizable_bswap): Adjust.
12023 (vectorizable_call): Likewise.
12024 (vectorizable_conversion): Likewise.
12025 (vectorizable_assignment): Likewise.
12026 (vectorizable_shift): Likewise.
12027 (vectorizable_operation): Likewise.
12028 (vectorizable_store): Likewise.
12029 (vectorizable_condition): Likewise.
12030 (vectorizable_comparison): Likewise.
12032 2017-06-28 Michael Collison <michael.collison@arm.com>
12035 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
12037 (arm_gen_movmemqi): Removed unused variable 'i'.
12038 Convert 'for' loop into 'while' loop.
12039 (arm_expand_prologue): Remove last unnecessary set of insn.
12040 (thumb_pop): Remove unused variable 'pushed_words'.
12041 (thumb_exit): Remove last unnecessary set of regs_to_pop.
12043 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12045 * config/s390/predicates.md: Use s390_rel_address_ok_p.
12046 * config/s390/s390-protos.h: Add prototype of
12047 s390_rel_address_ok_p.
12048 * config/s390/s390.c (s390_got_symbol): New function.
12049 (s390_rel_address_ok_p): New function.
12050 (legitimize_pic_address): Use s390_rel_address_ok_p.
12051 (s390_load_got): Use s390_got_symbol.
12052 (s390_option_override): Issue error if
12053 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
12054 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
12056 * config/s390/s390.opt: New option mpic-data-is-text-relative.
12058 2017-06-27 Andrew Pinski <apinski@cavium.com>
12060 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
12061 (X * copysign (1.0, X)): New pattern.
12062 (X * copysign (1.0, -X)): New pattern.
12063 (copysign (-1.0, CST)): New pattern.
12065 2017-06-27 Joseph Myers <joseph@codesourcery.com>
12067 * genmultilib (combination_space): Remove variable.
12068 Validate reuse rules against regular expression for any sequence
12069 of multilib options in any order.
12071 2017-06-27 Michael Collison <michael.collison@arm.com>
12073 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
12074 call aarch64_split_simd_combine.
12075 * (aarch64_combine_internal<mode>): Delete pattern.
12076 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
12077 Allow register and subreg operands.
12079 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
12081 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
12082 specific need, just fallback on defaults.
12083 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
12085 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
12086 Olivier Hainque <hainque@adacore.com>
12088 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
12090 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
12091 targets. Pick a default if no particular attempt applied.
12092 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
12095 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
12097 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
12098 (x86_64-wrs-vxworks7): Likewise.
12100 2017-06-27 Marek Polacek <polacek@redhat.com>
12103 * ubsan.c (instrument_null): Check get_base_address's result for null.
12105 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
12107 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
12109 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
12111 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
12112 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
12113 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
12114 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
12115 New function types.
12116 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
12117 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
12118 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
12119 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
12120 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
12121 BUILT_IN_FEUPDATEENV): New builtins.
12122 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
12123 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
12124 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
12125 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
12127 (builtin_structptr_types): Adjust size.
12128 * tree.c (builtin_structptr_types): Add four entries.
12130 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
12131 Olivier Hainque <hainque@adacore.com>
12133 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
12134 (TLS_SYM): New local macro, forcing reference to __tls__ on
12135 link command lines for VxWorks 7 RTPs, triggering initialization
12137 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
12138 OS features TLS support, true for RTPs on VxWorks 7.
12139 * config/vxworks.c (vxworks_override_options): Setup emutls
12142 2017-06-27 Jakub Jelinek <jakub@redhat.com>
12144 * predict.c (test_prediction_value_range): Use -1U instead of -1
12145 to avoid narrowing conversion warning.
12146 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
12147 to avoid narrowing conversion warning.
12148 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
12150 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
12152 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
12154 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
12155 64bit configurations.
12156 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
12157 (SIZE_TYPE): Likewise.
12158 * config/vxworks.c (vxworks_emutls_var_fields): Use
12159 long_unsigned_type_node instead of unsigned_type_node as the offset
12160 field type, which is "pointer" mode in emutls.c.
12162 2017-06-27 Jakub Jelinek <jakub@redhat.com>
12165 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
12167 PR middle-end/81207
12168 * gimple-fold.c (replace_call_with_call_and_fold): Handle
12169 gimple_vuse copying separately from gimple_vdef copying.
12171 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
12173 * value-prof.c (free_hist): Remove call to memset and the enclosing if
12176 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
12177 Olivier Hainque <hainque@adacore.com>
12179 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
12180 for all vxworks7 targets.
12181 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
12182 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
12183 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
12184 variations for VX6/VX7 and 32/64bits later on in ...
12185 (VXWORKS_LIB_SPEC): Leverage new macros.
12186 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
12187 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
12189 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
12191 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
12192 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
12194 2017-06-26 Carl Love <cel@us.ibm.com>
12196 * config/rs6000/rs6000-c.c: Add support for built-in functions
12197 vector bool char vec_reve (vector bool char);
12198 vector signed char vec_reve (vector signed char);
12199 vector unsigned char vec_reve (vector unsigned char);
12200 vector bool int vec_reve (vector bool int);
12201 vector signed int vec_reve (vector signed int);
12202 vector unsigned int vec_reve (vector unsigned int);
12203 vector bool long long vec_reve (vector bool long long);
12204 vector signed long long vec_reve (vector signed long long);
12205 vector unsigned long long vec_reve (vector unsigned long long);
12206 vector bool short vec_reve (vector bool short);
12207 vector signed short vec_reve (vector signed short);
12208 vector double vec_reve (vector double);
12209 vector float vec_reve (vector float);
12210 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
12211 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
12212 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
12213 (altivec_vreve): New pattern.
12214 * config/rs6000/altivec.h (vec_reve): New define.
12215 * doc/extend.texi (vec_rev): Update the built-in documentation file
12216 for the new built-in functions.
12218 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12220 PR tree-optimization/71815
12221 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
12223 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
12225 (slsr_process_phi): Likewise.
12226 (replace_uncond_cands_and_profitable_phis): Don't replace a
12227 multiply candidate with a stride of 1 (copy or cast).
12228 (phi_incr_cost): Call uses_consumed_by_stmt rather than
12230 (lowest_cost_path): Likewise.
12231 (total_savings): Likewise.
12233 2017-06-26 Richard Biener <rguenther@suse.de>
12236 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
12237 Use def_builtin_pure for all gather builtins.
12239 2017-06-26 Richard Biener <rguenther@suse.de>
12241 PR tree-optimization/81203
12242 * tree-tailcall.c (find_tail_calls): Do not move stmts into
12243 non-dominating BBs.
12245 2017-06-26 Marek Polacek <polacek@redhat.com>
12248 * doc/invoke.texi: Document -Wmultistatement-macros.
12250 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
12252 * doc/sourcebuild.texi (ARM-specific attributes): Document new
12253 arm_neon_ok_no_float_abi effective target.
12255 2017-06-26 Richard Biener <rguenther@suse.de>
12257 PR tree-optimization/80928
12258 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
12259 (copy_bbs): Set BB_DUPLICATED flag early.
12260 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
12262 (execute_on_shrinking_pred): Likewise.
12263 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
12264 BB_DUPLICATED blocks.
12265 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
12266 iterate over all PHIs considering removal of *gsi.
12268 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
12270 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
12273 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
12275 * config/rs6000/rs6000-string.c: (expand_block_clear,
12276 do_load_for_compare, select_block_compare_mode,
12277 compute_current_alignment, expand_block_compare,
12278 expand_strncmp_align_check, expand_strn_compare,
12279 expand_block_move, rs6000_output_load_multiple)
12280 Move functions related to string/block move/compare
12281 to a separate file.
12282 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
12283 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
12284 for this function which is now used in two files.
12285 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
12286 * config.gcc: Add rs6000-string.o to extra_objs for
12287 targets powerpc*-*-* and rs6000*-*-*.
12289 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
12292 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
12293 32-bit, since indexed is not valid for DImode.
12294 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
12295 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
12296 (define_peephole2 for Altivec d-form load): Add 32-bit support.
12297 (define_peephole2 for Altivec d-form store): Likewise.
12300 * multiple_target.c (create_dispatcher_calls): Only create the
12301 dispatcher call if the function is the default clone of a
12302 versioned function.
12304 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
12306 PR middle-end/80902
12307 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
12308 a call, force the call to not be a tail call.
12310 2017-06-23 Jeff Law <law@redhat.com>
12312 * doc/contrib.texi: Add entry for Steven Pemberton's work on
12315 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
12317 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
12318 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
12319 handling for early expansion of vector shifts (sl,sr,sra,rl).
12320 (builtin_function_type): Add vector shift right instructions
12321 to the unsigned argument list.
12323 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
12325 rtl-optimizatoin/79286
12326 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
12327 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
12328 trap. PIC register plus a const unspec without offset can never trap.
12330 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
12332 * tree.h (builtin_structptr_type): New type.
12333 (builtin_structptr_types): Declare new array.
12334 * tree.c (builtin_structptr_types): New array.
12335 (free_lang_data, build_common_tree_nodes): Use it.
12337 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
12340 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
12343 2017-06-22 Matt Turner <mattst88@gmail.com>
12345 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
12346 Lake models to skylake case. Assume skylake for unknown
12347 models with clflushopt.
12349 2017-06-22 Jeff Law <law@redhat.com>
12351 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
12352 frame sizes that do not satisfy aarch64_uimm12_shift.
12354 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
12356 * profile-count.h (apply_probability,
12357 apply_scale, probability_in): Fix checks for zero.
12359 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12361 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
12362 * doc/cppdiropts.texi (-I @var{dir}): Document it.
12364 2016-06-22 Richard Biener <rguenther@suse.de>
12366 * tree-vect-loop.c (vect_model_reduction_cost): Handle
12367 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
12368 REDUC_MAX_EXPR support.
12369 (vectorizable_reduction): Likewise.
12370 (vect_create_epilog_for_reduction): Likewise.
12372 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
12374 * match.pd (A / (1 << B) -> A >> B): New.
12375 * generic-match-head.c: Include optabs-tree.h.
12376 * gimple-match-head.c: Likewise.
12377 * optabs-tree.h (target_supports_op_p): New.
12378 * optabs-tree.c (target_supports_op_p): New.
12380 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12382 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
12383 $gcc_cv_ld --help output.
12384 (gcc_cv_ld_demangle): Likewise.
12385 (gcc_cv_ld_eh_frame_hdr): Likewise.
12386 (gcc_cv_ld_pie): Likewise.
12387 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
12388 (gcc_cv_ld_buildid): Likewise.
12389 (gcc_cv_ld_sysroot): Likewise.
12390 (ld_bndplt_support): Likewise.
12391 (ld_pushpopstate_support): Likewise.
12392 * configure: Regenerate.
12393 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
12395 2017-06-21 Jakub Jelinek <jakub@redhat.com>
12398 * config/i386/sse.md (round<mode>2): Renumber match_dup and
12399 operands indexes to avoid gap between operands and match_dups.
12401 2017-06-21 Andrew Pinski <apinski@cavium.com>
12403 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
12404 Increment Arith_shift and Arith_shift_reg by 1.
12405 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
12407 * config/aarch64/aarch64.c (thunderx_tunings): Enable
12408 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
12409 (aarch64_strip_extend): Add new argument and test for it.
12410 (aarch64_cheap_mult_shift_p): New function.
12411 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
12412 add a cost if it is true.
12413 Update calls to aarch64_strip_extend.
12414 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
12416 2017-06-21 Andrew Pinski <apinski@cavium.com>
12418 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
12420 (thunderxt88): Likewise.
12421 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
12422 (thunderx_prefetch_tune): New variable.
12423 (thunderx2t99_prefetch_tune): Update for the correct values.
12424 (thunderxt88_tunings): New variable.
12425 (thunderx_tunings): Use thunderx_prefetch_tune instead of
12426 generic_prefetch_tune.
12427 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
12429 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12431 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
12432 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
12433 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
12434 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
12435 (aarch64_atomic_cas<mode>, GPI): Likewise.
12437 2017-06-21 Martin Liska <mliska@suse.cz>
12439 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
12440 statements on cold and hot labels.
12441 * predict.c (tree_estimate_probability_bb): Remove the
12442 prediction from this place.
12444 2017-06-21 Martin Liska <mliska@suse.cz>
12446 PR tree-optimization/79489
12447 * gimplify.c (maybe_add_early_return_predict_stmt): New
12449 (gimplify_return_expr): Call the function.
12450 * predict.c (tree_estimate_probability_bb): Remove handling
12452 * predict.def: Update comment about early return predictor.
12453 * gimple-predict.h (is_gimple_predict): New function.
12454 * predict.def: Change default value of early return to 66.
12455 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
12457 * passes.def: Put pass_strip_predict_hints to the beginning of
12460 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
12462 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
12463 FUNCTION_DECL declarations.
12464 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
12466 (dwaf2out_decl): Likewise.
12467 * godump.c (go_early_global_decl): Skip call to the real debug hook
12468 for FUNCTION_DECL declarations.
12469 * passes.c (rest_of_decl_compilation): Skip call to the
12470 early_global_decl debug hook for FUNCTION_DECL declarations, unless
12471 -fdump-go-spec is passed.
12473 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
12475 * config/i386/i386.c (struct builtin_isa): New field pure_p.
12476 Reorder for compactness.
12477 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
12478 (def_builtin_pure, def_builtin_pure2): New functions.
12479 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
12481 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
12483 * match.pd (nop_convert): New predicate.
12484 ((A +- CST1) +- CST2): Allow some NOP conversions.
12486 2017-06-21 Jakub Jelinek <jakub@redhat.com>
12489 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
12490 with ctors/dtors if GOVD_SHARED is set.
12492 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
12494 * config/aarch64/aarch64.md (movti_aarch64):
12495 Emit mov rather than orr.
12496 (movtf_aarch64): Likewise.
12497 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
12498 Emit mov rather than orr.
12500 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
12502 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
12503 Swap alternatives, make integer dup more expensive.
12505 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
12507 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
12508 Return true for non-tls symbols.
12510 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
12512 * config/aarch64/aarch64-cores.def (cortex-a55): New.
12513 (cortex-a75): Likewise.
12514 (cortex-a75.cortex-a55): Likewise.
12515 * config/aarch64/aarch64-tune.md: Regenerate.
12516 * doc/invoke.texi (-mtune): Document new values for -mtune.
12518 2017-06-21 Tom de Vries <tom@codesourcery.com>
12520 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
12521 stack_size feature.
12522 (Effective-Target Keywords, Other attributes): Suggest using
12523 dg-add-options stack_size feature to get stack limit in stack_size
12524 effective target documentation.
12526 2017-06-21 Julian Brown <julian@codesourcery.com>
12527 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
12529 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
12530 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
12531 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
12533 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
12534 attribute type list for neon_multiply.
12535 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
12536 attribute type list for neon_multiply.
12537 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
12538 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
12539 attribute type list for neon_multiply.
12540 * config/arm/types.md (crypto_pmull): Add.
12541 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
12542 attribute type list.
12544 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
12546 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
12549 2017-06-20 Jakub Jelinek <jakub@redhat.com>
12551 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
12552 to make sure not to dereference a NULL cost_classes_ptr pointer.
12554 2017-06-20 Carl Love <cel@us.ibm.com>
12556 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12557 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
12558 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
12559 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
12560 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
12561 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
12562 VMULOSW): New enum "unspec" values.
12563 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
12564 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
12565 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
12566 altivec_vmulosw): New patterns.
12567 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
12568 VMULOSW): Add definitions.
12570 2017-06-20 Julia Koval <julia.koval@intel.com>
12572 * config/i386/i386.c: Fix rounding expand for new pattern.
12573 * config/i386/subst.md: Fix pattern (parallel -> unspec).
12575 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
12577 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
12578 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
12580 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
12582 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
12585 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
12587 * config/aarch64/aarch64-cores.def: Rearrange to sort by
12588 architecture, then by implementer ID.
12589 * config/aarch64/aarch64-tune.md: Regenerate.
12591 2017-06-20 Richard Biener <rguenther@suse.de>
12593 PR middle-end/81097
12594 * fold-const.c (split_tree): Fold to type before negating.
12596 2017-06-20 David Malcolm <dmalcolm@redhat.com>
12598 * diagnostic-show-locus.c
12599 (selftest::test_fixit_deletion_affecting_newline): New function.
12600 (selftest::diagnostic_show_locus_c_tests): Call it.
12602 2017-06-20 Andreas Schwab <schwab@suse.de>
12605 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
12608 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
12610 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
12611 __ARM_FEATURE_COPROC according to support.
12613 2017-06-20 Jakub Jelinek <jakub@redhat.com>
12615 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
12616 Rewritten to avoid overflow for > 32-bit pointers.
12619 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
12620 by removing enum keyword.
12621 (ubsan_type_descriptor): Likewise. Formatting fix.
12624 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
12625 splitter): Require TARGET_SSE2 in the condition.
12627 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
12630 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
12631 for doing vector set of SFmode on ISA 3.0.
12632 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
12633 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
12635 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
12636 SFmode value into a V4SF variable that was extracted from another
12637 V4SF variable without converting the element to double precision
12638 and back to single precision vector format.
12639 (vsx_insert_extract_v4sf_p9_2): Likewise.
12641 2017-06-19 Jakub Jelinek <jakub@redhat.com>
12643 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
12644 in UWHI to avoid undefined overflow.
12647 * ubsan.h (enum ubsan_encode_value_phase): New.
12648 (ubsan_encode_value): Change second argument to
12649 enum ubsan_encode_value_phase with default value of
12650 UBSAN_ENCODE_VALUE_GENERIC.
12651 * ubsan.c (ubsan_encode_value): Change second argument to
12652 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
12653 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
12654 create_tmp_var_raw instead of create_tmp_var and use a
12656 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
12657 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
12658 ubsan_encode_value callers.
12661 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
12662 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
12663 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
12665 2017-06-19 Richard Biener <rguenther@suse.de>
12667 PR middle-end/81118
12668 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
12669 estimates if we changed anything.
12671 2017-06-19 Richard Biener <rguenther@suse.de>
12673 PR tree-optimization/80887
12674 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
12675 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
12676 simplified lookups, then reset mprts_hook.
12677 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
12679 (try_to_simplify): Likewise.
12681 2017-06-19 Martin Liska <mliska@suse.cz>
12684 * gimplify.c (gimplify_switch_expr):
12685 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
12687 2017-06-19 Martin Liska <mliska@suse.cz>
12689 * doc/install.texi: Document that PGO runs in 4 stages.
12691 2017-06-19 Martin Liska <mliska@suse.cz>
12694 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
12695 to dispatcher function name.
12696 * multiple_target.c (replace_function_decl): New function.
12697 (create_dispatcher_calls): Redirect both edges and references.
12699 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
12701 * profile-count.c (profile_count::dump): Dump quality.
12702 (profile_count::differs_from_p): Update for unsigned val.
12703 * profile-count.h (profile_count_quality): New enum.
12704 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
12706 2017-06-19 Richard Biener <rguenther@suse.de>
12708 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
12709 struct function as arg.
12710 (estimate_numbers_of_iterations): Export overload with loop arg.
12711 (free_numbers_of_iterations_estimates_loop): Use an overload of
12712 free_numbers_of_iterations_estimates instead.
12713 * tree-cfg.c (remove_bb): Adjust.
12714 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
12715 * tree-parloops.c (gen_parallel_loop): Likewise.
12716 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
12718 (tree_unroll_loops_completely): Likewise.
12719 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
12720 Use an overload instead and export.
12721 (estimated_loop_iterations): Adjust.
12722 (max_loop_iterations): Likewise.
12723 (likely_max_loop_iterations): Likewise.
12724 (estimate_numbers_of_iterations): Take struct function as arg
12726 (loop_exits_before_overflow): Adjust.
12727 (free_numbers_of_iterations_estimates_loop): Use an overload.
12728 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
12729 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
12731 2017-06-19 Richard Biener <rguenther@suse.de>
12734 * ipa-prop.c (find_constructor_constant_at_offset): Handle
12735 RANGE_EXPR conservatively.
12737 2017-06-16 Carl Love <cel@us.ibm.com>
12739 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12740 definitions for vec_float, vec_float2, vec_floato,
12741 vec_floate built-ins.
12742 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
12743 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
12745 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
12746 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
12747 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
12748 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
12749 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
12750 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
12751 vec_floato): Add builtin defines.
12752 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
12753 Update the built-in documentation file for the new built-in
12756 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12758 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
12759 (mthumb): Mark as the negative of -marm.
12761 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12763 * doc/invoke.texi (ARM Options, -mcpu): Document supported
12765 (ARM Options, -mtune): Document that this accepts the same
12766 extension options as -mcpu.
12767 (ARM Options, -mfpu): Document addition of -mfpu=auto.
12769 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12771 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
12772 permitted extensions.
12774 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12776 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
12777 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
12778 (armv8-m.main): Add option +nodsp.
12779 * config/arm/arm-cpu-cdata.h: Regenerated.
12781 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12783 * config/arm/t-fuchsia: New file.
12784 * config.gcc (arm*-*-fuchsia*): Use it.
12786 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12788 * config/arm/t-symbian: Rewrite for new option infrastructure.
12790 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12792 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
12793 (MULTILIB_REQUIRED): Likewise.
12795 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12797 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
12798 (MULTILIB_RESUE): Likewise.
12799 (MULTILIB_MATCHES): Likewise.
12800 (MULTLIB_REQUIRED): Likewise.
12802 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12804 * config/arm/t-rtems: Rewrite for new option framework.
12806 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12808 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
12809 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
12810 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
12811 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
12812 * config/arm/t-multilib: ... here.
12813 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
12814 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
12815 armv7-a and armv8*-a when A-profile libraries have not been built.
12816 * config/arm/t-rmprofile: Rewrite.
12818 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12820 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
12821 with a backslash. Remove the backslash after substituting unescaped
12823 * doc/fragments.texi (MULTILIB_REUSE): Document it.
12825 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12827 * config.gcc: (arm*-*-*): When building a-profile libraries, force
12828 the driver to pass through the default setting of -mfloat-abi.
12829 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
12831 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
12832 (all_feat_combs): New rule.
12833 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
12835 * config/arm/t-aprofile: Rewrite.
12837 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12839 * config/arm/arm.h (FPUTYPE_AUTO): Define.
12840 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
12841 fpu is not specified by the user/command-line.
12842 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
12843 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
12844 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
12845 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
12846 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
12847 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
12849 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12851 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
12852 * config/arm/t-arm-elf: Rewritten.
12854 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12856 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
12857 have some floating-point instructions.
12858 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
12859 (TARGET_MAYBE_HARD_FLOAT): New macro.
12860 * config/arm/arm-builtins.c (arm_init_builtins): Use
12861 TARGET_MAYBE_HARD_FLOAT.
12862 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
12864 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12866 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
12867 (configargs.h): Include it.
12868 (arm_print_hint_for_fpu_option): New function.
12869 (arm_parse_fpu_option): New function.
12870 (candidate_extension): New class.
12871 (arm_canon_for_multilib): New function.
12872 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
12873 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
12874 (ARCH_CANONICAL_SPECS): New macro.
12875 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
12877 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12879 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
12880 are set after handling multilib fragments. Set target_cpu_default2
12883 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12885 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
12887 (arm*-*-*): Set target_cpu_default2 to a quoted string.
12888 * config/arm/parsecpu.awk (check_cpu): Validate any extension
12890 (check_arch): Likewise.
12891 * config/arm/arm.c (arm_configure_build_target): Handle
12892 TARGET_CPU_DEFAULT being a string constant. Scan any feature
12893 options in the default.
12895 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12897 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
12898 when an option is an alias of another.
12899 * config/arm/parsecpu.awk (optalias): New parser token.
12900 (gen_comm_data): Mark non-alias options as such. Emit entries
12901 for extension aliases.
12902 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
12903 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
12904 (armv6kz, armv6zk, armv6t2): Likewise.
12905 (armv7): Make vfpv3-d16 an alias.
12906 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
12908 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
12909 Sort in canonical order.
12910 (armv8-a): Sort in canonical order.
12911 (armv8.1-a, armv8.2-a): Likewise.
12912 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
12914 (cortex-a9): Sort in canonical order.
12915 * config/arm/arm.c (selftests.h): Include it.
12916 (arm_test_cpu_arch_data): New function.
12917 (arm_run_self_tests): New function.
12918 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
12919 (targetm): Move declaration to the end of the file.
12920 * arm-cpu-cdata.h: Regenerated.
12922 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12924 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
12925 call to target_mode_check describing the type of option passed.
12926 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
12927 (arm_target_thumb_only): Use arm_parse_arch_option_name or
12928 arm_parse_cpu_option_name to match parameters against list of
12930 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
12931 arm_arch_core_flags data structure.
12932 * config/arm/arm-cpu_cdata.h: Regenerated.
12934 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12936 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
12938 (arm_print_hint_for_cpu_option): Likewise.
12939 (arm_print_hint_for_arch_option): Likewise.
12940 (arm_parse_cpu_option_name): Likewise.
12941 (arm_parse_arch_option_name): Likewise.
12942 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
12943 of entries in the all_fpus list.
12944 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
12945 (arm_parse_cpu_option_name): Declare.
12946 (arm_parse_arch_option_name): Declare.
12947 (arm_parse_option_features): Declare.
12948 (arm_intialize_isa): Declare.
12949 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
12951 (gen_comm_data): ... here. Make definitions non-static.
12952 * config/arm/arm-cpu-data.h: Regenerated.
12953 * config/arm/arm-cpu-cdata.h: Regenerated.
12955 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12957 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
12958 (cpu_arch_extension): New structure.
12959 (cpu_arch_option, arch_option, cpu_option): New structures.
12960 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
12961 architecture types.
12962 (gen_data): Generate new format data tables.
12963 * config/arm/arm.c (cpu_tune): New structure.
12964 (cpu_option, processors): Delete.
12965 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
12966 (arm_print_hint_for_cpu_option): ... this and ...
12967 (arm_print_hint_for_arch_option): ... this.
12968 (arm_parse_arch_cpu_name): Delete. Replace with ...
12969 (arm_parse_cpu_option_name): ... this and ...
12970 (arm_parse_arch_option_name): ... this.
12971 (arm_unrecognized_feature): Change type of target parameter to
12973 (arm_parse_arch_cpu_features): Delete. Replace with ...
12974 (arm_parse_option_features): ... this.
12975 (arm_configure_build_target): Rework to use new configuration data
12977 (arm_print_tune_info): Rework for new configuration data tables.
12978 * config/arm/arm-cpu-data.h: Regenerated.
12979 * config/arm/arm-cpu.h: Regenerated.
12981 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12983 * Makefile.in (OBJS): Move sbitmap.o from here ...
12984 (OBJS-libcommon): ... to here.
12986 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12988 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
12989 (ISA_ALL_CRYPTO): New macro.
12990 (ISA_ALL_SIMD): New macro
12991 (ISA_ALL_FP): New macro.
12992 * config/arm/arm.c (fpu_bitlist): Update initializer.
12993 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
12995 (arm9e): Add fpu. Add option for nofp
12996 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
12997 (arm926ej-s, arm1026ej-s): Likewise.
12998 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
12999 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
13000 neon-fp16, neon-vfpv4, nofp and nosimd.
13001 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
13002 (cortex-a8): Add fpu. Add option for nofp.
13003 (cortex-a9): Add fpu. Add options for nosimd and nofp.
13004 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
13005 (cortex-r4f): Add fpu.
13006 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
13007 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
13009 (cortex-r8): Likewise.
13010 (cortex-m4): Add fpu. Add option for nofp.
13011 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
13012 (cortex-a17.cortex-a7): Likewise.
13013 (cortex-a32): Add fpu. Add options for crypto and nofp.
13014 (cortex-a35, cortex-a53): Likewise.
13015 (cortex-a57): Add fpu. Add option for crypto.
13016 (cortex-a72, cortex-a73): Likewise.
13017 (exynos-m1): Likewise.
13018 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
13019 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
13020 (cortex-m33): Add fpu. Add option for nofp.
13021 * config/arm/arm-cpu-cdata.h: Regenerated
13022 * config/arm/arm-cpu-data.h: Regenerated.
13024 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13026 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
13027 (armv5te, armv5tej): Likewise.
13028 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
13029 (armv7): Add options fp and vfpv3-d16.
13030 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
13031 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
13033 (armv7ve): Likewise.
13034 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
13035 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
13036 (armv8-a): Add nocrypto option.
13037 (armv8.1-a, armv8.2-a): Likewise.
13038 (armv8-m.main): add options fp, fp.dp and nofp.
13040 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13042 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
13044 (armv8-a+crc): Delete.
13045 (armv8.1-a): Add options simd, crypto and nofp.
13046 (armv8.2-a): Add options fp16, simd, crypto and nofp.
13047 (armv8.2-a+fp16): Delete.
13048 (armv8-m.main): Add option dsp.
13049 (armv8-m.main+dsp): Delete.
13050 (cortex-a8): Add fpu. Add nofp option.
13051 (cortex-a9): Add fpu. Add nofp and nosimd options.
13052 * config/arm/parsecpu.awk (gen_data): Generate option tables and
13053 link to main cpu and architecture data structures.
13054 (gen_comm_data): Only put isa attributes from the main architecture
13056 (option): New statement for architecture and CPU entries.
13057 * arm.c (struct cpu_option): New structure.
13058 (struct processors): Add entry for options.
13059 (arm_unrecognized_feature): New function.
13060 (arm_parse_arch_cpu_name): Ignore any characters after the first
13062 (arm_parse_arch_cpu_feature): New function.
13063 (arm_configure_build_target): Separate out any CPU and architecture
13064 features and parse separately. Don't error out if -mfpu=auto is
13065 used with only an architecture string.
13066 (arm_print_asm_arch_directives): New function.
13067 (arm_file_start): Call it.
13068 * config/arm/arm-cpu-cdata.h: Regenerated.
13069 * config/arm/arm-cpu-data.h: Likewise.
13070 * config/arm/arm-tables.opt: Likewise.
13072 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13074 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
13075 assembler when it is not -mfpu=auto.
13077 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13079 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
13080 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
13081 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
13082 (ASM_CPU_SPEC): Rewrite.
13083 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
13084 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
13085 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
13086 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
13087 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
13088 copied string is NUL-terminated. Also strip any characters prefixed
13090 (arm_rewrite_selected_arch): New function.
13091 (arm_rewrite_march): New function.
13093 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13095 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
13096 (x_arm_cpu_string, x_arm_tune_string): Likewise.
13097 (march, mcpu, mtune): Convert to string-based options.
13098 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
13099 (arm_parse_arch_cpu_name): New function.
13100 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
13101 identify selected architecture or CPU.
13102 (arm_option_save): New function.
13103 (TARGET_OPTION_SAVE): Redefine.
13104 (arm_option_restore): Restore string options.
13105 (arm_option_print): Print string options.
13107 2017-06-16 Martin Sebor <msebor@redhat.com>
13109 PR tree-optimization/80933
13110 PR tree-optimization/80934
13111 * builtins.c (fold_builtin_3): Do not handle bcmp here.
13112 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
13113 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
13114 (gimple_fold_builtin): Call them.
13116 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13118 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
13119 as unlikely; update profile.
13121 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13123 * predict.c (force_edge_cold): Handle declaring edges impossible
13126 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13128 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
13130 (try_unroll_loop_completely): Fix reporting.
13132 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13134 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
13136 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
13139 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
13140 if given a non-constant argument for an intrinsic which requires a
13143 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13145 * profile.c (compare_freqs): New function.
13146 (branch_prob): Sort edge list.
13147 (find_spanning_tree): Assume that the list is priority sorted.
13149 2017-06-16 Richard Biener <rguenther@suse.de>
13151 PR tree-optimization/81090
13152 * passes.def (pass_record_bounds): Remove.
13153 * tree-pass.h (make_pass_record_bounds): Likewise.
13154 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
13155 make_pass_record_bounds): Likewise.
13156 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
13157 not free niter estimates at the beginning but at the end.
13158 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
13160 2017-06-16 Richard Biener <rguenther@suse.de>
13162 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
13163 initializer to workaround ICE in host GCC 4.8.
13165 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13167 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
13169 (clone_inlined_nodes): Update.
13171 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
13173 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
13174 prefetch settings, and enable prefetching by default at -O3.
13176 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
13178 * config/aarch64/aarch64.c (aarch64_override_options_internal):
13179 Set flag_prefetch_loop_arrays according to tuning data.
13181 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
13183 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
13184 New tune structure.
13185 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
13186 [Unrelated to main purpose of the patch] Place the pointer field last
13187 to enable type checking errors when tune structure are wrongly merged.
13188 * config/aarch64/aarch64.c (generic_prefetch_tune,)
13189 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
13190 (thunderx2t99_prefetch_tune): New tune constants.
13191 (tune_params *_tunings): Update all tunings (no functional change).
13192 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
13193 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
13194 from tunings structures.
13196 2017-06-16 Jakub Jelinek <jakub@redhat.com>
13199 * ubsan.c (instrument_null): Add T argument, use it instead
13200 of computing it based on IS_LHS.
13201 (instrument_object_size): Likewise.
13202 (pass_ubsan::execute): Adjust instrument_null and
13203 instrument_object_size callers to pass gimple_get_lhs or
13204 gimple_assign_rhs1 result to it. Use instrument_null instead of
13205 calling get_base_address and instrument_mem_ref. Handle
13206 aggregate call arguments for object-size sanitization.
13208 2017-06-16 Yury Gribov <tetra2005@gmail.com>
13210 PR tree-optimization/81089
13211 * tree-vrp.c (is_masked_range_test): Validate operands of
13214 2017-06-15 Martin Sebor <msebor@redhat.com>
13217 * dumpfile.c (dump_register): Avoid calling memset to initialize
13218 a class with a default ctor.
13219 * gcc.c (struct compiler): Remove const qualification.
13220 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
13221 * hash-table.h: Ditto.
13222 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
13224 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
13225 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
13227 * params.h (struct param_info): Make struct members non-const.
13228 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
13229 with default initialization.
13230 * vec.h (vec_copy_construct, vec_default_construct): New helper
13232 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
13233 with vec_copy_construct.
13234 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
13235 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
13236 * doc/invoke.texi (-Wclass-memaccess): Document.
13238 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
13240 * emit-rtl.h (is_leaf): Update comment about local
13241 register allocator.
13243 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13246 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
13247 for a variable to have a section before checking if the section has a
13249 Set section to.persistent if persistent attribute is set.
13250 Warn if .persistent attribute is used on an automatic variable.
13252 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
13254 PR rtl-optimization/80474
13255 * reorg.c (update_block): Do not ignore instructions in a delay slot.
13257 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
13259 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
13262 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
13264 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
13265 (casesi): Emit bounds checking as RTL.
13266 (casesi_internal_mips16_<mode>): Remove bounds checking.
13268 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
13270 * config/xtensa/xtensa.c (xtensa_option_override): Append
13271 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
13272 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
13273 xtensa_doloop_hooks): Define unconditionally.
13274 (xtensa_reorg_loops): Only call reorg_loops in the presence of
13276 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
13277 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
13278 for it in xtensa_option_override.
13279 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
13280 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
13282 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
13284 * doc/cppopts.texi: Document '-' special value to -MF.
13286 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
13288 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
13289 (cortex_a53_fconst): Likewise.
13290 (cortex_a53_fpmul): Likewise.
13291 (cortex_a53_f_load_64): Likewise.
13292 (cortex_a53_f_load_many): Likewise.
13293 (cortex_a53_advsimd_alu): Likewise.
13294 (cortex_a53_advsimd_alu_q): Likewise.
13295 (cortex_a53_advsimd_mul): Likewise.
13296 (cortex_a53_advsimd_mul_q): Likewise.
13297 (fpmac bypass): Add new bypass for fpmac-fpmac case.
13298 Add missing fmul, r2f_cvt and fconst cases.
13300 2017-06-14 Richard Biener <rguenther@suse.de>
13302 PR middle-end/81088
13303 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
13305 (fold_binary_loc): When associating do not treat pre-existing
13306 TREE_OVERFLOW on literal constants as a reason to allow
13307 TREE_OVERFLOW on associated literal constants.
13309 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
13311 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
13312 (MASK_FEATURES): New macro.
13313 * config/sparc/sparc.c (sparc_option_override): Remove the special
13314 handling of -mfpu and generalize it to all MASK_FEATURES switches.
13316 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
13318 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
13319 a division of 0 if non-call exceptions are enabled.
13321 2017-06-14 Andrew Pinski <apinski@cavium.com>
13322 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
13325 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
13326 Improve vector initialization code gen for only variable case.
13328 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
13330 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
13332 2017-06-14 Richard Biener <rguenther@suse.de>
13334 PR tree-optimization/81083
13335 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
13338 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13340 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
13341 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
13342 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
13343 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
13344 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
13345 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
13347 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13349 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
13350 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
13352 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13354 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
13356 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13358 * config/rs6000/t-rtems: Don't handle SPE.
13360 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13362 * config/rs6000/t-linux: Don't handle SPE.
13364 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13366 * config/rs6000/eabispe.h: Delete file.
13368 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13370 * config/rs6000/t-spe: Delete file.
13372 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13374 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
13375 (rs6000_legitimate_offset_address_p): Return false for anything in
13376 V2SImode or V2SFmode.
13378 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13380 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
13381 except V2SF and V2SI. Rearrange the vector modes, and add comments.
13382 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
13384 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
13385 (rs6000_legitimate_offset_address_p): Ditto.
13386 (rs6000_emit_move): Ditto.
13387 (rs6000_init_builtins): Remove V4HI_type_node.
13389 2017-06-13 Martin Liska <mliska@suse.cz>
13392 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
13393 (gate_asan): Likewise.
13394 * asan.h (asan_no_sanitize_address_p): Remove the function.
13395 (sanitize_flags_p): New function.
13396 * builtins.def: Fix coding style.
13397 * common.opt: Use renamed enum value.
13398 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
13399 * doc/extend.texi: Document no_sanitize attribute.
13400 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
13401 to SANITIZE_UNDEFINED_NONDEFAULT.
13402 * gcc.c (sanitize_spec_function): Use the renamed enum value.
13403 * gimple-fold.c (optimize_atomic_compare_exchange_p):
13404 Use sanitize_flags_p.
13405 * gimplify.c (gimplify_function_tree): Likewise.
13406 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
13407 * opts.c (parse_no_sanitize_attribute): New function.
13408 (common_handle_option): Use renamed enum value.
13409 * opts.h (parse_no_sanitize_attribute): Declare.
13410 * tree.c (sanitize_flags_p): New function.
13411 * tree.h: Declared here.
13412 * tsan.c: Use sanitize_flags_p.
13413 * ubsan.c (ubsan_expand_null_ifn): Likewise.
13414 (instrument_mem_ref): Likewise.
13415 (instrument_bool_enum_load): Likewise.
13416 (do_ubsan_in_current_function): Remove the function.
13417 (pass_ubsan::execute): Use sanitize_flags_p.
13418 * ubsan.h: Remove do_ubsan_in_current_function
13419 * tree-cfg.c (print_no_sanitize_attr_value): New function.
13420 (dump_function_to_file): Use it here.
13422 2017-06-13 Martin Jambor <mjambor@suse.cz>
13424 PR tree-optimization/80803
13425 PR tree-optimization/81063
13426 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
13427 (propagate_subaccesses_across_link): Enqueue subtree whenever
13428 necessary instead of relying on the caller.
13430 2017-06-13 Martin Jambor <mjambor@suse.cz>
13432 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
13433 that have a first_link.
13434 (sort_and_splice_var_accesses): Do not check first_link before
13436 (subtree_mark_written_and_enqueue): Likewise.
13437 (propagate_all_subaccesses): Likewise and do not stop at first
13438 parent with a first_link.
13440 2017-06-13 Martin Jambor <mjambor@suse.cz>
13442 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
13445 2017-06-13 Yury Gribov <tetra2005@gmail.com>
13447 * match.pd: New pattern.
13449 2017-06-13 Yury Gribov <tetra2005@gmail.com>
13451 * tree-vrp.c (is_masked_range_test): New function.
13452 (register_edge_assert_for): Determine ranges for
13455 2017-06-13 Yury Gribov <tetra2005@gmail.com>
13457 PR tree-optimization/67328
13458 * fold-const.c (maskable_range_p): New function.
13459 (build_range_check): Generate bittests if possible.
13461 2017-06-13 Martin Liska <mliska@suse.cz>
13463 * gimple-pretty-print.c (dump_probability): Add new argument.
13464 (dump_edge_probability): Dump both probability and count.
13465 (dump_gimple_label): Likewise.
13466 (dump_gimple_bb_header): Likewise.
13468 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
13471 * config/avr/avr-devices.c: Fix indentation.
13472 * config/avr/gen-avr-mmcu-specs.c: Dito.
13474 2017-06-13 Richard Biener <rguenther@suse.de>
13476 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
13477 instead get vector type from stmt_info.
13478 (vectorizable_reduction): Adjust. Remove dead code.
13480 2017-06-13 Richard Biener <rguenther@suse.de>
13482 PR middle-end/81065
13483 * fold-const.c (extract_muldiv_1): Remove bogus distribution
13484 case of C * (x * C2 + C3).
13485 (fold_addr_of_array_ref_difference): Properly fold index difference.
13487 2017-06-12 David S. Miller <davem@davemloft.net>
13490 * config/sparc/sparc.md (return expander): Emit frame blockage if
13491 function uses alloca.
13493 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
13495 * combine.c (make_field_assignment): Check len rather than the mode
13496 precision when calling force_to_mode.
13498 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
13500 Support multilibs and devices that see flash in RAM address range.
13503 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
13504 (avr_mcu_t) <flash_pm_offset>: New field.
13505 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
13506 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
13507 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
13508 (AVR_TINY_PM_OFFSET): Remove macro.
13509 * config/avr/avr.opt (-mshort-calls): New option.
13510 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
13511 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
13512 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
13513 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
13514 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
13515 instead of avr_arch->have_jmp_call.
13516 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
13517 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
13518 avr_arch->flash_pm_offset to define.
13519 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
13520 new field flash_pm_offset. Add entry for avrxmega3.
13521 (avr_texinfo): Add entry for avrxmega3.
13522 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
13523 attiny212, attiny214,
13524 attiny412, attiny414, attiny416, attiny417,
13525 attiny814, attiny816, attiny817,
13526 attiny1614, attiny1616, attiny1617,
13527 attiny3214, attiny3216, attiny3217.
13528 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
13529 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
13530 (avr_print_operand_address) [AVR_TINY]: Same.
13531 (avr_asm_init_sections) <readonly_data_section>: Only patch
13532 callback if avr_arch->flash_pm_offset = 0.
13533 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
13534 for rodata if avr_arch->flash_pm_offset != 0.
13535 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
13536 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
13537 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
13538 (m_options): Append opt_rcall.
13539 (m_dirnames): Append dir_rcall.
13540 * config/avr/t-multilib: Regenerate.
13542 * configure.ac [target=avr]: Check whether avrxmega3 default
13543 linker description file works as needed.
13544 * configure: Regenerate.
13545 * doc/avr-mmcu.texi: Regenerate.
13546 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
13547 <__AVR_ARCH__>: Document avrxmega3 and 103.
13548 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
13549 <__AVR_SHORT_CALLS__>: Document it.
13550 <__AVR_PM_BASE_ADDRESS__>: Document it.
13551 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
13552 (AVR Variable Attributes) <progmem>: Document this is
13553 not needed for avrxmega3.
13554 (AVR Named Address Spaces) <__flash>: Dito.
13556 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
13558 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
13560 2017-06-12 Doug Rupp <rupp@adacore.com>
13562 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
13563 Append vxworks-stdint.h to the tm_file list.
13564 * config/vxworks-stdint.h: New file.
13566 2017-06-12 Martin Liska <mliska@suse.cz>
13568 PR tree-optimization/81041
13569 * tree-profile.c (gimple_gen_ic_func_profiler):
13570 Create an extra BB in profile-generate
13571 (gimple_gen_time_profiler): Likewise.
13573 2017-06-12 Jakub Jelinek <jakub@redhat.com>
13575 PR tree-optimization/81003
13576 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
13577 (update_range_test): Use it instead of force_gimple_operand_gsi.
13579 2017-06-12 Richard Biener <rguenther@suse.de>
13581 PR tree-optimization/81053
13582 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
13583 with backedge value not defined in loop. Simplify def stmt
13586 2017-06-11 Tom de Vries <tom@codesourcery.com>
13589 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
13591 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
13592 nvptx_cannot_force_const_mem.
13594 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
13596 * opts.c (finish_options): Move test for flag_split_stack after
13597 it has been initialized.
13599 2017-06-11 Jason Merrill <jason@redhat.com>
13601 * tree.h (id_equal): New.
13602 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
13603 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
13604 instead of strcmp of IDENTIFIER_POINTER.
13606 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
13608 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
13609 (mark_all_inlined_calls_cdtor): Fix formating.
13610 (inline_transform): Rescale profile before inlining.
13612 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
13614 * cgraph.h (cgraph_edge::clone): Update prototype.
13615 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
13616 (cgraph_node::create_clone): Update.
13617 (cgraph_node::create_version_clone): Update.
13618 * tree-inline.c (copy_bb): Update.
13619 (expand_call_inline): Update.
13621 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
13623 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
13624 factored out from ...
13625 (rs6000_emit_prologue): ... here.
13627 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
13629 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
13630 factored out from ...
13631 (rs6000_emit_prologue): ... here.
13633 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
13635 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
13637 (handle_missing_profiles): Fix computation of tp_first_run.
13638 (counts_to_freqs): Do not touch freqs when count is 0.
13640 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
13642 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
13645 2017-06-10 Tom de Vries <tom@codesourcery.com>
13647 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
13648 attributes): Document signal effective target.
13650 2017-06-10 Tom de Vries <tom@codesourcery.com>
13652 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
13653 Document effective target stack_size.
13655 2017-06-09 David Malcolm <dmalcolm@redhat.com>
13657 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
13658 to the edit_context if they can be auto-applied.
13660 2017-06-9 Ian Lance Taylor <iant@golang.org>
13662 * opts.c (finish_options): If -fsplit-stack, disable implicit
13663 -forder-blocks-and-partition.
13664 * doc/invoke.texi (Optimize Options): Document that when using
13665 -fsplit-stack -forder-blocks-and-partition is not implicitly
13668 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
13670 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
13671 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
13672 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
13673 * builtins.def (abort, trap, unreachable): Declare cold.
13674 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
13675 * tree-core.h (ECF_COLD): New.
13676 * tree.c (set_call_expr_flags): Handle ECF_COLD.
13677 (build_common_builtin_nodes): Mark unreachable and abort as cold.
13679 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
13681 * predict.c (unlikely_executed_stmt_p): Cleanup.
13683 2017-06-09 Richard Biener <rguenther@suse.de>
13685 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
13686 model if the ref is always written to.
13688 2017-06-09 Tamar Christina <tamar.christina@arm.com>
13690 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
13692 2017-06-09 Tamar Christina <tamar.christina@arm.com>
13694 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
13697 2017-06-09 Tom de Vries <tom@codesourcery.com>
13700 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
13701 "target cannot support label values" when encountering LABEL_REF.
13703 2017-06-09 Martin Liska <mliska@suse.cz>
13705 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
13706 (gimple_gen_ic_func_profiler): Emit direct comparison
13707 of __gcov_indirect_call_callee with NULL.
13708 (gimple_gen_time_profiler): Change probability from
13709 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
13711 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
13713 * profile.c (edge_gcov_counts): Turn to pointer.
13714 (compute_branch_probabilities, compute_branch_probabilities): Update.
13715 (branch_prob): Do not clear edge_gcov_count.
13716 * profile.h (edge_gcov_counts): Turn to pointer.
13717 (edge_gcov_count): Update.
13719 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
13721 * gimple.h (gimple_check_failed): Mark cold.
13723 2017-06-09 Richard Biener <rguenther@suse.de>
13725 PR tree-optimization/66623
13726 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
13727 refactor check_reduction into two parts, properly computing
13728 whether we have to check reduction validity for outer loop
13731 2017-06-09 Richard Biener <rguenther@suse.de>
13733 PR tree-optimization/79483
13734 * graphite-scop-detection.c (order): New global.
13735 (get_order): Compute bb to order mapping that satisfies code
13736 generation constraints.
13737 (cmp_pbbs): New helper.
13738 (build_scops): Start domwalk at entry block, sort generated
13741 2017-06-09 Richard Biener <rguenther@suse.de>
13743 PR middle-end/81007
13744 * ipa-polymorphic-call.c
13745 (ipa_polymorphic_call_context::restrict_to_inner_class):
13746 Skip FIELD_DECLs with error_mark_node type.
13747 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
13750 2017-06-09 Martin Liska <mliska@suse.cz>
13752 * predict.c (struct branch_predictor): New struct.
13753 (test_prediction_value_range): New test.
13754 (predict_c_tests): New function.
13755 * selftest-run-tests.c (selftest::run_tests): Run the function.
13756 * selftest.h: Declare new tests.
13758 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
13761 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
13762 gen_add3_insn did not fail.
13763 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
13764 r0, construct that number in a temporary reg and add that reg to r0.
13765 If asked to put the result in r0 as well, fail.
13767 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
13769 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13770 for early expansion of vec_eqv.
13772 2017-06-08 Jakub Jelinek <jakub@redhat.com>
13774 PR middle-end/81005
13775 * ubsan.c (instrument_null): Avoid pointless code temporary.
13776 (pass_ubsan::execute): Instrument aggregate arguments of calls.
13778 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
13782 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
13785 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
13788 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13790 * predict.c (unlikely_executed_edge_p): Move ahead.
13791 (probably_never_executed_edge_p): Use it.
13793 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
13795 PR middle-end/79988
13796 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
13797 gimple_call_builtin_p call.
13799 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13801 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
13802 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
13803 rtl_check_failed_type2, rtl_check_failed_code1,
13804 rtl_check_failed_code2, rtl_check_failed_code_mode,
13805 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
13806 rtvec_check_failed_bounds, rtl_check_failed_flag,
13807 _fatal_insn_not_found, _fatal_insn): Likewise.
13808 * tree.h (tree_contains_struct_check_failed,
13809 tree_check_failed, tree_not_check_failed,
13810 tree_class_check_failed, tree_range_check_failed,
13811 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
13812 tree_vec_elt_check_failed, phi_node_elt_check_failed,
13813 tree_operand_check_failed, omp_clause_check_failed,
13814 omp_clause_operand_check_failed, omp_clause_range_check_failed):
13817 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13819 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
13820 flag_branch_probabilities.
13821 * ipa-inline.c (edge_badness): Likewise.
13822 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
13823 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
13824 * predict.c (maybe_hot_frequency_p): Likewise.
13825 (probably_never_executed): Likewise.
13826 * sched-ebb.c (schedule_ebbs): Likewise.
13827 * sched-rgn.c (find_single_block_region): Likewise.
13828 * tracer.c (tail_duplicate): Likewise.
13830 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13832 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
13833 longer requires x_flag_profile_use.
13835 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13837 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
13838 instead of flag_reorder_blocks_and_partition.
13839 * dbxout.c (dbxout_function_end): Likewise.
13840 * dwarf2out.c (gen_subprogram_die): Likewise.
13841 * haifa-sched.c (sched_create_recovery_edges): Likewise.
13842 * hw-doloop.c (reorg_loops): Likewise.
13843 * varasm.c (assemble_start_function,
13844 assemble_end_function): Likewise.
13845 (decide_function_section): Do not check for
13846 flag_reorder_blocks_and_partition.
13848 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
13850 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
13852 (chkp_get_hard_register_fake_addr_expr): Ditto.
13853 (chkp_build_addr_expr): Add check for hard reg case.
13854 (chkp_parse_array_and_component_ref): Ditto.
13855 (chkp_find_bounds_1): Ditto.
13856 (chkp_process_stmt): Don't generate bounds store for
13859 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13861 * predict.c (maybe_hot_bb_p): Do not check profile status.
13862 (maybe_hot_edge_p): Likewise.
13863 (probably_never_executed): Check for zero counts even if profile
13865 (unlikely_executed_edge_p): New function.
13866 (unlikely_executed_stmt_p): New function.
13867 (unlikely_executed_bb_p): New function.
13868 (set_even_probabilities): Use unlikely predicates.
13869 (combine_predictions_for_bb): Likewise.
13870 (predict_paths_for_bb): Likewise.
13871 (predict_paths_leading_to_edge): Likewise.
13872 (determine_unlikely_bbs): New function.
13873 (estimate_bb_frequencies): Use it.
13874 (compute_function_frequency): Use zero counts even if profile is
13876 * profile-count.h: Fix typo.
13878 2017-08-08 Julia Koval <julia.koval@intel.com>
13880 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
13881 _mm512_mask_cvtsepi16_storeu_epi8,
13882 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
13883 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
13884 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
13885 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
13886 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
13887 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
13888 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
13889 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
13890 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
13891 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
13892 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
13893 __builtin_ia32_pmovuswb256mem_mask,
13894 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
13895 __builtin_ia32_pmovwb512mem_mask): New builtins.
13897 2017-08-08 Julia Koval <julia.koval@intel.com>
13899 PR target/73350,80862
13900 * config/i386/subst.md (round): Fix round pattern.
13901 * config/i386/i386.c (ix86_erase_embedded_rounding):
13902 Fix erasing rounding for the fixed pattern.
13904 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13906 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
13908 2017-06-08 Martin Liska <mliska@suse.cz>
13910 PR gcov-profile/80911
13911 * gcov.c (block_info::block_info): New constructor.
13913 2017-06-07 Carl Love <cel@us.ibm.com>
13915 * config/rs6000/rs6000-c: The return type of the following
13916 built-in functions was implemented as int not long long. Fix sign
13917 of return value for the unsigned version of vec_mulo and vec_mule.
13918 vector unsigned long long vec_bperm (vector unsigned long long,
13919 vector unsigned char)
13920 vector signed long long vec_mule (vector signed int,
13922 vector unsigned long long vec_mule (vector unsigned int,
13923 vector unsigned int)
13924 vector signed long long vec_mulo (vector signed int,
13926 vector unsigned long long vec_mulo (vector unsigned int,
13927 vector unsigned int)
13928 * doc/extend.texi: Fix the documentation for the built-in
13931 2017-06-07 Carl Love <cel@us.ibm.com>
13934 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
13937 2017-06-07 Carl Love <cel@us.ibm.com>
13939 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
13940 support, Generate doublehv for signed int/float for BE case only.
13942 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
13944 * doc/invoke.texi (mcx16): Rewrite.
13946 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13948 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
13949 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
13950 *mov<mode>_softfloat, and an anonymous splitter): Use
13951 nonimmediate_operand instead of rs6000_nonimmediate_operand.
13953 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13955 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
13957 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
13958 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
13959 (rs6000_debug_reg_global): Adjust.
13960 (rs6000_init_hard_regno_mode_ok): Adjust.
13961 (rs6000_dbx_register_number): Adjust.
13962 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
13963 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
13964 Remove SPE_ACC and SPEFSCR.
13965 (REG_ALLOC_ORDER): Ditto.
13966 (FRAME_POINTER_REGNUM): Change to 111.
13967 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
13968 (REG_CLASS_NAMES): Ditto.
13969 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
13970 (REGISTER_NAMES): Ditto.
13971 (ADDITIONAL_REG_NAMES): Ditto.
13972 (rs6000_reg_names): Ditto.
13973 * config/rs6000/rs6000.md: Renumber some register number
13976 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13978 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
13980 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
13981 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
13983 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
13984 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
13985 Delete the SPE high registers.
13986 (REG_ALLOC_ORDER): Ditto.
13987 (enum reg_class): Remove SPE_HIGH_REGS.
13988 (REG_CLASS_NAMES): Ditto.
13989 (REG_CLASS_CONTENTS): Delete the SPE high registers.
13990 (REGISTER_NAMES): Ditto.
13991 (rs6000_reg_names): Ditto.
13992 * doc/tm.texi.in: Remove SPE as example.
13993 * doc/tm.texi: Regenerate.
13995 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13997 * config/rs6000/8540.md (ppc8540_brinc): Delete.
13998 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
13999 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
14000 * config/rs6000/rs6000.md (type): Remove "brinc".
14002 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14004 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
14005 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
14006 * config/rs6000/linuxspe.h: Delete file.
14007 * config/rs6000/rs6000.md: Don't include spe.md.
14008 * config/rs6000/spe.h: Delete file.
14009 * config/rs6000/spe.md: Delete file.
14010 * config/rs6000/t-rs6000: Remove spe.md.
14012 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14014 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
14015 (reg_or_none500mem_operand): Delete.
14016 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
14017 instead of reg_or_none500mem_operand.
14019 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14021 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
14022 handling of SPE flags.
14023 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
14025 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14027 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
14029 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
14030 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
14031 paired_divv2sf3): Similar.
14032 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
14033 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
14034 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
14036 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
14037 Rename the paired_* instruction patterns.
14038 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
14040 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
14041 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
14042 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
14043 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
14044 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
14045 PAIRED_VECTOR_MODE.
14046 (struct machine_function): Delete field spe_insn_chain_scanned_p.
14047 (spe_func_has_64bit_regs_p): Delete.
14048 (spe_expand_predicate_builtin): Delete.
14049 (spe_expand_evsel_builtin): Delete.
14050 (TARGET_DWARF_REGISTER_SPAN): Do not define.
14051 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
14052 (invalid_e500_subreg): Delete.
14053 (rs6000_legitimize_address): Always force_reg op2 as well, for
14054 paired single memory accesses.
14055 (rs6000_member_type_forces_blk): Delete.
14056 (rs6000_spe_function_arg): Delete.
14057 (rs6000_expand_unop_builtin): Delete SPE handling.
14058 (rs6000_expand_binop_builtin): Ditto.
14059 (spe_expand_stv_builtin): Delete.
14060 (bdesc_2arg_spe): Delete.
14061 (spe_expand_builtin): Delete.
14062 (spe_expand_predicate_builtin): Delete.
14063 (spe_expand_evsel_builtin): Delete.
14064 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
14065 (spe_init_builtins): Delete.
14066 (spe_func_has_64bit_regs_p): Delete.
14067 (savres_routine_name): Delete "info" parameter. Adjust callers.
14068 (rs6000_emit_stack_reset): Ditto.
14069 (rs6000_dwarf_register_span): Delete.
14070 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
14071 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
14072 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
14074 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
14076 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
14077 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
14078 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
14079 mulv2sf3, divv2sf3): Delete expanders.
14081 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14083 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
14085 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14087 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
14088 * config/rs6000/rs6000.c: Ditto.
14090 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14092 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
14093 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
14094 comparison_operator.
14096 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14098 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
14099 * config/rs6000/rs6000.opt: Ditto.
14100 * config/rs6000/t-rtems: Ditto.
14102 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14104 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
14105 TARGET_E500_SINGLE by 0, simplify.
14106 * config/rs6000/rs6000.c: Ditto.
14107 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
14108 (spe_build_register_parallel): Delete.
14109 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
14110 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
14111 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
14112 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
14113 (E500_CONVERT): Delete.
14114 * config/rs6000/spe.md: Remove many patterns and all define_constants.
14116 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14118 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
14119 * config/rs6000/dfp.md: Ditto.
14120 (negdd2, *negdd2_fpr): Merge.
14121 (absdd2, *absdd2_fpr): Merge.
14122 (negtd2, *negtd2_fpr): Merge.
14123 (abstd2, *abstd2_fpr): Merge.
14124 * config/rs6000/e500.h: Delete file.
14125 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
14126 TARGET_FPRS by 1 and simplify.
14127 * config/rs6000/rs6000-c.c: Ditto.
14128 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
14129 TARGET_DF_SPE by 0.
14130 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
14132 * config/rs6000/rs6000.md: Ditto.
14133 (floatdidf2, *floatdidf2_fpr): Merge.
14134 (move_from_CR_gt_bit): Delete.
14135 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
14136 (E500_CR_IOR_COMPARE): Delete.
14137 (All patterns that require !TARGET_FPRS): Delete.
14138 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
14140 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14142 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
14144 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14146 * graphds.c (add_edge): Intitialize edge's attached data.
14147 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
14148 pointer parameter. Call pointed function on each edge during
14149 graph traversing. Skip traversing the edge when the function
14151 (graphds_dfs, graphds_scc): Ditto.
14152 (for_each_edge): New parameter. Pass the new parameter to callback
14154 * graphds.h (skip_edge_callback): New function pointer type.
14155 (graphds_dfs, graphds_scc): New function pointer parameter.
14156 (graphds_edge_callback, for_each_edge): New parameter.
14158 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14160 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
14161 out code checking if runtime alias check is possible to below ...
14162 Call the new function.
14163 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
14164 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
14166 2017-06-07 Marek Polacek <polacek@redhat.com>
14169 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
14170 TYPE_OVERFLOW_WRAPS check.
14172 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14174 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
14175 if versioning is required.
14176 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
14177 peeling with the check for versioning.
14179 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14181 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
14182 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
14183 Set true to new parameter if new ssa variable is defined.
14184 (vect_gen_vector_loop_niters): Refactor. Set range information
14185 for the new vector loop bound variable.
14186 (vect_do_peeling): Ditto.
14188 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14190 * tree-affine.c (ssa.h): Include header file.
14191 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
14192 has wrapping overflow behavior.
14194 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14196 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
14198 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14200 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
14201 (tree_to_aff_combination): ... here.
14203 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14205 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
14206 reg_pressure model function.
14207 (ivopts_global_cost_for_size): Delete.
14208 (determine_set_costs, iv_ca_recount_cost): Call new model function
14209 ivopts_estimate_reg_pressure.
14211 2017-06-07 Tamar Christina <tamar.christina@arm.com>
14213 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
14214 expensive than udiv. Remove floating point cases from mod.
14216 2017-06-07 Tamar Christina <tamar.christina@arm.com>
14218 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
14219 Increase idiv cost.
14221 2017-06-07 Tamar Christina <tamar.christina@arm.com>
14223 * config/aarch64/aarch64.md
14224 (copysignsf3): Fix mask generation.
14226 2017-06-07 Jakub Jelinek <jakub@redhat.com>
14228 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
14230 (class dump_manager): Add register_dumps method.
14231 * dumpfile.c: Include langhooks.h.
14232 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
14233 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
14234 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
14235 (dump_manager::dump_register): Start with 512 entries instead of 32.
14236 (dump_manager::register_dumps): New method.
14237 * toplev.c (general_init): Instead of invoking register_dumps
14238 langhook, invoke register_dumps method on the dump manager.
14239 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
14242 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
14244 * doc/md.texi: Clarify the restrictions on a define_insn condition.
14245 Say that # requires an associated define_split to exist, and that
14246 the define_split must be suitable for use after register allocation.
14248 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14250 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
14251 (compute_outgoing_frequencies): Also initialize zero counts.
14252 (find_many_sub_basic_blocks): Do not produce uninitialized profile
14253 around loops; preserve more of profile when nothing changes.
14255 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
14257 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
14259 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
14260 * config/arm/arm-cpu-cdata.h: Regenerate.
14261 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
14262 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
14263 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
14264 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
14265 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
14267 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
14269 * config/arm/t-rmprofile: Likewise.
14270 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
14272 2017-06-06 David S. Miller <davem@davemloft.net>
14275 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
14276 blockage if function uses alloca.
14278 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
14280 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
14281 New "uid" fields to hold pretty-print IDs of group and ref.
14282 Memory references are now identified as <group_id>:<ref_id>
14283 instead of using [random] addresses.
14284 (dump_mem_details): Simplify, no functional change.
14285 (dump_mem_ref): Simplify and make output more concise.
14286 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
14287 (find_or_create_group): Initialize group uid.
14288 (record_ref): Initialize ref uid. Improve debug output.
14289 (prune_group_by_reuse, should_issue_prefetch_p,)
14290 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
14291 (mark_nontemporal_store, determine_loop_nest_reuse):
14292 Improve debug output.
14294 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
14296 * dbgcnt.def (prefetch): New debug counter.
14297 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
14298 (schedule_prefetches): Stop issueing prefetches if debug counter
14301 2017-06-06 Tom de Vries <tom@codesourcery.com>
14303 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
14304 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
14306 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14308 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
14309 Use aarch64_reg_or_zero predicate for operand 4.
14310 (aarch64_compare_and_swap<mode> define_insn_and_split):
14311 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
14312 (aarch64_store_exclusive<mode>): Likewise for operand 2.
14314 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
14316 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
14317 (arm_compute_save_core_reg_mask): This.
14318 (thumb1_compute_save_reg_mask): Rename into ...
14319 (thumb1_compute_save_core_reg_mask): This.
14320 (arm_compute_save_reg0_reg12_mask): Adapt comment.
14321 (arm_compute_frame_layout): Likewise.
14323 2017-06-06 Richard Biener <rguenther@suse.de>
14325 PR tree-optimization/80974
14326 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
14327 keep or clear leaders SSA info.
14329 2017-06-06 Tom de Vries <tom@codesourcery.com>
14331 * config/nvptx/nvptx.c (split_mode_p): New function.
14332 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
14334 2017-06-06 Tom de Vries <tom@codesourcery.com>
14336 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
14338 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14341 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
14344 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14346 * shrink-wrap.c (handle_simple_exit): Update profile.
14347 (try_shrink_wrapping): Upate profile.
14349 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14351 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
14352 (tree_guess_outgoing_edge_probabilities): New.
14353 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
14354 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
14356 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14358 * ipa-split.c (split_function): Initialize return bb profile.
14360 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14362 * profile.c (compute_branch_probabilities): Also initialize
14363 EXIT_BLOCK profile.
14365 2017-06-06 Richard Biener <rguenther@suse.de>
14367 PR tree-optimization/80928
14368 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
14369 (vect_analyze_loop_operations): Properly guard analysis for
14371 (vect_transform_loop): Likewise.
14372 (vect_analyze_loop_2): Also reset SLP type on PHIs.
14373 (vect_model_induction_cost): Do not cost for pure SLP.
14374 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
14375 of induction in inner loop vectorization.
14376 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
14377 (vect_get_and_check_slp_defs): Handle vect_induction_def.
14378 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
14380 (vect_analyze_slp_cost_1): Cost induction.
14381 (vect_detect_hybrid_slp_stmts): Handle PHIs.
14382 (vect_get_slp_vect_defs): Likewise.
14383 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
14384 (vect_transform_stmt): Handle SLP reductions.
14385 * tree-vectorizer.h (vectorizable_induction): Adjust.
14387 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
14389 * config/rs6000/rs6000.c (make_resolver_func): Update
14390 init_lowered_empty_function call.
14392 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
14394 * doc/invoke.texi: Document the -fprofile-abs-path option.
14395 * common.opt (fprofile-abs-path): New option.
14396 * gcov-io.h (gcov_write_filename): Declare.
14397 * gcov-io.c (gcov_write_filename): New function.
14398 * coverage.c (coverage_begin_function): Use gcov_write_filename.
14399 * profile.c (output_location): Likewise.
14401 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
14403 * shring-wrap.c: Revert accidental commit.
14405 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
14407 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
14409 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
14411 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
14413 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
14414 profile in callgraph edge.
14415 * profile-count.h (apply_probability): If THIS is 0, then result is 0
14416 (apply_scale): Likewise.
14417 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
14418 Also scale profile when inlining function with zero profile.
14419 (initialize_cfun): Update exit block profile even when it is zero.
14420 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
14421 when profile is read.
14423 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
14425 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
14426 (CLONE_*): New constants to define the processors we can generate
14427 code for with the target_clone attribute.
14428 (rs6000_clone_map): New array to identify which clone processors
14429 the current program is running on.
14430 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
14431 target_clone attribute.
14432 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
14433 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
14434 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
14435 (cpu_expand_builtin): Add support for target_clone attribute.
14436 (rs6000_valid_attribute_p): Allow "default" attribute.
14437 (get_decl_name): New debug function to simplify printing the
14438 current function name in debugging statements.
14439 (rs6000_clone_priority): New functions to support the target_clone
14440 attribute, and be able to generate code to switch between ISA 2.05
14441 through ISA 3.0 (power6 through power9).
14442 (rs6000_compare_version_priority): Likewise.
14443 (rs6000_get_function_versions_dispatcher): Likewise.
14444 (make_resolver_func): Likewise.
14445 (add_condition_to_bb): Likewise.
14446 (dispatch_function_versions): Likewise.
14447 (rs6000_generate_version_dispatcher_body): Likewise.
14448 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
14449 (fusion_gpr_load_p): Fix a spacing issue.
14450 * doc/extend.texi (Common Function Attributes): Document that the
14451 PowerPC supports the target_clone attribute.
14453 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
14455 * config/arm/arm.h: explain F symbol found in description of ARM
14456 register allocation in its legend.
14458 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
14460 * config/mips/frame-header-opt.c: Include profile-count.h.
14461 * config/riscv/riscv.c: Include profile-count.h
14463 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
14465 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
14467 (sm_set_flag_if_changed): Add bbs field.
14468 (execute_sm_if_changed_flag_set): Pass BBS.
14469 (execute_sm): Update.
14471 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14473 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
14476 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14478 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
14479 (peephole2): New peephole2 to emit the above.
14480 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
14482 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14484 * config/aarch64/aarch64.c (define_peephole2 above
14485 *sub_<shift>_<mode>): New peephole.
14487 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
14489 * config/i386/i386.c (make_resolver_func): Update.
14490 * Makefile.in: Add profile-count.h and profile-count.o
14491 * auto-profile.c (afdo_indirect_call): Update to new API.
14492 (afdo_set_bb_count): Update.
14493 (afdo_propagate_edge): Update.
14494 (afdo_propagate_circuit): Update.
14495 (afdo_calculate_branch_prob): Update.
14496 (afdo_annotate_cfg): Update.
14497 * basic-block.h: Include profile-count.h
14498 (struct edge_def): Turn count to profile_count.
14499 (struct basic_block_def): Likewie.
14500 (REG_BR_PROB_BASE): Move to profile-count.h
14501 (RDIV): Move to profile-count.h
14502 * bb-reorder.c (max_entry_count): Turn to profile_count.
14503 (find_traces): Update.
14504 (rotate_loop):Update.
14505 (connect_traces):Update.
14506 (sanitize_hot_paths):Update.
14507 * bt-load.c (migrate_btr_defs): Update.
14508 * cfg.c (RDIV): Remove.
14509 (init_flow): Use alloc_block.
14510 (alloc_block): Uninitialize count.
14511 (unchecked_make_edge): Uninitialize count.
14512 (check_bb_profile): Update.
14513 (dump_edge_info): Update.
14514 (dump_bb_info): Update.
14515 (update_bb_profile_for_threading): Update.
14516 (scale_bbs_frequencies_int): Update.
14517 (scale_bbs_frequencies_gcov_type): Update.
14518 (scale_bbs_frequencies_profile_count): New.
14519 * cfg.h (update_bb_profile_for_threading): Update.
14520 (scale_bbs_frequencies_profile_count): Declare.
14521 * cfgbuild.c (compute_outgoing_frequencies): Update.
14522 (find_many_sub_basic_blocks): Update.
14523 * cfgcleanup.c (try_forward_edges): Update.
14524 (try_crossjump_to_edge): Update.
14525 * cfgexpand.c (expand_gimple_tailcall): Update.
14526 (construct_exit_block): Update.
14527 * cfghooks.c (verify_flow_info): Update.
14528 (dump_bb_for_graph): Update.
14529 (split_edge): Update.
14530 (make_forwarder_block): Update.
14531 (duplicate_block): Update.
14532 (account_profile_record): Update.
14533 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
14534 (get_estimated_loop_iterations): Update.
14535 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
14536 (single_likely_exit): Update.
14537 * cfgloopmanip.c (scale_loop_profile): Update.
14539 (set_zero_probability): Update.
14540 (lv_adjust_loop_entry_edge): Update.
14541 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
14542 (purge_dead_edges): Update.
14543 (rtl_account_profile_record): Update.
14544 * cgraph.c (cgraph_node::create): Uninitialize count.
14545 (symbol_table::create_edge): Uninitialize count.
14546 (cgraph_update_edges_for_call_stmt_node): Update.
14547 (cgraph_edge::dump_edge_flags): Update.
14548 (cgraph_node::dump): Update.
14549 (cgraph_edge::maybe_hot_p): Update.
14550 * cgraph.h: Include profile-count.h
14551 (create_clone), create_edge, create_indirect_edge): Update.
14552 (cgraph_node): Turn count to profile_count.
14553 (cgraph_edge0: Likewise.
14554 (make_speculative, clone): Update.
14555 (create_edge): Update.
14556 (init_lowered_empty_function): Update.
14557 * cgraphclones.c (cgraph_edge::clone): Update.
14558 (duplicate_thunk_for_node): Update.
14559 (cgraph_node::create_clone): Update.
14560 * cgraphunit.c (cgraph_node::analyze): Update.
14561 (cgraph_node::expand_thunk): Update.
14562 * final.c (dump_basic_block_info): Update.
14563 * gimple-streamer-in.c (input_bb): Update.
14564 * gimple-streamer-out.c (output_bb): Update.
14565 * graphite.c (print_global_statistics): Update.
14566 (print_graphite_scop_statistics): Update.
14567 * hsa-brig.c: Include basic-block.h.
14568 * hsa-dump.c: Include basic-block.h.
14569 * hsa-gen.c (T sum_slice): Update.
14570 (convert_switch_statements):Update.
14571 * hsa-regalloc.c: Include basic-block.h.
14572 * ipa-chkp.c (chkp_produce_thunks): Update.
14573 * ipa-cp.c (struct caller_statistics): Update.
14574 (init_caller_stats): Update.
14575 (gather_caller_stats): Update.
14576 (ipcp_cloning_candidate_p): Update.
14577 (good_cloning_opportunity_p): Update.
14578 (get_info_about_necessary_edges): Update.
14579 (dump_profile_updates): Update.
14580 (update_profiling_info): Update.
14581 (update_specialized_profile): Update.
14582 (perhaps_add_new_callers): Update.
14583 (decide_about_value): Update.
14584 (ipa_cp_c_finalize): Update.
14585 * ipa-devirt.c (struct odr_type_warn_count): Update.
14586 (struct decl_warn_count): Update.
14587 (struct final_warning_record): Update.
14588 (possible_polymorphic_call_targets): Update.
14589 (ipa_devirt): Update.
14590 * ipa-fnsummary.c (redirect_to_unreachable): Update.
14591 * ipa-icf.c (sem_function::merge): Update.
14592 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
14593 * ipa-inline.c (compute_uninlined_call_time): Update.
14594 (compute_inlined_call_time): Update.
14595 (want_inline_small_function_p): Update.
14596 (want_inline_self_recursive_call_p): Update.
14597 (edge_badness): Update.
14598 (lookup_recursive_calls): Update.
14599 (recursive_inlining): Update.
14600 (inline_small_functions): Update.
14601 (dump_overall_stats): Update.
14602 (dump_inline_stats): Update.
14603 * ipa-profile.c (ipa_profile_generate_summary): Update.
14604 (ipa_propagate_frequency): Update.
14605 (ipa_profile): Update.
14606 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
14607 * ipa-utils.c (ipa_merge_profiles): Update.
14608 * loop-doloop.c (doloop_modify): Update.
14609 * loop-unroll.c (report_unroll): Update.
14610 (unroll_loop_runtime_iterations): Update.
14611 * lto-cgraph.c (lto_output_edge): Update.
14612 (lto_output_node): Update.
14613 (input_node): Update.
14614 (input_edge): Update.
14615 (merge_profile_summaries): Update.
14616 * lto-streamer-in.c (input_cfg): Update.
14617 * lto-streamer-out.c (output_cfg): Update.
14618 * mcf.c (create_fixup_graph): Update.
14619 (adjust_cfg_counts): Update.
14620 (sum_edge_counts): Update.
14621 * modulo-sched.c (sms_schedule): Update.
14622 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
14623 * predict.c (maybe_hot_count_p): Update.
14624 (probably_never_executed): Update.
14625 (dump_prediction): Update.
14626 (combine_predictions_for_bb): Update.
14627 (propagate_freq): Update.
14628 (handle_missing_profiles): Update.
14629 (counts_to_freqs): Update.
14630 (rebuild_frequencies): Update.
14631 (force_edge_cold): Update.
14632 * predict.h: Include profile-count.h
14633 (maybe_hot_count_p, counts_to_freqs): UPdate.
14634 * print-rtl-function.c: Do not include cfg.h
14635 * print-rtl.c: Include basic-block.h
14636 * profile-count.c: New file.
14637 * profile-count.h: New file.
14638 * profile.c (is_edge_inconsistent): Update.
14639 (correct_negative_edge_counts): Update.
14640 (is_inconsistent): Update.
14641 (set_bb_counts): Update.
14642 (read_profile_edge_counts): Update.
14643 (compute_frequency_overlap): Update.
14644 (compute_branch_probabilities): Update; Initialize and deinitialize
14646 (branch_prob): Update.
14647 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
14648 (edge_gcov_count): New.
14649 (bb_gcov_count): New.
14650 * shrink-wrap.c (try_shrink_wrapping): Update.
14651 * tracer.c (better_p): Update.
14652 * trans-mem.c (expand_transaction): Update.
14653 (ipa_tm_insert_irr_call): Update.
14654 (ipa_tm_insert_gettmclone_call): Update.
14655 * tree-call-cdce.c: Update.
14656 * tree-cfg.c (gimple_duplicate_sese_region): Update.
14657 (gimple_duplicate_sese_tail): Update.
14658 (gimple_account_profile_record): Update.
14659 (execute_fixup_cfg): Update.
14660 * tree-inline.c (copy_bb): Update.
14661 (copy_edges_for_bb): Update.
14662 (initialize_cfun): Update.
14663 (freqs_to_counts): Update.
14664 (copy_cfg_body): Update.
14665 (expand_call_inline): Update.
14666 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
14667 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
14668 (try_unroll_loop_completely): Update.
14669 (try_peel_loop): Update.
14670 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
14671 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
14672 * tree-ssa-loop-split.c (connect_loops): Update.
14673 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
14674 * tree-ssa-reassoc.c (branch_fixup): Update.
14675 * tree-ssa-tail-merge.c (replace_block_by): Update.
14676 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
14677 (compute_path_counts): Update.
14678 (update_profile): Update.
14679 (recompute_probabilities): Update.
14680 (update_joiner_offpath_counts): Update.
14681 (estimated_freqs_path): Update.
14682 (freqs_to_counts_path): Update.
14683 (clear_counts_path): Update.
14684 (ssa_fix_duplicate_block_edges): Update.
14685 (duplicate_thread_path): Update.
14686 * tree-switch-conversion.c (case_bit_test_cmp): Update.
14687 (struct switch_conv_info): Update.
14688 * tree-tailcall.c (decrease_profile): Update.
14689 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
14690 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
14691 * value-prof.c (check_counter): Update.
14692 (gimple_divmod_fixed_value): Update.
14693 (gimple_mod_pow2): Update.
14694 (gimple_mod_subtract): Update.
14695 (gimple_ic_transform): Update.
14696 (gimple_stringop_fixed_value): Update.
14697 * value-prof.h (gimple_ic): Update.
14699 2017-06-02 Carl Love <cel@us.ibm.com>
14701 * config/rs6000/rs6000-c: Add support for built-in functions
14702 vector double vec_doublee (vector signed int);
14703 vector double vec_doublee (vector unsigned int);
14704 vector double vec_doublee (vector float);
14705 vector double vec_doubleh (vector signed int);
14706 vector double vec_doubleh (vector unsigned int);
14707 vector double vec_doubleh (vector float);
14708 vector double vec_doublel (vector signed int);
14709 vector double vec_doublel (vector unsigned int);
14710 vector double vec_doublel (vector float);
14711 vector double vec_doubleo (vector signed int);
14712 vector double vec_doubleo (vector unsigned int);
14713 vector double vec_doubleo (vector float);.
14714 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
14715 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
14717 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
14718 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
14719 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
14721 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
14722 vec_doublel, vec_doubleh.
14723 * doc/extend.texi: Update the built-in documentation file for the
14724 new built-in functions.
14726 2017-06-02 David Malcolm <dmalcolm@redhat.com>
14729 * ipa-inline-analysis.c (free_growth_caches): Set
14730 edge_removal_hook_holder to NULL after removing it.
14732 2017-06-02 Sudakshina Das <sudi.das@arm.com>
14734 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
14735 comparision with zero.
14737 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
14738 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14739 for early expansion of vec_min and vec_max builtins.
14740 (builtin_function_type): Add min/max unsigned variants to those
14741 identified as having unsigned arguments.
14743 2017-06-02 Olivier Hainque <hainque@adacore.com>
14745 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
14747 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14749 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
14750 Use VALL_F16 iterator rather than VALL.
14752 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14754 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
14755 Emit CBNZ inside loop when doing a strong exchange and comparing
14756 against zero. Generate the CC flags after the loop.
14758 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
14760 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
14761 (dl_section_ref): New.
14762 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
14763 On AIX, append an expression to subtract the size of the
14764 section length to dl_section_ref.
14766 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
14768 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14769 for early expansion of vector absolute builtins.
14771 2017-06-02 Richard Biener <rguenther@suse.de>
14773 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
14774 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
14776 2017-06-02 Richard Biener <rguenther@suse.de>
14778 PR tree-optimization/80948
14779 * tree-tailcall.c (find_tail_calls): Track stmts to move in
14780 stmt order as well.
14782 2017-06-02 Richard Biener <rguenther@suse.de>
14784 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
14786 * tree-vect-stmts.c (process_use): Do not mark backedge defs
14787 for inductions as relevant.
14789 2017-06-02 Richard Biener <rguenther@suse.de>
14791 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
14792 (vectorizable_induction): ... this. Remove dead code.
14794 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
14796 * builtins. (expand_builtin_alloca): Remove second parameter and
14797 infer its value from the first parameter instead.
14798 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
14800 2017-06-02 Jakub Jelinek <jakub@redhat.com>
14802 PR rtl-optimization/80903
14803 * loop-doloop.c (add_test): Unshare sequence.
14805 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
14807 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
14809 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
14811 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
14813 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
14814 xlogue_layout::get_instance, logue_layout::xlogue_layout,
14815 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
14816 (xlogue_layout::get_stub_rtx): Make static.
14817 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
14818 (xlogue_layout::compute_stub_managed_regs): Rename to...
14819 (xlogue_layout::count_stub_managed_regs): ...this.
14820 (xlogue_layout::is_stub_managed_reg): New function.
14821 (xlogue_layout::m_stub_names): Rename to...
14822 (xlogue_layout::s_stub_names): ...this, make static.
14823 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
14824 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
14825 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
14826 xlogue_layout::s_stub_names): Instantiate statics.
14827 (stub_managed_regs): Remove.
14828 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
14829 (disable_call_ms2sysv_xlogues): Rename to...
14830 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
14831 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
14833 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
14834 change after reload_completed.
14835 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
14837 (ix86_expand_prologue): Likewise.
14838 (ix86_expand_epilogue): Likewise.
14839 (ix86_expand_split_stack_prologue): Likewise.
14840 (ix86_compute_frame_layout): Remove frame parameter ...
14841 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
14842 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
14844 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
14845 (ix86_frame): Move from here ...
14846 * config/i386/i386.h (ix86_frame): ... to here.
14847 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
14848 complete ix86_frame data structure instead. Remove some_ld_name.
14850 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
14852 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
14853 symbols that hold a DECL_VALUE_EXPR.
14855 2017-06-01 Martin Jambor <mjambor@suse.cz>
14857 PR tree-optimization/80898
14858 * tree-sra.c (process_subtree_disqualification): Removed.
14859 (disqualify_candidate): Do not acll
14860 process_subtree_disqualification.
14861 (subtree_mark_written_and_enqueue): New function.
14862 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
14863 RHS has been disqualified and re-queue LHS if necessary. Apart
14864 from that, ignore disqualified RHS.
14866 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14868 * config/s390/s390.c (s390_emit_epilogue): Disable early return
14869 address fetch for z10 or later.
14871 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14873 * config/arc/arc.md (tst_movb): Add guard when splitting.
14875 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14877 * config/arc/arc.c (arc_can_eliminate): Test against
14878 arc_frame_pointer_needed.
14880 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14882 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
14883 to prevent store reordering.
14884 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
14885 (type): Add block type.
14886 (stack_tie): Define special instruction to be used in
14889 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14891 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
14892 constraint. It is not valid for the pattern.
14893 (noncommutative_binary_comparison): Likewise.
14895 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14897 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
14900 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14902 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
14903 be used by the reg-alloc.
14905 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14907 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
14908 reg-alloc when having mul64 or mul32x16 instructions.
14909 (mulsidi3): Likewise.
14910 (umulsidi3): Likewise.
14911 (mulsi32x16): New pattern.
14912 (mulsi64): Likewise.
14913 (mulsidi64): Likewise.
14914 (umulsidi64): Likewise.
14915 (MUL32x16_REG): Define.
14916 (mul64_600): Use MUL32x16_REG.
14917 (mac64_600): Likewise.
14918 (umul64_600): Likewise.
14919 (umac64_600): Likewise.
14921 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14923 * config/arc/arc.md (mulsi3_700): Make it commutative.
14925 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
14927 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
14929 (*sign_extendsidi2_insn): Likewise for movstosw.
14931 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
14933 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
14934 the type of the input discriminant value. Convert the
14935 discriminant value of signedness vary.
14937 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
14939 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
14940 Add to -Wall section.
14942 2017-06-01 Richard Biener <rguenther@suse.de>
14944 PR middle-end/66313
14945 * fold-const.c (fold_plusminus_mult_expr): If the factored
14946 factor may be zero use a wrapping type for the inner operation.
14947 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
14948 and handle moved defs.
14949 (process_assignment): Properly guard the unary op case. Return a
14950 tri-state indicating that moving the stmt before the call may allow
14951 to continue. Pass through to_move.
14952 (find_tail_calls): Handle moving unrelated defs before
14955 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
14958 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
14959 splitter result in the canonical way.
14961 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
14963 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
14964 also for 32bit target. Update insn attributes.
14965 (zero-extendsidi2 splitter): Allow all registers for operand 1.
14967 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
14969 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
14970 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
14971 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
14972 (_mm_maskz_min_ss): New intrinsics.
14974 2017-05-31 Martin Liska <mliska@suse.cz>
14976 * tree-vect-loop.c (vect_create_epilog_for_reduction):
14977 Change comment style to one we normally use.
14978 (vectorizable_reduction): Likewise.
14979 (vectorizable_induction): Likewise.
14980 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
14981 (vectorizable_call): Likewise.
14982 (vectorizable_simd_clone_call): Likewise.
14983 (vectorizable_conversion): Likewise.
14984 (vectorizable_assignment): Likewise.
14985 (vectorizable_shift): Likewise.
14986 (vectorizable_operation): Likewise.
14987 (vectorizable_store): Likewise.
14988 (vectorizable_load): Likewise.
14989 * tree-vectorizer.h: Likewise.
14991 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
14993 * passes.c (emergency_dump_function): New.
14994 * tree-pass.h (emergency_dump_function): Declare.
14995 * plugin.c (plugins_internal_error_function): Remove.
14996 * plugin.h (plugins_internal_error_function): Remove declaration.
14997 * toplev.c (internal_error_function): New static function. Use it...
14998 (general_init): ...here.
15000 2017-05-31 Graham Markall <graham.markall@embecosm.com>
15002 * config/arc/arc.c (arc_print_operand): Handle constant operands.
15003 (arc_rtx_costs): Add costs for new patterns.
15004 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
15005 * config/arc/predicates.md: Add _1_2_3_operand predicate.
15007 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
15009 * tree-ssa-strlen.c (get_next_strinfo): New function.
15010 (get_stridx_plus_constant): Use it.
15011 (zero_length_string): Likewise.
15012 (adjust_related_strinfos): Likewise.
15013 (adjust_last_stmt): Likewise.
15015 2017-05-31 Richard Biener <rguenther@suse.de>
15018 * config/i386/i386.c (ix86_expand_builtin): Remove assert
15019 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
15021 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
15023 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
15024 loop_vinfo argument and use of dependence distance vectors.
15025 Check instead whether the two references differ only in their
15026 initial value and assume that they have the same alignment if the
15027 difference is a multiple of the vector alignment.
15028 (vect_analyze_data_refs_alignment): Update call accordingly.
15030 2017-05-31 Martin Liska <mliska@suse.cz>
15033 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
15035 2017-05-31 Bin Cheng <bin.cheng@arm.com>
15037 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
15038 (create_intersect_range_checks): Move from ...
15039 * tree-data-ref.c (create_intersect_range_checks_index)
15040 (create_intersect_range_checks): ... to here.
15041 (create_runtime_alias_checks): New function factored from ...
15042 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
15043 here. Call above function.
15044 * tree-data-ref.h (create_runtime_alias_checks): New function.
15046 2017-05-31 Bin Cheng <bin.cheng@arm.com>
15048 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
15049 segment length for dr_b and compute it in wide_int.
15051 2017-05-31 Richard Biener <rguenther@suse.de>
15053 PR tree-optimization/80906
15054 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
15055 and pass through iv_map.
15056 (copy_bb_and_scalar_dependences): Adjust.
15057 (translate_pending_phi_nodes): Likewise.
15058 (copy_loop_close_phi_args): Handle code-generating IVs instead
15061 2017-05-30 David Malcolm <dmalcolm@redhat.com>
15063 * diagnostic-color.c (color_dict): Add "type-diff".
15064 (parse_gcc_colors): Update comment.
15065 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
15066 -fdiagnostics-show-template-tree and -fno-elide-type.
15067 (GCC_COLORS): Add type-diff to example.
15069 (-fdiagnostics-show-template-tree): New.
15070 (-fno-elide-type): New.
15071 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
15072 the pp_format_decoder callback. Call any m_format_postprocessor's
15074 (pretty_printer::pretty_printer): Initialize
15075 m_format_postprocessor.
15076 (pretty_printer::~pretty_printer): Delete any
15077 m_format_postprocessor.
15078 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
15079 (class format_postprocessor): New class.
15080 (struct pretty_printer::format_decoder): Document the new parameters.
15081 (struct pretty_printer::m_format_postprocessor): New field.
15082 * tree-diagnostic.c (default_tree_printer): Update for new
15083 bool and const char ** params.
15084 * tree-diagnostic.h (default_tree_printer): Likewise.
15086 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
15088 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
15089 (lwa_operand): Delete rs6000_gen_cell_microcode test.
15090 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
15091 rs6000_gen_cell_microcode code.
15092 (rs6000_final_prescan_insn): Delete.
15093 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
15094 "warn-cell-microcode" entries.
15095 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
15096 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
15097 throughout. Change cc_reg_not_micro_cr0_operand to
15098 cc_reg_not_cr0_operand throughout.
15099 (*extendhi<mode>2_noload): Delete.
15100 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
15101 (mwarn-cell-microcode): Delete.
15102 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
15103 -mgen-cell-microcode and -mwarn-cell-microcode.
15105 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
15108 * config/i386/constraints.md (Yd): New constraint.
15110 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
15111 and (?Yd, r) alternatives. Update insn attributes.
15112 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
15113 and (?*Yd, r) alternatives. Update insn attributes.
15114 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
15116 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
15118 * gimplify.c (gimplify_modify_expr): Don't create a
15119 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
15122 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
15124 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
15126 2017-05-30 Richard Biener <rguenther@suse.de>
15128 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
15129 and reduc_def fields.
15130 (STMT_VINFO_REDUC_TYPE): New define.
15131 (STMT_VINFO_REDUC_DEF): Likewise.
15132 (vect_force_simple_reduction): Adjust prototype.
15133 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
15134 (vect_is_simple_reduction): Remove check_reduction argument.
15135 (vect_force_simple_reduction): Adjust and set
15136 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
15137 (vectorizable_reduction): Do not re-do reduction analysis
15138 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
15139 * tree-parloops.c (gather_scalar_reductions): Adjust.
15141 2017-05-30 Richard Biener <rguenther@suse.de>
15143 PR middle-end/80901
15144 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
15147 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
15149 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
15150 Introduce unknown_misalignment parameter and remove vf.
15151 (vect_peeling_hash_get_lowest_cost):
15152 Pass unknown_misalignment parameter.
15153 (vect_enhance_data_refs_alignment):
15154 Fix unsupportable data ref treatment.
15156 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
15158 * tree-vect-data-refs.c (vect_get_data_access_cost):
15159 Workaround for SLP handling.
15160 (vect_enhance_data_refs_alignment):
15161 Compute costs for doing no peeling at all, compare to the best
15162 peeling costs so far and avoid peeling if cheaper.
15164 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
15166 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
15167 Return peeling info and set costs to zero for unlimited cost
15169 (vect_enhance_data_refs_alignment): Also inspect all datarefs
15170 with unknown misalignment. Compute and costs for unknown
15171 misalignment, compare them to the costs for known misalignment
15172 and choose the cheapest for peeling.
15174 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
15176 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
15177 (vect_get_peeling_costs_all_drs): Create function.
15178 (vect_peeling_hash_get_lowest_cost):
15179 Use vect_get_peeling_costs_all_drs.
15180 (vect_peeling_supportable): Create function.
15181 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
15183 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
15185 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
15186 DR_HAS_NEGATIVE_STEP.
15187 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
15188 (vect_enhance_data_refs_alignment): Use.
15189 (vect_duplicate_ssa_name_ptr_info): Use.
15190 * tree-vectorizer.h (dr_misalignment): Use.
15191 (known_alignment_for_access_p): Use.
15193 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
15196 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
15198 (has_section_name): New function.
15200 2017-05-30 Martin Liska <mliska@suse.cz>
15203 * auto-profile.c (get_function_decl_from_block): Fix
15206 2017-05-30 Richard Biener <rguenther@suse.de>
15208 PR middle-end/80876
15209 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
15211 2017-05-30 Martin Liska <mliska@suse.cz>
15213 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
15214 * dumpfile.h (struct dump_file_info): Remove ctors.
15216 2017-05-30 Martin Liska <mliska@suse.cz>
15218 * predict.def: Fix GNU coding style.
15220 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
15222 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
15223 Mark 'to' argument with ATTRIBUTE_UNUSED.
15225 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
15227 * config/xtensa/xtensa.c (xtensa_emit_call): Use
15228 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
15229 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
15232 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
15234 * doc/install.texi (Options specification): Restore entry of
15235 --enable-sjlj-exceptions.
15237 2017-05-27 Michael Eager <eager@eagercon.com>
15240 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
15242 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
15244 * config/microblaze/microblaze.h
15245 (FIXED_REGISTERS): Update in macro.
15246 (CALL_USED_REGISTERS): Update in macro.
15248 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
15250 * doc/install.texi: Add links to macOS binary distributions.
15252 2017-05-27 Jakub Jelinek <jakub@redhat.com>
15256 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
15258 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
15260 2017-05-26 Martin Liska <mliska@suse.cz>
15262 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
15264 2017-05-26 Martin Liska <mliska@suse.cz>
15266 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
15267 always leading ';; '.
15268 (dump_bb_info): Likewise.
15269 (brief_dump_cfg): Likewise.
15270 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
15271 * dumpfile.c: Remove usage of TDF_VERBOSE.
15272 * dumpfile.h (enum dump_kind): Likewise.
15273 (dump_gimple_bb_header): Do not use TDF_COMMENT.
15274 * print-tree.c (debug_verbose): Remove.
15275 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
15276 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
15277 * tree-diagnostic.c (default_tree_printer): Replace
15278 TDF_DIAGNOSTIC with TDF_SLIM.
15280 2017-05-26 Bin Cheng <bin.cheng@arm.com>
15282 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
15283 in parameter loop, rather than loop_vinfo.
15284 (create_intersect_range_checks): Ditto.
15285 (vect_create_cond_for_alias_checks): Update call to above functions.
15287 2017-05-26 Bin Cheng <bin.cheng@arm.com>
15289 PR tree-optimization/80815
15290 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
15291 for merging runtime alias checks. Handle negative DR_STEPs.
15293 2017-05-26 Bin Cheng <bin.cheng@arm.com>
15295 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
15297 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
15298 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
15299 out code pruning runtime alias checks.
15300 * tree-data-ref.c (prune_runtime_alias_test_list): New function
15301 factored out from above.
15302 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
15304 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
15306 (prune_runtime_alias_test_list): New decalaration.
15308 2017-05-26 Bin Cheng <bin.cheng@arm.com>
15310 * tree-vect-data-refs.c (compare_tree): Rename and move ...
15311 * tree-data-ref.c (data_ref_compare_tree): ... to here.
15312 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
15313 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
15314 (operator==, comp_dr_with_seg_len_pair): Ditto.
15315 (vect_prune_runtime_alias_test_list): Ditto.
15317 2017-05-26 Martin Liska <mliska@suse.cz>
15320 * params.def: Bound partial-inlining-entry-probability param.
15322 2017-05-26 Marek Polacek <polacek@redhat.com>
15325 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
15328 2017-05-26 Richard Biener <rguenther@suse.de>
15330 PR tree-optimization/80842
15331 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
15334 2017-05-26 Richard Biener <rguenther@suse.de>
15336 PR tree-optimization/80844
15337 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
15339 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
15341 * doc/md.texi (Machine Constraints): Update x86 family
15342 machine constraints section to match 'config/i386/constraints.md'.
15344 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
15346 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
15348 2017-05-25 Nathan Sidwell <nathan@acm.org>
15350 * doc/invoke.texi (--enable-languages): Update documentation.
15352 2017-05-25 Martin Liska <mliska@suse.cz>
15354 * dumpfile.c: Add TDF_FOLDING.
15355 * dumpfile.h (enum dump_kind): Likewise.
15356 * genmatch.c (dt_simplify::gen_1): Use it.
15358 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
15360 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
15362 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
15364 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
15365 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
15367 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
15369 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
15370 * match.pd (X == C): Rewrite it here.
15371 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
15372 with_certain_nonzero_bits2): New predicates.
15373 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
15375 2017-05-24 Nathan Sidwell <nathan@acm.org>
15377 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
15380 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
15383 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
15385 * config/powerpcspe: New port. Files are copied from the rs6000
15386 port, with "rs6000" in filenames replaced by "powerpcspe".
15388 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
15390 PR rtl-optimization/80754
15391 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
15393 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
15395 * config/sparc/sparc.md (length): Return the correct value for -mflat
15396 sibcalls to match output_sibcall.
15398 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
15402 * config/rs6000/rs6000.c (struct machine_function): Add new field
15404 (rs6000_get_separate_components): Init that field, use it.
15405 (rs6000_components_for_bb): Use the field.
15407 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
15409 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
15411 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
15413 PR middle-end/80823
15414 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
15416 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15419 * config/s390/s390.c (s390_check_qrst_address): Check incoming
15420 address against address_operand predicate.
15421 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
15423 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
15425 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
15426 parameters passed indirectly.
15428 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
15430 * config/i386/i386.md (*movdi_internal): Remove SSE4
15431 alternative 18 (?r, *v). Update insn attributes.
15432 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
15433 Update insn attributes.
15434 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
15435 Update insn attributes.
15436 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
15437 alternative 1 (r, v). Remove isa attribute.
15438 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
15439 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
15440 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
15442 2017-05-23 Tom de Vries <tom@codesourcery.com>
15444 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
15447 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
15449 * cgraphunit.c (symbol_table::process_new_functions): Update.
15450 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
15451 (inline_generate_summary): Rename to ...
15452 (ipa_fn_summary_generate): ... this one.
15453 (inline_read_summary): Rename to ...
15454 (ipa_fn_summary_read): ... this one.
15455 (inline_write_summary): Rename to ...
15456 (ipa_fn_summary_write): ... this one.
15457 (inline_free_summary): Rename to ...
15458 (ipa_free_fn_summary): ... this one.
15459 (pass_data_local_fn_summary, pass_local_fn_summary,
15460 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
15461 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
15462 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
15463 make_pass_ipa_fn_summary): New.
15464 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
15465 inline_write_summary, inline_free_summary): Remove.
15466 (ipa_free_fn_summary) : New.
15467 * ipa-inline.c (ipa_inline): Update.
15468 (pass_ipa_inline): Do not generate summaries.
15469 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
15471 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
15472 and add pass_ipa_fn_summary.
15473 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
15475 (make_pass_inline_parameters): Remove.
15477 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
15479 * omp-low.c (struct omp_context): Remove "default_kind" member.
15482 * omp-offload.c (execute_oacc_device_lower): Remove the
15483 parallelism dimensions function attributes for unparallelized
15484 OpenACC kernels constructs.
15486 2017-05-23 Martin Liska <mliska@suse.cz>
15488 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
15490 (cgraph_edge::make_speculative): Likewise.
15491 (cgraph_edge::resolve_speculation): Likewise.
15492 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
15493 (cgraph_node::dump): Likewise.
15494 * cgraph.h: Likewise.
15495 * cgraphunit.c (analyze_functions): Likewise.
15496 (symbol_table::compile): Likewise.
15497 * ipa-cp.c (print_all_lattices): Likewise.
15498 (determine_versionability): Likewise.
15499 (initialize_node_lattices): Likewise.
15500 (ipcp_verify_propagated_values): Likewise.
15501 (estimate_local_effects): Likewise.
15502 (update_profiling_info): Likewise.
15503 (create_specialized_node): Likewise.
15504 (perhaps_add_new_callers): Likewise.
15505 (decide_about_value): Likewise.
15506 (decide_whether_version_node): Likewise.
15507 (identify_dead_nodes): Likewise.
15508 (ipcp_store_bits_results): Likewise.
15509 * ipa-devirt.c (dump_targets): Likewise.
15510 (ipa_devirt): Likewise.
15511 * ipa-icf.c (sem_item::dump): Likewise.
15512 (sem_function::equals): Likewise.
15513 (sem_variable::equals): Likewise.
15514 (sem_item_optimizer::read_section): Likewise.
15515 (sem_item_optimizer::execute): Likewise.
15516 (congruence_class::dump): Likewise.
15517 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
15518 (dump_inline_summary): Likewise.
15519 (estimate_node_size_and_time): Likewise.
15520 (inline_analyze_function): Likewise.
15521 * ipa-inline-transform.c (inline_call): Likewise.
15522 * ipa-inline.c (report_inline_failed_reason): Likewise.
15523 (want_early_inline_function_p): Likewise.
15524 (edge_badness): Likewise.
15525 (update_edge_key): Likewise.
15526 (inline_small_functions): Likewise.
15527 * ipa-profile.c (ipa_profile): Likewise.
15528 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
15529 (ipa_make_edge_direct_to_target): Likewise.
15530 (remove_described_reference): Likewise.
15531 (ipa_impossible_devirt_target): Likewise.
15532 (propagate_controlled_uses): Likewise.
15533 (ipa_print_node_params): Likewise.
15534 (ipcp_transform_function): Likewise.
15535 * ipa-pure-const.c (pure_const_read_summary): Likewise.
15536 (propagate_pure_const): Likewise.
15537 * ipa-reference.c (generate_summary): Likewise.
15538 (read_write_all_from_decl): Likewise.
15539 (propagate): Likewise.
15540 (ipa_reference_read_optimization_summary): Likewise.
15541 * ipa-utils.c (ipa_merge_profiles): Likewise.
15542 * ipa.c (walk_polymorphic_call_targets): Likewise.
15543 (symbol_table::remove_unreachable_nodes): Likewise.
15544 (ipa_single_use): Likewise.
15545 * passes.c (execute_todo): Likewise.
15546 * predict.c (drop_profile): Likewise.
15547 * symtab.c (symtab_node::get_dump_name): New function.
15548 (symtab_node::dump_name): Likewise.
15549 (symtab_node::dump_asm_name): Likewise.
15550 (symtab_node::dump_references): Likewise.
15551 (symtab_node::dump_referring): Likewise.
15552 (symtab_node::dump_base): Likewise.
15553 (symtab_node::debug_symtab): Likewise.
15554 * tree-sra.c (convert_callers_for_node): Likewise.
15555 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
15556 * value-prof.c (init_node_map): Likewise.
15558 2017-05-23 Martin Liska <mliska@suse.cz>
15560 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
15561 and symtab_node::debug_symtab to symbol_table::debug.
15562 * cgraphunit.c (analyze_functions): Use the renamed function.
15563 (symbol_table::compile): Likewise.
15564 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
15565 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
15566 * passes.c (execute_todo): Likewise.
15567 * symtab.c (symbol_table::dump): New function.
15568 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
15570 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
15572 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
15573 that nonconst implies exec.
15575 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
15577 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
15578 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
15579 (inline_edge_summary_vec): Turn into ...
15580 (ipa_call_summaries): ... this one.
15581 (redirect_to_unreachable, edge_set_predicate,
15582 evaluate_properties_for_edge, inline_summary_alloc,
15583 reset_ipa_call_summary, reset_inline_summary,
15584 inline_summary_t::duplicate): Update.
15585 (inline_edge_duplication_hook): Turn to ...
15586 (ipa_call_summary_t::duplicate): ... this one.
15587 (inline_edge_removal_hook): Turn to ...
15588 (ipa_call_summary_t::remove): ... this one.
15589 (dump_inline_edge_summary): Turn to ...
15590 (dump_ipa_call_summary): ... this one.
15591 (estimate_function_body_sizes): Update.
15592 (inline_update_callee_summaries): Update.
15593 (remap_edge_change_prob): Update.
15594 (remap_edge_summaries): Update.
15595 (inline_merge_summary): Update.
15596 (do_estimate_edge_time): Update.
15597 (inline_generate_summary): Update.
15598 (inline_read_section): Update.
15599 (inline_read_summary): Update.
15600 (inline_free_summary): Update.
15601 * ipa-inline.c (can_inline_edge_p): Update.
15602 (compute_inlined_call_time): Update.
15603 (want_inline_small_function_p): Update.
15604 (edge_badness): Update.
15605 (early_inliner): Update.
15606 * ipa-inline.h (inline_edge_summary): Turn to ...
15607 (ipa_call_summary): ... this one.
15608 (ipa_call_summary_t): New class.
15609 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
15610 (ipa_call_summaries): New.
15611 (inline_edge_summary): Remove.
15612 (estimate_edge_growth): Update.
15613 * ipa-profile.c (ipa_propagate_frequency_1): Update.
15614 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
15615 * ipa-split.c (execute_split_functions): Update.
15616 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
15618 2017-05-23 Tom de Vries <tom@codesourcery.com>
15620 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
15621 attributes): Document rdrand effective target.
15623 2017-05-23 Tom de Vries <tom@codesourcery.com>
15625 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
15626 attributes): Sort alphabetically.
15628 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
15630 * config/avr/genmultilib.awk: Use gsub instead of gensub.
15632 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
15635 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
15636 V2DF/V2DI splat into two separate patterns, one that handles
15637 registers, and the other that only handles memory. Drop support
15638 for splatting from a GPR on ISA 2.07 and then splitting the
15639 splat into direct move and splat.
15640 (vsx_splat_<mode>_reg): Likewise.
15641 (vsx_splat_<mode>_mem): Likewise.
15643 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
15645 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
15647 2017-05-22 Jakub Jelinek <jakub@redhat.com>
15649 PR middle-end/80809
15650 * omp-low.c (finish_taskreg_remap): New function.
15651 (finish_taskreg_scan): If unit size of ctx->record_type
15652 is non-constant, unshare the size expression and replace
15653 decls in it with possible outer var refs.
15655 PR middle-end/80809
15656 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
15657 GOVD_SHARED rather than GOVD_PRIVATE with it.
15658 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
15659 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
15661 PR middle-end/80853
15662 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
15663 as last argument to build_outer_var_ref for pointer bases of array
15664 section reductions.
15666 2017-05-19 Martin Sebor <msebor@redhat.com>
15668 * print-tree.c (print_node): Print DECL_READ_P flag.
15670 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15672 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
15673 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
15674 * cgraph.c: Likewise.
15675 * cgraphunit.c: Likewise.
15676 * gengtype.c: Likewise.
15677 * ipa-cp.c: Likewise.
15678 * ipa-devirt.c: Likewise.
15679 * ipa-icf.c: Likewise.
15680 * ipa-predicate.c: Likewise.
15681 * ipa-profile.c: Likewise.
15682 * ipa-prop.c: Likewise.
15683 * ipa-split.c: Likewise.
15685 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
15686 edge_predicate_pool, dump_inline_hints,
15687 inline_summary::account_size_time, redirect_to_unreachable,
15688 edge_set_predicate, set_hint_predicate,
15689 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
15690 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
15691 inline_summary_t::remove, remap_hint_predicate_after_duplication,
15692 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
15693 ipa_call_summary_t::remove, initialize_growth_caches,
15694 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
15695 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
15696 mark_modified, unmodified_parm_1, unmodified_parm,
15697 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
15698 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
15699 compute_bb_predicates, will_be_nonconstant_expr_predicate,
15700 will_be_nonconstant_predicate, record_modified_bb_info,
15701 get_minimal_bb, record_modified, param_change_prob,
15702 phi_result_unknown_predicate, predicate_for_phi_result,
15703 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
15704 estimate_function_body_sizes, compute_inline_parameters,
15705 compute_inline_parameters_for_curren, pass_data_inline_parameters,
15706 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
15707 inline_update_callee_summaries, remap_edge_change_prob,
15708 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
15709 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
15710 inline_analyze_function, inline_summary_t::insert,
15711 inline_generate_summary, read_ipa_call_summary, inline_read_section,
15712 inline_read_summary, write_ipa_call_summary, inline_write_summary,
15713 inline_free_summary): Move to ipa-fnsummary.h
15714 (predicate_t): Remove.
15715 * ipa-fnsummary.c: New file.
15716 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
15717 (enum inline_hints_vals, inline_hints, agg_position_info,
15718 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
15719 inline_summaries, ipa_call_summary, ipa_call_summary_t,
15720 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
15721 dump_inline_summary, dump_inline_hints, inline_generate_summary,
15722 inline_read_summary, inline_write_summary, inline_free_summary,
15723 inline_analyze_function, initialize_inline_failed,
15724 inline_merge_summary, inline_update_overall_summary,
15725 compute_inline_parameters): Move to ipa-fnsummary.h
15726 * ipa-fnsummary.h: New file.
15727 * ipa-inline-transform.h: Include ipa-inline.h.
15728 * ipa-inline.c: LIkewise.
15730 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15732 * ipa-inline.c (edge_badness): Use inlined_time instead of
15733 inline_summaries->get.
15735 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15737 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
15739 2017-05-22 Nathan Sidwell <nathan@acm.org>
15741 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
15742 (fdump-lang): Document 'raw' option.
15743 * dumpfile.h (TDI_tu): Delete.
15744 * dumpfile.c (dump_files): Remove translation-unit.
15745 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
15747 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
15749 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
15750 command option from $(AWK) call.
15751 * config/avr/genmultilib.awk: Simplify and rewrite so that it
15752 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
15753 [FORMAT]: Remove handling of variable.
15754 * config/avr/t-multilib: Regenerate.
15756 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15758 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
15760 (dump_inline_summary): Do not print self_time.
15761 (estimate_function_body_sizes): Do not set self_time.
15762 (compute_inline_parameters): Likewise.
15763 (inline_read_section, inline_write_summary): Do not stream self_time.
15764 * ipa-inline.h (inline_summary): Drop self_time.
15766 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15768 * ipa-inline-analysis.c (account_size_time): Rename to ...
15769 (inline_summary::account_size_time): ... this one.
15770 (reset_ipa_call_summary): Turn to ...
15771 (ipa_call_summary::reset): ... this one.
15772 (reset_inline_summary): Turn to ...
15773 (inline_summary::reset): ... this one.
15774 (inline_summary_t::remove): Update.
15775 (inline_summary_t::duplicate): Update.
15776 (ipa_call_summary_t::remove): Update.
15777 (dump_inline_summary): Update.
15778 (estimate_function_body_sizes): Update.
15779 (compute_inline_parameters): Update.
15780 (estimate_node_size_and_time): Update.
15781 (inline_merge_summary): Update.
15782 (inline_update_overall_summary): Update.
15783 (inline_read_section): Update.
15784 (inline_write_summary): Update.
15785 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
15786 add account_size_time and reset member functions.
15787 (ipa_call_summary): Add reset function.
15788 * ipa-predicate.h (predicate::operator &): Constify.
15790 2017-05-22 Richard Biener <rguenther@suse.de>
15792 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
15794 2017-05-19 Jason Merrill <jason@redhat.com>
15796 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
15798 2017-05-19 Marek Polacek <polacek@redhat.com>
15801 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
15802 TYPE_OVERFLOW_WRAPS checks.
15804 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
15806 * tree-core.h (enum omp_clause_default_kind): Add
15807 "OMP_CLAUSE_DEFAULT_PRESENT".
15808 * tree-pretty-print.c (dump_omp_clause): Handle it.
15809 * gimplify.c (enum gimplify_omp_var_data): Add
15810 "GOVD_MAP_FORCE_PRESENT".
15811 (gimplify_adjust_omp_clauses_1): Map it to
15812 "GOMP_MAP_FORCE_PRESENT".
15813 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
15815 * gimplify.c (oacc_default_clause): Clarify.
15817 2017-05-19 Nathan Sidwell <nathan@acm.org>
15819 LANG_HOOK_REGISTER_DUMPS
15820 * toplev.c (general_init): Call register dump lang hook.
15821 * doc/invoke.texi: Document -fdump-lang option family.
15822 * dumpfile.c (dump_files): Remove class dump here.
15823 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
15824 * dumpfile.h (tree_dump_index): Remove TDI_class.
15825 * langhooks-def.h (lhd_register_dumps): Declare.
15826 (LANG_HOOKS_REGISTER_DUMPS): Define.
15827 (LANG_HOOKS_INITIALIZER): Add it.
15828 * langhooks.c (lhd_register_dumps): Define.
15829 * langhooks.h (struct lang_hooks): Add register_dumps.
15831 2017-05-19 Nathan Sidwell <nathan@acm.org>
15833 * context.h (context::set_passes): New.
15834 * context.c (context::context): Do not create pass manager.
15835 * toplev.c (general_init): Create pass manager here.
15837 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
15839 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
15840 use this splitter if two add or or instructions would also work for
15841 the constant we want to generate.
15843 2017-05-19 Richard Biener <rguenther@suse.de>
15846 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
15847 predicate evaluation.
15849 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
15851 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
15853 * ipa-inline.c (want_inline_small_function_p): Do not cast to
15856 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
15858 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
15859 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
15860 (inline_edge_summary_vec): Turn into ...
15861 (ipa_call_summaries): ... this one.
15862 (redirect_to_unreachable, edge_set_predicate,
15863 evaluate_properties_for_edge, inline_summary_alloc,
15864 reset_ipa_call_summary, reset_inline_summary,
15865 inline_summary_t::duplicate): Update.
15866 (inline_edge_duplication_hook): Turn to ...
15867 (ipa_call_summary_t::duplicate): ... this one.
15868 (inline_edge_removal_hook): Turn to ...
15869 (ipa_call_summary_t::remove): ... this one.
15870 (dump_inline_edge_summary): Turn to ...
15871 (dump_ipa_call_summary): ... this one.
15872 (estimate_function_body_sizes): Update.
15873 (inline_update_callee_summaries): Update.
15874 (remap_edge_change_prob): Update.
15875 (remap_edge_summaries): Update.
15876 (inline_merge_summary): Update.
15877 (do_estimate_edge_time): Update.
15878 (inline_generate_summary): Update.
15879 (inline_read_section): Update.
15880 (inline_read_summary): Update.
15881 (inline_free_summary): Update.
15882 * ipa-inline.c (can_inline_edge_p): Update.
15883 (compute_inlined_call_time): Update.
15884 (want_inline_small_function_p): Update.
15885 (edge_badness): Update.
15886 (early_inliner): Update.
15887 * ipa-inline.h (inline_edge_summary): Turn to ...
15888 (ipa_call_summary): ... this one.
15889 (ipa_call_summary_t): New class.
15890 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
15891 (ipa_call_summaries): New.
15892 (inline_edge_summary): Remove.
15893 (estimate_edge_growth): Update.
15894 * ipa-profile.c (ipa_propagate_frequency_1): Update.
15895 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
15896 * ipa-split.c (execute_split_functions): Update.
15897 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
15899 2017-05-19 Richard Biener <rguenther@suse.de>
15901 PR middle-end/80764
15902 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
15904 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
15906 * config/rs6000/rs6000.c (struct machine_function): Add field
15907 fpr_is_wrapped_separately.
15908 (rs6000_get_separate_components): Use 64 components. Handle the
15909 new FPR components.
15910 (rs6000_components_for_bb): Handle the FPR components.
15911 (rs6000_emit_prologue_components): Handle the FPR components.
15912 (rs6000_emit_epilogue_components): Handle the FPR components.
15913 (rs6000_set_handled_components): Handle the FPR components.
15914 (rs6000_emit_prologue): Don't output prologue code for those FPRs
15915 that are already separately shrink-wrapped.
15916 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
15917 that are already separately shrink-wrapped.
15919 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
15922 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
15925 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
15926 (define_peephole2 for Altivec d-form load): Add peepholes to catch
15927 cases where the register allocator uses a move and an offsettable
15928 memory operation to/from a FPR register on ISA 2.06/2.07.
15929 (define_peephole2 for Altivec d-form store): Likewise.
15931 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
15934 * config/i386/mmx.md (*mov<mode>_internal): Enable
15935 alternatives 11, 12, 13 and 14 also for 32bit targets.
15936 Remove alternatives 15, 16, 17 and 18.
15937 * config/i386/sse.md (vec_concatv2di): Change
15938 alternative (!x, *y) to (x, ?!*Yn).
15940 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
15942 * dumpfile.h (enum dump_kind): Remove stray comma.
15944 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
15946 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
15947 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
15948 predicate::num_conditions
15949 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
15950 (CHANGED): turn into predicate::changed.
15951 (agg_position_info): Move to ipa-predicate.h
15952 (add_condition, predicate::add_clause, predicate::operator &=,
15953 predicate::or_with, predicate::evaluate, predicate::probability,
15954 dump_condition, dump_clause, predicate::dump,
15955 predicate::remap_after_duplication, predicate::remap_after_inlining,
15956 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
15957 (evaluate_conditions_for_known_args): Update.
15958 (set_cond_stmt_execution_predicate): Update.
15959 * ipa-inline.h: Include ipa-predicate.h
15960 (condition, inline_param_summary, conditions, agg_position_info,
15961 predicate): Move to ipa-predicate.h
15962 * ipa-predicate.c: New file.
15963 * ipa-predicate.h: New file.
15965 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
15967 * final.c (leaf_function_p): Check we are not in a sequence.
15969 2017-05-18 Martin Liska <mliska@suse.cz>
15971 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
15972 * dumpfile.c (dump_register): Use new enum dump_kind.
15973 (get_dump_file_name): Likewise.
15974 (dump_enable_all): Likewise.
15975 (dump_switch_p_1): Likewise.
15976 (enable_rtl_dump_file): Remove usage of TDF_RTL.
15977 * dumpfile.h (enum dump_kind): New enum type.
15978 (struct dump_file_info): Create constructor and
15979 format fields and comments.
15980 * passes.c (pass_manager::register_one_dump_file):
15982 * statistics.c (statistics_early_init): Likewise.
15983 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
15984 TDF_TREE with TDF_SLIM.
15985 (gather_memory_references_ref): Likewise.
15987 2017-05-18 Martin Liska <mliska@suse.cz>
15989 * vec.h (struct vnull): Use it.
15991 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
15993 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
15994 (true_predicate, false_predicate, true_predicate_p,
15995 false_predicate_p): Remove.
15996 (single_cond_predicate, not_inlined_predicate): Turn to member function
15998 (add_condition): Update.
15999 (add_clause): Turn to...
16000 (predicate::add_clause): ... this one; update; allow passing NULL
16002 (and_predicates): Turn to ...
16003 (predicate::operator &=): ... this one.
16004 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
16005 (or_predicates): Turn to ...
16006 (predicate::or_with): ... this one.
16007 (evaluate_predicate): Turn to ...
16008 (predicate::evaluate): ... this one.
16009 (predicate_probability): Turn to ...
16010 (predicate::probability): ... this one.
16011 (dump_condition): Update.
16012 (dump_predicate): Turn to ...
16013 (predicate::dump): ... this one.
16014 (account_size_time): Update.
16015 (edge_set_predicate): Update.
16016 (set_hint_predicate): UPdate.
16017 (evaluate_conditions_for_known_args): Update.
16018 (evaluate_properties_for_edge): Update.
16019 (remap_predicate_after_duplication): Turn to...
16020 (predicate::remap_after_duplication): ... this one.
16021 (remap_hint_predicate_after_duplication): Update.
16022 (inline_summary_t::duplicate): UPdate.
16023 (dump_inline_edge_summary): Update.
16024 (dump_inline_summary): Update.
16025 (set_cond_stmt_execution_predicate): Update.
16026 (set_switch_stmt_execution_predicate): Update.
16027 (compute_bb_predicates): Update.
16028 (will_be_nonconstant_expr_predicate): Update.
16029 (will_be_nonconstant_predicate): Update.
16030 (phi_result_unknown_predicate): Update.
16031 (predicate_for_phi_result): Update.
16032 (array_index_predicate): Update.
16033 (estimate_function_body_sizes): Update.
16034 (estimate_node_size_and_time): Update.
16035 (estimate_ipcp_clone_size_and_time): Update.
16036 (remap_predicate): Rename to ...
16037 (predicate::remap_after_inlining): ... this one.
16038 (remap_hint_predicate): Update.
16039 (inline_merge_summary): Update.
16040 (inline_update_overall_summary): Update.
16041 (estimate_size_after_inlining): Update.
16042 (read_predicate): Rename to ...
16043 (predicate::stream_in): ... this one.
16044 (read_inline_edge_summary): Update.
16045 (write_predicate): Rename to ...
16046 (predicate::stream_out): ... this one.
16047 (write_inline_edge_summary): Update.
16048 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
16049 (clause_t): Turn to uint32_t
16050 (predicate): Turn to class; implement constructor and operators
16052 (size_time_entry): Update.
16053 (inline_summary): Update.
16054 (inline_edge_summary): Update.
16056 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
16058 * fold-const.c (fold_binary_loc): Move transformation...
16059 * match.pd (C - X CMP X): ... here.
16061 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
16063 * config/sparc/sparc.c (sparc_option_override): Set function
16064 alignment for -mcpu=niagara7 to 64 to match the I$ line.
16065 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
16067 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
16069 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
16071 2017-05-18 Marek Polacek <polacek@redhat.com>
16074 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
16075 (pass_ubsan::execute): Call gimple_assign_single_p instead of
16076 gimple_assign_load_p.
16078 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
16080 PR middle-end/80692
16081 * real.c (do_compare): Give decimal_do_compare preference over
16082 comparing just the signs.
16084 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
16086 * doc/md.texi (Canonicalization of Instructions): Describe the
16087 canonical form of instructions that inherently set a condition
16090 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
16092 PR middle-end/80775
16093 * tree-cfg.c: Move deletion of unreachable case statements to after
16094 the merging of consecutive case labels.
16096 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
16098 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
16099 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
16100 restoring of callee-saved registers.
16102 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
16104 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
16105 * config/visium/visium.c (single_set_and_flags): Likewise.
16106 * config/visium/visium.md (Substitutions): Likewise.
16108 2017-05-17 Martin Liska <mliska@suse.cz>
16110 * cfg.c: Introduce dump_flags_t type and
16111 use it instead of int type.
16113 * cfghooks.c: Likewise.
16114 * cfghooks.h (struct cfg_hooks): Likewise.
16115 * cfgrtl.c: Likewise.
16116 * cfgrtl.h: Likewise.
16117 * cgraph.c (cgraph_node::get_body): Likewise.
16118 * coretypes.h: Likewise.
16119 * domwalk.c: Likewise.
16120 * domwalk.h: Likewise.
16121 * dumpfile.c (struct dump_option_value_info): Likewise.
16122 (dump_enable_all): Likewise.
16123 (dump_switch_p_1): Likewise.
16124 (opt_info_switch_p): Likewise.
16125 * dumpfile.h (enum tree_dump_index): Likewise.
16126 (struct dump_file_info): Likewise.
16127 * genemit.c: Likewise.
16128 * generic-match-head.c: Likewise.
16129 * gengtype.c (open_base_files): Likewise.
16130 * gimple-pretty-print.c: Likewise.
16131 * gimple-pretty-print.h: Likewise.
16132 * graph.c (print_graph_cfg): Likewise.
16133 * graphite-scop-detection.c (dot_all_sese): Likewise.
16134 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
16135 * loop-unroll.c (report_unroll): Likewise.
16136 * passes.c (pass_manager::register_one_dump_file): Likewise.
16137 * print-tree.c: Likewise.
16138 * statistics.c: Likewise.
16139 * tree-cfg.c: Likewise.
16140 * tree-cfg.h: Likewise.
16141 * tree-dfa.c: Likewise.
16142 * tree-dfa.h: Likewise.
16143 * tree-dump.c (dump_function): Likewise.
16144 * tree-dump.h (struct dump_info): Likewise.
16145 * tree-pretty-print.c: Likewise.
16146 * tree-pretty-print.h: Likewise.
16147 * tree-ssa-live.c: Likewise.
16148 * tree-ssa-live.h: Likewise.
16149 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
16150 * tree-vect-loop.c: Likewise.
16151 * tree-vect-slp.c: Likewise.
16153 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
16154 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16156 PR tree-optimization/80457
16157 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
16158 of all arguments to a statement as scalar_to_vec operations.
16159 (vectorizable_call): Adjust call to vect_model_simple_cost for
16161 (vectorizable_conversion): Likewise.
16162 (vectorizable_assignment): Likewise.
16163 (vectorizable_shift): Likewise.
16164 (vectorizable_operation): Likewise.
16165 (vectorizable_comparison): Likewise.
16166 (vect_is_simple_cond): Record the def types for operands.
16167 (vectorizable_condition): Likewise, call vect_model_simple_cost.
16168 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
16169 for statement argument count.
16171 2017-05-16 Carl Love <cel@us.ibm.com>
16173 * config/rs6000/rs6000-c: Add support for built-in functions
16174 vector unsigned long long vec_bperm (vector unsigned long long,
16175 vector unsigned char)
16176 vector signed long long vec_mule (vector signed int,
16178 vector unsigned long long vec_mule (vector unsigned int,
16179 vector unsigned int)
16180 vector signed long long vec_mulo (vector signed int,
16182 vector unsigned long long vec_mulo (vector unsigned int,
16183 vector unsigned int)
16184 vector signed char vec_sldw (vector signed char,
16185 vector signed char,
16187 vector unsigned char vec_sldw (vector unsigned char,
16188 vector unsigned char,
16190 vector signed short vec_sldw (vector signed short,
16191 vector signed short,
16193 vector unsigned short vec_sldw (vector unsigned short,
16194 vector unsigned short,
16196 vector signed int vec_sldw (vector signed int,
16199 vector unsigned int vec_sldw (vector unsigned int,
16200 vector unsigned int,
16202 vector signed long long vec_sldw (vector signed long long,
16203 vector signed long long,
16205 vector unsigned long long vec_sldw (vector unsigned long long,
16206 vector unsigned long long,
16208 * config/rs6000/rs6000-c: Add support for built-in functions
16209 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
16210 * config/rs6000/altivec.h: Add defintion for vec_sldw.
16211 * doc/extend.texi: Update the built-in documentation for the
16212 new built-in functions.
16214 2017-05-16 Marek Polacek <polacek@redhat.com>
16218 * tree.c (save_expr): Don't fold the expression.
16220 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
16222 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
16223 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
16224 and (?*y,m). Update insn attributes.
16226 2017-05-16 Martin Liska <mliska@suse.cz>
16228 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
16229 flags argument of print_gimple_stmt, print_gimple_expr,
16230 print_generic_stmt and print_generic_expr.
16231 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
16232 * coretypes.h: Likewise.
16233 * except.c (dump_eh_tree): Likewise.
16234 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
16235 * gimple-pretty-print.h: Likewise.
16236 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
16237 (backprop::push_to_worklist): Likewise.
16238 (backprop::pop_from_worklist): Likewise.
16239 (backprop::process_use): Likewise.
16240 (backprop::intersect_uses): Likewise.
16241 (note_replacement): Likewise.
16242 * gimple-ssa-store-merging.c
16243 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
16244 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
16245 (pass_store_merging::execute): Likewise.
16246 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
16247 (ssa_base_cand_dump_callback): Likewise.
16248 (dump_incr_vec): Likewise.
16249 (replace_refs): Likewise.
16250 (replace_mult_candidate): Likewise.
16251 (create_add_on_incoming_edge): Likewise.
16252 (create_phi_basis): Likewise.
16253 (insert_initializers): Likewise.
16254 (all_phi_incrs_profitable): Likewise.
16255 (introduce_cast_before_cand): Likewise.
16256 (replace_one_candidate): Likewise.
16257 * gimplify.c (gimplify_expr): Likewise.
16258 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
16259 (set_rename): Likewise.
16260 (rename_uses): Likewise.
16261 (copy_loop_phi_nodes): Likewise.
16262 (add_close_phis_to_merge_points): Likewise.
16263 (copy_loop_close_phi_args): Likewise.
16264 (copy_cond_phi_args): Likewise.
16265 (graphite_copy_stmts_from_block): Likewise.
16266 (translate_pending_phi_nodes): Likewise.
16267 * graphite-poly.c (print_pdr): Likewise.
16268 (dump_gbb_cases): Likewise.
16269 (dump_gbb_conditions): Likewise.
16270 (print_scop_params): Likewise.
16271 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
16272 (build_cross_bb_scalars_use): Likewise.
16273 (gather_bbs::before_dom_children): Likewise.
16274 * hsa-dump.c (dump_hsa_immed): Likewise.
16275 * ipa-cp.c (print_ipcp_constant_value): Likewise.
16276 (get_replacement_map): Likewise.
16277 * ipa-inline-analysis.c (dump_condition): Likewise.
16278 (estimate_function_body_sizes): Likewise.
16279 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
16280 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
16281 * ipa-prop.c (ipa_dump_param): Likewise.
16282 (ipa_print_node_jump_functions_for_edge): Likewise.
16283 (ipa_modify_call_arguments): Likewise.
16284 (ipa_modify_expr): Likewise.
16285 (ipa_dump_param_adjustments): Likewise.
16286 (ipa_dump_agg_replacement_values): Likewise.
16287 (ipcp_modif_dom_walker::before_dom_children): Likewise.
16288 * ipa-pure-const.c (check_stmt): Likewise.
16289 (pass_nothrow::execute): Likewise.
16290 * ipa-split.c (execute_split_functions): Likewise.
16291 * omp-offload.c (dump_oacc_loop_part): Likewise.
16292 (dump_oacc_loop): Likewise.
16293 * trans-mem.c (tm_log_emit): Likewise.
16294 (tm_memopt_accumulate_memops): Likewise.
16295 (dump_tm_memopt_set): Likewise.
16296 (dump_tm_memopt_transform): Likewise.
16297 * tree-cfg.c (gimple_verify_flow_info): Likewise.
16298 (print_loop): Likewise.
16299 * tree-chkp-opt.c (chkp_print_addr): Likewise.
16300 (chkp_gather_checks_info): Likewise.
16301 (chkp_get_check_result): Likewise.
16302 (chkp_remove_check_if_pass): Likewise.
16303 (chkp_use_outer_bounds_if_possible): Likewise.
16304 (chkp_reduce_bounds_lifetime): Likewise.
16305 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
16306 (chkp_mark_completed_bounds): Likewise.
16307 (chkp_register_incomplete_bounds): Likewise.
16308 (chkp_mark_invalid_bounds): Likewise.
16309 (chkp_maybe_copy_and_register_bounds): Likewise.
16310 (chkp_build_returned_bound): Likewise.
16311 (chkp_get_bound_for_parm): Likewise.
16312 (chkp_build_bndldx): Likewise.
16313 (chkp_get_bounds_by_definition): Likewise.
16314 (chkp_generate_extern_var_bounds): Likewise.
16315 (chkp_get_bounds_for_decl_addr): Likewise.
16316 * tree-chrec.c (chrec_apply): Likewise.
16317 * tree-data-ref.c (dump_data_reference): Likewise.
16318 (dump_subscript): Likewise.
16319 (dump_data_dependence_relation): Likewise.
16320 (analyze_overlapping_iterations): Likewise.
16321 * tree-inline.c (expand_call_inline): Likewise.
16322 (tree_function_versioning): Likewise.
16323 * tree-into-ssa.c (dump_defs_stack): Likewise.
16324 (dump_currdefs): Likewise.
16325 (dump_names_replaced_by): Likewise.
16326 (dump_update_ssa): Likewise.
16327 (update_ssa): Likewise.
16328 * tree-object-size.c (pass_object_sizes::execute): Likewise.
16329 * tree-parloops.c (build_new_reduction): Likewise.
16330 (try_create_reduction_list): Likewise.
16331 (ref_conflicts_with_region): Likewise.
16332 (oacc_entry_exit_ok_1): Likewise.
16333 (oacc_entry_exit_single_gang): Likewise.
16334 * tree-pretty-print.h: Likewise.
16335 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
16336 (get_scalar_evolution): Likewise.
16337 (add_to_evolution): Likewise.
16338 (get_loop_exit_condition): Likewise.
16339 (analyze_evolution_in_loop): Likewise.
16340 (analyze_initial_condition): Likewise.
16341 (analyze_scalar_evolution): Likewise.
16342 (instantiate_scev): Likewise.
16343 (number_of_latch_executions): Likewise.
16344 (gather_chrec_stats): Likewise.
16345 (final_value_replacement_loop): Likewise.
16346 (scev_const_prop): Likewise.
16347 * tree-sra.c (dump_access): Likewise.
16348 (disqualify_candidate): Likewise.
16349 (create_access): Likewise.
16350 (reject): Likewise.
16351 (maybe_add_sra_candidate): Likewise.
16352 (create_access_replacement): Likewise.
16353 (analyze_access_subtree): Likewise.
16354 (analyze_all_variable_accesses): Likewise.
16355 (sra_modify_assign): Likewise.
16356 (initialize_constant_pool_replacements): Likewise.
16357 (find_param_candidates): Likewise.
16358 (decide_one_param_reduction): Likewise.
16359 (replace_removed_params_ssa_names): Likewise.
16360 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
16361 * tree-ssa-copy.c (dump_copy_of): Likewise.
16362 (copy_prop_visit_cond_stmt): Likewise.
16363 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
16364 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
16365 (record_equivalences_from_stmt): Likewise.
16366 * tree-ssa-dse.c (compute_trims): Likewise.
16367 (delete_dead_call): Likewise.
16368 (delete_dead_assignment): Likewise.
16369 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
16370 (forward_propagate_into_cond): Likewise.
16371 (pass_forwprop::execute): Likewise.
16372 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
16373 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
16375 (move_computations_worker): Likewise.
16376 (execute_sm): Likewise.
16377 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
16378 (remove_exits_and_undefined_stmts): Likewise.
16379 (remove_redundant_iv_tests): Likewise.
16380 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
16381 (adjust_iv_update_pos): Likewise.
16382 * tree-ssa-math-opts.c (bswap_replace): Likewise.
16383 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
16384 (value_replacement): Likewise.
16385 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
16386 * tree-ssa-pre.c (print_pre_expr): Likewise.
16387 (get_representative_for): Likewise.
16388 (create_expression_by_pieces): Likewise.
16389 (insert_into_preds_of_block): Likewise.
16390 (eliminate_insert): Likewise.
16391 (eliminate_dom_walker::before_dom_children): Likewise.
16392 (eliminate): Likewise.
16393 (remove_dead_inserted_code): Likewise.
16394 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
16395 * tree-ssa-reassoc.c (get_rank): Likewise.
16396 (eliminate_duplicate_pair): Likewise.
16397 (eliminate_plus_minus_pair): Likewise.
16398 (eliminate_not_pairs): Likewise.
16399 (undistribute_ops_list): Likewise.
16400 (eliminate_redundant_comparison): Likewise.
16401 (update_range_test): Likewise.
16402 (optimize_range_tests_var_bound): Likewise.
16403 (optimize_vec_cond_expr): Likewise.
16404 (rewrite_expr_tree): Likewise.
16405 (rewrite_expr_tree_parallel): Likewise.
16406 (linearize_expr): Likewise.
16407 (break_up_subtract): Likewise.
16408 (linearize_expr_tree): Likewise.
16409 (attempt_builtin_powi): Likewise.
16410 (attempt_builtin_copysign): Likewise.
16411 (transform_stmt_to_copy): Likewise.
16412 (transform_stmt_to_multiply): Likewise.
16413 (dump_ops_vector): Likewise.
16414 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
16415 (print_scc): Likewise.
16416 (set_ssa_val_to): Likewise.
16417 (visit_reference_op_store): Likewise.
16418 (visit_use): Likewise.
16419 (sccvn_dom_walker::before_dom_children): Likewise.
16420 (run_scc_vn): Likewise.
16421 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
16423 (expr_hash_elt::print): Likewise.
16424 (const_and_copies::pop_to_marker): Likewise.
16425 (const_and_copies::record_const_or_copy_raw): Likewise.
16426 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
16427 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
16428 (dump_predicates): Likewise.
16429 (find_uninit_use): Likewise.
16430 (warn_uninitialized_phi): Likewise.
16431 (pass_late_warn_uninitialized::execute): Likewise.
16432 * tree-ssa.c (verify_vssa): Likewise.
16433 (verify_ssa): Likewise.
16434 (maybe_optimize_var): Likewise.
16435 * tree-vrp.c (dump_value_range): Likewise.
16436 (dump_all_value_ranges): Likewise.
16437 (dump_asserts_for): Likewise.
16438 (register_edge_assert_for_2): Likewise.
16439 (vrp_visit_cond_stmt): Likewise.
16440 (vrp_visit_switch_stmt): Likewise.
16441 (vrp_visit_stmt): Likewise.
16442 (vrp_visit_phi_node): Likewise.
16443 (simplify_cond_using_ranges_1): Likewise.
16444 (fold_predicate_in): Likewise.
16445 (evrp_dom_walker::before_dom_children): Likewise.
16446 (evrp_dom_walker::push_value_range): Likewise.
16447 (evrp_dom_walker::pop_value_range): Likewise.
16448 (execute_early_vrp): Likewise.
16450 2017-05-16 Richard Biener <rguenther@suse.de>
16452 * dwarf2out.c (loc_list_from_tree_1): Do not create
16453 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
16455 2017-05-16 Richard Biener <rguenther@suse.de>
16457 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
16459 (note_variable_value_in_expr): If we resolved the decl ref
16460 do not push to the stack.
16462 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
16464 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
16465 operations in fast-math mode.
16466 (vaddq_f16): Likewise.
16467 (vmul_f16): Likewise.
16468 (vmulq_f16): Likewise.
16469 (vsub_f16): Likewise.
16470 (vsubq_f16): Likewise.
16471 * config/arm/neon.md (add<mode>3): New.
16473 (fma:<VH:mode>3): New. Also remove outdated comment.
16476 2017-05-16 Martin Liska <mliska@suse.cz>
16480 * ipa-devirt.c (warn_types_mismatch): Fix typo.
16481 (odr_types_equivalent_p): Likewise.
16483 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
16485 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
16487 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
16490 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
16491 non-interunit SSE move alternatives with '?'.
16492 (zero-extendsidi peephole2): New peephole to skip intermediate
16493 general register in SSE zero-extend sequence.
16495 2017-05-15 Jeff Law <law@redhat.com>
16497 * reorg.c (relax_delay_slots): Create a new variable to hold
16498 the temporary target rather than clobbering TARGET_LABEL.
16500 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
16501 missing argument to extract_bit_field call.
16502 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
16504 2017-05-15 Martin Liska <mliska@suse.cz>
16507 * gcc.c (process_command): Do not allow empty argument of -o option.
16509 2017-05-15 Renlin Li <renlin.li@arm.com>
16511 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
16512 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
16513 * config/aarch64/constraints.md (Usf): Add long call check.
16514 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
16515 (call_value): Likewise.
16516 (sibcall): Likewise.
16517 (sibcall_value): Likewise.
16519 (call_value_insn): New.
16520 (sibcall_insn): Update rtx pattern.
16521 (sibcall_value_insn): Likewise.
16522 (call_internal): Remove.
16523 (call_value_internal): Likewise.
16524 (sibcall_internal): Likewise.
16525 (sibcall_value_internal): Likewise.
16526 (call_reg): Likewise.
16527 (call_symbol): Likewise.
16528 (call_value_reg): Likewise.
16529 (call_value_symbol): Likewise.
16531 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
16534 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
16536 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
16538 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
16539 compatible with CCGOCmode and with CCZmode.
16541 2017-05-14 Martin Sebor <msebor@redhat.com>
16543 PR middle-end/77671
16544 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
16545 (gimple_fold_builtin_snprintf): Same.
16546 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
16547 (gimple_fold_builtin_snprintf): Same.
16548 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
16549 of character types.
16550 (is_call_safe): New function.
16551 (try_substitute_return_value): Call it.
16552 (try_simplify_call): New function.
16553 (pass_sprintf_length::handle_gimple_call): Call it.
16555 2017-05-14 Martin Sebor <msebor@redhat.com>
16557 PR middle-end/80669
16558 * builtins.c (expand_builtin_stpncpy): Simplify.
16560 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
16562 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
16563 * config/i386/i386.h
16564 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
16565 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
16566 (struct machine_function): Add new members call_ms2sysv,
16567 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
16568 (struct machine_frame_state): New fields sp_realigned and
16569 sp_realigned_offset.
16570 * config/i386/i386.c
16571 (enum xlogue_stub): New enum.
16572 (enum xlogue_stub_sets): New enum.
16573 (class xlogue_layout): New class.
16574 (struct ix86_frame): New fields stack_realign_allocate_offset,
16575 stack_realign_offset and outlined_save_offset. Modify comments to
16576 detail stack layout when using out-of-line stubs.
16577 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
16578 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
16579 -mcall-ms2sysv-xlogues.
16580 (stub_managed_regs): New static variable.
16581 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
16582 registers managed by out-of-line stub.
16583 (disable_call_ms2sysv_xlogues): New function.
16584 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
16585 m->call_ms2sysv when appropriate and compute frame layout for
16587 (sp_valid_at, fp_valid_at): New inline functions.
16588 (choose_basereg): New function.
16589 (choose_baseaddr): Add align parameter, use choose_basereg and modify
16591 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
16592 Use align parameter of choose_baseaddr to generated aligned SSE movs
16594 (pro_epilogue_adjust_stack): Modify to track
16595 machine_frame_state::sp_realigned.
16596 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
16597 (ix86_nsaved_sseregs): Likewise.
16598 (ix86_emit_save_regs): Likewise.
16599 (ix86_emit_save_regs_using_mov): Likewise.
16600 (ix86_emit_save_sse_regs_using_mov): Likewise.
16601 (get_scratch_register_on_entry): Likewise.
16602 (gen_frame_set): New function.
16603 (gen_frame_load): Likewise.
16604 (gen_frame_store): Likewise.
16605 (emit_outlined_ms2sysv_save): Likewise.
16606 (emit_outlined_ms2sysv_restore): Likewise.
16607 (ix86_expand_prologue): Modify stack re-alignment code and call
16608 emit_outlined_ms2sysv_save when appropriate.
16609 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
16610 parameter rtx_insn *insn, which allows the function to be used to only
16611 generate the notes.
16612 (ix86_expand_epilogue): Modify validity checks of frame and stack
16613 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
16614 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
16615 * config/i386/predicates.md
16616 (save_multiple): New predicate.
16617 (restore_multiple): Likewise.
16618 * config/i386/sse.md
16619 (save_multiple<mode>): New pattern.
16620 (save_multiple_realign<mode>): Likewise.
16621 (restore_multiple<mode>): Likewise.
16622 (restore_multiple_and_return<mode>): Likewise.
16623 (restore_multiple_leave_return<mode>): Likewise.
16624 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
16626 2017-05-14 Julia Koval <julia.koval@intel.com>
16628 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
16629 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
16630 (__builtin_ia32_xsetbv): New builtins.
16631 * config/i386/i386.c (ix86_expand_special_args_builtin):
16633 (ix86_expand_builtin): Special expand for new intrinsics.
16634 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
16635 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
16636 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
16638 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16640 * cfganal.c (inverted_post_order_compute): Change argument type
16642 * cfganal.h (inverted_post_order_compute): Adjust prototype.
16643 * df-core.c (rest_of_handle_df_initialize): Adjust.
16644 (rest_of_handle_df_finish): Likewise.
16645 (df_analyze_1): Likewise.
16646 (df_analyze): Likewise.
16647 (loop_inverted_post_order_compute): Change argument to be a vec *.
16648 (df_analyze_loop): Adjust.
16649 (df_get_n_blocks): Likewise.
16650 (df_get_postorder): Likewise.
16651 * df.h (struct df_d): Change field to be a vec.
16652 * lcm.c (compute_laterin): Adjust.
16653 (compute_available): Likewise.
16654 * lra-lives.c (lra_create_live_ranges_1): Likewise.
16655 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
16656 * tree-ssa-pre.c (compute_antic): Likewise.
16658 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16660 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
16661 (depth_first_search::depth_first_search): Change structure init
16662 function to this constructor.
16663 (depth_first_search::add_bb): Rename function to this member.
16664 (depth_first_search::execute): Likewise.
16665 (flow_dfs_compute_reverse_finish): Adjust.
16667 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16669 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
16670 (longest_simple_path): Likewise.
16671 * shrink-wrap.c (spread_components): Likewise.
16672 (disqualify_problematic_components): Likewise.
16673 (emit_common_heads_for_components): Likewise.
16674 (emit_common_tails_for_components): Likewise.
16675 (insert_prologue_epilogue_for_components): Likewise.
16677 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16679 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
16682 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16684 * df-core.c (df_set_blocks): Start using auto_bitmap.
16685 (df_compact_blocks): Likewise.
16686 * df-problems.c (df_rd_confluence_n): Likewise.
16687 * df-scan.c (df_insn_rescan_all): Likewise.
16688 (df_process_deferred_rescans): Likewise.
16689 (df_update_entry_block_defs): Likewise.
16690 (df_update_exit_block_uses): Likewise.
16691 (df_entry_block_bitmap_verify): Likewise.
16692 (df_exit_block_bitmap_verify): Likewise.
16693 (df_scan_verify): Likewise.
16694 * lra-constraints.c (lra_constraints): Likewise.
16695 (undo_optional_reloads): Likewise.
16696 (lra_undo_inheritance): Likewise.
16697 * lra-remat.c (calculate_gen_cands): Likewise.
16698 (do_remat): Likewise.
16699 * lra-spills.c (assign_spill_hard_regs): Likewise.
16700 (spill_pseudos): Likewise.
16701 * tree-ssa-pre.c (bitmap_set_and): Likewise.
16702 (bitmap_set_subtract_values): Likewise.
16704 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16706 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
16707 management with auto_bitmap.
16708 (fix_inter_tick): Likewise.
16709 (fix_recovery_deps): Likewise.
16710 * ira.c (add_store_equivs): Likewise.
16711 (find_moveable_pseudos): Likewise.
16712 (split_live_ranges_for_shrink_wrap): Likewise.
16713 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
16714 (rtx_reuse_manager::seen_def_p): Likewise.
16715 (rtx_reuse_manager::set_seen_def): Likewise.
16716 * print-rtl.h (class rtx_reuse_manager): Likewise.
16718 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16720 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
16722 (migrate_btr_def): Likewise.
16723 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
16724 * df-core.c (loop_post_order_compute): Likewise.
16725 (loop_inverted_post_order_compute): Likewise.
16726 * hsa-common.h: Likewise.
16727 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
16728 * init-regs.c (initialize_uninitialized_regs): Likewise.
16729 * ipa-inline.c (resolve_noninline_speculation): Likewise.
16730 (inline_small_functions): Likewise.
16731 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
16732 * ira.c (combine_and_move_insns): Likewise.
16733 (build_insn_chain): Likewise.
16734 * loop-invariant.c (find_invariants): Likewise.
16735 * lower-subreg.c (propagate_pseudo_copies): Likewise.
16736 * predict.c (tree_predict_by_opcode): Likewise.
16737 (predict_paths_leading_to): Likewise.
16738 (predict_paths_leading_to_edge): Likewise.
16739 (estimate_loops_at_level): Likewise.
16740 (estimate_loops): Likewise.
16741 * shrink-wrap.c (try_shrink_wrapping): Likewise.
16742 (spread_components): Likewise.
16743 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
16744 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
16745 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
16746 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
16747 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
16748 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
16749 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
16750 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
16751 (mark_threaded_blocks): Likewise.
16752 (thread_through_all_blocks): Likewise.
16753 * tree-ssa.c (verify_ssa): Likewise.
16754 (execute_update_addresses_taken): Likewise.
16755 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
16757 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16759 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
16761 (post_order_compute): Likewise.
16762 (inverted_post_order_compute): Likewise.
16763 (pre_and_rev_post_order_compute_fn): Likewise.
16765 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16767 * genrecog.c (int_set::int_set): Explicitly construct our
16768 auto_vec base class.
16769 * vec.h (auto_vec::auto_vec): New constructor.
16771 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16773 * bitmap.h (class auto_bitmap): New constructor taking
16774 bitmap_obstack * argument.
16776 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16778 * bitmap.h (class auto_bitmap): Change type of m_bits to
16779 bitmap_head, and adjust ctor / dtor and member operators.
16781 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
16783 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
16784 when returned register mode doesn't match original mode.
16786 2017-05-12 Jeff Law <law@redhat.com>
16787 Jakub Jelinek <jakub@redhat.com>
16789 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
16790 we look for cc setter after the compare-elim changes.
16791 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
16792 within the vector to match what compare-elim now expects.
16793 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
16794 (xorsi3_flags, one_cmplsi2_flags): Likewise.
16796 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
16797 after the compare-elim changes.
16798 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
16799 the vector to match what compare-elim now expects.
16800 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
16801 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
16802 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
16803 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
16804 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
16806 * config/visium/visium.c (single_set_and_flags): Fix where
16807 we look for cc setter after the compare-elim changes.
16808 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
16809 with the vector to match what compare-elim now expects.
16810 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
16811 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
16812 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
16813 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
16814 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
16815 (neg<mode>2_insn_set_overflow): Likewise.
16817 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
16819 PR middle-end/79794
16820 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
16821 maybe_expand_insn call, set ops[0].target. If still set after call,
16822 set alt_rtl. Add extra arg to recursive calls.
16823 (extract_bit_field): Add alt_rtl argument. Pass to
16825 * expmed.h (extract_bit_field): Fix prototype.
16826 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
16827 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
16828 to extract_bit_field_calls.
16829 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
16830 Pass alt_rtl to extract_bit_field calls.
16831 * calls.c (store_unaligned_arguments_into_psuedos)
16832 load_register_parameters): Pass extra NULL to extract_bit_field calls.
16833 * optabs.c (maybe_legitimize_operand): Clear op->target when call
16835 * optabs.h (struct expand_operand): Add target bitfield.
16837 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
16839 * compare-elim.c (try_eliminate_compare): Canonicalize
16840 operation with embedded compare to
16841 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
16842 (set (reg) (operation)].
16844 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
16846 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
16849 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
16850 cost of adding a carry flag for ADC instruction.
16851 [case MINUS]: Ignore the cost of subtracting a carry flag
16852 for SBB instruction.
16854 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
16856 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
16858 * config/rs6000/bmiintrin.h: New file.
16859 * config/rs6000/bmi2intrin.h: New file.
16860 * config/rs6000/x86intrin.h: New file.
16862 2017-05-12 Jeff Law <law@redhat.com>
16864 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
16867 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
16869 PR middle-end/80707
16870 * tree-cfg.c: Remove cfg edges of unreachable case statements.
16872 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
16874 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
16875 early expansion of vector divide builtins.
16876 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
16877 builtins identified as having unsigned arguments.
16879 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
16881 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
16882 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
16883 expansion of vector logical operations (and, andc, or, xor,
16886 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
16888 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
16889 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
16891 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
16893 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
16894 early GIMPLE expansion of vector multiplies.
16896 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
16898 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
16899 TARGET_HAVE_MOVT conditional.
16900 (movt splitter): Likewise.
16902 2017-05-12 Richard Biener <rguenther@suse.de>
16904 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
16905 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
16906 Fold all stmts not inplace.
16908 2017-05-12 Richard Biener <rguenther@suse.de>
16910 PR tree-optimization/80713
16911 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
16912 inserted_exprs bit for not removed stmts.
16914 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
16916 PR middle-end/69921
16917 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
16918 parallelized" attribute for parallelized OpenACC kernels.
16919 * omp-offload.c (execute_oacc_device_lower): Use it.
16921 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
16922 Set "oacc kernels" attribute.
16923 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
16924 parameter. Adjust all users.
16925 (oacc_fn_attrib_kernels_p): Remove function.
16926 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
16927 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
16928 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
16929 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
16930 assert "oacc kernels" attribute is set.
16932 2017-05-11 Carl Love <cel@us.ibm.com>
16934 * config/rs6000/rs6000-c: Add support for built-in functions
16935 vector unsigned char vec_popcnt (vector signed char)
16936 vector unsigned char vec_popcnt (vector unsigned char)
16937 vector unsigned short vec_popcnt (vector signed short)
16938 vector unsigned short vec_popcnt (vector unsigned short)
16939 vector unsigned int vec_popcnt (vector signed int)
16940 vector unsigned int vec_popcnt (vector unsigned int)
16941 vector unsigned long long vec_popcnt (vector signed long long)
16942 vector unsigned long long vec_popcnt (vector unsigned long long)
16943 vector signed long long vec_slo (vector signed long long,
16944 vector signed char)
16945 vector signed long long vec_slo (vector signed long long,
16946 vector unsigned char)
16947 vector unsigned long long vec_slo (vector unsigned long long,
16948 vector signed char)
16949 vector unsigned long long vec_slo (vector unsigned long long,
16950 vector unsigned char)
16951 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
16952 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
16953 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
16954 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
16955 * doc/extend.texi: Update the built-in documentation file for the
16956 new built-in functions.
16958 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
16960 * attribs.h (sorted_attr_string): Move machine independent
16961 functions for target clone support from the i386 port to common
16962 code. Rename ix86_function_versions to common_function_versions.
16963 Rename make_name to make_unique_name.
16964 (common_function_versions): Likewise.
16965 (make_unique_name): Likewise.
16966 (make_dispatcher_decl): Likewise.
16967 (is_function_default_version): Likewise.
16968 * attribs.c (attr_strcmp): Likewise.
16969 (sorted_attr_string): Likewise.
16970 (common_function_versions): Likewise.
16971 (make_unique_name): Likewise.
16972 (make_dispatcher_decl): Likewise.
16973 (is_function_default_version): Likewise.
16974 * config/i386/i386.c (attr_strcmp): Likewise.
16975 (sorted_attr_string): Likewise.
16976 (ix86_function_versions): Likewise.
16977 (make_name): Likewise.
16978 (make_dispatcher_decl): Likewise.
16979 (is_function_default_version): Likewise.
16980 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
16982 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16985 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
16986 Account for direct move costs for vec_construct of integer
16989 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
16992 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
16993 (UNSPEC_STX_ATOMIC): Ditto.
16994 (loaddi_via_sse): New insn.
16995 (storedi_via_sse): Ditto.
16996 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
16997 Update corresponding peephole2 patterns.
16998 (atomic_storedi_fpu): Ditto.
17000 2017-05-11 Julia Koval <julia.koval@intel.com>
17002 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
17003 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
17005 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
17006 (__builtin_ia32_rsqrt14ss_mask): New builtins.
17007 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
17009 2017-05-11 Nathan Sidwell <nathan@acm.org>
17011 * graphite-poly.c: Include dumpfile.h.
17013 * dumpfle.h (dump_function): Declare here ...
17014 * tree-dump.h (dump_function): ... not here.
17015 * dumpfile.c: #include tree-cfg.h.
17016 (dump_function): Move here from ...
17017 * tree-dump.c (dump_function): ... here.
17018 * gimplify.c: #include splay-tree.h, not tree-dump.h.
17019 * graphite-poly.c: Don't include tree-dump.h.
17020 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
17021 * print-tree.c: Likewise.
17022 * stor-layout.c: Likewise.
17023 * tree-nested.c: Likewise.
17025 * dumpfile.c (dump_start): Use TDF_FLAGS.
17026 (dump_enable_all): Fix TDF_KIND check thinko.
17028 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
17030 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17031 array entries to represent two legal parameterizations of the
17032 overloaded __builtin_cmpb function, as represented by the
17033 P6_OV_BUILTIN_CMPB constant.
17034 (altivec_resolve_overloaded_builtin): Add special case handling
17035 for the __builtin_cmpb function, as represented by the
17036 P6_OV_BUILTIN_CMPB constant.
17037 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
17038 (BU_P6_64BIT_2): New macro.
17039 (BU_P6_OVERLOAD_2): New macro
17040 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
17041 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
17042 (CMPB): Add overload support to represent both 32-bit and 64-bit
17043 compare-bytes function.
17044 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
17045 support for TARGET_CMPB.
17046 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
17047 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
17048 documentation of the __builtin_cmpb overloaded built-in function.
17050 2017-05-11 Richard Biener <rguenther@suse.de>
17052 PR tree-optimization/80705
17053 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
17054 bases are not vectorizable.
17056 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17058 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
17059 when counting register pressure.
17061 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17063 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
17064 (iv_ca_more_deps): Renamed to ...
17065 (iv_ca_compare_deps): ... this.
17066 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
17068 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17070 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
17072 (determine_group_iv_costs): ... here.
17073 (find_inv_vars_cb): Record inv var if it's not recorded before.
17075 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17077 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
17078 (get_shiftadd_cost): Ditto.
17080 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17082 * tree-ssa-address.c: Include header file.
17083 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
17085 (add_to_parts): Refactor.
17086 (addr_to_parts): New parameter. Update use of move_hint_to_base.
17087 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
17090 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17092 PR tree-optimization/53090
17093 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
17095 (extract_cond_operands): Detect condition with IV on both sides
17096 and return COMP_IV_EXPR_2.
17097 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
17098 (rewrite_use_compare): Simplify by removing call to function
17099 extract_cond_operands.
17101 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17103 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
17104 (extract_cond_operands): Detect condition comparing against non-
17105 invariant bound and return appropriate enum value.
17106 (find_interesting_uses_cond): Update use of extract_cond_operands.
17107 Handle its return value accordingly.
17108 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
17110 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17112 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
17113 nonlinear iv_use computation in loop invariant sensitive way.
17115 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17117 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
17118 (find_iv_candidates): Call relate_compare_use_with_all_cands.
17120 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17122 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
17123 (dump_cand): Support iv_cand.inv_exprs.
17124 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
17126 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
17129 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17131 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
17133 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
17134 as local function. Include necessary header files.
17135 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
17137 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17139 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
17141 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17143 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
17144 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
17145 RSHIFT_EXPR and BIT_NOT_EXPR.
17147 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17149 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
17150 (adjust_setup_cost): New parameter supporting round up adjustment.
17151 (struct address_cost_data): Delete.
17152 (force_expr_to_var_cost): Don't bound cost with spill_cost.
17153 (split_address_cost, ptr_difference_cost): Delete.
17154 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
17155 (struct ainc_cost_data): New struct.
17156 (get_address_cost_ainc): New function.
17157 (get_address_cost, get_computation_cost): Reimplement.
17158 (determine_group_iv_cost_address): Record inv_expr for all uses of
17160 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
17161 (iv_ca_has_deps): Reimplemented to ...
17162 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
17164 (iv_ca_extend): Call iv_ca_more_deps.
17166 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17168 * tree-ssa-address.c (struct mem_address): Move to header file.
17169 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
17170 * tree-ssa-address.h (struct mem_address): Move from C file.
17171 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
17173 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17175 * tree-affine.h (aff_combination_type): New interface.
17176 (aff_combination_zero_p): Remove static.
17177 (aff_combination_const_p): New interface.
17178 (aff_combination_singleton_var_p): New interfaces.
17180 2017-05-11 Richard Biener <rguenther@suse.de>
17182 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
17183 Skip unreachable blocks and destinations.
17184 (eliminate): Move stmt removal and fixup ...
17185 (fini_eliminate): ... here. Skip inserted exprs.
17186 (pass_pre::execute): Move fini_pre after fini_eliminate.
17187 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
17188 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
17189 PRE to get rid of dead code that has invalid SSA form and
17190 split critical edges again.
17192 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17194 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
17196 2017-05-11 Richard Biener <rguenther@suse.de>
17198 * passes.c (execute_function_todo): Verify loops if they are
17199 said to be up-to-date.
17200 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
17201 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
17203 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
17206 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
17207 handle calling assemble_external ourself.
17210 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
17211 modes with zero size. Enhance comment.
17213 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17215 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
17216 built-ins for vec_xl and vec_xst with short and char pointer
17219 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
17221 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
17222 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
17223 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
17224 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
17225 (_mm_maskz_min_round_ss): New intrinsics.
17226 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
17227 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
17228 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
17229 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
17230 (__builtin_ia32_minss_mask_round): New builtins.
17231 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
17232 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
17233 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
17235 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
17236 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
17238 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
17241 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
17243 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
17244 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
17245 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
17246 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
17247 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
17248 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
17249 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
17250 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
17251 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
17252 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
17253 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
17254 (__builtin_ia32_mulss_mask_round): New builtins.
17255 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
17256 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
17257 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
17259 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
17260 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
17262 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
17265 2017-05-10 Julia Koval <julia.koval@intel.com>
17267 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
17268 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
17269 (_mm256_setr_m128i): New intrinsics.
17271 2017-05-10 Julia Koval <julia.koval@intel.com>
17273 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
17274 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
17275 (_mm_maskz_rcp14_ss): New intrinsics.
17276 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
17277 (__builtin_ia32_rcp14ss_mask): New builtins.
17278 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
17280 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
17282 PR tree-optimization/51513
17283 * tree-cfg.c (gimple_seq_unreachable_p): New function.
17284 (assert_unreachable_fallthru_edge_p): Use it.
17285 (group_case_labels_stmt): Likewise.
17286 * tree-cfg.h: Prototype it.
17287 * stmt.c: Include cfghooks.h and tree-cfg.h.
17288 (emit_case_dispatch_table) <gap_label>: New local variable.
17289 Use it to fill dispatch table gaps.
17290 Test for default_label before updating probabilities.
17291 (expand_case) <default_label>: Remove unneeded initialization.
17292 Test for unreachable default case statement and remove its edge.
17293 Set default_label accordingly.
17294 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
17296 2017-05-10 Carl Love <cel@us.ibm.com>
17298 * config/rs6000/rs6000-c: Add support for built-in functions
17299 vector signed char vec_neg (vector signed char)
17300 vector signed short int vec_neg (vector short int)
17301 vector signed int vec_neg (vector signed int)
17302 vector signed long long vec_neg (vector signed long long)
17303 vector float vec_neg (vector float)
17304 vector double vec_neg (vector double)
17305 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
17307 * config/rs6000/altivec.h: Add define for vec_neg
17308 * doc/extend.texi: Update the built-in documentation for the
17309 new built-in functions.
17311 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17313 PR tree-optimization/77644
17314 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
17316 2017-05-10 Nathan Sidwell <nathan@acm.org>
17318 * dumpfile.h (TDI_lang_all): New.
17319 (TDF_KIND): New. Renumber others
17320 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
17322 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
17324 (get_dump_file_name): Adjust suffix generation.
17325 (dump_enable_all): Use TDF_KIND.
17326 * doc/invoke.texi (-fdump-lang-all): Document.
17328 * dumpfile.h: Tabify.
17330 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
17333 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
17334 Move member access before delete.
17336 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
17338 * tree-inline.c (expand_call_inline): Split block at stmt
17341 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
17344 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
17345 are now unused after splitting mov{sf,sd}_hardfloat.
17346 (f32_lr2): Likewise.
17347 (f32_lm): Likewise.
17348 (f32_lm2): Likewise.
17349 (f32_li): Likewise.
17350 (f32_li2): Likewise.
17351 (f32_lv): Likewise.
17352 (f32_sr): Likewise.
17353 (f32_sr2): Likewise.
17354 (f32_sm): Likewise.
17355 (f32_sm2): Likewise.
17356 (f32_si): Likewise.
17357 (f32_si2): Likewise.
17358 (f32_sv): Likewise.
17359 (f32_dm): Likewise.
17360 (f32_vsx): Likewise.
17361 (f32_av): Likewise.
17362 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
17363 For movsf, order stores so the VSX stores occur before the GPR
17364 store which encourages the register allocator to use a traditional
17365 FPR instead of a GPR. For movsd, order the stores so that the GPR
17366 store comes before the VSX stores to allow the power6 to work.
17367 This is due to the power6 not having a 32-bit integer store
17368 instruction from a FPR.
17369 (movsf_hardfloat): Likewise.
17370 (movsd_hardfloat): Likewise.
17372 2017-05-09 Martin Sebor <msebor@redhat.com>
17374 PR translation/80280
17375 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
17378 PR translation/80280
17379 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
17380 data member added in r247778.
17381 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
17383 2017-05-09 Nathan Sidwell <nathan@acm.org>
17385 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
17387 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
17390 2017-05-09 Marek Polacek <polacek@redhat.com>
17392 * doc/invoke.texi: Fix typo.
17394 2017-05-09 Richard Biener <rguenther@suse.de>
17396 * tree-vrp.c (vrp_val_is_max): Adjust comment.
17397 (vrp_val_is_min): Likewise.
17398 (set_value_range_to_value): Likewise.
17399 (set_value_range_to_nonnegative): Likewise.
17400 (gimple_assign_nonzero_p): Likewise.
17401 (gimple_stmt_nonzero_p): Likewise.
17402 (vrp_int_const_binop): Likewise. Remove unreachable case.
17403 (adjust_range_with_scev): Adjust comments.
17404 (compare_range_with_value): Likewise.
17405 (extract_range_from_phi_node): Likewise.
17406 (test_for_singularity): Likewise.
17408 2017-05-09 Richard Biener <rguenther@suse.de>
17410 * tree-vrp.c (get_single_symbol): Add assert that we don't
17411 get overflowed constants as invariant part.
17412 (compare_values_warnv): Add comment before the TREE_NO_WARNING
17413 checks. Use wi::cmp instead of recursing for integer constants.
17414 (compare_values): Just ignore whether we assumed undefined
17415 overflow instead of failing the compare.
17416 (extract_range_for_var_from_comparison_expr): Add comment before the
17417 TREE_NO_WARNING sets.
17418 (test_for_singularity): Likewise.
17419 (extract_range_from_comparison): Do not disable optimization
17420 when we assumed undefined overflow.
17421 (extract_range_basic): Remove init of unused var.
17423 2017-05-09 Richard Biener <rguenther@suse.de>
17425 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
17426 (extract_range_from_multiplicative_op_1): Adjust.
17427 (extract_range_from_binary_expr_1): Use int_const_binop.
17429 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
17432 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
17433 rs6000_store_data_bypass_p in seven define_bypass directives and
17434 in several comments.
17435 * config/rs6000/rs6000-protos.h: Add prototype for
17436 rs6000_store_data_bypass_p function.
17437 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
17438 function implements slightly different (rs6000-specific) semantics
17439 than store_data_bypass_p, returning false rather than aborting
17440 with assertion error when arguments do not satisfy the
17441 requirements of store data bypass.
17442 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
17443 rs6000_store_data_bypass_p.
17445 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
17447 * config/xtensa/xtensa-protos.h
17448 (xtensa_initial_elimination_offset): New declaration.
17449 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
17450 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
17451 macro definition, add case for FRAME_POINTER_REGNUM when
17452 FRAME_GROWS_DOWNWARD.
17453 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
17454 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
17455 xtensa_initial_elimination_offset.
17457 2017-05-08 Nathan Sidwell <nathan@acm.org>
17459 * doc/invoke.texi: Alphabetize -fdump options.
17461 2017-05-08 Martin Sebor <msebor@redhat.com>
17463 PR translation/80280
17464 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
17466 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
17468 * target.def (compute_frame_layout): New optional target hook.
17469 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
17470 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
17471 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
17473 * reload1.c (verify_initial_elim_offsets): Likewise.
17474 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
17475 (use_simple_return_p): Call arm_compute_frame_layout if needed.
17476 (arm_get_frame_offsets): Split up into this ...
17477 (arm_compute_frame_layout): ... and this function.
17479 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
17481 * config/aarch64/constraints.md (Usa): New constraint.
17482 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
17484 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
17486 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
17487 with_multilib_list after it has been checked.
17489 2017-05-08 Richard Biener <rguenther@suse.de>
17491 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
17492 (bitmap_set_subtract_values): Likewise.
17494 2017-05-08 Richard Biener <rguenther@suse.de>
17496 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
17497 (gimple_assign_nonzero): ... this and remove strict_overflow_p
17499 (gimple_stmt_nonzero_warnv_p): Rename to ...
17500 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
17502 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
17503 (extract_range_basic): Adjust, do not disable propagation on
17504 strict overflow sensitive simplification.
17505 (vrp_visit_cond_stmt): Likewise.
17507 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
17509 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
17510 body size unconditionally.
17512 2017-05-07 Jeff Law <law@redhat.com>
17515 2017-05-06 Jeff Law <law@redhat.com>
17516 PR tree-optimization/78496
17517 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
17520 PR tree-optimization/78496
17521 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
17522 (simplify_stmt_using_ranges): Call it.
17523 (vrp_dom_walker::before_dom_children): Extract equivalences
17524 from an ASSERT_EXPR with an equality comparison against a
17527 2017-05-06 Jeff Law <law@redhat.com>
17529 PR tree-optimization/78496
17530 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
17533 PR tree-optimization/78496
17534 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
17535 (simplify_stmt_using_ranges): Call it.
17536 (vrp_dom_walker::before_dom_children): Extract equivalences
17537 from an ASSERT_EXPR with an equality comparison against a
17540 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
17542 * lra-constraints.c (lra_copy_reg_equiv): New function.
17543 (split_reg): Use it to copy equivalence information from the
17544 original register to the spill register.
17546 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
17548 PR rtl-optimization/75964
17549 * simplify-rtx.c (simplify_const_relational_operation): Remove
17550 invalid handling of comparisons of integer ABS.
17552 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
17554 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
17555 initialize to zero.
17556 (init_regs): Remove declaration.
17557 (function_arg_advance_32): Initialize error_p as boolean variable.
17559 2017-05-05 Nathan Sidwell <nathan@acm.org>
17561 * store-motion.c (remove_reachable_equiv_notes): Reformat long
17562 lines. Use for (;;).
17564 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17566 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
17567 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
17568 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
17569 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
17570 VF=2 that require versioning.
17572 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17574 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
17577 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17579 * diagnostic.h (diagnostic_override_option_index): Convert from
17580 macro to inline function.
17582 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17584 * diagnostic.c (last_module_changed_p): New function.
17585 (set_last_module): New function.
17586 (diagnostic_report_current_module): Convert macro usage to
17587 the above functions.
17588 * diagnostic.h (diagnostic_context::last_module): Strengthen
17589 from const line_map * to const line_map_ordinary *.
17590 (diagnostic_last_module_changed): Delete macro.
17591 (diagnostic_set_last_module): Delete macro.
17593 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17595 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
17596 with diagnostic_report_diagnostic.
17597 (diagnostic_n_impl_richloc): Likewise.
17598 * diagnostic.h (report_diagnostic): Delete macro.
17599 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
17600 with diagnostic_report_diagnostic.
17601 * substring-locations.c (format_warning_va): Likewise.
17603 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17605 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
17606 save/restor of format_spec. Move option-printing code to...
17607 (print_option_information): ...this new function, and
17608 reimplement by simply printing to the pretty_printer,
17609 rather than appending to the format string.
17611 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17613 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
17614 handling logic into...
17615 (update_effective_level_from_pragmas): ...this new function.
17617 2017-05-04 Andrew Waterman <andrew@sifive.com>
17619 * config/riscv/riscv.opt (mstrict-align): New option.
17620 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
17621 (SLOW_UNALIGNED_ACCESS): Define.
17622 (riscv_slow_unaligned_access): Declare.
17623 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
17625 (riscv_slow_unaligned_access): New variable.
17626 (rocket_tune_info): Set slow_unaligned_access to true.
17627 (optimize_size_tune_info): Set slow_unaligned_access to false.
17628 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
17629 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
17630 (riscv_option_override): Set riscv_slow_unaligned_access.
17631 * doc/invoke.texi: Add -mstrict-align to RISC-V.
17633 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
17635 * config/riscv/riscv.md: Unify indentation.
17637 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
17642 * config/rs6000/rs6000.md (u code attribute): Add FIX and
17644 (extendsi<mode>2): Add support for doing sign extension via
17645 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
17646 don't have ISA 3.0 instructions.
17647 (extendsi<mode>2 splitter): Likewise.
17648 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
17649 generate the normal insns since SImode can now go in vector
17650 registers. Disallow the special UNSPECs needed for previous
17651 machines to hide SImode being used. Add new insns
17652 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
17653 (fix_trunc<mode>si2_stfiwx): Likewise.
17654 (fix_trunc<mode>si2_internal): Likewise.
17655 (fixuns_trunc<mode>si2): Likewise.
17656 (fixuns_trunc<mode>si2_stfiwx): Likewise.
17657 (fctiw<u>z_<mode>_smallint): Likewise.
17658 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
17659 of floating point to 32-bit integer from doing a direct move to
17660 the GPR registers to do a store.
17661 (fctiwz_<mode>): Break long line.
17663 2017-05-05 Bin Cheng <bin.cheng@arm.com>
17665 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
17666 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
17667 (addr_list, addr_offset_valid_p): New.
17668 (split_address_groups): Check offset validity with above function.
17669 (gt-tree-ssa-loop-ivopts.h): Include header file.
17671 2017-05-05 Nathan Sidwell <nathan@acm.org>
17673 * config.gcc (arm*-*-*): Add missing 'fi'.
17675 2017-05-05 Steve Ellcey <sellcey@cavium.com>
17677 * doc/invoke.texi (-fopt-info): Explicitly say order of options
17678 included in -fopt-info does not matter.
17679 * doc/optinfo.texi (-fopt-info): Fix description of default
17680 behavour. Explicitly say order of options included in -fopt-info
17683 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
17685 * config.gcc: Allow combinations of aprofile and rmprofile values for
17686 --with-multilib-list.
17687 * config/arm/t-multilib: New file.
17688 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
17689 variables. Remove setting of ISA and floating-point ABI in
17690 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
17691 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
17692 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
17693 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
17695 * config/arm/t-rmprofile: Likewise except for the matches changes.
17696 * doc/install.texi (--with-multilib-list): Document the combination of
17697 aprofile and rmprofile values and warn about pitfalls in doing that.
17699 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
17701 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
17702 (movdi_aarch64): Likewise.
17704 2017-05-05 Jakub Jelinek <jakub@redhat.com>
17706 PR tree-optimization/80632
17707 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
17709 (build_arrays): Initialize it for virtual phis.
17710 (fix_phi_nodes): Use it for virtual phis.
17712 PR tree-optimization/80558
17713 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
17714 [x, y] op z into [x op, y op z] for op & or | if conditions
17717 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
17718 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
17721 * config/arm/arm.md (use_literal_pool): Remove.
17722 (64-bit immediate split): No longer takes cost into consideration
17723 if arm_disable_literal_pool is enabled.
17724 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
17725 used when arm_disable_literal_pool is enabled.
17726 (arm_max_const_double_inline_cost): Remove use of
17727 arm_disable_literal_pool.
17728 (push_minipool_fix): Add assert.
17729 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
17730 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
17731 (no_literal_pool_sf_immediate): New.
17733 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17735 PR tree-optimization/80613
17736 * tree-ssa-dce.c (propagate_necessity): Remove cases for
17737 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
17739 2017-05-05 Richard Biener <rguenther@suse.de>
17741 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
17743 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
17745 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
17746 of this flag from insn conditions due to removal from r247495.
17748 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
17750 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
17752 (arm_early_store_addr_dep_ptr): Likewise.
17753 * config/arm/aarch-common-protos.h
17754 (arm_early_load_addr_dep_ptr): Add prototype.
17755 (arm_early_store_addr_dep_ptr): Likewise.
17756 * config/arm/cortex-a53.md: Add new bypasses.
17758 2017-05-05 Jakub Jelinek <jakub@redhat.com>
17760 * tree.c (next_type_uid): Change type to unsigned.
17761 (type_hash_canon): Decrement back next_type_uid if
17762 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
17763 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
17766 2017-05-04 Martin Sebor <msebor@redhat.com>
17768 * builtins.c: Fix a trivial typo in a comment.
17770 PR middle-end/79234
17771 * builtins.c (check_sizes): Adjust to handle reading past the end.
17772 Avoid printing excessive upper bound of ranges. Use %E to print
17773 tree nodes instead of converting them to %wu.
17774 (expand_builtin_memchr): New function.
17775 (compute_dest_size): Rename...
17776 (compute_objsize): ...to this.
17777 (expand_builtin_memcpy): Adjust.
17778 (expand_builtin_mempcpy): Adjust.
17779 (expand_builtin_strcat): Adjust.
17780 (expand_builtin_strcpy): Adjust.
17781 (check_strncat_sizes): Adjust.
17782 (expand_builtin_strncat): Adjust.
17783 (expand_builtin_strncpy): Adjust and simplify.
17784 (expand_builtin_memset): Adjust.
17785 (expand_builtin_bzero): Adjust.
17786 (expand_builtin_memcmp): Adjust.
17787 (expand_builtin): Handle memcmp.
17788 (maybe_emit_chk_warning): Check strncat just once.
17790 2017-05-04 Martin Sebor <msebor@redhat.com>
17792 PR preprocessor/79214
17793 PR middle-end/79222
17794 PR middle-end/79223
17795 * builtins.c (check_sizes): Add inlining context and issue
17796 warnings even when -Wno-system-headers is set.
17797 (check_strncat_sizes): Same.
17798 (expand_builtin_strncat): Same.
17799 (expand_builtin_memmove): New function.
17800 (expand_builtin_stpncpy): Same.
17801 (expand_builtin): Handle memmove and stpncpy.
17803 2017-05-04 Bin Cheng <bin.cheng@arm.com>
17805 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
17806 which is not used any more.
17808 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
17810 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
17812 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
17814 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
17815 (cortexa53_tunings): Likewise.
17816 (cortexa57_tunings): Likewise.
17817 (cortexa72_tunings): Likewise.
17818 (cortexa73_tunings): Likewise.
17820 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
17822 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
17823 Set loop alignment to 8.
17825 2017-05-04 Martin Sebor <msebor@redhat.com>
17827 PR translation/80280
17828 * builtins.c (expand_builtin_object_size): Add missing quoting to
17829 %D and like directives.
17830 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
17831 (hsa_type_for_tree_type): Same.
17832 (verify_function_arguments): Same.
17833 * symtab.c (symbol_table::change_decl_assembler_name): Same.
17834 * varasm.c (get_section): Same.
17837 2017-05-04 Martin Sebor <msebor@redhat.com>
17839 PR translation/80280
17840 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
17842 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
17844 * config/aarch64/aarch64.c (generic_addrcost_table):
17845 Change HI/TI mode setting.
17847 2017-05-04 Martin Jambor <mjambor@suse.cz>
17849 PR tree-optimization/80622
17850 * tree-sra.c (comes_initialized_p): New function.
17851 (build_accesses_from_assign): Only set write lazily when
17852 comes_initialized_p is false.
17853 (analyze_access_subtree): Use comes_initialized_p.
17854 (propagate_subaccesses_across_link): Assert !comes_initialized_p
17855 instead of testing for PARM_DECL.
17857 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17859 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
17860 constraint on operand 0 to allow more general addressing modes.
17861 Adjust output template.
17862 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
17864 * config/aarch64/aarch64-protos.h
17865 (aarch64_address_valid_for_prefetch_p): Declare prototype.
17866 * config/aarch64/constraints.md (Dp): New address constraint.
17867 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
17870 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
17872 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
17873 update use of estimate_ipcp_clone_size_and_time.
17874 (estimate_local_effects): Update use of
17875 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
17876 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
17877 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
17878 Return nonspecialized time.
17880 2017-05-04 Richard Biener <rguenther@suse.de>
17882 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
17883 for the last VUSE which def dominates the PHI. Directly call
17885 (get_continuation_for_phi_1): Remove.
17887 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
17889 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
17890 to explain the use of truncating division. Cap the number of
17891 iterations to the maximum given by nb_iterations_upper_bound,
17894 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
17896 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
17897 * configure: Regenerate.
17898 * config.in: Regenerate.
17899 * config/i386/driver-mingw32.c: new file.
17900 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
17901 * config.host: Link driver-mingw32.o on MinGW host.
17902 * doc/install.texi: Document new --enable-mingw-wildcard configure
17905 2017-05-04 Marek Polacek <polacek@redhat.com>
17907 PR tree-optimization/80612
17908 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
17910 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
17911 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
17913 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
17914 (movt splitter): Likewise.
17915 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
17916 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
17917 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
17918 block for Thumb-1 with MOVT.
17919 (thumb2_legitimate_address_p): Move code block ...
17920 (can_avoid_literal_pool_for_label_p): ... into this new function.
17921 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
17923 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
17924 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
17925 "M-profile targets with the MOVT instruction".
17927 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
17929 * config/arm/arm-builtins.c (arm_init_builtins): Rename
17930 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
17931 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
17933 2017-05-04 Martin Liska <mliska@suse.cz>
17935 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
17936 variable cond_code.
17938 2017-05-04 Richard Biener <rguenther@suse.de>
17940 * tree.c (array_at_struct_end_p): Handle arrays at struct
17941 end with flexarrays more conservatively. Refactor and treat
17942 arrays of arrays or aggregates more strict. Fix
17943 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
17944 * tree.c (array_at_struct_end_p): Adjust prototype.
17945 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
17946 * gimple-fold.c (get_range_strlen): Likewise.
17947 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
17949 2017-05-04 Richard Biener <rguenther@suse.de>
17951 PR tree-optimization/31130
17952 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
17954 (supports_overflow_infinity): Likewise.
17955 (is_negative_overflow_infinity): Likewise.
17956 (is_positive_overflow_infinity): Likewise.
17957 (is_overflow_infinity): Likewise.
17958 (stmt_overflow_infinity): Likewise.
17959 (overflow_infinity_range_p): Likewise.
17960 (usable_range_p): Remove as always returning true.
17961 (make_overflow_infinity): Remove.
17962 (negative_overflow_infinity): Likewise.
17963 (positive_overflow_infinity): Likewise.
17964 (avoid_overflow_infinity): Likewise.
17965 (set_value_range): Adjust accordingly.
17966 (set_value_range_to_nonnegative): Likewise, remove now unused
17967 overflow_infinity arg.
17968 (vrp_operand_equal_p): Adjust.
17969 (update_value_range): Likewise.
17970 (range_int_cst_singleton_p): Likewise.
17971 (operand_less_p): Likewise.
17972 (compare_values_warnv): Likewise.
17973 (extract_range_for_var_from_comparison_expr): Likewise.
17974 (vrp_int_const_binop): Likewise.
17975 (zero_nonzero_bits_from_vr): Likewise.
17976 (extract_range_from_multiplicative_op_1): Likewise.
17977 (extract_range_from_binary_expr_1): Likewise.
17978 (extract_range_from_unary_expr): Likewise.
17979 (extract_range_from_comparison): Likewise.
17980 (extract_range_basic): Likewise.
17981 (adjust_range_with_scev): Likewise.
17982 (compare_ranges): Likewise.
17983 (compare_range_with_value): Likewise.
17984 (dump_value_range): Likewise.
17985 (test_for_singularity): Likewise, remove strict_overflow_p parameter
17987 (simplify_cond_using_ranges): Adjust.
17989 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
17991 * brig-builtins.def: Added a builtin for class_f64.
17992 * builtin-types.def: Added a builtin type needed by class_f64.
17994 2017-05-03 Jason Merrill <jason@redhat.com>
17996 * timevar.def: Add TV_CONSTEXPR.
17998 2017-05-03 David Malcolm <dmalcolm@redhat.com>
18000 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
18002 2017-05-03 Martin Jambor <mjambor@suse.cz>
18004 * ipa-prop.c (ipa_update_after_lto_read): Removed.
18005 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
18006 * ipa-cp.c (ipcp_propagate_stage): Do not call
18007 ipa_update_after_lto_read.
18008 * ipa-inline.c (ipa_inline): Likewise.
18010 2017-05-03 Martin Jambor <mjambor@suse.cz>
18012 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
18013 tag. Added a default constructor and a destructor.
18014 (ipa_edge_args_sum_t): New class;
18015 (ipa_edge_args_sum): Declare.
18016 (ipa_edge_args_vector): Remove declaration.
18017 (IPA_EDGE_REF): Use ipa_edge_args_sum.
18018 (ipa_free_edge_args_substructures): Remove declaration.
18019 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
18020 (ipa_edge_args_info_available_for_edge_p): Likewise.
18021 * ipa-prop.c (ipa_edge_args_vector): Removed.
18022 (edge_removal_hook_holder): Likewise.
18023 (edge_duplication_hook_holder): Likewise.
18024 (ipa_edge_args_sum): New variable.
18025 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
18026 ipa_edge_args_vector.
18027 (ipa_free_edge_args_substructures): Likewise.
18028 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
18029 ipa_edge_args_vector.
18030 (ipa_edge_removal_hook): Turned into method
18031 ipa_edge_args_sum_t::remove.
18032 (ipa_edge_duplication_hook): Turned into method
18033 ipa_edge_args_sum_t::duplicate.
18034 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
18035 registering edge hooks.
18036 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
18037 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
18038 ipa_edge_args_sum instead of ipa_edge_args_vector.
18039 * ipa-profile.c (ipa_profile): Likewise.
18041 2017-05-03 Martin Jambor <mjambor@suse.cz>
18043 * symbol-summary.h (function_summary): New method exists.
18044 (function_summary::symtab_removal): Deallocate through release.
18045 (call_summary): New class.
18046 (gt_ggc_mx): New overload.
18047 (gt_pch_nx): Likewise.
18048 (gt_pch_nx): Likewise.
18050 2017-05-03 Jeff Law <law@redhat.com>
18052 PR tree-optimization/78496
18053 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
18054 from simplify_cond_using_ranges. Split off code to walk
18055 backwards through casts into ...
18056 (simplify_cond_using_ranges_2): New function.
18057 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
18058 (execute_vrp): After identifying jump threads, call
18059 simplify_cond_using_ranges_2.
18061 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
18064 * ipa-inline.h (inline_summary): Add ctor.
18065 (create_ggc): Do not use ggc_cleared_alloc.
18067 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
18068 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18070 * gcc.c (handle_braces): Support escaping in switch matching
18072 * doc/invoke.texi (Spec Files): Document it.
18073 Remove superfluous @code markup in items.
18075 2017-05-03 David Malcolm <dmalcolm@redhat.com>
18077 * diagnostic-show-locus.c (struct column_range): New struct.
18078 (get_affected_columns): New function.
18079 (get_printed_columns): New function.
18080 (struct correction): New struct.
18081 (correction::ensure_capacity): New function.
18082 (correction::ensure_terminated): New function.
18083 (struct line_corrections): New struct.
18084 (line_corrections::~line_corrections): New dtor.
18085 (line_corrections::add_hint): New function.
18086 (layout::print_trailing_fixits): Reimplement in terms of the new
18088 (selftest::test_overlapped_fixit_printing): New function.
18089 (selftest::diagnostic_show_locus_c_tests): Call it.
18091 2017-05-03 Nathan Sidwell <nathan@acm.org>
18093 Canonicalize canonical type hashing
18094 * tree.h (type_hash_canon_hash): Declare.
18095 * tree.c (type_hash_list, attribute_hash_list): Move into
18096 type_hash_canon_hash.
18097 (build_type_attribute_qual_variant): Break out hash code calc into
18098 type_hash_canon_hash.
18099 (type_hash_canon_hash): New. Generic type hash computation.
18100 (build_range_type_1, build_array_type_1, build_function_type,
18101 build_method_type_directly, build_offset_type, build_complex_type,
18102 make_vector_type): Call it.
18104 2017-05-03 Richard Biener <rguenther@suse.de>
18106 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
18107 When all DRs have unknown misaligned do not always peel
18108 when there is a store but apply the same costing model as if
18109 there were only loads.
18111 2017-05-03 Richard Biener <rguenther@suse.de>
18114 PR tree-optimization/80492
18115 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
18116 compare_base_decls returning dont-know properly.
18118 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
18120 * config/arm/iterators.md (CCSI): New mode iterator.
18121 (arch): New mode attribute.
18122 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
18123 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
18124 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
18125 code iterator for success result mode.
18126 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
18127 the corresponding new insn generators.
18129 2017-05-03 Bin Cheng <bin.cheng@arm.com>
18132 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18133 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
18135 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
18137 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
18138 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
18139 (DDR_A): Wrap DDR argument in brackets.
18140 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
18141 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
18142 (DDR_REVERSED_P): Likewise.
18144 2017-05-03 Jakub Jelinek <jakub@redhat.com>
18146 PR tree-optimization/79472
18147 * tree-switch-conversion.c (struct switch_conv_info): Add
18148 contiguous_range and default_case_nonstandard fields.
18149 (collect_switch_conv_info): Compute contiguous_range and
18150 default_case_nonstandard fields, don't clear final_bb if
18151 contiguous_range and only the default case doesn't have the required
18153 (check_all_empty_except_final): Set default_case_nonstandard instead
18154 of failing if contiguous_range and the default case doesn't have empty
18156 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
18157 and only the default case doesn't have the required constants. Skip
18159 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
18160 if default_case_nonstandard.
18161 (build_constructors): Build constant 1 just once. Assert that default
18162 values aren't inserted in between cases if contiguous_range. Skip
18164 (build_arrays): Skip virtual phis.
18165 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
18166 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
18167 Handle virtual phis.
18168 (gen_inbound_check): Handle default_case_nonstandard case.
18169 (process_switch): Adjust check_final_bb caller. Call
18170 gather_default_values with the first non-default case instead of
18171 default case if default_case_nonstandard.
18173 2017-05-02 Nathan Sidwell <nathan@acm.org>
18175 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
18176 check. Fix formatting.
18178 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
18180 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
18181 errors when comparing specialized and unspecialized times.
18183 2017-05-02 David Malcolm <dmalcolm@redhat.com>
18185 * diagnostic-show-locus.c
18186 (layout::should_print_annotation_line_p): Make private.
18187 (layout::print_annotation_line): Make private.
18188 (layout::annotation_line_showed_range_p): Make private.
18189 (layout::show_ruler): Make private.
18190 (layout::print_source_line): Make private. Pass in line and
18191 line_width, rather than calling location_get_source_line. Drop
18193 (layout::print_leading_fixits): New method.
18194 (layout::print_any_fixits): Rename to...
18195 (layout::print_trailing_fixits): ...this, and make private.
18196 Don't print newline fixits.
18197 (diagnostic_show_locus): Move logic for printing one row into...
18198 (layout::print_line): ...this new function. Move the
18199 location_get_source_line call and error-handling from
18200 print_source_line to here. Call print_leading_fixits, and rename
18201 print_any_fixits to print_trailing_fixits.
18202 (selftest::test_fixit_insert_containing_newline): Update now that
18203 newlines are partially supported.
18204 (selftest::test_fixit_insert_containing_newline_2): New test.
18205 (selftest::test_fixit_replace_containing_newline): Update comments.
18206 (selftest::diagnostic_show_locus_c_tests): Call the new test.
18207 * edit-context.c (class added_line): New class.
18208 (class edited_line): Describe newline handling in comment.
18209 (edited_line::actually_edited_p): New method.
18210 (edited_line::print_content): Delete redundant decl.
18211 (edited_line::m_predecessors): New field.
18212 (edited_file::print_content): Call edited_line::print_content.
18213 (edited_file::print_diff): Update to support newlines.
18214 (edited_file::print_diff_hunk): Likewise.
18215 (edited_file::print_run_of_changed_lines): New function.
18216 (edited_file::print_diff_line): Convert to...
18217 (print_diff_line): ...this.
18218 (edited_file::get_effective_line_count): New function.
18219 (edited_line::edited_line): Initialize new field m_predecessors.
18220 (edited_line::~edited_line): Clean up m_predecessors.
18221 (edited_line::apply_fixit): Handle newlines.
18222 (edited_line::get_effective_line_count): New function.
18223 (edited_line::print_content): New function.
18224 (edited_line::print_diff_lines): New function.
18225 (selftest::test_applying_fixits_insert_containing_newline): New
18227 (selftest::test_applying_fixits_replace_containing_newline): New
18229 (selftest::insert_line): New function.
18230 (selftest::test_applying_fixits_multiple_lines): Add example of
18232 (selftest::edit_context_c_tests): Call the new tests.
18234 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18236 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
18237 parameter cand. Update dump information.
18238 (get_computation_cost): Update uses.
18240 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18242 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
18243 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
18244 (get_computation_at, rewrite_use_address): Update use of
18245 get_computation_aff.
18247 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18249 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
18250 (get_computation): Delete.
18251 (get_computation_cost): Implement like get_computation_cost_at.
18252 Use get_computation_at.
18253 (get_computation_cost_at): Delete.
18254 (rewrite_use_nonlinear_expr): Use get_computation_at.
18255 (rewrite_use_compare, remove_unused_ivs): Ditto.
18257 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18259 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
18261 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18263 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
18264 (ivopts_global_cost_for_size): Rename parameter and update uses.
18265 (iv_ca_recount_cost): Update uses.
18266 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
18267 candidates seperately in n_invs and n_cands.
18268 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
18270 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18272 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
18273 (find_inv_vars_cb): New.
18274 (find_depends): Renamed to ...
18275 (find_inv_vars): ... this.
18276 (add_candidate_1, force_var_cost): Call find_inv_vars.
18277 (split_address_cost, determine_group_iv_cost_cond): Ditto.
18279 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18281 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
18282 inv_vars. Add inv_exprs.
18283 (struct iv_cand): Rename depends_on to inv_vars.
18284 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
18285 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
18286 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
18287 (dump_cand): Dump inv_vars.
18288 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
18289 (record_invariant, find_depends, add_candidate_1): Ditto.
18290 (set_group_iv_cost, force_var_cost): Ditto.
18291 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
18292 (get_computation_cost_at, get_computation_cost): Ditto.
18293 (determine_group_iv_cost_generic): Ditto.
18294 (determine_group_iv_cost_address): Ditto.
18295 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
18296 (determine_group_iv_costs): Ditto.
18297 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
18298 (iv_ca_set_remove_invariants): Renamed to ...
18299 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
18300 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
18301 (iv_ca_set_add_invariants): Renamed to ...
18302 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
18303 (iv_ca_set_cp): Use iv_ca_set_add_invs.
18304 (iv_ca_has_deps): Support inv_vars and inv_exprs.
18305 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
18306 (create_new_ivs): Remove useless dump.
18308 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18310 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
18312 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
18313 (iv_ca_set_no_cp, create_new_iv): Ditto.
18315 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18317 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
18319 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
18321 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
18322 function tree_check2.
18324 2017-05-02 Martin Liska <mliska@suse.cz>
18326 * doc/gcov.texi: Add missing preposition.
18327 * gcov.c (function_info::function_info): Properly fill up
18328 all member variables.
18330 2017-05-02 Tamar Christina <tamar.christina@arm.com>
18332 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
18334 2017-05-02 Tamar Christina <tamar.christina@arm.com>
18336 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
18338 2017-05-02 Martin Liska <mliska@suse.cz>
18341 * lto-streamer-in.c (lto_read_tree_1): Remove
18342 LTO_STREAMER_DEBUG.
18343 * lto-streamer.c (struct tree_hash_entry): Likewise.
18344 (struct tree_entry_hasher): Likewise.
18345 (tree_entry_hasher::hash): Likewise.
18346 (tree_entry_hasher::equal): Likewise.
18347 (lto_streamer_init): Likewise.
18348 (lto_orig_address_map): Likewise.
18349 (lto_orig_address_get): Likewise.
18350 (lto_orig_address_remove): Likewise.
18351 * lto-streamer.h: Likewise.
18352 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
18353 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
18355 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
18357 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
18358 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
18359 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
18360 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
18361 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
18362 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
18363 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
18364 (mm_maskz_sub_ss): New intrinsics.
18365 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
18366 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
18367 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
18368 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
18369 (__builtin_ia32_subss_mask_round): New builtins.
18370 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
18371 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
18372 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
18374 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
18375 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
18377 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
18380 2017-05-02 Martin Jambor <mjambor@suse.cz>
18382 PR tree-optimization/78687
18383 * tree-sra.c (access): New field parent.
18384 (process_subtree_disqualification): New function.
18385 (disqualify_candidate): Call it.
18386 (build_accesses_from_assign): Reset write flag if creating an
18388 (build_access_subtree): Fill in parent field and also prpagate
18389 down grp_write flag.
18390 (create_artificial_child_access): New parameter set_grp_write, set
18391 grp_write to its value.
18392 (propagate_subaccesses_across_link): Also propagate grp_write flag
18394 (propagate_all_subaccesses): Push the closest parent back to work
18395 queue if add_access_to_work_queue returned true.
18397 2017-05-02 Richard Biener <rguenther@suse.de>
18399 * common.opt (fstrict-overflow): Alias negative to fwrapv.
18400 * doc/invoke.texi (fstrict-overflow): Remove all traces of
18401 -fstrict-overflow documentation.
18402 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
18403 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
18404 flag_strict_overflow.
18405 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
18406 * lto-opts.c (lto_write_options): Do not stream it.
18407 * lto-wrapper.c (merge_and_complain): Do not handle it.
18408 * opts.c (default_options_table): Do not set -fstrict-overflow.
18409 (finish_options): Likewise do not clear it when sanitizing.
18410 * simplify-rtx.c (simplify_const_relational_operation): Do not
18411 test flag_strict_overflow.
18413 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
18415 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
18416 using enabled attribute.
18417 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
18418 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
18419 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
18420 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
18421 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
18422 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
18423 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
18424 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
18425 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
18426 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
18428 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
18430 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
18432 2017-05-02 Richard Biener <rguenther@suse.de>
18434 PR tree-optimization/80591
18436 2017-04-10 Richard Biener <rguenther@suse.de>
18438 * tree-ssa-structalias.c (find_func_aliases): Properly handle
18441 2017-05-02 Richard Biener <rguenther@suse.de>
18443 PR tree-optimization/80549
18444 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
18445 (cleanup_tree_cfg_noloop): Create forwarders to known loop
18446 headers if they do not have a preheader.
18448 2017-05-02 Martin Liska <mliska@suse.cz>
18451 * common.opt: Fix typo.
18452 * doc/invoke.texi: Likewise.
18454 2017-05-01 Jan Beulich <jbeulich@suse.com>
18456 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
18457 swapping, add (x,x,m,x,n) alternative.
18459 2017-05-01 Nathan Sidwell <nathan@acm.org>
18461 * calls.c (combine_pending_stack_adjustment_and_call): Remove
18462 unnecessary unadjusted_alignment check.
18464 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
18467 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
18469 * gimplify.c (gimplify_cilk_detach): New function.
18470 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
18471 * tree-core.h: Document EXPR_CILK_SPAWN.
18472 * tree.h (EXPR_CILK_SPAWN): Define.
18474 2017-05-01 David Malcolm <dmalcolm@redhat.com>
18476 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
18477 to use new fixit_hint representation, using the "replace" logic.
18478 (get_line_span_for_fixit_hint): Likewise.
18479 (layout::print_any_fixits): Likewise.
18480 (selftest::test_one_liner_many_fixits): Rename to...
18481 (selftest::test_one_liner_many_fixits_1): ...this, and update
18482 comment and expected output to reflect that the multiple fix-it
18483 hints are now consolidated into one insertion.
18484 (selftest::test_one_liner_many_fixits_2): New test.
18485 (selftest::test_diagnostic_show_locus_one_liner): Update for
18487 (selftest::test_fixit_consolidation): Update for fix-it API
18489 * diagnostic.c (print_parseable_fixits): Likewise.
18490 * edit-context.c (edited_line::m_line_events): Convert from
18491 auto_vec <line_event *> to auto_vec <line_event>.
18492 (class line_event): Convert from abstract base class to a concrete
18493 class, taking over the role of replace_event.
18494 (class insert_event): Delete.
18495 (class replace_event): Rename to class line_event. Convert to
18497 (edit_context::add_fixits): Reimplement.
18498 (edit_context::apply_insert): Delete.
18499 (edit_context::apply_replace): Rename to...
18500 (edit_context::apply_fixit): ...this. Convert to half-open range.
18501 (edited_file::apply_insert): Delete.
18502 (edited_file::apply_replace): Rename to...
18503 (edited_file::apply_fixit): ...this.
18504 (edited_line::~edited_line): Drop deletion of events.
18505 (edited_line::apply_insert): Delete.
18506 (edited_line::apply_replace): Rename to...
18507 (edited_line::apply_fixit): ...this. Convert to half-open range.
18508 Update for change to type of m_line_events.
18509 * edit-context.h (edit_context::apply_insert): Delete.
18510 (edit_context::apply_replace): Rename to...
18511 (edit_context::apply_fixit): ...this.
18513 2017-05-01 Martin Sebor <msebor@redhat.com>
18515 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
18518 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
18521 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
18522 __get_cpuid_max returns 0.
18523 (__get_cpuid_count): Ditto.
18525 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
18527 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
18528 replacement expression is another instance of one of its arguments.
18530 2017-05-01 Jakub Jelinek <jakub@redhat.com>
18533 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
18534 check for stack push/pop autoinc.
18535 * config/i386/i386.c (ix86_agi_dependent): Return false
18536 if the only reason why modified_in_p returned true is that
18537 addr is SP based and set_insn is a push or pop.
18539 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
18541 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
18544 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
18547 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
18548 (account_size_time): Use two predicates - exec_pred and
18550 (evaluate_conditions_for_known_args): Compute both clause and
18552 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
18553 (inline_summary_t::duplicate): Update.
18554 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
18556 (compute_inline_parameters): Likewise.
18557 (estimate_edge_size_and_time): Update caluclation of time.
18558 (estimate_node_size_and_time): Compute both time and nonspecialized
18560 (estimate_ipcp_clone_size_and_time): Update.
18561 (inline_merge_summary): Update.
18562 (do_estimate_edge_time): Update.
18563 (do_estimate_edge_size): Update.
18564 (do_estimate_edge_hints): Update.
18565 (inline_read_section, inline_write_summary): Stream both new predicates.
18566 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
18568 (compute_inlined_call_time): Cleanup.
18569 (big_speedup_p): Update.
18570 (edge_badness): Update.
18571 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
18572 (size_time_entry): Replace predicate by exec_predicate and
18573 nonconst_predicate.
18574 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
18575 (estimate_edge_time): Return also nonspec_time.
18576 (reset_edge_growth_cache): Update.
18578 2017-04-29 Jakub Jelinek <jakub@redhat.com>
18580 PR rtl-optimization/80491
18581 * ifcvt.c (noce_process_if_block): When looking for x setter
18582 with missing else_bb, don't check only the insn right before
18583 cond_earliest, but look for the last insn that x is modified in
18584 within the same bb.
18586 PR rtl-optimization/80491
18587 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
18589 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
18591 PR tree-optimization/80487
18592 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
18594 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18596 PR tree-optimization/79697
18597 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
18598 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
18599 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
18601 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
18602 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
18604 2017-04-28 Martin Sebor <msebor@redhat.com>
18606 PR tree-optimization/80523
18607 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
18608 (init_target_to_host_charmap, target_to_host, target_strtol10): New
18610 (maybe_warn, format_directive, parse_directive): Use new functions.
18611 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
18613 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
18615 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
18617 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
18619 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
18620 target_header_dir): Set correctly.
18621 * configure: Regenerated.
18622 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
18623 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
18624 instead of SYSTEM_HEADER_DIR.
18626 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
18628 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
18629 (estimate_local_effects): Likewise.
18630 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
18631 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
18632 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
18633 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
18634 do_estimate_edge_time, estimate_edge_time): Likewise.
18635 * ipa-inline-analysis.c (estimate_node_size_and_time,
18636 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
18637 (estimate_time_after_inlining): Remove.
18639 2017-04-28 Martin Liska <mliska@suse.cz>
18641 * doc/gcov.texi: Enhance documentation of gcov.
18643 2017-04-28 Martin Liska <mliska@suse.cz>
18645 * doc/gcov.texi: Sort options in alphabetic order.
18646 * doc/gcov-dump.texi: Likewise.
18647 * doc/gcov-tool.texi: Likewise.
18648 * gcov.c (print_usage): Likewise.
18649 * gcov-dump.c (print_usage): Likewise.
18650 * gcov-tool.c (print_merge_usage_message): Likewise.
18651 (print_rewrite_usage_message): Likewise.
18652 (print_overlap_usage_message): Likewise.
18654 2017-04-28 Martin Liska <mliska@suse.cz>
18656 PR gcov-profile/53915
18657 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
18659 2017-04-28 Martin Liska <mliska@suse.cz>
18661 PR gcov-profile/79891
18662 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
18663 is marked by compiler as living on a line.
18664 (get_cycles_count): Remove usage of the union.
18665 (output_intermediate_file): Likewise.
18666 (find_source): Fix GNU coding style.
18667 (accumulate_line_counts): Remove old non-all block mode.
18668 (output_lines): Remove usage of the union.
18669 * profile.c (output_location): Include all BBs, even if
18670 belonging to a same line (and file) as a previous BB.
18672 2017-04-28 Martin Liska <mliska@suse.cz>
18674 * gcov.c (process_args): Handle new argument 'w'.
18675 (read_graph_file): Assign ID to BBs.
18676 (output_branch_count): Display BB # if verbose flag is set.
18677 (output_lines): Likewise for arcs.
18678 (print_usage): Add '--verbose' option help.
18679 * doc/gcov.texi: Document --verbose (-w) option.
18681 2017-04-28 Martin Liska <mliska@suse.cz>
18683 * gcov.c (struct block_location_info): New struct.
18684 (process_file): Fill up the new structure.
18685 (read_graph_file): Replace usage of encoding by the newly added
18687 (add_line_counts): Likewise.
18688 (accumulate_line_counts): Remove usage of the union.
18689 (function_info::function_info): New function.
18690 (function_info::~function_info): Likewise.
18691 (process_file): Call delete instead of release_function.
18692 (release_function): Release the function.
18693 (release_structures): Call delete instead of release_function.
18694 (solve_flow_graph): Replace usage of num_blocks.
18695 (find_exception_blocks): Likewise.
18696 (output_lines): Fix GNU coding style.
18698 2017-04-28 Martin Liska <mliska@suse.cz>
18701 * coverage.c (coverage_remove_note_file): New function.
18702 * coverage.h: Declare the function.
18703 * toplev.c (finalize): Clean if an error has been seen.
18705 2017-04-28 Martin Liska <mliska@suse.cz>
18707 PR gcov-profile/80031
18708 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
18709 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
18710 * gcov.c (read_graph_file): Read just number of blocks.
18711 * profile.c (branch_prob): Do not stream 0 flags per a basic
18714 2017-04-28 Martin Liska <mliska@suse.cz>
18716 * gcov-dump.c (tag_*): Add new argument to declarations.
18717 (dump_gcov_file): Likewise.
18718 (tag_blocks): Add and use new argument depth.
18719 (tag_arcs): Likewise.
18720 (tag_lines): Likewise.
18721 (tag_counters): Likewise.
18722 (tag_summary): Likewise.
18723 (dump_working_sets): Use depth to do a proper indentation.
18725 2017-04-28 Jakub Jelinek <jakub@redhat.com>
18728 * cgraph.h (symtab_node::debug_symtab): No longer inline.
18729 * symtab.c (symtab_node::debug_symtab): Move definition here.
18731 2017-04-28 Richard Biener <rguenther@suse.de>
18733 * lto-streamer.h (LTO_major_version): Bump to 7.
18735 2017-04-28 Richard Biener <rguenther@suse.de>
18737 * tree-vrp.c (assert_info): New struct.
18738 (add_assert_info): New helper.
18739 (register_edge_assert_for_2): Refactor to add asserts to a vector
18741 (register_edge_assert_for_1): Likewise.
18742 (register_edge_assert_for): Likewise.
18743 (finish_register_edge_assert_for): New helper actually registering
18744 asserts where live on edge.
18745 (find_conditional_asserts): Adjust.
18746 (find_switch_asserts): Likewise.
18747 (evrp_dom_walker::try_find_new_range): Generalize.
18748 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
18750 2017-04-27 Marek Polacek <polacek@redhat.com>
18753 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
18754 arg10 and arg11 to itype.
18756 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
18758 * doc/extend.texi (Object Size Checking): Improve grammar.
18760 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
18763 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
18764 that the logic for permitting reciprocal estimates matches that
18767 2017-04-27 Jakub Jelinek <jakub@redhat.com>
18770 * tree.c (type_cache_hasher::equal): Only compare
18771 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
18772 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
18773 non-aggregate element types.
18774 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
18775 about the flag on ARRAY_TYPEs in the comment, formatting fix.
18777 2017-04-27 Richard Biener <rguenther@suse.de>
18779 PR middle-end/80533
18780 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
18781 stripping ARRAY_REFs from MEM_EXPR make sure we're not
18782 keeping a reference to a trailing array.
18784 2017-04-27 Richard Biener <rguenther@suse.de>
18786 PR middle-end/80539
18787 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
18788 being in loop-closed SSA form conservatively.
18789 (chrec_fold_multiply_poly_poly): Likewise.
18791 2017-04-27 Tamar Christina <tamar.christina@arm.com>
18793 PR middle-end/79665
18794 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
18795 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
18797 2017-04-27 Jakub Jelinek <jakub@redhat.com>
18800 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
18801 (aarch64_function_arg_alignment): Return unsigned int again, but still
18802 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
18803 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
18804 Don't emit -Wpsabi note.
18805 (aarch64_function_arg_boundary): Likewise.
18806 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
18809 2017-04-26 Nathan Sidwell <nathan@acm.org>
18811 * tree.h (crc32_unsigned_n): Declare.
18812 (crc32_unsigned, crc32_unsigned): Make inline.
18813 * tree.c (crc32_unsigned_bits): Replace with ...
18814 (crc32_unsigned_n): ... this.
18815 (crc32_unsigned, crc32_byte): Remove.
18816 (crc32_string): Remove unnecessary braces.
18818 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
18820 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
18821 * ipa-inline-analysis.c (MAX_TIME): Remove.
18822 (account_size_time): Use sreal for time.
18823 (dump_inline_summary): Update.
18824 (estimate_function_body_sizes): Update.
18825 (estimate_edge_size_and_time): Update.
18826 (estimate_calls_size_and_time): Update.
18827 (estimate_node_size_and_time): Update.
18828 (inline_merge_summary): Update.
18829 (inline_update_overall_summary): Update.
18830 (estimate_time_after_inlining): Update.
18831 (inline_read_section): Update.
18832 (inline_write_summary): Update.
18833 * ipa-inline.c (compute_uninlined_call_time): Update.
18834 (compute_inlined_call_time): Update.
18835 (recursive_inlining): Update.
18836 (inline_small_functions): Update.
18837 (dump_overall_stats): Update.
18838 * ipa-inline.h: Include sreal.h.
18839 (size_time_entry): Turn time to sreal.
18840 (inline_summary): Turn self_time nad time to sreal.
18842 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
18844 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
18846 (sreal::stream_out, sreal::stream_in): New.
18847 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
18849 2017-04-25 Jakub Jelinek <jakub@redhat.com>
18851 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
18854 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
18857 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
18859 (dimode_scalar_chain::compute_convert_gain): Ditto.
18860 (dimode_scalar_chain::make_vector_copies): Ditto.
18861 (dimode_scalar_chain::convert_reg): Ditto.
18862 (dimode_scalar_chain::convert_insn): Ditto.
18863 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
18864 (VI248_AVX512BW_1): New mode iterator.
18865 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
18866 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
18869 2017-04-25 Martin Sebor <msebor@redhat.com>
18871 PR tree-optimization/80497
18872 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
18873 constants are representable in HOST_WIDE_INT.
18874 (parse_directive): Ditto.
18876 2017-04-25 Martin Sebor <msebor@redhat.com>
18879 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
18880 (new_zero_array): Adjust signature.
18881 (dom_info::dom_init): Used unsigned rather that size_t.
18882 (dom_info::dom_info): Same.
18884 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18885 Jakub Jelinek <jakub@redhat.com>
18888 * config/arm/arm.c: Include gimple.h.
18889 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
18890 returns negative, increment ncrn only if it returned positive.
18891 (arm_needs_doubleword_align): Return int instead of bool,
18892 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
18893 members, but if there is any such non-FIELD_DECL
18894 > PARM_BOUNDARY aligned decl, return -1 instead of false.
18895 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
18896 returns negative, increment nregs only if it returned positive.
18897 (arm_setup_incoming_varargs): Likewise.
18898 (arm_function_arg_boundary): Emit -Wpsabi note if
18899 arm_needs_doubleword_align returns negative, return
18900 DOUBLEWORD_ALIGNMENT only if it returned positive.
18902 2017-04-25 Marek Polacek <polacek@redhat.com>
18905 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
18906 first argument to type.
18908 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
18911 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
18912 type checks to test for compatibility instead of equality.
18914 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18915 Jakub Jelinek <jakub@redhat.com>
18918 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
18920 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
18921 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
18922 the alignment computation, but return their maximum in warn_alignment.
18923 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
18924 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
18926 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
18927 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
18930 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18932 * config/arc/simdext.md (dmpyh): Fix typo.
18934 2017-04-25 Richard Biener <rguenther@suse.de>
18936 PR tree-optimization/80492
18937 * alias.c (compare_base_decls): Handle registers with asm
18938 specification conservatively.
18939 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
18940 compare_base_decls returning dont-know properly.
18942 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18944 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
18945 (legitimate_offset_address_p): New function.
18946 (arc_legitimate_address_p): Use above function.
18948 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18950 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
18952 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18954 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
18955 ACCH registers whenever they are available.
18957 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18959 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
18960 double regs fix when not used.
18962 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18964 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
18966 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
18967 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
18969 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18971 * config/arc/arc.c (arc_output_addsi): Check for h-register class
18972 when emitting short ADD instructions.
18974 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18976 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
18978 (cmpsi_cc_c_insn): Likewise.
18979 (cbranchsi4_scratch): Compute proper instruction length using
18980 compact_hreg_operand.
18981 * config/arc/predicates.md (compact_hreg_operand): New predicate.
18983 2017-04-25 Richard Biener <rguenther@suse.de>
18985 PR middle-end/80509
18986 * passes.c (pass_manager::pass_manager): Initialize
18987 m_name_to_pass_map.
18989 2017-04-25 Richard Biener <rguenther@suse.de>
18991 PR tree-optimization/79201
18992 * tree-ssa-sink.c (statement_sink_location): Handle calls.
18994 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18997 * config/s390/vector.md: Split MEM->GPR vector moves for
18998 non-s_operand addresses.
19000 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19003 * config/s390/predicates.md (reload_const_wide_int_operand): New
19005 * config/s390/s390.md ("movti"): Remove d/P alternative.
19006 ("movti_bigconst"): New pattern definition.
19008 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
19011 * s390-protos.h (s390_expand_cs_hqi): Removed.
19012 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
19013 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
19014 modes as well as CCZ1mode and CCZmode.
19015 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
19016 signature of s390_emit_compare_and_swap.
19017 (s390_expand_cs_hqi): Likewise, make static.
19018 (s390_expand_cs_tdsi): Generate an explicit compare before trying
19019 compare-and-swap, in some cases.
19020 (s390_expand_cs): Wrapper function.
19021 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
19023 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
19024 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
19025 patterns for small and large integers. Forbid symref memory operands.
19026 Move expander to s390.c. Require cc register.
19027 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
19028 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
19029 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
19030 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
19031 symref memory operands. Remove CC mode and call s390_match_ccmode
19033 ("atomic_exchange<mode>"): Allow and implement all integer modes.
19035 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
19037 * config/s390/s390.md (define_peephole2): New peephole to help
19038 combining the load-and-test pattern with volatile memory.
19040 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
19042 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
19043 with CCZmode for TARGET_Z196.
19045 2017-04-25 Jakub Jelinek <jakub@redhat.com>
19047 PR rtl-optimization/80501
19048 * combine.c (make_compound_operation_int): Set subreg_code to SET
19049 even for AND with mask of the sign bit of mode.
19051 PR rtl-optimization/80500
19052 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
19053 sum's initial value.
19055 2017-04-25 Julian Brown <julian@codesourcery.com>
19056 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19058 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
19060 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
19062 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
19064 2017-04-25 Julian Brown <julian@codesourcery.com>
19065 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19067 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
19068 (thunderx2t99_sha): New Reservation.
19070 2017-04-25 Julian Brown <julian@codesourcery.com>
19071 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19073 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
19074 type for 1-element load.
19076 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
19078 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
19080 2017-04-24 Martin Jambor <mjambor@suse.cz>
19082 PR tree-optimization/80293
19083 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
19084 char arrays not totally scalarizable if it is false.
19085 (analyze_all_variable_accesses): Pass correct value in the new
19086 parameter. Add a statistics counter.
19088 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
19090 PR middle-end/79931
19091 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
19093 2017-04-24 Richard Biener <rguenther@suse.de>
19095 PR tree-optimization/80494
19096 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
19097 out for complex types.
19099 2017-04-24 Richard Biener <rguenther@suse.de>
19101 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
19102 * tree-ssa-sccvn.c (print_scc): Print SCC size.
19103 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
19104 (DFS): Adjust and never fail.
19105 (sccvn_dom_walker::fail): Remove.
19106 (sccvn_dom_walker::before_dom_children): Adjust.
19107 (run_scc_vn): Likewise and never fail.
19108 * tree-ssa-pre.c (pass_pre::execute): Adjust.
19109 (pass_fre::execute): Likewise.
19111 2017-04-24 Richard Biener <rguenther@suse.de>
19113 PR tree-optimization/79725
19114 * tree-ssa-sink.c (statement_sink_location): Return whether
19115 failure reason was zero uses. Move that check later.
19116 (sink_code_in_bb): Deal with zero uses by removing the stmt
19119 2017-04-24 Richard Biener <rguenther@suse.de>
19122 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
19123 pointer-based references.
19125 2017-04-24 Richard Biener <rguenther@suse.de>
19128 * pass_manager.h (pass_manager::operator new): Remove.
19129 (pass_manager::operator delete): Likewise.
19130 * passes.c (pass_manager::operator new): Remove.
19131 (pass_manager::operator delete): Likewise.
19132 (pass_manager::pass_manager): Zero individual pass members.
19134 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
19137 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
19138 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
19139 Check "XEXP (src, 1)" operand here.
19140 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
19141 Check "XEXP (src, 1)" operand here.
19142 (dimode_scalar_chain::make_vector_copies): Detect count register
19143 of a shift instruction. Zero extend count register from QImode
19144 to DImode to satisfy vector shift pattern count operand predicate.
19145 Substitute vector shift count operand with a DImode copy.
19146 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
19149 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
19151 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
19152 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
19153 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
19154 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
19155 (UNSPEC_NOREX_MEM): Remove definition.
19157 2017-04-21 Richard Biener <rguenther@suse.de>
19159 PR tree-optimization/79547
19160 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
19161 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
19162 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
19163 without any constraints.
19165 2017-04-21 Richard Biener <rguenther@suse.de>
19167 PR tree-optimization/78847
19168 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
19170 2017-04-21 Richard Biener <rguenther@suse.de>
19172 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
19173 (build_distinct_type_copy): Likewise.
19174 (build_variant_type_copy): Likewise.
19175 * tree.c (build_qualified_type): Pass down mem-stat info.
19176 (build_distinct_type_copy): Likewise.
19177 (build_variant_type_copy): Likewise.
19179 2017-04-21 Richard Biener <rguenther@suse.de>
19181 PR tree-optimization/80237
19182 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
19184 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
19185 for a simplified result.
19187 2016-04-21 Richard Biener <rguenther@suse.de>
19189 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
19190 sth as strict as a simple_iv but a chrec without symbols and an
19191 operand defined in the loop we are peeling (and not some subloop).
19192 (propagate_constants_for_unrolling): Propagate all constants.
19194 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
19197 * config/i386/i386.c (print_reg): Remove assert for disalowed
19198 regno values, call output_operand_lossage instead.
19200 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
19203 * config/i386/constraints.md (Yc): New register constraint.
19204 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
19205 Use Yc constraint for alternative 2 of operand 0. Remove
19206 preferred_for_speed attribute.
19208 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
19210 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
19211 lastprivate clauses in SIMT case.
19213 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
19215 * doc/invoke.texi (-Wextra-semi): Document new warning option.
19217 2017-04-20 Richard Biener <rguenther@suse.de>
19219 PR tree-optimization/57796
19220 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
19221 as N scalar stores.
19222 (vect_model_load_cost): Cost gathers as N scalar loads.
19224 2017-04-20 Richard Biener <rguenther@suse.de>
19226 * ggc-page.c (ggc_allocated_p): Rename to ...
19227 (safe_lookup_page_table_entry): ... this and return the lookup
19229 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
19231 2017-04-20 Richard Biener <rguenther@suse.de>
19233 PR tree-optimization/80453
19234 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
19235 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
19236 from the conditions.
19237 (vn_phi_eq): Pass them down.
19238 (vn_phi_lookup): Record them.
19239 (vn_phi_insert): Likewise.
19241 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
19243 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
19244 uninitialized variable warning to avoid buffer overrun.
19246 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
19249 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
19250 is suppressed for '{ 0 }' in C.
19252 2017-04-20 Jakub Jelinek <jakub@redhat.com>
19254 * BASE-VER: Set to 8.0.0.
19256 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
19258 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
19259 priority .init_array and .fini_array section with SECTION_NOTYPE
19262 2017-04-20 Jakub Jelinek <jakub@redhat.com>
19264 PR middle-end/80423
19265 * tree.h (build_array_type): Add typeless_storage default argument.
19266 * tree.c (type_cache_hasher::equal): Also compare
19267 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
19268 (build_array_type): Add typeless_storage argument, set
19269 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
19271 (build_nonshared_array_type): Adjust build_array_type_1 caller.
19272 (build_array_type): Likewise. Add typeless_storage argument.
19274 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
19275 Jakub Jelinek <jakub@redhat.com>
19277 PR tree-optimization/80426
19278 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
19279 operation on symbolic operands, also compute the overflow for the
19280 invariant part when the operation degenerates into a negation.
19282 2017-04-19 Jakub Jelinek <jakub@redhat.com>
19285 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
19286 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
19289 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
19291 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
19294 * config/avr/avr.c (tree.h): Include it.
19295 (cgraph.h): Include it.
19296 (avr_encode_section_info): Don't warn for uninitialized progmem
19297 variable if it's just an alias.
19299 2017-04-19 Richard Biener <rguenther@suse.de>
19302 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
19303 when needed by AutoPGO.
19305 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
19308 * doc/lto.texi: Remove an extra 'that'.
19310 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
19312 PR rtl-optimization/80429
19313 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
19314 are only used in debug insns.
19316 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
19317 Vladimir Makarov <vmakarov@redhat.com>
19319 * config/sparc/predicates.md (input_operand): Add comment. Return
19320 true for any memory operand when LRA is in progress.
19321 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
19323 2017-04-18 Jeff Law <law@redhat.com>
19326 * mips.md ({return,simple_return}_internal): Do not overwrite
19329 2017-04-18 Jakub Jelinek <jakub@redhat.com>
19331 PR tree-optimization/80443
19332 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
19333 instead of adding 1, subtract -1 and similarly instead of subtracting
19336 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
19338 PR rtl-optimization/80357
19339 * haifa-sched.c (tmp_bitmap): New variable.
19340 (model_recompute): Handle duplicate use records.
19341 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
19342 (free_global_sched_pressure_data): Free it.
19344 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
19347 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
19348 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
19349 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
19350 instead of SYSTEM_HEADER_DIR.
19352 2017-04-18 Jeff Law <law@redhat.com>
19354 PR middle-end/80422
19355 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
19356 predecessors after walking up the insn chain.
19358 2017-04-18 Jakub Jelinek <jakub@redhat.com>
19361 * dwarf2out.c (modified_type_die): Try harder not to emit internal
19362 sizetype type into debug info.
19364 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
19367 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
19368 unneeded test for TARGET_UPPER_REGS_SF.
19369 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
19371 2017-04-18 Jakub Jelinek <jakub@redhat.com>
19374 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
19375 instead of gsi_after_labels.
19377 2017-04-18 Jeff Law <law@redhat.com>
19379 * regcprop.c (maybe_mode_change): Avoid creating copies of the
19383 2017-04-13 Jeff Law <law@redhat.com>
19384 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
19385 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
19387 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
19390 * config/avr/avr.c (intl.h): Include it.
19391 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
19393 2017-04-18 Martin Liska <mliska@suse.cz>
19395 PR gcov-profile/78783
19396 * gcov-tool.c (gcov_output_files): Validate that destination
19397 file is either removed by the tool or by a user.
19399 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
19400 Guy Benyei <guybe@mellanox.com>
19402 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
19403 block, and do not negate it, the stored id is already negative.
19405 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
19407 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
19409 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
19412 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
19413 masks of options that should be turned off if the VSX vector
19414 options are turned off.
19415 (OTHER_P8_VECTOR_MASKS): Likewise.
19416 (OTHER_VSX_VECTOR_MASKS): Likewise.
19417 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
19418 rs6000_disable_incompatible_switches to validate no type switches
19420 (rs6000_incompatible_switch): New function to disallow turning on
19421 other vector options if -mno-vsx, -mno-power8-vector, or
19422 -mno-power9-vector are specified.
19424 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
19426 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
19428 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
19430 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
19431 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
19432 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
19433 (ARG_POINTER_CFA_OFFSET): Likewise.
19435 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
19437 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
19438 conditions to take advantage of various optimizations.
19440 2017-04-13 Jeff Law <law@redhat.com>
19442 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
19443 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
19444 (zero_extendsidi2_dext): Likewise.
19446 2017-04-13 Jakub Jelinek <jakub@redhat.com>
19449 * fold-const.c (fold_ternary_loc): Revert
19450 use op0 instead of fold_convert_loc (loc, type, arg0) part of
19453 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
19455 PR rtl-optimization/80343
19456 * lra-remat.c (update_scratch_ops): Assign original hard reg to
19457 new scratch pseudo.
19459 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
19462 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
19463 to ubsan_encode_value.
19465 2017-04-13 Jeff Law <law@redhat.com>
19467 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
19468 appearing in DEBUG_INSNs.
19470 2017-04-13 Martin Liska <mliska@suse.cz>
19472 PR gcov-profile/80413
19473 * gcov-io.c (gcov_write_string): Copy to buffer just when
19474 allocated size is greater than zero.
19476 2017-04-13 Jakub Jelinek <jakub@redhat.com>
19479 * dwarf2out.c (decls_for_scope): Ignore declarations of
19480 current_function_decl in BLOCK_NONLOCALIZED_VARS.
19482 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
19485 * ipa-visibility.c (non_local_p): Fix typos.
19486 (localize_node): When localizing symbol in same comdat group,
19487 dissolve the group only when we know external symbols are going
19489 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
19491 2017-04-12 Jakub Jelinek <jakub@redhat.com>
19493 PR tree-optimization/79390
19494 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
19495 order does not result in usable sequence, retry with reversed operand
19501 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
19502 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
19503 op0 instead of fold_convert_loc (loc, type, arg0).
19505 2017-04-12 Jeff Law <law@redhat.com>
19507 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
19508 has a delay slot in the generated code.
19510 * config/cris/cris.md (cris_preferred_reload_class): Return
19511 GENNONACR_REGS rather than GENERAL_REGS.
19513 2017-04-12 Jakub Jelinek <jakub@redhat.com>
19516 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
19517 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
19518 signedness of the result type.
19520 2017-04-12 Richard Biener <rguenther@suse.de>
19521 Jeff Law <law@redhat.com>
19523 PR tree-optimization/80359
19524 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
19525 trim stores to TARGET_MEM_REFs.
19527 2017-04-12 Richard Biener <rguenther@suse.de>
19529 PR tree-optimization/79390
19530 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
19531 threading case even more.
19533 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
19536 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
19537 for quad_address_p for TImode, instead of just not indexed_address.
19539 2017-04-12 Richard Biener <rguenther@suse.de>
19540 Bernd Edlinger <bernd.edlinger@hotmail.de>
19542 PR middle-end/79671
19543 * alias.c (component_uses_parent_alias_set_from): Handle
19544 TYPE_TYPELESS_STORAGE.
19545 (get_alias_set): Likewise.
19546 * tree-core.h (tree_type_common): Add typeless_storage flag.
19547 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
19548 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
19549 for types containing members with TYPE_TYPELESS_STORAGE.
19550 (place_field): Likewise.
19551 (layout_type): Likewise for ARRAY_TYPE.
19552 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
19553 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
19554 TYPE_TYPELESS_STORAGE.
19555 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
19557 2017-04-12 Jakub Jelinek <jakub@redhat.com>
19560 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
19561 first argument to type.
19563 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19567 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
19568 CONST0_RTX (mode) rather than const0_rtx where appropriate.
19569 (rs6000_expand_binop_builtin): Likewise.
19570 (rs6000_expand_ternop_builtin): Likewise; also add missing
19571 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
19572 vshasigma built-ins.
19573 * doc/extend.texi: Document that vec_xxpermdi's third argument
19574 must be a constant.
19576 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
19578 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
19579 Use shift_const cost parameter when calculating gain of STV shifts.
19581 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
19583 PR rtl-optimization/70478
19584 * lra-constraints.c (process_alt_operands): Check memory for
19585 disfavoring memory insn operand.
19587 2017-04-11 Jakub Jelinek <jakub@redhat.com>
19589 PR middle-end/80100
19590 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
19591 left shift in unsigned HOST_WIDE_INT type.
19593 PR rtl-optimization/80385
19594 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
19595 (not (neg X)) into (plus X -1) for complex or non-integral modes.
19598 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
19599 if they have any depend clauses.
19601 2017-04-11 Martin Liska <mliska@suse.cz>
19604 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
19605 * ipa-split.c (split_function): Create a local comdat symbol
19606 if caller is in a comdat group.
19608 2017-04-11 Martin Liska <mliska@suse.cz>
19611 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
19614 2017-04-11 Martin Sebor <msebor@redhat.com>
19616 PR middle-end/80364
19617 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
19618 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
19620 (directive::set_width, directive::set_precision, format_character):
19622 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
19625 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
19628 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
19629 conflict, set target->arch_name instead of target->cpu_name.
19631 2017-04-11 Richard Biener <rguenther@suse.de>
19633 PR tree-optimization/80374
19634 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
19635 build_zero_cst, remove fold_convertible_p check again.
19637 2017-04-11 Martin Liska <mliska@suse.cz>
19640 * ubsan.c (instrument_object_size): Do not instrument register
19643 2017-04-11 Jakub Jelinek <jakub@redhat.com>
19646 * config/i386/i386-builtin-types.def
19647 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
19648 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
19649 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
19650 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
19651 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
19652 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
19653 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
19654 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
19655 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
19656 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
19657 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
19658 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
19659 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
19660 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
19661 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
19662 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
19663 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
19664 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
19665 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
19666 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
19667 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
19668 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
19669 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
19670 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
19671 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
19672 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
19673 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
19674 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
19675 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
19676 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
19677 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
19678 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
19679 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
19680 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
19681 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
19682 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
19683 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
19684 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
19685 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
19686 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
19687 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
19688 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
19689 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
19690 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
19691 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
19692 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
19694 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
19695 flag to second_arg_count, handle 4 argument function type _COUNT
19696 aliases, handle second_arg_count on second argument rather than last.
19698 2017-04-10 Jeff Law <law@redhat.com>
19700 PR tree-optimization/80374
19701 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
19702 record anything if we can not convert integer_zero_node to the
19705 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
19708 * config/rs6000/rs6000.c (rs6000_option_override_internal):
19709 Enhance special handling given to the TARGET_P9_MINMAX option in
19710 relation to certain other options.
19712 2017-04-10 Bin Cheng <bin.cheng@arm.com>
19714 PR tree-optimization/80153
19715 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
19716 remove POINTER_PLUS_EXPR's base part directly, rather than through
19719 2017-04-10 Richard Biener <rguenther@suse.de>
19720 Bin Cheng <bin.cheng@arm.com>
19722 PR tree-optimization/80153
19723 * tree-affine.c (aff_combination_to_tree): Get base pointer from
19724 the first element of pointer type aff_tree. Build result expr in
19726 (add_elt_to_tree): Convert to type unconditionally. Remove other
19727 fold_convert calls.
19728 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
19729 (rewrite_use_nonlinear_expr): Check invariant using iv information.
19731 2017-04-10 Richard Biener <rguenther@suse.de>
19733 * tree-ssa-structalias.c (find_func_aliases): Properly handle
19736 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
19738 PR rtl-optimization/70478
19739 * lra-constraints.c (curr_small_class_check): New.
19740 (update_and_check_small_class_inputs): New.
19741 (process_alt_operands): Update curr_small_class_check. Disfavor
19742 alternative insn memory operands. Check available regs for small
19745 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
19748 * config/mips/mips.opt (-mvirt): Update description.
19749 * doc/invoke.texi (-mvirt): Likewise.
19751 2017-04-10 Richard Biener <rguenther@suse.de>
19753 PR middle-end/80362
19754 * fold-const.c (fold_binary_loc): Look at unstripped ops when
19755 looking for NEGATE_EXPR in -A / -B to A / B folding.
19757 2017-04-10 Martin Liska <mliska@suse.cz>
19759 PR gcov-profile/80224
19760 * gcov.c (print_usage): Fix usage string.
19761 (get_gcov_intermediate_filename): Remove.
19762 (output_gcov_file): Use both for normal and intermediate format.
19763 (generate_results): Do not initialize special file for
19764 intermediate format.
19766 2017-04-10 Richard Biener <rguenther@suse.de>
19768 PR tree-optimization/80304
19769 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
19772 2017-04-10 Nathan Sidwell <nathan@acm.org>
19775 * config/rs6000/rs6000.c (rs6000_vector_type): New.
19776 (rs6000_init_builtins): Use it.
19778 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19780 * config/arm/arm.md (<mrc>): Add mode to SET source.
19781 (<mrrc>): Likewise.
19783 2017-04-10 Richard Biener <rguenther@suse.de>
19785 PR middle-end/80344
19786 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
19788 2017-04-10 Jakub Jelinek <jakub@redhat.com>
19791 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
19792 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
19793 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
19794 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
19795 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
19796 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
19797 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
19798 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
19799 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
19800 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
19801 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
19802 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
19803 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
19804 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
19805 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
19806 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
19807 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
19808 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
19809 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
19810 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
19811 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
19812 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
19813 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
19815 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
19817 PR rtl-optimization/70478
19818 * lra-constraints.c: Reverse the last patch.
19820 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
19822 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
19823 Add comment for WCHAR_T.
19825 2017-04-08 Martin Liska <mliska@suse.cz>
19828 2017-04-07 Martin Liska <mliska@suse.cz>
19831 * ipa-split.c (split_function): Add function part to a same comdat
19834 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19837 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
19839 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
19841 * rs6000/rs6000.c (vec_load_pendulum): Rename...
19842 (vec_pairing): ...to this.
19843 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
19844 (rs6000_sched_init): Adjust for name change.
19845 (struct rs6000_sched_context): Likewise.
19846 (rs6000_init_sched_context): Likewise.
19847 (rs6000_set_sched_context): Likewise.
19849 2017-04-07 Jakub Jelinek <jakub@redhat.com>
19855 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
19857 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
19858 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
19859 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
19861 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
19863 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
19865 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
19867 PR rtl-optimization/70703
19868 * ira-color.c (update_conflict_hard_regno_costs): Use
19869 int64_t instead of HOST_WIDE_INT.
19871 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
19873 PR rtl-optimization/70478
19874 * lra-constraints.c (process_alt_operands): Disfavor alternative
19875 insn memory operands.
19877 2017-04-07 Jeff Law <law@redhat.com>
19879 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
19880 CALL and NOTE_INSN_CALL_ARG_LOCATION.
19882 2017-04-07 Martin Liska <mliska@suse.cz>
19885 * config/aarch64/aarch64.c (aarch64_process_target_attr):
19886 Show error message instead of an ICE.
19888 2017-04-07 Martin Liska <mliska@suse.cz>
19891 * ipa-split.c (split_function): Add function part to a same comdat
19894 2017-04-07 Richard Biener <rguenther@suse.de>
19896 PR middle-end/80341
19897 * tree.c (get_unwidened): Also handle ! for_type case for
19899 * convert.c (do_narrow): Split out from ...
19900 (convert_to_integer_1): ... here. Do not pass final truncation
19901 type to get_unwidened for TRUNC_DIV_EXPR.
19903 2017-04-07 Richard Biener <rguenther@suse.de>
19905 * tree-affine.c (wide_int_ext_for_comb): Take type rather
19907 (aff_combination_const): Adjust.
19908 (aff_combination_scale): Likewise.
19909 (aff_combination_add_elt): Likewise.
19910 (aff_combination_add_cst): Likewise.
19911 (aff_combination_convert): Likewise.
19912 (add_elt_to_tree): Likewise. Remove unused argument.
19913 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
19915 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
19917 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
19919 * config/arm/arm.c (arm_default_short_enums): Use
19920 ARM_DEFAULT_SHORT_ENUMS.
19921 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
19923 2017-04-06 Jakub Jelinek <jakub@redhat.com>
19926 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
19927 members with redundant out-of-class redeclaration.
19929 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
19932 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
19933 * config/i386/i386.md (*zero_extendsidi2):
19934 Add (?*x,*x) and (?*v,*v) alternatives.
19936 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
19939 * config/i386/i386.c (ix86_expand_builtin)
19940 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
19941 mode from insn data. Convert operands to insn operand mode.
19942 Copy operands that don't satisfy insn predicate to a register.
19944 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
19946 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
19949 2017-04-06 Richard Biener <rguenther@suse.de>
19951 PR tree-optimization/80334
19952 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
19953 preserve alignment of accesses.
19955 2017-04-06 Richard Biener <rguenther@suse.de>
19957 PR tree-optimization/80262
19958 * tree-sra.c (build_ref_for_offset): Preserve address-space
19960 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
19961 Drop useless address-space information on MEM_REF offsets.
19963 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
19965 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
19967 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
19969 PR rtl-optimization/70703
19970 * ira-color.c (update_conflict_hard_regno_costs): Use
19971 HOST_WIDE_INT instead of long.
19973 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
19976 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
19977 not defined for x86_64 target. Add -mmmx target option when __SSE2__
19979 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
19980 for x86_64 target. Handle -m3dnowa option.
19982 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
19984 PR rtl-optimization/70703
19985 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
19986 (update_conflict_hard_regno_costs): Use long instead of unsigned
19987 arithmetic for cost calculation.
19989 2017-04-05 Jakub Jelinek <jakub@redhat.com>
19990 Bernd Edlinger <bernd.edlinger@hotmail.de>
19993 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
19996 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
19999 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
20000 ptr_mode with Pmode throughout.
20001 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
20002 into probe_stack_range and use DImode.
20004 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
20007 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
20008 call_eh_return is true.
20010 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20012 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
20013 Initialize last_match_fntype_index.
20015 2017-04-05 Jakub Jelinek <jakub@redhat.com>
20018 * tree-nvr.c: Include internal-fn.h.
20019 (pass_return_slot::execute): Ignore internal calls without
20022 2017-04-04 Jakub Jelinek <jakub@redhat.com>
20023 Richard Biener <rguenther@suse.de>
20026 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
20027 captures used multiple times, except for the last use.
20028 * generic-match-head.c: Include gimplify.h.
20030 2017-04-04 Jakub Jelinek <jakub@redhat.com>
20032 PR tree-optimization/79390
20033 * target.h (struct noce_if_info): Declare.
20034 * targhooks.h (default_noce_conversion_profitable_p): Declare.
20035 * target.def (noce_conversion_profitable_p): New target hook.
20036 * ifcvt.h (struct noce_if_info): New type, moved from ...
20037 * ifcvt.c (struct noce_if_info): ... here.
20038 (noce_conversion_profitable_p): Renamed to ...
20039 (default_noce_conversion_profitable_p): ... this. No longer
20041 (noce_try_store_flag_constants, noce_try_addcc,
20042 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
20043 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
20044 instead of noce_conversion_profitable_p.
20045 * config/i386/i386.c: Include ifcvt.h.
20046 (ix86_option_override_internal): Don't override
20047 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
20048 (ix86_noce_conversion_profitable_p): New function.
20049 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
20050 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
20051 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
20052 * doc/tm.texi: Regenerated.
20054 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20056 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
20059 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
20062 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
20063 instructions for small multiply cores.
20065 2017-04-04 Jeff Law <law@redhat.com>
20067 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
20069 (mips_expand_vec_perm_const): Initialize elements in orig_perm
20070 that are not set by the loop over the elements.
20072 2017-04-04 Jakub Jelinek <jakub@redhat.com>
20075 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
20076 int mode, convert_modes it to mode as unsigned, otherwise use
20077 lowpart_subreg to mode rather than SImode.
20078 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
20079 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
20080 Use DImode instead of SImode for the shift count operand.
20081 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
20084 2017-04-04 Richard Biener <rguenther@suse.de>
20086 PR middle-end/80281
20087 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
20088 arithmetic done for the negate or the plus. Simplify.
20089 (A - (-B) -> A + B): Likewise.
20090 * fold-const.c (split_tree): Make sure to not negate pointers.
20092 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
20094 PR rtl-optimization/60818
20095 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
20096 a compare of comparisons with the thing compared if this results
20097 in a different machine mode.
20099 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
20101 * alias.c (base_alias_check): Fix typo in comment.
20102 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
20103 * cgraphunit.c (symbol_table::compile): Likewise.
20104 * collect2.c (maybe_run_lto_and_relink): Likewise.
20105 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
20106 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
20107 * config/avr/avr.c (avr_map_op_t): Likewise.
20108 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
20109 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
20110 * config/epiphany/epiphany.md (movcc): Likewise.
20111 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
20112 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
20114 * config/mips/mips.c (mips_save_restore_reg): Likewise.
20115 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
20116 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
20117 * config/sh/sh.c (sh_rtx_costs): Likewise.
20118 * fold-const.c (fold_truth_andor): Likewise.
20119 * genautomata.c (collapse_flag): Likewise.
20120 * gengtype.h (struct type::u::s): Likewise.
20121 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
20122 * input.c (FORMAT_AMOUNT): Likewise.
20123 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
20124 (known_aggs_to_agg_replacement_list): Likewise.
20125 * ipa-inline-analysis.c: Likewise.
20126 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
20127 * ipa-polymorphic-call.c
20128 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
20129 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
20130 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
20132 * modulo-sched.c (apply_reg_moves): Likewise.
20133 * omp-expand.c (build_omp_regions_1): Likewise.
20134 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
20135 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
20136 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
20137 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
20138 * value-prof.c: Likewise.
20139 * var-tracking.c (val_reset): Likewise.
20141 2017-04-03 Richard Biener <rguenther@suse.de>
20143 PR tree-optimization/80275
20144 * fold-const.c (split_address_to_core_and_offset): Handle
20147 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
20149 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
20150 descriptors is at least equal to that of functions.
20152 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
20154 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
20156 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
20159 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
20160 (mov<IMOD4:mode>): New expander.
20161 (*mov<IMOD4:mode>_internal): New insn and split pattern.
20163 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
20165 PR rtl-optimization/79405
20166 * fwprop.c (propagations_left): New variable.
20167 (forward_propagate_into): Decrement it.
20168 (fwprop_init): Initialize it.
20169 (fw_prop): If the variable has reached zero, stop propagating.
20170 (fwprop_addr): Ditto.
20172 2017-03-31 Jakub Jelinek <jakub@redhat.com>
20175 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
20176 a FUNCTION_DECL, pass it as decl instead of origin to
20179 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
20181 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
20184 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
20187 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
20188 TARGET_VSX_SMALL_INTEGER.
20190 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20192 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
20193 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
20195 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
20197 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
20198 extraction from odd-numbered MSA register.
20200 2017-03-31 Jakub Jelinek <jakub@redhat.com>
20202 PR middle-end/80173
20203 * expmed.c (store_bit_field_1): Don't attempt to create
20204 a word subreg out of hard registers wider than word if they
20205 have HARD_REGNO_NREGS of 1 for their mode.
20207 PR middle-end/80163
20208 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
20209 conversions to integer types wider than word and pointer.
20212 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
20213 (rtx_equal_for_cselib_p): Pass 0 to it.
20214 * cselib.c (cselib_hasher::equal): Likewise.
20215 (rtx_equal_for_cselib_1): Add depth argument. If depth
20216 is 128, don't look up VALUE locs and punt. Increment
20217 depth in recursive calls when walking VALUE locs.
20219 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
20221 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
20222 (make_gcov_file_name): Use the canonical path name for generating
20224 (read_line): Fix handling of files with ascii null bytes.
20226 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
20228 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
20229 to initialise a vector register instead
20230 of using a const_int.
20232 2017-03-30 Jakub Jelinek <jakub@redhat.com>
20234 PR translation/80189
20235 * gimplify.c (omp_default_clause): Use %qs instead of %s in
20236 diagnostic messages.
20238 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
20241 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
20242 (dfp_diex_<mode>): Update mode of operand 1.
20243 * doc/extend.texi (dxex, dxexq): Document change to return type.
20244 (diex, diexq): Document change to argument type.
20246 2017-03-30 Martin Jambor <mjambor@suse.cz>
20249 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
20250 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
20251 it reflects the signature changes performed at the callee side.
20252 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
20253 to cgraph_build_function_type_skip_args.
20254 (build_function_decl_skip_args): Adjust call to the above function.
20256 2017-03-30 Jakub Jelinek <jakub@redhat.com>
20259 * config/i386/sse.md
20260 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
20261 register as dest whenever it is a MEM not rtx_equal_p to the
20262 corresponding dup operand, and when forcing into reg move the
20263 reg into the memory afterwards.
20264 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
20265 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
20266 for the force_reg mode.
20267 (avx512vl_vextractf128<mode>): Use register as dest either
20268 always when a MEM, or when it is a MEM not rtx_equal_p to the
20269 corresponding dup operand, or even not when it is a CONST_VECTOR
20270 depending on the mode and lo vs. hi.
20271 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
20273 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
20274 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
20275 Likewise. Require that operands[2] is even.
20276 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
20277 Remove extraneous parens. Require that operands[2] is a multiple
20279 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
20280 operands[0] is a MEM if <mask_applied>, the predicates/constraints
20281 disallow memory then.
20283 2017-03-30 Richard Biener <rguenther@suse.de>
20285 PR tree-optimization/77498
20286 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
20287 to non-constants over backedges.
20289 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
20291 PR rtl-optimization/80233
20292 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
20293 as last_combined_insn. Do not test for BARRIER_P separately.
20295 2017-03-29 Andreas Schwab <schwab@suse.de>
20298 * calls.c (prepare_call_address): Convert funexp to Pmode before
20299 copying to temp reg.
20301 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20303 PR tree-optimization/80158
20304 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
20305 Handle possible future case of more than one alternate
20307 (replace_rhs_if_not_dup): Likewise.
20308 (replace_one_candidate): Likewise.
20310 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
20312 PR rtl-optimization/80193
20313 * ira.c (ira): Do not check allocation for LRA.
20315 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
20317 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
20318 (nvptx_output_simt_exit): Declare.
20319 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
20320 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
20321 (init_softstack_frame): Move initialization of crtl->is_leaf to...
20322 (nvptx_declare_function_name): ...here. Emit declaration of local
20323 memory space buffer for omp_simt_enter insn.
20324 (nvptx_output_unisimt_switch): New.
20325 (nvptx_output_softstack_switch): New.
20326 (nvptx_output_simt_enter): New.
20327 (nvptx_output_simt_exit): New.
20328 * config/nvptx/nvptx.h (struct machine_function): New fields
20329 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
20330 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
20331 (UNSPECV_SIMT_EXIT): Ditto.
20332 (omp_simt_enter_insn): New insn.
20333 (omp_simt_enter): New expansion.
20334 (omp_simt_exit): New insn.
20335 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
20337 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
20338 (expand_GOMP_SIMT_ENTER_ALLOC): New.
20339 (expand_GOMP_SIMT_EXIT): New.
20340 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
20341 (GOMP_SIMT_ENTER_ALLOC): Ditto.
20342 (GOMP_SIMT_EXIT): Ditto.
20343 * target-insns.def (omp_simt_enter): New insn.
20344 (omp_simt_exit): Ditto.
20345 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
20347 (lower_rec_simd_input_clauses): Implement SIMT privatization.
20348 (lower_rec_input_clauses): Likewise.
20349 (lower_lastprivate_clauses): Handle SIMT privatization.
20351 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
20352 (ompdevlow_adjust_simt_enter): New.
20353 (find_simtpriv_var_op): New.
20354 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
20355 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
20357 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
20358 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
20359 (copy_decl_for_dup_finish): Ditto.
20361 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
20363 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
20366 * config/i386/i386.c (ix86_option_override_internal): Always
20367 allow -mpreferred-stack-boundary=3 for 64-bit targets.
20369 2017-03-28 Bin Cheng <bin.cheng@arm.com>
20371 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
20373 2017-03-28 Bin Cheng <bin.cheng@arm.com>
20375 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
20376 mark new edge's irreducible flag accordign to it.
20377 (vect_do_peeling): Check loop preheader edge's irreducible flag
20378 and pass it to function slpeel_add_loop_guard.
20380 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
20382 PR tree-optimization/80218
20383 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
20384 Update block frequencies and counts.
20386 2017-03-28 Richard Biener <rguenther@suse.de>
20388 PR tree-optimization/78644
20389 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
20390 of a simplification result we may not use it at all.
20392 2017-03-28 Richard Biener <rguenther@suse.de>
20395 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
20396 without arguments, generate default definition of a SSA name.
20398 2017-03-28 Richard Biener <rguenther@suse.de>
20400 PR middle-end/80222
20401 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
20402 TYPE_REF_CAN_ALIAS_ALL references.
20403 * fold-const.c (fold_indirect_ref_1): Likewise.
20405 2017-03-28 Martin Liska <mliska@suse.cz>
20408 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
20409 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
20411 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
20412 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
20414 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
20415 (EXTRA_SPECS): Define.
20416 (SUBTARGET_EXTRA_SPECS): Likewise.
20417 (SUBTARGET_CPP_SPEC): Likewise.
20418 * config/arc/elf.h (EXTRA_SPECS): Renamed to
20419 SUBTARGET_EXTRA_SPECS.
20420 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
20422 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
20424 * config/arc/simdext.md (vst64_insn): Update pattern.
20425 (vld32wh_insn): Likewise.
20426 (vld32wl_insn): Likewise.
20427 (vld64_insn): Likewise.
20428 (vld32_insn): Likewise.
20430 2017-03-28 Marek Polacek <polacek@redhat.com>
20433 * fold-const.c (fold_comparison): Use protected_set_expr_location
20434 instead of SET_EXPR_LOCATION.
20436 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
20438 * tree.c (add_expr): Avoid name lookup warning.
20440 2017-03-27 Jeff Law <law@redhat.com>
20442 PR tree-optimization/80216
20443 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
20444 function name. Limit recursion depth.
20445 (record_temporary_equivalences): Corresponding changes.
20447 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
20449 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
20452 2017-03-27 Jakub Jelinek <jakub@redhat.com>
20455 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
20457 * cfgcleanup.c (reg_note_cfa_p): New array.
20458 (insns_have_identical_cfa_notes): New function.
20459 (old_insns_match_p): Don't cross-jump in between /f
20460 and non-/f instructions. If both i1 and i2 are frame related,
20461 verify all CFA notes, their order and content.
20463 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
20466 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
20467 HImode and SImode with zero extend to DImode to one insn.
20468 (bswap<mode>2_extenddi): Likewise.
20469 (bswapsi2_extenddi): Likewise.
20470 (bswaphi2_extendsi): Likewise.
20471 (bswaphi2): Combine bswap HImode and SImode into one insn.
20472 Separate memory insns from swapping register.
20473 (bswapsi2): Likewise.
20474 (bswap<mode>2): Likewise.
20475 (bswaphi2_internal): Delete, no longer used.
20476 (bswapsi2_internal): Likewise.
20477 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
20478 store, and gpr<-gpr swap insns.
20479 (bswap<mode>2_store): Likewise.
20480 (bswaphi2_reg): Register only splitter, combine with the splitter.
20481 (bswaphi2 splitter): Likewise.
20482 (bswapsi2_reg): Likewise.
20483 (bswapsi2 splitter): Likewise.
20484 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
20485 the insns into load, store, and register/register insns.
20486 (bswapdi2_ldbrx): Likewise.
20487 (bswapdi2_load): Likewise.
20488 (bswapdi2_store): Likewise.
20489 (bswapdi2_reg): Likewise.
20491 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
20493 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
20494 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
20496 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
20499 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
20501 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
20502 special handling for target option conflicts between dform
20503 options (-mpower9-dform, -mpower9-dform-vector,
20504 -mpower9-dform-scalar) and -mno-direct-move.
20506 2017-03-27 Richard Biener <rguenther@suse.de>
20508 PR tree-optimization/80181
20509 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
20511 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
20513 * config/arc/predicates.md (move_double_src_operand): Replace the
20514 call to move_double_src_operand with a call to address_operand.
20516 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
20518 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
20519 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
20520 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
20522 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
20524 * config/arc/predicates.md (long_immediate_loadstore_operand):
20525 Consider scaled addresses cases.
20527 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
20529 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
20530 restored when in interrupt.
20531 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
20532 doesn't have delay slot.
20534 2017-03-27 Richard Biener <rguenther@suse.de>
20537 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
20538 inlined thunk clones.
20540 2017-03-27 Jakub Jelinek <jakub@redhat.com>
20543 * asan.c (instrument_derefs): Copy over last operand from
20544 original COMPONENT_REF to the new COMPONENT_REF with
20545 DECL_BIT_FIELD_REPRESENTATIVE.
20546 * ubsan.c (instrument_object_size): Likewise.
20548 2017-03-27 Richard Biener <rguenther@suse.de>
20550 PR tree-optimization/80170
20551 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
20552 sure DR/SCEV didnt fold in constants we do not see when looking
20553 at the reference base alignment.
20555 2017-03-27 Richard Biener <rguenther@suse.de>
20557 PR middle-end/80171
20558 * gimple-fold.c (fold_ctor_reference): Properly guard against
20559 NULL return value from canonicalize_constructor_val.
20561 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
20564 * config/i386/i386.c (ix86_expand_builtin)
20565 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
20566 flags reg setting and flags reg using instructions.
20567 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
20568 clobbering instructions to zero extend op2.
20570 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
20572 * doc/install.texi (Configuration) <--with-aix-soname>:
20573 Update link to AIX ld.
20575 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
20577 PR rtl-optimization/80160
20578 PR rtl-optimization/80159
20579 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
20580 reg_alternate_class into account.
20582 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
20585 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
20586 to consider in curr_insn_transform.
20588 2017-03-24 Jakub Jelinek <jakub@redhat.com>
20590 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
20591 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
20592 and emit_mode_inner.
20594 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20596 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
20597 argument to the overloaded builtin variants. Use the new flag to
20598 deprecate certain builtin variants.
20599 * config/s390/s390-builtin-types.def: Add new builtin types.
20600 * config/s390/s390-builtins.h: Support new flags field for
20601 overloaded builtins.
20602 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
20603 (s390_macro_to_expand): Enable vector float data type.
20604 (s390_cpu_cpp_builtins_internal): Indicate support of the new
20605 builtins by incrementing the __VEC__ version number.
20606 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
20608 (s390_resolve_overloaded_builtin): Emit error messages depending
20609 on the builtin flags.
20610 * config/s390/s390.c (s390_expand_builtin): Support additional
20611 flags argument. Change error message to match the messages
20612 emitted in s390-c.c.
20613 * config/s390/s390.md: New UNSPEC_* constants.
20614 (op_type): Add new instruction types.
20615 * config/s390/vecintrin.h: Add new builtins and test data class
20617 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
20618 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
20619 (VEC_INEXACT, VEC_NOINEXACT): New constants.
20620 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
20621 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
20622 ("vec_mergel<mode>"): V_HW -> VEC_HW.
20624 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
20625 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
20626 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
20627 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
20629 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
20630 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
20631 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
20632 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
20634 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
20635 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
20636 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
20637 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
20638 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
20639 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
20640 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
20642 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
20643 ("vec_scatter_element<V_HW_4:mode>_DI")
20644 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
20645 ("vec_fpint<mode>", "vflls")
20646 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
20647 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
20648 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
20649 ("*vec_cmphe<mode>_cc"): ... these.
20651 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
20652 mode constant instead of magic value.
20654 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20656 * config/s390/s390.c (s390_expand_vec_compare): Support other
20657 vector floating point modes than just V2DF.
20658 (s390_expand_vcond): Likewise.
20659 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
20660 (s390_cannot_change_mode_class): Prevent mode changes between TF
20661 and V1TF in vector registers.
20662 * config/s390/s390.md (DF, SF): New mode attributes.
20663 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
20664 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
20665 SFmode support for VRs.
20666 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
20668 (VFT, VF_HW): New mode iterators.
20669 (vw, sdx): New mode attributes.
20670 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
20671 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
20672 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
20673 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
20674 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
20675 also the new vector floating point modes. Renaming to ...
20677 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
20678 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
20679 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
20680 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
20681 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
20682 ("vec_unordered<mode>"): ... these.
20684 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
20685 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
20686 ("*vec_extendv2df"): New insn definitions.
20688 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20690 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
20691 ("mulditi3_2", "*muldi3_sign"): New patterns.
20692 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
20693 rename the pattern definition.
20695 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20697 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
20699 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
20701 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20703 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
20704 instruction if possible.
20705 * config/s390/vector.md (vec_halfnumelts): New mode
20707 ("*vec_vllezlf<mode>"): New pattern.
20709 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20711 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
20712 ("popcountv4si2", "popcountv2di2"): Rename to ...
20713 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
20714 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
20716 ("popcount<mode>2_vxe"): New pattern.
20718 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20720 * common/config/s390/s390-common.c (processor_flags_table): Add
20722 * config.gcc: Add arch12.
20723 * config/s390/driver-native.c (s390_host_detect_local_cpu):
20724 Default to arch12 for unknown CPU model numbers.
20725 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
20726 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
20727 PROCESSOR_max sanity check.
20728 * config/s390/s390-opts.h (enum processor_type): Add
20730 * config/s390/s390.c (processor_table): Add arch12.
20731 (s390_expand_builtin): Add check for B_VXE flag.
20732 (s390_issue_rate): Add PROCESSOR_ARCH12.
20733 (s390_get_sched_attrmask): Likewise.
20734 (s390_get_unit_mask): Likewise.
20735 (s390_sched_score): Enable z13 scheduling for arch12.
20736 (s390_sched_reorder): Likewise.
20737 (s390_sched_variable_issue): Likewise.
20738 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
20740 (s390_tune_attr): Use z13 scheduling also for arch12.
20741 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
20742 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
20743 (TARGET_VXE_P): New macros.
20744 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
20745 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
20746 * config/s390/s390.opt: Add arch12 as processor_type.
20748 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20750 * config/s390/s390.md
20751 ("fixuns_truncdddi2", "fixuns_trunctddi2")
20752 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
20753 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
20755 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
20756 Rename expanders to ...
20758 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
20759 ("fixuns_truncdddi2_emu"): ... these.
20761 ("fixuns_trunc<mode>si2_emu"): New expander.
20763 ("*fixuns_truncdfdi2_z13"): Rename to ...
20764 ("*fixuns_truncdfdi2_vx"): ... this.
20766 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20768 * config/s390/2964.md: Remove the single element vector compare
20769 instructions which are no longer used.
20770 * config/s390/s390.c (s390_select_ccmode): Remove handling of
20772 (s390_canonicalize_comparison): Remove handling of DFmode
20774 (s390_expand_vec_compare_scalar): Remove function.
20775 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
20776 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
20778 ("*cmp<mode>_ccs"): Add wfcdb instruction.
20780 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20782 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
20784 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
20785 will anyway by matched by mov<mode>_64dfp.
20787 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20789 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
20790 vlef/vstef. Add missing operand to vleif.
20792 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20794 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
20795 pair for all vector types with 64 bit elements.
20796 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
20797 * config/s390/vector.md (V_HW_64): ... here.
20798 (V_128_NOSINGLE): New mode iterator.
20799 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
20800 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
20801 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
20802 ("*vec_load_pairv2di"): Change to ...
20803 ("*vec_load_pair<mode>"): ... this one.
20805 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20807 * config/s390/constraints.md: Add comments.
20808 (jKK): Reject element sizes > 8 bytes.
20809 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
20811 * config/s390/s390.md: Add the s_operand checks formerly in
20812 s390_split_ok_p to various splitters where they are still
20814 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
20815 for 128 bit vectors. Plus two splitters.
20817 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20819 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
20822 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20825 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
20826 error if the boundary argument is not constant.
20828 2017-03-24 Jakub Jelinek <jakub@redhat.com>
20830 PR rtl-optimization/80112
20831 * loop-doloop.c (doloop_condition_get): Don't check condition
20832 if cmp isn't SET with IF_THEN_ELSE src.
20834 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20836 PR tree-optimization/80158
20837 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
20838 replacing a candidate statement, also replace it for the
20839 candidate's alternate interpretation.
20840 (replace_rhs_if_not_dup): Likewise.
20841 (replace_one_candidate): Likewise.
20843 2017-03-24 Richard Biener <rguenther@suse.de>
20845 PR tree-optimization/80167
20846 * graphite-isl-ast-to-gimple.c
20847 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
20849 (translate_isl_ast_to_gimple::get_rename): Likewise.
20851 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
20853 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
20854 handling of certain combinations of target options, including the
20855 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
20856 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
20858 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20861 * config/arm/arm.md (*load_multiple): Add reload_completed to
20862 matching condition.
20864 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20865 Richard Biener <rguenth@suse.de>
20867 PR tree-optimization/79908
20868 PR tree-optimization/80136
20869 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
20870 been cast away, gimplify_and_add suffices.
20872 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
20874 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
20876 2017-03-23 Richard Biener <rguenther@suse.de>
20878 PR tree-optimization/80032
20879 * gimplify.c (gimple_push_cleanup): Forced unconditional
20880 cleanups still have to go to the conditional_cleanups
20883 2017-03-22 Jakub Jelinek <jakub@redhat.com>
20885 PR tree-optimization/80072
20886 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
20888 (next_operand_entry_id): Change type to unsigned int.
20889 (sort_by_operand_rank): Make sure to return the right return value
20890 even if unsigned fields are bigger than INT_MAX.
20891 (struct oecount): Change cnt and id type to unsigned int.
20892 (oecount_hasher::equal): Formatting fix.
20893 (oecount_cmp): Make sure to return the right return value
20894 even if unsigned fields are bigger than INT_MAX.
20895 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
20898 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
20899 TREE_READONLY on result if writing it more than once.
20902 * doc/invoke.texi (-fsanitize=thread): Document that with
20903 -fnon-call-exceptions atomics are not able to throw
20907 * tsan.c: Include tree-eh.h.
20908 (instrument_builtin_call): Call maybe_clean_eh_stmt or
20909 maybe_clean_or_replace_eh_stmt where needed.
20910 (instrument_memory_accesses): Add cfg_changed argument.
20911 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
20912 if it returned true.
20913 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
20915 PR rtl-optimization/63191
20916 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
20917 wrapper function, moved the whole old content into ...
20918 (ix86_delegitimize_address_1): ... this. New inline function.
20919 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
20920 true as last argument instead of ix86_delegitimize_address.
20922 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
20924 * config/aarch64/aarch64.c (generic_branch_cost): Copy
20925 cortexa57_branch_cost.
20927 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
20929 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
20931 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20934 * doc/md.texi (Constraints): Document wA constraint.
20935 * config/rs6000/constraints.md (wA): New.
20936 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
20937 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
20938 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
20939 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
20941 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
20944 * gimplify.c (is_oacc_declared): New function.
20945 (oacc_default_clause): Use it to set default flags for acc declared
20946 variables inside parallel regions.
20947 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
20948 declared variables.
20949 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
20950 declare attribute to any decl as necessary.
20952 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
20955 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
20956 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
20957 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
20958 (arm_arch_lpae): This.
20959 * config/arm/arm.c (arm_arch7ve): Rename into ...
20960 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
20961 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
20964 2017-03-22 Martin Liska <mliska@suse.cz>
20967 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
20968 error message instead of an ICE.
20970 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20972 * doc/extend.texi (6.11 Additional Floating Types): Revise.
20974 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
20976 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
20978 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
20981 2017-03-21 Martin Sebor <msebor@redhat.com>
20983 * doc/extend.texi: Use "cannot" instead of "can't."
20984 * doc/hostconfig.texi: Same.
20985 * doc/install.texi: Same.
20986 * doc/invoke.texi: Same.
20987 * doc/loop.texi: Same.
20988 * doc/md.texi: Same.
20989 * doc/objc.texi: Same.
20990 * doc/rtl.texi: Same.
20991 * doc/tm.texi: Same.
20992 * doc/tm.texi.in: Same.
20993 * doc/trouble.texi: Same.
20995 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
20998 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
20999 (collect_checksum_attributes): Set it.
21000 (die_checksum_ordered): Use it.
21002 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21004 PR tree-optimization/79908
21005 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
21006 change: For a VA_ARG whose LHS has been cast away, use
21007 force_gimple_operand to construct the side effects.
21009 2017-03-21 David Malcolm <dmalcolm@redhat.com>
21011 PR translation/80001
21012 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
21013 more amenable to translation.
21014 (oacc_loop_auto_partitions): Likewise.
21016 2017-03-21 Marek Polacek <polacek@redhat.com>
21017 Martin Sebor <msebor@redhat.com>
21019 PR tree-optimization/80109
21020 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
21021 on INTEGRAL_TYPE_P.
21023 2017-03-21 Jakub Jelinek <jakub@redhat.com>
21024 Segher Boessenkool <segher@kernel.crashing.org>
21027 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
21028 check reg_used_between_p between insn and one of succ or succ2
21029 depending on if succ is artificial insn not inserted into insn
21032 2017-03-21 Martin Liska <mliska@suse.cz>
21034 PR gcov-profile/80081
21035 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
21036 * doc/gcc.texi: Include gcov-dump stuff.
21037 * doc/gcov-dump.texi: New file.
21039 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
21041 PR rtl-optimization/79150
21042 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
21043 conditional jump, if the jump is the last insn of the loop.
21045 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21046 Richard Biener <rguenth@suse.de>
21048 PR tree-optimization/79908
21049 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
21050 been cast away, use force_gimple_operand to construct the side
21053 2017-03-21 Martin Liska <mliska@suse.cz>
21055 PR libfortran/79956
21056 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
21059 2017-03-21 Brad Spengler <spender@grsecurity.net>
21062 * plugin.c (htab_hash_plugin): New function.
21063 (add_new_plugin): Use it and adjust.
21064 (parse_plugin_arg_opt): Adjust.
21065 (init_one_plugin): Likewise.
21067 2017-03-21 Richard Biener <rguenther@suse.de>
21069 PR tree-optimization/80032
21070 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
21071 if set force the cleanup to happen unconditionally.
21072 (gimplify_target_expr): Push inserted clobbers with force_uncond
21073 to avoid them being removed by control-dependent DCE.
21075 2017-03-21 Richard Biener <rguenther@suse.de>
21077 PR tree-optimization/80122
21078 * tree-inline.c (copy_bb): Do not expans va-arg packs or
21079 va_arg_pack_len when the inlined call stmt requires pack
21081 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
21083 2017-03-21 Jakub Jelinek <jakub@redhat.com>
21086 * tsan.c (instrument_builtin_call): If the memory model argument
21087 is not a constant, assume it is valid.
21090 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
21093 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
21095 PR rtl-optimization/79910
21096 * combine.c (can_combine_p): Do not allow combining an I0 or I1
21097 if its dest is used by an insn before I2 (other than the combined
21098 insns themselves, which are properly handled already).
21100 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
21103 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
21105 * combine.c (record_used_regs): New static function.
21106 (try_combine): Handle situations where there is an additional
21107 instruction between I2 and I3 which needs to have a LOG_LINK
21111 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
21113 * combine.c (try_combine): Delete redundant i1 test. Call
21114 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
21116 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21119 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
21120 alternatives 13/14.
21122 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21124 PR tree-optimization/80054
21125 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
21126 the optimization if a PHI or any of its arguments is not dominated
21127 by the candidate's basis. Use gphi* rather than gimple* as
21129 (replace_profitable_candidates): Clean up a gimple* variable that
21130 should be a gphi* variable.
21132 2017-03-20 Martin Sebor <msebor@redhat.com>
21135 * doc/extend.texi (attribute constructor): Document present limitation.
21137 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
21140 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
21141 __POWER9_VECTOR__ #ifdef control, change template definition to
21142 use Power9-specific built-in function.
21143 (vec_any_eq): Likewise.
21144 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
21145 to control outcomes from this test.
21146 (vector_ae_<mode>p): For VEC_F modes, likewise.
21148 2017-03-20 Ian Lance Taylor <iant@google.com>
21150 * config/i386/i386.c (ix86_function_regparm): Save an extra
21151 register for -fsplit-stack with DECL_STATIC_CHAIN.
21153 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
21156 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
21157 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
21159 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
21161 * config/riscv/riscv.c (riscv_print_operand): Use "fence
21163 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
21166 2017-03-20 Marek Polacek <polacek@redhat.com>
21169 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
21171 2017-03-20 Richard Biener <rguenther@suse.de>
21173 PR tree-optimization/80113
21174 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
21175 allocate extra SSA name for PHI def.
21176 (add_close_phis_to_outer_loops): Likewise.
21177 (add_close_phis_to_merge_points): Likewise.
21178 (copy_loop_close_phi_args): Likewise.
21179 (copy_cond_phi_nodes): Likewise.
21181 2017-03-20 Martin Liska <mliska@suse.cz>
21183 PR middle-end/79753
21184 * tree-chkp.c (chkp_build_returned_bound): Do not build
21185 returned bounds for a LHS that's not a BOUNDED_P type.
21187 2017-03-20 Martin Liska <mliska@suse.cz>
21191 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
21192 COMPLEX_CST and VECTOR_CST.
21194 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21197 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
21198 target operand. A new splitter adds the clobber statement in case
21199 the target operand is dead anyway.
21201 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
21203 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
21204 to age-old versions of binutils and glibc.
21206 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
21208 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
21210 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
21212 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
21214 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
21216 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
21217 requirement for binutils 2.13.
21219 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
21221 * combine.c (try_combine): Delete redundant i1 test. Call
21222 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
21224 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
21226 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
21227 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
21229 <riscv64-*-elf>: Re-arrange section
21230 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
21231 <riscv32-*-linux>: Likewise.
21232 <riscv64-*-elf>: Likewise
21233 <riscv64-*-linux>: Likewise.
21235 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
21238 * aarch64.opt(verbose-cost-dump): Fix typo.
21240 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
21243 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
21244 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
21246 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
21248 * reload.c (find_reloads): When reloading a nonoffsettable address,
21249 use RELOAD_OTHER for it and its address reloads.
21251 PR rtl-optimization/79910
21252 * combine.c (record_used_regs): New static function.
21253 (try_combine): Handle situations where there is an additional
21254 instruction between I2 and I3 which needs to have a LOG_LINK
21257 2017-03-17 Jeff Law <law@redhat.com>
21259 PR tree-optimization/71437
21260 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
21261 conditional in the hash table first.
21262 (vrp_dom_walker::before_dom_children): Extract condition from
21263 ASSERT_EXPR. Record condition, its inverion and any implied
21264 conditions as well.
21266 2017-03-17 Marek Polacek <polacek@redhat.com>
21267 Markus Trippelsdorf <markus@trippelsdorf.de>
21269 PR tree-optimization/80079
21270 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
21273 2017-03-17 Richard Biener <rguenther@suse.de>
21275 PR middle-end/80075
21276 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
21277 Properly verify the LHS before the RHS possibly claims to be
21279 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
21282 2017-03-17 Martin Jambor <mjambor@suse.cz>
21284 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
21285 (List of -O2 options): Likewise.
21286 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
21289 2017-03-17 Tom de Vries <tom@codesourcery.com>
21291 * gcov-dump.c (print_usage): Print bug_report_url.
21293 2017-03-17 Richard Biener <rguenther@suse.de>
21295 PR middle-end/80050
21296 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
21297 (parser::peek): Likewise.
21299 2017-03-17 Richard Biener <rguenther@suse.de>
21301 PR tree-optimization/80048
21302 * sese.c (free_sese_info): Properly release rename_map and
21303 copied_bb_map elements.
21305 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
21307 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
21308 Add linked-list forward and backlinks. Insert on
21309 construction, remove on destruction.
21310 (class pass_store_merging): Add m_stores_head field.
21311 (pass_store_merging::terminate_and_process_all_chains):
21312 Iterate over m_stores_head list.
21313 (pass_store_merging::terminate_all_aliasing_chains):
21315 (pass_store_merging::execute): Check for debug stmts first.
21316 Push new chains onto the m_stores_head stack.
21318 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
21321 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
21322 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
21323 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
21325 2017-03-16 Jeff Law <law@redhat.com>
21327 PR tree-optimization/71437
21328 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
21329 member function. Implementation moved into after_dom_children
21330 member function and into the threader's thread_outgoing_edges
21332 (dom_opt_dom_walker::after_dom_children): Simplify by moving
21333 some code into new thread_outgoing_edges.
21334 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
21335 definition. Simplify marker handling (do it here). Assume we always
21336 have the available expression and the const/copies tables.
21337 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
21339 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
21340 * tree-vrp.c (equiv_stack): No longer file scoped.
21341 (vrp_dom_walker): New class.
21342 (vrp_dom_walker::before_dom_children): New member function.
21343 (vrp_dom_walker::after_dom_children): Likewise.
21344 (identify_jump_threads): Setup domwalker. Use it rather than
21345 walking edges in a random order by hand. Simplify setup/finalization.
21346 (finalize_jump_threads): Remove.
21347 (vrp_finalize): Do not call identify_jump_threads here.
21348 (execute_vrp): Do it here instead and call thread_through_all_blocks
21351 PR tree-optimization/71437
21352 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
21354 (simplify_stmt_for_jump_threading): Add basic_block argument. All
21356 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
21357 (dom_opt_dom_walker::thread_across_edge): Remove
21358 handle_dominating_asserts argument. All callers changed.
21359 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
21360 changes. Remove calls to lhs_of_dominating_assert. Other
21361 uses of handle_dominating_asserts turn into unconditional code
21362 (simplify_control_stmt_condition_1): Likewise.
21363 (simplify_control_stmt_condition): Likewise.
21364 (thread_through_normal_block, thread_across_edge): Likewise.
21365 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
21366 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
21367 object if it is not an SSA_NAME.
21368 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
21369 before calling into the VRP specific simplifiers.
21370 (identify_jump_threads): Remove handle_dominating_asserts
21373 2017-03-16 Jakub Jelinek <jakub@redhat.com>
21376 * tree-diagnostic.c (default_tree_printer): No longer static.
21377 * tree-diagnostic.h (default_tree_printer): New prototype.
21379 2017-03-16 Tamar Christina <tamar.christina@arm.com>
21381 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
21382 Change ins into fmov.
21384 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21386 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
21387 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
21388 Use h_con constraint for operand 1.
21389 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
21390 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
21392 2017-03-15 Jeff Law <law@redhat.com>
21394 PR tree-optimization/71437
21395 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
21396 (record_temporary_equivalences): Use it.
21398 PR tree-optimization/71437
21399 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
21400 tree-ssa-scopedtables.
21401 (lookup_avail_expr, build_and_record_new_cond): Likewise.
21402 (record_conditions, record_cond, vuse_eq): Likewise.
21403 (record_edge_info): Adjust to API tweak of record_conditions.
21404 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
21405 (record_temporary_equivalences, optimize_stmt): Likewise.
21406 (eliminate_redundant_computations): Likewise.
21407 (record_equivalences_from_stmt): Likewise.
21408 * tree-ssa-scopedtables.c: Include options.h and params.h.
21409 (vuse_eq): New function, moved from tree-ssa-dom.c
21410 (build_and_record_new_cond): Likewise.
21411 (record_conditions): Likewise. Accept vector of conditions rather
21412 than edge_equivalence structure for first argument.
21413 for the first argument.
21414 (avail_exprs_stack::lookup_avail_expr): New member function, moved
21415 from tree-ssa-dom.c.
21416 (avail_exprs_stack::record_cond): Likewise.
21417 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
21418 from tree-ssa-dom.c.
21419 (avail_exprs_stack): Add new member functions lookup_avail_expr
21421 (record_conditions): Declare.
21423 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
21426 * lra-constraints.c (process_alt_operands): Increase reject for
21427 reloading an input/output operand.
21429 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
21432 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
21433 insns to convert from signed/unsigned char/short to IEEE 128-bit
21435 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
21437 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
21440 * config/i386/i386.c (ix86_vector_duplicate_value): Create
21441 subreg of inner mode for values already in registers.
21443 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
21445 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
21446 iteration reg is used after the loop.
21448 2017-03-14 Martin Sebor <msebor@redhat.com>
21450 PR tree-optimization/79800
21451 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
21452 precision in negative-positive range.
21453 (format_floating): Call non-const overload with adjusted precision.
21455 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
21458 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
21461 2017-03-14 Martin Sebor <msebor@redhat.com>
21463 PR middle-end/80020
21464 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
21465 * builtins.def (aligned_alloc): Use it.
21468 * Makefile.in (GTFILES): Add calls.c.
21469 * calls.c: Include "gt-calls.h".
21471 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
21473 PR rtl-optimization/79728
21474 * regs.h (struct target_regs): New field
21475 x_contains_allocatable_regs_of_mode.
21476 (contains_allocatable_regs_of_mode): New macro.
21477 * reginfo.c (init_reg_sets_1): Initialize it, and change
21478 contains_reg_of_mode so it includes global regs as well.
21479 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
21480 rather than contains_regs_of_mode.
21482 2017-03-14 Martin Liska <mliska@suse.cz>
21484 * doc/invoke.texi: Document options that can't be combined with
21485 -fcheck-pointer-bounds.
21487 2017-03-14 Martin Liska <mliska@suse.cz>
21489 PR middle-end/79831
21490 * doc/invoke.texi (-Wchkp): Document the option.
21492 2017-03-14 Martin Liska <mliska@suse.cz>
21494 * Makefile.in: Install gcov-dump.
21496 2017-03-14 Martin Liska <mliska@suse.cz>
21498 * multiple_target.c (expand_target_clones): Bail out for
21499 an invalid attribute.
21501 2017-03-14 Richard Biener <rguenther@suse.de>
21503 * alias.c (struct alias_set_entry): Pack properly.
21504 * cfgloop.h (struct loop): Likewise.
21505 * cse.c (struct set): Likewise.
21506 * ipa-utils.c (struct searchc_env): Likewise.
21507 * loop-invariant.c (struct invariant): Likewise.
21508 * lra-remat.c (struct cand): Likewise.
21509 * recog.c (struct change_t): Likewise.
21510 * rtl.h (struct address_info): Likewise.
21511 * symbol-summary.h (function_summary): Likewise.
21512 * tree-loop-distribution.c (struct partition): Likewise.
21513 * tree-object-size.c (struct object_size_info): Likewise.
21514 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
21515 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
21516 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
21517 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
21518 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
21519 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
21520 (struct _stmt_vec_info): Likewise.
21522 2017-03-14 Martin Liska <mliska@suse.cz>
21525 * multiple_target.c (create_dispatcher_calls): Check that
21526 a target can create a function dispatcher.
21528 2017-03-14 Martin Liska <mliska@suse.cz>
21531 * multiple_target.c (expand_target_clones): Drop local.local
21532 flag for default implementation.
21534 2017-03-14 Richard Biener <rguenther@suse.de>
21536 PR tree-optimization/80030
21537 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
21539 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
21541 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
21542 gcc_fallthrough() instead of __attribute__((fallthrough));
21544 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
21546 * doc/gcc.texi: Remove "up" link to (DIR).
21547 * doc/gccint.texi: Ditto.
21549 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
21551 * doc/install.texi (Specific) <avr>: Remove reference to
21554 2017-03-13 Jeff Law <law@redhat.com>
21556 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
21557 attribute rather than comments.
21559 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
21560 match_scratch operand is highest.
21562 2017-03-13 Martin Liska <mliska@suse.cz>
21564 PR middle-end/78339
21565 * ipa-pure-const.c (warn_function_noreturn): If the declarations
21566 is a CHKP clone, use original declaration.
21568 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
21570 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
21571 (arc_conditional_register_usage): Use a different allocation order
21572 when optimizing for size.
21573 * common/config/arc/arc-common.c (arc_option_optimization_table):
21574 Section anchors default on when optimizing for size.
21576 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
21578 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
21580 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
21582 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
21583 * config/arc/arc.md (cpu_facility): Add cd variant.
21584 (*movqi_insn): Add code density variant.
21585 (*movhi_insn): Likewise.
21586 (*movqi_insn): Likewise.
21587 (*addsi3_mixed): Likewise.
21588 (subsi3_insn): Likewise.
21590 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
21592 * config/arc/arc.md (movsi_cond_exec): Update constraint.
21594 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
21596 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
21597 expressions with MINUS and UNARY ops.
21599 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21602 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
21604 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
21605 between vec_select and vector argument.
21606 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
21607 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
21608 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
21609 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
21610 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
21611 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
21613 2017-03-13 Richard Biener <rguenther@suse.de>
21616 * params.def (vect-max-peeling-for-alignment): Fix typo.
21618 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
21620 * doc/install.texi (Specific) <mips-*-*>: Remove description of
21621 issue that only occurred with binutils below 2.18.
21623 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
21625 * doc/install.texi (Specific) <cris-axis-elf>: No longer
21626 refer to binutils 2.11/2.12 minimum.
21628 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
21630 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
21631 ftp.kernel.org and simplify binutils requirement.
21633 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
21635 * doc/invoke.texi (Warning Options): Fix spelling of link-time
21637 (Optimize Options): Ditto. Also remove redundancy.
21639 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21641 PR translation/79848
21642 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
21644 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
21645 to G_ to avoid double translation.
21647 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21649 PR translation/79923
21650 * auto-profile.c (get_combined_location): Convert leading
21651 character of diagnostics to lower case and remove trailing period.
21652 (read_profile): Likewise for various diagnostics.
21653 * config/arm/arm.c (arm_option_override): Remove trailing period
21654 from various diagnostics.
21655 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
21656 (msp430_expand_delay_cycles): Likewise.
21658 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21661 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
21662 full command-line argument, rather than just "str".
21663 (aarch64_validate_march): Likewise.
21664 (aarch64_validate_mtune): Likewise.
21666 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
21668 PR rtl-optimization/78911
21669 * lra-assigns.c (must_not_spill_p): New function.
21670 (spill_for): Use it.
21672 2017-03-10 Jakub Jelinek <jakub@redhat.com>
21674 PR tree-optimization/79981
21675 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
21676 ATOMIC_COMPARE_EXCHANGE ifn result.
21677 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
21678 IFN_ATOMIC_COMPARE_EXCHANGE.
21680 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21683 * opts.c (parse_sanitizer_options): Add missing question mark to
21684 "did you mean" message.
21686 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21688 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
21690 (VMULEUH_UNS): Likewise.
21691 (VMULOUB_UNS): Likewise.
21692 (VMULOUH_UNS): Likewise.
21693 * config/rs6000/rs6000.c (builtin_function_type): Remove
21694 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
21696 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21699 * read-rtl-function.c (function_reader::read_rtx_operand): Update
21700 x with result of extra_parsing_for_operand_code_0.
21701 (function_reader::extra_parsing_for_operand_code_0): Convert
21702 return type from void to rtx, returning x. When reading
21703 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
21704 larger size containing struct block_symbol.
21706 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
21708 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
21709 -mfloat128-hardware without -m64.
21711 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
21714 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
21715 entries to the case statement that marks unsigned arguments to
21716 overloaded functions.
21718 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
21720 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
21721 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
21723 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
21726 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
21727 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
21729 2017-03-10 Martin Liska <mliska@suse.cz>
21733 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
21734 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
21737 2017-03-10 Olivier Hainque <hainque@adacore.com>
21739 * tree-switch-conversion (array_value_type): Start by resetting
21740 candidate type to it's main variant.
21742 2017-03-10 Jakub Jelinek <jakub@redhat.com>
21744 PR rtl-optimization/79909
21745 * combine.c (try_combine): Use simplify_replace_rtx on individual
21746 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
21747 of the whole CALL_INSN_FUNCTION_USAGE.
21749 PR tree-optimization/79972
21750 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
21751 get_range_info on SSA_NAMEs. Formatting fixes.
21753 2017-03-10 Richard Biener <rguenther@suse.de>
21754 Jakub Jelinek <jakub@redhat.com>
21756 PR tree-optimization/77975
21757 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
21758 edge to be constant.
21759 (get_val_for): For constant x return it. Formatting fix.
21760 (loop_niter_by_eval): Avoid pointless looping if the next iteration
21761 would use the same bases as the current one.
21763 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21765 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
21766 instead of vec_select for V1TImode.
21767 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
21769 (VSX_LE_128): Add V1TI to this mode iterator.
21770 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
21771 (*vsx_le_perm_store_<mode>): Likewise.
21772 (pre-reload splitter for VSX stores): Likewise.
21773 (post-reload splitter for VSX stores): Likewise.
21774 (*vsx_xxpermdi2_le_<mode>): Likewise.
21775 (*vsx_lxvd2x2_le_<mode>): Likewise.
21776 (*vsx_stxvd2x2_le_<mode>): Likewise.
21778 2017-03-09 Michael Eager <eager@eagercon.com>
21780 Correct failures with --enable-checking=yes,rtl.
21782 * config/microblaze/microblaze.c (microblaze_expand_shift):
21783 Replace GET_CODE test with CONST_INT_P and INTVAL test with
21784 test for const0_rtx.
21785 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
21786 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
21788 2017-03-09 Richard Biener <rguenther@suse.de>
21790 PR tree-optimization/79977
21791 * graphite-scop-detection.c (scop_detection::merge_sese):
21792 Handle the case of extra exits to blocks dominating the entry.
21794 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
21796 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
21799 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
21801 PR rtl-optimization/79949
21802 * lra-constraints.c (process_alt_operands): Check memory when
21803 trying to predict a cycle. Print about the overall increase.
21805 2017-03-09 Richard Biener <rguenther@suse.de>
21807 PR middle-end/79971
21808 * gimple-expr.c (useless_type_conversion_p): Preserve
21809 TYPE_SATURATING for fixed-point types.
21811 2017-03-09 Richard Biener <rguenther@suse.de>
21814 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
21815 alignment of BLKmode params.
21817 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21820 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
21821 (VALL_NO_V2Q): Likewise.
21823 * config/aarch64/aarch64-simd.md
21824 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
21826 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
21827 VALL_NO_V2Q mode iterator.
21828 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
21830 2017-03-09 Martin Liska <mliska@suse.cz>
21832 PR tree-optimization/79631
21833 * tree-chkp-opt.c (chkp_is_constant_addr): Call
21834 tree_int_cst_sign_bit just for INTEGER constants.
21836 2017-03-09 Martin Liska <mliska@suse.cz>
21840 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
21843 2017-03-09 Marek Polacek <polacek@redhat.com>
21846 * tree.c (inchash::add_expr): Handle TREE_VEC.
21848 2017-03-09 Martin Liska <mliska@suse.cz>
21851 (chkp_narrow_size_and_offset): New function.
21852 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
21853 (void chkp_parse_bit_field_ref): New function.
21854 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
21855 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
21857 2017-03-09 Martin Liska <mliska@suse.cz>
21860 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
21861 (chkp_find_bounds_1): Remove gcc_unreachable.
21863 2017-03-09 Jakub Jelinek <jakub@redhat.com>
21866 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
21867 BUILT_IN_SYNC*, determine the access type from the size suffix and
21868 always build a MEM_REF with that type. Handle forgotten
21869 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
21872 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
21873 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
21874 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
21875 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
21876 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
21877 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
21878 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
21879 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
21880 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
21881 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
21882 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
21883 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
21884 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
21885 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
21886 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
21887 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
21888 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
21889 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
21890 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
21891 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
21892 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
21893 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
21894 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
21895 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
21896 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
21897 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
21898 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
21899 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
21900 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
21901 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
21902 definitions outside of __OPTIMIZE__ guarded section.
21905 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
21906 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
21907 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
21908 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
21911 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21913 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
21914 ("vfenez<mode>"): Add missing constraints.
21916 2017-03-08 Martin Sebor <msebor@redhat.com>
21919 * config/nds32/nds32.c (nds32_option_override):
21920 Fix misspelled diagnostic.
21922 2017-03-08 Jakub Jelinek <jakub@redhat.com>
21925 * gimplify.c (gimplify_omp_for): Replace index var in outer
21926 taskloop statement with an artificial variable and add
21927 OMP_CLAUSE_PRIVATE clause for it.
21929 2017-03-08 Richard Biener <rguenther@suse.de>
21931 PR tree-optimization/79955
21932 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
21933 for accesses that are completely outside of the variable.
21935 2017-03-08 Andrew Haley <aph@redhat.com>
21937 PR tree-optimization/79943
21938 * tree-ssa-loop-split.c (compute_new_first_bound): When
21939 calculating the new upper bound, (END-BEG) should be added, not
21942 2017-03-08 Jakub Jelinek <jakub@redhat.com>
21944 * config/avr/avr.md (setmemhi): Make sure match_dup
21945 operand number comes before match_scratch.
21947 2017-03-08 Richard Biener <rguenther@suse.de>
21949 PR tree-optimization/79920
21950 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
21951 with ncopies == 1 to ...
21952 (vect_transform_slp_perm_load): ... here. Properly compute
21953 all element loads by iterating VF times over the group. Do
21954 not handle ncopies (computed in a broken way) in
21955 vect_create_mask_and_perm.
21957 2017-03-08 Jakub Jelinek <jakub@redhat.com>
21960 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
21961 is a uniform vector, use uniform_vector_p return value instead of
21962 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
21964 2017-03-07 Marek Polacek <polacek@redhat.com>
21966 PR middle-end/79809
21967 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
21968 (alloca_call_type): Likewise.
21970 2017-03-07 Martin Liska <mliska@suse.cz>
21972 * gcov.c (process_args): Put comment to correct location.
21974 2017-03-07 Martin Liska <mliska@suse.cz>
21976 PR middle-end/68270
21977 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
21978 Use array_at_struct_end_p instead of DECL_CHAIN (field).
21979 (chkp_narrow_bounds_for_field): Likewise.
21980 (chkp_parse_array_and_component_ref): Pass one more argument to
21983 2017-03-07 Richard Biener <rguenther@suse.de>
21985 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
21988 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
21990 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
21991 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
21993 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21996 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
21997 to end of description.
21998 (PARAM_MAX_STORES_TO_MERGE): Likewise.
22000 2017-03-07 Jakub Jelinek <jakub@redhat.com>
22002 PR rtl-optimization/79901
22003 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
22005 (*avx512f_<code><mode>3<mask_name>): ... this.
22006 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
22007 iterator instead of VI8_AVX2_AVX512BW.
22009 PR rtl-optimization/79901
22010 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
22011 min/max expander, expand it using expand_vec_cond_expr.
22014 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
22017 2017-03-06 Jakub Jelinek <jakub@redhat.com>
22020 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
22021 to void * for PCH reasons.
22022 * dwarf2out.c (output_loc_operands, output_die): Cast
22023 v.val_vec.array to unsigned char *.
22025 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
22028 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
22031 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
22033 PR rtl-optimization/79571
22034 * lra-constraints.c (process_alt_operands): Calculate static
22035 reject and subtract it from overall when only addresses will be
22038 2017-03-06 Julia Koval <julia.koval@intel.com>
22041 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
22042 incoming stack boundary to 128 for 64-bit targets.
22044 2017-03-06 Richard Biener <rguenther@suse.de>
22046 PR tree-optimization/79894
22047 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
22048 to NULL after folding it.
22050 2017-03-06 Richard Biener <rguenther@suse.de>
22052 PR tree-optimization/79824
22053 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
22054 check disabling peeling for gaps.
22056 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
22058 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
22059 attributes): Document gettimeofday.
22061 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
22063 * config/s390/s390.c (s390_option_override_internal): Set
22064 PARAM_MIN_VECT_LOOP_BOUND
22066 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
22068 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
22069 * config/s390/s390.md: Likewise.
22071 2017-03-06 Jakub Jelinek <jakub@redhat.com>
22074 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
22075 (<avx2_avx512>_perm<mode>): Rename to ...
22076 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
22078 (<avx512>_perm<mode>_mask): Rename to ...
22079 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
22081 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
22082 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
22083 instead of VI8F_256_512.
22084 (avx512f_perm<mode>): New define_expand.
22085 (avx512f_perm<mode>_mask): Likewise.
22086 (avx512f_perm<mode>_1<mask_name>): New define_insn.
22087 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
22089 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
22091 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
22092 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
22094 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
22096 2017-03-06 Martin Liska <mliska@suse.cz>
22099 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
22100 when having a SSA NAME w/o VAR_DECL assigned to it.
22102 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
22104 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
22105 msa_dpsub_<su>_d): Fix MODE for vec_select.
22107 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
22109 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
22111 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
22113 2017-03-06 Richard Biener <rguenther@suse.de>
22115 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
22116 * plugin.c (register_plugin_info): Likewise.
22117 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
22119 2017-03-05 Jakub Jelinek <jakub@redhat.com>
22121 * config/i386/sse.md (sse_storehps, sse_storelps,
22122 avx_<castmode><avxsizesuffix>_<castmode>,
22123 avx512f_<castmode><avxsizesuffix>_<castmode>,
22124 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
22125 in condition that at least one operand is not a MEM.
22127 2017-03-03 Jakub Jelinek <jakub@redhat.com>
22129 PR middle-end/79805
22130 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
22131 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
22133 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
22134 gimple_call_nothrow_p flag based on whether original builtin can throw.
22135 If it can, emit following stmts on the fallthrough edge.
22136 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
22137 don't create new bb if inserting just debug stmts on the edge, try to
22138 insert them on the fallthru bb or just reset debug stmts.
22140 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
22143 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
22144 restore recog_data (including the operand rtxes inside it) around
22145 the call to get_insn_template.
22147 2017-03-03 Martin Sebor <msebor@redhat.com>
22149 PR tree-optimization/79699
22150 * context.c (context::~context): Free MPFR caches to avoid
22151 a memory leak on program exit.
22153 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22155 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
22156 Use wide_int::ulow () instead of .elt (0).
22158 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
22160 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
22161 (*pushxf): Limit oF constraint to 32bit targets and add oC
22162 constraint for 64bit targets.
22163 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
22164 (*pushdf): Change rmF constraint to rmC.
22166 2017-03-03 Martin Liska <mliska@suse.cz>
22168 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
22169 Remove unused variable.
22171 2017-03-03 Jakub Jelinek <jakub@redhat.com>
22174 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
22175 is a memory operand, increase num_memory.
22176 (ix86_expand_args_builtin): Likewise.
22178 2017-03-03 Jan Hubicka <jh@suse.cz>
22181 * ipa-devirt.c (maybe_record_node): Properly handle
22182 __cxa_pure_virtual visibility.
22184 2017-03-03 Martin Liska <mliska@suse.cz>
22186 PR tree-optimization/79803
22187 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
22189 (pass_loop_prefetch::execute): Disabled optimization if an
22190 assumption about L1 cache size is not met.
22192 2017-03-03 Martin Liska <mliska@suse.cz>
22194 PR rtl-optimization/79574
22195 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
22196 (hash_scan_set): Likewise.
22197 (dump_hash_table): Likewise.
22198 (hoist_code): Likewise.
22200 2017-03-03 Richard Biener <rguenther@suse.de>
22202 * fixed-value.c (fixed_from_string): Restore use of elt (1)
22203 in place of uhigh ().
22204 (fixed_convert_from_real): Likewise.
22206 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
22209 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
22211 2017-03-03 Richard Biener <rguenther@suse.de>
22213 PR middle-end/79818
22214 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
22215 TYPE_OVERFLOW_UNDEFINED check.
22217 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22219 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
22221 (vector_ae_<mode>_p): Likewise.
22222 (vector_nez_<mode>_p): Likewise.
22223 (vector_ne_v2di_p): Likewise.
22224 (vector_ae_v2di_p): Likewise.
22225 (vector_ne_<mode>_p): Likewise.
22226 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
22228 (vsx_tsqrt<mode>2_fe): Likewise.
22230 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
22233 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
22235 2017-03-02 Jakub Jelinek <jakub@redhat.com>
22237 PR rtl-optimization/79780
22238 * cprop.c (one_cprop_pass): When second and further conditional trap
22239 in a single basic block is turned into an unconditional trap, turn it
22240 into a deleted note to avoid RTL verification failures.
22242 2017-03-02 Richard Biener <rguenther@suse.de>
22244 * fold-const.c (const_binop): Use ulow () instead of elt (0).
22246 2017-03-02 Richard Biener <rguenther@suse.de>
22248 PR tree-optimization/79345
22250 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
22251 param and abort the walk, returning -1 if it is hit.
22252 (walk_aliased_vdefs): Take a limit param and pass it on.
22253 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
22254 defaulting to 0 and return a signed int.
22255 * tree-ssa-uninit.c (struct check_defs_data): New struct.
22256 (check_defs): New helper.
22257 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
22258 about uninitialized memory.
22259 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
22260 bogus uninitialized warning.
22261 (fixed_convert_from_real): Likewise.
22263 2017-03-02 Bin Cheng <bin.cheng@arm.com>
22265 PR tree-optimization/66768
22266 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
22267 iv_use if base object can't be determined.
22269 2017-03-02 Jakub Jelinek <jakub@redhat.com>
22271 PR tree-optimization/79345
22272 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
22273 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
22274 (get_pattern_stats): Initialize it.
22275 * genemit.c (gen_expand): Verify match_scratch numbers come after
22276 match_operand/match_dup numbers.
22277 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
22278 match_scratch numbers.
22279 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
22281 * config/s390/s390.md (trunctdsd2): Likewise.
22283 2017-03-02 Richard Biener <rguenther@suse.de>
22285 * wide-int.h (wide_int_storage::operator=): Implement in terms
22288 2017-03-02 Richard Biener <rguenther@suse.de>
22290 PR tree-optimization/79777
22291 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
22292 the to insert expression to sth existing.
22294 2017-03-01 Martin Sebor <msebor@redhat.com>
22296 PR middle-end/79692
22297 * gimple-ssa-sprintf.c
22298 (directive::known_width_and_precision): New function.
22299 (format_integer): Use it.
22300 (get_mpfr_format_length): Consider the full range of precision
22301 when computing %g output with the # flag. Set the likely byte
22302 count to 3 rather than 1 when precision is indeterminate.
22303 (format_floating): Correct the lower bound of precision.
22305 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22307 * doc/invoke.texi: Document default code model for 64-bit Linux.
22309 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22312 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
22313 udiv rather than div since input pattern is unsigned.
22315 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
22317 * config/i386/i386.c (print_reg): Warn for values of
22318 unsupported size in integer register.
22320 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
22323 * config/rs6000/predicates.md (current_file_function_operand): Do
22324 not allow self calls to be local if the function is replaceable.
22326 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
22329 * config/rs6000/altivec.h (vec_ctz and others): Change the
22330 preprocessor macro that controls conditional compilation from
22331 _ARCH_PWR9 to __POWER9_VECTOR__.
22332 (vec_all_ne): Change parameterization of __altivec_scalar_pred
22333 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
22334 control (instead of _ARCH_PWR9 control) so that template
22335 definition uses power9-specific function.
22336 (vec_any_eq): Likewise.
22337 (vec_all_ne): Change macro definition to use a power9-specific
22338 expansion under #ifdef __POWER9_VECTOR__ control (instead of
22339 _ARCH_PWR9 control).
22340 (vec_any_eq) Likewise.
22341 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
22342 expansion for CMPNEF to remove support for xvcmpnesp instruction.
22343 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
22344 support for xvcmpnedp instruction.
22345 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
22346 macro expansion so that Power9 implementation of vec_all_ne does
22347 not use the AltiVec predicate framework.
22348 (VCMPNEH_P): Likewise.
22349 (VCMPNEW_P): Likewise.
22350 (VCMPNED_P): Likewise.
22351 (VCMPNEFP_P): Likewise.
22352 (VCMPNEDP_P): Likewise.
22353 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
22354 implementation of vec_any_eq to not use AltiVec predicate
22356 (VCMPAEH_P): Likewise.
22357 (VCMPAEW_P): Likewise.
22358 (VCMPAED_P): Likewise.
22359 (VCMPAEFP_P): Likewise.
22360 (VCMPAEDP_P): Likewise.
22361 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
22362 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
22363 not use the AltiVec predicate framework.
22364 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
22365 of vec_any_eq to not use AltiVec predicate framework.
22366 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
22367 support for predefined __POWER9_VECTOR__ macro to indicate that
22368 Power9 instruction selection is enabled.
22369 (altivec_overloaded_builtins): Remove extraneous
22370 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
22371 function argument types RS6000_BTI_bool_V16QI and
22372 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
22373 entry for overloaded function argument types RS6000_BTI_bool_V4SI
22374 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
22375 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
22376 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
22377 Power9 for implementations of vec_cmpne. Change the signature for
22378 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
22379 (representing vec_all_ne) to remove the previously described first
22380 argument of type RS6000_BTI_INTSI, as this was an artifact of
22381 reliance on the AltiVec predicate framework, which is no longer
22382 used in the implementation of these functions. Add
22383 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
22384 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
22385 since, unlike the AltiVec predicate framework implementation, we
22386 do not share function descriptors between vec_alle and vec_anyeq.
22387 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
22388 set of modes that receive special treatment even when
22389 TARGET_P9_VECTOR is true. The special treatment emits code that
22390 does not depend on Power9 instructions.
22391 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
22392 define_expand to not rely on AltiVec predicate framework.
22393 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
22395 (vector_ne_v2di_p): Change this define_expand to not rely on
22396 AltiVec predicate framework.
22397 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
22399 (vector_ne_<mode>_p): Change this define_expand to not rely on
22400 AltiVec predicate framework.
22401 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
22403 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
22404 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
22405 define_insn pattern.
22406 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
22407 define_insn pattern because the xvcmpne<VSs>. instruction is not
22409 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
22410 instruction is not supported.
22412 2017-03-01 Jakub Jelinek <jakub@redhat.com>
22414 * config/nvptx/nvptx.c: Include intl.h.
22416 2017-03-01 Martin Jambor <mjambor@suse.cz>
22419 * ipa-prop.h (ipa_bits): Removed field known.
22420 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
22421 to pointers. Adjusted their comments to warn about their sharing.
22422 (ipcp_transformation_summary): Change bits to a vector of pointers.
22423 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
22424 (ipa_get_ipa_bits_for_value): Declare.
22425 * tree-vrp.h (value_range): Mark as GTY((for_user)).
22426 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
22427 (ipa_bits_hash_table): Likewise.
22428 (ipa_vr_ggc_hash_traits): Likewise.
22429 (ipa_vr_hash_table): Likewise.
22430 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
22431 being pointers and vr_known being removed.
22432 (ipa_set_jf_unknown): Likewise.
22433 (ipa_get_ipa_bits_for_value): New function.
22434 (ipa_set_jfunc_bits): Likewise.
22435 (ipa_get_value_range): New overloaded functions.
22436 (ipa_set_jfunc_vr): Likewise.
22437 (ipa_compute_jump_functions_for_edge): Use the above functions to
22438 construct bits and vr parts of jump functions.
22439 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
22440 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
22442 (ipcp_grow_transformations_if_necessary): Also allocate
22443 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
22445 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
22446 them. Fix too long lines.
22447 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
22448 vr_known being removed.
22449 (ipa_read_jump_function): Use new setter functions to construct bits
22450 and vr parts of jump functions or set them to NULL.
22451 (write_ipcp_transformation_info): Adjust for bits being pointers.
22452 (read_ipcp_transformation_info): Likewise.
22453 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
22455 Include gt-ipa-prop.h.
22456 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
22458 (ipcp_store_bits_results): Likewise.
22459 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
22460 Do not write to existing jump functions but use a temporary instead.
22462 2017-03-01 Jakub Jelinek <jakub@redhat.com>
22465 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
22466 attempt to use its first operand as BIT_FIELD_REF base.
22468 2017-03-01 Richard Biener <rguenther@suse.de>
22470 PR middle-end/79721
22471 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
22472 interpolating formula in wrapping arithmetic.
22473 (chrec_apply): Convert chrec_evaluate return value to wanted type.
22475 2017-03-01 Jakub Jelinek <jakub@redhat.com>
22477 PR tree-optimization/79734
22478 * tree-vect-generic.c (expand_vector_condition): Optimize
22479 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
22480 Handle VEC_COND_EXPR where comparison has different inner width from
22481 type's inner width.
22483 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
22485 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
22486 markup, and similar issues. Remove @opindex entries for things
22487 that aren't options. Add missing -mmpy-option entries.
22489 2017-02-28 Jakub Jelinek <jakub@redhat.com>
22491 PR tree-optimization/79737
22492 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
22493 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
22494 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
22495 instead of byte_size. Formatting fix.
22496 (shift_bytes_in_array_right): Formatting fix.
22498 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
22501 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
22502 condition on optimize for the leaf function test.
22504 2017-02-28 Martin Liska <mliska@suse.cz>
22507 * read-rtl-function.c (function_reader::handle_unknown_directive):
22508 Bail out when one uses -flto.
22510 2017-02-28 Martin Liska <mliska@suse.cz>
22512 * common.opt: Replace space with tabular for options of <number>
22514 * config/i386/i386.opt: Show <number> value for
22515 -mlarge-data-threshold.
22516 * opts.c (print_filtered_help): Do not display number in hexadecimal
22519 2017-02-28 Martin Liska <mliska@suse.cz>
22521 * common.opt: Fix --help=option -Q for options which are of
22524 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
22526 * config/i386/i386.c (print_reg): Error out for values
22527 of 8-bit size in invalid integer register.
22529 2017-02-28 Martin Sebor <msebor@redhat.com>
22531 PR tree-optimization/79691
22532 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
22534 2017-02-28 Jakub Jelinek <jakub@redhat.com>
22537 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
22538 gcc_unreachable with output_operand_lossage.
22540 2017-02-28 Richard Biener <rguenther@suse.de>
22542 PR tree-optimization/79740
22543 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
22545 (visit_nary_op): Insert the nary into the hashtable if we
22546 pattern-matched sth.
22547 * tree-ssa-pre.c (eliminate_insert): Robustify.
22549 2017-02-28 Richard Biener <rguenther@suse.de>
22551 PR middle-end/79731
22552 * fold-const.c (decode_field_reference): Reject out-of-bound
22555 2017-02-28 Jakub Jelinek <jakub@redhat.com>
22557 * config/i386/i386.c: Include intl.h.
22558 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
22559 instead of just cond ? "..." : "...".
22560 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
22561 * coverage.c (read_counts_file): Likewise.
22562 * omp-offload.c: Include intl.h.
22563 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
22564 of just cond ? "..." : "...".
22565 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
22566 of just cond ? "..." : "...".
22568 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
22571 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
22573 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
22574 'tune for' CPU name.
22575 * config/arm/arm-cpu-data.h: Regenerated.
22577 2017-02-28 Richard Biener <rguenther@suse.de>
22579 PR tree-optimization/79732
22580 * tree-inline.c (expand_call_inline): Do not shadow var.
22582 2017-02-28 Richard Biener <rguenther@suse.de>
22584 PR tree-optimization/79723
22585 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
22586 address-space properly.
22588 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
22590 * doc/optinfo.texi (Optimization groups): Fix option used for
22592 * doc/invoke.texi (-fopt-info): Document "omp".
22593 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
22594 (OPTGROUP_ALL): Add OPTGROUP_OMP.
22595 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
22596 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
22597 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
22599 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
22601 * dumpfile.c (optgroup_options): Instead of "openmp", associate
22602 OPTGROUP_OMP with "omp".
22604 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
22607 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
22608 for arithmetic shift of unsigned V2DI.
22610 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
22612 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
22613 arc/linux.h headers.
22614 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
22615 (LINK_SPEC): Likewise.
22616 (ARC_TLS_EXTRA_START_SPEC): Likewise.
22617 (EXTRA_SPECS): Likewise.
22618 (STARTFILE_SPEC): Likewise.
22619 (ENDFILE_SPEC): Likewise.
22620 (LIB_SPEC): Likewise.
22621 (TARGET_SDATA_DEFAULT): Likewise.
22622 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
22623 (MULTILIB_DEFAULTS): Likewise.
22624 (DWARF2_UNWIND_INFO): Likewise.
22625 * config/arc/big.h: New file.
22626 * config/arc/elf.h: Likewise.
22627 * config/arc/linux.h: Likewise.
22628 * config/arc/t-uClibc: Remove.
22630 2017-02-27 Bin Cheng <bin.cheng@arm.com>
22632 PR tree-optimization/77536
22633 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
22634 (tree_transform_and_unroll_loop): Use above function to compute the
22635 estimated niter of unrolled loop and use it when scaling profile.
22636 Also use count info rather than frequency if it's non-zero.
22637 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
22638 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
22639 (vect_transform_loop): Call above function.
22641 2017-02-27 Richard Biener <rguenther@suse.de>
22643 PR tree-optimization/45397
22644 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
22645 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
22646 (visit_nary_op): Add pattern matching for CSEing sign-changed
22647 or truncated operations with wider ones.
22649 2017-02-27 Richard Biener <rguenther@suse.de>
22651 PR tree-optimization/79690
22652 * tree-vect-stmts.c (vectorizable_store): Use vector type
22653 built from the DR with address-space.
22655 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
22657 * doc/invoke.texi (Optimize Options): Refine the description
22658 of asan-use-after-return.
22660 2017-02-25 Alan Modra <amodra@gmail.com>
22662 PR rtl-optimization/79584
22663 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
22664 base, not ad->base_term, the reg within base. Remove assertion
22665 that ad->base == ad->base_term. Replace gen_int_mode using
22666 bogus mode with const0_rtx.
22668 2017-02-25 Jakub Jelinek <jakub@redhat.com>
22670 PR middle-end/79396
22671 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
22672 FMA_EXPR like tcc_binary or tcc_unary.
22674 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
22677 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
22679 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
22680 (output_loc_operands): Handle DW_OP_call_ref and
22681 DW_OP_GNU_variable_value.
22682 (struct variable_value_struct): New type.
22683 (struct variable_value_hasher): Likewise.
22684 (variable_value_hash): New variable.
22685 (string_types): Remove.
22686 (copy_loc_descr): New function.
22687 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
22688 (prepend_loc_descr_to_each): New function.
22689 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
22690 instead of add_loc_descr_to_each if the first argument is single
22691 location list and the second has multiple.
22692 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
22693 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
22694 when looking for variable value which doesn't have other location info.
22695 (loc_list_from_tree): Formatting fix.
22696 (gen_array_type_die): Simplify DW_AT_string_length handling.
22697 (adjust_string_types): Remove.
22698 (gen_subprogram_die): Don't call adjust_string_types nor test/set
22699 string_types. Call resolve_variable_values.
22700 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
22701 (resolve_addr_in_expr): Likewise. Add A argument.
22702 (copy_deref_exprloc): Remove deref argument. Adjust for the
22703 original expression being DW_OP_GNU_variable_value with optionally
22704 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
22705 optionally after it.
22706 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
22707 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
22708 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
22709 (variable_value_hasher::hash, variable_value_hasher::equal): New
22711 (resolve_variable_value_in_expr, resolve_variable_value,
22712 resolve_variable_values, note_variable_value_in_expr,
22713 note_variable_value): New functions.
22714 (dwarf2out_early_finish): Call note_variable_value on all toplevel
22717 2017-02-24 Jakub Jelinek <jakub@redhat.com>
22720 * opts.h (handle_generated_option): Add GENERATED_P argument.
22721 * opts-common.c (handle_option): Adjust function comment.
22722 (handle_generated_option): Add GENERATED_P argument, pass it to
22724 (control_warning_option): Pass false to handle_generated_option
22726 * opts.c (maybe_default_option): Pass true to handle_generated_option
22728 * optc-gen.awk: Likewise.
22730 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22732 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
22733 a REG, look at the REG it is a SUBREG of.
22734 (splitter for cmpeqsi_t): Ditto.
22736 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22738 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
22739 the special USEs with the pattern of the insn, not the insn itself.
22741 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
22744 * doc/invoke.texi: Document -mload-store-pairs.
22746 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22747 Sandra Loosemore <sandra@codesourcery.com>
22749 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
22750 argument isn't a CONST_INT.
22751 (nios2_alternate_compare_const): Assert op is a CONST_INT.
22752 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
22753 (nios2_validate_compare): Bypass alternate compare logic if *op2
22754 is not a CONST_INT.
22755 (ldstwm_operation_p): Return false if first_base is not a REG or
22756 if first_offset is not a CONST_INT.
22758 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22760 * config/cris/cris.md: Use correct operand in a define_peephole2.
22762 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22764 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
22766 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22768 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
22769 this_insn if it is an INSN or JUMP_INSN.
22770 (force_offsettable): Look at base, not at addr.
22771 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
22772 on things that aren't necessarily CONST_INTs.
22774 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
22776 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
22777 -mfpmath=sse is the default also for x86-32 targets with SSE2
22778 instruction set when @option{-ffast-math} is enabled
22780 2017-02-24 Jeff Law <law@redhat.com>
22782 PR rtl-optimizatoin/79286
22783 * ira.c (update_equiv_regs): Drop may_trap_p exception to
22786 2017-02-24 Richard Biener <rguenther@suse.de>
22788 PR tree-optimization/79389
22789 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
22792 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
22794 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
22795 function comment to reflect reality.
22796 (loop_exits_before_overflow): Fix typo in function description.
22798 2017-02-24 Richard Biener <rguenther@suse.de>
22800 PR tree-optimization/79389
22801 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
22802 properly that a threading opportunity exists. Detect conditional
22803 copy/constant propagation opportunities.
22805 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
22807 * config/visium/visium.md (type): Add trap.
22808 (b): New mode attribute.
22809 (*btst): Rename into...
22810 (*btst<mode>): ...this and adjust.
22811 (*cbranchsi4_btst_insn): Rename into...
22812 (*cbranch<mode>4_btst_insn): ...this and adjust.
22813 (trap): New define_insn.
22815 2017-02-23 Jakub Jelinek <jakub@redhat.com>
22817 PR tree-optimization/79389
22818 * ifcvt.c (struct noce_if_info): Add rev_cond field.
22819 (noce_reversed_cond_code): New function.
22820 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
22821 reversed_comparison_code. Formatting fix.
22822 (noce_try_store_flag): Test rev_cond != NULL in addition to
22823 reversed_comparison_code.
22824 (noce_try_store_flag_constants): Likewise.
22825 (noce_try_store_flag_mask): Likewise.
22826 (noce_try_addcc): Use rev_cond if non-NULL instead of
22827 reversed_comparison_code.
22828 (noce_try_cmove_arith): Likewise. Formatting fixes.
22829 (noce_try_minmax, noce_try_abs): Clear rev_cond.
22830 (noce_find_if_block): Initialize rev_cond.
22831 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
22832 instead of false as last argument never attempt to reverse it
22835 2017-02-23 Bin Cheng <bin.cheng@arm.com>
22837 PR tree-optimization/79663
22838 * tree-predcom.c (combine_chains): Process refs in reverse order
22839 only for ZERO length chains, and add explaining comment.
22841 2017-02-23 Jeff Law <law@redhat.com>
22843 PR tree-optimization/79578
22844 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
22845 in call to operand_equal_p.
22847 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
22850 * config/i386/cpuid.h: Fix another undefined behavior.
22852 2017-02-23 Richard Biener <rguenther@suse.de>
22854 PR tree-optimization/79683
22855 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
22856 vector types for data-refs.
22858 2017-02-23 Martin Liska <mliska@suse.cz>
22860 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
22862 2017-02-23 Jakub Jelinek <jakub@redhat.com>
22864 PR middle-end/79665
22865 * internal-fn.c (get_range_pos_neg): Moved to ...
22866 * tree.c (get_range_pos_neg): ... here. No longer static.
22867 * tree.h (get_range_pos_neg): New prototype.
22868 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
22869 are known to be in between 0 and signed maximum inclusive, try to
22870 expand both unsigned and signed divmod and use the cheaper one from
22873 2017-02-22 Jeff Law <law@redhat.com>
22875 PR tree-optimization/79578
22876 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
22877 to compare base operands.
22879 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
22882 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
22883 gpc_reg_operand instead of fpr_reg_operand.
22885 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
22887 * config/mips/mips.c (mips_return_in_memory): Force FP
22888 vector types to be returned in memory for o32 ABI.
22890 2017-02-22 Jakub Jelinek <jakub@redhat.com>
22892 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
22893 instead of DW_TAG_member for static data member declarations and don't
22894 set no_linkage_name for static inline data members.
22895 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
22898 2017-02-22 Martin Liska <mliska@suse.cz>
22900 * doc/invoke.texi: Replace inequality signs with square brackets
22903 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22905 PR tree-optimization/68644
22906 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
22908 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
22911 * lra-constraints.c (simplify_operand_subreg): Handle
22912 WORD_REGISTER_OPERATIONS targets.
22914 2017-02-22 Jakub Jelinek <jakub@redhat.com>
22917 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
22918 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
22919 elimination by swapping fld*.
22921 2017-02-22 Richard Biener <rguenther@suse.de>
22923 PR tree-optimization/79673
22924 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
22925 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
22926 irrelevant address-space qualifiers and avoiding a
22927 ADDR_SPACE_CONVERT_EXPR from fold_convert.
22929 2017-02-22 Richard Biener <rguenther@suse.de>
22931 PR tree-optimization/79666
22932 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
22933 to not symbolically negate if that may introduce undefined
22936 2017-02-22 Martin Liska <mliska@suse.cz>
22939 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
22940 * data-streamer-out.c (streamer_write_gcov_count_stream):
22942 * value-prof.c (stream_out_histogram_value): Make assert more
22943 precise based on type of counter.
22945 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
22948 * config/i386/i386.md (standard_x87sse_constant_load splitter):
22949 Use nonimmediate_operand instead of memory_operand for operand 1.
22950 (float-extend standard_x87sse_constant_load splitter): Ditto.
22952 2017-02-21 Jeff Law <law@redhat.com>
22954 PR tree-optimization/79621
22955 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
22956 blocks with edges to themselves.
22958 2017-02-21 Jakub Jelinek <jakub@redhat.com>
22961 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
22962 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
22963 Use gimple_call_builtin_p.
22966 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
22967 on temporarily removed DEBUG_INSNs.
22969 PR tree-optimization/79649
22970 * tree-loop-distribution.c (classify_partition): Give up on
22971 non-generic address space loads/stores.
22973 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
22975 * doc/loop.texi (Loop manipulation): Remove nonexistent
22976 tree_ssa_loop_version from the documentation.
22977 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
22979 2017-02-21 Jakub Jelinek <jakub@redhat.com>
22982 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
22983 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
22984 * config/rs6000/rs6000.c: Include except.h.
22985 (rs6000_expand_split_stack_prologue): Call
22986 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
22988 2017-02-21 Martin Jambor <mjambor@suse.cz>
22991 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
22992 have been analyzed.
22994 2017-02-21 Martin Jambor <mjambor@suse.cz>
22996 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
22997 for backward compatibility only.
22998 * doc/invoke.texi (Option Summary): Remove all references to
22999 -fipa-cp-alignment.
23001 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
23005 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
23007 * lra-constraints.c (curr_insn_transform): Handle
23008 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
23010 2017-02-21 Martin Liska <mliska@suse.cz>
23012 * config/i386/i386.opt: Replace -masm-dialect with -masm.
23014 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
23016 PR translation/79638
23017 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
23019 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
23022 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
23023 (arm_function_ok_for_sibcall): Return false for an indirect call by
23024 descriptor if all the argument registers are used.
23025 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
23026 alignment of the function.
23028 2017-02-21 Jakub Jelinek <jakub@redhat.com>
23030 PR tree-optimization/61441
23031 * simplify-rtx.c (simplify_const_unary_operation): For
23032 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
23033 the sNaN unmodified.
23035 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
23037 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
23038 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
23039 instead of SYSTEM_HEADER_DIR.
23041 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
23042 Martin Liška <mliska@suse.cz>
23044 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
23045 Fix typos and grammar, use active voice, and clarify.
23047 2017-02-20 Marek Polacek <polacek@redhat.com>
23049 PR middle-end/79537
23050 * gimplify.c (gimplify_expr): Handle unused *&&L;.
23053 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
23055 2017-02-20 Jakub Jelinek <jakub@redhat.com>
23058 * config/i386/i386.c (ix86_expand_builtin): Handle
23059 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
23060 ix86_builtins_isa[fcode].isa as a requirement of those
23061 flags and any other flag in the bitmask.
23062 (ix86_init_mmx_sse_builtins): Use 0 instead of
23063 ~OPTION_MASK_ISA_64BIT as mask.
23064 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
23065 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
23066 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
23067 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
23069 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
23072 * lra-constraints.c (split_reg): Check requested split mode
23073 is supported by the register.
23075 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
23077 * lra-constraints.c (simplify_operand_subreg): Remove early
23080 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
23083 * lra-constraints.c (curr_insn_transform): Tighten condition
23084 for converting SUBREG reloads from OP_OUT to OP_INOUT.
23086 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
23089 * lra-constraints.c (curr_insn_transform): Handle
23090 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
23092 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
23095 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
23097 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
23099 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
23102 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
23105 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23107 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
23110 2017-02-18 Jakub Jelinek <jakub@redhat.com>
23112 * final.c (last_columnnum, override_columnnum): New variables.
23113 (final_start_function): Set last_columnnum, pass it to begin_prologue
23114 hook and pass 0 to dwarf2out_begin_prologue.
23115 (final_scan_insn): Update override_columnnum. Pass last_columnnum
23116 to source_line debug hook.
23117 (notice_source_line): Compute last_columnnum and for debug_column_info
23118 return true on column changes.
23119 * debug.h (struct gcc_debug_hooks): Add column argument to
23120 source_line and begin_prologue hooks.
23121 (debug_nothing_int_charstar_int_bool): Remove prototype.
23122 (debug_nothing_int_int_charstar,
23123 debug_nothing_int_int_charstar_int_bool): New prototypes.
23124 (dwarf2out_begin_prologue): Add column argument.
23125 * debug.c (do_nothing_debug_hooks): Adjust source_line and
23126 begin_prologue hooks.
23127 (debug_nothing_int_charstar_int_bool): Remove.
23128 (debug_nothing_int_int_charstar,
23129 debug_nothing_int_int_charstar_int_bool): New functions.
23130 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
23131 through to dwarf2out_source_line.
23132 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
23133 (dwarf2out_source_line): Add column argument, emit it if requested.
23134 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
23136 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
23137 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
23138 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
23139 through to dwarf2out_begin_prologue.
23140 (vmsdbgout_source_line): Add column argument, pass it through to
23141 dwarf2out_source_line.
23142 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
23143 dbxout_source_line caller.
23144 (dbxout_source_line): Add column argument.
23146 * common.opt (gno-column-info, gcolumn-info): New options.
23147 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
23148 (check_die): Also test for multiple DW_AT_decl_column attributes.
23149 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
23150 DW_AT_decl_column if requested.
23151 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
23153 (gen_variable_die): Likewise.
23154 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
23155 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
23158 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
23159 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
23160 (ix86_handle_option): Handle OPT_m3dnowa.
23161 * doc/invoke.texi (-m3dnowa): Document.
23162 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
23163 -m3dnowa instead of -m3dnow -march=athlon.
23166 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
23167 instead of gcc_assert for K, r and R code checks. Formatting fixes.
23169 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23172 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
23173 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
23174 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
23175 generator for vsx_xxpermdi_<mode>_be.
23176 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
23177 force big-endian semantics.
23178 (vsx_xxpermdi_<mode>_be): New define_expand with same
23179 implementation as previous version of vsx_xxpermdi_<mode>.
23181 2017-02-17 Jakub Jelinek <jakub@redhat.com>
23183 PR tree-optimization/79327
23184 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
23185 variable, its initialization and use.
23187 2017-02-17 Julia Koval <julia.koval@intel.com>
23189 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
23190 (OPTION_MASK_ISA_PKU_UNSET): New.
23191 (ix86_handle_option): Handle -mrdpid.
23192 * config/i386/cpuid.h (bit_RDPID): New.
23193 * config/i386/driver-i386.c (host_detect_local_cpu):
23194 Detect RDPID feature.
23195 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
23196 * config/i386/i386-c.c (ix86_target_macros_internal):
23198 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
23199 (ix86_valid_target_attribute_inner_p): Add "rdpid".
23200 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
23201 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
23202 * config/i386/i386.md (define_insn "rdpid"): New.
23203 * config/i386/i386.opt Add -mrdpid.
23204 * config/i386/immintrin.h (_rdpid_u32): New.
23206 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
23208 PR rtl-optimization/79541
23209 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
23210 instead of transforming it into USE.
23212 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
23214 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
23215 If HONOR_SNANS (SFmode) force the input to a register.
23216 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
23217 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
23218 an frsp or similar insn.
23220 2017-02-17 Martin Liska <mliska@suse.cz>
23222 PR rtl-optimization/79577
23223 * params.def (selsched-max-sched-times): Increase minimum to 1.
23225 2017-02-17 Martin Liska <mliska@suse.cz>
23227 PR rtl-optimization/79574
23228 * gcse.c (want_to_gcse_p): Prevent integer overflow.
23230 2017-02-17 Martin Liska <mliska@suse.cz>
23232 PR tree-optimization/79529
23233 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
23234 ssa_defined_default_def_p to handle cases which are implicitly
23236 * tree-ssa.c (ssa_defined_default_def_p): New function.
23237 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
23238 which are implicitly defined.
23239 * tree-ssa.h (ssa_defined_default_def_p): Declare.
23241 2017-02-17 Richard Biener <rguenther@suse.de>
23243 PR middle-end/79576
23244 * params.def (max-ssa-name-query-depth): Limit to 10.
23246 2017-02-17 Richard Biener <rguenther@suse.de>
23248 PR tree-optimization/79552
23249 * tree-ssa-structalias.c (visit_loadstore): Properly verify
23252 2017-02-17 Richard Biener <rguenther@suse.de>
23255 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
23257 2017-02-17 Marek Polacek <polacek@redhat.com>
23259 PR middle-end/79536
23260 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
23261 (fold_negate_expr): New wrapper.
23263 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
23265 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
23266 Correct terminology and de-emphasize pre-standard behavior.
23268 2017-02-16 Alan Modra <amodra@gmail.com>
23270 PR rtl-optimization/79286
23271 * ira.c (def_dominates_uses): New function.
23272 (update_equiv_regs): Don't create an equivalence for insns that
23273 may trap where the register def does not dominate the use.
23275 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
23277 PR rtl-optimization/78127
23278 * lra.c (lra): Call lra_eliminate before finish the loop after
23281 2017-02-16 Richard Biener <rguenther@suse.de>
23283 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
23285 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
23286 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
23287 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
23288 (isl_val_int_from_wi): New function.
23289 (extract_affine_gmp): Rename to ...
23290 (extract_affine_wi): ... this, take a widest_int.
23291 (extract_affine_int): Just wrap extract_affine_wi.
23292 (add_param_constraints): Use isl_val_int_from_wi.
23293 (add_loop_constraints): Likewise, and extract_affine_wi.
23295 2017-02-15 Jeff Law <law@redhat.com>
23297 PR middle-end/79521
23298 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
23299 ira_init_register_move_cost_if_necessary.
23301 2017-02-15 Martin Sebor <msebor@redhat.com>
23303 PR middle-end/32003
23304 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
23305 removed in a prior commit.
23307 2017-02-15 Bin Cheng <bin.cheng@arm.com>
23309 PR tree-optimization/79347
23310 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
23311 counters during peeling.
23313 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
23315 * Makefile.in (site.exp): Remove "set ISLVER".
23317 2017-02-15 Jakub Jelinek <jakub@redhat.com>
23320 * real.c (real_from_integer): Call real_convert even for decimal.
23322 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
23325 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
23327 2017-02-14 Andrew Pinski <apinski@cavium.com>
23329 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
23330 cores and change the partno/implementer to be correct.
23331 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
23332 the 'B" as the implementer.
23333 * config/aarch64/aarch64-tune.md: Regenerate.
23335 2017-02-14 Carl Love <cel@us.ibm.com>
23337 * config/rs6000/rs6000.c: Add case statement entry to make the
23338 xvcvuxdsp built-in argument unsigned.
23339 * config/rs6000/vsx.md: Fix the source and return operand types so they
23340 match the instruction definitions from the ISA document. Fix typo
23341 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
23344 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
23347 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
23348 member early_clobber_alts.
23349 * lra-lives.c (reg_early_clobber_p): New.
23350 (process_bb_lives): Use it.
23351 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
23352 (debug_operand_data): Initialize early_clobber_alts.
23353 (setup_operand_alternative): Set up early_clobber_alts.
23354 (collect_non_operand_hard_regs): Ditto. Pass early clobber
23355 alternatives to new_insn_reg.
23356 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
23358 (lra_update_insn_regno_info): Pass the new arg.
23360 2017-02-14 Jakub Jelinek <jakub@redhat.com>
23362 PR middle-end/79505
23363 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
23364 (new_oacc_loop_raw): Don't clear already cleared fields.
23367 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
23368 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
23369 _mm512_prefetch_i64gather_ps): New inline functions and macros.
23371 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
23374 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
23376 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
23379 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
23380 the extra instruction to the right place to store 128-bit constant
23383 2017-02-14 Martin Sebor <msebor@redhat.com>
23385 PR middle-end/79448
23386 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
23387 warning for strings of unknown length.
23389 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
23391 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
23393 2017-02-14 Jeff Law <law@redhat.com>
23396 * ira-costs.c (scan_one_insn): Initialize register move costs
23397 for pseudos seen in USE/CLOBBER insns.
23399 PR tree-optimization/79095
23400 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
23401 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
23402 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
23403 if the operands are known to be not equal, then the resulting range
23405 (intersect_ranges): If the new range is ~[0,0] and the old range is
23406 wide, then prefer ~[0,0].
23407 * tree-vrp.c (overflow_comparison_p_1): New function.
23408 (overflow_comparison_p): New function.
23409 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
23410 if NAME is used in an overflow test.
23411 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
23412 overflow check that can be expressed as an equality test, then adjust
23413 ops to be that equality test.
23415 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23417 * config/s390/s390-builtin-types.def: Remove flags argument.
23418 * config/s390/s390.c (s390_init_builtins): Likewise.
23420 2017-02-14 Martin Liska <mliska@suse.cz>
23422 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
23423 vector. Fix trailing white spaces.
23425 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
23427 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
23430 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23432 PR rtl-optimization/68664
23433 * config/arm/arm.c (arm_sched_can_speculate_insn):
23434 New function. Declare prototype.
23435 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
23437 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23439 PR rtl-optimization/68664
23440 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
23442 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
23444 2017-02-14 Amit Pawar <amit.pawar@amd.com>
23446 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
23447 max skip bytes for function, loop and jump.
23449 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23451 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
23452 ABS_EXPR for gimple dump.
23454 2017-02-14 Jakub Jelinek <jakub@redhat.com>
23457 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
23459 PR tree-optimization/79408
23460 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
23461 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
23462 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
23463 also if rhs1 is INTEGER_CST.
23465 2017-02-14 Richard Biener <rguenther@suse.de>
23467 PR middle-end/79432
23468 * tree-into-ssa.c (insert_phi_nodes): When the function can
23469 have abnormal edges rewrite SSA names with broken use-def
23470 dominance out of SSA and register them for PHI insertion.
23472 2017-02-13 Martin Sebor <msebor@redhat.com>
23474 PR middle-end/79496
23475 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
23476 clearing info.nowrite flag when snprintf size argument is a range.
23478 2017-02-13 Jakub Jelinek <jakub@redhat.com>
23480 * cprop.c (cprop_jump): Add missing space in string literal.
23481 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
23482 (get_constraint_for_component_ref): Likewise.
23483 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
23484 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
23485 * lra-constraints.c (process_alt_operands): Likewise.
23486 * ipa-inline.c (inline_small_functions): Likewise.
23487 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
23488 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
23489 * trans-mem.c (diagnose_tm_1_op): Likewise.
23490 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
23491 (grid_parallel_clauses_gridifiable): Likewise.
23493 * config/nvptx/mkoffload.c (process): Add space in between
23496 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
23497 "MOD4_SSE_REGS" and "ALL_REGS".
23499 * spellcheck.c (test_data): Add , in between "foo" and "food".
23501 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23504 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
23505 boundary crossing check and subsequent code generation agree.
23507 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23509 * config/aarch64/aarch64.c (has_memory_op): Delete.
23510 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
23513 2017-02-13 Jakub Jelinek <jakub@redhat.com>
23515 PR rtl-optimization/79388
23516 PR rtl-optimization/79450
23517 * combine.c (distribute_notes): When removing TEM_INSN for which
23518 corresponding dest has last value recorded, invalidate that last
23521 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23523 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
23524 of explicit '@'. Add missing assembly comment marker on branch costs
23527 2017-02-13 Nathan Sidwell <nathan@acm.org>
23529 * gengtype-lex.l (<in_struct>): Add '/'.
23531 2017-02-13 Martin Liska <mliska@suse.cz>
23534 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
23536 2017-02-13 Richard Biener <rguenther@suse.de>
23538 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
23540 * configure: Re-generate.
23541 * config.in: Likewise.
23542 * graphite-dependences.c: Simplify as if
23543 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
23544 * graphite-isl-ast-to-gimple.c: Likewise.
23545 * graphite-optimize-isl.c: Likewise.
23546 * graphite-poly.c: Likewise.
23547 * graphite-sese-to-poly.c: Likewise.
23548 * graphite.h: Likewise.
23549 * toplev.c: Include isl/version.h and use isl_version () for
23550 printing the ISL version.
23551 * doc/install.texi: Update ISL requirement.
23553 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
23555 * doc/standards.texi (Standards): Update reference to
23558 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
23560 * doc/extend.texi (Named Address Spaces): sourceware.org now
23562 * doc/install.texi (Binaries): Ditto.
23565 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
23567 * doc/cpp.texi: Replace "stringify"/"stringification" with C
23568 standard terminology "stringize"/"stringizing" throughout.
23569 * doc/cppinternals.texi: Likewise.
23571 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
23573 * doc/extend.texi: Fix some spelling mistakes and typos.
23574 * doc/invoke.texi: Likewise.
23576 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
23579 * params.def (inline-min-speedup) Change from 10 to 8.
23581 2017-02-11 Jakub Jelinek <jakub@redhat.com>
23583 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
23586 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
23589 * ipa-inline-analysis.c (get_minimal_bb): New function.
23590 (record_modified): Use it.
23591 (remap_edge_change_prob): Handle also ancestor functions.
23593 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
23595 * doc/contrib.texi (Contributors): Remove broken link into
23596 the Mauve CVS repository.
23598 2017-02-11 Jakub Jelinek <jakub@redhat.com>
23600 PR middle-end/79454
23601 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
23602 result computation whenever lhs doesn't have vector mode, not
23603 just when it has BLKmode.
23605 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
23607 * doc/makefile.texi (profiledbootstrap): Refer to the
23608 installation instructions only in textual form.
23610 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23613 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
23615 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
23617 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
23618 (Specific): Update mingw-w64 reference.
23620 (Specific): Remove broken link to Renesas RX processor.
23622 2017-02-10 Richard Biener <rguenther@suse.de>
23624 * toplev.c (process_options): Do not mention obsolete graphite
23625 options when printing sorry message about missing graphite support.
23626 Mention -floop-nest-optimize.
23628 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
23630 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
23631 (vtst_p16): Likewise.
23632 (vtstq_p8): Likewise.
23633 (vtstq_p16): Likewise.
23635 (vtstq_p64): Likewise.
23636 * config/arm/arm_neon.h (vgetq_lane_p64): New.
23637 (vset_lane_p64): New.
23638 (vsetq_lane_p64): New.
23640 2017-02-10 Jakub Jelinek <jakub@redhat.com>
23642 PR tree-optimization/79411
23643 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
23644 stmt operands are SSA_NAMEs used in abnormal phis.
23645 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
23648 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
23651 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
23654 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
23656 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
23658 2017-02-09 Jakub Jelinek <jakub@redhat.com>
23660 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
23664 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
23665 not arbitrary TREE_CONSTANT.
23668 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
23669 "omp declare target link" attribute unless is_global_var.
23670 * omp-offload.c (find_link_var_op): Likewise.
23672 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
23673 Chung-Lin Tang <cltang@codesourcery.com>
23675 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
23677 (gimplify_adjust_omp_clauses): Don't delete TILE.
23678 (gimplify_omp_for): Deal with TILE.
23679 * internal-fn.c (expand_GOACC_TILE): New function.
23680 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
23682 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
23683 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
23685 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
23686 avoid DIV for outermost collapse var.
23687 (expand_oacc_for): Insert tile element loop as needed. Adjust.
23688 Remove out of date comments, fix whitespace.
23689 * omp-general.c (omp_extract_for_data): Deal with tiling.
23690 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
23691 adjust OLF_DIM_BASE value.
23692 (struct omp_for_data): Add tiling field.
23693 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
23694 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
23695 for auto loops. Remove default auto determining, moved to
23696 oacc_loop_fixed_partitions.
23697 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
23698 stmts, add e_mask field.
23699 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
23700 (oacc_thread_numbers): Use oacc_dim_call.
23701 (oacc_xform_tile): New.
23702 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
23703 (finish_oacc_loop): Adjust for ifns vector.
23704 (oacc_loop_discover_walk): Append loop abstraction sites to list,
23705 add case for GOACC_TILE fns.
23706 (oacc_loop_xform_loop): Delete.
23707 (oacc_loop_process): Iterate over call list directly, and add
23708 handling for GOACC_TILE fns.
23709 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
23711 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
23712 vector partitioning to outer loops. Assign 2 partitions to loops
23713 when available. Add TILE handling.
23714 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
23715 (execite_oacc_device_lower): Process GOACC_TILE fns,
23716 ignore unknown specs.
23717 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
23718 * tree.c (omp_clause_num_ops): Adjust TILE ops.
23719 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
23721 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
23723 * configure.ac (ACX_BUGURL): Update.
23724 * configure: Regenerate.
23726 2017-02-09 Richard Biener <rguenther@suse.de>
23728 PR tree-optimization/69823
23729 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
23730 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
23732 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
23734 * config/arc/arc-c.def: Add __NPS400__ definition.
23735 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
23736 (TARGET_NPS400): Define.
23738 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
23740 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
23742 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
23743 pointer, arch_info.
23744 (arc_cpu_types): Fill the arch_info field with a pointer into the
23745 arc_arch_types table.
23746 (arc_selected_cpu): Declare.
23747 * config/arc/arc.c (arc_selected_cpu): Make global.
23748 (arc_selected_arch): Delete.
23749 (arc_base_cpu): Delete.
23750 (arc_override_options): Remove references to deleted variables,
23751 update access to arch information.
23752 (ARC_OPT): Update access to arch information.
23753 (ARC_OPTX): Likewise.
23754 * config/arc/arc.h (arc_base_cpu): Remove declaration.
23755 (TARGET_ARC600): Update access to arch information.
23756 (TARGET_ARC601): Likewise.
23757 (TARGET_ARC700): Likewise.
23758 (TARGET_EM): Likewise.
23759 (TARGET_HS): Likewise.
23760 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
23763 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
23766 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
23767 condition/operands for integer GE/LE/GEU/LEU operations.
23769 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
23771 PR translation/79397
23772 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
23775 2017-02-08 Martin Jambor <mjambor@suse.cz>
23778 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
23779 whether allocation happened.
23780 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
23781 nothing was allocated.
23783 2017-02-08 Jakub Jelinek <jakub@redhat.com>
23785 PR tree-optimization/79408
23786 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
23787 constant, but SSA_NAME with a known integer range, use the minimum
23788 of that range instead of op1 to determine if modulo can be replaced
23789 with its first operand.
23791 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23793 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
23795 2017-02-08 Richard Biener <rguenther@suse.de>
23797 PR tree-optimization/71824
23798 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
23799 Check all loops contained in the merged region.
23801 2017-02-07 Andrew Pinski <apinski@cavium.com>
23803 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
23805 2017-02-07 Andrew Pinski <apinski@cavium.com>
23807 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
23808 (thunderxt88): Likewise.
23809 (thunderxt81): Disable LSE and change v8.1 to v8.
23810 (thunderxt83): Likewise.
23812 2017-02-07 Jakub Jelinek <jakub@redhat.com>
23813 Richard Biener <rguenther@suse.de>
23815 PR middle-end/79399
23816 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
23817 type from int to size_t.
23818 * ira-costs.c (struct_costs_size): Change type from int to size_t.
23820 2017-02-07 Jakub Jelinek <jakub@redhat.com>
23822 PR rtl-optimization/79386
23823 * cprop.c (bypass_conditional_jumps): Initialize
23824 bypass_last_basic_block already before splitting bbs after
23825 unconditional traps...
23826 (bypass_conditional_jumps): ... rather than here.
23829 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
23830 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
23831 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
23832 fix -masm=intel patterns.
23834 2017-02-07 Richard Biener <rguenther@suse.de>
23836 PR tree-optimization/79256
23837 PR middle-end/79278
23838 * builtins.c (get_object_alignment_2): Use min_align_of_type
23839 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
23840 and ADJUST_FIELD_ALIGN.
23842 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
23844 * doc/tm.texi: Regenerate.
23845 * stor-layout.c (layout_decl): Adjust.
23846 (update_alignment_for_field): Likewise.
23847 (place_field): Likewise.
23848 (min_align_of_type): Likewise.
23849 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
23850 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
23851 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
23852 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
23853 * config/frv/frv.c (frv_adjust_field_align): Likewise.
23854 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
23855 * config/i386/i386.c (x86_field_alignment): Likewise.
23856 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
23857 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
23858 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
23859 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
23860 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
23861 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
23865 2017-01-30 Richard Biener <rguenther@suse.de>
23867 PR tree-optimization/79256
23868 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
23869 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
23872 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
23874 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
23875 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
23876 builtins to SImode and emit a zero-extend, if necessary.
23878 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
23880 * docs/invoke.texi (RISC-V Options): Alphabetize.
23882 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
23884 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
23887 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
23889 * config/riscv/riscv.c: New file.
23890 * common/config/riscv/riscv-common.c: Likewise.
23891 * config.gcc: Likewise.
23892 * config/riscv/constraints.md: Likewise.
23893 * config/riscv/elf.h: Likewise.
23894 * config/riscv/generic.md: Likewise.
23895 * config/riscv/linux.h: Likewise.
23896 * config/riscv/multilib-generator: Likewise.
23897 * config/riscv/peephole.md: Likewise.
23898 * config/riscv/pic.md: Likewise.
23899 * config/riscv/predicates.md: Likewise.
23900 * config/riscv/riscv-builtins.c: Likewise.
23901 * config/riscv/riscv-c.c: Likewise.
23902 * config/riscv/riscv-ftypes.def: Likewise.
23903 * config/riscv/riscv-modes.def: Likewise.
23904 * config/riscv/riscv-opts.h: Likewise.
23905 * config/riscv/riscv-protos.h: Likewise.
23906 * config/riscv/riscv.h: Likewise.
23907 * config/riscv/riscv.md: Likewise.
23908 * config/riscv/riscv.opt: Likewise.
23909 * config/riscv/sync.md: Likewise.
23910 * config/riscv/t-elf-multilib: Likewise.
23911 * config/riscv/t-linux: Likewise.
23912 * config/riscv/t-linux-multilib: Likewise.
23913 * config/riscv/t-riscv: Likewise.
23914 * configure.ac: Likewise.
23915 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
23916 Waterman as RISC-V maintainers.
23917 * doc/install.texi: Add RISC-V entries.
23918 * doc/invoke.texi: Add RISC-V options section.
23919 * doc/md.texi: Add RISC-V constraints section.
23920 * configure: Regenerated.
23922 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
23925 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
23926 false values to be constant vectors with all 0 or all 1 bits set.
23927 (vcondu<mode><mode>): Likewise.
23928 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
23930 (fpmask_comparison_operator): Update comment.
23931 (vecint_comparison_operator): New predicate.
23932 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
23933 vector conditionals when the true and false values are constant
23934 vectors with all 0 bits or all 1 bits set.
23936 2017-02-06 Martin Sebor <msebor@redhat.com>
23938 PR tree-optimization/79376
23939 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
23941 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
23943 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
23944 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
23945 to simplify split condition.
23947 2017-02-06 Jakub Jelinek <jakub@redhat.com>
23949 * omp-expand.c (oxpand_omp_atomic_fetch_op,
23950 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
23953 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
23955 PR rtl-optimization/68664
23956 * target.def (can_speculate_insn): New hook.
23957 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
23958 * doc/tm.texi: Regenerate.
23959 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
23960 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
23961 (rs6000_sched_can_speculate_insn): New function.
23963 2017-02-06 Jakub Jelinek <jakub@redhat.com>
23965 PR tree-optimization/79284
23966 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
23967 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
23968 vectorizable_mask_load_store, vectorizable_operation,
23969 vect_is_simple_cond, get_same_sized_vectype): Use it instead
23970 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
23971 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
23972 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
23973 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
23974 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
23975 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
23976 is_gimple_assign (stmt). Replace another such test with
23977 is_gimple_assign (stmt).
23979 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
23982 * config/avr/avr.c (rtl-iter.h): Include it.
23983 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
23984 (avr_legitimate_combined_insn): ...and implementation.
23986 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
23988 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
23989 * config/s390/s390.c (s390_const_operand_ok)
23990 (s390_canonicalize_comparison, s390_extract_part)
23991 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
23992 (s390_contiguous_bitmask_p, s390_rtx_costs)
23993 (legitimize_pic_address): Likewise.
23994 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
23995 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
23996 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
23997 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
23998 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
24000 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
24002 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
24003 REGNO($0) == REGNO($1).
24005 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24007 * config/s390/linux.h(SIZE_TYPE): Add comment.
24009 2017-02-06 Julian Brown <julian@codesourcery.com>
24010 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24011 Virendra Pathak <virendra.pathak@broadcom.com>
24013 * config/aarch64/aarch64-cores.def: Change the scheduler
24015 * config/aarch64/aarch64.md: Include thunderx2t99.md.
24016 * config/aarch64/thunderx2t99.md: New file.
24018 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
24020 * doc/standards.texi (Go Language): Update link to language
24023 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
24025 * tree-eh.c (lower_resx): Sanitize profile.
24026 (cleanup_empty_eh_move_lp): Likewise.
24028 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
24031 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
24033 * cfgloopmanip.h (loop_version): Update prototype.
24034 * modulo-sched.c (sms_schedule): Update call of loop_version.
24035 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
24036 * tree-parloops.c (gen_parallel_loop): Likewise.
24037 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
24038 * tree-ssa-loop-split.c (split_loop): Likewise.
24039 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
24040 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
24042 2017-02-05 Martin Liska <mliska@suse.cz>
24045 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
24047 (print_operand_address): Initialize a struct to zero.
24049 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
24051 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
24052 garbage collector only in textual form.
24054 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
24056 * doc/extend.texi (x86 specific memory model extensions for
24057 transactional memory): Simplify a phrase.
24059 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
24062 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
24063 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
24064 (atomic_storedi_1): Likewise.
24066 2017-02-04 Jakub Jelinek <jakub@redhat.com>
24068 PR tree-optimization/79338
24069 * tree-parloops.c (gather_scalar_reductions): Don't call
24070 vect_analyze_loop_form for loop->inner before destroying loop's
24073 2017-02-03 Martin Sebor <msebor@redhat.com>
24075 PR tree-optimization/79327
24076 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
24077 when precision has resulted in leading zeros.
24078 (format_integer): Adjust the likely counter to assume an unknown
24079 argument that may be zero is non-zero.
24081 2017-02-03 Jason Merrill <jason@redhat.com>
24084 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
24085 avoid copying non-taken branch.
24087 2017-02-03 Jakub Jelinek <jakub@redhat.com>
24089 PR tree-optimization/79340
24090 * tree-vect-loop.c (vectorizable_reduction): Release
24091 vec_defs elements after safe_splicing them into other vectors.
24094 PR tree-optimization/79327
24095 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
24096 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
24098 (format_integer): Use wide_int_to_tree instead of build_int_cst
24099 + to_?hwi. If argmin is NULL, just set argmin and argmax to
24100 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
24101 of shortest and longest sequence.
24103 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
24105 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
24106 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
24108 2017-02-03 Walter Lee <walt@tilera.com>
24111 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
24112 after initial stackframe link reg save.
24113 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
24115 2017-02-03 Jakub Jelinek <jakub@redhat.com>
24118 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
24119 wu for stxssp alternative.
24121 2017-02-03 Martin Sebor <msebor@redhat.com>
24123 PR tree-optimization/79352
24124 * gimple-fold.c (get_range_strlen): Add argument.
24125 (get_range_strlen): Change return type to bool.
24126 (get_maxval_strlen): Pass in a dummy argument.
24127 * gimple-fold.h (get_range_strlen): Change return type to bool.
24128 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
24129 * tree.h (array_at_struct_end_p): Add argument.
24130 * tree.c (array_at_struct_end_p): Handle it.
24132 2017-02-03 Martin Liska <mliska@suse.cz>
24135 * multiple_target.c (create_dispatcher_calls): Redirect edge
24136 from a caller of a dispatcher.
24137 (expand_target_clones): Make the clones local.
24138 (ipa_target_clone): Do both target clones and resolvers.
24139 (ipa_dispatcher_calls): Remove the pass.
24140 (pass_dispatcher_calls::gate): Likewise.
24141 (make_pass_dispatcher_calls): Likewise.
24142 * passes.def (pass_target_clone): Put as very first IPA early
24145 2017-02-03 Martin Liska <mliska@suse.cz>
24147 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
24148 in case of a function with ifunc attribute.
24150 2017-02-03 Martin Liska <mliska@suse.cz>
24152 * cgraph.c (cgraph_node::dump): Dump function version info.
24153 * symtab.c (symtab_node::dump_base): Add missing new line.
24155 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
24157 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
24158 (ifcombine_ifandif): Use it.
24160 2017-02-03 Martin Liska <mliska@suse.cz>
24162 * doc/invoke.texi: Document default value for
24163 use-after-scope-direct-emission-threshold.
24165 2017-02-03 Martin Liska <mliska@suse.cz>
24167 PR tree-optimization/79339
24168 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
24169 (format_floating): Likewise.
24171 2017-02-03 Martin Liska <mliska@suse.cz>
24174 * ipa-prop.c (ipa_node_params_t::insert): Remove current
24176 (ipa_node_params_t::remove): Likewise.
24177 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
24178 initialization from removed ipa_node_params_t::insert.
24179 (ipa_node_params::~ipa_node_params): Move from removed
24180 ipa_node_params_t::release.
24181 * symbol-summary.h (symbol_summary::m_released): New member.
24182 Do not release a summary twice. Do not allow to call finalizer
24183 for types of a summary that live in GGC memory.
24185 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24187 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
24190 2017-02-02 Martin Sebor <msebor@redhat.com>
24192 PR middle-end/79275
24193 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
24194 (format_string): Tighten up the range of output for non-constant
24195 strings and correct the expected range for wide non-constant strings.
24197 2017-02-02 Martin Sebor <msebor@redhat.com>
24199 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
24201 PR middle-end/32003
24202 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
24204 (-fdump-tree-@var): Add to index and document how to come up
24205 with pass-specific option and dump file names.
24206 (-fdump-passes): Clarify where to look for output.
24208 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
24210 PR middle-end/77445
24211 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
24212 statistics of the analyzed path; allow threading for speed when
24213 any of BBs along the path are optimized for speed.
24215 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
24217 PR middle-end/78468
24218 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
24219 settings of the virtual registers.
24222 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
24224 * explow.c (get_dynamic_stack_size): Take known alignment of stack
24225 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
24228 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24230 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
24231 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
24233 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24235 * config/s390/s390.md: Add missing comments with the expanded
24237 * config/s390/vector.md: Likewise.
24238 * config/s390/vx-builtins.md: Likewise.
24240 2017-02-02 Jakub Jelinek <jakub@redhat.com>
24243 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
24244 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
24245 conditions on a single line.
24247 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24249 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
24250 __S390_VX__ to __VX__.
24252 2017-02-01 Andrew Pinski <apinski@cavium.com>
24254 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
24255 stmt_info to record_stmt_cost.
24256 (vect_get_known_peeling_cost): Pass stmt_info if known to
24258 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
24259 cpu_vector_cost field into
24260 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
24261 field into vec_int_stmt_cost and vec_fp_stmt_cost.
24262 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
24263 splitting of scalar_stmt_cost and vec_stmt_cost.
24264 (thunderx_vector_cost): Likewise.
24265 (cortexa57_vector_cost): LIkewise.
24266 (exynosm1_vector_cost): Likewise.
24267 (xgene1_vector_cost): Likewise.
24268 (thunderx2t99_vector_cost): Improve after the splitting of the two
24270 (aarch64_builtin_vectorization_cost): Update for the splitting of
24271 scalar_stmt_cost and vec_stmt_cost.
24273 2017-02-01 Torvald Riegel <triegel@redhat.com>
24274 Richard Henderson <rth@redhat.com>
24276 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
24277 conditional on existance of a fast atomic load.
24278 * optabs-query.c (can_atomic_load_p): New function.
24279 * optabs-query.h (can_atomic_load_p): Declare it.
24280 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
24281 no fast atomic load is available for the particular size of access.
24282 (expand_atomic_compare_and_swap): Likewise.
24283 (expand_atomic_load): Likewise.
24284 (expand_atomic_store): Likewise.
24285 (expand_atomic_fetch_op): Likewise.
24286 * testsuite/lib/target-supports.exp
24287 (check_effective_target_sync_int_128): Remove x86 because it provides
24288 no fast atomic load.
24289 (check_effective_target_sync_int_128_runtime): Likewise.
24291 2017-02-01 Richard Biener <rguenther@suse.de>
24293 * graphite.c: Include tree-vectorizer.h for find_loop_location.
24294 (graphite_transform_loops): Provide opt-info for optimized nests.
24295 * tree-parloop.c (parallelize_loops): Provide opt-info for
24296 parallelized loops.
24298 2017-02-01 Richard Biener <rguenther@suse.de>
24300 PR middle-end/79315
24301 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
24302 was not set before.
24304 2017-02-01 Richard Biener <rguenther@suse.de>
24306 PR tree-optimization/71824
24307 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
24308 Verify the loops are valid in the merged SESE region.
24309 (scop_detection::can_represent_loop_1): Check analyzing the
24310 evolution of the number of iterations in the region succeeds.
24312 2017-01-31 Ian Lance Taylor <iant@golang.org>
24314 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
24315 REG_ARGS_SIZE note to 32-bit push insns and call insn.
24317 2017-01-31 David Malcolm <dmalcolm@redhat.com>
24319 PR preprocessor/79210
24320 * input.c (get_substring_ranges_for_loc): Replace line_width
24321 assertion with error-handling.
24323 2017-01-31 Richard Biener <rguenther@suse.de>
24325 PR tree-optimization/77318
24326 * graphite-sese-to-poly.c (extract_affine): Fix assert.
24327 (create_pw_aff_from_tree): Take loop parameter.
24328 (add_condition_to_pbb): Pass loop of the condition to
24329 create_pw_aff_from_tree.
24331 2017-01-31 Jakub Jelinek <jakub@redhat.com>
24333 * config/s390/s390.c (s390_asan_shadow_offset): New function.
24334 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
24336 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
24340 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
24342 (convert_int_to_float128): Likewise.
24343 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
24344 (convert_int_to_float128): Likewise.
24345 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
24346 (UNSPEC_IEEE128_CONVERT): Likewise.
24347 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
24348 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
24349 Use local variables for IBM extended format.
24350 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
24351 (fix_trunc<mode>si2_fprs): Likewise.
24352 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
24353 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
24354 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
24355 to know that we can now have integers of all sizes in vector
24357 (fix<uns>_<mode>di2_hw): Likewise.
24358 (float<uns>_<mode>si2_hw): Likewise.
24359 (fix_<mode>si2_hw): Likewise.
24360 (fixuns_<mode>si2_hw): Likewise.
24361 (float<uns>_<mode>di2_hw): Likewise.
24362 (float_<mode>di2_hw): Likewise.
24363 (float_<mode>si2_hw): Likewise.
24364 (floatuns_<mode>di2_hw): Likewise.
24365 (floatuns_<mode>si2_hw): Likewise.
24366 (xscvqp<su>wz_<mode>): Delete, no longer used.
24367 (xscvqp<su>dz_<mode>): Likewise.
24368 (xscv<su>dqp_<mode>): Likewise.
24369 (ieee128_mfvsrd_64bit): Likewise.
24370 (ieee128_mfvsrd_32bit): Likewise.
24371 (ieee128_mfvsrwz): Likewise.
24372 (ieee128_mtvsrw): Likewise.
24373 (ieee128_mtvsrd_64bit): Likewise.
24374 (ieee128_mtvsrd_32bit): Likewise.
24376 2017-01-31 Martin Liska <mliska@suse.cz>
24379 * ipa-prop.c (ipa_free_all_node_params): Call release method
24380 instead of ~sumbol_summary to not to trigger double times
24383 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
24385 PR tree-optimization/71691
24386 * bitmap.h (class auto_bitmap): New.
24387 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
24388 is_maybe_undefined instead of ssa_undefined_value_p.
24390 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24392 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
24393 __S390_ARCH_LEVEL__ to __ARCH__.
24395 2017-01-31 Jakub Jelinek <jakub@redhat.com>
24397 PR tree-optimization/79267
24398 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
24399 if should_remove_lhs_p is true.
24401 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
24404 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
24405 (add_alignment_attribute): New.
24406 (base_type_die): Add alignment attribute.
24407 (subrange_type_die): Likewise.
24408 (modified_type_die): Likewise.
24409 (gen_array_type_die): Likewise.
24410 (gen_descr_array_type_die: Likewise.
24411 (gen_enumeration_type_die): Likewise.
24412 (gen_subprogram_die): Likewise.
24413 (gen_variable_die): Likewise.
24414 (gen_field_die): Likewise.
24415 (gen_ptr_to_mbr_type_die): Likewise.
24416 (gen_struct_or_union_type_die): Likewise.
24417 (gen_subroutine_type_die): Likewise.
24418 (gen_typedef_die): Likewise.
24419 (base_type_cmp): Compare alignment attribute.
24421 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24424 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
24425 (setb_unsigned) New pattern for setb with CCUNS.
24426 * config/rs6000/rs6000.c (expand_block_compare): Use a different
24427 subfc./subfe sequence to avoid overflow problems. Generate a
24428 shorter sequence with cmpld/setb for power9.
24429 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
24430 for generating subfc. instruction.
24431 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
24432 now uses this instruction.
24434 2017-01-30 Ian Lance Taylor <iant@google.com>
24437 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
24438 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
24440 2017-01-30 Martin Sebor <msebor@redhat.com>
24442 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
24443 Move constant to the right of a relational operator.
24444 (get_mpfr_format_length, format_character, format_string): Ditto.
24445 (should_warn_p, maybe_warn): Same.
24447 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
24449 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
24452 * asan.c (get_translation_unit_decl): Remove function.
24453 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
24455 2017-01-30 Martin Liska <mliska@suse.cz>
24457 PR gcov-profile/79259
24458 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
24459 -fprofile-generate.
24461 2017-01-30 Martin Liska <mliska@suse.cz>
24464 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
24465 Initialize variables with NULL value.
24467 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
24470 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
24472 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
24474 2017-01-30 Richard Biener <rguenther@suse.de>
24476 PR tree-optimization/79276
24477 * tree-vrp.c (process_assert_insertions): Properly adjust common
24478 when removing a duplicate.
24480 * gcc.dg/torture/pr79276.c: New testcase.
24482 2017-01-30 Richard Biener <rguenther@suse.de>
24484 PR tree-optimization/79256
24485 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
24486 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
24488 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
24490 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
24493 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
24494 ("*r<noxa>sbg_<mode>_sll_bitmask")
24495 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
24496 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
24497 Use contiguous_bitmask_nowrap_operand.
24499 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24502 * config/rs6000/altivec.h (vec_xl): Revise #define.
24503 (vec_xst): Likewise.
24505 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
24507 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
24509 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
24511 PR rtl-optimization/79194
24512 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
24513 traps before call to bypass_conditional_jumps.
24515 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
24517 PR tree-optimization/71374
24518 * lra-constraints.c (check_conflict_input_operands): New.
24519 (match_reload): Use it.
24521 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
24524 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
24525 account to calculate conflict_set.
24527 2017-01-27 Bin Cheng <bin.cheng@arm.com>
24529 PR rtl-optimization/78559
24530 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
24531 other_insn in combine.
24533 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
24535 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
24536 uint16_type_node for BT_UINT16.
24538 2017-01-27 David Malcolm <dmalcolm@redhat.com>
24540 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
24541 "RTL Tests" to menu.
24542 (GIMPLE Tests): New node.
24543 (RTL Tests): New node.
24545 2017-01-27 Richard Biener <rguenther@suse.de>
24547 PR tree-optimization/79245
24548 * tree-loop-distribution.c (distribute_loop): Apply cost
24549 modeling also to detected patterns.
24551 2017-01-27 Richard Biener <rguenther@suse.de>
24553 PR tree-optimization/71433
24554 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
24555 (compare_assert_loc): New function.
24556 (process_assert_insertions): Sort and optimize assert locations
24557 to remove duplicates and push down identical assertions on
24558 edges to their destination block.
24560 2017-01-27 Richard Biener <rguenther@suse.de>
24562 PR tree-optimization/79244
24563 * tree-vrp.c (remove_range_assertions): Forcefully propagate
24564 out SSA names even if abnormal.
24566 2017-01-27 Jakub Jelinek <jakub@redhat.com>
24568 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
24569 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
24570 instead of MPFR_RNDN.
24572 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
24575 * arm.c (arm_option_override): Don't call build_target_option_node
24576 until after doing all option overrides.
24577 (arm_valid_target_attribute_tree): Likewise.
24579 2017-01-27 Martin Liska <mliska@suse.cz>
24581 * doc/invoke.texi (-fprofile-arcs): Document profiling support
24582 for {cd}tors and C++ {cd}tors.
24584 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
24586 * config/s390/s390.md ("*setmem_long_and")
24587 ("*setmem_long_and_31z"): Use zero_extend instead of and.
24589 2017-01-26 Martin Sebor <msebor@redhat.com>
24591 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
24594 2017-01-26 Martin Sebor <msebor@redhat.com>
24596 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
24597 HAVE_DFmode before using XFmode or DFmode.
24598 (parse_directive): Avoid using the z length modifier to avoid
24599 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
24601 PR middle-end/78703
24602 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
24603 to accept adjustment as an array.
24604 (get_int_range): New function.
24605 (struct directive): Make width and prec arrays.
24606 (directive::set_width, directive::set_precision): Call get_int_range.
24607 (format_integer, format_floating): Handle width and precision ranges.
24608 (format_string, parse_directive): Same.
24610 2017-01-26 Jakub Jelinek <jakub@redhat.com>
24613 * dwarf2out.c (generate_skeleton_bottom_up): For children with
24614 comdat_type_p set, just clone them, but keep the children in the
24618 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
24619 which have direct callers with -fvar-tracking-assignments enabled
24621 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
24622 inside of type units.
24624 2017-01-26 Martin Sebor <msebor@redhat.com>
24626 PR middle-end/78703
24627 * gimple-ssa-sprintf.c (struct result_range): Add likely and
24629 (struct format_result): Replace number_chars, number_chars_min,
24630 and number_chars_max with a single member of struct result_range.
24632 (format_result::operator+=): Adjust.
24633 (struct fmtresult): Remove bounded. Handle likely and unlikely
24635 (fmtresult::adjust_for_width_or_precision): New function.
24636 (fmtresult:type_max_digits): New function.
24637 (bytes_remaining): Handle likely and unlikely counters.
24638 (min_bytes_remaining): Remove.
24639 (format_percent): Simplify.
24640 (format_integer, format_floating): Set likely and unlikely counters.
24641 (get_string_length, format_character, format_string): Same.
24642 (format_plain, should_warn_p): New function.
24643 (maybe_warn): Call should_warn_p. Update diagnostic messages
24644 and handle those for all directives, including plain strings.
24645 (format_directive): Handle likely and unlikely counters.
24646 Remove unnecessary quoting from diagnostics. Add an informational
24648 (add_bytes): Remove.
24649 (pass_sprintf_length::compute_format_length): Simplify.
24650 (try_substitute_return_value): Handle likely and unlikely counters.
24652 2017-01-26 Carl Love <cel@us.ibm.com>
24654 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
24655 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
24657 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
24660 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
24661 endianess for subregs into account.
24662 * lra-constraints.c (lra_constraints): Do risky transformations
24663 always on the first iteration.
24664 * lra-lives.c (check_pseudos_live_through_calls): Add arg
24665 last_call_used_reg_set.
24666 (process_bb_lives): Define and use last_call_used_reg_set.
24667 * lra.c (lra): Always continue after lra_constraints on the first
24670 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
24672 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
24674 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
24676 2017-01-26 Jakub Jelinek <jakub@redhat.com>
24678 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
24679 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
24680 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
24681 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
24682 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
24683 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
24684 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
24685 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
24686 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
24688 2017-01-26 Marek Polacek <polacek@redhat.com>
24691 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
24692 for the third operand.
24694 2017-01-26 Jakub Jelinek <jakub@redhat.com>
24696 PR middle-end/79236
24697 * omp-low.c (struct omp_context): Add simt_stmt field.
24698 (scan_omp_for): Return omp_context *.
24699 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
24700 context to the _simt_ SIMD stmt.
24701 (lower_omp_for): For combined SIMD with sibling _simt_
24702 SIMD, make sure to use the same decls in _looptemp_
24703 clauses as in the sibling.
24705 2017-01-26 David Sherwood <david.sherwood@arm.com>
24707 PR middle-end/79212
24708 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
24711 2017-01-26 Jakub Jelinek <jakub@redhat.com>
24714 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
24715 emit fld b; fld a; if possible.
24717 * brig-builtins.def: Update copyright years.
24718 * config/arm/arm_acle_builtins.def: Update copyright years.
24720 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
24723 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
24724 constraint instead of o for the stxsd instruction.
24726 2017-01-25 Carl Love <cel@us.ibm.com>
24728 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
24729 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
24731 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
24733 * doc/invoke.texi (C++ Dialect Options): Fix typo.
24735 2017-01-25 Richard Biener <rguenther@suse.de>
24737 PR tree-optimization/69264
24738 * target.def (vector_alignment_reachable): Improve documentation.
24739 * doc/tm.texi: Regenerate.
24740 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
24742 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
24743 earlier changes with respect to TYPE_USER_ALIGN.
24744 (vector_alignment_reachable_p): Likewise. Improve dumping.
24746 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24749 * config/arm/arm.md (xordi3): Force constant operand into a register
24752 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24754 * doc/invoke.texi (-fstore-merging): Correct default optimization
24755 levels at which it is enabled.
24756 (-O): Move -fstore-merging from list to...
24759 2017-01-25 Richard Biener <rguenther@suse.de>
24762 * omp-expand.c: Include debug.h.
24763 (expand_omp_taskreg): Make sure to generate early debug before
24764 outlining anything from a function.
24765 (expand_omp_target): Likewise.
24766 (grid_expand_target_grid_body): Likewise.
24768 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
24771 * asan.c (get_translation_unit_decl): New function.
24772 (asan_add_global): Extract modules file name from globals
24773 TRANSLATION_UNIT_DECL name.
24775 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
24778 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
24779 for long calls with APCS frame and VFP.
24781 2017-01-24 David Malcolm <dmalcolm@redhat.com>
24783 * cfg.c (original_copy_tables_initialized_p): New function.
24784 * cfg.h (original_copy_tables_initialized_p): New decl.
24785 * cfgrtl.c (relink_block_chain): Guard the call to
24786 free_original_copy_tables with a call to
24787 original_copy_tables_initialized_p.
24788 * cgraph.h (symtab_node::native_rtl_p): New decl.
24789 * cgraphunit.c (symtab_node::native_rtl_p): New function.
24790 (symtab_node::needed_p): Don't assert for early assembly output
24791 for __RTL functions.
24792 (cgraph_node::finalize_function): Set "force_output" for __RTL
24794 (cgraph_node::analyze): Bail out early for __RTL functions.
24795 (analyze_functions): Update assertion to support __RTL functions.
24796 (cgraph_node::expand): Bail out early for __RTL functions.
24797 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
24799 * function.h (struct function): Update comment for field
24801 * gimple-expr.c: Include "tree-pass.h".
24802 (gimple_has_body_p): Return false for __RTL functions.
24803 * Makefile.in (OBJS): Add run-rtl-passes.o.
24804 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
24806 (gcc::pass_manager::get_clean_slate): New accessor.
24807 * passes.c: Include "insn-addr.h".
24808 (should_skip_pass_p): Add logging. Update logic for running
24809 "expand" to be compatible with both __GIMPLE and __RTL. Guard
24810 property-provider override so it is only done for gimple passes.
24812 (skip_pass): New function.
24813 (execute_one_pass): Call skip_pass when skipping passes.
24814 * read-md.c (md_reader::read_char): Support filtering
24815 the input to a subset of line numbers.
24816 (md_reader::md_reader): Initialize fields
24817 m_first_line and m_last_line.
24818 (md_reader::read_file_fragment): New function.
24819 * read-md.h (md_reader::read_file_fragment): New decl.
24820 (md_reader::m_first_line): New field.
24821 (md_reader::m_last_line): New field.
24822 * read-rtl-function.c (function_reader::create_function): Only
24823 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
24824 curr_properties. Set DECL_INITIAL to a dummy block.
24825 (read_rtl_function_body_from_file_range): New function.
24826 * read-rtl-function.h (read_rtl_function_body_from_file_range):
24828 * run-rtl-passes.c: New file.
24829 * run-rtl-passes.h: New file.
24831 2017-01-24 Jeff Law <law@redhat.com>
24833 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
24836 2017-01-24 Bin Cheng <bin.cheng@arm.com>
24838 PR tree-optimization/79159
24839 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
24840 (record_nonwrapping_iv): Improve boundary using above function if no
24841 value range information.
24843 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
24844 Martin Jambor <mjambor@suse.cz>
24846 * brig-builtins.def: New file.
24847 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
24848 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
24849 (DEF_HSAIL_SAT_BUILTIN): Likewise.
24850 (DEF_HSAIL_INTR_BUILTIN): Likewise.
24851 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
24852 * builtin-types.def (BT_INT8): New.
24853 (BT_INT16): Likewise.
24854 (BT_UINT8): Likewise.
24855 (BT_UINT16): Likewise.
24856 (BT_FN_ULONG): Likewise.
24857 (BT_FN_UINT_INT): Likewise.
24858 (BT_FN_UINT_ULONG): Likewise.
24859 (BT_FN_UINT_LONG): Likewise.
24860 (BT_FN_UINT_PTR): Likewise.
24861 (BT_FN_ULONG_PTR): Likewise.
24862 (BT_FN_INT8_FLOAT): Likewise.
24863 (BT_FN_INT16_FLOAT): Likewise.
24864 (BT_FN_UINT32_FLOAT): Likewise.
24865 (BT_FN_UINT16_FLOAT): Likewise.
24866 (BT_FN_UINT8_FLOAT): Likewise.
24867 (BT_FN_UINT64_FLOAT): Likewise.
24868 (BT_FN_UINT16_UINT32): Likewise.
24869 (BT_FN_UINT32_UINT16): Likewise.
24870 (BT_FN_UINT16_UINT16_UINT16): Likewise.
24871 (BT_FN_INT_PTR_INT): Likewise.
24872 (BT_FN_UINT_PTR_UINT): Likewise.
24873 (BT_FN_LONG_PTR_LONG): Likewise.
24874 (BT_FN_ULONG_PTR_ULONG): Likewise.
24875 (BT_FN_VOID_UINT64_UINT64): Likewise.
24876 (BT_FN_UINT8_UINT8_UINT8): Likewise.
24877 (BT_FN_INT8_INT8_INT8): Likewise.
24878 (BT_FN_INT16_INT16_INT16): Likewise.
24879 (BT_FN_INT_INT_INT): Likewise.
24880 (BT_FN_UINT_FLOAT_UINT): Likewise.
24881 (BT_FN_FLOAT_UINT_UINT): Likewise.
24882 (BT_FN_ULONG_UINT_UINT): Likewise.
24883 (BT_FN_ULONG_UINT_PTR): Likewise.
24884 (BT_FN_ULONG_ULONG_ULONG): Likewise.
24885 (BT_FN_UINT_UINT_UINT): Likewise.
24886 (BT_FN_VOID_UINT_PTR): Likewise.
24887 (BT_FN_UINT_UINT_PTR: Likewise.
24888 (BT_FN_UINT32_UINT64_PTR): Likewise.
24889 (BT_FN_INT_INT_UINT_UINT): Likewise.
24890 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
24891 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
24892 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
24893 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
24894 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
24895 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
24896 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
24897 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
24898 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
24899 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
24900 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
24901 * doc/frontends.texi: List BRIG FE.
24902 * doc/install.texi (Testing): Add BRIG tesring requirements.
24903 * doc/invoke.texi (Overall Options): Mention BRIG.
24904 * doc/standards.texi (Standards): Doucment BRIG HSA version.
24906 2017-01-24 Richard Biener <rguenther@suse.de>
24908 PR translation/79208
24909 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
24911 2017-01-24 Martin Jambor <mjambor@suse.cz>
24914 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
24915 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
24916 and known_contexts.
24918 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
24920 PR middle-end/79123
24921 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
24922 casts from signed to unsigned really don't have a range.
24924 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
24926 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
24927 GMP_RNDx for compatiblity.
24929 2017-01-24 Martin Liska <mliska@suse.cz>
24932 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
24933 that would prevent us to call alloca with -1 as argument.
24935 2017-01-24 Jakub Jelinek <jakub@redhat.com>
24937 * dwarf2out.c (output_compilation_unit_header, output_file_names):
24938 Avoid -Wformat-security warning.
24940 2017-01-23 Andrew Pinski <apinski@cavium.com>
24942 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
24945 2017-01-23 Martin Sebor <msebor@redhat.com>
24947 PR middle-end/78703
24948 * gimple-ssa-sprintf.c (warn_level): New global.
24949 (format_integer): Use it here and throughout the rest of the file.
24950 Use the same switch to compute sign as base.
24951 (maybe_warn): New function.
24952 (format_directive): Factor out warnings into maybe_warn.
24953 Add debugging output. Use warn_level.
24954 (add_bytes): Use warn_level.
24955 (pass_sprintf_length::compute_format_length): Add debugging output.
24956 (try_substitute_return_value): Same.
24957 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
24959 PR middle-end/78703
24960 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
24961 (struct fmtresult, format_integer, format_floating): Adjust.
24962 (fmtresult::fmtresult): Set max correctly in two argument ctor.
24963 (get_string_length, format_string,format_directive): Same.
24964 (pass_sprintf_length::compute_format_length): Same.
24965 (try_substitute_return_value): Simplify slightly.
24967 PR middle-end/78703
24968 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
24969 (fmtresult::operator+=): Outlined.
24970 (struct fmtresult): Add ctors.
24971 (struct conversion_spec): Rename...
24972 (struct directive): ...to this. Add and remove data members.
24973 (directive::set_width, directive::set_precision): New functions.
24974 (format_percent): Use fmtresult ctor.
24975 (get_width_and_precision): Remove.
24976 (format_integer): Make naming changes. Avoid computing width and
24978 (format_floating): Same. Adjust indentation.
24979 (format_character, format_none): New functions.
24980 (format_string): Moved character handling to format_character.
24981 (format_directive): Remove arguments, change return type.
24982 (parse_directive): New function.
24983 (pass_sprintf_length::compute_format_length): Move directive
24984 parsing to parse_directive.
24986 2017-01-23 Jakub Jelinek <jakub@redhat.com>
24988 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
24989 (assign_assembler_name_if_needed): ... this.
24990 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
24991 (assign_assembler_name_if_needed): ... this.
24992 (free_lang_data_in_cgraph): Adjust callers.
24993 * cgraphunit.c (cgraph_node::analyze): Likewise.
24994 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
24997 2017-01-23 Richard Biener <rguenther@suse.de>
24999 PR tree-optimization/79088
25000 PR tree-optimization/79188
25001 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
25002 resetting loop bounds after last path deletion. Reset loop
25003 bounds of the target loop, make code match the comments.
25004 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
25005 Make sure loops need no fixups.
25007 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
25009 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
25010 exponent support with double type for first argument.
25011 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
25012 type returned by __builtin_vec_extract_sig,
25013 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
25014 functions from "vector int" to "vector unsigned int" or from
25015 "vector long long int" to "vector unsigned long long int".
25016 Changed type returned by __builtin_vec_extract_exp,
25017 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
25018 functions from "vector int" to "vector unsigned int" or from
25019 "vector long long int" to "vector unsigned long long int".
25020 Changed return type of __builtin_vec_test_data_class,
25021 __builtin_vec_test_data_class_sp, and
25022 __builtin_vec_test_data_class_dp from "vector int" to
25023 "vector bool int" or from "vector long long int" to "vector bool
25024 long long int" and changed second argument type from "unsigned
25025 int" to "int". Added new overloaded function forms "vector float
25026 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
25027 "vector float __builtin_vec_insert_exp_sp (vector float, vector
25028 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
25029 double, vector unsigned long long int)" and "vector double
25030 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
25031 long int)". Changed return type of
25032 __builtin_scalar_test_data_class and
25033 __builtin_scalar_test_data_class_sp and
25034 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
25035 int" and changed second argument from "unsigned int" to "int".
25036 Changed type returned by __builtin_scalar_test_neg,
25037 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
25038 from "int" to "bool int". Added new overloaded function form
25039 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
25040 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
25041 exponent double-precision with floating point first argument.
25042 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
25043 documentation of scalar_test_data_class, scalar_test_neg,
25044 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
25045 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
25046 vec_test_data_class built-in functions to reflect refinements in
25047 their type signatures.
25049 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
25051 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
25053 (aarch64_elf_asm_destructor): Likewise.
25055 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
25057 PR rtl-optimization/78634
25058 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
25059 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
25060 * ifcvt.c (noce_try_cmove): Add missing cost check.
25062 PR rtl-optimization/71724
25063 * combine.c (if_then_else_cond): Look for situations where it is
25064 beneficial to undo the work of one of the recursive calls.
25066 2017-01-23 Bin Cheng <bin.cheng@arm.com>
25068 PR tree-optimization/70754
25069 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
25070 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
25071 combined stmt before it if not NULL.
25072 (combine_chains): Process refs reversely and compute dominance point
25075 2017-01-23 Martin Liska <mliska@suse.cz>
25077 PR tree-optimization/79196
25078 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
25079 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
25081 (strlen_optimize_stmt): Call the renamed function.
25083 2017-01-23 Michael Matz <matz@suse.de>
25085 PR tree-optimization/78384
25086 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
25088 2017-01-23 Richard Biener <rguenther@suse.de>
25090 PR tree-optimization/79186
25091 * tree-vrp.c (register_new_assert_for): Make sure we've seen
25092 both incoming edges before moving an assert.
25094 2017-01-23 Martin Jambor <mjambor@suse.cz>
25096 * ipa-prop.c (load_from_param_1): Removed.
25097 (load_from_unmodified_param): Bits from load_from_param_1 put back
25099 (load_from_param): Removed.
25100 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
25101 with stmt. Reverted back to use of load_from_unmodified_param.
25103 2017-01-23 Martin Jambor <mjambor@suse.cz>
25106 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
25107 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
25108 field a pointer to garbage collected vector, mark lattices and
25109 ipcp_orig_node with GTY((skip)).
25110 (ipa_get_param_count): Adjust to descriptors being a pointer.
25111 (ipa_get_param): Likewise.
25112 (ipa_get_type): Likewise.
25113 (ipa_get_param_move_cost): Likewise.
25114 (ipa_set_param_used): Likewise.
25115 (ipa_get_controlled_uses): Likewise.
25116 (ipa_set_controlled_uses): Likewise.
25117 (ipa_is_param_used): Likewise.
25118 (ipa_node_params_t): Move into garbage collector. New methods insert
25120 (ipa_node_params_sum): Annotate wth GTY(()).
25121 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
25123 (ipa_load_from_parm_agg): Adjust declaration.
25124 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
25125 * ipa-profile.c (ipa_profile): Likewise.
25126 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
25127 (ipa_populate_param_decls): Make descriptors parameter garbage
25129 (ipa_dump_param): Adjust to descriptors being a pointer.
25130 (ipa_alloc_node_params): Likewise.
25131 (ipa_initialize_node_params): Likewise.
25132 (load_from_param_1): Make descriptors parameter garbage collected.
25133 (load_from_unmodified_param): Likewise.
25134 (load_from_param): Likewise.
25135 (ipa_load_from_parm_agg): Likewise.
25136 (ipa_node_params::~ipa_node_params): Removed.
25137 (ipa_free_all_node_params): Remove call to delete operator.
25138 (ipa_node_params_t::insert): New.
25139 (ipa_node_params_t::remove): Likewise.
25140 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
25141 copy known_csts and known_contexts vectors.
25142 (ipa_read_node_info): Adjust to descriptors being a pointer.
25143 (ipcp_modif_dom_walker): Make m_descriptors field garbage
25145 (ipcp_transform_function): Make descriptors variable garbage
25148 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
25150 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
25151 * config/i386/avx512dqintrin.h: Ditto.
25152 * config/i386/avx512fintrin.h: Ditto.
25153 * config/i386/i386.c: Handle new builtins.
25154 * config/i386/i386-builtin.def: Add new builtins.
25155 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
25156 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
25158 2017-01-23 Jakub Jelinek <jakub@redhat.com>
25159 Martin Liska <mliska@suse.cz>
25161 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
25162 * asan.c (asan_expand_poison_ifn): Support stores and use
25163 appropriate ASAN report function.
25164 * internal-fn.c (expand_ASAN_POISON_USE): New function.
25165 * internal-fn.def (ASAN_POISON_USE): Declare.
25166 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
25167 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
25168 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
25169 ASAN_POISON calls w/o LHS.
25170 * tree-ssa.c (execute_update_addresses_taken): Create clobber
25171 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
25172 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
25173 * gimplify.c (asan_poison_variables): Add attribute
25174 use_after_scope_memory to variables that really needs to live
25176 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
25177 having the attribute.
25179 2017-01-23 Martin Liska <mliska@suse.cz>
25181 * asan.c (create_asan_shadow_var): New function.
25182 (asan_expand_poison_ifn): Likewise.
25183 * asan.h (asan_expand_poison_ifn): New declaration.
25184 * internal-fn.c (expand_ASAN_POISON): Likewise.
25185 * internal-fn.def (ASAN_POISON): New builtin.
25186 * sanopt.c (pass_sanopt::execute): Expand
25187 asan_expand_poison_ifn.
25188 * tree-inline.c (copy_decl_for_dup_finish): Make function
25190 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
25191 * tree-ssa.c (is_asan_mark_p): New function.
25192 (execute_update_addresses_taken): Rewrite local variables
25193 (identified just by use-after-scope as addressable) into SSA.
25195 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
25197 * doc/install.texi (Specific): opensource.apple.com uses https
25198 now. Remove trailing slash.
25200 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
25202 * README.Portability: Remove note on an Irix compatibility issue.
25204 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
25206 * gcov.c (INCLUDE_ALGORITHM): Define.
25207 (INCLUDE_VECTOR): Define.
25208 No longer include <vector> and <algorithm> directly.
25210 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
25212 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
25214 * doc/invoke.texi (Code Gen Options): Ditto.
25216 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
25219 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
25221 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
25223 rtl-optimization/79125
25224 * cprop.c (local_cprop_pass): Handle cases where we make an
25225 unconditional trap.
25227 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
25231 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
25232 read from, for big endian.
25234 2017-01-20 Jiong Wang <jiong.wang@arm.com>
25236 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
25237 register pauth builtins for LP64 only.
25239 2017-01-20 Marek Polacek <polacek@redhat.com>
25242 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
25245 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
25247 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
25249 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
25250 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
25251 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
25253 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25256 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
25257 in big-endian mode when they are not a single duplicated value.
25259 2017-01-20 Richard Biener <rguenther@suse.de>
25261 * BASE-VER: Bump to 7.0.1.
25263 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
25265 * omp-low.c (omplow_simd_context): New struct. Use it...
25266 (lower_rec_simd_input_clauses): ...here and...
25267 (lower_rec_input_clauses): ...here to hold common data. Adjust all
25268 references to idx, lane, max_vf, is_simt.
25270 2017-01-20 Graham Markall <graham.markall@embecosm.com>
25272 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
25275 2017-01-20 Martin Jambor <mjambor@suse.cz>
25277 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
25278 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
25280 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
25281 (GTFILES): Rename hsa.c to hsa-common.c.
25282 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
25283 * hsa-dump.c: Likewise.
25284 * hsa-gen.c: Likewise.
25285 * hsa-regalloc.c: Likewise.
25286 * ipa-hsa.c: Likewise.
25287 * omp-expand.c: Likewise.
25288 * omp-low.c: Likewise.
25289 * toplev.c: Likewise.
25291 2017-01-20 Marek Polacek <polacek@redhat.com>
25294 * doc/invoke.texi: Document -Wduplicated-branches.
25295 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
25296 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
25297 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
25298 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
25299 return 0 only when not OEP_LEXICOGRAPHIC.
25300 (fold_build_cleanup_point_expr): Use the expression
25301 location when building CLEANUP_POINT_EXPR.
25302 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
25303 * tree.c (add_expr): Handle error_mark_node.
25305 2017-01-20 Martin Liska <mliska@suse.cz>
25308 * tree-profile.c (init_ic_make_global_vars): Do not call
25310 (gimple_init_gcov_profiler): Likewise.
25312 2017-01-20 Martin Liska <mliska@suse.cz>
25315 * cgraph.h (maybe_create_reference): Remove argument and
25317 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
25319 * ipa-cp.c (create_specialized_node): Likewise.
25320 * symtab.c (symtab_node::maybe_create_reference): Handle
25321 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
25323 2017-01-20 Martin Liska <mliska@suse.cz>
25325 * read-rtl-function.c (function_reader::create_function): Use
25326 build_decl instread of build_decl_stat.
25328 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
25330 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
25331 * config/i386/avx512dqintrin.h: Ditto.
25332 * config/i386/avx512fintrin.h: Ditto.
25333 * config/i386/i386-builtin-types.def: Add new types.
25334 * config/i386/i386.c: Handle new types.
25335 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
25336 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
25337 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
25338 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
25339 (__builtin_ia32_kshiftridi): New.
25340 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
25342 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
25346 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
25347 define to rs6000_init_stack_protect_guard.
25348 (rs6000_init_stack_protect_guard): New function.
25350 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
25351 Yunqiang Su <yunqiang.su@imgtec.com>
25353 * config.gcc (supported_defaults): Add madd4.
25354 (with_madd4): Add validation.
25355 (all_defaults): Add madd4.
25356 * config/mips/mips.opt (mmadd4): New option.
25357 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
25359 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
25361 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
25362 (ISA_HAS_FUSED_MADD4): Likewise.
25363 * doc/invoke.texi (-mmadd4): Document the new option.
25364 * doc/install.texi (--with-madd4): Document the new option.
25366 2017-01-19 Jiong Wang <jiong.wang@arm.com>
25368 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
25369 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
25370 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
25371 (aarch64_init_pauth_hint_builtins): New.
25372 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
25373 (aarch64_expand_builtin): Expand new builtins.
25375 2017-01-19 Jiong Wang <jiong.wang@arm.com>
25377 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
25378 * combine-stack-adj.c (no_unhandled_cfa): Handle
25379 REG_CFA_TOGGLE_RA_MANGLE.
25380 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
25381 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
25382 info for return address signing.
25383 (aarch64_expand_epilogue): Likewise.
25385 2017-01-19 Jiong Wang <jiong.wang@arm.com>
25387 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
25388 * config/aarch64/aarch64-protos.h
25389 (aarch64_return_address_signing_enabled): New declaration.
25390 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
25392 (aarch64_expand_prologue): Sign return address before it's pushed onto
25394 (aarch64_expand_epilogue): Authenticate return address fetched from
25396 (aarch64_override_options): Sanity check for ILP32 and ISA level.
25397 (aarch64_attributes): New function attributes for "sign-return-address".
25398 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
25399 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
25400 ("*do_return"): Generate combined instructions according to key index.
25401 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
25402 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
25404 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
25405 * config/aarch64/aarch64.opt (msign-return-address=): New.
25406 * doc/extend.texi (AArch64 Function Attributes): Documents
25407 "sign-return-address=".
25408 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
25410 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
25412 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
25413 overall option summary.
25415 2017-01-19 Jiong Wang <jiong.wang@arm.com>
25417 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
25418 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
25419 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
25420 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
25422 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
25424 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
25425 -mpower9-minmax by default for -mcpu=power9.
25426 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
25427 128-bit floating point.
25429 2017-01-20 Alan Modra <amodra@gmail.com>
25431 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
25432 optimizing for size.
25434 2017-01-20 Alan Modra <amodra@gmail.com>
25437 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
25438 for strcmp and strncmp from corresponding builtin decl.
25440 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
25442 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
25443 instead of i386/rtems-64.h.
25444 * config/i386/rtems-64.h: Remove.
25446 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
25450 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
25452 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
25454 2017-01-19 Tamar Christina <tamar.christina@arm.com>
25456 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
25457 Change int to HOST_WIDE_INT.
25458 * config/aarch64/aarch64-protos.h
25459 (aarch64_simd_gen_const_vector_dup): Likewise.
25460 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
25462 2017-01-19 David Malcolm <dmalcolm@redhat.com>
25464 * langhooks-def.h (lhd_type_for_size): New decl.
25465 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
25466 * langhooks.c (lhd_type_for_size): New function, taken from
25469 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
25471 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
25472 define_bypass for CR latency.
25473 (power9-cracked-alu): Update bypass latency and remove power9-branch.
25474 (power9-alu2): Add define_bypass for CR latency.
25476 (power9-mul): Update insn latency.
25477 (power9-mul-compare): Update insn latency, bypass latency and remove
25480 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25482 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
25484 * config/aarch64/aarch64.md
25485 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
25486 aarch64_nopcrelative_literal_loads.
25487 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
25489 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
25491 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
25492 TARGET_LOONGSON_3A.
25493 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
25495 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
25498 * config.gcc (supported_defaults): Add lxc1-sxc1.
25499 (with_lxc1_sxc1): Add validation.
25500 (all_defaults): Add lxc1-sxc1.
25501 * config/mips/mips.opt (mlxc1-sxc1): New option.
25502 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
25504 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
25505 __mips_no_lxc1_sxc1.
25506 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
25507 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
25508 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
25510 2017-01-19 Richard Biener <rguenther@suse.de>
25512 PR tree-optimization/72488
25513 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
25514 sure to restore SSA info.
25515 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
25517 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
25519 PR rtl-optimization/79121
25520 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
25521 of the inner type when shifting an extended value.
25523 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
25526 * symtab.c (symtab_node::equal_address_to): Fix comparing of
25527 interposable aliases.
25529 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
25532 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
25533 Use the evmergelohi instruction.
25534 (mov_si<mode>_e500_subreg4_2_le): Likewise.
25535 (mov_sitf_e500_subreg8_2_be): Likewise.
25536 (mov_sitf_e500_subreg12_2_le): Likewise.
25537 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
25538 (mov_si<mode>_e500_subreg4_2_be): Likewise.
25539 (mov_sitf_e500_subreg8_2_le): Likewise.
25540 (mov_sitf_e500_subreg12_2_be): Likewise.
25542 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25544 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
25545 attribute from vecsimple to vecperm.
25546 (altivec_vbpermq2): Likewise.
25548 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25551 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
25553 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25554 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
25555 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
25556 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
25557 case where N arg is SIZE_MAX.
25558 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
25559 (cmpstrsi): Add pattern.
25561 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
25563 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
25564 __builtin_vec_revb builtins.
25565 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
25566 built-in functions to support generation of the ISA 3.0 XXBR<x>
25567 vector byte reverse instructions.
25568 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
25569 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
25570 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
25571 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
25572 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
25573 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
25574 (P9V_BUILTIN_VEC_REVB): Likewise.
25575 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
25576 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
25577 (p9_xxbrq_v16qi): Likewise.
25578 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
25579 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
25580 (p9_xxbrh_v8hi): Likewise.
25581 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
25582 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
25583 vec_revb built-in functions.
25585 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
25587 PR rtl-optimization/78952
25588 * config/i386/i386.md (any_extract): New code iterator.
25589 (*insvqi_2): Use any_extract for source operand.
25590 (*insvqi_3): Use any_shiftrt for source operand.
25592 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
25594 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
25596 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
25598 2017-01-18 Matthias Klose <doko@ubuntu.com>
25600 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
25602 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25604 * config/rs6000/altivec.h (vec_bperm): Change #define.
25605 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
25606 (altivec_vbpermq2): New define_insn.
25607 (altivec_vbpermd): Likewise.
25608 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
25609 function interface.
25610 (VBPERMD): Likewise.
25611 (VBPERM): New polymorphic function interface.
25612 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
25613 Add entries for P9V_BUILTIN_VEC_VBPERM.
25614 * doc/extend.texi: Add interfaces for vec_bperm.
25616 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25618 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
25619 first letter of error messages.
25620 (s390_resolve_overloaded_builtin): Likewise.
25621 * config/s390/s390.c (s390_expand_builtin): Likewise.
25622 (s390_invalid_arg_for_unprototyped_fn): Likewise.
25623 (s390_valid_target_attribute_inner_p): Likewise.
25624 * config/s390/s390.md ("tabort"): Likewise.
25626 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
25628 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
25629 (ISA_AVOID_DIV_HILO): New macro.
25630 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
25631 (ISA_HAS_DDIV): Likewise.
25633 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
25635 * doc/invoke.texi (fabi-version): Correct number of occurrences.
25637 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
25639 * doc/invoke.texi (fabi-version): Spelling fix.
25641 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
25644 * doc/invoke.texi (fabi-version): Mention mangling fix for
25647 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
25650 * doc/invoke.texi (fabi-version): Document discriminator mangling.
25652 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
25655 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
25656 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
25658 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
25660 (stack_protect_test): Ditto.
25661 * config/rs6000/rs6000.opt (mstack-protector-guard=,
25662 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
25664 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
25665 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
25666 -mstack-protector-guard-offset=.
25667 (RS/6000 and PowerPC Options): Ditto.
25669 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
25671 * config/i386/i386.h (MASK_CLASS_P): New define.
25672 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
25673 there are no registers from different register sets also when
25674 mask registers are used. Update function comment.
25675 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
25676 to (*k/*r) and (*k/*km) alternatives.
25678 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
25680 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
25681 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
25682 (EH_RETURN_HANDLER_RTX): New define.
25683 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
25684 Force frame pointer in EH return functions.
25685 (aarch64_expand_epilogue): Add barrier for eh_return.
25686 (aarch64_final_eh_return_addr): Remove.
25687 (aarch64_eh_return_handler_rtx): New function.
25688 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
25690 (aarch64_eh_return_handler_rtx): New prototype.
25692 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25694 * config/rs6000/altivec.h (vec_rlmi): New #define.
25695 (vec_vrlnm): Likewise.
25696 (vec_rlnm): Likewise.
25697 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
25698 (UNSPEC_VRLNM): Likewise.
25699 (VIlong): New mode iterator.
25700 (altivec_vrl<VI_char>mi): New define_insn.
25701 (altivec_vrl<VI_char>nm): Likewise.
25702 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
25704 (VRLDNM): Likewise.
25705 (RLNM): New polymorphic function entry.
25706 (VRLWMI): New monomorphic function entry.
25707 (VRLDMI): Likewise.
25708 (RLMI): New polymorphic function entry.
25709 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
25710 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
25711 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
25714 2017-01-17 Jakub Jelinek <jakub@redhat.com>
25717 * dwarf2out.c (field_byte_offset): Restore the
25718 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
25719 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
25720 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
25723 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
25726 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
25728 2017-01-17 Jakub Jelinek <jakub@redhat.com>
25731 * dwarf2out.c (add_data_member_location_attribute): For constant
25732 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
25733 instead of DW_AT_data_member_location, DW_AT_bit_offset and
25734 DW_AT_byte_size attributes.
25736 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
25738 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
25739 after forcing to constant memory when the code model is medium.
25741 2017-01-17 Julia Koval <julia.koval@intel.com>
25744 * config/i386/avx512fintrin.h
25745 (_mm512_i32gather_ps): Change __addr type to void const*.
25746 (_mm512_mask_i32gather_ps): Ditto.
25747 (_mm512_i32gather_pd): Ditto.
25748 (_mm512_mask_i32gather_pd): Ditto.
25749 (_mm512_i64gather_ps): Ditto.
25750 (_mm512_mask_i64gather_ps): Ditto.
25751 (_mm512_i64gather_pd): Ditto.
25752 (_mm512_mask_i64gather_pd): Ditto.
25753 (_mm512_i32gather_epi32): Ditto.
25754 (_mm512_mask_i32gather_epi32): Ditto.
25755 (_mm512_i32gather_epi64): Ditto.
25756 (_mm512_mask_i32gather_epi64): Ditto.
25757 (_mm512_i64gather_epi32): Ditto.
25758 (_mm512_mask_i64gather_epi32): Ditto.
25759 (_mm512_i64gather_epi64): Ditto.
25760 (_mm512_mask_i64gather_epi64): Ditto.
25761 (_mm512_i32scatter_ps): Change __addr type to void*.
25762 (_mm512_mask_i32scatter_ps): Ditto.
25763 (_mm512_i32scatter_pd): Ditto.
25764 (_mm512_mask_i32scatter_pd): Ditto.
25765 (_mm512_i64scatter_ps): Ditto.
25766 (_mm512_mask_i64scatter_ps): Ditto.
25767 (_mm512_i64scatter_pd): Ditto.
25768 (_mm512_mask_i64scatter_pd): Ditto.
25769 (_mm512_i32scatter_epi32): Ditto.
25770 (_mm512_mask_i32scatter_epi32): Ditto.
25771 (_mm512_i32scatter_epi64): Ditto.
25772 (_mm512_mask_i32scatter_epi64): Ditto.
25773 (_mm512_i64scatter_epi32): Ditto.
25774 (_mm512_mask_i64scatter_epi32): Ditto.
25775 (_mm512_i64scatter_epi64): Ditto.
25776 (_mm512_mask_i64scatter_epi64): Ditto.
25777 * config/i386/avx512pfintrin.h
25778 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
25779 (_mm512_mask_prefetch_i32gather_ps): Ditto.
25780 (_mm512_mask_prefetch_i64gather_pd): Ditto.
25781 (_mm512_mask_prefetch_i64gather_ps): Ditto.
25782 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
25783 (_mm512_prefetch_i32scatter_ps): Ditto.
25784 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
25785 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
25786 (_mm512_prefetch_i64scatter_pd): Ditto.
25787 (_mm512_prefetch_i64scatter_ps): Ditto.
25788 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
25789 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
25790 * config/i386/avx512vlintrin.h
25791 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
25792 (_mm_mmask_i32gather_ps): Ditto.
25793 (_mm256_mmask_i32gather_pd): Ditto.
25794 (_mm_mmask_i32gather_pd): Ditto.
25795 (_mm256_mmask_i64gather_ps): Ditto.
25796 (_mm_mmask_i64gather_ps): Ditto.
25797 (_mm256_mmask_i64gather_pd): Ditto.
25798 (_mm_mmask_i64gather_pd): Ditto.
25799 (_mm256_mmask_i32gather_epi32): Ditto.
25800 (_mm_mmask_i32gather_epi32): Ditto.
25801 (_mm256_mmask_i32gather_epi64): Ditto.
25802 (_mm_mmask_i32gather_epi64): Ditto.
25803 (_mm256_mmask_i64gather_epi32): Ditto.
25804 (_mm_mmask_i64gather_epi32): Ditto.
25805 (_mm256_mmask_i64gather_epi64): Ditto.
25806 (_mm_mmask_i64gather_epi64): Ditto.
25807 (_mm256_i32scatter_ps): Change __addr type to void*.
25808 (_mm256_mask_i32scatter_ps): Ditto.
25809 (_mm_i32scatter_ps): Ditto.
25810 (_mm_mask_i32scatter_ps): Ditto.
25811 (_mm256_i32scatter_pd): Ditto.
25812 (_mm256_mask_i32scatter_pd): Ditto.
25813 (_mm_i32scatter_pd): Ditto.
25814 (_mm_mask_i32scatter_pd): Ditto.
25815 (_mm256_i64scatter_ps): Ditto.
25816 (_mm256_mask_i64scatter_ps): Ditto.
25817 (_mm_i64scatter_ps): Ditto.
25818 (_mm_mask_i64scatter_ps): Ditto.
25819 (_mm256_i64scatter_pd): Ditto.
25820 (_mm256_mask_i64scatter_pd): Ditto.
25821 (_mm_i64scatter_pd): Ditto.
25822 (_mm_mask_i64scatter_pd): Ditto.
25823 (_mm256_i32scatter_epi32): Ditto.
25824 (_mm256_mask_i32scatter_epi32): Ditto.
25825 (_mm_i32scatter_epi32): Ditto.
25826 (_mm_mask_i32scatter_epi32): Ditto.
25827 (_mm256_i32scatter_epi64): Ditto.
25828 (_mm256_mask_i32scatter_epi64): Ditto.
25829 (_mm_i32scatter_epi64): Ditto.
25830 (_mm_mask_i32scatter_epi64): Ditto.
25831 (_mm256_i64scatter_epi32): Ditto.
25832 (_mm256_mask_i64scatter_epi32): Ditto.
25833 (_mm_i64scatter_epi32): Ditto.
25834 (_mm_mask_i64scatter_epi32): Ditto.
25835 (_mm256_i64scatter_epi64): Ditto.
25836 (_mm256_mask_i64scatter_epi64): Ditto.
25837 (_mm_i64scatter_epi64): Ditto.
25838 (_mm_mask_i64scatter_epi64): Ditto.
25839 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
25840 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
25841 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
25842 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
25843 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
25844 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
25845 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
25846 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
25847 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
25848 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
25849 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
25850 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
25851 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
25852 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
25853 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
25854 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
25855 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
25856 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
25857 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
25858 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
25859 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
25860 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
25861 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
25862 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
25863 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
25864 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
25865 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
25866 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
25867 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
25868 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
25869 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
25870 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
25871 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
25872 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
25873 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
25874 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
25875 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
25876 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
25877 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
25878 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
25879 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
25880 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
25881 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
25882 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
25883 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
25884 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
25885 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
25886 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
25887 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
25888 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
25889 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
25890 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
25891 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
25892 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
25893 definitions accordingly.
25895 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
25896 Kuan-Lin Chen <kuanlinchentw@gmail.com>
25899 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
25902 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
25905 * ira-conflicts.c (ira_build_conflicts): Update total conflict
25906 hard regs for inner regno.
25908 2017-01-17 Martin Liska <mliska@suse.cz>
25911 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
25912 assumption and add comment.
25914 2017-01-17 Nathan Sidwell <nathan@acm.org>
25916 * ipa-visibility.c (localize_node): New function, broken out of ...
25917 (function_and_variable_visibility): ... here. Call it.
25919 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
25921 PR middle-end/77445
25922 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
25923 correctly set frequency of oudgoing edge.
25924 (duplicate_thread_path): Fix profile updating.
25926 2017-01-17 Jakub Jelinek <jakub@redhat.com>
25929 * configure.ac: Add GCC_BASE_VER.
25930 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
25931 version from BASE-VER file.
25932 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
25933 (gcc.o): Depend on $(BASEVER).
25934 * common.opt (dumpfullversion): New option.
25935 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
25936 * doc/invoke.texi: Document -dumpfullversion.
25937 * doc/install.texi: Document --with-gcc-major-version-only.
25938 * configure: Regenerated.
25940 2017-01-17 Richard Biener <rguenther@suse.de>
25942 PR tree-optimization/71433
25943 * tree-vrp.c (register_new_assert_for): Merge same asserts
25944 on all incoming edges.
25945 (process_assert_insertions_for): Handle insertions at the
25948 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
25950 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
25951 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
25953 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
25956 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
25959 2017-01-17 Alan Modra <amodra@gmail.com>
25962 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
25963 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
25964 symbolic stack limit when pic.
25966 2017-01-16 Martin Sebor <msebor@redhat.com>
25968 PR tree-optimization/78608
25969 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
25971 2017-01-16 Jeff Law <law@redhat.com>
25974 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
25975 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
25976 for several include directories that may be relative to sysroot.
25977 * config/i386/x-mingw32 (gplus_includedir): Define.
25978 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
25979 (native_system_includedir): Likewise.
25980 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
25981 override if TARGET_SYSTEM_ROOT is defined.
25982 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
25984 PR tree-optimization/79090
25985 PR tree-optimization/33562
25986 PR tree-optimization/61912
25987 PR tree-optimization/77485
25988 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
25989 and computed trims into the dump file.
25991 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
25993 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
25995 2017-01-16 Jakub Jelinek <jakub@redhat.com>
25998 * gimplify.c (gimplify_init_constructor): If want_value and
25999 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
26003 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
26004 sequence. Formatting fixes.
26005 (doloop_optimize): Formatting fixes.
26008 * gcc.c (debug_level_greater_than_spec_func): New function.
26009 (static_spec_functions): Add debug-level-gt spec function.
26010 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
26012 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
26013 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
26014 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
26015 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
26016 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
26017 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
26019 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
26021 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
26022 QImode fixups to general and mask registers only.
26024 2017-01-16 Carl Love <cel@us.ibm.com>
26026 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
26027 for built-in functions
26028 vector signed char vec_nabs (vector signed char)
26029 vector signed short vec_nabs (vector signed short)
26030 vector signed int vec_nabs (vector signed int)
26031 vector signed long long vec_nabs (vector signed long long)
26032 vector float vec_nabs (vector float)
26033 vector double vec_nabs (vector double)
26034 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
26036 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
26037 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
26038 * doc/extend.texi: Update the documentation file for the new built-in
26041 2017-01-16 Martin Sebor <msebor@redhat.com>
26043 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
26046 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26048 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
26049 UNSPEC_VSX__XXSPLTD to require special splat handling.
26051 2017-01-16 David Malcolm <dmalcolm@redhat.com>
26054 * system.h: Poison strndup.
26056 2017-01-16 Alan Modra <amodra@gmail.com>
26059 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
26062 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
26064 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
26066 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
26068 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
26069 call recog here. Assert that INSN_CODE (insn) is non-negative.
26071 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
26074 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
26076 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
26077 in the currently scheduled RTL fragment.
26079 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
26081 PR rtl-optimization/78751
26082 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
26085 2017-01-14 Jeff Law <law@redhat.com>
26087 PR tree-optimization/79090
26088 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
26089 variable length stores.
26090 (compute_trims): Delete dead assignment to *trim_tail.
26091 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
26094 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
26096 PR rtl-optimization/78626
26097 PR rtl-optimization/78727
26098 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
26099 of a block, and split such blocks after everything else is finished.
26101 2017-01-14 Alan Modra <amodra@gmail.com>
26104 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
26105 target legitimate_combined_insn.
26106 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
26107 (rs6000_legitimate_combined_insn): New function.
26108 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
26110 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
26111 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
26112 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
26114 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
26116 * doc/frontends.texi (G++ and GCC): Remove references to Java.
26118 2017-01-13 Jeff Law <law@redhat.com>
26120 PR tree-optimization/33562
26121 PR tree-optimization/61912
26122 PR tree-optimization/77485
26123 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
26125 (delete_dead_assignment): Likewise.
26126 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
26127 statement to delete_dead_call and delete_dead_assignment.
26129 2017-01-13 David Malcolm <dmalcolm@redhat.com>
26132 * substring-locations.c (format_warning_va): Strengthen case 1 so
26133 that both endpoints of the substring must be within the format
26134 range for just the substring to be printed.
26136 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
26138 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
26139 * config/i386/i386.c (ix86_target_string): Add missing options
26140 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
26141 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
26142 flags_other and ix86_target_other to flags2_other. Display unknown
26144 (ix86_valid_target_attribute_inner_p): Add missing options and
26145 reorder options by implied ISAs, as in ix86_target_string.
26147 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
26149 * hash-table.h (hash_table::too_empty_p): New function.
26150 (hash_table::expand): Use it.
26151 (hash_table::traverse): Likewise.
26152 (hash_table::empty_slot): Use sizeof (value_type) instead of
26153 sizeof (PTR) to convert bytes to elements. Shrink the table
26154 if the current size is excessive for the current number of
26157 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
26159 * ira-costs.c (record_reg_classes): Break from the inner loop
26160 early once alt_fail is known to be true. Update outer loop
26161 handling accordingly.
26163 2017-01-13 Jeff Law <law@redhat.com>
26165 * tree-ssa-dse.c (decrement_count): New function.
26166 (increment_start_addr, maybe_trim_memstar_call): Likewise.
26167 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
26168 when we know the partially dead statement is a mem* function.
26170 PR tree-optimization/61912
26171 PR tree-optimization/77485
26172 * tree-ssa-dse.c: Include expr.h.
26173 (maybe_trim_constructor_store): New function.
26174 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
26176 PR tree-optimization/33562
26177 PR tree-optimization/61912
26178 PR tree-optimization/77485
26179 * doc/invoke.texi: Document new dse-max-object-size param.
26180 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
26181 * tree-ssa-dse.c: Include params.h.
26182 (dse_store_status): New enum.
26183 (initialize_ao_ref_for_dse): New, partially extracted from
26185 (valid_ao_ref_for_dse, normalize_ref): New.
26186 (setup_live_bytes_from_ref, compute_trims): Likewise.
26187 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
26188 (maybe_trim_partially_dead_store): Likewise.
26189 (maybe_trim_complex_store): Likewise.
26190 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
26191 Track what bytes live from the original store. Return tri-state
26192 for dead, partially dead or live.
26193 (dse_dom_walker): Add constructor, destructor and new private members.
26194 (delete_dead_call, delete_dead_assignment): New extracted from
26196 (dse_optimize_stmt): Make a member of dse_dom_walker.
26197 Use initialize_ao_ref_for_dse.
26199 PR tree-optimization/33562
26200 PR tree-optimization/61912
26201 PR tree-optimization/77485
26202 * sbitmap.h (bitmap_count_bits): Prototype.
26203 (bitmap_clear_range, bitmap_set_range): Likewise.
26204 * sbitmap.c (bitmap_clear_range): New function.
26205 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
26207 2017-01-13 Martin Liska <mliska@suse.cz>
26210 * function.c (set_cfun): Add new argument force.
26211 * function.h (set_cfun): Likewise.
26212 * ipa-inline-transform.c (inline_call): Use the function when
26213 strict alising from is dropped for function we inline to.
26215 2017-01-13 Richard Biener <rguenther@suse.de>
26217 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
26218 for dumping GIMPLE INTEGER_CSTs.
26220 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26222 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
26223 to 201112L since C++17.
26225 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
26228 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
26229 if -fsanitize=kernel-address is present.
26231 2017-01-13 Richard Biener <rguenther@suse.de>
26233 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
26234 as _Literal ( type ) number in case usual suffixes do not
26235 preserve all information.
26237 2017-01-13 Richard Biener <rguenther@suse.de>
26239 PR tree-optimization/77283
26240 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
26241 and ssa-iterators.h.
26242 (is_feasible_trace): Implement a cost model based on joiner
26245 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
26248 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
26249 char or short to __float128/_Float128 directly.
26251 2017-01-12 Martin Sebor <msebor@redhat.com>
26253 to -Wformat-overflow.
26254 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
26255 (min_bytes_remaining): Same.
26256 (get_string_length): Same.
26257 (format_string): Same.
26258 (format_directive): Same.
26260 (pass_sprintf_length::handle_gimple_call): Same.
26262 2017-01-12 Jakub Jelinek <jakub@redhat.com>
26264 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
26265 info.nowrite calls with no lhs that can't throw. Return bool
26266 whether gsi_remove has been called or not.
26267 (pass_sprintf_length::handle_gimple_call): Return bool whether
26268 try_substitute_return_value called gsi_remove. Formatting fix.
26269 (pass_sprintf_length::execute): Don't use gsi_remove if
26270 handle_gimple_call returned true.
26273 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
26274 be removed due to side-effects, don't remove following barrier nor
26275 turn the successor edge into fallthru edge.
26277 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26280 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
26281 element-reversing loads and stores as not swappable.
26283 2017-01-12 Nathan Sidwell <nathan@acm.org>
26284 Nicolai Stange <nicstange@gmail.com>
26286 * combine.c (try_combine): Don't ignore result of overlap checking
26287 loop. Combine overlap & asm check into single loop.
26289 2017-01-12 Richard Biener <rguenther@suse.de>
26291 * tree-pretty-print.c (dump_generic_node): Provide -gimple
26292 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
26294 2017-01-12 Richard Biener <rguenther@suse.de>
26296 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
26297 and TS_TARGET_OPTION directly derive from TS_BASE.
26298 * tree-core.h (tree_optimization_option): Derive from tree_base.
26299 (tree_target_option): Likewise.
26301 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
26303 * config/i386/i386.c (memory_address_length): Increase len
26304 only when rip_relative_addr_p returns false.
26306 2017-01-11 Julia Koval <julia.koval@intel.com>
26308 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
26309 (OPTION_MASK_ISA_SGX_SET): New.
26310 (ix86_handle_option): Handle OPT_msgx.
26311 * config.gcc: Added sgxintrin.h.
26312 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
26313 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
26314 * config/i386/i386.c (ix86_target_string): Add -msgx.
26316 (ix86_option_override_internal): Handle new options.
26317 (ix86_valid_target_attribute_inner_p): Add sgx.
26318 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
26319 * config/i386/i386.opt: Add msgx.
26320 * config/i386/sgxintrin.h: New file.
26321 * config/i386/x86intrin.h: Add sgxintrin.h.
26323 2017-01-11 Jakub Jelinek <jakub@redhat.com>
26326 * fold-const.c (maybe_nonzero_address): Return 1 for function
26328 (tree_single_nonzero_warnv_p): Don't handle function local objects
26332 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
26335 2017-01-11 David Malcolm <dmalcolm@redhat.com>
26338 * opts.c: Include "spellcheck.h"
26339 (struct string_fragment): New struct.
26340 (struct edit_distance_traits<const string_fragment &>): New
26342 (get_closest_sanitizer_option): New function.
26343 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
26345 2017-01-11 Jakub Jelinek <jakub@redhat.com>
26347 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
26349 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
26350 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
26351 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
26352 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
26353 for initial die_offset if dwarf_split_debug_info.
26354 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
26355 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
26357 (output_skeleton_debug_sections): Formatting fix. Use
26358 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
26359 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
26361 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
26363 * config/arm/cortex-a53.md: Add bypasses for
26364 cortex_a53_r2f_cvt.
26365 (cortex_a53_r2f): Only use for transfers.
26366 (cortex_a53_f2r): Likewise.
26367 (cortex_a53_r2f_cvt): Add reservation for conversions.
26368 (cortex_a53_f2r_cvt): Likewise.
26370 2017-01-11 Tamar Christina <tamar.christina@arm.com>
26372 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
26373 to all inlined functions, change static to extern.
26375 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
26378 * config/arm/arm.c (legitimize_pic_address): Handle reference to
26380 (arm_assemble_integer): Likewise.
26382 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
26384 * config.gcc: Use new awk script to check CPU, FPU and architecture
26385 parameters for --with-... options.
26386 * config/arm/parsecpu.awk: New file
26387 * config/arm/arm-cpus.in: New file.
26388 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
26390 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
26392 * config/arm/t-arm: Update dependency rules.
26393 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
26394 of processing .def files.
26395 * config/arm/genopt.sh: Deleted.
26396 * config/arm/gentune.sh: Deleted.
26397 * config/arm/arm-cores.def: Deleted.
26398 * config/arm/arm-arches.def: Deleted.
26399 * config/arm/arm-fpus.def: Deleted.
26400 * config/arm/arm-tune.md: Regenerated.
26401 * config/arm/arm-tables.opt: Regenerated.
26402 * config/arm/arm-cpu.h: New generated file.
26403 * config/arm/arm-cpu-data.h: New generated file.
26404 * config/arm/arm-cpu-cdata.h: New generated file.
26406 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
26409 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
26411 (input_varpool_node): Unpack dynamically_initialized bit.
26413 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
26415 PR rtl-optimization/79032
26416 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
26417 the alignment of the adjusted memory reference against that of MODE,
26418 instead of the alignment of the original memory reference.
26420 2017-01-11 Martin Jambor <mjambor@suse.cz>
26422 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
26424 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
26425 decorated functions.
26427 2017-01-11 Richard Biener <rguenther@suse.de>
26429 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
26430 set range/nonnull info for PHI results. Do not set it on
26431 stmts marked for removal.
26433 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
26435 * expr.c (store_field): In the bitfield case, fetch the return value
26436 from the registers before applying a single big-endian adjustment.
26437 Always do a final load for a BLKmode value not larger than a word.
26439 2017-01-10 David Malcolm <dmalcolm@redhat.com>
26442 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
26443 that we correctly handle column numbers greater than
26444 LINE_MAP_MAX_COLUMN_NUMBER.
26446 2017-01-10 Martin Sebor <msebor@redhat.com>
26448 PR middle-end/78245
26449 * gimple-ssa-sprintf.c (get_destination_size): Call
26450 {init,fini}object_sizes.
26451 * tree-object-size.c (addr_object_size): Adjust.
26452 (pass_through_call): Adjust.
26453 (pass_object_sizes::execute): Adjust.
26454 * tree-object-size.h (fini_object_sizes): Declare.
26456 2017-01-10 Martin Sebor <msebor@redhat.com>
26458 PR tree-optimization/78775
26459 * builtins.c (get_size_range): Move...
26460 * calls.c: ...to here.
26461 (alloc_max_size): Accept zero argument.
26462 (operand_signed_p): Remove.
26463 (maybe_warn_alloc_args_overflow): Call get_size_range.
26464 * calls.h (get_size_range): Declare.
26466 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
26468 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
26469 from TI's devices.csv file as of September 2016.
26470 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
26472 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
26474 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
26475 * doc/invoke.texi: Likewise.
26476 * doc/md.texi: Likewise.
26477 * doc/objc.texi: Likewise.
26479 2017-01-10 Joshua Conner <joshconner@google.com>
26481 * config/arm/fuchsia-elf.h: New file.
26482 * config/fuchsia.h: New file.
26483 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
26484 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
26486 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
26488 2016-01-10 Richard Biener <rguenther@suse.de>
26490 PR tree-optimization/79034
26491 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
26492 Propagate out degenerate PHIs in the joiner.
26494 2017-01-10 Martin Liska <mliska@suse.cz>
26496 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
26497 (sort_congruence_classes_by_decl_uid): Likewise.
26498 (sort_congruence_class_groups_by_decl_uid): Likewise.
26499 (sem_item_optimizer::merge_classes): Sort class, groups in these
26500 classes and members in the groups by DECL_UID of declarations.
26501 This would make merge operations stable.
26503 2017-01-10 Martin Liska <mliska@suse.cz>
26505 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
26506 usage of m_classes_vec.
26507 (sem_item_optimizer::~sem_item_optimizer): Likewise.
26508 (sem_item_optimizer::get_group_by_hash): Likewise.
26509 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
26510 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
26511 (sem_item_optimizer::verify_classes): Likewise.
26512 (sem_item_optimizer::process_cong_reduction): Likewise.
26513 (sem_item_optimizer::dump_cong_classes): Likewise.
26514 (sem_item_optimizer::merge_classes): Likewise.
26515 * ipa-icf.h (congruence_class_hash): Rename from
26516 congruence_class_group_hash. Remove declaration of m_classes_vec.
26518 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
26520 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
26521 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
26522 * config.gcc: Add avx512vpopcntdqintrin.h.
26523 * config/i386/avx512vpopcntdqintrin.h: New.
26524 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
26525 * config/i386/i386-builtin-types.def: Add new types.
26526 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
26527 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
26528 __builtin_ia32_vpopcountq_v8di_mask): New.
26529 * config/i386/i386-c.c (ix86_target_macros_internal): Define
26530 __AVX512VPOPCNTDQ__.
26531 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
26532 (PTA_AVX512VPOPCNTDQ): Define.
26533 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
26534 TARGET_AVX512VPOPCNTDQ_P): Define.
26535 * config/i386/i386.opt: Add mavx512vpopcntdq.
26536 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
26537 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
26539 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
26541 PR middle-end/77484
26542 * predict.def (PRED_CALL): Set to 67.
26544 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
26546 * expr.c (store_field): In the bitfield case, if the value comes from
26547 a function call and is of an aggregate type returned in registers, do
26548 not modify the field mode; extract the value in all cases if the mode
26549 is BLKmode and the size is not larger than a word.
26551 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
26554 * config/i386/cpuid.h: Fix undefined behavior.
26556 2017-01-04 Jeff Law <law@redhat.com>
26558 PR tree-optimization/79007
26559 PR tree-optimization/67955
26560 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
26561 conservative for pt.null when flag_non_call_exceptions is on.
26563 2017-01-09 Jakub Jelinek <jakub@redhat.com>
26565 PR translation/79019
26566 PR translation/79020
26567 * params.def (PARAM_INLINE_MIN_SPEEDUP,
26568 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
26569 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
26571 * config/avr/avr.opt (maccumulate-args): Likewise.
26572 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
26573 * common.opt (freport-bug): Likewise.
26574 * cif-code.def (CIF_FINAL_ERROR): Likewise.
26575 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
26576 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
26577 translatable string.
26578 * config/i386/i386.c (function_value_32): Likewise.
26579 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
26580 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
26582 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
26583 * common/config/msp430/msp430-common.c (msp430_handle_option):
26585 * symtab.c (symtab_node::verify_base): Likewise.
26586 * opts.c (set_debug_level): Likewise.
26587 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
26588 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
26589 missing whitespace to translatable strings.
26590 * config/avr/avr.md (bswapsi2): Fix typo in comment.
26591 * config/sh/superh.h: Likewise.
26592 * config/i386/xopintrin.h: Likewise.
26593 * config/i386/znver1.md: Likewise.
26594 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
26595 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
26596 * double-int.h (struct double_int): Likewise.
26597 * double-int.c (div_and_round_double): Likewise.
26598 * wide-int.cc: Likewise.
26599 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
26600 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
26601 * cfgcleanup.c (crossjumps_occured): Renamed to ...
26602 (crossjumps_occurred): ... this.
26603 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
26606 PR tree-optimization/78899
26607 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
26608 returning bool return struct loop *, NULL for failure and the new
26610 (versionable_outer_loop_p): Don't version outer loop if it has
26611 dont_vectorized bit set.
26612 (tree_if_conversion): When versioning outer loop, ensure
26613 tree_if_conversion is performed also on the inner loop of the
26614 non-vectorizable outer loop copy.
26615 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
26616 LOOP_VECTORIZED in inner loop of the scalar outer loop and
26617 prevent vectorization of it.
26618 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
26619 the outer loop vectorization of the non-scalar version is attempted
26620 before vectorization of the inner loop in scalar version. If
26621 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
26622 vectorization of its inner loop.
26623 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
26624 has 2 inner loops, rename also on edges from bb whose single pred
26625 is outer_loop->header. Fix typo in function comment.
26627 2017-01-09 Martin Sebor <msebor@redhat.com>
26630 * asan.c (asan_emit_stack_protection): Increase local buffer size
26631 to avoid snprintf truncation warning.
26633 2017-01-09 Andrew Pinski <apinski@cavium.com>
26635 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
26636 to reference thunderx2t99 for the tuning structure
26637 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
26639 (thunderx2t99_extra_costs): This.
26640 * config/aarch64/aarch64-tune.md: Regenerate.
26641 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
26642 (vulcan_addrcost_table): This.
26643 (vulcan_regmove_cost): Rename to ...
26644 (thunderx2t99_regmove_cost): This.
26645 (vulcan_vector_cost): Rename to ...
26646 (thunderx2t99_vector_cost): this.
26647 (vulcan_branch_cost): Rename to ...
26648 (thunderx2t99_branch_cost): This.
26649 (vulcan_tunings): Rename to ...
26650 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
26651 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
26653 2017-01-09 Martin Jambor <mjambor@suse.cz>
26657 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
26658 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
26659 (propagate_vr_accross_jump_function): Use the above function for all
26660 value range computations for pass-through jump functions and type
26661 converasion from explicit value range values.
26662 (ipcp_propagate_stage): Do not attempt to deduce types of formal
26663 parameters from TYPE_ARG_TYPES.
26664 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
26665 (ipa_write_node_info): Stream type of the actual argument.
26666 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
26668 2017-01-09 Martin Liska <mliska@suse.cz>
26671 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
26672 (lookup_compiler): Do not show error message with have_E.
26674 2017-01-09 Jakub Jelinek <jakub@redhat.com>
26676 PR tree-optimization/78938
26677 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
26678 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
26679 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
26680 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
26683 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26685 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
26688 2017-01-09 Richard Biener <rguenther@suse.de>
26690 PR tree-optimization/78997
26691 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
26692 name condition properly.
26694 2017-01-09 Richard Biener <rguenther@suse.de>
26697 * dwarf2out.c (is_cxx): New overload with context.
26698 (is_naming_typedef_decl): Use it.
26700 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
26702 * invoke.texi (Option Summary): Correct spacing in option lists
26703 and add line breaks to fix over-long lines.
26705 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
26707 PR middle-end/17660
26709 * extend.texi (Common Variable Attributes): Add xref to GCC
26710 Internals manual to explain mode attribute keywords.
26712 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
26715 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
26716 and Preprocessor Options.
26717 (Options for Linking): Document -pthread here....
26718 (RS/6000 and PowerPC Options): ...not here.
26719 (Solaris 2 Options): ...or here.
26720 * doc/cppopts.texi: Document -pthread.
26722 2017-01-08 Martin Sebor <msebor@redhat.com>
26724 PR middle-end/77708
26725 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
26726 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
26727 New member functions.
26728 (format_directive): Used them.
26730 (pass_sprintf_length::handle_gimple_call): Same.
26731 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
26732 to avoid truncation for any argument.
26733 (extract_affine_mul): Same.
26734 * tree.c (get_file_function_name): Same.
26736 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
26738 PR middle-end/77484
26739 * predict.def (PRED_INDIR_CALL): Set to 86.
26741 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
26743 PR preprocessor/54124
26744 * doc/cppopts.texi: Reformat -d subtable to list the full name
26745 of the options. Add cross-reference to the docs for the general
26746 compiler -d options.
26747 * doc/invoke.texi (Developer Options): Add cross-reference to the
26748 preprocessor-specific -d option documentation.
26750 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
26752 PR preprocessor/13498
26753 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
26754 redudant material, and reflect new command-line options.
26755 (System Headers): Likewise.
26757 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
26759 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
26760 -isystem, and -idirafter. Copy-edit.
26761 * doc/cppopts.texi: Copy-edit. Remove contradiction about
26762 default for -ftrack-macro-expansion. Delete obsolete and
26763 badly-formatted implementation details about -fdebug-cpp output.
26764 * doc/cppwarnopts.texi: Copy-edit.
26766 2017-01-07 David Malcolm <dmalcolm@redhat.com>
26769 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
26770 that the transition from a max line width >= 1<<10 to narrower
26771 lines works correctly.
26773 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
26775 * doc/options.texi (PerFunction): New.
26776 * opt-functions.awk (switch_flags): Map both Optimization and
26777 PerFunction to CL_OPTIMIZATION.
26778 * opth-gen.awk: Test for PerFunction flag along with
26780 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
26781 it only when the latter is present. Skip those that don't in
26782 the hash function generator.
26783 * common.opt (fvar-tracking): Mark as PerFunction instead of
26785 (fvar-tracking-assignments): Likewise.
26786 (fvar-tracking-assignments-toggle): Likewise.
26787 (fvar-tracking-uninit): Likewise.
26789 2017-01-07 Jakub Jelinek <jakub@redhat.com>
26791 PR translation/79018
26792 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
26795 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
26798 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
26799 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
26800 TARGET_LONG_JUMP_TABLE_OFFSETS.
26801 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
26802 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
26803 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
26804 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
26805 * config/m68k/m68k.md (tablejump expander): Likewise.
26806 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
26807 TARGET_LONG_JUMP_TABLE_OFFSETS.
26808 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
26809 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
26811 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
26812 David Holsgrove <david.holsgrove@xilinx.com>
26814 * common/config/microblaze/microblaze-common.c
26815 (TARGET_EXCEPT_UNWIND_INFO): Remove.
26816 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
26818 * config/microblaze/microblaze.c (microblaze_must_save_register)
26819 (microblaze_expand_epilogue, microblaze_return_addr): Handle
26821 (microblaze_eh_return): New function.
26822 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
26823 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
26824 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
26825 * config/microblaze/microblaze.md (eh_return): New pattern.
26827 2017-01-06 Jakub Jelinek <jakub@redhat.com>
26829 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
26830 GCC_DIAGNOSTIC_STRINGIFY): Define.
26832 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
26834 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26836 * config/arm/arm.md (<mcrr>): New.
26838 * config/arm/arm.c (arm_arch5te): New.
26839 (arm_option_override): Set arm_arch5te.
26840 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
26842 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
26843 (arm_mcrr_qualifiers): ... this. New.
26844 (MRRC_QUALIFIERS): Define to...
26845 (arm_mrrc_qualifiers): ... this. New.
26846 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
26848 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
26849 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
26850 (MRRCI, mrrc, MRRC): New.
26851 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
26852 VUNSPEC_MRRC2): New.
26854 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26856 * config/arm/arm.md (<mcr>): New.
26858 * config/arm/arm.c (arm_coproc_builtin_available): Add
26859 support for mcr, mrc, mcr2 and mrc2.
26860 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
26861 (arm_mcr_qualifiers): ... this. New.
26862 (MRC_QUALIFIERS): Define to ...
26863 (arm_mrc_qualifiers): ... this. New.
26864 (MCR_QUALIFIERS): Define to ...
26865 (arm_mcr_qualifiers): ... this. New.
26866 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
26868 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
26869 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
26870 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
26871 VUNSPEC_MRC2): New.
26873 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26875 * config/arm/arm.md (*ldc): New.
26879 * config/arm/arm.c (arm_coproc_builtin_available): Add
26880 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
26881 (arm_coproc_ldc_stc_legitimate_address): New.
26882 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
26883 'qualifier_const_pointer'.
26884 (LDC_QUALIFIERS): Define to...
26885 (arm_ldc_qualifiers): ... this. New.
26886 (STC_QUALIFIERS): Define to...
26887 (arm_stc_qualifiers): ... this. New.
26888 * config/arm/arm-protos.h
26889 (arm_coproc_ldc_stc_legitimate_address): New.
26890 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
26891 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
26892 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
26893 stc2, stcl, stc2l): New.
26894 * config/arm/constraints.md (Uz): New.
26895 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
26896 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
26897 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
26898 VUNSPEC_STC2L): New.
26900 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26902 * config/arm/arm.md (<cdp>): New.
26903 * config/arm/arm.c (neon_const_bounds): Rename this ...
26904 (arm_const_bounds): ... this.
26905 (arm_coproc_builtin_available): New.
26906 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
26907 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
26908 (CDP_QUALIFIERS): Define to...
26909 (arm_cdp_qualifiers): ... this. New.
26911 (arm_expand_builtin_args): Add case for 6 arguments.
26912 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
26913 (arm_const_bounds): ... this.
26914 (arm_coproc_builtin_available): New.
26915 * config/arm/arm_acle.h (__arm_cdp): New.
26917 * config/arm/arm_acle_builtins.def (cdp): New.
26919 * config/arm/iterators.md (CDPI,CDP,cdp): New.
26920 * config/arm/neon.md: Rename all 'neon_const_bounds' to
26921 'arm_const_bounds'.
26922 * config/arm/types.md (coproc): New.
26923 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
26924 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
26925 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
26926 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
26928 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26930 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
26931 (UBINOP_QUALIFIERS): New.
26933 (acle_builtin_data): New. Change comment.
26934 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
26935 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
26936 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
26937 arm_acle_builtins.def.
26938 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
26939 (arm_init_acle_builtins): New.
26940 (CRC32_BUILTIN): Remove.
26941 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
26942 crc32cb, crc32ch and crc32cw.
26943 (arm_init_crc32_builtins): Remove.
26944 (arm_init_builtins): Use arm_init_acle_builtins rather
26945 than arm_init_crc32_builtins.
26946 (arm_expand_acle_builtin): New.
26947 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
26948 * config/arm/arm_acle_builtins.def: New.
26950 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26952 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
26953 (arm_builtin_datum): ... this.
26954 (arm_init_neon_builtin): Rename to ...
26955 (arm_init_builtin): ... this. Add a new parameters PREFIX
26956 and USE_SIG_IN_NAME.
26957 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
26958 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
26959 'arm_builtin_datum'.
26960 (arm_init_vfp_builtins): Likewise.
26961 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
26962 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
26963 (arm_expand_neon_args): Rename to ...
26964 (arm_expand_builtin_args): ... this. Rename builtin_arg
26965 enum values and differentiate between ARG_BUILTIN_MEMORY
26966 and ARG_BUILTIN_NEON_MEMORY.
26967 (arm_expand_neon_builtin_1): Rename to ...
26968 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
26969 values, arm_expand_builtin_args and add bool parameter NEON.
26970 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
26971 (arm_expand_vfp_builtin): Likewise.
26972 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
26974 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
26976 PR middle-end/77484
26977 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
26978 * predict.c (tree_estimate_probability_bb): Reverse direction of
26979 polymorphic call predictor.
26981 2017-01-06 David Malcolm <dmalcolm@redhat.com>
26983 * passes.c (execute_one_pass): Split out pass-skipping logic into...
26984 (determine_pass_name_match): ...this new function and...
26985 (should_skip_pass_p): ...this new function.
26987 2017-01-06 Nathan Sidwell <nathan@acm.org>
26989 * ipa-visibility.c (function_and_variable_visibility): Reformat
26990 comments and long lines. Remove extrneous if.
26991 * symtab.c (symtab_node::make_decl_local): Fix code format.
26992 (symtab_node::set_section_for_node): Fix comment typo.
26994 2017-01-06 Martin Liska <mliska@suse.cz>
26997 * lra-constraints.c: Rename invariant to lra_invariant.
26998 * predict.c (set_even_probabilities): Initialize e to NULL.
27000 2017-01-05 Martin Sebor <msebor@redhat.com>
27002 PR tree-optimization/78910
27003 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
27004 (format_integer): Correct off-by-one error in the handling
27005 of precision with negative numbers in signed conversions..
27007 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
27009 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
27011 2017-01-05 Jakub Jelinek <jakub@redhat.com>
27013 PR tree-optimization/71016
27014 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
27015 factor_out_conditional_conversion. Formatting fix.
27016 (factor_out_conditional_conversion): Add cond_stmt argument.
27017 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
27018 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
27021 2017-01-05 David Malcolm <dmalcolm@redhat.com>
27023 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
27024 read-rtl-function.o, and selftest-rtl.o.
27025 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
27026 (selftest::aarch64_test_loading_full_dump): New function.
27027 (selftest::aarch64_run_selftests): New function.
27028 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
27029 selftest::aarch64_run_selftests.
27030 * config/i386/i386.c
27031 (selftest::ix86_test_loading_dump_fragment_1): New function.
27032 (selftest::ix86_test_loading_call_insn): New function.
27033 (selftest::ix86_test_loading_full_dump): New function.
27034 (selftest::ix86_test_loading_unspec): New function.
27035 (selftest::ix86_run_selftests): Call the new functions.
27036 * emit-rtl.c (maybe_set_max_label_num): New function.
27037 * emit-rtl.h (maybe_set_max_label_num): New decl.
27038 * function.c (instantiate_decls): Guard call to
27039 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
27040 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
27042 * gensupport.c (gen_reader::gen_reader): Pass "false"
27043 for new "compact" param of rtx_reader.
27044 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
27045 rather than an empty string for NULL strings.
27046 * read-md.c: Potentially include config.h rather than bconfig.h.
27047 Wrap include of errors.h with #ifdef GENERATOR_FILE.
27048 (have_error): New global, copied from errors.c.
27049 (md_reader::read_name): Rename to...
27050 (md_reader::read_name_1): ...this, adding "out_loc" param,
27051 and converting "missing name or number" to returning false, rather
27053 (md_reader::read_name): Reimplement in terms of read_name_1.
27054 (md_reader::read_name_or_nil): New function.
27055 (md_reader::read_string): Handle "(nil)" by returning NULL.
27056 (md_reader::md_reader): Add new param "compact".
27057 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
27058 (md_reader::read_file): New method.
27059 * read-md.h (md_reader::md_reader): Add new param "compact".
27060 (md_reader::read_file): New method.
27061 (md_reader::is_compact): New accessor.
27062 (md_reader::read_name): Convert return type from void to file_location.
27063 (md_reader::read_name_or_nil): New decl.
27064 (md_reader::read_name_1): New decl.
27065 (md_reader::m_compact): New field.
27066 (noop_reader::noop_reader): Pass "false" for new "compact" param
27068 (rtx_reader::rtx_reader): Add new "compact" param.
27069 (rtx_reader::read_rtx_operand): Make virtual and convert return
27070 type from void to rtx.
27071 (rtx_reader::read_until): New decl.
27072 (rtx_reader::handle_any_trailing_information): New virtual function.
27073 (rtx_reader::postprocess): New virtual function.
27074 (rtx_reader::finalize_string): New virtual function.
27075 (rtx_reader::m_in_call_function_usage): New field.
27076 (rtx_reader::m_reuse_rtx_by_id): New field.
27077 * read-rtl-function.c: New file.
27078 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
27079 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
27080 (selftest::verify_three_block_rtl_cfg): New decl.
27081 * read-rtl-function.h: New file.
27082 * read-rtl.c: Potentially include config.h rather than bconfig.h.
27083 For host, include function.h, memmodel.h, and emit-rtl.h.
27084 (one_time_initialization): New function.
27085 (struct compact_insn_name): New struct.
27086 (compact_insn_names): New array.
27087 (find_code): Handle insn codes in compact dumps.
27088 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
27089 (bind_subst_iter_and_attr): Likewise.
27090 (add_condition_to_string): Likewise.
27091 (add_condition_to_rtx): Likewise.
27092 (apply_attribute_uses): Likewise.
27093 (add_current_iterators): Likewise.
27094 (apply_iterators): Likewise.
27095 (initialize_iterators): Guard usage of apply_subst_iterator with
27096 #ifdef GENERATOR_FILE.
27097 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
27098 (md_reader::read_mapping): Likewise.
27099 (add_define_attr_for_define_subst): Likewise.
27100 (add_define_subst_attr): Likewise.
27101 (read_subst_mapping): Likewise.
27102 (check_code_iterator): Likewise.
27103 (rtx_reader::read_rtx): Likewise. Move one-time initialization
27105 (one_time_initialization): New function.
27106 (rtx_reader::read_until): New method.
27107 (read_flags): New function.
27108 (parse_reg_note_name): New function.
27109 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
27110 Handle reuse_rtx ids.
27111 Wrap iterator lookup within #ifdef GENERATOR_FILE.
27112 Add parsing support for RTL dumps, mirroring the special-cases in
27113 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
27114 values, and calling handle_any_trailing_information.
27115 (rtx_reader::read_rtx_operand): Convert return type from void
27116 to rtx, returning return_rtx. Handle case 'e'. Call
27117 finalize_string on XSTR and XTMPL fields.
27118 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
27119 "(nil)" values were omitted. Call the postprocess vfunc on the
27121 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
27122 class ctor. Initialize m_in_call_function_usage. Call
27123 one_time_initialization.
27124 * rtl-tests.c (selftest::test_uncond_jump): Call
27125 set_new_first_and_last_insn.
27126 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
27127 * selftest-rtl.c: New file.
27128 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
27129 (selftest::get_insn_by_uid): New decl.
27130 * selftest-run-tests.c (selftest::run_tests): Call
27131 read_rtl_function_c_tests.
27132 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
27133 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
27136 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
27138 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
27139 operands in a special way. Assert that pos+len <= mode precision.
27141 2017-01-05 Jakub Jelinek <jakub@redhat.com>
27143 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
27144 3 argument Alias with unlimited for the negative form.
27145 (fno-vect-cost-model): Removed.
27147 2017-01-05 Martin Liska <mliska@suse.cz>
27149 * hsa-gen.c (gen_hsa_divmod): New function.
27150 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
27152 2017-01-05 Martin Liska <mliska@suse.cz>
27155 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
27158 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27160 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
27161 small constant length operands.
27163 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27165 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
27166 between loop iterations.
27168 2017-01-05 Martin Liska <mliska@suse.cz>
27171 * gimplify.c (gimplify_decl_expr): Compare to
27172 asan_poisoned_variables instread of checking flags.
27173 (gimplify_target_expr): Likewise.
27174 (gimplify_expr): Likewise.
27175 (gimplify_function_tree): Conditionally initialize
27176 asan_poisoned_variables.
27178 2017-01-04 Jeff Law <law@redhat.com>
27180 PR tree-optimizatin/78812
27181 * rtl.h (contains_mem_rtx_p): Prototype.
27182 * ifcvt.c (containts_mem_rtx_p): Move from here to...
27183 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
27184 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
27185 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
27186 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
27188 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
27190 * input.c (assert_char_at_range): Default-initialize actual_range.
27192 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
27194 * df-scan.c (df_ref_create_structure): Make regno unsigned,
27195 to match the caller.
27197 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
27199 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
27200 insns after final jump in test to emit dummy move.
27202 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
27204 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
27205 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
27207 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
27209 * multiple_target.c (create_dispatcher_calls): Init e_next.
27210 * tree-ssa-loop-split.c (split_loop): Init border.
27211 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
27214 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
27219 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
27220 (altivec_register_operand): Do not return true if the operand
27221 contains a SUBREG mixing SImode and SFmode.
27222 (vsx_register_operand): Likewise.
27223 (vsx_reg_sfsubreg_ok): New predicate.
27224 (vfloat_operand): Do not return true if the operand contains a
27225 SUBREG mixing SImode and SFmode.
27226 (vint_operand): Likewise.
27227 (vlogical_operand): Likewise.
27228 (gpc_reg_operand): Likewise.
27229 (int_reg_operand): Likewise.
27230 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
27231 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
27232 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
27234 (rs6000_emit_move_si_sf_subreg): New helper function.
27235 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
27236 fixup SUBREGs involving SImode and SFmode.
27237 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
27238 numbers for the new peephole2 optimization.
27239 (peephole2 for SFmode unions): New peephole2 to optimize cases in
27240 the GLIBC math library that do AND/IOR/XOR operations on single
27241 precision floating point.
27242 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
27243 target macros to say whether we need to avoid SUBREGs mixing
27245 (TARGET_ALLOW_SF_SUBREG): Likewise.
27246 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
27247 (UNSPEC_SI_FROM_SF): Likewise.
27248 (iorxor): Change spacing.
27249 (and_ior_xor): New iterator for AND, IOR, and XOR.
27250 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
27251 (movdi_from_sf_zero_ext): Likewise.
27252 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
27253 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
27254 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
27255 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
27256 (fms<mode>4): Likewise.
27257 (fnma<mode>4): Likewise.
27258 (fnms<mode>4): Likewise.
27259 (nfma<mode>4): Likewise.
27260 (nfms<mode>4): Likewise.
27262 2017-01-04 Marek Polacek <polacek@redhat.com>
27265 * doc/invoke.texi: Document -Wpointer-compare.
27267 2017-01-04 Jakub Jelinek <jakub@redhat.com>
27269 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
27272 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
27273 descriptions for -gdwarf-5 and emit them as uleb128 instead of
27276 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
27279 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
27280 documentation of the powerpc_popcntb_ok attribute.
27281 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
27282 code to issue warning messages if a requested CPU configuration is
27283 not supported by the binary (assembler and loader) toolchain.
27284 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
27285 made to define a built-in function that has been disabled.
27286 (paired_init_builtins): Add assertion to prevent ICE if attempt is
27287 made to define a built-in function that has been disabled.
27288 (altivec_init_builtins): Add comment explaining why definition
27289 of the DST built-in functions is not preceded by an assertion
27290 check. Add assertions to prevent ICE if attempts are made to
27291 define an altivec predicate or an abs* built-in function that has
27293 (htm_init_builtins): Add comment explaining why definition of the
27294 htm built-in functions is not preceded by an assertion check.
27296 2017-01-04 Jeff Law <law@redhat.com>
27298 PR tree-optimizatin/67955
27299 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
27300 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
27301 the points-to solution does not include pt_null. Use DECL_PT_UID
27304 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
27306 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
27307 Use gen_int_mode instead of gen_lopwart for const_int operands.
27309 2017-01-04 Jakub Jelinek <jakub@redhat.com>
27311 PR tree-optimization/71563
27312 * match.pd: Simplify X << Y into X if Y is known to be 0 or
27313 out of range value - has low bits known to be zero.
27315 2017-01-04 Alan Modra <amodra@gmail.com>
27317 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
27318 * configure: Regenerate.
27319 * config.in: Regenerate.
27321 2017-01-04 Jakub Jelinek <jakub@redhat.com>
27324 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
27325 a substring of the message, but strcmp with the whole message. Ifdef
27326 ENABLE_NLS, translate the message first using dgettext.
27328 2017-01-03 Jeff Law <law@redhat.com>
27330 PR tree-optimizatin/78856
27331 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
27332 (mark_threaded_blocks): Remove code to truncate thread paths that
27333 cross multiple loop headers. Instead invalidate the cached loop
27334 iteration information and handle case of a thread path walking
27335 into an irreducible region.
27337 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
27340 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
27341 assertions. Add support for doing the signbit if the IEEE 128-bit
27342 floating point value is in a GPR.
27343 * config/rs6000/rs6000.md (Fsignbit): Delete.
27344 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
27345 Update the length attribute if the value is in a GPR.
27346 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
27347 the sign or zero extension instruction, since the value is always 0/1.
27348 (signbit<mode>2_dm2): Delete using <Fsignbit>.
27351 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
27352 extracting SImode to a GPR register so that we can generate a
27353 store, limit the vector to be in a traditional Altivec register
27354 for the vextuwrx instruction.
27356 2017-01-03 Ian Lance Taylor <iant@google.com>
27358 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
27360 2017-01-03 Martin Sebor <msebor@redhat.com>
27362 PR tree-optimization/78696
27363 * gimple-ssa-sprintf.c (format_floating): Correct handling of
27364 precision. Use MPFR for %f for greater fidelity. Correct handling
27366 (pass_sprintf_length::compute_format_length): Set width and precision
27367 specified by asrerisk to void_node for vararg functions.
27368 (try_substitute_return_value): Adjust dump output.
27370 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
27372 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
27374 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
27376 * doc/invoke.texi (SPARC options): Document -mlra as the default.
27377 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
27378 -mlra/-mno-lra was passed to the compiler.
27380 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
27382 PR rtl-optimization/65618
27383 * emit-rtl.c (try_split): Move initialization of "before" and
27384 "after" to just before the call to emit_insn_after_setloc.
27386 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
27388 * doc/md.texi (Standard Names): Remove reference to Java frontend.
27390 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
27392 * dwarf2out.c (gen_enumeration_type_die): When
27393 -gno-strict-dwarf, add a DW_AT_encoding attribute.
27395 2017-01-03 Jakub Jelinek <jakub@redhat.com>
27397 PR tree-optimization/78965
27398 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
27399 Change first argument from const call_info & to call_info &. For %n
27400 set info.nowrite to false.
27402 PR middle-end/78901
27403 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
27404 possibly throwing calls.
27406 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
27407 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
27408 and fns handling, rather than in a separate case SSA_NAME.
27410 2017-01-02 Jeff Law <law@redhat.com>
27412 * config/darwin-driver.c (darwin_driver_init): Const-correctness
27413 fixes for first_period and second_period variables.
27415 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
27418 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
27419 (*insvqi_1): New insn pattern.
27420 (*insvqi_1_mem_rex64): Ditto.
27421 (*insvqi_2): Ditto.
27422 (*insvqi_3): Rename from *insvqi.
27424 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
27426 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
27428 * doc/cfg.texi (Edges): Remove reference to Java.
27429 (Maintaining the CFG): Ditto.
27431 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
27433 PR middle-end/77674
27434 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
27435 transparent aliases.
27437 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
27439 PR middle-end/77484
27440 * predict.def (PRED_CALL): Update hitrate.
27441 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
27442 * predict.c (tree_estimate_probability_bb): Split CALL predictor
27443 into direct/indirect/polymorphic variants.
27445 2017-01-01 Jakub Jelinek <jakub@redhat.com>
27447 Update copyright years.
27449 * gcc.c (process_command): Update copyright notice dates.
27450 * gcov-dump.c (print_version): Ditto.
27451 * gcov.c (print_version): Ditto.
27452 * gcov-tool.c (print_version): Ditto.
27453 * gengtype.c (create_file): Ditto.
27454 * doc/cpp.texi: Bump @copying's copyright year.
27455 * doc/cppinternals.texi: Ditto.
27456 * doc/gcc.texi: Ditto.
27457 * doc/gccint.texi: Ditto.
27458 * doc/gcov.texi: Ditto.
27459 * doc/install.texi: Ditto.
27460 * doc/invoke.texi: Ditto.
27462 Copyright (C) 2017 Free Software Foundation, Inc.
27464 Copying and distribution of this file, with or without modification,
27465 are permitted in any medium without royalty provided the copyright
27466 notice and this notice are preserved.