1 2017-09-19 Jeff Law <law@redhat.com>
3 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
4 (ix86_expand_prologue): Dump stack clash info as needed.
5 Call ix86_adjust_stack_and_probe_stack_clash as needed.
7 * function.c (dump_stack_clash_frame_info): New function.
8 * function.h (dump_stack_clash_frame_info): Prototype.
9 (enum stack_clash_probes): New enum.
11 * config/alpha/alpha.c (alpha_expand_prologue): Also check
12 flag_stack_clash_protection.
13 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
14 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
15 (arm_frame_pointer_required): Likewise.
16 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
17 (ia64_expand_prologue): Likewise.
18 * config/mips/mips.c (mips_expand_prologue): Likewise.
19 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
20 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
21 (sparc_flat_expand_prologue): Likewise.
22 * config/spu/spu.c (spu_expand_prologue): Likewise.
24 * explow.c: Include "params.h".
25 (anti_adjust_stack_and_probe_stack_clash): New function.
26 (get_stack_check_protect): Likewise.
27 (compute_stack_clash_protection_loop_data): Likewise.
28 (emit_stack_clash_protection_loop_start): Likewise.
29 (emit_stack_clash_protection_loop_end): Likewise.
30 (allocate_dynamic_stack_space): Use get_stack_check_protect.
31 Use anti_adjust_stack_and_probe_stack_clash.
32 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
33 (emit_stack_clash_protection_loop_start): Likewise.
34 (emit_stack_clash_protection_loop_end): Likewise.
35 * rtl.h (get_stack_check_protect): Prototype.
36 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
37 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
38 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
40 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
42 * doc/tm.texi: Rebuilt.
43 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
44 get_stack_check_protect.
45 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
46 * config/arm/arm.c (arm_expand_prologue): Likewise.
47 (arm_frame_pointer_required): Likewise.
48 * config/i386/i386.c (ix86_expand_prologue): Likewise.
49 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
50 * config/mips/mips.c (mips_expand_prologue): Likewise.
51 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
52 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
53 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
54 (sparc_flat_expand_prologue): Likewise.
56 * common.opt (-fstack-clash-protection): New option.
57 * flag-types.h (enum stack_check_type): Note difference between
58 -fstack-check= and -fstack-clash-protection.
59 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
60 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
61 * toplev.c (process_options): Issue warnings/errors for cases
62 not handled with -fstack-clash-protection.
63 * doc/invoke.texi (-fstack-clash-protection): Document new option.
64 (-fstack-check): Note additional problem with -fstack-check=generic.
65 Note that -fstack-check is primarily for Ada and refer users
66 to -fstack-clash-protection for stack-clash-protection.
67 Document new params for stack clash protection.
69 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
71 * config/i386/i386.md (*scc_bt<mode>): New insn_and_split pattern.
72 (*scc_bt<mode>_1): Ditto.
73 (*scc_bt<mode>_mask): Ditto.
75 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
77 * config/i386/i386.c (ix86_split_long_move): Do not handle
78 address used for LEA in a special way.
80 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
82 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
84 2017-09-19 Martin Sebor <msebor@redhat.com>
87 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
88 of incompatible types.
90 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
92 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
93 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
94 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
95 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
97 2017-09-19 Richard Biener <rguenther@suse.de>
99 PR tree-optimization/82244
100 * tree-vrp.c (remove_range_assertions): Do not propagate
101 a constant to abnormals but replace the assert with a copy.
103 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
105 PR rtl-optimization/57878
106 PR rtl-optimization/68988
107 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
108 avoidance test involving non_reload_pseudos. Move frequency test
109 below the general fragmentation avoidance test.
111 2017-09-19 Richard Biener <rguenther@suse.de>
113 PR tree-optimization/69728
114 * graphite-sese-to-poly.c (schedule_error): New global.
115 (add_loop_schedule): Handle empty domain by failing the
117 (build_original_schedule): Handle schedule_error.
119 2017-09-19 Richard Biener <rguenther@suse.de>
121 * graphite-scop-detection.c (scop_detection::can_represent_loop):
122 Do not iterate to sibling loops but only to siblings of inner
125 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
128 * config/m68k/m68k.md (moveq feeding equality comparison): Check
129 that the registers are different.
131 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
133 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
134 to processor_model and "amdfam17h" to arch_names_table.
135 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
137 2017-09-18 Jakub Jelinek <jakub@redhat.com>
140 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
142 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
143 Alan Hayward <alan.hayward@arm.com>
144 David Sherwood <david.sherwood@arm.com>
146 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
148 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
150 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
151 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
152 vect_schedule_slp_instance.
153 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
154 Update call to vect_slp_analyze_node_operations. Simplify return
156 (vect_slp_analyze_bb_1): Update call accordingly.
157 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
158 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
159 (vect_schedule_slp): Update call accordingly.
161 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
162 Alan Hayward <alan.hayward@arm.com>
163 David Sherwood <david.sherwood@arm.com>
165 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
166 with types that aren't in fact scalar.
168 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
170 * tree-vect-slp.c (vect_record_max_nunits): New function,
172 (vect_build_slp_tree_1): ...here.
173 (vect_build_slp_tree_2): Call it for phis too.
175 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
177 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
178 to vect_get_vec_def_for_operand when getting the mask operand.
180 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
181 Alan Hayward <alan.hayward@arm.com>
182 David Sherwood <david.sherwood@arm.com>
184 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
187 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
188 Alan Hayward <alan.hayward@arm.com>
189 David Sherwood <david.sherwood@arm.com>
191 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
192 calculation for vector booleans.
194 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
195 Alan Hayward <alan.hayward@arm.com>
196 David Sherwood <david.sherwood@arm.com>
198 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
200 (vect_transform_stmt): ...here.
201 (vect_analyze_stmt): Use it instead of calling
202 vectorizable_live_operation directly.
204 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
206 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
207 non-SIMT targets in acc vector loops.
209 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
211 * configure.ac: Add arc and check if assembler supports gdwarf2.
212 * configure: Regenerate.
214 2017-09-18 Richard Biener <rguenther@suse.de>
216 PR tree-optimization/82220
217 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
218 epilogue niters from the min_profitable_iters compute.
220 2017-09-18 Jakub Jelinek <jakub@redhat.com>
223 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
224 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
225 (ix86_init_pic_reg): Revert 2017-09-01 changes.
227 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
230 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
231 switching to a new text section.
233 2017-09-18 Richard Biener <rguenther@suse.de>
235 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
237 (build_alias_set): Reject aliases with no access function.
239 2017-09-18 Richard Biener <rguenther@suse.de>
241 PR tree-optimization/79622
242 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
244 (build_cross_bb_scalars_use): Likewise.
246 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
248 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
250 2017-09-18 Alan Modra <amodra@gmail.com>
253 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
254 stack_pointer_rtx for count 0. Update comments. Break up
255 large rtl expression.
257 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
259 config/i386/i386.c: (xlogue_layout::STUB_NAME_MAX_LEN): Increase to 20
261 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
262 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
263 or avx version of the stub.
265 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
268 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
269 compute the minimum stack alignment. Also update preferred stack
270 boundary for leaf functions.
272 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
274 PR tree-optimization/82228
275 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
278 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
280 * common/config/nds32/nds32-common.c
281 (nds32_option_optimization_table): Refine formatting.
282 (nds32_option_optimization_table): Use -fsched-pressure and
283 -fomit-frame-pointer for specific optimization level.
285 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
287 * config/nds32/nds32.c: Refine formatting and comments.
288 * config/nds32/nds32.h: Likewise.
289 * config/nds32/nds32.md: Likewise.
290 * config/nds32/nds32-cost.c: Likewise.
291 * config/nds32/nds32-isr.c: Likewise.
292 * config/nds32/nds32-md-auxiliary.c: Likewise.
293 * config/nds32/nds32-multiple.md: Likewise.
294 * config/nds32/nds32-predicates.c: Likewise.
296 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
297 Jakub Jelinek <jakub@redhat.com>
299 Add support for -std=c++2a.
300 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
302 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
304 2017-09-15 Steve Ellcey <sellcey@cavium.com>
307 * doc/extend.texi (Common Function Attributes): Add
308 references to ARM, AArch64, and S/390 specific attributes.
309 (Function Specific Option Pragmas): Add AArch64 and S/390
310 to list of back ends that support the target pragma.
312 2017-09-15 Nathan Sidwell <nathan@acm.org>
314 * doc/standards.texi: Fix C++17 description. Update URLs for
317 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
319 * common.opt (Wcast-align=strict): New warning option.
320 * doc/invoke.texi: Document -Wcast-align=strict.
322 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
324 * cgraph.h (cgraph_thunk_info): Add comments.
325 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
326 assert for VIRTUAL_* arguments stricter.
328 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
330 PR tree-optimization/71026
331 * match.pd: Move RDIV patterns from fold-const.c
332 * fold-const.c (distribute_real_division): Removed.
333 (fold_binary_loc): Remove calls to distribute_real_divison.
335 2017-09-15 Jakub Jelinek <jakub@redhat.com>
337 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
338 c++1z and gnu++1z as deprecated. Change other references to
339 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
340 Change -Wc++1z-compat to -Wc++17-compat.
341 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
342 * dwarf2out.c (highest_c_language): Handle C++17.
343 (gen_compile_unit_die): Likewise.
345 2017-09-15 Jakub Jelinek <jakub@redhat.com>
347 PR rtl-optimization/82192
348 * combine.c (make_extraction): Don't look through non-paradoxical
349 SUBREGs or TRUNCATE if pos + len is or might be bigger than
352 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
353 Alan Hayward <alan.hayward@arm.com>
354 David Sherwood <david.sherwood@arm.com>
356 * target.def (function_arg_offset): New hook.
357 * targhooks.h (default_function_arg_offset): Declare.
358 * targhooks.c (default_function_arg_offset): New function.
359 * function.c (locate_and_pad_parm): Use
360 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
361 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
362 (TARGET_FUNCTION_ARG_OFFSET): ...this.
363 * doc/tm.texi: Regenerate.
364 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
365 * config/spu/spu.c (spu_function_arg_offset): New function.
366 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
367 * system.h (FUNCTION_ARG_OFFSET): Poison.
369 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
370 Alan Hayard <alan.hayward@arm.com>
371 David Sherwood <david.sherwood@arm.com>
373 * target.def (truly_noop_truncation): New hook.
374 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
375 than TRULY_NOOP_TRUNCATION.
376 * hooks.h (hook_bool_uint_uint_true): Declare.
377 * hooks.c (hook_bool_uint_uint_true): New function.
378 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
379 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
380 * doc/tm.texi: Regenerate.
381 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
382 rather than TRULY_NOOP_TRUNCATION in comments.
383 (simplify_comparison): Likewise.
384 (record_truncated_value): Likewise.
385 * expmed.c (extract_bit_field_1): Likewise.
386 (extract_split_bit_field): Likewise.
387 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
388 instead of TRULY_NOOP_TRUNCATION.
389 * function.c (assign_parm_setup_block): Likewise.
390 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
391 * rtlhooks.c: Include target.h.
392 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
393 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
394 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
395 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
396 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
397 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
398 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
399 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
400 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
401 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
402 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
403 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
404 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
405 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
406 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
407 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
408 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
409 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
410 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
411 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
412 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
413 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
414 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
415 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
416 * config/mips/mips.c (mips_truly_noop_truncation): New function.
417 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
418 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
419 rather than TRULY_NOOP_TRUNCATION in comments.
420 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
421 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
422 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
423 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
424 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
425 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
426 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
427 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
428 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
429 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
430 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
431 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
432 rather than TRULY_NOOP_TRUNCATION in comments.
433 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
434 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
435 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
436 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
437 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
438 TRULY_NOOP_TRUNCATION condition.
439 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
440 (TRULY_NOOP_TRUNCATION): Delete.
441 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
442 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
443 * config/spu/spu.c (spu_truly_noop_truncation): New function.
444 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
445 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
446 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
447 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
448 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
449 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
450 rather than TRULY_NOOP_TRUNCATION in comments.
451 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
452 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
453 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
454 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
455 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
456 * system.h (TRULY_NOOP_TRUNCATION): Poison.
458 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
461 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
462 (*cmp_ior): Likewise.
463 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
464 (*ior_scc_scc_cmp): Likewise.
465 (*and_scc_scc): Likewise.
466 (*and_scc_scc_cmp): Likewise.
468 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
469 Alan Hayard <alan.hayward@arm.com>
470 David Sherwood <david.sherwood@arm.com>
472 * target.def (can_change_mode_class): New hook.
473 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
474 (hard_regno_nregs): Likewise.
475 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
476 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
477 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
478 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
479 (LOAD_EXTEND_OP): Update accordingly.
480 * doc/tm.texi: Regenerate.
481 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
482 CANNOT_CHANGE_MODE_CLASS.
483 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
484 (REG_CAN_CHANGE_MODE_P): ...this new macro.
485 * combine.c (simplify_set): Update accordingly.
486 * emit-rtl.c (validate_subreg): Likewise.
487 * recog.c (general_operand): Likewise.
488 * regcprop.c (mode_change_ok): Likewise.
489 * reload1.c (choose_reload_regs): Likewise.
490 (inherit_piecemeal_p): Likewise.
491 * rtlanal.c (simplify_subreg_regno): Likewise.
492 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
493 instead of CANNOT_CHANGE_MODE_CLASS.
494 (reload_cse_simplify_operands): Likewise.
495 * reload.c (push_reload): Use targetm.can_change_mode_class
496 instead of CANNOT_CHANGE_MODE_CLASS.
497 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
498 REG_CANNOT_CHANGE_MODE_P.
499 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
500 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
501 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
502 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
503 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
504 (arm_can_change_mode_class): New function.
505 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
506 than CANNOT_CHANGE_MODE_CLASS in comments.
507 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
508 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
509 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
510 (ix86_can_change_mode_class): ...this new function, inverting the
511 sense of the return value.
512 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
513 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
514 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
515 (ia64_can_change_mode_class): New function.
516 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
517 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
518 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
519 (m32c_can_change_mode_class): ...this new function, inverting the
520 sense of the return value.
521 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
522 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
523 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
524 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
525 (mips_can_change_mode_class): ...this new function, inverting the
526 sense of the return value.
527 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
528 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
529 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
530 (msp430_can_change_mode_class): New function.
531 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
532 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
533 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
534 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
535 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
536 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
537 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
538 (pa_cannot_change_mode_class): Replace with...
539 (pa_can_change_mode_class): ...this new function, inverting the
540 sense of the return value.
541 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
542 than CANNOT_CHANGE_MODE_CLASS in comments.
543 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
544 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
545 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
546 (pdp11_cannot_change_mode_class): Replace with...
547 (pdp11_can_change_mode_class): ...this new function, inverting the
548 sense of the return value.
549 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
550 * config/powerpcspe/powerpcspe-protos.h
551 (rs6000_cannot_change_mode_class_ptr): Delete.
552 * config/powerpcspe/powerpcspe.c
553 (rs6000_cannot_change_mode_class_ptr): Delete.
554 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
555 (rs6000_option_override_internal): Assign to
556 targetm.can_change_mode_class instead of
557 rs6000_cannot_change_mode_class_ptr.
558 (rs6000_cannot_change_mode_class): Replace with...
559 (rs6000_can_change_mode_class): ...this new function, inverting the
560 sense of the return value.
561 (rs6000_debug_cannot_change_mode_class): Replace with...
562 (rs6000_debug_can_change_mode_class): ...this new function.
563 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
564 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
565 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
566 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
567 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
569 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
570 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
571 (rs6000_option_override_internal): Assign to
572 targetm.can_change_mode_class instead of
573 rs6000_cannot_change_mode_class_ptr.
574 (rs6000_cannot_change_mode_class): Replace with...
575 (rs6000_can_change_mode_class): ...this new function, inverting the
576 sense of the return value.
577 (rs6000_debug_cannot_change_mode_class): Replace with...
578 (rs6000_debug_can_change_mode_class): ...this new function.
579 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
580 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
581 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
582 (s390_can_change_mode_class): ...this new function, inverting the
583 sense of the return value.
584 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
585 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
586 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
587 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
588 (sh_cannot_change_mode_class): Replace with...
589 (sh_can_change_mode_class): ...this new function, inverting the
590 sense of the return value.
591 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
592 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
593 (sparc_can_change_mode_class): New function.
594 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
595 * config/spu/spu.c (spu_can_change_mode_class): New function.
596 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
597 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
598 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
599 (visium_can_change_mode_class): New function.
600 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
602 2017-09-15 Richard Biener <rguenther@suse.de>
604 PR tree-optimization/82217
605 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
606 but not undefined case.
608 2017-09-15 Jakub Jelinek <jakub@redhat.com>
611 * postreload.c (reload_cse_simplify_operands): Skip
612 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
614 2017-09-15 Richard Biener <rguenther@suse.de>
616 PR tree-optimization/68823
617 * graphite-scop-detection.c (build_alias_set): If we have a
618 possible dependence check whether we can handle them by just
619 looking at the DRs DR_ACCESS_FNs.
620 (build_scops): If build_alias_set fails, fail the SCOP.
622 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
624 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
625 to support float128 built-in functions that require the ISA 3.0
627 (BU_FLOAT128_3_HW): Likewise.
628 (SQRTF128): Add support for the IEEE 128-bit square root and fma
632 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
633 support for built-in functions that need the ISA 3.0 IEEE 128-bit
634 floating point instructions.
635 (rs6000_invalid_builtin): Likewise.
636 (rs6000_builtin_mask_names): Likewise.
637 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
638 (RS6000_BTM_FLOAT128_HW): Likewise.
639 (RS6000_BTM_COMMON): Likewise.
640 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
642 * doc/extend.texi (RS/6000 built-in functions): Document the
643 IEEE 128-bit floating point square root and fused multiply-add
646 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
648 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
649 reg (r2) isn't in the set of registers defined in the prologue.
651 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
652 Alan Hayward <alan.hayward@arm.com>
653 David Sherwood <david.sherwood@arm.com>
655 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
656 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
657 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
658 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
659 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
661 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
662 max_vectorization_factor.
663 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
665 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
666 Alan Hayward <alan.hayward@arm.com>
667 David Sherwood <david.sherwood@arm.com>
669 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
670 (vect_worthwhile_without_simd_p): Declare.
671 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
672 (vectorizable_reduction): Use it.
673 * tree-vect-stmts.c (vectorizable_shift): Likewise.
674 (vectorizable_operation): Likewise.
676 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
677 Alan Hayward <alan.hayward@arm.com>
678 David Sherwood <david.sherwood@arm.com>
680 * tree-vectorizer.h (vect_get_num_copies): New function.
681 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
682 * tree-vect-loop.c (vectorizable_reduction): Likewise.
683 (vectorizable_induction): Likewise.
684 (vectorizable_live_operation): Likewise.
685 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
686 (vectorizable_bswap): Likewise.
687 (vectorizable_call): Likewise.
688 (vectorizable_conversion): Likewise.
689 (vectorizable_assignment): Likewise.
690 (vectorizable_shift): Likewise.
691 (vectorizable_operation): Likewise.
692 (vectorizable_store): Likewise.
693 (vectorizable_load): Likewise.
694 (vectorizable_condition): Likewise.
695 (vectorizable_comparison): Likewise.
696 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
698 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
699 Alan Hayward <alan.hayward@arm.com>
700 David Sherwood <david.sherwood@arm.com>
702 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
705 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
706 Alan Hayward <alan.hayward@arm.com>
707 David Sherwood <david.sherwood@arm.com>
709 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
710 an inline wrapper that provides a location.
711 (gimple_build_vector): Likewise.
712 * gimple-fold.c (gimple_build_vector_from_val): New function.
713 (gimple_build_vector): Likewise.
714 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
715 functions to build the initial value. Always return a gimple value.
716 (get_initial_defs_for_reduction): Likewise. Only compute
718 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
719 vect_init_vector on the results from get_initial_def(s)_for_reduction.
720 (vectorizable_induction): Use gimple_build_vector rather than
723 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
724 Alan Hayward <alan.hayward@arm.com>
725 David Sherwood <david.sherwood@arm.com>
727 * target.h (vec_perm_indices): New typedef.
728 (auto_vec_perm_indices): Likewise.
729 * optabs-query.h: Include target.h
730 (can_vec_perm_p): Take a vec_perm_indices *.
731 * optabs-query.c (can_vec_perm_p): Likewise.
732 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
733 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
734 * tree-vect-generic.c (lower_vec_perm): Likewise.
735 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
736 (vect_grouped_load_supported): Likewise.
737 (vect_shift_permute_load_chain): Likewise.
738 (vect_permute_store_chain): Use auto_vec_perm_indices.
739 (vect_permute_load_chain): Likewise.
740 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
741 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
742 Update uses of can_vec_perm_p.
743 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
744 mode with a number of elements. Take a vec_perm_indices *.
745 (vect_create_epilog_for_reduction): Update accordingly.
746 Use auto_vec_perm_indices.
747 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
748 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
749 (vect_transform_slp_perm_load): Likewise.
750 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
751 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
752 (vect_gen_perm_mask_checked): Likewise.
753 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
754 (vect_gen_perm_mask_checked): Likewise.
755 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
756 (vectorizable_store): Likewise.
757 (vectorizable_load): Likewise.
758 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
759 (vectorizable_bswap): Likewise.
761 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
762 Alan Hayward <alan.hayward@arm.com>
763 David Sherwood <david.sherwood@arm.com>
765 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
766 * tree.c (build_vector): Likewise.
767 (build_vector_from_ctor): Update accordingly.
768 (build_vector_from_val): Likewise.
769 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
770 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
771 * tree-vect-generic.c (add_rshift): Likewise.
772 (expand_vector_divmod): Likewise.
773 (optimize_vector_constructor): Likewise.
774 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
775 (vect_transform_slp_perm_load): Likewise.
776 (vect_schedule_slp_instance): Likewise.
777 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
778 (vectorizable_call): Likewise.
779 (vect_gen_perm_mask_any): Likewise. Add elements in order.
780 * expmed.c (make_tree): Likewise.
781 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
782 a vector passed to build_vector.
783 (fold_convert_const): Likewise.
784 (exact_inverse): Likewise.
785 (fold_ternary_loc): Likewise.
786 (fold_relational_const): Likewise.
787 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
788 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
789 (const_unop): Likewise. Store the reduction accumulator in a
790 variable rather than an array.
791 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
792 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
793 the new vector, rather than constructing it after the input arrays.
794 (native_interpret_vector): Use auto_vec<tree> when building
795 a vector passed to build_vector. Add elements in order.
796 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
797 auto_vec<tree> when building a vector passed to build_vector.
798 (vect_create_epilog_for_reduction): Likewise.
799 (vectorizable_induction): Likewise.
800 (get_initial_def_for_reduction): Likewise. Fix indentation of
802 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
804 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
805 passed to build_vector.
807 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
808 Alan Hayward <alan.hayward@arm.com>
809 David Sherwood <david.sherwood@arm.com>
811 * tree-core.h (tree_base::u): Add an "nelts" field.
812 (tree_vector): Use VECTOR_CST_NELTS as the length.
813 * tree.c (tree_size): Likewise.
814 (make_vector): Initialize VECTOR_CST_NELTS.
815 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
816 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
817 TYPE_VECTOR_SUBPARTS.
818 * expr.c (const_vector_mask_from_tree): Consistently use "units"
819 as the number of units, setting it from VECTOR_CST_NELTS.
820 (const_vector_from_tree): Likewise.
821 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
822 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
823 (fold_negate_expr_1): Likewise.
824 (fold_convert_const): Likewise.
825 (const_binop): Likewise. Differentiate the number of output and
827 (const_unop): Likewise.
828 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
829 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
830 in cases that did the opposite.
832 2017-09-14 Richard Biener <rguenther@suse.de>
834 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
837 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
839 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
841 2017-09-14 Jakub Jelinek <jakub@redhat.com>
844 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
845 if and where to split a bb, except for splitting before debug insn
846 sequences followed by non-label real insn. Delete debug insns
847 in between basic blocks.
849 * combine.c (make_compound_operation_int): Formatting fixes.
851 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
852 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
853 * config/netbsd.h (LINK_EH_SPEC): Likewise.
854 * config/sol2.h (LINK_EH_SPEC): Likewise.
855 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
856 * config/s390/linux.h (LINK_SPEC): Likewise.
857 * config/freebsd.h (LINK_EH_SPEC): Likewise.
858 * config/openbsd.h (LINK_EH_SPEC): Likewise.
859 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
860 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
861 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
862 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
863 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
864 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
866 2017-09-13 Jakub Jelinek <jakub@redhat.com>
868 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
870 (ENDFILE_LINUX_SPEC): Likewise.
871 (LINK_EH_SPEC): Likewise.
872 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
873 (LINK_OS_LINUX_SPEC32): Likewise.
874 (LINK_OS_LINUX_SPEC64): Likewise.
875 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
876 (LINK_OS_LINUX_SPEC): Likewise.
878 2017-09-13 Martin Liska <mliska@suse.cz>
881 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
882 CASE_HIGH is NULL_TREE.
884 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
885 Alan Hayward <alan.hayward@arm.com>
886 David Sherwood <david.sherwood@arm.com>
888 * target.def (secondary_memory_needed): New hook.
889 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
890 instead of SECONDARY_MEMORY_NEEDED.
891 (secondary_memory_needed_mode): Likewise.
892 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
893 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
894 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
895 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
896 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
897 * doc/tm.texi: Regenerate.
898 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
899 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
900 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
901 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
902 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
903 * config/i386/i386.c (inline_secondary_memory_needed): Put the
904 mode argument first and change the reg_class arguments to reg_class_t.
905 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
906 Make static. Update the call to inline_secondary_memory_needed.
907 (ix86_register_move_cost): Update the call to
908 inline_secondary_memory_needed.
909 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
910 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
912 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
913 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
915 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
916 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
917 * config/mips/mips.c (mips_secondary_memory_needed): Make static
918 and match hook interface. Add comment from mips.h.
919 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
920 * config/mmix/mmix.md (truncdfsf2): Refer to
921 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
923 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
924 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
925 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
926 (pa_secondary_memory_needed): New function.
927 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
928 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
929 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
930 (pdp11_secondary_memory_needed): Make static and match hook interface.
931 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
932 * config/powerpcspe/powerpcspe-protos.h
933 (rs6000_secondary_memory_needed_ptr): Delete.
934 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
936 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
937 (rs6000_option_override_internal): Assign to
938 targetm.secondary_memory_needed rather than
939 rs6000_secondary_memory_needed_ptr.
940 (rs6000_secondary_memory_needed): Match hook interface.
941 (rs6000_debug_secondary_memory_needed): Likewise.
942 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
943 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
944 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
945 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
946 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
947 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
949 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
950 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
951 (rs6000_option_override_internal): Assign to
952 targetm.secondary_memory_needed rather than
953 rs6000_secondary_memory_needed_ptr.
954 (rs6000_secondary_memory_needed): Match hook interface.
955 (rs6000_debug_secondary_memory_needed): Likewise.
956 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
957 * config/s390/s390.c (s390_secondary_memory_needed): New function.
958 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
959 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
960 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
961 (sparc_secondary_memory_needed): New function.
962 * lra-constraints.c (check_and_process_move): Refer to
963 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
965 (curr_insn_transform): Likewise.
966 (process_alt_operands): Use targetm.secondary_memory_needed
967 instead of TARGET_SECONDARY_MEMORY_NEEDED.
968 (check_secondary_memory_needed_p): Likewise.
969 (choose_split_class): Likewise.
970 * reload.c: Unconditionally include code that was previously
971 conditional on SECONDARY_MEMORY_NEEDED.
972 (push_secondary_reload): Use targetm.secondary_memory_needed
973 instead of TARGET_SECONDARY_MEMORY_NEEDED.
974 (push_reload): Likewise.
975 * reload1.c: Unconditionally include code that was previously
976 conditional on SECONDARY_MEMORY_NEEDED.
977 (choose_reload_regs): Use targetm.secondary_memory_needed
978 instead of TARGET_SECONDARY_MEMORY_NEEDED.
979 (gen_reload): Likewise.
980 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
982 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
983 Alan Hayward <alan.hayward@arm.com>
984 David Sherwood <david.sherwood@arm.com>
986 * target.def (secondary_memory_needed_mode): New hook:
987 * targhooks.c (default_secondary_memory_needed_mode): Declare.
988 * targhooks.h (default_secondary_memory_needed_mode): New function.
989 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
990 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
991 * doc/tm.texi: Regenerate.
992 * lra-constraints.c (check_and_process_move): Use
993 targetm.secondary_memory_needed_mode instead of
994 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
995 (curr_insn_transform): Likewise.
996 * reload.c (get_secondary_mem): Likewise.
997 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
998 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
1000 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
1001 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
1002 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
1003 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
1004 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
1006 * config/powerpcspe/powerpcspe-protos.h
1007 (rs6000_secondary_memory_needed_mode): Delete.
1008 * config/powerpcspe/powerpcspe.c
1009 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
1010 (rs6000_secondary_memory_needed_mode): Make static.
1011 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
1012 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
1014 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
1016 (rs6000_secondary_memory_needed_mode): Make static.
1017 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
1018 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
1019 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
1020 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
1021 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
1023 (sparc_secondary_memory_needed_mode): New function.
1024 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
1026 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
1028 * config/aarch64/constraints.md (Umq): New constraint.
1029 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
1031 (mov<mode>): Update condition.
1033 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1035 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
1036 when bitposition is the same.
1038 2017-09-13 Richard Biener <rguenther@suse.de>
1040 * dwarf2out.c (output_die_symbol): Remove.
1041 (output_die): Do not output a DIEs symbol.
1043 2017-09-13 Richard Biener <rguenther@suse.de>
1046 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
1047 default-def to avoid breaking iterator update with the weird
1048 interaction with cgraph_update_edges_for_call_stmt_node.
1050 2017-09-13 Richard Biener <rguenther@suse.de>
1052 * tree-cfg.c (verify_gimple_assign_binary): Add verification
1053 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
1054 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
1055 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
1057 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
1059 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
1060 Disable pc relative literal load irrespective of
1061 TARGET_FIX_ERR_A53_84341 for default.
1063 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
1065 * config/sparc/sparc.c (output_return): Output the source location of
1066 the insn in the delay slot, if any.
1067 (output_sibcall): Likewise.
1069 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
1072 * common.opt (-static-pie): New alias.
1073 (shared): Negate static-pie.
1074 (-no-pie): Update help text.
1076 (static-pie): New option.
1077 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
1078 -static-pie support.
1079 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
1080 (LINK_EH_SPEC): Likewise.
1081 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
1082 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
1083 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
1084 * gcc.c (LINK_COMMAND_SPEC): Likewise.
1085 (init_gcc_specs): Likewise.
1086 (init_spec): Likewise.
1087 (display_help): Update help message for -pie.
1088 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
1091 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
1093 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
1094 (movdi_aarch64): Likewise.
1095 (movti_aarch64): Likewise.
1097 2017-09-12 Simon Wright <simon@pushface.org>
1100 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
1101 calculation of the minor version, always output as 0.
1103 2017-09-12 Jakub Jelinek <jakub@redhat.com>
1106 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
1107 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
1108 on it early, rather than manual conversion late. For
1109 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
1110 instead of performing manual conversion.
1112 2017-09-12 Carl Love <cel@us.ibm.com>
1114 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
1115 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
1117 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
1118 VMULOSW): Add definitions.
1119 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
1120 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
1121 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
1122 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
1123 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
1125 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
1127 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
1129 (movti_aarch64): Likewise.
1130 (movdf_aarch64): Likewise.
1131 (movtf_aarch64): Likewise.
1132 (load_pairdi): Likewise.
1133 (store_pairdi): Likewise.
1134 (load_pairdf): Likewise.
1135 (store_pairdf): Likewise.
1136 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
1137 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
1138 (ldr_got_small_<mode>): Likewise.
1139 (ldr_got_small_28k_<mode>): Likewise.
1140 (ldr_got_tiny): Likewise.
1141 * config/aarch64/iterators.md (ldst_sz): New.
1142 (ldpstp_sz): Likewise.
1143 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
1145 (thunderx_load): Split load_8 to load_16.
1146 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
1148 (thunderx2t99_storepair_basic): Split store_8 to store_16.
1149 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
1150 (xgene1_store_pair): Split store_8 to store_16.
1151 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
1152 (falkor_st_0_st_sd): Split store_8 to store_16.
1154 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
1156 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
1157 and store1/2/3/4 to store_4/8/12/16.
1158 * config/aarch64/aarch64.md: Update for rename.
1159 * config/arm/arm.md: Likewise.: Likewise.
1160 * config/arm/arm.c: Likewise.
1161 * config/arm/thumb1.md: Likewise.
1162 * config/arm/thumb2.md: Likewise.
1163 * config/arm/vfp.md: Likewise.
1164 * config/arm/arm-generic.md: Likewise.
1165 * config/arm/arm1020e.md: Likewise.
1166 * config/arm/arm1026ejs.md: Likewise.
1167 * config/arm/arm1136jfs.md: Likewise.
1168 * config/arm/arm926ejs.md: Likewise.
1169 * config/arm/cortex-a15.md: Likewise.
1170 * config/arm/cortex-a17.md: Likewise.
1171 * config/arm/cortex-a5.md: Likewise.
1172 * config/arm/cortex-a53.md: Likewise.
1173 * config/arm/cortex-a57.md: Likewise.
1174 * config/arm/cortex-a7.md: Likewise.
1175 * config/arm/cortex-a8.md: Likewise.
1176 * config/arm/cortex-a9.md: Likewise.
1177 * config/arm/cortex-m4.md: Likewise.
1178 * config/arm/cortex-m7.md: Likewise.
1179 * config/arm/cortex-r4.md: Likewise.
1180 * config/arm/exynos-m1.md: Likewise.
1181 * config/arm/fa526.md: Likewise.
1182 * config/arm/fa606te.md: Likewise.
1183 * config/arm/fa626te.md: Likewise.
1184 * config/arm/fa726te.md: Likewise.
1185 * config/arm/fmp626.md: Likewise.
1186 * config/arm/iwmmxt.md: Likewise.
1187 * config/arm/ldmstm.md: Likewise.
1188 * config/arm/marvell-pj4.md: Likewise.
1189 * config/arm/xgene1.md: Likewise.
1190 * config/aarch64/thunderx.md: Likewise.
1191 * config/aarch64/thunderx2t99.md: Likewise.
1192 * config/aarch64/falkor.md: Likewise.
1194 2017-09-12 Martin Liska <mliska@suse.cz>
1196 * attribs.c (private_lookup_attribute): New function.
1197 * attribs.h (private_lookup_attribute): Declared here.
1198 (lookup_attribute): Called from this place.
1200 2017-09-12 Richard Biener <rguenther@suse.de>
1202 PR tree-optimization/82157
1203 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
1204 stmts with side-effects.
1206 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1207 Alan Hayward <alan.hayward@arm.com>
1208 David Sherwood <david.sherwood@arm.com>
1210 * target.def (hard_regno_nregs): New hook.
1211 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
1212 * targhooks.h (default_hard_regno_nregs): Declare.
1213 * targhooks.c (default_hard_regno_nregs): New function.
1214 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
1215 (TARGET_HARD_REGNO_NREGS): ...this hook.
1216 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
1217 (CLASS_MAX_NREGS): Likewise.
1218 * doc/tm.texi: Regenerate.
1219 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
1220 instead of HARD_REGNO_NREGS.
1221 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
1222 HARD_REGNO_NREGS in the comment.
1223 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
1224 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
1225 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
1226 Return an unsigned int.
1227 (TARGET_HARD_REGNO_NREGS): Redefine.
1228 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
1229 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
1230 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
1231 (arc_hard_regno_nregs): New function.
1232 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
1233 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
1234 (arm_hard_regno_nregs): New function.
1235 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
1236 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
1237 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
1238 (TARGET_HARD_REGNO_NREGS): Redefine.
1239 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
1240 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
1241 (HARD_REGNO_NREGS): Delete.
1242 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
1243 (cr16_hard_regno_nregs): New function.
1244 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
1245 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
1246 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
1247 (cris_hard_regno_nregs): New function.
1248 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
1249 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
1250 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
1251 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
1252 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
1253 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
1254 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
1255 (frv_hard_regno_nregs): Make static. Take and return an
1257 (frv_class_max_nregs): Remove outdated copy of documentation.
1258 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
1259 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
1260 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
1261 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
1262 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
1263 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
1264 (TARGET_HARD_REGNO_NREGS): Redefine.
1265 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
1266 (CLASS_MAX_NREGS): Update comment.
1267 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
1268 (ia64_hard_regno_nregs): New function.
1269 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
1270 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
1271 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
1272 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
1273 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
1275 (m32c_hard_regno_nregs): Likewise. Make static.
1276 (TARGET_HARD_REGNO_NREGS): Redefine.
1277 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
1278 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
1279 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
1280 (m68k_hard_regno_nregs): New function.
1281 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
1282 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
1283 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
1284 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
1285 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
1286 Take and return an unsigned int.
1287 (TARGET_HARD_REGNO_NREGS): Redefine.
1288 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
1289 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
1290 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
1291 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
1292 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
1293 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
1294 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
1295 (msp430_hard_regno_nregs): Make static. Take and return an
1297 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
1298 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
1299 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
1300 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
1301 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
1302 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
1303 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
1304 (TARGET_HARD_REGNO_NREGS): Redefine.
1305 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
1306 (PA_HARD_REGNO_NREGS): ...this.
1307 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
1308 (PA_HARD_REGNO_NREGS): ...this.
1309 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
1310 (pa_hard_regno_nregs): New function.
1311 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
1312 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
1313 (pdp11_hard_regno_nregs): New function.
1314 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
1315 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
1316 (rs6000_hard_regno_nregs_hook): New function.
1317 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
1318 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
1319 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
1320 Take and return an unsigned int. Move earlier in file.
1321 (TARGET_HARD_REGNO_NREGS): Redefine.
1322 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
1323 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
1324 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
1325 (rl78_hard_regno_nregs): Make static. Take and return an
1327 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
1328 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
1329 (rs6000_hard_regno_nregs_hook): New function.
1330 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
1331 * config/rx/rx.c (rx_hard_regno_nregs): New function.
1332 (TARGET_HARD_REGNO_NREGS): Redefine.
1333 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
1334 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
1335 instead of HARD_REGNO_NREGS.
1336 (s390_hard_regno_nregs): New function.
1337 (s390_hard_regno_mode_ok): Add comment from s390.h.
1338 (TARGET_HARD_REGNO_NREGS): Redefine.
1339 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
1340 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
1341 (sh_hard_regno_nregs): New function.
1342 (sh_pass_in_reg_p): Use it.
1343 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
1344 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
1345 (sparc_hard_regno_nregs): New function.
1346 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
1347 * config/spu/spu.c (spu_hard_regno_nregs): New function.
1348 (spu_function_arg_advance): Use it, supplying a valid register number.
1349 (TARGET_HARD_REGNO_NREGS): Redefine.
1350 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
1351 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
1352 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
1353 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
1354 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
1355 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
1356 (CLASS_MAX_NREGS): Remove copy of old documentation.
1357 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
1358 (visium_hard_regno_nregs): New function.
1359 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
1360 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
1361 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
1362 xtensa_hard_regno_nregs): New function.
1363 * system.h (HARD_REGNO_NREGS): Poison.
1365 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1367 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
1368 hard_regno_nregs instead of HARD_REGNO_NREGS.
1369 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
1370 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
1371 (c6x_expand_epilogue): Likewise.
1372 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
1373 (frv_read_iacc_argument): Likewise.
1374 * config/sh/sh.c: Include regs.h.
1375 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
1376 (regs_used): Likewise.
1377 (output_stack_adjust): Likewise.
1378 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
1379 * expmed.c: Include regs.h.
1380 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
1381 * ree.c: Include regs.h.
1382 (combine_reaching_defs): Use hard_regno_nregs instead of
1384 (add_removable_extension): Likewise.
1386 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1388 * regs.h (hard_regno_nregs): Turn into a function.
1389 (end_hard_regno): Update accordingly.
1390 * caller-save.c (setup_save_areas): Likewise.
1391 (save_call_clobbered_regs): Likewise.
1392 (replace_reg_with_saved_mem): Likewise.
1393 (insert_restore): Likewise.
1394 (insert_save): Likewise.
1395 * combine.c (can_change_dest_mode): Likewise.
1396 (move_deaths): Likewise.
1397 (distribute_notes): Likewise.
1398 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
1399 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
1400 (rs6000_split_multireg_move): Likewise.
1401 (rs6000_register_move_cost): Likewise.
1402 (rs6000_memory_move_cost): Likewise.
1403 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
1404 (rs6000_split_multireg_move): Likewise.
1405 (rs6000_register_move_cost): Likewise.
1406 (rs6000_memory_move_cost): Likewise.
1407 * cselib.c (cselib_reset_table): Likewise.
1408 (cselib_lookup_1): Likewise.
1409 * emit-rtl.c (set_mode_and_regno): Likewise.
1410 * function.c (aggregate_value_p): Likewise.
1411 * ira-color.c (setup_profitable_hard_regs): Likewise.
1412 (check_hard_reg_p): Likewise.
1413 (calculate_saved_nregs): Likewise.
1414 (assign_hard_reg): Likewise.
1415 (improve_allocation): Likewise.
1416 (calculate_spill_cost): Likewise.
1417 * ira-emit.c (modify_move_list): Likewise.
1418 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
1419 (ira_hard_reg_in_set_p): Likewise.
1420 * ira.c (setup_reg_mode_hard_regset): Likewise.
1421 (clarify_prohibited_class_mode_regs): Likewise.
1422 (check_allocation): Likewise.
1423 * lra-assigns.c (find_hard_regno_for_1): Likewise.
1424 (lra_setup_reg_renumber): Likewise.
1425 (setup_try_hard_regno_pseudos): Likewise.
1426 (spill_for): Likewise.
1427 (assign_hard_regno): Likewise.
1428 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
1429 * lra-constraints.c (in_class_p): Likewise.
1430 (lra_constraint_offset): Likewise.
1431 (simplify_operand_subreg): Likewise.
1432 (lra_constraints): Likewise.
1433 (split_reg): Likewise.
1434 (split_if_necessary): Likewise.
1435 (invariant_p): Likewise.
1436 (inherit_in_ebb): Likewise.
1437 * lra-lives.c (process_bb_lives): Likewise.
1438 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
1439 (get_hard_regs): Likewise.
1440 (do_remat): Likewise.
1441 * lra-spills.c (assign_spill_hard_regs): Likewise.
1442 * mode-switching.c (create_pre_exit): Likewise.
1443 * postreload.c (reload_combine_recognize_pattern): Likewise.
1444 * recog.c (peep2_find_free_register): Likewise.
1445 * regcprop.c (kill_value_regno): Likewise.
1446 (set_value_regno): Likewise.
1447 (copy_value): Likewise.
1448 (maybe_mode_change): Likewise.
1449 (find_oldest_value_reg): Likewise.
1450 (copyprop_hardreg_forward_1): Likewise.
1451 * regrename.c (check_new_reg_p): Likewise.
1452 (regrename_do_replace): Likewise.
1453 * reload.c (push_reload): Likewise.
1454 (combine_reloads): Likewise.
1455 (find_dummy_reload): Likewise.
1456 (operands_match_p): Likewise.
1457 (find_reloads): Likewise.
1458 (find_equiv_reg): Likewise.
1459 (reload_adjust_reg_for_mode): Likewise.
1460 * reload1.c (count_pseudo): Likewise.
1461 (count_spilled_pseudo): Likewise.
1462 (find_reg): Likewise.
1463 (clear_reload_reg_in_use): Likewise.
1464 (free_for_value_p): Likewise.
1465 (allocate_reload_reg): Likewise.
1466 (choose_reload_regs): Likewise.
1467 (reload_adjust_reg_for_temp): Likewise.
1468 (emit_reload_insns): Likewise.
1469 (delete_output_reload): Likewise.
1470 * rtlanal.c (subreg_get_info): Likewise.
1471 * sched-deps.c (sched_analyze_reg): Likewise.
1472 * sel-sched.c (init_regs_for_mode): Likewise.
1473 (mark_unavailable_hard_regs): Likewise.
1474 (choose_best_reg_1): Likewise.
1475 (verify_target_availability): Likewise.
1476 * valtrack.c (dead_debug_insert_temp): Likewise.
1477 * var-tracking.c (track_loc_p): Likewise.
1478 (emit_note_insn_var_location): Likewise.
1479 * varasm.c (make_decl_rtl): Likewise.
1480 * reginfo.c (choose_hard_reg_mode): Likewise.
1481 (init_reg_modes_target): Refer directly to
1482 this_target_regs->x_hard_regno_nregs.
1484 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1486 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
1487 instead of hard_regno_nregs.
1489 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1491 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
1492 end_hard_regno instead of hard_regno_nregs.
1493 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
1494 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
1495 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
1496 * ira-color.c (improve_allocation): Likewise.
1497 * lra-assigns.c (find_hard_regno_for_1): Likewise.
1498 * lra-lives.c (mark_regno_live): Likewise.
1499 (mark_regno_dead): Likewise.
1500 * lra-remat.c (operand_to_remat): Likewise.
1501 * lra.c (collect_non_operand_hard_regs): Likewise.
1502 * postreload.c (reload_combine_note_store): Likewise.
1503 (move2add_valid_value_p): Likewise.
1504 * reload.c (regno_clobbered_p): Likewise.
1506 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1508 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
1510 * config/v850/v850.c (v850_reorg): Likewise.
1511 * reload.c (refers_to_regno_for_reload_p): Likewise.
1512 (find_equiv_reg): Likewise.
1513 * reload1.c (reload_reg_reaches_end_p): Likewise.
1515 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1517 * caller-save.c (add_used_regs): Use REG_NREGS instead of
1519 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
1520 * config/arm/arm.c (output_move_neon): Likewise.
1521 (arm_attr_length_move_neon): Likewise.
1522 (neon_split_vcombine): Likewise.
1523 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
1524 (c6x_mark_reg_written): Likewise.
1525 (c6x_dwarf_register_span): Likewise.
1526 * config/i386/i386.c (ix86_save_reg): Likewise.
1527 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
1528 (rws_access_reg): Likewise.
1529 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
1530 * mode-switching.c (create_pre_exit): Likewise.
1531 * ree.c (combine_reaching_defs): Likewise.
1532 (add_removable_extension): Likewise.
1533 * regcprop.c (find_oldest_value_reg): Likewise.
1534 (copyprop_hardreg_forward_1): Likewise.
1535 * reload.c (reload_inner_reg_of_subreg): Likewise.
1536 (push_reload): Likewise.
1537 (combine_reloads): Likewise.
1538 (find_dummy_reload): Likewise.
1539 (reload_adjust_reg_for_mode): Likewise.
1540 * reload1.c (find_reload_regs): Likewise.
1541 (forget_old_reloads_1): Likewise.
1542 (reload_reg_free_for_value_p): Likewise.
1543 (reload_adjust_reg_for_temp): Likewise.
1544 (emit_reload_insns): Likewise.
1545 (delete_output_reload): Likewise.
1546 * sel-sched.c (choose_best_reg_1): Likewise.
1547 (choose_best_pseudo_reg): Likewise.
1549 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1550 Alan Hayward <alan.hayward@arm.com>
1551 David Sherwood <david.sherwood@arm.com>
1553 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
1554 * target.def (slow_unaligned_access): New hook.
1555 * targhooks.h (default_slow_unaligned_access): Declare.
1556 * targhooks.c (default_slow_unaligned_access): New function.
1557 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
1558 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
1559 * doc/tm.texi: Regenerate.
1560 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
1561 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
1562 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
1564 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
1565 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
1567 (rs6000_slow_unaligned_access): New function.
1568 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
1569 (expand_block_compare): Likewise.
1570 (expand_strn_compare): Likewise.
1571 (rs6000_rtx_costs): Likewise.
1572 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
1573 (riscv_slow_unaligned_access): Likewise.
1574 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
1575 (riscv_slow_unaligned_access_p): ...this and make static.
1576 (riscv_option_override): Update accordingly.
1577 (riscv_slow_unaligned_access): New function.
1578 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
1579 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
1580 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
1581 (rs6000_slow_unaligned_access): New function.
1582 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
1583 (rs6000_rtx_costs): Likewise.
1584 * config/rs6000/rs6000-string.c (expand_block_compare)
1585 (expand_strn_compare): Use targetm.slow_unaligned_access instead
1586 of SLOW_UNALIGNED_ACCESS.
1587 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
1588 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
1589 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
1590 of SLOW_UNALIGNED_ACCESS.
1591 * expmed.c (simple_mem_bitfield_p): Likewise.
1592 * expr.c (alignment_for_piecewise_move): Likewise.
1593 (emit_group_load_1): Likewise.
1594 (emit_group_store): Likewise.
1595 (copy_blkmode_from_reg): Likewise.
1596 (emit_push_insn): Likewise.
1597 (expand_assignment): Likewise.
1598 (store_field): Likewise.
1599 (expand_expr_real_1): Likewise.
1600 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
1601 * lra-constraints.c (simplify_operand_subreg): Likewise.
1602 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
1603 * gimple-ssa-store-merging.c: Likewise in block comment at start
1605 * tree-ssa-strlen.c: Include target.h.
1606 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
1607 of SLOW_UNALIGNED_ACCESS.
1608 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
1610 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1612 PR rtl-optimization/82185
1613 * expmed.c (emit_store_flag_int): Only test tem if it has been
1616 2017-09-12 Richard Biener <rguenther@suse.de>
1619 * match.pd ((FTYPE) N CMP CST): Fix typo.
1621 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
1623 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
1625 (mips_near_type_p): Add 'short_call' attribute as a synonym
1627 * doc/extend.texi (short_call): Document new function attribute.
1629 2017-09-12 Jakub Jelinek <jakub@redhat.com>
1632 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
1633 assertion check that in the condition.
1634 (get_atomic_generic_size): Likewise. Before testing if parameter
1635 has pointer type, if it has array type, call for C++
1636 default_conversion to perform array-to-pointer conversion.
1638 2017-09-12 Richard Biener <rguenther@suse.de>
1640 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
1641 for operations we cannot scalarize.
1643 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
1645 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
1646 vectors heap vectors. Clean up comments.
1647 Make visited_bbs a reference.
1648 (profitable_jump_thread_path): Make GC
1649 vectors heap vectors. Clean up comments.
1651 (convert_and_register_jump_thread_path): Make GC vectors heap
1653 (check_subpath_and_update_thread_path): Same. Clean up comments.
1654 Make visited_bbs a reference.
1655 (handle_phi): Abstract common code to to
1656 register_jump_thread_path_if_profitable.
1657 Rename VAR_BB to DEF_BB.
1659 Make GC vectors heap vectors.
1660 Make visited_bbs a reference.
1661 (handle_assignment): Same.
1662 (register_jump_thread_path_if_profitable): New.
1663 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
1665 Make GC vectors heap vectors. Clean up comments.
1666 Make visited_bbs a reference.
1667 (find_jump_threads_backwards): Make visited_bbs live in the stack.
1668 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
1671 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
1674 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
1675 words of E_DImode object are reachable by xtensa_uimm8x4 access.
1677 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
1679 Revert r251800 and r251799.
1681 2017-09-11 Martin Jambor <mjambor@suse.cz>
1684 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
1685 arguments in advance.
1686 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
1687 use it to find predecessor edges.
1688 (naive_outof_ssa): Collect vector of predecessors.
1690 2017-09-08 Jason Merrill <jason@redhat.com>
1692 PR c++/70029 - ICE with ref-qualifier and -flto
1693 * langhooks.h (struct lang_hooks_for_types): Add
1694 copy_lang_qualifiers.
1695 * attribs.c (build_type_attribute_qual_variant): Use it.
1696 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
1698 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
1699 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
1701 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
1704 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
1705 (*mulsi3_sp64): New instruction.
1706 (mulsi3): New expander.
1708 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
1710 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
1712 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1714 * sancov.c: Include memmodel.h.
1716 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
1719 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
1722 2017-09-07 Carl Love <cel@us.ibm.com>
1724 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
1725 the sldi instruction.
1727 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
1729 * sancov.c: Include tm_p.h.
1731 2017-09-07 Jakub Jelinek <jakub@redhat.com>
1734 * output.h (switch_to_other_text_partition): New declaration.
1735 * varasm.c (switch_to_other_text_partition): New function.
1736 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
1737 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
1738 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
1739 to the other text partition before emitting LCL label and switch back
1740 after emitting the word after it.
1742 2017-09-07 Richard Biener <rguenther@suse.de>
1744 * passes.def (pass_split_crit_edges): Remove instance before PRE.
1745 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
1746 critical edges here, after loop init.
1747 (pass_data_pre): Remove PROP_no_crit_edges flags.
1748 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
1749 for valueization of call args to avoid leaking VN_TOP.
1750 (visit_use): Assert we do not visit default defs.
1751 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
1752 Use error_mark_node to more easily detect leaking VN_TOP.
1753 All default-defs are varying, not VN_TOP. Mark them visited.
1754 (run_scc_vn): Make code match comment.
1756 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
1758 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
1759 OPTION_MASK_FLOAT128_KEYWORD.
1760 (POWERPC_MASKS): Likewise.
1761 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
1762 support for the -mfloat128-type option, and make -mfloat128
1763 default on PowerPC Linux systems. Define or undefine
1764 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
1765 Define __float128 to be __ieee128 if IEEE 128-bit support is
1766 enabled, or undefine it.
1767 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
1768 Delete defining __FLOAT128_TYPE__.
1769 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
1770 -mfloat128-type option and make -mfloat128 default on PowerPC
1772 (TARGET_FLOAT128_TYPE): Likewise.
1773 (-mfloat128-type): Likewise.
1774 * config/rs6000/rs6000.c (rs6000_option_override_internal):
1775 Delete the -mfloat128-type option and make -mfloat128 default on
1776 PowerPC Linux systems. Always use __ieee128 to be the keyword for
1777 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
1778 128-bit floating point is enabled. Change tests from using
1779 -mfloat128-type to -mfloat128.
1780 (rs6000_mangle_type): Use the correct mangling for the __float128
1781 type even if normal long double is restricted to 64-bits.
1782 (floatn_mode): Enable the _Float128 type by default on VSX Linux
1784 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
1785 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
1786 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
1788 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
1789 documentation for -mfloat128.
1791 2017-09-06 Olivier Hainque <hainque@adacore.com>
1793 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
1795 2017-09-06 Wish Wu <wishwu007@gmail.com>
1796 Jakub Jelinek <jakub@redhat.com>
1798 * asan.c (initialize_sanitizer_builtins): Add
1799 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
1800 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
1801 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
1802 BT_FN_VOID_UINT64_PTR variables.
1803 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
1804 (BT_FN_VOID_UINT16_UINT16): Likewise.
1805 (BT_FN_VOID_UINT32_UINT32): Likewise.
1806 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
1807 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
1808 (BT_FN_VOID_UINT64_PTR): Likewise.
1809 * common.opt (flag_sanitize_coverage): New variable.
1810 (fsanitize-coverage=trace-pc): Remove.
1811 (fsanitize-coverage=): Add.
1812 * flag-types.h (enum sanitize_coverage_code): New enum.
1813 * fold-const.c (fold_range_test): Disable non-short-circuit
1814 optimization if flag_sanitize_coverage.
1815 (fold_truth_andor): Likewise.
1816 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
1817 * opts.c (COVERAGE_SANITIZER_OPT): Define.
1818 (coverage_sanitizer_opts): New array.
1819 (get_closest_sanitizer_option): Add OPTS argument, handle also
1820 OPT_fsanitize_coverage_.
1821 (parse_sanitizer_options): Adjusted to also handle
1822 OPT_fsanitize_coverage_.
1823 (common_handle_option): Add OPT_fsanitize_coverage_.
1824 * sancov.c (instrument_comparison, instrument_switch): New function.
1825 (sancov_pass): Add trace-cmp support.
1826 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
1827 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
1828 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
1829 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
1830 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
1831 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
1832 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
1833 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
1834 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
1835 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
1837 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
1839 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
1840 error. Only quit immediately if parsing is complete.
1841 (BEGIN): Initialize fatal_err and parse_done.
1842 (begin fpu, end fpu): Check number of arguments.
1843 (begin arch, end arch): Likewise.
1844 (begin cpu, end cpu): Likewise.
1845 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
1846 (optalias): Likewise.
1848 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
1850 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
1851 * config/arm/arm-isa.h: Delete. Move definitions to ...
1852 * arm-cpus.in: ... here. Use new feature and fgroup values.
1853 * config/arm/arm.c (arm_option_override): Use lower case for feature
1855 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
1856 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
1857 * config/arm/parsecpu.awk (END): Add new command 'isa'.
1859 (print_isa_bits_for): New function.
1860 (gen_isa): New function.
1861 (gen_comm_data): Use print_isa_bits_for.
1862 (define feature): New keyword.
1863 (define fgroup): New keyword.
1864 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
1865 (arm-isa.h): Add rule to generate file.
1866 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
1867 case for feature bit names.
1869 2017-09-06 Richard Biener <rguenther@suse.de>
1871 * tree-ssa-pre.c (NECESSARY): Remove.
1872 (create_expression_by_pieces): Do not touch pass-local flags.
1873 (insert_into_preds_of_block): Likewise.
1874 (do_pre_regular_insertion): Likewise.
1875 (eliminate_insert): Likewise.
1876 (eliminate_dom_walker::before_dom_children): Likewise.
1877 (fini_eliminate): Do not look at inserted_exprs.
1878 (mark_operand_necessary): Remove.
1879 (remove_dead_inserted_code): Replace with simple work-list
1880 algorithm based on inserted_exprs and SSA uses.
1881 (pass_pre::execute): Re-order fini_eliminate and
1882 remove_dead_inserted_code.
1884 2017-09-06 Olivier Hainque <hainque@adacore.com>
1886 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
1887 for VxWorks 7. Adjust surrounding comments.
1889 2017-09-06 Richard Biener <rguenther@suse.de>
1891 * gimple-ssa-strength-reduction.c
1892 (find_candidates_dom_walker::before_dom_children): Also allow
1895 2017-09-06 Richard Biener <rguenther@suse.de>
1897 PR tree-optimization/82108
1898 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
1899 for gap in the non-permutation SLP case.
1901 2017-09-06 Martin Jambor <mjambor@suse.cz>
1903 PR tree-optimization/82078
1904 * tree-sra.c (sort_and_splice_var_accesses): Move call to
1905 add_access_to_work_queue...
1906 (build_accesses_from_assign): ...here.
1907 (propagate_all_subaccesses): Make sure racc is the group
1908 representative, if there is one.
1910 2017-09-06 Jakub Jelinek <jakub@redhat.com>
1913 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
1916 2017-09-06 Richard Biener <rguenther@suse.de>
1918 * gimple-ssa-strength-reduction.c
1919 (find_candidates_dom_walker::before_doom_children): Use a
1920 type and not a mode check.
1922 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
1925 * config/arm/predicates.md (arm_general_adddi_operand): Create new
1927 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
1929 2017-09-05 Jeff Law <law@redhat.com>
1931 PR tree-optimization/64910
1932 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
1933 cases where we have 3 or more operands.
1935 2017-09-05 Jakub Jelinek <jakub@redhat.com>
1938 * omp-low.c (lower_omp_for): Recompute tree invariant if
1939 gimple_omp_for_initial/final is ADDR_EXPR.
1942 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
1943 into gimple val before gimplification fo the COND_EXPR.
1945 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
1947 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
1948 REGION_COPY argument.
1949 (thread_through_all_blocks): Remove unused argument to
1950 duplicate_thread_path.
1952 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1953 Alan Hayward <alan.hayward@arm.com>
1954 David Sherwood <david.sherwood@arm.com>
1956 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
1957 Take a scalar_mode rather than a machine_mode.
1958 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
1959 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
1960 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
1961 (aarch64_gen_adjusted_ldpstp): Likewise.
1962 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
1964 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1965 Alan Hayward <alan.hayward@arm.com>
1966 David Sherwood <david.sherwood@arm.com>
1968 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
1969 Take a scalar_int_mode instead of a machine_mode.
1970 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
1971 (aarch64_output_scalar_simd_mov_immediate): Likewise.
1972 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
1973 (aarch64_simd_attr_length_rglist): Delete.
1974 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
1975 a scalar_int_mode instead of a machine_mode.
1976 (aarch64_add_offset): Likewise.
1977 (aarch64_internal_mov_immediate): Likewise
1978 (aarch64_add_constant_internal): Likewise.
1979 (aarch64_add_constant): Likewise.
1980 (aarch64_movw_imm): Likewise.
1981 (aarch64_rtx_arith_op_extract_p): Likewise.
1982 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
1983 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
1984 Remove assert that the mode isn't a vector.
1985 (aarch64_output_scalar_simd_mov_immediate): Likewise.
1986 (aarch64_expand_mov_immediate): Update calls after above changes.
1987 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
1988 (aarch64_and_bitmask_imm): Check for scalar integer modes.
1989 (aarch64_move_imm): Likewise.
1990 (aarch64_can_const_movi_rtx_p): Likewise.
1991 (aarch64_strip_extend): Likewise.
1992 (aarch64_extr_rtx_p): Likewise.
1993 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
1994 a CONST_INT when the mode parameter is VOIDmode.
1995 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
1997 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1999 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
2000 * stor-layout.c (bitwise_mode_for_mode): Likewise.
2001 (bitwise_type_for_mode): Update accordingly.
2003 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2005 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
2006 * stor-layout.c (mode_for_size_tree): Likewise.
2007 (mode_for_array): Update accordingly.
2008 (layout_decl): Likewise.
2009 (compute_record_mode): Likewise. Only set the mode once.
2011 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2013 * target.def (get_mask_mode): Change return type to opt_mode.
2015 * doc/tm.texi: Regenerate.
2016 * targhooks.h (default_get_mask_mode): Return an opt_mode.
2017 * targhooks.c (default_get_mask_mode): Likewise.
2018 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
2019 * optabs-query.c (can_vec_mask_load_store_p): Update use of
2020 targetm.get_mask_mode.
2021 * tree.c (build_truth_vector_type): Likewise.
2023 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2025 * machmode.h (mode_for_vector): Return an opt_mode.
2026 * stor-layout.c (mode_for_vector): Likewise.
2027 (mode_for_int_vector): Update accordingly.
2028 (layout_type): Likewise.
2029 * config/i386/i386.c (emit_memmov): Likewise.
2030 (ix86_expand_set_or_movmem): Likewise.
2031 (ix86_expand_vector_init): Likewise.
2032 (ix86_get_mask_mode): Likewise.
2033 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
2035 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
2036 * expmed.c (extract_bit_field_1): Likewise.
2037 * expr.c (expand_expr_real_2): Likewise.
2038 * optabs-query.c (can_vec_perm_p): Likewise.
2039 (can_vec_mask_load_store_p): Likewise.
2040 * optabs.c (expand_vec_perm): Likewise.
2041 * targhooks.c (default_get_mask_mode): Likewise.
2042 * tree-vect-stmts.c (vectorizable_store): Likewise.
2043 (vectorizable_load): Likewise.
2044 (get_vectype_for_scalar_type_and_size): Likewise.
2046 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2048 * machmode.h (mode_for_int_vector): New function.
2049 * stor-layout.c (mode_for_int_vector): Likewise.
2050 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
2051 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
2052 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
2053 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
2054 (s390_expand_vcond): Likewise.
2056 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2058 * machmode.h (opt_machine_mode): New type.
2059 (opt_mode<T>): Allow construction from anything that can be
2061 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
2062 (mode_for_size): Return an opt_machine_mode.
2063 * stor-layout.c (mode_for_size): Likewise.
2064 (mode_for_size_tree): Update call accordingly.
2065 (bitwise_mode_for_mode): Likewise.
2066 (make_fract_type): Likewise.
2067 (make_accum_type): Likewise.
2068 * caller-save.c (replace_reg_with_saved_mem): Update call
2070 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
2071 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
2072 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
2073 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
2074 * expmed.c (extract_bit_field_1): Likewise.
2075 * reload.c (get_secondary_mem): Likewise.
2076 * varasm.c (assemble_integer): Likewise.
2077 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
2080 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2082 * machmode.h (decimal_float_mode_for_size): New function.
2083 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
2084 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
2085 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
2086 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
2087 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
2088 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
2090 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2092 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
2093 (get_builtin_sync_mode): Likewise.
2094 (expand_ifn_atomic_compare_exchange): Likewise.
2095 (expand_builtin_atomic_clear): Likewise.
2096 (expand_builtin_atomic_test_and_set): Likewise.
2097 (fold_builtin_atomic_always_lock_free): Likewise.
2098 * calls.c (compute_argument_addresses): Likewise.
2099 (emit_library_call_value_1): Likewise.
2100 (store_one_arg): Likewise.
2101 * combine.c (combine_instructions): Likewise.
2102 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
2103 * config/arm/arm.c (arm_function_value): Likewise.
2104 (aapcs_allocate_return_reg): Likewise.
2105 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
2106 * config/i386/i386.c (construct_container): Likewise.
2107 (ix86_gimplify_va_arg): Likewise.
2108 (ix86_expand_sse_cmp): Likewise.
2109 (emit_memmov): Likewise.
2110 (emit_memset): Likewise.
2111 (expand_small_movmem_or_setmem): Likewise.
2112 (ix86_expand_pextr): Likewise.
2113 (ix86_expand_pinsr): Likewise.
2114 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
2115 * config/microblaze/microblaze.c (microblaze_block_move_straight):
2117 * config/mips/mips.c (mips_function_value_1) Likewise.
2118 (mips_block_move_straight): Likewise.
2119 (mips_expand_ins_as_unaligned_store): Likewise.
2120 * config/powerpcspe/powerpcspe.c
2121 (rs6000_darwin64_record_arg_advance_flush): Likewise.
2122 (rs6000_darwin64_record_arg_flush): Likewise.
2123 * config/rs6000/rs6000.c
2124 (rs6000_darwin64_record_arg_advance_flush): Likewise.
2125 (rs6000_darwin64_record_arg_flush): Likewise.
2126 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
2127 (sparc_function_value_1): Likewise.
2128 * config/spu/spu.c (adjust_operand): Likewise.
2129 (spu_emit_branch_or_set): Likewise.
2130 (arith_immediate_p): Likewise.
2131 * emit-rtl.c (gen_lowpart_common): Likewise.
2132 * expr.c (expand_expr_real_1): Likewise.
2133 * function.c (assign_parm_setup_block): Likewise.
2134 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
2135 * reload1.c (alter_reg): Likewise.
2136 * stor-layout.c (mode_for_vector): Likewise.
2137 (layout_type): Likewise.
2139 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2141 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
2142 (spu_convert_move): Likewise.
2143 * lower-subreg.c (resolve_simple_move): Likewise.
2145 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2148 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
2149 define_insn to a define_expand.
2150 (altivec_vsum2sws_direct): New define_insn.
2151 (altivec_vsumsws): Convert from a define_insn to a define_expand.
2153 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
2155 * config/arm/arm.c (arm_option_params_internal): Improve setting of
2158 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
2163 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
2164 realign stack if stack alignment needed is less than incoming
2167 2017-09-05 Marek Polacek <polacek@redhat.com>
2170 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
2173 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
2175 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
2177 2017-09-05 Richard Biener <rguenther@suse.de>
2179 PR tree-optimization/82084
2180 * fold-const.c (can_native_encode_string_p): Handle wide characters.
2182 2017-09-05 Richard Biener <rguenther@suse.de>
2184 PR tree-optimization/82102
2185 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
2187 2017-09-05 Martin Liska <mliska@suse.cz>
2189 PR tree-optimization/82032
2190 * tree-cfg.c (generate_range_test): New function.
2191 * tree-cfg.h (generate_range_test): Declared here.
2192 * tree-cfgcleanup.c (convert_single_case_switch): New function.
2193 (cleanup_control_expr_graph): Use it.
2194 * tree-switch-conversion.c (try_switch_expansion): Remove
2196 (emit_case_nodes): Use generate_range_test.
2198 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
2201 * config/i386/i386.md (*<btsc><mode>_mask): Add
2202 TARGET_USE_BT to insn constraint.
2203 (*btr<mode>_mask): Ditto.
2205 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
2207 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
2208 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
2210 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2213 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
2214 TARGET_NEON and TARGET_IWMMXT.
2215 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
2216 TARGET_NEON and TARGET_IWMMXT.
2217 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
2219 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
2221 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
2222 (ix86_rewrite_tls_address): Ditto.
2223 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
2224 (ix86_rewrite_tls_address_1): Ditto.
2225 (ix86_rewrite_tls_address): Ditto.
2226 * config/i386/predicates.md (tls_address_pattern): New predicate.
2227 * config/i386/i386.md (TLS address splitter): New splitter.
2229 2017-09-04 Richard Biener <rguenther@suse.de>
2231 PR tree-optimization/82084
2232 * fold-const.h (can_native_encode_string_p): Declare.
2233 * fold-const.c (can_native_encode_string_p): Factor out from ...
2234 (native_encode_string): ... here.
2235 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
2236 vectorizing stores from constants we later cannot handle.
2238 2017-09-04 Marek Polacek <polacek@redhat.com>
2241 * doc/invoke.texi: Update -Wtautological-compare documentation.
2243 2017-09-04 Jeff Law <law@redhat.com>
2245 PR tree-optimization/64910
2246 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
2247 swap the first and last operand if the last is a constant.
2249 2017-09-04 Marek Polacek <polacek@redhat.com>
2252 * convert.c (do_narrow): When sanitizing signed integer overflows,
2253 bail out for signed types.
2254 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
2256 2017-09-04 Richard Biener <rguenther@suse.de>
2258 PR tree-optimization/82060
2259 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
2260 Move devirtualization after stmt folding and before EH/AB/noreturn
2261 cleanup to get the stmt refs canonicalized. Use a bool instead
2262 of gimple_modified_p since that doesn't work for NOPs. Schedule
2263 NOPs generated by folding for removal.
2265 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2266 Alan Hayward <alan.hayward@arm.com>
2267 David Sherwood <david.sherwood@arm.com>
2269 * coretypes.h (pad_direction): New enum.
2270 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
2271 (FUNCTION_ARG_PADDING): Likewise.
2272 * target.def (function_arg_padding): New hook.
2273 * targhooks.h (default_function_arg_padding): Declare.
2274 * targhooks.c (default_function_arg_padding): New function.
2275 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
2276 (TARGET_FUNCTION_ARG_PADDING): ...this.
2277 * doc/tm.texi: Regenerate.
2278 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
2279 instead of direction.
2280 (compute_argument_addresses): Likewise.
2281 (load_register_parameters): Likewise.
2282 (emit_library_call_value_1): Likewise.
2283 (store_one_arg): Use targetm.calls.function_arg_padding instead
2284 of FUNCTION_ARG_PADDING.
2285 (must_pass_in_stack_var_size_or_pad): Likewise.
2286 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
2287 (emit_group_store): Likewise.
2288 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
2289 instead of FUNCTION_ARG_PADDING.
2290 (emit_push_insn): Likewise, and propagate enum change throughout
2292 * function.h (direction): Delete.
2293 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
2295 * function.c (assign_parm_find_stack_rtl): Likewise.
2296 (assign_parm_setup_block_p): Likewise.
2297 (assign_parm_setup_block): Likewise.
2298 (gimplify_parameters): Likewise.
2299 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
2300 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
2302 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
2303 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
2304 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
2305 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
2306 (aarch64_function_arg_padding): ...this new function.
2307 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
2308 (TARGET_FUNCTION_ARG_PADDING): Redefine.
2309 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
2310 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
2311 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
2312 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2313 (arm_pad_arg_upward): Replace with...
2314 (arm_function_arg_padding): ...this new function.
2315 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
2317 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
2318 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
2319 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2320 (ia64_hpux_function_arg_padding): Replace with...
2321 (ia64_function_arg_padding): ...this new function. Use pad_direction
2322 instead of direction. Check for TARGET_HPUX.
2323 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
2324 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2325 (iq2000_function_arg_padding): New function.
2326 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
2327 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
2328 (mips_function_arg_padding): ...this new function.
2329 (mips_pad_reg_upward): Update accordingly.
2330 (TARGET_FUNCTION_ARG_PADDING): Redefine.
2331 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
2332 targetm.calls.function_arg_padding.
2333 (FUNCTION_ARG_PADDING): Delete.
2334 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
2335 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
2336 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
2337 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
2338 (nios2_block_reg_padding): Return pad_direction instead of direction.
2339 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
2340 instead of direction.
2341 (nios2_function_arg_padding): Likewise. Make static.
2342 (TARGET_FUNCTION_ARG_PADDING): Redefine.
2343 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
2344 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
2345 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
2346 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2347 (pa_function_arg_padding): Make static. Return pad_direction instead
2349 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
2350 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
2351 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
2352 instead of direction. Use targetm.calls.function_arg_padding.
2353 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
2354 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
2355 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
2356 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
2357 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
2359 (function_arg_padding): Rename to...
2360 (rs6000_function_arg_padding): ...this. Make static. Return
2361 pad_direction instead of direction.
2362 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
2363 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
2364 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
2365 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
2366 instead of direction. Use targetm.calls.function_arg_padding.
2367 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
2368 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
2369 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
2370 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
2371 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2372 (function_arg_padding): Rename to...
2373 (rs6000_function_arg_padding): ...this. Make static. Return
2374 pad_direction instead of direction.
2375 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
2376 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
2377 * config/s390/s390.c (s390_function_arg_padding): New function.
2378 (TARGET_FUNCTION_ARG_PADDING): Redefine.
2379 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
2380 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
2381 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2382 (function_arg_padding): Rename to...
2383 (sparc_function_arg_padding): ...this. Make static. Return
2384 pad_direction instead of direction.
2385 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
2386 * config/spu/spu.c (spu_function_arg_padding): New function.
2387 (TARGET_FUNCTION_ARG_PADDING): Redefine.
2388 * system.h (FUNCTION_ARG_PADDING): Poison.
2390 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2391 Alan Hayward <alan.hayward@arm.com>
2392 David Sherwood <david.sherwood@arm.com>
2394 * target.def (modes_tieable_p): New hook.
2395 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
2396 (TARGET_MODES_TIEABLE_P): ...this.
2397 * doc/tm.texi.in: Regenerate.
2398 * hooks.h (hook_bool_mode_mode_true): Declare.
2399 * hooks.c (hook_bool_mode_mode_true): New function.
2400 * combine.c (subst): Use targetm.modes_tieable_p instead of
2402 * dse.c (find_shift_sequence): Likewise.
2403 * expmed.c (extract_low_bits): Likewise.
2404 * lower-subreg.c: Include target.h.
2405 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
2407 * rtlanal.c (rtx_cost): Likewise.
2408 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
2409 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
2410 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
2411 (TARGET_MODES_TIEABLE_P): Redefine.
2412 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
2413 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
2414 (TARGET_MODES_TIEABLE_P): Redefine.
2415 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
2416 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
2417 (arc_modes_tieable_p): New function.
2418 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
2419 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
2420 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
2421 (arm_modes_tieable_p): Make static.
2422 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
2423 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
2424 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
2425 (TARGET_MODES_TIEABLE_P): Redefine.
2426 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
2427 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
2428 (TARGET_MODES_TIEABLE_P): Redefine.
2429 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
2430 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
2431 (cr16_modes_tieable_p): New function.
2432 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
2433 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
2434 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
2435 (TRULY_NOOP_TRUNCATION): Update comment.
2436 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
2437 (TRULY_NOOP_TRUNCATION): Update comment.
2438 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
2439 (frv_modes_tieable_p): New function.
2440 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
2441 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
2442 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
2443 (TARGET_MODES_TIEABLE_P): Redefine.
2444 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
2445 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
2446 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
2447 (TARGET_MODES_TIEABLE_P): Redefine.
2448 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
2449 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
2450 (ia64_modes_tieable_p): New function.
2451 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
2452 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
2453 (iq2000_modes_tieable_p): New function.
2454 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
2455 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
2456 (lm32_modes_tieable_p): New function.
2457 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
2458 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
2459 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
2460 (TARGET_MODES_TIEABLE_P): Redefine.
2461 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
2462 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
2463 (m32r_modes_tieable_p): New function.
2464 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
2465 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
2466 (m68k_modes_tieable_p): New function.
2467 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
2468 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
2469 (mcore_modes_tieable_p): New function.
2470 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
2471 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
2473 (TARGET_MODES_TIEABLE_P): Redefine.
2474 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
2475 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
2476 * config/mips/mips.c (mips_modes_tieable_p): Make static.
2477 (TARGET_MODES_TIEABLE_P): Redefine.
2478 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
2479 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
2480 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
2481 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
2482 (mn10300_modes_tieable_p): ...this and make static.
2483 (TARGET_MODES_TIEABLE_P): Redefine.
2484 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
2485 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
2486 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
2487 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
2488 (msp430_modes_tieable_p): Make static.
2489 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
2490 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
2491 (TARGET_MODES_TIEABLE_P): Redefine.
2492 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
2493 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
2494 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
2495 (TARGET_MODES_TIEABLE_P): Redefine.
2496 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
2497 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
2498 * config/pa/pa.c (pa_modes_tieable_p): Make static.
2499 (TARGET_MODES_TIEABLE_P): Redefine.
2500 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
2501 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
2502 (pdp11_modes_tieable_p): New function.
2503 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
2504 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
2505 (rs6000_modes_tieable_p): New function.
2506 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
2507 * config/powerpcspe/powerpcspe.md: Update comment.
2508 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
2509 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
2510 (TARGET_MODES_TIEABLE_P): Redefine.
2511 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
2512 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
2513 (rl78_modes_tieable_p): New function.
2514 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
2515 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
2516 (rs6000_modes_tieable_p): New function.
2517 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
2518 * config/rs6000/rs6000.md: Update comment.
2519 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
2520 * config/rx/rx.c (rx_modes_tieable_p): New function.
2521 (TARGET_MODES_TIEABLE_P): Redefine.
2522 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
2523 * config/s390/s390.c (s390_modes_tieable_p): New function.
2524 (TARGET_MODES_TIEABLE_P): Redefine.
2525 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
2526 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
2527 (sh_modes_tieable_p): New function.
2528 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
2529 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
2530 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
2531 (sparc_modes_tieable_p): Make static.
2532 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
2533 * config/spu/spu.c (spu_modes_tieable_p): New function.
2534 (TARGET_MODES_TIEABLE_P): Redefine.
2535 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
2536 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
2537 (TARGET_MODES_TIEABLE_P): Redefine.
2538 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
2539 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
2540 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
2541 * config/v850/v850.c (v850_modes_tieable_p): New function.
2542 (TARGET_MODES_TIEABLE_P): Redefine.
2543 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
2544 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
2545 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
2546 (visium_modes_tieable_p): New function.
2547 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
2548 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
2549 (xtensa_modes_tieable_p): New function.
2550 * system.h (MODES_TIEABLE_P): Poison.
2552 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2553 Alan Hayward <alan.hayward@arm.com>
2554 David Sherwood <david.sherwood@arm.com>
2556 * target.def (hard_regno_mode_ok): New hook.
2557 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
2558 (TARGET_HARD_REGNO_MODE_OK): ...this.
2559 * doc/tm.texi.in: Regenerate.
2560 * hooks.h (hook_bool_uint_mode_true): Declare.
2561 * hooks.c (hook_bool_uint_mode_true): New function.
2562 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
2564 * genpreds.c (write_insn_preds_c): Add an include of target.h.
2565 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
2566 instead of HARD_REGNO_MODE_OK.
2567 * caller-save.c: Include target.h.
2568 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
2570 * combine.c (can_combine_p): Likewise.
2571 (combinable_i3pat): Likewise.
2572 (can_change_dest_mode): Likewise.
2573 * expr.c (init_expr_target): Likewise.
2574 (convert_move): Likewise.
2575 (convert_modes): Likewise.
2576 * ira.c (setup_prohibited_class_mode_regs): Likewise.
2577 (setup_prohibited_mode_move_regs): Likewise.
2578 * ira.h (target_ira): Likewise.
2579 * lra-assigns.c (find_hard_regno_for_1): Likewise.
2580 * lra-constraints.c (process_alt_operands): Likewise.
2581 (split_reg): Likewise.
2582 * recog.c (peep2_find_free_register): Likewise.
2583 * ree.c (combine_reaching_defs): Likewise.
2584 * regcprop.c (maybe_mode_change): Likewise.
2585 * reginfo.c (init_reg_sets_1): Likewise.
2586 (choose_hard_reg_mode): Likewise.
2587 (simplifiable_subregs): Likewise.
2588 * regrename.c (check_new_reg_p): Likewise.
2589 * reload.c (find_valid_class): Likewise.
2590 (find_valid_class_1): Likewise.
2591 (reload_inner_reg_of_subreg): Likewise.
2592 (push_reload): Likewise.
2593 (combine_reloads): Likewise.
2594 (find_dummy_reload): Likewise.
2595 (find_reloads): Likewise.
2596 * reload1.c (find_reg): Likewise.
2597 (set_reload_reg): Likewise.
2598 (allocate_reload_reg): Likewise.
2599 (choose_reload_regs): Likewise.
2600 (reload_adjust_reg_for_temp): Likewise.
2601 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
2602 (simplify_subreg_regno): Likewise.
2603 * sel-sched.c (init_regs_for_mode): Likewise.
2604 * varasm.c (make_decl_rtl): Likewise.
2605 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
2606 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
2608 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
2609 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
2610 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2611 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
2612 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
2613 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2614 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
2615 (arc_mode_class): Delete.
2616 (HARD_REGNO_MODE_OK): Delete.
2617 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2618 (arc_hard_regno_mode_ok): Rename old array to...
2619 (arc_hard_regno_mode_ok_modes): ...this.
2620 (arc_conditional_register_usage): Update accordingly.
2621 (arc_mode_class): Make static.
2622 (arc_hard_regno_mode_ok): New function.
2623 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
2624 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
2625 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2626 (arm_hard_regno_mode_ok): Make static.
2627 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
2629 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
2630 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
2631 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
2633 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2634 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
2635 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
2636 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
2637 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
2638 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2639 * config/bfin/predicates.md (valid_reg_operand): Use
2640 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
2641 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
2642 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
2643 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2644 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
2645 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
2646 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2647 (cr16_hard_regno_mode_ok): Make static and return a bool.
2648 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
2649 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2650 (cris_hard_regno_mode_ok): New function.
2651 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
2652 (epiphany_mode_class): Delete.
2653 (HARD_REGNO_MODE_OK): Delete.
2654 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
2655 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2656 (hard_regno_mode_ok): Rename to...
2657 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
2658 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
2659 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
2661 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
2662 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
2663 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2664 (frv_hard_regno_mode_ok): Make static and return a bool.
2665 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
2667 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
2668 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
2669 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
2670 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
2672 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2673 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
2674 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
2675 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
2677 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2678 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
2679 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2680 (ia64_hard_regno_mode_ok): New function.
2681 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
2682 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2683 (iq2000_hard_regno_mode_ok): New function.
2684 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
2685 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2686 (lm32_hard_regno_mode_ok): New function.
2687 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
2688 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
2689 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
2690 instead of HARD_REGNO_MODE_OK.
2691 (m32c_hard_regno_ok): Rename to...
2692 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
2693 (m32c_cannot_change_mode_class): Update accordingly.
2694 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2695 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
2696 (m32r_mode_class): Delete.
2697 (HARD_REGNO_MODE_OK): Delete.
2698 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2699 (m32r_hard_regno_mode_ok): Rename to...
2700 (m32r_hard_regno_modes): ...this.
2701 (m32r_mode_class): Make static.
2702 (m32r_hard_regno_mode_ok): New function.
2703 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
2704 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
2705 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2706 (m68k_hard_regno_mode_ok): Make static.
2707 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
2708 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2709 (mcore_hard_regno_mode_ok): New function.
2710 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
2711 (HARD_REGNO_MODE_OK): Delete.
2712 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
2714 (microblaze_hard_regno_mode_ok_p): ...this and make static.
2715 (microblaze_hard_regno_mode_ok): New function.
2716 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2717 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
2718 (mips_hard_regno_mode_ok): Delete.
2719 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
2720 (mips_hard_regno_mode_ok_p): ...this and make static.
2721 (mips_hard_regno_mode_ok_p): Rename to...
2722 (mips_hard_regno_mode_ok_uncached): ...this.
2723 (mips_hard_regno_mode_ok): New function.
2724 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
2725 of HARD_REGNO_MODE_OK.
2726 (mips_option_override): Update after above name changes.
2727 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2728 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
2729 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
2730 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
2731 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
2732 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2733 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
2734 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
2735 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
2736 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2737 (msp430_hard_regno_mode_ok): Make static and return a bool.
2738 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
2739 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
2740 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
2742 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2743 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
2744 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
2745 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
2746 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
2747 (PA_HARD_REGNO_MODE_OK): ...this
2748 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
2749 (PA_HARD_REGNO_MODE_OK): ...this.
2750 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2751 (pa_hard_regno_mode_ok): New function.
2752 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
2753 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2754 (pdp11_hard_regno_mode_ok): New function.
2755 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
2756 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
2758 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
2760 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2761 (rs6000_hard_regno_mode_ok): Rename to...
2762 (rs6000_hard_regno_mode_ok_uncached): ...this.
2763 (rs6000_init_hard_regno_mode_ok): Update accordingly.
2764 (rs6000_hard_regno_mode_ok): New function.
2765 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
2766 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
2767 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
2768 (riscv_hard_regno_mode_ok): ...this and make static.
2769 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2770 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
2771 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
2772 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2773 (rl78_hard_regno_mode_ok): Make static and return bool.
2774 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
2775 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
2777 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
2778 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2779 (rs6000_hard_regno_mode_ok): Rename to...
2780 (rs6000_hard_regno_mode_ok_uncached): ...this.
2781 (rs6000_init_hard_regno_mode_ok): Update accordingly.
2782 (rs6000_hard_regno_mode_ok): New function.
2783 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
2784 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
2785 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2786 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
2787 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
2788 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
2789 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2790 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
2791 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
2792 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2793 (sh_hard_regno_mode_ok): Make static.
2794 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
2795 instead of HARD_REGNO_MODE_OK.
2796 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
2797 (sparc_mode_class): Delete.
2798 (HARD_REGNO_MODE_OK): Delete.
2799 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2800 (hard_regno_mode_classes): Make static.
2801 (sparc_mode_class): Likewise.
2802 (sparc_hard_regno_mode_ok): New function.
2803 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
2804 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
2805 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
2807 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2808 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
2809 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
2810 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
2811 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
2812 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2813 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
2814 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
2815 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2816 (visium_hard_regno_mode_ok): New function.
2817 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
2818 instead of HARD_REGNO_MODE_OK.
2819 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
2820 (HARD_REGNO_MODE_OK): Delete.
2821 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
2822 (xtensa_hard_regno_mode_ok_p): ...this and make static.
2823 (xtensa_option_override): Update accordingly.
2824 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2825 (xtensa_hard_regno_mode_ok): New function.
2826 * system.h (HARD_REGNO_MODE_OK): Poison.
2828 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2829 Alan Hayward <alan.hayward@arm.com>
2830 David Sherwood <david.sherwood@arm.com>
2832 * target.def (hard_regno_call_part_clobbered): New hook.
2833 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
2834 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
2835 * doc/tm.texi: Regenerate.
2836 * hooks.h (hook_bool_uint_mode_false): Declare.
2837 * hooks.c (hook_bool_uint_mode_false): New function.
2838 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2839 * cselib.c (cselib_process_insn): Use
2840 targetm.hard_regno_call_part_clobbered instead of
2841 HARD_REGNO_CALL_PART_CLOBBERED.
2842 * ira-conflicts.c (ira_build_conflicts): Likewise.
2843 * ira-costs.c (ira_tune_allocno_costs): Likewise.
2844 * lra-constraints.c (need_for_call_save_p): Likewise.
2845 * lra-lives.c: Include target.h.
2846 (check_pseudos_live_through_calls): Use
2847 targetm.hard_regno_call_part_clobbered instead of
2848 HARD_REGNO_CALL_PART_CLOBBERED.
2849 * regcprop.c: Include target.h.
2850 (copyprop_hardreg_forward_1): Use
2851 targetm.hard_regno_call_part_clobbered instead of
2852 HARD_REGNO_CALL_PART_CLOBBERED.
2853 * reginfo.c (choose_hard_reg_mode): Likewise.
2854 * regrename.c (check_new_reg_p): Likewise.
2855 * reload.c (find_equiv_reg): Likewise.
2856 * reload1.c (emit_reload_insns): Likewise.
2857 * sched-deps.c (deps_analyze_insn): Likewise.
2858 * sel-sched.c (init_regs_for_mode): Likewise.
2859 (mark_unavailable_hard_regs): Likewise.
2860 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
2861 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2862 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
2864 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2865 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2866 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
2868 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
2870 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2871 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2872 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
2874 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2875 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2876 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
2878 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2879 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
2881 * config/powerpcspe/powerpcspe.c
2882 (rs6000_hard_regno_call_part_clobbered): New function.
2883 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2884 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2885 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
2887 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2888 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2889 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
2891 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2892 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2893 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
2895 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2896 Alan Hayward <alan.hayward@arm.com>
2897 David Sherwood <david.sherwood@arm.com>
2899 * rtl.h (subreg_memory_offset): Declare.
2900 * emit-rtl.c (subreg_memory_offset): New function.
2901 * expmed.c (store_bit_field_1): Use it.
2902 * expr.c (undefined_operand_subword_p): Likewise.
2903 * simplify-rtx.c (simplify_subreg): Likewise.
2905 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
2907 PR rtl-optimization/57448
2910 * optabs.c (expand_atomic_load): Place compiler memory barriers if
2911 using atomic_load pattern.
2912 (expand_atomic_store): Likewise.
2914 2017-09-04 Jakub Jelinek <jakub@redhat.com>
2917 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
2918 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
2919 handling. Use replace_call_with_value with NULL instead of
2920 gsi_replace, unlink_stmt_vdef and release_defs.
2922 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
2925 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
2927 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2930 * rtl.h (emit_library_call_value_1): Declare.
2931 (emit_library_call): Replace declaration with a series of overloads.
2932 Remove the parameter count argument.
2933 (emit_library_call_value): Likewise.
2934 * calls.c (emit_library_call_value_1): Make global. Replace varargs
2935 with an "rtx_mode_t *".
2936 (emit_library_call_value): Delete.
2937 (emit_library_call): Likewise.
2938 * asan.c (asan_emit_stack_protection): Update calls accordingly.
2939 (asan_emit_allocas_unpoison): Likewise.
2940 * builtins.c (expand_builtin_powi): Likewise.
2941 (expand_asan_emit_allocas_unpoison): Likewise.
2942 * cfgexpand.c (expand_main_function): Likewise.
2943 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
2944 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
2945 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
2946 * config/arm/arm.c (arm_trampoline_init): Likewise.
2947 (arm_call_tls_get_addr): Likewise.
2948 (arm_expand_divmod_libfunc): Likewise.
2949 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
2950 (smulsi3_highpart): Likewise.
2951 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
2952 (c6x_expand_compare): Likewise.
2953 (c6x_expand_movmem): Likewise.
2954 * config/frv/frv.c (frv_trampoline_init): Likewise.
2955 * config/i386/i386.c (ix86_trampoline_init): Likewise.
2956 (ix86_expand_divmod_libfunc): Likewise.
2957 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
2958 (ia64_expand_compare): Likewise.
2959 (ia64_profile_hook): Likewise.
2960 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
2961 (nonlocal_goto): Likewise.
2962 (restore_stack_nonlocal): Likewise.
2963 * config/m32r/m32r.c (block_move_call): Likewise.
2964 (m32r_trampoline_init): Likewise.
2965 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
2966 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
2967 (m68k_call_m68k_read_tp): Likewise.
2968 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
2969 (microblaze_expand_divide): Likewise.
2970 * config/mips/mips.h (mips_args): Likewise.
2971 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
2972 (MIPS_ICACHE_SYNC): Likewise.
2973 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
2974 (nios2_trampoline_init): Likewise.
2975 * config/pa/pa.c (hppa_tls_call): Likewise.
2976 (pa_trampoline_init): Likewise.
2977 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
2978 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
2979 (expand_strn_compare): Likewise.
2980 (rs6000_generate_compare): Likewise.
2981 (rs6000_expand_float128_convert): Likewise.
2982 (output_profile_hook): Likewise.
2983 (rs6000_trampoline_init): Likewise.
2984 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
2985 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
2986 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
2987 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
2988 (rs6000_generate_compare): Likewise.
2989 (rs6000_expand_float128_convert): Likewise.
2990 (output_profile_hook): Likewise.
2991 (rs6000_trampoline_init): Likewise.
2992 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
2993 * config/sh/sh.c (sh_trampoline_init): Likewise.
2994 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
2995 (sparc_emit_float_lib_cmp): Likewise.
2996 (sparc32_initialize_trampoline): Likewise.
2997 (sparc64_initialize_trampoline): Likewise.
2998 (sparc_profile_hook): Likewise.
2999 * config/spu/spu.c (ea_load_store): Likewise.
3000 * config/spu/spu.md (floatunssidf2): Likewise.
3001 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
3002 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
3003 * config/visium/visium.c (expand_block_move_4): Likewise.
3004 (expand_block_move_2): Likewise.
3005 (expand_block_move_1): Likewise.
3006 (expand_block_set_4): Likewise.
3007 (expand_block_set_2): Likewise.
3008 (expand_block_set_1): Likewise.
3009 (visium_trampoline_init): Likewise.
3010 (visium_profile_hook): Likewise.
3011 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
3012 (xtensa_setup_frame_addresses): Likewise.
3013 (xtensa_trampoline_init): Likewise.
3014 * except.c (sjlj_emit_function_enter): Likewise.
3015 (sjlj_emit_function_exit): Likewise.
3016 * explow.c (allocate_dynamic_stack_space): Likewise.
3017 (probe_stack_range): Likewise.
3018 * expr.c (convert_mode_scalar): Likewise.
3019 * optabs.c (expand_binop): Likewise.
3020 (expand_twoval_binop_libfunc): Likewise.
3021 (expand_unop): Likewise.
3022 (prepare_cmp_insn): Likewise.
3023 (prepare_float_lib_cmp): Likewise.
3024 (expand_float): Likewise.
3025 (expand_fix): Likewise.
3026 (expand_fixed_convert): Likewise.
3027 (maybe_emit_sync_lock_test_and_set): Likewise.
3028 (expand_atomic_compare_and_swap): Likewise.
3029 (expand_mem_thread_fence): Likewise.
3030 (expand_atomic_fetch_op): Likewise.
3032 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
3034 * doc/generic.texi (OpenACC): Adjust URL.
3035 * doc/invoke.texi (C Dialect Options): Ditto.
3037 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
3039 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
3040 predicate for operand 1. Add (m,<S>) constraint.
3041 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
3042 Prevent memory operand 1 with register operand 2.
3044 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
3046 PR rtl-optimization/82024
3047 * combine.c (try_combine): If the combination result is a PARALLEL,
3048 and we only need to retain the SET in there that would be placed
3049 at I2, check that we can place that at I3 instead, before doing so.
3051 2017-09-01 Jakub Jelinek <jakub@redhat.com>
3054 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
3056 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
3057 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
3060 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
3061 Jeff Law <law@redhat.com>
3063 * varasm.c (bss_initializer_p): Do not put constants into .bss
3064 (categorize_decl_for_section): Handle bss_initializer_p returning
3065 false when DECL_INITIAL is NULL.
3067 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3070 * config/s390/s390.c (s390_can_inline_p): New function.
3072 2017-09-01 Jeff Law <law@redhat.com>
3074 PR tree-optimization/82052
3075 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
3076 Always initialize the returned slot after a hash table miss
3077 when INSERT is true.
3079 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
3081 * config/s390/s390.md (mem_signal_fence): Remove.
3082 * doc/md.texi (mem_signal_fence): Remove.
3083 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
3085 * target-insns.def (mem_signal_fence): Remove.
3087 2017-09-01 Jakub Jelinek <jakub@redhat.com>
3090 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
3093 * asan.c (create_odr_indicator): Strip name encoding from assembler
3094 name before appending it after __odr_asan_.
3096 2017-09-01 Martin Liska <mliska@suse.cz>
3098 PR tree-optimization/82059
3099 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
3100 frequency only when an edge is redirected.
3102 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
3104 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
3105 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
3106 (arc_conditional_register_usage): Remove ARC600 lp_count
3108 (arc_file_start): Emit Tag_ARC_CPU_variation.
3109 (arc_can_use_doloop_p): New conditions to use ZOLs.
3110 (hwloop_fail): New function.
3111 (hwloop_optimize): Likewise.
3112 (hwloop_pattern_reg): Likewise.
3113 (arc_doloop_hooks): New struct, to be used with reorg_loops.
3114 (arc_reorg_loops): New function, calls reorg_loops.
3115 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
3116 (arc600_corereg_hazard): Remove ZOL checking, case handled by
3118 (arc_loop_hazard): Remove function, functionality moved into
3120 (arc_hazard): Remove arc_loop_hazard call.
3121 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
3122 into hwloop_optimize.
3123 (arc_label_align): Remove ZOL handling.
3124 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
3125 * config/arc/arc.md (doloop_begin): Remove pattern.
3126 (doloop_begin_i): Likewise.
3127 (doloop_end_i): Likewise.
3128 (doloop_fallback): Likewise.
3129 (doloop_fallback_m): Likewise.
3130 (doloop_end): Reimplement expand.
3131 (arc_lp): New pattern for LP instruction.
3132 (loop_end): New pattern.
3133 (loop_fail): Likewise.
3134 (decrement_and_branch_until_zero): Likewise.
3135 * config/arc/arc.opt (mlpc-width): New option.
3136 * doc/invoke.texi (mlpc-width): Document option.
3138 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
3140 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
3141 (arc_ccfsm_advance): Fix checking for delay slots.
3142 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
3144 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
3146 * config/arc/arc.md (movqi_insn): Add stores to save constant long
3148 (movhi_insn): Update store instruction constraint which are saving
3149 6-bit short immediates.
3150 (movsi_insn): Consider also short scaled load operations.
3151 (zero_extendhisi2_i): Use Usd constraint instead of T.
3152 (extendhisi2_i): Add q constraint.
3153 (arc_clzsi2): Add type and length attributes.
3154 (arc_ctzsi2): Likewise.
3155 * config/arc/constraints.md (Usc): Update constraint, the
3156 assembler can parse two relocations for a single instruction.
3158 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
3160 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
3161 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
3163 2017-08-31 Olivier Hainque <hainque@adacore.com>
3165 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
3166 match as powerpc-wrs-vxworks*.
3168 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
3170 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
3171 register constraint for by-element operand.
3172 (aarch64_mls_elt_merge<mode>): Likewise.
3174 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
3176 * config/arc/arc.c (arc_can_follow_jump): Check for short
3179 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
3181 * config.gcc: Use g.opt for arc.
3182 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
3183 functionality moved to ...
3184 (legitimate_scaled_address_p): New function, ...here.
3185 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
3186 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
3187 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
3189 (arc_override_options): Handle G option.
3190 (arc_output_pic_addr_const): Correct function definition.
3191 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
3192 (arc_decl_anon_ns_mem_p): Delete.
3193 (arc_in_small_data_p): Overhaul this function to take into
3194 consideration the value given via G option.
3195 (arc_rewrite_small_data_1): Renamed and corrected old
3196 arc_rewrite_small_data function.
3197 (arc_rewrite_small_data): New function.
3198 (small_data_pattern): Don't use pic_offset_table_rtx.
3199 * config/arc/arc.h (CC1_SPEC): Recognize G option.
3200 * config/arc/simdext.md (movmisalignv2hi): Use
3201 prepare_move_operands function.
3203 (movmisalign*): Likewise.
3204 * doc/invoke.texi (ARC options): Document -G option.
3206 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
3208 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
3210 * config/arc/arc.c (arc_print_operand): Output scalled address for
3211 sdata whenever is possible.
3212 (arc_in_small_data_p): Allow sdata for 64bit datum when double
3213 load/stores are available.
3214 (compact_sda_memory_operand): Check for the alignment required by
3215 code density instructions.
3216 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
3218 * config/arc/constraints.md (Usd): Update constraint.
3219 (Us0): New constraint.
3220 (Usc): Update constraint.
3222 2017-08-31 Richard Biener <rguenther@suse.de>
3225 * dwarf2out.c (dwarf2out_early_global_decl): Process each
3228 2017-08-31 Tamar Christina <tamar.christina@arm.com>
3230 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
3231 Resize type_signature.
3233 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
3234 Alan Hayward <alan.hayward@arm.com>
3235 David Sherwood <david.sherwood@arm.com>
3237 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
3238 subregs whose inner modes can be stored in GPRs.
3239 (aarch64_classify_index): Likewise.
3241 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
3242 Alan Hayward <alan.hayward@arm.com>
3243 David Sherwood <david.sherwood@arm.com>
3245 * config/aarch64/iterators.md (V_cmp_result): Rename to...
3246 (V_INT_EQUIV): ...this.
3247 (v_cmp_result): Rename to...
3248 (v_int_equiv): ...this.
3249 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
3250 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
3251 (copysign<mode>3): Likewise.
3252 (aarch64_simd_bsl<mode>_internal): Likewise.
3253 (aarch64_simd_bsl<mode>): Likewise.
3254 (vec_cmp<mode><mode>): Likewise.
3255 (vcond<mode><mode>): Likewise.
3256 (vcond<v_cmp_mixed><mode>): Likewise.
3257 (vcondu<mode><v_cmp_mixed>): Likewise.
3258 (aarch64_cm<optab><mode>): Likewise.
3259 (aarch64_cmtst<mode>): Likewise.
3260 (aarch64_fac<optab><mode>): Likewise.
3261 (vec_perm_const<mode>): Likewise.
3262 (vcond_mask_<mode><v_cmp_result>): Rename to...
3263 (vcond_mask_<mode><v_int_equiv>): ...this.
3264 (vec_cmp<mode><v_cmp_result>): Rename to...
3265 (vec_cmp<mode><v_int_equiv>): ...this.
3267 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
3268 Alan Hayward <alan.hayward@arm.com>
3269 David Sherwood <david.sherwood@arm.com>
3271 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
3273 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
3274 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
3275 (UNSPEC_LD4_DREG): New unspecs.
3276 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
3277 (aarch64_ld2<mode>_dreg_be): Replace with...
3278 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
3280 (aarch64_ld3<mode>_dreg_le)
3281 (aarch64_ld3<mode>_dreg_be): Replace with...
3282 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
3284 (aarch64_ld4<mode>_dreg_le)
3285 (aarch64_ld4<mode>_dreg_be): Replace with...
3286 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
3289 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3291 PR tree-optimization/81987
3292 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
3293 insert an initializer in a location not dominated by the stride
3296 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
3298 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
3299 on the entire header of the finally block in the fallthru case.
3301 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
3303 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
3305 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
3307 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
3308 rs6000_emit_move_from_cr and call renamed function.
3309 (rs6000_emit_prologue): Call renamed functions.
3310 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
3311 movesi_from_cr, remove volatile CRs.
3313 2017-08-30 Jon Beniston <jon@beniston.com>
3314 Richard Biener <rguenther@suse.de>
3316 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
3317 of VECTOR_MODE_P check.
3318 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
3319 element vector types.
3321 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3323 * df.h (df_read_modify_subreg_p): Remove in favor of...
3324 * rtl.h (read_modify_subreg_p): ...this new function. Take a
3325 const_rtx instead of an rtx.
3326 * cprop.c (local_cprop_find_used_regs): Update accordingly.
3327 * df-problems.c (df_word_lr_mark_ref): Likewise.
3328 * ira-lives.c (mark_pseudo_reg_live): Likewise.
3329 (mark_pseudo_reg_dead): Likewise.
3330 (mark_ref_dead): Likewise.
3331 * reginfo.c (init_subregs_of_mode): Likewise.
3332 * sched-deps.c (sched_analyze_1): Likewise.
3333 * df-scan.c (df_def_record_1): Likewise.
3334 (df_uses_record): Likewise.
3335 (df_read_modify_subreg_p): Remove in favor of...
3336 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
3337 const_rtx instead of an rtx.
3339 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3340 Alan Hayward <alan.hayward@arm.com>
3341 David Sherwood <david.sherwood@arm.com>
3343 * rtl.h (partial_subreg_p): New function.
3344 * caller-save.c (save_call_clobbered_regs): Use it.
3345 * calls.c (expand_call): Likewise.
3346 * combine.c (combinable_i3pat): Likewise.
3347 (simplify_set): Likewise.
3348 (make_extraction): Likewise.
3349 (make_compound_operation_int): Likewise.
3350 (gen_lowpart_or_truncate): Likewise.
3351 (force_to_mode): Likewise.
3352 (make_field_assignment): Likewise.
3353 (reg_truncated_to_mode): Likewise.
3354 (record_truncated_value): Likewise.
3355 (move_deaths): Likewise.
3356 * cse.c (record_jump_cond): Likewise.
3357 (cse_insn): Likewise.
3358 * cselib.c (cselib_lookup_1): Likewise.
3359 * expmed.c (extract_bit_field_using_extv): Likewise.
3360 * function.c (assign_parm_setup_reg): Likewise.
3361 * ifcvt.c (noce_convert_multiple_sets): Likewise.
3362 * ira-build.c (create_insn_allocnos): Likewise.
3363 * lra-coalesce.c (merge_pseudos): Likewise.
3364 * lra-constraints.c (match_reload): Likewise.
3365 (simplify_operand_subreg): Likewise.
3366 (curr_insn_transform): Likewise.
3367 * lra-lives.c (process_bb_lives): Likewise.
3368 * lra.c (new_insn_reg): Likewise.
3369 (lra_substitute_pseudo): Likewise.
3370 * regcprop.c (mode_change_ok): Likewise.
3371 (maybe_mode_change): Likewise.
3372 (copyprop_hardreg_forward_1): Likewise.
3373 * reload.c (push_reload): Likewise.
3374 (find_reloads): Likewise.
3375 (find_reloads_subreg_address): Likewise.
3376 * reload1.c (alter_reg): Likewise.
3377 (eliminate_regs_1): Likewise.
3378 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3380 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
3382 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
3383 back to if statements, including unpack.
3385 2017-08-30 Martin Liska <mliska@suse.cz>
3388 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
3390 (func_checker::compare_asm_inputs_outputs): ... this function.
3391 (func_checker::compare_gimple_asm): Use the function to compare
3392 also ASM constrains.
3393 * ipa-icf-gimple.h: Rename the function.
3395 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3396 Alan Hayward <alan.hayward@arm.com>
3397 David Sherwood <david.sherwood@arm.com>
3399 * coretypes.h (complex_mode): New type.
3400 * gdbhooks.py (build_pretty_printer): Handle it.
3401 * machmode.h (complex_mode): New class.
3402 (complex_mode::includes_p): New function.
3403 (is_complex_int_mode): Likewise.
3404 (is_complex_float_mode): Likewise.
3405 * genmodes.c (get_mode_class): Handle complex mode classes.
3406 * function.c (expand_function_end): Use is_complex_int_mode.
3408 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3409 Alan Hayward <alan.hayward@arm.com>
3410 David Sherwood <david.sherwood@arm.com>
3412 * coretypes.h (scalar_mode_pod): New typedef.
3413 * gdbhooks.py (build_pretty_printer): Handle it.
3414 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
3415 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
3416 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
3417 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
3420 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3421 Alan Hayward <alan.hayward@arm.com>
3422 David Sherwood <david.sherwood@arm.com>
3424 * machmode.h (mode_for_vector): Take a scalar_mode instead
3426 * stor-layout.c (mode_for_vector): Likewise.
3427 * explow.c (promote_mode): Use as_a <scalar_mode>.
3428 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
3430 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3431 Alan Hayward <alan.hayward@arm.com>
3432 David Sherwood <david.sherwood@arm.com>
3434 * target.def (preferred_simd_mode): Take a scalar_mode
3435 instead of a machine_mode.
3436 * targhooks.h (default_preferred_simd_mode): Likewise.
3437 * targhooks.c (default_preferred_simd_mode): Likewise.
3438 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
3439 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
3440 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
3441 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
3442 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
3443 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
3444 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
3445 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
3447 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
3448 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
3449 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
3450 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
3451 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
3452 * doc/tm.texi: Regenerate.
3453 * optabs-query.c (can_vec_mask_load_store_p): Return false for
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 * target.def (scalar_mode_supported_p): Take a scalar_mode
3461 instead of a machine_mode.
3462 * targhooks.h (default_scalar_mode_supported_p): Likewise.
3463 * targhooks.c (default_scalar_mode_supported_p): Likewise.
3464 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
3465 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
3466 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
3467 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
3468 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
3469 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
3470 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
3471 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
3472 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
3473 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
3474 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
3475 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
3477 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
3478 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
3479 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
3480 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
3481 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
3483 * doc/tm.texi: Regenerate.
3485 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3486 Alan Hayward <alan.hayward@arm.com>
3487 David Sherwood <david.sherwood@arm.com>
3489 * coretypes.h (opt_scalar_mode): New typedef.
3490 * gdbhooks.py (build_pretty_printers): Handle it.
3491 * machmode.h (mode_iterator::get_2xwider): Add overload for
3493 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
3495 * expr.c (convert_mode_scalar): Likewise.
3496 * omp-low.c (omp_clause_aligned_alignment): Likewise.
3497 * optabs.c (expand_float): Likewise.
3498 (expand_fix): Likewise.
3499 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
3501 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3502 Alan Hayward <alan.hayward@arm.com>
3503 David Sherwood <david.sherwood@arm.com>
3505 * optabs.c (expand_float): Explicitly check for scalars before
3506 using a branching expansion.
3507 (expand_fix): Likewise.
3509 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3510 Alan Hayward <alan.hayward@arm.com>
3511 David Sherwood <david.sherwood@arm.com>
3513 * expr.c (convert_mode): Split scalar handling out into...
3514 (convert_mode_scalar): ...this new function. Treat the modes
3517 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3518 Alan Hayward <alan.hayward@arm.com>
3519 David Sherwood <david.sherwood@arm.com>
3521 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
3523 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
3525 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3526 Alan Hayward <alan.hayward@arm.com>
3527 David Sherwood <david.sherwood@arm.com>
3529 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
3530 rather than a machine_mode.
3531 (fixed_from_string): Likewise.
3532 (fixed_convert): Likewise.
3533 (fixed_convert_from_int): Likewise.
3534 (fixed_convert_from_real): Likewise.
3535 (real_convert_from_fixed): Likewise.
3536 * fixed-value.c (fixed_from_double_int): Likewise.
3537 (fixed_from_string): Likewise.
3538 (fixed_convert): Likewise.
3539 (fixed_convert_from_int): Likewise.
3540 (fixed_convert_from_real): Likewise.
3541 (real_convert_from_fixed): Likewise.
3542 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
3544 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3545 Alan Hayward <alan.hayward@arm.com>
3546 David Sherwood <david.sherwood@arm.com>
3548 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
3549 of separate mode class checks. Do not allow vector modes here.
3550 (immed_wide_int_const): Use as_a <scalar_mode>.
3551 * explow.c (trunc_int_for_mode): Likewise.
3552 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
3553 (wi::shwi): Likewise.
3554 (wi::min_value): Likewise.
3555 (wi::max_value): Likewise.
3556 * dwarf2out.c (loc_descriptor): Likewise.
3557 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
3560 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3561 Alan Hayward <alan.hayward@arm.com>
3562 David Sherwood <david.sherwood@arm.com>
3564 * tree.h (SCALAR_TYPE_MODE): New macro.
3565 * expr.c (expand_expr_addr_expr_1): Use it.
3566 (expand_expr_real_2): Likewise.
3567 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
3568 (fold_convert_const_fixed_from_int): Likewise.
3569 (fold_convert_const_fixed_from_real): Likewise.
3570 (native_encode_fixed): Likewise
3571 (native_encode_complex): Likewise
3572 (native_encode_vector): Likewise.
3573 (native_interpret_fixed): Likewise.
3574 (native_interpret_real): Likewise.
3575 (native_interpret_complex): Likewise.
3576 (native_interpret_vector): Likewise.
3577 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
3578 (simd_clone_adjust_argument_types): Likewise.
3579 (simd_clone_init_simd_arrays): Likewise.
3580 (simd_clone_adjust): Likewise.
3581 * stor-layout.c (layout_type): Likewise.
3582 * tree.c (build_minus_one_cst): Likewise.
3583 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
3584 * tree-inline.c (estimate_move_cost): Likewise.
3585 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
3586 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
3587 (vectorizable_reduction): Likewise.
3588 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
3589 (vect_recog_mixed_size_cond_pattern): Likewise.
3590 (check_bool_pattern): Likewise.
3591 (adjust_bool_pattern): Likewise.
3592 (search_type_for_mask_1): Likewise.
3593 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
3594 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
3595 (vectorizable_load): Likewise.
3596 (vectorizable_store): Likewise.
3597 * ubsan.c (ubsan_encode_value): Likewise.
3598 * varasm.c (output_constant): Likewise.
3600 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3601 Alan Hayward <alan.hayward@arm.com>
3602 David Sherwood <david.sherwood@arm.com>
3604 * coretypes.h (scalar_mode): New class.
3605 * machmode.h (scalar_mode): Likewise.
3606 (scalar_mode::includes_p): New function.
3607 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
3608 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
3609 * genmodes.c (get_mode_class): Handle remaining scalar modes.
3610 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
3611 * expmed.c (store_bit_field_1): Likewise.
3612 (extract_bit_field_1): Likewise.
3613 * expr.c (write_complex_part): Likewise.
3614 (read_complex_part): Likewise.
3615 (emit_move_complex_push): Likewise.
3616 (expand_expr_real_2): Likewise.
3617 * function.c (assign_parm_setup_reg): Likewise.
3618 (assign_parms_unsplit_complex): Likewise.
3619 * optabs.c (expand_binop): Likewise.
3620 * rtlanal.c (subreg_get_info): Likewise.
3621 * simplify-rtx.c (simplify_immed_subreg): Likewise.
3622 * varasm.c (output_constant_pool_2): Likewise.
3624 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3625 Alan Hayward <alan.hayward@arm.com>
3626 David Sherwood <david.sherwood@arm.com>
3628 * expmed.c (extract_high_half): Use scalar_int_mode and remove
3630 (expmed_mult_highpart_optab): Likewise.
3631 (expmed_mult_highpart): Likewise.
3633 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3634 Alan Hayward <alan.hayward@arm.com>
3635 David Sherwood <david.sherwood@arm.com>
3637 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
3638 instead of a machine_mode.
3639 (builtin_memset_read_str): Likewise.
3640 * builtins.c (c_readstr): Likewise.
3641 (builtin_memcpy_read_str): Likewise.
3642 (builtin_strncpy_read_str): Likewise.
3643 (builtin_memset_read_str): Likewise.
3644 (builtin_memset_gen_str): Likewise.
3645 (expand_builtin_signbit): Use scalar_int_mode for local variables.
3646 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
3647 instead of a machine_mode.
3648 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
3650 (make_extraction): Likewise.
3651 (try_widen_shift_mode): Take and return scalar_int_modes instead
3653 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
3654 a scalar_int_mode instead of a machine_mode.
3655 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
3656 (avr_addr_space_pointer_mode): Likewise.
3657 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
3658 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
3659 (msp430_unwind_word_mode): Likewise.
3660 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
3661 (spu_addr_space_pointer_mode): Likewise.
3662 (spu_addr_space_address_mode): Likewise.
3663 (spu_libgcc_cmp_return_mode): Likewise.
3664 (spu_libgcc_shift_count_mode): Likewise.
3665 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
3666 (rl78_addr_space_pointer_mode): Likewise.
3667 (fl78_unwind_word_mode): Likewise.
3668 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
3670 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
3671 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
3672 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
3673 (mips_valid_pointer_mode): Likewise.
3674 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
3675 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
3676 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
3678 (ft32_addr_space_address_mode): Likewise.
3679 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
3680 scalar_int_mode instead of a machine_mode.
3681 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
3683 (m32c_addr_space_address_mode): Likewise.
3684 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
3685 (rs6000_eh_return_filter_mode): Likewise.
3686 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
3687 (rs6000_eh_return_filter_mode): Likewise.
3688 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
3689 (s390_libgcc_shift_count_mode): Likewise.
3690 (s390_unwind_word_mode): Likewise.
3691 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
3693 * target.def (mode_rep_extended): Likewise.
3694 (valid_pointer_mode): Likewise.
3695 (addr_space.valid_pointer_mode): Likewise.
3696 (eh_return_filter_mode): Return a scalar_int_mode rather than
3698 (libgcc_cmp_return_mode): Likewise.
3699 (libgcc_shift_count_mode): Likewise.
3700 (unwind_word_mode): Likewise.
3701 (addr_space.pointer_mode): Likewise.
3702 (addr_space.address_mode): Likewise.
3703 * doc/tm.texi: Regenerate.
3704 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
3706 (do_jump): Use scalar_int_mode for local variables.
3707 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
3708 rather than a machine_mode.
3709 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
3710 (scompare_loc_descriptor_wide): Likewise.
3711 (scompare_loc_descriptor_narrow): Likewise.
3712 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
3714 * except.c (sjlj_emit_dispatch_table): Likewise.
3715 (expand_builtin_eh_copy_values): Likewise.
3716 * explow.c (convert_memory_address_addr_space_1): Likewise.
3717 Take a scalar_int_mode rather than a machine_mode.
3718 (convert_memory_address_addr_space): Take a scalar_int_mode rather
3719 than a machine_mode.
3720 (memory_address_addr_space): Use scalar_int_mode for local variables.
3721 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
3722 rather than a machine_mode.
3723 * expmed.c (mask_rtx): Likewise.
3724 (init_expmed_one_conv): Likewise.
3725 (expand_mult_highpart_adjust): Likewise.
3726 (extract_high_half): Likewise.
3727 (expmed_mult_highpart_optab): Likewise.
3728 (expmed_mult_highpart): Likewise.
3729 (expand_smod_pow2): Likewise.
3730 (expand_sdiv_pow2): Likewise.
3731 (emit_store_flag_int): Likewise.
3732 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
3734 (extract_low_bits): Likewise.
3735 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
3737 * expr.c (pieces_addr::adjust): Likewise.
3738 (can_store_by_pieces): Likewise.
3739 (store_by_pieces): Likewise.
3740 (clear_by_pieces_1): Likewise.
3741 (expand_expr_addr_expr_1): Likewise.
3742 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
3743 (expand_expr_real_1): Likewise.
3744 (try_casesi): Likewise.
3745 * final.c (shorten_branches): Likewise.
3746 * fold-const.c (fold_convert_const_int_from_fixed): Change the
3747 type of "mode" to machine_mode.
3748 * internal-fn.c (expand_arith_overflow_result_store): Take a
3749 scalar_int_mode rather than a machine_mode.
3750 (expand_mul_overflow): Use scalar_int_mode for local variables.
3751 * loop-doloop.c (doloop_modify): Likewise.
3752 (doloop_optimize): Likewise.
3753 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
3754 than a machine_mode.
3755 (expand_doubleword_shift_condmove): Likewise.
3756 (expand_doubleword_shift): Likewise.
3757 (expand_doubleword_clz): Likewise.
3758 (expand_doubleword_popcount): Likewise.
3759 (expand_doubleword_parity): Likewise.
3760 (expand_absneg_bit): Use scalar_int_mode for local variables.
3761 (prepare_float_lib_cmp): Likewise.
3762 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
3763 rather than a machine_mode.
3764 (convert_memory_address_addr_space): Likewise.
3765 (get_mode_bounds): Likewise.
3766 (get_address_mode): Return a scalar_int_mode rather than a
3768 * rtlanal.c (get_address_mode): Likewise.
3769 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
3770 than a machine_mode.
3771 * targhooks.c (default_mode_rep_extended): Likewise.
3772 (default_valid_pointer_mode): Likewise.
3773 (default_addr_space_valid_pointer_mode): Likewise.
3774 (default_eh_return_filter_mode): Return a scalar_int_mode rather
3775 than a machine_mode.
3776 (default_libgcc_cmp_return_mode): Likewise.
3777 (default_libgcc_shift_count_mode): Likewise.
3778 (default_unwind_word_mode): Likewise.
3779 (default_addr_space_pointer_mode): Likewise.
3780 (default_addr_space_address_mode): Likewise.
3781 * targhooks.h (default_eh_return_filter_mode): Likewise.
3782 (default_libgcc_cmp_return_mode): Likewise.
3783 (default_libgcc_shift_count_mode): Likewise.
3784 (default_unwind_word_mode): Likewise.
3785 (default_addr_space_pointer_mode): Likewise.
3786 (default_addr_space_address_mode): Likewise.
3787 (default_mode_rep_extended): Take a scalar_int_mode rather than
3789 (default_valid_pointer_mode): Likewise.
3790 (default_addr_space_valid_pointer_mode): Likewise.
3791 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
3793 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
3794 rather than a machine_mode.
3795 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
3796 for local variables.
3797 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
3798 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
3799 than a machine_mode.
3801 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3802 Alan Hayward <alan.hayward@arm.com>
3803 David Sherwood <david.sherwood@arm.com>
3805 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
3806 the mode argument to scalar_int_mode.
3807 (do_jump_by_parts_zero_rtx): Likewise.
3808 (do_jump_by_parts_equality_rtx): Likewise.
3809 (do_jump_by_parts_greater): Take a mode argument.
3810 (do_jump_by_parts_equality): Likewise.
3811 (do_jump_1): Update calls accordingly.
3813 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3814 Alan Hayward <alan.hayward@arm.com>
3815 David Sherwood <david.sherwood@arm.com>
3817 * is-a.h (safe_dyn_cast): New function.
3818 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
3819 (jump_table_for_label): Likewise.
3820 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
3821 instead of an rtx_insn *.
3822 (shorten_branches): Use dyn_cast instead of LABEL_P and
3823 JUMP_TABLE_DATA_P. Use jump_table_for_label and
3824 rtx_jump_table_data::get_data_mode.
3825 (final_scan_insn): Likewise.
3827 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3828 Alan Hayward <alan.hayward@arm.com>
3829 David Sherwood <david.sherwood@arm.com>
3831 * combine.c (try_combine): Use is_a <scalar_int_mode> when
3832 trying to combine a full-register integer set with a subreg
3835 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3836 Alan Hayward <alan.hayward@arm.com>
3837 David Sherwood <david.sherwood@arm.com>
3839 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
3840 that is always either address_mode or pointer_mode.
3842 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3843 Alan Hayward <alan.hayward@arm.com>
3844 David Sherwood <david.sherwood@arm.com>
3846 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
3847 when the two are known to be equal.
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 * simplify-rtx.c (simplify_const_unary_operation): Use
3854 is_a <scalar_int_mode> instead of checking for a nonzero
3855 precision. Forcibly convert op_mode to a scalar_int_mode
3856 in that case. More clearly differentiate the operand and
3857 result modes and use the former when deciding what the value
3858 of a count-bits operation should be. Use is_int_mode instead
3859 of checking for a MODE_INT. Remove redundant check for whether
3860 this mode has a zero precision.
3862 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3863 Alan Hayward <alan.hayward@arm.com>
3864 David Sherwood <david.sherwood@arm.com>
3866 * optabs.c (widen_leading): Change the type of the mode argument
3867 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
3868 (widen_bswap): Likewise.
3869 (expand_parity): Likewise.
3870 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
3871 (expand_ffs): Likewise.
3872 (epand_unop): Check for scalar integer modes before calling the
3875 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3876 Alan Hayward <alan.hayward@arm.com>
3877 David Sherwood <david.sherwood@arm.com>
3879 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
3881 (expand_expr_real_1): Update call accordingly.
3883 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3884 Alan Hayward <alan.hayward@arm.com>
3885 David Sherwood <david.sherwood@arm.com>
3887 * expmed.c (store_bit_field_using_insv): Add op0_mode and
3888 value_mode arguments. Use scalar_int_mode internally.
3889 (store_bit_field_1): Rename the new integer mode from imode
3890 to op0_mode and use it instead of GET_MODE (op0). Update calls
3891 to store_split_bit_field, store_bit_field_using_insv and
3892 store_fixed_bit_field.
3893 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
3894 Use scalar_int_mode internally. Use a bit count rather than a mode
3895 when calculating the largest bit size for get_best_mode.
3896 Update calls to store_split_bit_field and store_fixed_bit_field_1.
3897 (store_fixed_bit_field_1): Add mode and value_mode arguments.
3898 Remove assertion that OP0 has a scalar integer mode.
3899 (store_split_bit_field): Add op0_mode and value_mode arguments.
3900 Update calls to extract_fixed_bit_field.
3901 (extract_bit_field_using_extv): Add an op0_mode argument.
3902 Use scalar_int_mode internally.
3903 (extract_bit_field_1): Rename the new integer mode from imode to
3904 op0_mode and use it instead of GET_MODE (op0). Update calls to
3905 extract_split_bit_field, extract_bit_field_using_extv and
3906 extract_fixed_bit_field.
3907 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
3908 to extract_split_bit_field and extract_fixed_bit_field_1.
3909 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
3910 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
3912 (extract_split_bit_field): Add an op0_mode argument. Update call
3913 to extract_fixed_bit_field.
3915 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3916 Alan Hayward <alan.hayward@arm.com>
3917 David Sherwood <david.sherwood@arm.com>
3919 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
3920 * explow.c (hard_function_value): Likewise.
3921 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
3922 convert_to_mode call outside the loop.
3923 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
3924 for the mode iterator. Require the mode specified by max_pieces
3926 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
3928 (copy_blkmode_to_reg): Likewise.
3929 (set_storage_via_setmem): Likewise.
3930 * optabs.c (prepare_cmp_insn): Likewise.
3931 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
3932 * stor-layout.c (finish_bitfield_representative): Likewise.
3934 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3935 Alan Hayward <alan.hayward@arm.com>
3936 David Sherwood <david.sherwood@arm.com>
3938 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
3939 * expr.c (convert_move): Use them.
3940 (convert_modes): Likewise.
3941 (store_expr_with_bounds): Likewise.
3943 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3944 Alan Hayward <alan.hayward@arm.com>
3945 David Sherwood <david.sherwood@arm.com>
3947 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
3948 parameter for the mode of "x". Remove the "known_x", "known_mode"
3949 and "known_ret" arguments. Change the type of the mode argument
3951 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
3952 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
3953 (reg_num_sign_bit_copies_for_combine): Likewise.
3954 * rtlanal.c (nonzero_bits1): Likewise.
3955 (num_sign_bit_copies1): Likewise.
3956 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
3957 (reg_num_sign_bit_copies_general): Likewise.
3958 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
3959 (reg_nonzero_bits_general): Likewise.
3961 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3962 Alan Hayward <alan.hayward@arm.com>
3963 David Sherwood <david.sherwood@arm.com>
3965 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
3966 than in subroutines. Return 1 for non-integer modes.
3967 (cached_num_sign_bit_copies): Change the type of the mode parameter
3969 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
3970 classes. Handle CONST_INT_P first and then check whether X also
3971 has a scalar integer mode. Check the same thing for inner registers
3972 of a SUBREG and for values that are being extended or truncated.
3974 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3975 Alan Hayward <alan.hayward@arm.com>
3976 David Sherwood <david.sherwood@arm.com>
3978 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
3979 in subroutines. Return the mode mask for non-integer modes.
3980 (cached_nonzero_bits): Change the type of the mode parameter
3982 (nonzero_bits1): Likewise. Remove early exit for other mode
3983 classes. Handle CONST_INT_P first and then check whether X
3984 also has a scalar integer mode.
3986 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3987 Alan Hayward <alan.hayward@arm.com>
3988 David Sherwood <david.sherwood@arm.com>
3990 * expr.c (widest_int_mode_for_size): Make the comment match the code.
3991 Return a scalar_int_mode and assert that the size is greater than
3993 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
3994 (op_by_pieces_d::op_by_pieces_d): Likewise.
3995 (op_by_pieces_d::run): Likewise.
3996 (can_store_by_pieces): Likewise.
3998 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3999 Alan Hayward <alan.hayward@arm.com>
4000 David Sherwood <david.sherwood@arm.com>
4002 * combine.c (extract_left_shift): Add a mode argument and update
4004 (make_compound_operation_int): Change the type of the mode parameter
4005 to scalar_int_mode and update the call to extract_left_shift.
4007 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4008 Alan Hayward <alan.hayward@arm.com>
4009 David Sherwood <david.sherwood@arm.com>
4011 * combine.c (simplify_and_const_int): Change the type of the mode
4012 parameter to scalar_int_mode.
4013 (simplify_and_const_int_1): Likewise. Update recursive call.
4015 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4016 Alan Hayward <alan.hayward@arm.com>
4017 David Sherwood <david.sherwood@arm.com>
4019 * combine.c (simplify_compare_const): Check that the mode is a
4020 scalar_int_mode (rather than VOIDmode) before testing its
4022 (simplify_comparison): Move COMPARISON_P handling out of the
4023 loop and restrict the latter part of the loop to scalar_int_modes.
4024 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
4025 and when considering SUBREG_REGs. Use is_int_mode instead of
4026 checking GET_MODE_CLASS against MODE_INT.
4028 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4029 Alan Hayward <alan.hayward@arm.com>
4030 David Sherwood <david.sherwood@arm.com>
4032 * combine.c (try_widen_shift_mode): Move check for equal modes to...
4033 (simplify_shift_const_1): ...here. Use scalar_int_mode for
4034 shift_unit_mode and for modes involved in scalar shifts.
4036 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4037 Alan Hayward <alan.hayward@arm.com>
4038 David Sherwood <david.sherwood@arm.com>
4040 * combine.c (force_int_to_mode): New function, split out from...
4041 (force_to_mode): ...here. Keep xmode up-to-date and use it
4042 instead of GET_MODE (x).
4044 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4045 Alan Hayward <alan.hayward@arm.com>
4046 David Sherwood <david.sherwood@arm.com>
4048 * optabs-query.h (extraction_insn::struct_mode): Change type to
4049 opt_scalar_int_mode and update comment.
4050 (extraction_insn::field_mode): Change type to scalar_int_mode.
4051 (extraction_insn::pos_mode): Likewise.
4052 * combine.c (make_extraction): Update accordingly.
4053 * optabs-query.c (get_traditional_extraction_insn): Likewise.
4054 (get_optab_extraction_insn): Likewise.
4055 * recog.c (simplify_while_replacing): Likewise.
4056 * expmed.c (narrow_bit_field_mem): Change the type of the mode
4057 parameter to opt_scalar_int_mode.
4059 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4060 Alan Hayward <alan.hayward@arm.com>
4061 David Sherwood <david.sherwood@arm.com>
4063 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
4064 to a scalar_int_mode instead of a machine_mode.
4065 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
4066 (get_best_mode): Return a boolean and use a pointer argument to store
4067 the selected mode. Replace the limit mode parameter with a bit limit.
4068 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
4069 for the values returned by bit_field_mode_iterator::next_mode.
4070 (store_bit_field): Update call to get_best_mode.
4071 (store_fixed_bit_field): Likewise.
4072 (extract_fixed_bit_field): Likewise.
4073 * expr.c (optimize_bitfield_assignment_op): Likewise.
4074 * fold-const.c (optimize_bit_field_compare): Likewise.
4075 (fold_truth_andor_1): Likewise.
4076 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
4077 Update for new type of m_mode.
4078 (get_best_mode): As above.
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 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
4085 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
4086 (store_bit_field): Check is_a <scalar_int_mode> before calling
4087 strict_volatile_bitfield_p.
4088 (extract_bit_field): Likewise.
4090 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4091 Alan Hayward <alan.hayward@arm.com>
4092 David Sherwood <david.sherwood@arm.com>
4094 * target.def (cstore_mode): Return a scalar_int_mode.
4095 * doc/tm.texi: Regenerate.
4096 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
4097 * targhooks.h (default_cstore_mode): Likewise.
4098 * targhooks.c (default_cstore_mode): Likewise, using a forced
4100 * expmed.c (emit_cstore): Expect the target of the cstore to be
4103 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4104 Alan Hayward <alan.hayward@arm.com>
4105 David Sherwood <david.sherwood@arm.com>
4107 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
4109 (niter_desc): Likewise mode.
4110 (iv_analyze): Add a mode parameter.
4112 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
4113 and change its type to scalar_int_mode.
4114 * loop-iv.c: Update commentary at head of file.
4115 (iv_constant): Pass the mode paraeter before the rtx it describes
4116 and change its type to scalar_int_mode. Remove VOIDmode handling.
4117 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
4118 (iv_extend): Likewise.
4119 (shorten_into_mode): Likewise.
4120 (iv_add): Use scalar_int_mode.
4121 (iv_mult): Likewise.
4122 (iv_shift): Likewise.
4123 (canonicalize_iv_subregs): Likewise.
4124 (get_biv_step_1): Pass the outer_mode parameter before the rtx
4125 it describes and change its mode to scalar_int_mode. Also change
4126 the type of the returned inner_mode to scalar_int_mode.
4127 (get_biv_step): Likewise, turning outer_mode from a pointer
4128 into a direct parameter. Update call to get_biv_step_1.
4129 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
4130 iv_constant and get_biv_step.
4131 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
4132 and change its type to scalar_int_mode. Don't initialise iv->mode
4133 to VOIDmode and remove later checks for its still being VOIDmode.
4134 Update calls to iv_analyze_op and iv_analyze_expr. Check
4135 is_a <scalar_int_mode> when changing the mode under consideration.
4136 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
4137 Update call to iv_analyze_expr.
4138 (iv_analyze_op): Add a mode parameter. Reject subregs whose
4139 inner register is not also a scalar_int_mode. Update call to
4141 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
4142 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
4143 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
4144 separate mode class checks. Update calls to iv_analyze. Remove
4145 fix-up of VOIDmodes after iv_analyze_biv.
4146 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
4147 don't have a scalar_int_mode. Update call to biv_p.
4149 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4150 Alan Hayward <alan.hayward@arm.com>
4151 David Sherwood <david.sherwood@arm.com>
4153 * cfgexpand.c (convert_debug_memory_address): Use
4154 as_a <scalar_int_mode>.
4155 * combine.c (expand_compound_operation): Likewise.
4156 (make_extraction): Likewise.
4157 (change_zero_ext): Likewise.
4158 (simplify_comparison): Likewise.
4159 * cse.c (cse_insn): Likewise.
4160 * dwarf2out.c (minmax_loc_descriptor): Likewise.
4161 (mem_loc_descriptor): Likewise.
4162 (loc_descriptor): Likewise.
4163 * expmed.c (init_expmed_one_mode): Likewise.
4164 (synth_mult): Likewise.
4165 (emit_store_flag_1): Likewise.
4166 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
4167 of a comparison with size.
4168 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
4169 (reduce_to_bit_field_precision): Likewise.
4170 * function.c (expand_function_end): Likewise.
4171 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
4172 * loop-doloop.c (doloop_modify): Likewise.
4173 * optabs.c (expand_binop): Likewise.
4174 (expand_unop): Likewise.
4175 (expand_copysign_absneg): Likewise.
4176 (prepare_cmp_insn): Likewise.
4177 (maybe_legitimize_operand): Likewise.
4178 * recog.c (const_scalar_int_operand): Likewise.
4179 * rtlanal.c (get_address_mode): Likewise.
4180 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4181 (simplify_cond_clz_ctz): Likewise.
4182 * tree-nested.c (get_nl_goto_field): Likewise.
4183 * tree.c (build_vector_type_for_mode): Likewise.
4184 * var-tracking.c (use_narrower_mode): Likewise.
4186 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4187 Alan Hayward <alan.hayward@arm.com>
4188 David Sherwood <david.sherwood@arm.com>
4190 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
4191 * builtins.c (expand_builtin_signbit): Use it.
4192 * cfgexpand.c (expand_debug_expr): Likewise.
4193 * dojump.c (do_jump): Likewise.
4194 (do_compare_and_jump): Likewise.
4195 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
4196 * expmed.c (make_tree): Likewise.
4197 * expr.c (expand_expr_real_2): Likewise.
4198 (expand_expr_real_1): Likewise.
4199 (try_casesi): Likewise.
4200 * fold-const-call.c (fold_const_call_ss): Likewise.
4201 * fold-const.c (unextend): Likewise.
4202 (extract_muldiv_1): Likewise.
4203 (fold_single_bit_test): Likewise.
4204 (native_encode_int): Likewise.
4205 (native_encode_string): Likewise.
4206 (native_interpret_int): Likewise.
4207 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
4208 * internal-fn.c (expand_addsub_overflow): Likewise.
4209 (expand_neg_overflow): Likewise.
4210 (expand_mul_overflow): Likewise.
4211 (expand_arith_overflow): Likewise.
4212 * match.pd: Likewise.
4213 * stor-layout.c (layout_type): Likewise.
4214 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
4215 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
4216 * tree-ssanames.c (get_range_info): Likewise.
4217 * tree-switch-conversion.c (array_value_type) Likewise.
4218 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
4219 (vect_recog_divmod_pattern): Likewise.
4220 (vect_recog_mixed_size_cond_pattern): Likewise.
4221 * tree-vrp.c (extract_range_basic): Likewise.
4222 (simplify_float_conversion_using_ranges): Likewise.
4223 * tree.c (int_fits_type_p): Likewise.
4224 * ubsan.c (instrument_bool_enum_load): Likewise.
4225 * varasm.c (mergeable_string_section): Likewise.
4226 (narrowing_initializer_constant_valid_p): Likewise.
4227 (output_constant): Likewise.
4229 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4230 Alan Hayward <alan.hayward@arm.com>
4231 David Sherwood <david.sherwood@arm.com>
4233 * machmode.h (NARROWEST_INT_MODE): New macro.
4234 * expr.c (alignment_for_piecewise_move): Use it instead of
4235 GET_CLASS_NARROWEST_MODE (MODE_INT).
4236 (push_block): Likewise.
4237 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
4239 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
4241 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4242 Alan Hayward <alan.hayward@arm.com>
4243 David Sherwood <david.sherwood@arm.com>
4245 * postreload.c (move2add_valid_value_p): Change the type of the
4246 mode parameter to scalar_int_mode.
4247 (move2add_use_add2_insn): Add a mode parameter and use it instead
4249 (move2add_use_add3_insn): Likewise.
4250 (reload_cse_move2add): Update accordingly.
4252 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4253 Alan Hayward <alan.hayward@arm.com>
4254 David Sherwood <david.sherwood@arm.com>
4256 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
4258 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
4259 * optabs.c (expand_unop): Likewise.
4261 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4262 Alan Hayward <alan.hayward@arm.com>
4263 David Sherwood <david.sherwood@arm.com>
4265 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
4266 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
4267 (popcount_loc_descriptor): Likewise.
4268 (bswap_loc_descriptor): Likewise.
4269 (rotate_loc_descriptor): Likewise.
4270 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
4271 calling the functions above.
4273 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4274 Alan Hayward <alan.hayward@arm.com>
4275 David Sherwood <david.sherwood@arm.com>
4277 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
4279 (try_combine): Likewise.
4280 (simplify_if_then_else): Likewise.
4281 * cse.c (cse_insn): Likewise.
4282 * dwarf2out.c (mem_loc_descriptor): Likewise.
4283 * emit-rtl.c (gen_lowpart_common): Likewise.
4284 * simplify-rtx.c (simplify_truncation): Likewise.
4285 (simplify_binary_operation_1): Likewise.
4286 (simplify_const_relational_operation): Likewise.
4287 (simplify_ternary_operation): Likewise.
4288 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
4290 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4291 Alan Hayward <alan.hayward@arm.com>
4292 David Sherwood <david.sherwood@arm.com>
4294 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
4295 * reload.c (push_reload): Likewise.
4296 (find_reloads): Likewise.
4298 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4299 Alan Hayward <alan.hayward@arm.com>
4300 David Sherwood <david.sherwood@arm.com>
4302 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
4303 (make_compound_operation_int): Likewise.
4304 (change_zero_ext): Likewise.
4305 * expr.c (convert_move): Likewise.
4306 (convert_modes): Likewise.
4307 * fwprop.c (forward_propagate_subreg): Likewise.
4308 * loop-iv.c (get_biv_step_1): Likewise.
4309 * optabs.c (widen_operand): Likewise.
4310 * postreload.c (move2add_valid_value_p): Likewise.
4311 * recog.c (simplify_while_replacing): Likewise.
4312 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4313 (simplify_binary_operation_1): Likewise. Remove redundant
4314 mode equality check.
4316 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4317 Alan Hayward <alan.hayward@arm.com>
4318 David Sherwood <david.sherwood@arm.com>
4320 * combine.c (combine_simplify_rtx): Add checks for
4321 is_a <scalar_int_mode>.
4322 (simplify_if_then_else): Likewise.
4323 (make_field_assignment): Likewise.
4324 (simplify_comparison): Likewise.
4325 * ifcvt.c (noce_try_bitop): Likewise.
4326 * loop-invariant.c (canonicalize_address_mult): Likewise.
4327 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4329 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4330 Alan Hayward <alan.hayward@arm.com>
4331 David Sherwood <david.sherwood@arm.com>
4333 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
4334 is_a <scalar_int_mode> instead of != BLKmode.
4336 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4337 Alan Hayward <alan.hayward@arm.com>
4338 David Sherwood <david.sherwood@arm.com>
4340 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
4341 instead of != VOIDmode.
4342 * combine.c (if_then_else_cond): Likewise.
4343 (change_zero_ext): Likewise.
4344 * dwarf2out.c (mem_loc_descriptor): Likewise.
4345 (loc_descriptor): Likewise.
4346 * rtlanal.c (canonicalize_condition): Likewise.
4347 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
4349 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4350 Alan Hayward <alan.hayward@arm.com>
4351 David Sherwood <david.sherwood@arm.com>
4353 * simplify-rtx.c (simplify_binary_operation_1): Use
4354 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
4356 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4357 Alan Hayward <alan.hayward@arm.com>
4358 David Sherwood <david.sherwood@arm.com>
4360 * wide-int.h (int_traits<unsigned char>) New class.
4361 (int_traits<unsigned short>) Likewise.
4362 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
4363 Use GET_MODE_UNIT_PRECISION and remove redundant test for
4365 * combine.c (set_nonzero_bits_and_sign_copies): Use
4366 is_a <scalar_int_mode>.
4367 (find_split_point): Likewise.
4368 (combine_simplify_rtx): Likewise.
4369 (simplify_logical): Likewise.
4370 (expand_compound_operation): Likewise.
4371 (expand_field_assignment): Likewise.
4372 (make_compound_operation): Likewise.
4373 (extended_count): Likewise.
4374 (change_zero_ext): Likewise.
4375 (simplify_comparison): Likewise.
4376 * dwarf2out.c (scompare_loc_descriptor): Likewise.
4377 (ucompare_loc_descriptor): Likewise.
4378 (minmax_loc_descriptor): Likewise.
4379 (mem_loc_descriptor): Likewise.
4380 (loc_descriptor): Likewise.
4381 * expmed.c (init_expmed_one_mode): Likewise.
4382 * lra-constraints.c (lra_constraint_offset): Likewise.
4383 * optabs.c (prepare_libcall_arg): Likewise.
4384 * postreload.c (move2add_note_store): Likewise.
4385 * reload.c (operands_match_p): Likewise.
4386 * rtl.h (load_extend_op): Likewise.
4387 * rtlhooks.c (gen_lowpart_general): Likewise.
4388 * simplify-rtx.c (simplify_truncation): Likewise.
4389 (simplify_unary_operation_1): Likewise.
4390 (simplify_binary_operation_1): Likewise.
4391 (simplify_const_binary_operation): Likewise.
4392 (simplify_const_relational_operation): Likewise.
4393 (simplify_subreg): Likewise.
4394 * stor-layout.c (bitwise_mode_for_mode): Likewise.
4395 * var-tracking.c (adjust_mems): Likewise.
4396 (prepare_call_arguments): Likewise.
4398 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4399 Alan Hayward <alan.hayward@arm.com>
4400 David Sherwood <david.sherwood@arm.com>
4402 * machmode.h (is_int_mode): New fuction.
4403 * combine.c (find_split_point): Use it.
4404 (combine_simplify_rtx): Likewise.
4405 (simplify_if_then_else): Likewise.
4406 (simplify_set): Likewise.
4407 (simplify_shift_const_1): Likewise.
4408 (simplify_comparison): Likewise.
4409 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
4410 * cse.c (notreg_cost): Likewise.
4411 (cse_insn): Likewise.
4412 * cselib.c (cselib_lookup_1): Likewise.
4413 * dojump.c (do_jump_1): Likewise.
4414 (do_compare_rtx_and_jump): Likewise.
4415 * dse.c (get_call_args): Likewise.
4416 * dwarf2out.c (rtl_for_decl_init): Likewise.
4417 (native_encode_initializer): Likewise.
4418 * expmed.c (emit_store_flag_1): Likewise.
4419 (emit_store_flag): Likewise.
4420 * expr.c (convert_modes): Likewise.
4421 (store_field): Likewise.
4422 (expand_expr_real_1): Likewise.
4423 * fold-const.c (fold_read_from_constant_string): Likewise.
4424 * gimple-ssa-sprintf.c (get_format_string): Likewise.
4425 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
4426 * optabs.c (expand_binop): Likewise.
4427 (expand_unop): Likewise.
4428 (expand_abs_nojump): Likewise.
4429 (expand_one_cmpl_abs_nojump): Likewise.
4430 * simplify-rtx.c (mode_signbit_p): Likewise.
4431 (val_signbit_p): Likewise.
4432 (val_signbit_known_set_p): Likewise.
4433 (val_signbit_known_clear_p): Likewise.
4434 (simplify_relational_operation_1): Likewise.
4435 * tree.c (vector_type_mode): Likewise.
4437 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4438 Alan Hayward <alan.hayward@arm.com>
4439 David Sherwood <david.sherwood@arm.com>
4441 * machmode.h (smallest_mode_for_size): Fix formatting.
4442 (smallest_int_mode_for_size): New function.
4443 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
4444 instead of smallest_mode_for_size.
4445 * combine.c (make_extraction): Likewise.
4446 * config/arc/arc.c (arc_expand_movmem): Likewise.
4447 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
4448 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
4449 * config/s390/s390.c (s390_expand_insv): Likewise.
4450 * config/sparc/sparc.c (assign_int_registers): Likewise.
4451 * config/spu/spu.c (spu_function_value): Likewise.
4452 (spu_function_arg): Likewise.
4453 * coverage.c (get_gcov_type): Likewise.
4454 (get_gcov_unsigned_t): Likewise.
4455 * dse.c (find_shift_sequence): Likewise.
4456 * expmed.c (store_bit_field_1): Likewise.
4457 * expr.c (convert_move): Likewise.
4458 (store_field): Likewise.
4459 * internal-fn.c (expand_arith_overflow): Likewise.
4460 * optabs-query.c (get_best_extraction_insn): Likewise.
4461 * optabs.c (expand_twoval_binop_libfunc): Likewise.
4462 * stor-layout.c (layout_type): Likewise.
4463 (initialize_sizetypes): Likewise.
4464 * targhooks.c (default_get_mask_mode): Likewise.
4465 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
4467 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4468 Alan Hayward <alan.hayward@arm.com>
4469 David Sherwood <david.sherwood@arm.com>
4471 * machmode.h (opt_mode::else_blk): New function.
4472 (int_mode_for_mode): Declare.
4473 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
4474 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
4476 * cfgexpand.c (expand_debug_expr): Likewise.
4477 * combine.c (gen_lowpart_or_truncate): Likewise.
4478 (gen_lowpart_for_combine): Likewise.
4479 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
4480 * config/avr/avr.c (avr_to_int_mode): Likewise.
4481 (avr_out_plus_1): Likewise.
4482 (avr_out_plus): Likewise.
4483 (avr_out_round): Likewise.
4484 * config/i386/i386.c (ix86_split_to_parts): Likewise.
4485 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
4486 (s390_expand_vcond): Likewise.
4487 * config/spu/spu.c (spu_split_immediate): Likewise.
4488 (spu_expand_mov): Likewise.
4489 * dse.c (get_stored_val): Likewise.
4490 * expmed.c (store_bit_field_1): Likewise.
4491 (convert_extracted_bit_field): Use int_mode_for_mode instead of
4493 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
4494 (extract_low_bits): Likewise.
4495 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
4496 handling rather than repeating the check.
4497 (emit_group_store): Likewise.
4498 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
4499 * optabs.c (expand_absneg_bit): Likewise.
4500 (expand_copysign_absneg): Likewise.
4501 (expand_copysign_bit): Likewise.
4502 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
4503 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
4504 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
4505 * var-tracking.c (prepare_call_arguments): Likewise.
4506 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
4507 int_mode_for_mode instead of mode_for_size.
4508 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
4510 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4511 Alan Hayward <alan.hayward@arm.com>
4512 David Sherwood <david.sherwood@arm.com>
4514 * machmode.h (int_mode_for_size): New function.
4515 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
4516 instead of mode_for_size.
4517 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
4519 * combine.c (expand_field_assignment): Use int_mode_for_size
4520 instead of mode_for_size.
4521 (make_extraction): Likewise.
4522 (simplify_shift_const_1): Likewise.
4523 (simplify_comparison): Likewise.
4524 * dojump.c (do_jump): Likewise.
4525 * dwarf2out.c (mem_loc_descriptor): Likewise.
4526 * emit-rtl.c (init_derived_machine_modes): Likewise.
4527 * expmed.c (flip_storage_order): Likewise.
4528 (convert_extracted_bit_field): Likewise.
4529 * expr.c (copy_blkmode_from_reg): Likewise.
4530 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
4531 * internal-fn.c (expand_mul_overflow): Likewise.
4532 * lower-subreg.c (simple_move): Likewise.
4533 * optabs-libfuncs.c (init_optabs): Likewise.
4534 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4535 * tree.c (vector_type_mode): Likewise.
4536 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
4537 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
4538 * tree-vect-generic.c (expand_vector_parallel): Likewise.
4539 * tree-vect-stmts.c (vectorizable_load): Likewise.
4540 (vectorizable_store): Likewise.
4542 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4543 Alan Hayward <alan.hayward@arm.com>
4544 David Sherwood <david.sherwood@arm.com>
4546 * coretypes.h (pod_mode): New type.
4547 (scalar_int_mode_pod): New typedef.
4548 * machmode.h (pod_mode): New class.
4549 (int_n_data_t::m): Change type to scalar_int_mode_pod.
4550 * genmodes.c (emit_mode_int_n): Update accordingly.
4551 * lower-subreg.h (target_lower_subreg): Change type to
4552 scalar_int_mode_pod.
4553 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
4554 scalar_int_mode_pod.
4556 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4557 Alan Hayward <alan.hayward@arm.com>
4558 David Sherwood <david.sherwood@arm.com>
4560 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
4561 machine_mode to scalar_int_mode.
4562 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
4563 (rs6000_option_override_internal): Remove cast to int.
4564 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
4565 machine_mode to scalar_int_mode.
4566 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
4567 (rs6000_option_override_internal): Remove cast to int.
4568 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
4569 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
4571 * config/s390/s390.c (s390_expand_builtin): Likewise.
4572 * coretypes.h (scalar_int_mode): New type.
4573 (opt_scalar_int_mode): New typedef.
4574 * machmode.h (scalar_int_mode): New class.
4575 (scalar_int_mode::includes_p): New function.
4576 (byte_mode): Change type to scalar_int_mode.
4577 (word_mode): Likewise.
4578 (ptr_mode): Likewise.
4579 * emit-rtl.c (byte_mode): Likewise.
4580 (word_mode): Likewise.
4581 (ptr_mode): Likewise.
4582 (init_derived_machine_modes): Update accordingly.
4583 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
4584 and MODE_PARTIAL_INT.
4585 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
4586 opt_scalar_int_mode.
4588 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4589 Alan Hayward <alan.hayward@arm.com>
4590 David Sherwood <david.sherwood@arm.com>
4592 * target.def (libgcc_floating_mode_supported_p): Take a
4594 * doc/tm.texi: Regenerate.
4595 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
4597 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
4598 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
4601 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4602 Alan Hayward <alan.hayward@arm.com>
4603 David Sherwood <david.sherwood@arm.com>
4605 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
4606 * doc/tm.texi: Regenerate.
4607 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
4608 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
4609 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
4610 * targhooks.h (default_floatn_mode): Likewise.
4611 * targhooks.c (default_floatn_mode): Likewise.
4612 * tree.c (build_common_tree_nodes): Update accordingly.
4614 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4615 Alan Hayward <alan.hayward@arm.com>
4616 David Sherwood <david.sherwood@arm.com>
4618 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
4619 (mode_iterator::iterate_p): Likewise.
4620 (mode_iterator::get_wider): Likewise.
4621 * expr.c (init_expr_target): Use opt_scalar_float_mode.
4623 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4624 Alan Hayward <alan.hayward@arm.com>
4625 David Sherwood <david.sherwood@arm.com>
4627 * coretypes.h (opt_scalar_float_mode): New typedef.
4628 * machmode.h (float_mode_for_size): New function.
4629 * emit-rtl.c (double_mode): Delete.
4630 (init_emit_once): Use float_mode_for_size.
4631 * stor-layout.c (layout_type): Likewise.
4632 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
4634 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4635 Alan Hayward <alan.hayward@arm.com>
4636 David Sherwood <david.sherwood@arm.com>
4638 * output.h (assemble_real): Take a scalar_float_mode.
4639 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
4640 * config/arm/arm.md (consttable_4): Likewise.
4641 (consttable_8): Likewise.
4642 (consttable_16): Likewise.
4643 * config/mips/mips.md (consttable_float): Likewise.
4644 * config/s390/s390.c (s390_output_pool_entry): Likewise.
4645 * varasm.c (assemble_real): Take a scalar_float_mode.
4646 (output_constant_pool_2): Update accordingly.
4647 (output_constant): Likewise.
4649 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4650 Alan Hayward <alan.hayward@arm.com>
4651 David Sherwood <david.sherwood@arm.com>
4653 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
4654 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
4655 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
4656 (native_encode_real): Likewise.
4657 (native_interpret_real): Likewise.
4658 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
4659 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
4661 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4662 Alan Hayward <alan.hayward@arm.com>
4663 David Sherwood <david.sherwood@arm.com>
4665 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
4666 <scalar_float_mode>. Simplify.
4667 (gen_extend_conv_libfunc): Likewise.
4669 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4670 Alan Hayward <alan.hayward@arm.com>
4671 David Sherwood <david.sherwood@arm.com>
4673 * coretypes.h (scalar_float_mode): New type.
4674 * machmode.h (mode_traits::from_int): Use machine_mode if
4675 USE_ENUM_MODES is defined.
4676 (is_a): New function.
4678 (dyn_cast): Likewise.
4679 (scalar_float_mode): New class.
4680 (scalar_float_mode::includes_p): New function.
4681 (is_float_mode): Likewise.
4682 * gdbhooks.py (MachineModePrinter): New class.
4683 (build_pretty_printer): Use it for scalar_float_mode.
4684 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
4685 (format_helper::format_helper): Turn into a template.
4686 * genmodes.c (get_mode_class): New function.
4687 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
4688 or machine_mode if none.
4689 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
4690 as_a <scalar_float_mode>.
4691 * dwarf2out.c (mem_loc_descriptor): Likewise.
4692 (insert_float): Likewise.
4693 (add_const_value_attribute): Likewise.
4694 * simplify-rtx.c (simplify_immed_subreg): Likewise.
4695 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
4696 (expand_unop): Update accordingly.
4697 (expand_abs_nojump): Likewise.
4698 (expand_copysign_absneg): Take a scalar_float_mode.
4699 (expand_copysign_bit): Likewise.
4700 (expand_copysign): Update accordingly.
4702 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4703 Alan Hayward <alan.hayward@arm.com>
4704 David Sherwood <david.sherwood@arm.com>
4706 * coretypes.h (opt_mode): New class.
4707 * machmode.h (opt_mode): Likewise.
4708 (opt_mode::else_void): New function.
4709 (opt_mode::require): Likewise.
4710 (opt_mode::exists): Likewise.
4711 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
4712 (GET_MODE_2XWIDER_MODE): Likewise.
4713 (mode_iterator::get_wider): Update accordingly.
4714 (mode_iterator::get_2xwider): Likewise.
4715 (mode_iterator::get_known_wider): Likewise, turning into a template.
4716 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
4717 forcing a wider mode to exist.
4718 * config/cr16/cr16.h (LONG_REG_P): Likewise.
4719 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
4720 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
4721 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
4722 * lower-subreg.c (init_lower_subreg): Likewise.
4723 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
4724 on the final iteration.
4725 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
4726 a wider mode exists before asking for a move pattern.
4727 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
4728 forcing a wider mode to exist.
4729 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
4730 returning false if no such mode exists.
4731 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
4732 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
4733 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
4734 Avoid checking for a MODE_INT if we already know the mode is not a
4736 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
4737 forcing a wider mode to exist.
4738 (expmed_mult_highpart_optab): Likewise.
4739 (expmed_mult_highpart): Likewise.
4740 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
4742 * lto-streamer-in.c (lto_input_mode_table): Likewise.
4743 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
4744 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
4745 * internal-fn.c (expand_mul_overflow): Update use of
4746 GET_MODE_2XWIDER_MODE.
4747 * omp-low.c (omp_clause_aligned_alignment): Likewise.
4748 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
4749 GET_MODE_WIDER_MODE.
4750 (convert_plusminus_to_widen): Likewise.
4751 * tree-switch-conversion.c (array_value_type): Likewise.
4752 * var-tracking.c (emit_note_insn_var_location): Likewise.
4753 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
4754 Return false inside rather than outside the loop if no wider mode
4756 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
4757 and GET_MODE_2XWIDER_MODE
4758 (can_compare_p): Use else_void.
4759 * gdbhooks.py (OptMachineModePrinter): New class.
4760 (build_pretty_printer): Use it for opt_mode.
4762 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4763 Alan Hayward <alan.hayward@arm.com>
4764 David Sherwood <david.sherwood@arm.com>
4766 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
4767 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
4769 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4770 Alan Hayward <alan.hayward@arm.com>
4771 David Sherwood <david.sherwood@arm.com>
4773 * machmode.h (mode_traits): New structure.
4774 (get_narrowest_mode): New function.
4775 (mode_iterator::start): Likewise.
4776 (mode_iterator::iterate_p): Likewise.
4777 (mode_iterator::get_wider): Likewise.
4778 (mode_iterator::get_known_wider): Likewise.
4779 (mode_iterator::get_2xwider): Likewise.
4780 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
4781 (FOR_EACH_MODE): Likewise.
4782 (FOR_EACH_MODE_FROM): Likewise.
4783 (FOR_EACH_MODE_UNTIL): Likewise.
4784 (FOR_EACH_WIDER_MODE): Likewise.
4785 (FOR_EACH_2XWIDER_MODE): Likewise.
4786 * builtins.c (expand_builtin_strlen): Use new mode iterators.
4787 * combine.c (simplify_comparison): Likewise
4788 * config/i386/i386.c (type_natural_mode): Likewise.
4789 * cse.c (cse_insn): Likewise.
4790 * dse.c (find_shift_sequence): Likewise.
4791 * emit-rtl.c (init_derived_machine_modes): Likewise.
4792 (init_emit_once): Likewise.
4793 * explow.c (hard_function_value): Likewise.
4794 * expmed.c (extract_fixed_bit_field_1): Likewise.
4795 (extract_bit_field_1): Likewise.
4796 (expand_divmod): Likewise.
4797 (emit_store_flag_1): Likewise.
4798 * expr.c (init_expr_target): Likewise.
4799 (convert_move): Likewise.
4800 (alignment_for_piecewise_move): Likewise.
4801 (widest_int_mode_for_size): Likewise.
4802 (emit_block_move_via_movmem): Likewise.
4803 (copy_blkmode_to_reg): Likewise.
4804 (set_storage_via_setmem): Likewise.
4805 (compress_float_constant): Likewise.
4806 * omp-low.c (omp_clause_aligned_alignment): Likewise.
4807 * optabs-query.c (get_best_extraction_insn): Likewise.
4808 * optabs.c (expand_binop): Likewise.
4809 (expand_twoval_unop): Likewise.
4810 (expand_twoval_binop): Likewise.
4811 (widen_leading): Likewise.
4812 (widen_bswap): Likewise.
4813 (expand_parity): Likewise.
4814 (expand_unop): Likewise.
4815 (prepare_cmp_insn): Likewise.
4816 (prepare_float_lib_cmp): Likewise.
4817 (expand_float): Likewise.
4818 (expand_fix): Likewise.
4819 (expand_sfix_optab): Likewise.
4820 * postreload.c (move2add_use_add2_insn): Likewise.
4821 * reg-stack.c (reg_to_stack): Likewise.
4822 * reginfo.c (choose_hard_reg_mode): Likewise.
4823 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
4824 * stor-layout.c (mode_for_size): Likewise.
4825 (smallest_mode_for_size): Likewise.
4826 (mode_for_vector): Likewise.
4827 (finish_bitfield_representative): Likewise.
4828 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
4829 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
4830 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
4831 * var-tracking.c (prepare_call_arguments): Likewise.
4833 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4834 Alan Hayward <alan.hayward@arm.com>
4835 David Sherwood <david.sherwood@arm.com>
4837 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
4838 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
4839 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
4840 * machmode.h (mode_size): Move earlier in file.
4841 (mode_precision): Likewise.
4842 (mode_inner): Likewise.
4843 (mode_nunits): Likewise.
4844 (mode_unit_size): Likewise.
4845 (unit_unit_precision): Likewise.
4846 (mode_wider): Likewise.
4847 (mode_2xwider): Likewise.
4848 (machine_mode): New class.
4849 (mode_to_bytes): New function.
4850 (mode_to_bits): Likewise.
4851 (mode_to_precision): Likewise.
4852 (mode_to_inner): Likewise.
4853 (mode_to_unit_size): Likewise.
4854 (mode_to_unit_precision): Likewise.
4855 (mode_to_nunits): Likewise.
4856 (GET_MODE_SIZE): Use mode_to_bytes.
4857 (GET_MODE_BITSIZE): Use mode_to_bits.
4858 (GET_MODE_PRECISION): Use mode_to_precision.
4859 (GET_MODE_INNER): Use mode_to_inner.
4860 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
4861 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
4862 (GET_MODE_NUNITS): Use mode_to_nunits.
4863 * system.h (ALWAYS_INLINE): New macro.
4864 * config/powerpcspe/powerpcspe-c.c
4865 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
4866 int for arg1_mode and arg2_mode.
4868 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4869 Alan Hayward <alan.hayward@arm.com>
4870 David Sherwood <david.sherwood@arm.com>
4872 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
4873 Prefix mode names with E_ in case statements.
4874 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4875 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
4876 (aarch64_split_simd_move): Likewise.
4877 (aarch64_gen_storewb_pair): Likewise.
4878 (aarch64_gen_loadwb_pair): Likewise.
4879 (aarch64_gen_store_pair): Likewise.
4880 (aarch64_gen_load_pair): Likewise.
4881 (aarch64_get_condition_code_1): Likewise.
4882 (aarch64_constant_pool_reload_icode): Likewise.
4883 (get_rsqrte_type): Likewise.
4884 (get_rsqrts_type): Likewise.
4885 (get_recpe_type): Likewise.
4886 (get_recps_type): Likewise.
4887 (aarch64_gimplify_va_arg_expr): Likewise.
4888 (aarch64_simd_container_mode): Likewise.
4889 (aarch64_emit_load_exclusive): Likewise.
4890 (aarch64_emit_store_exclusive): Likewise.
4891 (aarch64_expand_compare_and_swap): Likewise.
4892 (aarch64_gen_atomic_cas): Likewise.
4893 (aarch64_emit_bic): Likewise.
4894 (aarch64_emit_atomic_swap): Likewise.
4895 (aarch64_emit_atomic_load_op): Likewise.
4896 (aarch64_evpc_trn): Likewise.
4897 (aarch64_evpc_uzp): Likewise.
4898 (aarch64_evpc_zip): Likewise.
4899 (aarch64_evpc_ext): Likewise.
4900 (aarch64_evpc_rev): Likewise.
4901 (aarch64_evpc_dup): Likewise.
4902 (aarch64_gen_ccmp_first): Likewise.
4903 (aarch64_gen_ccmp_next): Likewise.
4904 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
4905 (alpha_emit_xfloating_libcall): Likewise.
4906 (emit_insxl): Likewise.
4907 (alpha_arg_type): Likewise.
4908 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
4909 (arc_preferred_simd_mode): Likewise.
4910 (arc_secondary_reload): Likewise.
4911 (get_arc_condition_code): Likewise.
4912 (arc_print_operand): Likewise.
4913 (arc_legitimate_constant_p): Likewise.
4914 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4915 * config/arc/arc.md (casesi_load): Likewise.
4916 (casesi_compact_jump): Likewise.
4917 * config/arc/predicates.md (proper_comparison_operator): Likewise.
4918 (cc_use_register): Likewise.
4919 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4920 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
4921 (arm_init_iwmmxt_builtins): Likewise.
4922 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
4923 (neon_expand_vector_init): Likewise.
4924 (arm_attr_length_move_neon): Likewise.
4925 (maybe_get_arm_condition_code): Likewise.
4926 (arm_emit_vector_const): Likewise.
4927 (arm_preferred_simd_mode): Likewise.
4928 (arm_output_iwmmxt_tinsr): Likewise.
4929 (thumb1_output_casesi): Likewise.
4930 (thumb2_output_casesi): Likewise.
4931 (arm_emit_load_exclusive): Likewise.
4932 (arm_emit_store_exclusive): Likewise.
4933 (arm_expand_compare_and_swap): Likewise.
4934 (arm_evpc_neon_vuzp): Likewise.
4935 (arm_evpc_neon_vzip): Likewise.
4936 (arm_evpc_neon_vrev): Likewise.
4937 (arm_evpc_neon_vtrn): Likewise.
4938 (arm_evpc_neon_vext): Likewise.
4939 (arm_validize_comparison): Likewise.
4940 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
4941 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
4942 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
4943 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
4944 (c6x_preferred_simd_mode): Likewise.
4945 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
4946 (epiphany_rtx_costs): Likewise.
4947 * config/epiphany/predicates.md (proper_comparison_operator):
4949 * config/frv/frv.c (condexec_memory_operand): Likewise.
4950 (frv_emit_move): Likewise.
4951 (output_move_single): Likewise.
4952 (output_condmove_single): Likewise.
4953 (frv_hard_regno_mode_ok): Likewise.
4954 (frv_matching_accg_mode): Likewise.
4955 * config/h8300/h8300.c (split_adds_subs): Likewise.
4956 (h8300_rtx_costs): Likewise.
4957 (h8300_print_operand): Likewise.
4958 (compute_mov_length): Likewise.
4959 (output_logical_op): Likewise.
4960 (compute_logical_op_length): Likewise.
4961 (compute_logical_op_cc): Likewise.
4962 (h8300_shift_needs_scratch_p): Likewise.
4963 (output_a_shift): Likewise.
4964 (compute_a_shift_length): Likewise.
4965 (compute_a_shift_cc): Likewise.
4966 (expand_a_rotate): Likewise.
4967 (output_a_rotate): Likewise.
4968 * config/i386/i386.c (classify_argument): Likewise.
4969 (function_arg_advance_32): Likewise.
4970 (function_arg_32): Likewise.
4971 (function_arg_64): Likewise.
4972 (function_value_64): Likewise.
4973 (ix86_gimplify_va_arg): Likewise.
4974 (ix86_legitimate_constant_p): Likewise.
4975 (put_condition_code): Likewise.
4976 (split_double_mode): Likewise.
4977 (ix86_avx256_split_vector_move_misalign): Likewise.
4978 (ix86_expand_vector_logical_operator): Likewise.
4979 (ix86_split_idivmod): Likewise.
4980 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
4981 (ix86_build_const_vector): Likewise.
4982 (ix86_build_signbit_mask): Likewise.
4983 (ix86_match_ccmode): Likewise.
4984 (ix86_cc_modes_compatible): Likewise.
4985 (ix86_expand_branch): Likewise.
4986 (ix86_expand_sse_cmp): Likewise.
4987 (ix86_expand_sse_movcc): Likewise.
4988 (ix86_expand_int_sse_cmp): Likewise.
4989 (ix86_expand_vec_perm_vpermi2): Likewise.
4990 (ix86_expand_vec_perm): Likewise.
4991 (ix86_expand_sse_unpack): Likewise.
4992 (ix86_expand_int_addcc): Likewise.
4993 (ix86_split_to_parts): Likewise.
4994 (ix86_vectorize_builtin_gather): Likewise.
4995 (ix86_vectorize_builtin_scatter): Likewise.
4996 (avx_vpermilp_parallel): Likewise.
4997 (inline_memory_move_cost): Likewise.
4998 (ix86_tieable_integer_mode_p): Likewise.
4999 (x86_maybe_negate_const_int): Likewise.
5000 (ix86_expand_vector_init_duplicate): Likewise.
5001 (ix86_expand_vector_init_one_nonzero): Likewise.
5002 (ix86_expand_vector_init_one_var): Likewise.
5003 (ix86_expand_vector_init_concat): Likewise.
5004 (ix86_expand_vector_init_interleave): Likewise.
5005 (ix86_expand_vector_init_general): Likewise.
5006 (ix86_expand_vector_set): Likewise.
5007 (ix86_expand_vector_extract): Likewise.
5008 (emit_reduc_half): Likewise.
5009 (ix86_emit_i387_round): Likewise.
5010 (ix86_mangle_type): Likewise.
5011 (ix86_expand_round_sse4): Likewise.
5012 (expand_vec_perm_blend): Likewise.
5013 (canonicalize_vector_int_perm): Likewise.
5014 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
5015 (expand_vec_perm_1): Likewise.
5016 (expand_vec_perm_interleave3): Likewise.
5017 (expand_vec_perm_even_odd_pack): Likewise.
5018 (expand_vec_perm_even_odd_1): Likewise.
5019 (expand_vec_perm_broadcast_1): Likewise.
5020 (ix86_vectorize_vec_perm_const_ok): Likewise.
5021 (ix86_expand_vecop_qihi): Likewise.
5022 (ix86_expand_mul_widen_hilo): Likewise.
5023 (ix86_expand_sse2_abs): Likewise.
5024 (ix86_expand_pextr): Likewise.
5025 (ix86_expand_pinsr): Likewise.
5026 (ix86_preferred_simd_mode): Likewise.
5027 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
5028 * config/i386/sse.md (*andnot<mode>3): Likewise.
5029 (<mask_codefor><code><mode>3<mask_name>): Likewise.
5030 (*<code><mode>3): Likewise.
5031 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
5032 (ia64_expand_atomic_op): Likewise.
5033 (ia64_arg_type): Likewise.
5034 (ia64_mode_to_int): Likewise.
5035 (ia64_scalar_mode_supported_p): Likewise.
5036 (ia64_vector_mode_supported_p): Likewise.
5037 (expand_vec_perm_broadcast): Likewise.
5038 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
5039 (iq2000_function_arg_advance): Likewise.
5040 (iq2000_function_arg): Likewise.
5041 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
5042 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
5043 (m68k_libcall_value): Likewise.
5044 (m68k_function_value): Likewise.
5045 (sched_attr_op_type): Likewise.
5046 * config/mcore/mcore.c (mcore_output_move): Likewise.
5047 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
5049 (microblaze_function_arg): Likewise.
5050 * config/mips/mips.c (mips16_build_call_stub): Likewise.
5051 (mips_print_operand): Likewise.
5052 (mips_mode_ok_for_mov_fmt_p): Likewise.
5053 (mips_vector_mode_supported_p): Likewise.
5054 (mips_preferred_simd_mode): Likewise.
5055 (mips_expand_vpc_loongson_even_odd): Likewise.
5056 (mips_expand_vec_unpack): Likewise.
5057 (mips_expand_vi_broadcast): Likewise.
5058 (mips_expand_vector_init): Likewise.
5059 (mips_expand_vec_reduc): Likewise.
5060 (mips_expand_msa_cmp): Likewise.
5061 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
5062 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
5063 (cc_flags_for_mode): Likewise.
5064 * config/msp430/msp430.c (msp430_print_operand): Likewise.
5065 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
5066 (nds32_output_casesi_pc_relative): Likewise.
5067 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
5068 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
5069 (nvptx_gen_unpack): Likewise.
5070 (nvptx_gen_pack): Likewise.
5071 (nvptx_gen_shuffle): Likewise.
5072 (nvptx_gen_wcast): Likewise.
5073 (nvptx_preferred_simd_mode): Likewise.
5074 * config/pa/pa.c (pa_secondary_reload): Likewise.
5075 * config/pa/predicates.md (base14_operand): Likewise.
5076 * config/powerpcspe/powerpcspe-c.c
5077 (altivec_resolve_overloaded_builtin): Likewise.
5078 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
5080 (rs6000_preferred_simd_mode): Likewise.
5081 (output_vec_const_move): Likewise.
5082 (rs6000_expand_vector_extract): Likewise.
5083 (rs6000_split_vec_extract_var): Likewise.
5084 (reg_offset_addressing_ok_p): Likewise.
5085 (rs6000_legitimate_offset_address_p): Likewise.
5086 (rs6000_legitimize_address): Likewise.
5087 (rs6000_emit_set_const): Likewise.
5088 (rs6000_const_vec): Likewise.
5089 (rs6000_emit_move): Likewise.
5090 (spe_build_register_parallel): Likewise.
5091 (rs6000_darwin64_record_arg_recurse): Likewise.
5092 (swap_selector_for_mode): Likewise.
5093 (spe_init_builtins): Likewise.
5094 (paired_init_builtins): Likewise.
5095 (altivec_init_builtins): Likewise.
5096 (do_load_for_compare): Likewise.
5097 (rs6000_generate_compare): Likewise.
5098 (rs6000_expand_float128_convert): Likewise.
5099 (emit_load_locked): Likewise.
5100 (emit_store_conditional): Likewise.
5101 (rs6000_output_function_epilogue): Likewise.
5102 (rs6000_handle_altivec_attribute): Likewise.
5103 (rs6000_function_value): Likewise.
5104 (emit_fusion_gpr_load): Likewise.
5105 (emit_fusion_p9_load): Likewise.
5106 (emit_fusion_p9_store): Likewise.
5107 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
5108 (fusion_gpr_mem_load): Likewise.
5109 (fusion_addis_mem_combo_load): Likewise.
5110 (fusion_addis_mem_combo_store): Likewise.
5111 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
5112 (fusion_gpr_mem_load): Likewise.
5113 (fusion_addis_mem_combo_load): Likewise.
5114 (fusion_addis_mem_combo_store): Likewise.
5115 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5117 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
5118 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
5119 (rs6000_preferred_simd_mode): Likewise.
5120 (output_vec_const_move): Likewise.
5121 (rs6000_expand_vector_extract): Likewise.
5122 (rs6000_split_vec_extract_var): Likewise.
5123 (reg_offset_addressing_ok_p): Likewise.
5124 (rs6000_legitimate_offset_address_p): Likewise.
5125 (rs6000_legitimize_address): Likewise.
5126 (rs6000_emit_set_const): Likewise.
5127 (rs6000_const_vec): Likewise.
5128 (rs6000_emit_move): Likewise.
5129 (rs6000_darwin64_record_arg_recurse): Likewise.
5130 (swap_selector_for_mode): Likewise.
5131 (paired_init_builtins): Likewise.
5132 (altivec_init_builtins): Likewise.
5133 (rs6000_expand_float128_convert): Likewise.
5134 (emit_load_locked): Likewise.
5135 (emit_store_conditional): Likewise.
5136 (rs6000_output_function_epilogue): Likewise.
5137 (rs6000_handle_altivec_attribute): Likewise.
5138 (rs6000_function_value): Likewise.
5139 (emit_fusion_gpr_load): Likewise.
5140 (emit_fusion_p9_load): Likewise.
5141 (emit_fusion_p9_store): Likewise.
5142 * config/rx/rx.c (rx_gen_move_template): Likewise.
5143 (flags_from_mode): Likewise.
5144 * config/s390/predicates.md (s390_alc_comparison): Likewise.
5145 (s390_slb_comparison): Likewise.
5146 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
5147 (s390_vector_mode_supported_p): Likewise.
5148 (s390_cc_modes_compatible): Likewise.
5149 (s390_match_ccmode_set): Likewise.
5150 (s390_canonicalize_comparison): Likewise.
5151 (s390_emit_compare_and_swap): Likewise.
5152 (s390_branch_condition_mask): Likewise.
5153 (s390_rtx_costs): Likewise.
5154 (s390_secondary_reload): Likewise.
5155 (__SECONDARY_RELOAD_CASE): Likewise.
5156 (s390_expand_cs): Likewise.
5157 (s390_preferred_simd_mode): Likewise.
5158 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
5159 * config/sh/sh.c (sh_print_operand): Likewise.
5160 (dump_table): Likewise.
5161 (sh_secondary_reload): Likewise.
5162 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
5163 * config/sh/sh.md (casesi_worker_1): Likewise.
5164 (casesi_worker_2): Likewise.
5165 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
5166 (fcc_comparison_operator): Likewise.
5167 * config/sparc/sparc.c (sparc_expand_move): Likewise.
5168 (emit_soft_tfmode_cvt): Likewise.
5169 (sparc_preferred_simd_mode): Likewise.
5170 (output_cbranch): Likewise.
5171 (sparc_print_operand): Likewise.
5172 (sparc_expand_vec_perm_bmask): Likewise.
5173 (vector_init_bshuffle): Likewise.
5174 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
5175 (spu_vector_mode_supported_p): Likewise.
5176 (spu_expand_insv): Likewise.
5177 (spu_emit_branch_or_set): Likewise.
5178 (spu_handle_vector_attribute): Likewise.
5179 (spu_builtin_splats): Likewise.
5180 (spu_builtin_extract): Likewise.
5181 (spu_builtin_promote): Likewise.
5182 (spu_expand_sign_extend): Likewise.
5183 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
5184 (tilegx_simd_int): Likewise.
5185 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
5186 (tilepro_simd_int): Likewise.
5187 * config/v850/v850.c (const_double_split): Likewise.
5188 (v850_print_operand): Likewise.
5189 (ep_memory_offset): Likewise.
5190 * config/vax/vax.c (vax_rtx_costs): Likewise.
5191 (vax_output_int_move): Likewise.
5192 (vax_output_int_add): Likewise.
5193 (vax_output_int_subtract): Likewise.
5194 * config/visium/predicates.md (visium_branch_operator): Likewise.
5195 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
5196 (visium_print_operand_address): Likewise.
5197 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
5198 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
5199 (xtensa_expand_conditional_branch): Likewise.
5200 (xtensa_copy_incoming_a7): Likewise.
5201 (xtensa_output_literal): Likewise.
5202 * dfp.c (decimal_real_maxval): Likewise.
5203 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
5205 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5206 Alan Hayward <alan.hayward@arm.com>
5207 David Sherwood <david.sherwood@arm.com>
5209 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
5210 (mode_nunits_inline): Likewise.
5211 (mode_inner_inline): Likewise.
5212 (mode_unit_size_inline): Likewise.
5213 (mode_unit_precision_inline): Likewise.
5214 (emit_insn_modes_h): Likewise. Also emit a #define of the
5216 (emit_mode_wider): Add an E_ prefix to mode names.
5217 (emit_mode_complex): Likewise.
5218 (emit_mode_inner): Likewise.
5219 (emit_mode_adjustments): Likewise.
5220 (emit_mode_int_n): Likewise.
5221 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
5222 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
5223 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
5224 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
5225 (CRC32_BUILTIN, ENTRY): Likewise.
5226 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
5227 (aarch64_pop_regs): Likewise.
5228 (aarch64_process_components): Likewise.
5229 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
5230 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
5231 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
5232 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
5233 * config/arm/arm.c (arm_init_libfuncs): Likewise.
5234 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
5236 * config/i386/i386-builtin.def (pcmpestr): Likewise.
5237 (pcmpistr): Likewise.
5238 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
5239 * config/mmix/mmix.c (mmix_output_condition): Likewise.
5240 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
5242 * config/rl78/rl78.c (mduc_regs): Likewise.
5243 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
5244 (htm_expand_builtin): Likewise.
5245 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
5246 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
5247 * config/xtensa/xtensa.c (print_operand): Likewise.
5248 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
5249 (NUM_MODE_VECTOR_INT): Likewise.
5250 * genoutput.c (null_operand): Likewise.
5251 (output_operand_data): Likewise.
5252 * genrecog.c (print_parameter_value): Likewise.
5253 * lra.c (debug_operand_data): Likewise.
5255 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5257 * dwarf2out.c (scompare_loc_descriptor_wide)
5258 (scompare_loc_descriptor_narrow): New functions, split out from...
5259 (scompare_loc_descriptor): ...here.
5260 * expmed.c (emit_store_flag_int): New function, split out from...
5261 (emit_store_flag): ...here.
5263 2017-08-30 Richard Biener <rguenther@suse.de>
5265 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
5266 (dwarf2out_early_finish): Move setting of AT_pubnames from
5267 early debug output to early finish.
5269 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
5271 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
5272 and -mdata-region to the assembler.
5274 2017-08-30 Richard Biener <rguenther@suse.de>
5276 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
5278 (gen_subprogram_die): Add DW_AT_object_pointer only early.
5279 (dwarf2out_early_global_decl): Only generate a DIE for the
5280 abstract origin if it doesn't already exist or is a declaration DIE.
5281 (resolve_addr): Do not add the linkage name twice when
5282 generating a stub DIE for the DW_TAG_GNU_call_site target.
5284 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5286 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5287 Use machine_mode rather than int for arg1_mode.
5289 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
5292 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
5293 that the second argument of the built-in functions to unpack
5294 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
5295 switch statement instead a lot of if statements.
5296 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
5297 Allow 64-bit values to be in Altivec registers as well as
5298 traditional floating point registers.
5299 (pack<mode>, FMOVE128_VSX iterator): Likewise.
5301 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
5303 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
5304 MAX_REGS_PER_ADDRESS == 1.
5306 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
5308 * config/i386/i386.opt (flag_fentry): Do not init to -1.
5309 * config/i386/i386.c (ix86_option_override_internal): Simplify
5310 setting of opts->x_flag_entry.
5312 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5313 Jakub Jelinek <jakub@redhat.com>
5314 Richard Biener <rguenther@suse.de>
5316 PR tree-optimization/81503
5317 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
5318 folded constant fits in the target type; reorder tests for clarity.
5320 2017-08-29 Martin Liska <mliska@suse.cz>
5322 * passes.def: Include pass_lower_switch.
5323 * stmt.c (dump_case_nodes): Remove and move to
5324 tree-switch-conversion.
5325 (case_values_threshold): Likewise.
5326 (expand_switch_as_decision_tree_p): Likewise.
5327 (emit_case_decision_tree): Likewise.
5328 (expand_case): Likewise.
5329 (balance_case_nodes): Likewise.
5330 (node_has_low_bound): Likewise.
5331 (node_has_high_bound): Likewise.
5332 (node_is_bounded): Likewise.
5333 (emit_case_nodes): Likewise.
5334 (struct simple_case_node): New struct.
5335 (add_case_node): Remove.
5336 (emit_case_dispatch_table): Use vector instead of case_list.
5337 (reset_out_edges_aux): Remove.
5338 (compute_cases_per_edge): Likewise.
5339 (expand_case): Build list of simple_case_node.
5340 (expand_sjlj_dispatch_table): Use it.
5341 * tree-switch-conversion.c (struct case_node): Moved from
5342 stmt.c and adjusted.
5343 (emit_case_nodes): Likewise.
5344 (node_has_low_bound): Likewise.
5345 (node_has_high_bound): Likewise.
5346 (node_is_bounded): Likewise.
5347 (case_values_threshold): Likewise.
5348 (reset_out_edges_aux): Likewise.
5349 (compute_cases_per_edge): Likewise.
5350 (add_case_node): Likewise.
5351 (dump_case_nodes): Likewise.
5352 (balance_case_nodes): Likewise.
5353 (expand_switch_as_decision_tree_p): Likewise.
5354 (emit_jump): Likewise.
5355 (emit_case_decision_tree): Likewise.
5356 (try_switch_expansion): Likewise.
5357 (do_jump_if_equal): Likewise.
5358 (emit_cmp_and_jump_insns): Likewise.
5359 (fix_phi_operands_for_edge): New function.
5360 (record_phi_operand_mapping): Likewise.
5361 (class pass_lower_switch): New pass.
5362 (pass_lower_switch::execute): New function.
5363 (make_pass_lower_switch): Likewise.
5364 (conditional_probability):
5365 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
5366 * tree-pass.h: Add make_pass_lower_switch.
5368 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
5371 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
5374 2017-08-29 Richard Biener <rguenther@suse.de>
5376 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
5377 we do not add a DW_AT_inline attribute twice.
5378 (gen_subprogram_die): Remove code setting DW_AT_inline on
5379 DECL_ABSTRACT_P nodes.
5381 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
5383 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
5384 calls to internal functions.
5385 (gimplify_modify_expr): Likewise.
5386 * tree-call-cdce.c (use_internal_fn): Likewise.
5387 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
5388 (convert_to_divmod): Set the nothrow flag.
5389 * tree-if-conv.c (predicate_mem_writes): Likewise.
5390 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
5391 (vectorizable_call): Likewise.
5392 (vectorizable_store): Likewise.
5393 (vectorizable_load): Likewise.
5394 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
5395 (vect_recog_mask_conversion_pattern): Likewise.
5397 2017-08-29 Martin Liska <mliska@suse.cz>
5400 * gcc.c (driver_handle_option): Add new argument.
5401 * opts-common.c (handle_option): Pass
5402 target_option_override_hook.
5403 * opts-global.c (lang_handle_option): Add new option.
5404 (set_default_handlers): Add new argument.
5405 (decode_options): Likewise.
5406 * opts.c (target_handle_option): Likewise.
5407 (common_handle_option): Call target_option_override_hook.
5408 * opts.h (struct cl_option_handler_func): Add hook for
5409 target option override.
5410 (struct cl_option_handlers): Likewise.
5411 (set_default_handlers): Add new argument.
5412 (decode_options): Likewise.
5413 (common_handle_option): Likewise.
5414 (target_handle_option): Likewise.
5415 * toplev.c (toplev::main): Pass targetm.target_option.override
5418 2017-08-29 Richard Biener <rguenther@suse.de>
5419 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
5421 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
5422 life to the active subtree.
5424 2017-08-28 Jeff Law <law@redhat.com>
5426 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
5427 derive_equivalences.
5428 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
5430 (edge_info::derive_equivalences): New private member function
5432 * tree-ssa-dom.c (class edge_info): Changed from a struct
5433 to a class. Add ctor/dtor, methods and data members.
5434 (edge_info::edge_info): Renamed from allocate_edge_info.
5435 Initialize additional members.
5436 (edge_info::~edge_info): New.
5437 (free_dom_edge_info): Delete the edge info.
5438 (record_edge_info): Use new class & associated member functions.
5439 Tighten forms for testing for edge equivalences.
5440 (record_temporary_equivalences): Iterate over the simple
5441 equivalences rather than assuming there's only one per edge.
5442 (cprop_into_successor_phis): Iterate over the simple
5443 equivalences rather than assuming there's only one per edge.
5444 (optimize_stmt): Use operand_equal_p rather than pointer
5445 equality for mini-DSE code.
5447 2017-08-28 Nathan Sidwell <nathan@acm.org>
5449 * gcc.c (execute): Fold SIGPIPE handling into switch
5450 statement. Adjust internal error message.
5452 2017-08-28 Richard Biener <rguenther@suse.de>
5455 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
5456 Do nothing for removed DIEs.
5458 2017-08-28 Richard Biener <rguenther@suse.de>
5460 PR tree-optimization/81977
5461 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
5464 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
5467 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
5468 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
5469 using targetm.gen_mem_thread_fence.
5471 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
5474 * config/i386/i386.md (*<btsc><mode>): Change operand 2
5475 predicate to register_operand. Reorder operands.
5476 (*btr<mode>): Ditto.
5477 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
5478 (*btr<mode>_mask): Ditto.
5480 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
5482 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
5483 * config/rs6000/xmmintrin.h: New file.
5484 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
5486 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5489 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
5490 parameter and_insn and return it.
5491 (recombine_lvx_pattern): Insert a copy to ensure availability of
5492 the base register of the copied masking operation at the point of
5493 the instruction replacement.
5494 (recombine_stvx_pattern): Likewise.
5496 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
5498 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
5499 undocumented switches.
5500 (-mpower9-dform-vector): Likewise.
5501 (-mpower9-dform): Likewise.
5502 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
5503 comments to delete references to -mpower9-dform* switches.
5504 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
5505 Delete reference to -mpower9-dform* switches, test for
5506 -mpower9-vector instead.
5507 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
5508 (OTHER_P9_VECTOR_MASKS): Likewise.
5509 (POWERPC_MASKS): Likewise.
5510 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
5511 tests against -mpower9-dform* to -mpower9-vector. Delete code
5512 that checked for -mpower9-dform* consistancy with other options.
5513 Add test for -mpower9-misc to enable other power9 switches.
5514 (rs6000_init_hard_regno_mode_ok): Likewise.
5515 (rs6000_option_override_internal): Likewise.
5516 (rs6000_emit_prologue): Likewise.
5517 (rs6000_emit_epilogue): Likewise.
5518 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
5519 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
5520 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
5522 (emit_fusion_p9_store): Likewise.
5523 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
5524 resetting these macros if the assembler does not support ISA 3.0
5526 (TARGET_P9_DFORM_VECTOR): Likewise.
5527 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
5528 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
5530 2017-08-25 Alan Modra <amodra@gmail.com>
5532 PR rtl-optimization/81747
5533 * cse.c (cse_extended_basic_block): Don't attempt to record
5534 equivalences for degenerate conditional branches that branch
5535 to their fall-through.
5537 2017-08-24 Martin Sebor <msebor@redhat.com>
5540 * gimple-fold.c (size_must_be_zero_p): New function.
5541 (gimple_fold_builtin_memory_op): Call it.
5543 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
5545 * config/rs6000/mm_malloc.h: New file.
5547 2017-08-24 Bin Cheng <bin.cheng@arm.com>
5549 PR tree-optimization/81913
5550 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
5551 analysis when either IVs in condition can wrap.
5553 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
5555 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
5556 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
5558 2017-08-24 Richard Biener <rguenther@suse.de>
5561 * targhooks.c (default_target_can_inline_p): Properly
5562 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
5563 is present and always compare.
5564 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
5565 infer -mfpmath=sse from TARGET_SSE_P.
5566 (ix86_can_inline_p): Properly use target_option_default_node when
5567 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
5569 2017-08-24 Richard Biener <rguenther@suse.de>
5572 * dwarf2out.c (output_die): Handle flag_generate_offload like
5574 (output_comp_unit): Likewise.
5575 (gen_array_type_die): Likewise.
5576 (dwarf2out_early_finish): Likewise.
5577 (note_variable_value_in_expr): Likewise.
5578 (dwarf2out_finish): Likewise. Adjust assert.
5579 * cgraphunit.c (symbol_table::compile): Move setting of
5580 flag_generate_offload earlier ...
5581 (symbol_table::finalize_compilation_unit): ... here, before
5582 early debug finalization.
5584 2017-08-24 Richard Biener <rguenther@suse.de>
5586 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
5587 and ipa-fnsummary.h.
5588 (ix86_can_inline_p): When ix86_fpmath flags do not match
5589 check whether the callee uses FP math at all.
5591 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
5594 * tree-ssanames.c (get_nonzero_bits): Use element_precision
5595 instead of TYPE_PRECISION.
5597 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
5598 Alan Hayward <alan.hayward@arm.com>
5599 David Sherwood <david.sherwood@arm.com>
5601 * combine.c (make_extraction): Use subreg_offset_from_lsb.
5603 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
5605 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
5607 (ix86_frame::stack_realign_allocate): New field.
5608 (struct machine_frame_state): Modify comments.
5609 (machine_frame_state::sp_realigned_fp_end): New field.
5610 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
5612 (sp_valid_at): Add assertion to assure no attempt to access invalid
5613 offset of a realigned stack.
5614 (fp_valid_at): Likewise.
5615 (choose_baseaddr): Modify comments.
5616 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
5617 ix86_expand_prologue.
5618 (ix86_expand_prologue): Modify stack realignment and allocation.
5619 (ix86_expand_epilogue): Modify comments.
5620 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
5621 avx2_runtime, avx512f, and avx512f_runtime.
5623 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
5625 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
5626 (mstackrealign): Do not init to -1.
5627 * config/i386/i386.c (ix86_option_override_internal):
5628 Check opts_set, not opts when setting default value of
5629 opts->x_ix86_force_align_arg_pointer.
5631 2017-08-23 Richard Biener <rguenther@suse.de>
5633 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
5634 lang_hooks.decl_printable_name.
5635 * print-rtl-function.c (print_rtx_function): Likewise.
5636 * tree-pretty-print.c (dump_function_header): Likewise.
5638 2017-08-23 Richard Biener <rguenther@suse.de>
5641 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
5642 -g0 at compile-time.
5644 2017-08-23 Tamar Christina <tamar.christina@arm.com>
5647 * doc/sourcebuild.texi (Other hardware attributes):
5650 2017-08-23 Tamar Christina <tamar.christina@arm.com>
5653 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
5654 Fix single-use check.
5656 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
5658 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
5660 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
5662 * doc/install.texi: Modify to add more details on running selected
5665 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
5667 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
5668 is combined with -mabi=ms.
5669 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
5672 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5674 PR tree-optimization/81488
5675 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
5676 and cached_basis fields.
5677 (MAX_SPREAD): New constant.
5678 (alloc_cand_and_find_basis): Initialize new fields.
5679 (clear_visited): New function.
5680 (create_phi_basis_1): Rename from create_phi_basis, set visited
5681 and cached_basis fields.
5682 (create_phi_basis): New wrapper function.
5683 (phi_add_costs_1): Rename from phi_add_costs, add spread
5684 parameter, set visited field, short-circuit when limits reached.
5685 (phi_add_costs): New wrapper function.
5686 (record_phi_increments_1): Rename from record_phi_increments, set
5688 (record_phi_increments): New wrapper function.
5689 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
5690 (phi_incr_cost): New wrapper function.
5691 (all_phi_incrs_profitable_1): Rename from
5692 all_phi_incrs_profitable, set visited field.
5693 (all_phi_incrs_profitable): New wrapper function.
5695 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
5696 Alan Hayward <alan.hayward@arm.com>
5697 David Sherwood <david.sherwood@arm.com>
5699 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
5700 that takes the outer and inner modes.
5701 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
5702 comparison as the canonical test for a paradoxical subreg.
5703 * combine.c (simplify_set): Use paradoxical_subreg_p.
5704 (make_extraction): Likewise.
5705 (force_to_mode): Likewise.
5706 (rtx_equal_for_field_assignment_p): Likewise.
5707 (gen_lowpart_for_combine): Likewise.
5708 (simplify_comparison): Likewise.
5709 * cse.c (equiv_constant): Likewise.
5710 * expmed.c (store_bit_field_1): Likewise.
5711 * final.c (alter_subreg): Likewise.
5712 * fwprop.c (propagate_rtx): Likewise.
5713 (forward_propagate_subreg): Likewise.
5714 * ira-conflicts.c (ira_build_conflicts): Likewise.
5715 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
5716 * lra-constraints.c (curr_insn_transform): Likewise.
5717 (split_reg): Likewise.
5718 * lra-eliminations.c (move_plus_up): Likewise.
5719 (lra_eliminate_regs_1): Likewise.
5720 * recog.c (general_operand): Likewise.
5721 * ree.c (combine_reaching_defs): Likewise.
5722 * reload.c (push_reload): Likewise.
5723 (find_reloads): Likewise.
5724 * reload1.c (elimination_effects): Likewise.
5725 (compute_reload_subreg_offset): Likewise.
5726 (choose_reload_regs): Likewise.
5727 * rtlanal.c (subreg_lsb_1): Likewise.
5728 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5729 (simplify_subreg): Likewise.
5730 * var-tracking.c (track_loc_p): Likewise.
5731 * emit-rtl.c (byte_lowpart_offset): Likewise.
5732 (paradoxical_subreg_p): Delete out-of-line definition.
5734 2017-08-22 Jeff Law <law@redhat.com>
5736 PR tree-optimization/81741
5737 PR tree-optimization/71947
5738 * tree-ssa-dom.c: Include tree-inline.h.
5739 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
5740 equivalences if one is more expensive to compute than the other.
5741 * tree-ssa-scopedtables.h (class const_or_copies): Make
5742 record_const_or_copy_raw method private.
5743 (class avail_exprs_stack): New method simplify_binary_operation.
5744 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
5745 avail_exprs_stack::simplify_binary_operation as needed.
5746 (avail_exprs_stack::simplify_binary_operation): New function.
5748 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
5750 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
5751 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
5752 (DOT_SYMBOLS): Likewise.
5753 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
5754 (RELOCATABLE_NEEDS_FIXUP): Likewise.
5755 (RS6000_ABI_NAME): Likewise.
5756 (TARGET_CMODEL): Likewise.
5757 (TOC_SECTION_ASM_OP): Likewise.
5758 (SET_CMODEL): New macro.
5759 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
5761 2017-08-22 Richard Biener <rguenther@suse.de>
5763 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
5764 to allow for free-lang-data replacements similar to verify_type_variant.
5766 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
5767 Alan Hayward <alan.hayward@arm.com>
5768 David Sherwood <david.sherwood@arm.com>
5770 * config/aarch64/aarch64.md (casesi): Use DImode rather than
5771 VOIDmode for the LABEL_REF.
5773 2017-08-22 Richard Biener <rguenther@suse.de>
5775 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
5777 2017-08-22 Richard Biener <rguenther@suse.de>
5779 * common.opt (feliminate-dwarf2-dups): Ignore.
5780 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
5781 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
5782 same_die_p_wrap, compute_section_prefix,
5783 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
5784 (comdat_symbol_id, comdat_symbol_number): Likewise.
5785 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
5787 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
5788 (output_die): Mark unreachable path unreachable.
5789 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
5790 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
5791 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
5792 (dwarf2out_early_finish): Likewise.
5794 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
5796 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
5798 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
5801 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
5802 not VAR_P. Filter attribute warnings with OPT_Wattributes.
5803 (avr_attribute_table) <io, io_low, address>: Initialize
5804 .decl_required with true.
5806 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
5808 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
5809 undocumented debugging options.
5810 (-mvsx-scalar-double): Likewise.
5811 (-mallow-df-permute): Likewise.
5812 (-mvectorize-builtins): Likewise.
5813 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
5814 (rs6000_builtin_vectorized_function): Likewise.
5815 (rs6000_builtin_md_vectorized_function): Likewise.
5816 (rs6000_opt_vars): Likewise.
5818 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
5821 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
5822 (*btrq_imm): Rename from *btrq.
5823 (*btcq_imm): Rename from *btcq.
5824 (btsc): New code attribute.
5825 (*<btsc><mode>): New insn pattern.
5826 (*btr<mode>): Ditto.
5827 (*<btsc><mode>_mask): New insn_and_split pattern.
5828 (*btr<mode>_mask): Ditto.
5830 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5831 Alan Hayward <alan.hayward@arm.com>
5832 David Sherwood <david.sherwood@arm.com>
5834 * function.c (pad_below): Simplify padding calculation.
5836 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5837 Alan Hayward <alan.hayward@arm.com>
5838 David Sherwood <david.sherwood@arm.com>
5840 * target.def (function_prologue): Remove frame size argument.
5841 (function_epilogue): Likewise.
5842 * doc/tm.texi: Regenerate.
5843 * final.c (final_start_function): Update call to function_prologue.
5844 (final_end_function): Update call to function_epilogue.
5845 (default_function_pro_epilogue): Remove frame size argument.
5846 * output.h (default_function_pro_epilogue): Likewise.
5847 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
5848 (arm_output_function_prologue): Likewise.
5849 * config/frv/frv.c (frv_function_prologue): Likewise.
5850 (frv_function_epilogue): Likewise.
5851 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
5852 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
5853 (ia64_output_function_epilogue): Likewise.
5854 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
5855 (m32r_output_function_epilogue): Likewise.
5856 * config/microblaze/microblaze.c (microblaze_function_prologue)
5857 (microblaze_function_epilogue): Likewise.
5858 * config/mips/mips.c (mips_output_function_prologue): Likewise.
5859 (mips_output_function_epilogue): Likewise.
5860 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
5861 (mmix_target_asm_function_epilogue): Likewise.
5862 * config/msp430/msp430.c (msp430_start_function): Likewise.
5863 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
5864 (nds32_asm_function_epilogue): Likewise.
5865 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
5866 * config/pa/pa.c (pa_output_function_prologue): Likewise.
5867 (pa_output_function_epilogue): Likewise.
5868 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
5869 (rs6000_output_function_epilogue): Likewise.
5870 * config/rl78/rl78.c (rl78_start_function): Likewise.
5871 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
5872 (rs6000_output_function_epilogue): Likewise.
5873 * config/rx/rx.c (rx_output_function_prologue): Likewise.
5874 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
5875 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
5876 (sparc_asm_function_epilogue): Likewise.
5878 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5880 * tree.h (type_has_mode_precision_p): New function.
5881 * convert.c (convert_to_integer_1): Use it.
5882 * expr.c (expand_expr_real_2): Likewise.
5883 (expand_expr_real_1): Likewise.
5884 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
5885 * match.pd: Likewise.
5886 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
5887 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
5888 * tree-tailcall.c (process_assignment): Likewise.
5889 * tree-vect-loop.c (vectorizable_reduction): Likewise.
5890 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
5891 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
5892 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
5893 (vectorizable_assignment): Likewise.
5894 (vectorizable_shift): Likewise.
5895 (vectorizable_operation): Likewise.
5896 * tree-vrp.c (register_edge_assert_for_2): Likewise.
5898 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
5900 * match.pd: Add pow (C, x) simplification.
5902 2017-08-21 Richard Biener <rguenther@suse.de>
5904 PR tree-optimization/81900
5905 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
5906 for blocks with abnormal predecessors.
5907 (compute_antic): Do not set visited flag prematurely.
5909 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
5912 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
5914 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5916 * stor-layout.h (vector_type_mode): Move to...
5917 * tree.h (vector_type_mode): ...here.
5918 * stor-layout.c (vector_type_mode): Move to...
5919 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
5921 2017-08-21 Richard Biener <rguenther@suse.de>
5923 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
5924 register_external_die hooks.
5925 (debug_false_tree_charstarstar_uhwistar): Declare.
5926 (debug_nothing_tree_charstar_uhwi): Likewise.
5927 * debug.c (do_nothing_debug_hooks): Adjust.
5928 (debug_false_tree_charstarstar_uhwistar): New do nothing.
5929 (debug_nothing_tree_charstar_uhwi): Likewise.
5930 * dbxout.c (dbx_debug_hooks): Adjust.
5931 (xcoff_debug_hooks): Likewise.
5932 * sdbout.c (sdb_debug_hooks): Likewise.
5933 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
5934 * dwarf2out.c (macinfo_label_base): New global.
5935 (dwarf2out_register_external_die): New function for the
5936 register_external_die hook.
5937 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
5938 (dwarf2_debug_hooks): Use them.
5939 (dwarf2_lineno_debug_hooks): Adjust.
5940 (struct die_struct): Add with_offset flag.
5941 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
5942 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
5943 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
5944 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
5945 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
5946 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
5947 defining section names for the early LTO debug variants.
5948 (reset_indirect_string): New helper.
5949 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
5950 (print_dw_val): Add support for offsetted symbol references.
5951 (get_ultimate_context): Split out from is_cxx.
5952 (is_cxx): Use get_ultimate_context.
5953 (is_fortran): Add decl overload.
5954 (compute_comp_unit_symbol): Split out worker from
5955 compute_section_prefix.
5956 (compute_section_prefix): Call compute_comp_unit_symbol and
5957 set comdat_type_p here.
5958 (output_die): Skip DIE symbol output for the LTO added one.
5959 Handle DIE symbol references with offset.
5960 (output_comp_unit): Guard section name mangling properly.
5961 For LTO debug sections emit a symbol at the section beginning
5962 which we use to refer to its DIEs.
5963 (add_abstract_origin_attribute): For DIEs registered via
5964 dwarf2out_register_external_die directly refer to the early
5965 DIE rather than indirectly through the shadow one we created.
5966 Remove obsolete call to dwarf2out_abstract_function for
5967 non-function/block origins.
5968 (gen_array_type_die): When generating early LTO debug do
5969 not emit DW_AT_string_length.
5970 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
5971 late when in LTO. As suggested place a gcc_unreachable for
5972 the DECL_ABSTRACT_P case.
5973 (gen_subprogram_die): Avoid another specification DIE
5974 for early built declarations/definitions for the late LTO case.
5975 (gen_variable_die): Add type references for late duplicated VLA dies
5977 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
5978 we have the abstract instance already.
5979 (process_scope_var): Adjust decl DIE contexts in LTO which
5980 first puts them in limbo.
5981 (gen_decl_die): Do not generate type DIEs late apart from
5982 types for VLAs or for decls we do not yet have a DIE. Do not
5983 call dwarf2out_abstract_function late.
5984 (dwarf2out_early_global_decl): Make sure to create DIEs
5985 for abstract instances of a decl first.
5986 (dwarf2out_late_global_decl): Adjust comment.
5987 (output_macinfo_op): With multiple macro sections use
5988 macinfo_label_base to distinguish labels.
5989 (output_macinfo): Likewise. Update macinfo_label_base.
5990 Pass in the line info label.
5991 (note_variable_value_in_expr): When generating LTO resolve
5992 all variable values here by generating DIEs as needed.
5993 (init_sections_and_labels): Add early LTO debug flag parameter
5994 and generate different sections and names if set. Add generation
5995 counter for the labels so we can have multiple of them.
5996 (reset_dies): Helper to allow DIEs to be output multiple times.
5997 (dwarf2out_finish): When outputting DIEs to the fat part of an
5998 LTO object first reset DIEs.
5999 (dwarf2out_early_finish): Output early DIEs when generating LTO.
6000 (modified_type_die): Check for decl_ultimate_origin being self
6002 (gen_type_die_with_usage): Likewise.
6003 (gen_typedef_die): Allow decl_ultimate_origin being self.
6004 (set_decl_abstract_flags): Remove.
6005 (set_block_abstract_flags): Likewise.
6006 (dwarf2out_abstract_function): Treat the early generated DIEs
6007 as the abstract copy and only add DW_AT_inline and
6008 DW_AT_artificial here and call set_decl_origin_self.
6009 If the DIE has an abstract origin don't do anything.
6010 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
6011 if we have none yet (Go fails to build one, PR78628).
6012 (variably_modified_type_p): Prevent endless recursion for Ada
6013 cyclic pointer types.
6014 * lto-streamer-in.c: Include debug.h.
6015 (dref_queue): New global.
6016 (lto_read_tree_1): Stream in DIE references.
6017 (lto_input_tree): Register DIE references.
6018 (input_function): Stream DECL_DEBUG_ARGS.
6019 * lto-streamer-out.c: Include debug.h.
6020 (lto_write_tree_1): Output DIE references.
6021 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
6022 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
6023 (output_function): Stream DECL_DEBUG_ARGS.
6024 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
6025 Stream DECL_ABSTRACT_ORIGIN.
6026 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
6027 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
6028 DECL_CONTEXT for file-scope decls.
6029 * lto-streamer.h (struct dref_entry): Declare.
6030 (dref_queue): Likewise.
6031 * cfgexpand.c (pass_expand::execute): Do not call the
6032 outlining_inline_function hook here.
6033 * lto-wrapper.c (debug_obj): New global.
6034 (tool_cleanup): Unlink it if required.
6035 (debug_objcopy): New function.
6036 (run_gcc): Handle early debug sections in the IL files by
6037 extracting them to separate files, partially linkin them and
6038 feeding the result back as result to the linker.
6039 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
6040 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
6041 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
6042 sections into a separate segment.
6043 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
6045 (darwin_asm_dwarf_section): Likewise.
6046 (darwin_asm_output_dwarf_offset): Likewise.
6047 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
6049 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
6050 Alan Hayward <alan.hayward@arm.com>
6051 David Sherwood <david.sherwood@arm.com>
6053 * read-md.h (md_reader::record_potential_iterator_use): Replace
6054 pointer argument with an rtx and an index.
6055 * read-rtl.c (iterator_group::apply_iterator): Likewise.
6056 (apply_mode_iterator): Likewise.
6057 (apply_code_iterator): Likewise.
6058 (apply_int_iterator): Likewise.
6059 (apply_subst_iterator): Likewise.
6060 (record_iterator_use): Likewise.
6061 (record_attribute_use): Likewise.
6062 (md_reader::record_potential_iterator_use): Likewise. Update calls
6063 to record_iterator_use and apply_iterator.
6064 (iterator_use): Replace ptr with x and index.
6065 (attribute_use): Likewise.
6066 (apply_attribute_uses): Update calls to apply_iterator.
6067 (apply_iterators): Likewise. Update initialization of iterator_use.
6068 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
6069 and record_potential_iterator_use.
6070 (rtx_reader::read_rtx_operand): Likewise.
6072 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
6073 Alan Hayward <alan.hayward@arm.com>
6074 David Sherwood <david.sherwood@arm.com>
6076 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
6079 2017-08-21 Richard Biener <rguenther@suse.de>
6082 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
6083 at struct end conservatively when comparing common bases.
6085 2017-08-21 Richard Biener <rguenther@suse.de>
6087 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
6088 (mem_ref_in_stmt): Remove.
6089 (determine_max_movement): Use ref index to get at the reference.
6090 (invariantness_dom_walker::before_dom_children): Deal with
6091 lim data already initialized.
6092 (gather_mem_refs_stmt): Initialize lim data and record ref index.
6094 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
6096 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
6097 (TARGET_ISA_ROUND): Ditto.
6098 (TARGET_ROUND): Ditto.
6099 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
6100 * config/i386/i386.md: Ditto.
6101 * config/i386/sse.md: Ditto.
6102 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
6103 with OPTION_MASK_ISA_SSE4_1.
6105 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
6108 * doc/extend.texi (x86 Built-in Functions): Correct the name of
6109 __builtin_ia32_lzcnt_u16.
6111 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
6114 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
6115 (rs6000_set_current_function): Rewrite function to use it.
6117 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
6120 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
6121 and TYPE_WARN_IF_NOT_ALIGN.
6122 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
6123 (handle_warn_if_not_align): New.
6124 (place_union_field): Call handle_warn_if_not_align.
6125 (place_field): Call handle_warn_if_not_align.
6126 Copy TYPE_WARN_IF_NOT_ALIGN.
6127 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
6128 (layout_type): Likewise.
6129 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
6131 (tree_decl_common): Add warn_if_not_align.
6132 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
6133 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
6134 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
6135 (DECL_WARN_IF_NOT_ALIGN): Likewise.
6136 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
6137 * doc/extend.texi: Document warn_if_not_aligned attribute.
6138 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
6140 2017-08-17 Martin Liska <mliska@suse.cz>
6143 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
6144 (get_data_dependence): Use it as pointer type.
6145 (distribute_loop): Likewise.
6147 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6149 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
6150 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
6151 (p8_vmrgow_<mode>_direct): New define_insn.
6152 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
6153 handle endianness for vmrgew and vmrgow permute patterns.
6155 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
6157 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
6158 * config/rs6000/rs6000-cpus.def: Remove comment.
6159 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
6160 (POWERPC_MASKS): Likewise.
6161 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
6162 use of TARGET_VSX_TIMODE.
6163 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
6164 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
6165 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
6166 (rs6000_option_override_internal): Remove dead code.
6167 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
6168 (rs6000_legitimize_reload_address): Likewise.
6169 (rs6000_legitimate_address_p): Likewise.
6170 (rs6000_opt_masks): Delete "vsx-timode".
6171 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
6172 from function comment.
6173 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
6174 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
6175 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
6177 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
6178 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
6179 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
6182 (define_peephole2): Likewise.
6184 2017-08-17 Martin Sebor <msebor@redhat.com>
6187 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
6188 past the end of an array.
6189 (test_pp_format): Add test cases.
6191 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
6193 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
6194 missing ECF_NOTHROW flags.
6196 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
6199 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
6200 operands residing in integer registers.
6201 (*vsx_le_perm_load_<mode>): Likewise.
6202 (*vsx_le_perm_store_<mode>): Likewise.
6203 (define_peephole2): Add peepholes to optimize the above.
6205 2017-08-17 Marek Polacek <polacek@redhat.com>
6208 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
6209 to mimic what shorten_compare did. Change the return type to bool.
6210 (fold_cond_expr_with_comparison): Update call to
6211 operand_equal_for_comparison_p.
6212 (fold_ternary_loc): Likewise.
6214 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
6216 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
6218 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
6220 2017-08-17 Richard Biener <rguenther@suse.de>
6222 * tree-ssa-structalias.c (solve_graph): When propagating
6223 to successors update the graphs succ edges and avoid duplicate work.
6225 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
6228 * config/i386/i386.c (ix86_option_override_internal): Save target
6229 specific options after ix86_stack_protector_guard_reg was changed.
6231 2017-08-17 Richard Biener <rguenther@suse.de>
6233 PR tree-optimization/81827
6234 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
6235 (new_var_info): Initialize it conservatively.
6236 (get_call_vi): Mark register vars.
6237 (new_scalar_tmp_constraint_exp): Likewise.
6238 (handle_rhs_call): Likewise.
6239 (handle_const_call): Likewise.
6240 (create_function_info_for): Likewise.
6241 (solve_constraints): Sort varinfos to separate register from
6242 non-register vars to pack points-to solution bitmaps during
6245 2017-08-17 Marek Polacek <polacek@redhat.com>
6247 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
6249 2017-08-17 Richard Biener <rguenther@suse.de>
6251 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
6252 to true when overflow is undefined and we saturated the result.
6254 2017-08-17 Alan Modra <amodra@gmail.com>
6257 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
6258 Don't use store multiple if only one reg needs saving.
6259 (interesting_frame_related_regno): New function.
6260 (rs6000_frame_related): Don't emit frame info for regs that
6262 (rs6000_emit_epilogue): Likewise.
6264 2017-08-16 Nathan Sidwell <nathan@acm.org>
6266 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
6267 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
6268 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
6269 * tree.c (free_lang_data_in_type): Use else-if chain. Always
6270 clear TYPE_LANG_1. Remove obsolete member-function stripping.
6271 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
6272 (verify_type): Adjust for TYPE_BINFO move.
6273 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
6274 process TYPE_BINFO directly.
6275 (hash_tree): Likewise.
6276 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
6278 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
6281 2017-08-16 David Malcolm <dmalcolm@redhat.com>
6283 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
6285 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
6288 * config/i386/i386.md (*anddi_1_btr): Change predicates of
6289 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
6290 Add ix86_binary_operator_ok to insn constraint.
6291 (*iordi_1_bts): Ditto.
6292 (*xordi_1_btc): Ditto.
6293 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
6294 Update corresponding peephole2 pattern.
6298 2017-08-16 Bin Cheng <bin.cheng@arm.com>
6300 PR tree-optimization/81832
6301 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
6302 copy loop header which has IFN_LOOP_DIST_ALIAS call.
6304 2017-08-16 Marek Polacek <polacek@redhat.com>
6307 * fold-const.c (fold_indirect_ref_1): Restore original behavior
6308 regarding size_zero_node.
6310 2017-08-16 Martin Liska <mliska@suse.cz>
6313 * config.gcc: Respect previously set extra_objs in case
6316 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
6318 PR tree-optimization/81835
6319 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
6320 the phi SSA_NAME. Check that the condition in a COND_EXPR does
6321 not depend on the phi.
6323 2017-08-16 Alan Modra <amodra@gmail.com>
6325 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
6328 2017-08-16 Alan Modra <amodra@gmail.com>
6330 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
6331 (save_reg_p): ..into this. Update all callers.
6332 (first_reg_to_save): Simplify.
6334 2017-08-16 Alan Modra <amodra@gmail.com>
6336 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
6339 2017-08-15 Joseph Myers <joseph@codesourcery.com>
6343 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
6344 constant count if that count is less than 32.
6346 2017-08-15 Nathan Sidwell <nathan@acm.org>
6348 * gcc.c (execute): Emit friendlier message if inferior is killed
6349 by an external cause.
6351 2017-08-15 Richard Biener <rguenther@suse.de>
6353 PR tree-optimization/81790
6354 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
6355 CONSTRUCTORs from simplifying and VN.
6357 2017-08-14 Martin Sebor <msebor@redhat.com>
6359 * builtin-attrs.def: Add comments.
6361 2017-08-14 Martin Sebor <msebor@redhat.com>
6364 * doc/extend.texi (attribute nonstring): Document new attribute.
6366 2017-08-14 Martin Sebor <msebor@redhat.com>
6369 * tree-diagnostic.c (default_tree_printer): Handle %G.
6370 * gimple-pretty-print.h (percent_G_format): Declare new function.
6371 * gimple-pretty-print.c (percent_G_format): Define.
6372 * tree-pretty-print.c (percent_K_format): Add argument.
6374 2017-08-14 Martin Sebor <msebor@redhat.com>
6376 PR translation/79998
6377 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
6378 Remove a stray space.
6380 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
6383 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
6384 (*iordi_1_bts): Ditto.
6385 (*xordi_1_btc): Ditto.
6387 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6390 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
6391 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6393 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
6394 quoted strings, and make more translator-friendly.
6395 (darwin_rs6000_override_options): Likewise.
6396 (rs6000_option_override_internal): Likewise.
6397 (rs6000_return_in_memory): Fix overlong line.
6398 (init_cmulative_args): Use quoted strings, and make more
6399 translator-friendly.
6400 (rs6000_pass_by_reference): Fix overlong line.
6401 (def_builtin): Use quoted strings.
6402 (altivec_expand_predicate_builtin): Use quoted strings, and make
6403 more translator-friendly.
6404 (htm_expand_builtin): Use quoted strings.
6405 (cpu_expand_builtin): Use quoted strings, and make more
6406 translator-friendly.
6407 (altivec_expand_builtin): Likewise.
6408 (paired_expand_predicate_builtin): Likewise.
6409 (rs6000_invalid_builtin): Likewise.
6410 (builtin_function_type): Use quoted strings.
6411 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
6412 more translator-friendly.
6413 (rs6000_trampoline_init): Likewise.
6414 (rs6000_handle_altivec_attribute): Likewise.
6415 (rs6000_inner_target_options): Use quoted strings.
6416 (rs6000_disable_incompatible_switches): Likewise.
6417 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
6418 strings, and make more translator-friendly.
6419 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
6421 2017-08-14 Bin Cheng <bin.cheng@arm.com>
6423 PR tree-optimization/81799
6424 * tree-loop-distribution.c (version_loop_by_alias_check): Force
6425 cond_expr to simple gimple operand.
6427 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
6430 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
6432 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
6436 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
6437 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
6438 TARGET_GASISR_PROLOGUES.
6439 * config/avr/avr.c (avr_option_override): Same.
6440 (avr_pass_pre_proep::execute): Same.
6442 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
6445 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
6446 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
6448 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
6450 * config/i386/i386.md (*load_tp_<mode>): Redefine as
6451 define_insn_and_split. Split to a memory load from 0 in
6452 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
6453 using PTR mode iterator.
6454 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
6455 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
6456 (*add_tp_<mode>): Redefine as define_insn_and_split.
6457 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
6458 address space. Merge with *add_tp_x32 using PTR mode iterator.
6459 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
6460 Split to an add with a memory load from 0 in
6461 DEFAULT_TLS_SEG_REG address space.
6463 2017-08-12 Andrew Pinski <apinski@cavium.com>
6465 * config/aarch64/aarch64-option-extensions.def (rdma):
6466 Fix feature string to what Linux prints out in /proc/cpuinfo.
6468 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
6471 * dwarf2out.c (modified_type_die): For C typedef types that have
6472 an ultimate origin, process the ultimate origin instead of the
6474 (gen_typedef_die): Assert that input DECLs have no ultimate
6476 (gen_type_die_with_usage): For typedef variants that have an
6477 ultimate origin, just call gen_decl_die on the original DECL.
6478 (process_scope_var): Avoid creating DIEs for local typedefs and
6479 concrete static variables.
6481 2017-08-12 Alan Modra <amodra@gmail.com>
6485 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
6486 match gnu-user.h startfile.
6487 (ENDFILE_LINUX_SPEC): Similarly.
6489 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
6492 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
6494 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
6496 2017-08-11 Tamar Christina <tamar.christina@arm.com>
6497 * config/aarch64/aarch64.md (mov<mode>): Change.
6498 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
6499 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
6500 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
6502 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
6504 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
6505 for storage order barriers.
6507 2017-08-11 Martin Liska <mliska@suse.cz>
6510 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
6511 variables of void type.
6513 2017-08-11 Martin Liska <mliska@suse.cz>
6515 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
6516 TARGET_SUPPORTS_ALIASES.
6517 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
6518 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
6519 (optimize_weakref): Likewise.
6520 * symtab.c (symtab_node::noninterposable_alias): Likewise.
6521 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
6522 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
6524 2017-08-11 Martin Liska <mliska@suse.cz>
6527 * config/i386/i386.c (make_resolver_func): Do complete
6528 refactoring of the function.
6530 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
6533 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
6534 * config/i386/i386.c (ix86_stack_protect_guard): Use
6535 ix86_stack_protect_guard_symbol_str to generate varible declaration.
6536 * doc/invoke.texi (x86 Options): Document
6537 -mstack-protector-guard-symbol= option.
6539 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
6541 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
6542 * config/i386/i386.c (ix86_split_stack_guard): New function.
6543 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
6544 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
6545 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
6546 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
6547 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
6548 (split_stack_space_check): Call ix86_split_stack_guard.
6550 2017-08-10 Martin Sebor <msebor@redhat.com>
6552 * print-tree.c (print_node): Print location using the established
6554 Replace spaces with colons.
6555 (debug_raw, debug): Ditto.
6557 2017-08-10 Martin Sebor <msebor@redhat.com>
6560 * pretty-print.c (pp_format): Correct the handling of %s precision.
6562 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
6565 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
6567 (ix86_finalize_stack_frame_flags): This. Also clear
6568 frame_pointer_needed if -fno-omit-frame-pointer is used without
6570 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
6571 with ix86_finalize_stack_frame_flags.
6572 (ix86_expand_epilogue): Likewise.
6573 (ix86_expand_split_stack_prologue): Likewise.
6574 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
6576 2017-08-10 Martin Liska <mliska@suse.cz>
6579 * c-attribs.c (handle_target_attribute):
6580 Report warning for an empty string argument of target attribute.
6582 2017-08-09 Jakub Jelinek <jakub@redhat.com>
6585 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
6587 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
6588 or DECL_NONLOCAL labels.
6589 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
6590 or DECL_NONLOCAL labels here.
6592 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
6594 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
6595 to indicate when early gimple folding has been disabled.
6596 (rs6000_gimple_fold_builtin): Add debug content.
6597 (rs6000_invalid_builtin): Fix whitespace.
6598 (rs6000_expand_builtin): Fix whitespace.
6599 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
6601 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
6604 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
6605 SAVE_MULTIPLE if not all the registers that saves, should be saved.
6607 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
6609 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
6610 (qdf24xx): Likewise.
6611 * config/aarch64/aarch64.md: Include falkor.md.
6612 * config/aarch64/falkor.md: New.
6614 2017-08-09 Marek Polacek <polacek@redhat.com>
6617 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
6618 * diagnostic.c (emit_diagnostic): Add a comment.
6619 (emit_diagnostic_valist): New function.
6621 2017-08-09 Marek Polacek <polacek@redhat.com>
6624 * input.c (make_location): New overload.
6625 * input.h (make_location): Declare.
6627 2017-08-08 Alan Modra <amodra@gmail.com>
6628 H.J. Lu <hongjiu.lu@intel.com>
6631 * gcc.c (NO_PIE_SPEC): Delete.
6632 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
6634 (LINK_PIE_SPEC): ..to here.
6635 (LINK_COMMAND_SPEC): Support -no-pie.
6636 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
6637 chain of crtbegin*.o selection, update for PIE_SPEC changes and
6639 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
6640 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
6641 (ENDFILE_CRTEND_SPEC): Similarly.
6643 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
6646 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
6647 (mstack-protector-guard-offset=): Ditto.
6648 * config/i386/i386.c (ix86_option_override): Handle
6649 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
6651 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
6652 ix86_stack_protect_guard_offset variables.
6653 (TARGET_STACK_PROTECT_GUARD): Always define.
6654 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
6655 and -mstack-protector-guard-offset= options.
6657 2017-08-08 Bin Cheng <bin.cheng@arm.com>
6659 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
6660 boundary case for the last candidate.
6662 2017-08-08 Bin Cheng <bin.cheng@arm.com>
6664 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
6665 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
6667 2017-08-08 Tamar Christina <tamar.christina@arm.com>
6670 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
6671 * config/aarch64/aarch64-builtins.c
6672 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
6673 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
6674 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
6676 2017-08-08 Tamar Christina <tamar.christina@arm.com>
6677 Andrew Pinski <pinskia@gmail.com>
6680 * internal-fn.def (XORSIGN): New.
6681 * optabs.def (xorsign_optab): New.
6682 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
6683 (convert_expand_mult_copysign): New.
6684 (pass_optimize_widening_mul::execute): Call
6685 convert_expand_mult_copysign.
6687 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6689 PR tree-optimization/81354
6690 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
6691 Insert on edges rather than explicitly creating landing pads.
6692 (analyze_candidates_and_replace): Commit edge inserts.
6694 2017-08-08 Richard Biener <rguenther@suse.de>
6697 * tree-ssa-loop-niter.c: Include tree-dfa.h.
6698 (expand_simple_operations): Also look through ADDR_EXPRs with
6699 MEM_REF bases treating them as POINTER_PLUS_EXPR.
6701 2017-08-08 Richard Biener <rguenther@suse.de>
6703 PR tree-optimization/81723
6704 * tree-vect-slp.c (struct bst_traits): New hash traits.
6705 (bst_fail): New global.
6706 (vect_build_slp_tree_2): New worker, split out from ...
6707 (vect_build_slp_tree): ... this now wrapping it with using
6708 bst_fail set to cache SLP tree build fails. Properly handle
6710 (vect_analyze_slp_instance): Allocate and free bst_fail.
6712 2017-08-08 Martin Liska <mliska@suse.cz>
6715 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
6716 LABEL_DECLs that can be from a different function.
6718 2017-08-08 Bin Cheng <bin.cheng@arm.com>
6720 PR tree-optimization/81744
6721 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
6722 loop's number of iterations.
6724 2017-08-08 Martin Liska <mliska@suse.cz>
6726 * asan.c: Include header files.
6727 * attribs.c (build_decl_attribute_variant): New function moved
6729 (build_type_attribute_qual_variant): Likewise.
6730 (cmp_attrib_identifiers): Likewise.
6731 (simple_cst_list_equal): Likewise.
6732 (omp_declare_simd_clauses_equal): Likewise.
6733 (attribute_value_equal): Likewise.
6734 (comp_type_attributes): Likewise.
6735 (build_type_attribute_variant): Likewise.
6736 (lookup_ident_attribute): Likewise.
6737 (remove_attribute): Likewise.
6738 (merge_attributes): Likewise.
6739 (merge_type_attributes): Likewise.
6740 (merge_decl_attributes): Likewise.
6741 (merge_dllimport_decl_attributes): Likewise.
6742 (handle_dll_attribute): Likewise.
6743 (attribute_list_equal): Likewise.
6744 (attribute_list_contained): Likewise.
6745 * attribs.h (lookup_attribute): New function moved from tree.[ch].
6746 (lookup_attribute_by_prefix): Likewise.
6747 * bb-reorder.c: Include header files.
6748 * builtins.c: Likewise.
6749 * calls.c: Likewise.
6750 * cfgexpand.c: Likewise.
6751 * cgraph.c: Likewise.
6752 * cgraphunit.c: Likewise.
6753 * convert.c: Likewise.
6754 * dwarf2out.c: Likewise.
6755 * final.c: Likewise.
6756 * fold-const.c: Likewise.
6757 * function.c: Likewise.
6758 * gimple-expr.c: Likewise.
6759 * gimple-fold.c: Likewise.
6760 * gimple-pretty-print.c: Likewise.
6761 * gimple.c: Likewise.
6762 * gimplify.c: Likewise.
6763 * hsa-common.c: Likewise.
6764 * hsa-gen.c: Likewise.
6765 * internal-fn.c: Likewise.
6766 * ipa-chkp.c: Likewise.
6767 * ipa-cp.c: Likewise.
6768 * ipa-devirt.c: Likewise.
6769 * ipa-fnsummary.c: Likewise.
6770 * ipa-inline.c: Likewise.
6771 * ipa-visibility.c: Likewise.
6773 * lto-cgraph.c: Likewise.
6774 * omp-expand.c: Likewise.
6775 * omp-general.c: Likewise.
6776 * omp-low.c: Likewise.
6777 * omp-offload.c: Likewise.
6778 * omp-simd-clone.c: Likewise.
6779 * opts-global.c: Likewise.
6780 * passes.c: Likewise.
6781 * predict.c: Likewise.
6782 * sancov.c: Likewise.
6783 * sanopt.c: Likewise.
6784 * symtab.c: Likewise.
6785 * toplev.c: Likewise.
6786 * trans-mem.c: Likewise.
6787 * tree-chkp.c: Likewise.
6788 * tree-eh.c: Likewise.
6789 * tree-into-ssa.c: Likewise.
6790 * tree-object-size.c: Likewise.
6791 * tree-parloops.c: Likewise.
6792 * tree-profile.c: Likewise.
6793 * tree-ssa-ccp.c: Likewise.
6794 * tree-ssa-live.c: Likewise.
6795 * tree-ssa-loop.c: Likewise.
6796 * tree-ssa-sccvn.c: Likewise.
6797 * tree-ssa-structalias.c: Likewise.
6798 * tree-ssa.c: Likewise.
6799 * tree-streamer-in.c: Likewise.
6800 * tree-vectorizer.c: Likewise.
6801 * tree-vrp.c: Likewise.
6803 * ubsan.c: Likewise.
6804 * varasm.c: Likewise.
6805 * varpool.c: Likewise.
6806 * tree.c: Remove functions moved to attribs.[ch].
6808 * config/aarch64/aarch64.c: Add attrs.h header file.
6809 * config/alpha/alpha.c: Likewise.
6810 * config/arc/arc.c: Likewise.
6811 * config/arm/arm.c: Likewise.
6812 * config/avr/avr.c: Likewise.
6813 * config/bfin/bfin.c: Likewise.
6814 * config/c6x/c6x.c: Likewise.
6815 * config/cr16/cr16.c: Likewise.
6816 * config/cris/cris.c: Likewise.
6817 * config/darwin.c: Likewise.
6818 * config/epiphany/epiphany.c: Likewise.
6819 * config/fr30/fr30.c: Likewise.
6820 * config/frv/frv.c: Likewise.
6821 * config/ft32/ft32.c: Likewise.
6822 * config/h8300/h8300.c: Likewise.
6823 * config/i386/winnt.c: Likewise.
6824 * config/ia64/ia64.c: Likewise.
6825 * config/iq2000/iq2000.c: Likewise.
6826 * config/lm32/lm32.c: Likewise.
6827 * config/m32c/m32c.c: Likewise.
6828 * config/m32r/m32r.c: Likewise.
6829 * config/m68k/m68k.c: Likewise.
6830 * config/mcore/mcore.c: Likewise.
6831 * config/microblaze/microblaze.c: Likewise.
6832 * config/mips/mips.c: Likewise.
6833 * config/mmix/mmix.c: Likewise.
6834 * config/mn10300/mn10300.c: Likewise.
6835 * config/moxie/moxie.c: Likewise.
6836 * config/msp430/msp430.c: Likewise.
6837 * config/nds32/nds32-isr.c: Likewise.
6838 * config/nds32/nds32.c: Likewise.
6839 * config/nios2/nios2.c: Likewise.
6840 * config/nvptx/nvptx.c: Likewise.
6841 * config/pa/pa.c: Likewise.
6842 * config/pdp11/pdp11.c: Likewise.
6843 * config/powerpcspe/powerpcspe.c: Likewise.
6844 * config/riscv/riscv.c: Likewise.
6845 * config/rl78/rl78.c: Likewise.
6846 * config/rx/rx.c: Likewise.
6847 * config/s390/s390.c: Likewise.
6848 * config/sh/sh.c: Likewise.
6849 * config/sol2.c: Likewise.
6850 * config/sparc/sparc.c: Likewise.
6851 * config/spu/spu.c: Likewise.
6852 * config/stormy16/stormy16.c: Likewise.
6853 * config/tilegx/tilegx.c: Likewise.
6854 * config/tilepro/tilepro.c: Likewise.
6855 * config/v850/v850.c: Likewise.
6856 * config/vax/vax.c: Likewise.
6857 * config/visium/visium.c: Likewise.
6858 * config/xtensa/xtensa.c: Likewise.
6860 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
6863 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
6864 constraints since the -mupper-regs-* switches have been
6866 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
6867 into a vector from a double word element that was extracted from
6868 another vector, and eliminate extra XXPERMDI instructions.
6869 (vsx_concat_<mode>_2): Likewise.
6870 (vsx_concat_<mode>_3): Likewise.
6871 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
6872 concat to allow optimizing inserts from previous extracts.
6874 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
6876 * config/i386/i386.c (ix86_stack_protect_guard): Generate
6877 memory reference to a SSP offset in TLS address space.
6878 (ix86_print_operand) <case '@'>: Remove.
6879 (ix86_print_operand_punct_valid_p): Remove '@' code.
6880 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
6882 (stack_tls_protect_set_<mode>): Remove.
6883 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
6884 (stack_tls_protect_test_<mode>): Remove.
6885 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
6887 2017-08-07 Olivier Hainque <hainque@adacore.com>
6890 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
6892 2017-08-07 Douglas Rupp <rupp@adacore.com>
6894 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
6895 variable was referenced as multidir in command.
6897 2017-08-07 Jakub Jelinek <jakub@redhat.com>
6900 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
6903 2017-08-07 Martin Liska <mliska@suse.cz>
6905 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
6906 * config/rl78/rl78.c: Add include of attribs.h.
6907 * config/sh/sh.c: Likewise.
6908 * config/v850/v850.c: Likewise.
6910 2017-08-07 Tom de Vries <tom@codesourcery.com>
6913 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
6915 2017-08-07 Martin Liska <mliska@suse.cz>
6917 * config/mips/mips.c: Include attribs.h.
6919 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
6922 * doc/invoke.texi: Document change in behvaior for -Ofast for
6925 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
6927 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
6929 (aarch64_pop_regs): Likewise.
6930 (aarch64_gen_load_pair): Likewise.
6931 (aarch64_save_callee_saves): Likewise.
6932 (aarch64_restore_callee_saves): Likewise.
6934 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
6936 * config/i386/i386.c: Revert the last change.
6938 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
6941 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
6943 (ix86_finalize_stack_frame_flags): This. Also clear
6944 frame_pointer_needed if -fno-omit-frame-pointer is used without
6946 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
6947 with ix86_finalize_stack_frame_flags.
6948 (ix86_expand_epilogue): Likewise.
6949 (ix86_expand_split_stack_prologue): Likewise.
6951 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
6954 * config/i386/i386.c (get_builtin_code_for_version): Set priority
6955 to P_AES for Westmere.
6957 2017-08-07 Jonathan Yong <10walls@gmail.com>
6959 * config/i386/mingw.opt (fset-stack-executable): Removed.
6960 * config/i386/cygming.opt (fset-stack-executable): Moved
6962 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
6964 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
6966 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
6968 2017-08-07 Marek Polacek <polacek@redhat.com>
6971 * fold-const.c (fold_indirect_ref_1): Check type_domain.
6973 2017-08-07 Martin Liska <mliska@suse.cz>
6975 * attribs.h (canonicalize_attr_name): New function.
6976 (cmp_attribs): Move from c-format.c and adjusted.
6977 (is_attribute_p): Moved from tree.h.
6978 * tree-inline.c: Add new includes.
6979 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
6980 (private_is_attribute_p): Remove.
6981 (private_lookup_attribute): Likewise.
6982 (private_lookup_attribute_by_prefix): Simplify.
6983 (remove_attribute): Use is_attribute_p.
6984 * tree.h: Remove removed declarations.
6986 2017-08-07 Jakub Jelinek <jakub@redhat.com>
6989 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
6990 instead of computing it in the function. Formatting fix.
6991 (expand_case): Don't rely on default_edge being the first edge,
6992 clear it if removing it, pass default_edge to
6993 emit_case_dispatch_table.
6994 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
6997 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
6999 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
7000 insn in the function, emit NOP after the insn.
7002 2017-08-06 Tom de Vries <tom@codesourcery.com>
7004 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
7007 2017-08-06 Tom de Vries <tom@codesourcery.com>
7009 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
7012 2017-08-04 Yury Gribov <tetra2005@gmail.com>
7014 PR tree-optimization/57371
7015 * match.pd: New pattern.
7017 2017-08-04 Marek Polacek <polacek@redhat.com>
7020 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
7021 perform the computation in offset_int.
7023 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
7025 PR tree-optimization/81136
7026 * tree-vectorizer.h: Include tree-hash-traits.h.
7027 (vec_base_alignments): New typedef.
7028 (vec_info): Add a base_alignments field.
7029 (vect_record_base_alignments): Declare.
7030 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
7032 (DR_IS_CONDITIONAL_IN_STMT): New macro.
7033 (create_data_ref): Add an is_conditional_in_stmt argument.
7034 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
7035 the is_conditional_in_stmt field.
7036 (data_ref_loc): Add an is_conditional_in_stmt field.
7037 (get_references_in_stmt): Set the is_conditional_in_stmt field.
7038 (find_data_references_in_stmt): Update call to create_data_ref.
7039 (graphite_find_data_references_in_stmt): Likewise.
7040 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
7041 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
7042 (vect_record_base_alignment): New function.
7043 (vect_record_base_alignments): Likewise.
7044 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
7045 for nested statements even if we fail to compute a misalignment.
7046 Use pooled base alignments for unconditional references.
7047 (vect_find_same_alignment_drs): Compare base addresses instead
7049 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
7050 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
7052 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
7054 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
7055 Add an explicit name for the enum. Use auto_vec for slp_instances
7057 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
7059 (_bb_vec_info): Add a constructor and destructor.
7060 (vinfo_for_stmt): Return NULL for uids of -1 as well.
7061 (destroy_loop_vec_info): Delete.
7062 (vect_destroy_datarefs): Likewise.
7063 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
7064 (vec_info::vec_info): New function.
7065 (vec_info::~vec_info): Likewise.
7066 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
7067 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
7068 destroy_loop_vec_info.
7069 * tree-vect-loop.c (new_loop_vec_info): Replace with...
7070 (_loop_vec_info::_loop_vec_info): ...this.
7071 (destroy_loop_vec_info): Replace with...
7072 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
7073 the stmt_vec_infos. Leave handling of vec_info information to its
7074 destructor. Remove explicit vector releases.
7075 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
7076 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
7077 * tree-vect-slp.c (new_bb_vec_info): Replace with...
7078 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
7079 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
7080 (destroy_bb_vec_info): Replace with...
7081 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
7082 information to its destructor.
7083 (vect_slp_analyze_bb_1): Use new and delete instead of
7084 new_bb_vec_info and destroy_bb_vec_info.
7085 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
7088 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
7090 * tree-data-ref.h (subscript): Add access_fn field.
7091 (data_dependence_relation): Add could_be_independent_p.
7092 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
7093 (same_access_functions): Move to tree-data-ref.c.
7094 * tree-data-ref.c (ref_contains_union_access_p): New function.
7095 (access_fn_component_p): Likewise.
7096 (access_fn_components_comparable_p): Likewise.
7097 (dr_analyze_indices): Add a reference to access_fn_component_p.
7098 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
7100 (constant_access_functions): Likewise.
7101 (add_other_self_distances): Likewise.
7102 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
7103 (initialize_data_dependence_relation): Use XCNEW and remove
7104 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
7105 of access functions that have the same type. Allow the
7106 subsequence to end with different bases in some circumstances.
7107 Record the chosen access functions in SUB_ACCESS_FN.
7108 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
7109 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
7110 (subscript_dependence_tester_1): Likewise dra and drb.
7111 (build_classic_dist_vector): Update calls accordingly.
7112 (subscript_dependence_tester): Likewise.
7113 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
7114 DDR_COULD_BE_INDEPENDENT_P.
7115 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
7116 comp_alias_ddrs instead of may_alias_ddrs.
7117 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
7119 (vect_analyze_data_ref_dependence): Use it if
7120 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
7121 distance vectors if that fails.
7122 (dependence_distance_ge_vf): New function.
7123 (vect_prune_runtime_alias_test_list): Use it. Don't clear
7124 LOOP_VINFO_MAY_ALIAS_DDRS.
7126 2017-08-04 Richard Biener <rguenther@suse.de>
7129 * fold-const.c (fold_binary_loc): Properly restrict
7130 minus_var0 && minus_var1 case when associating undefined overflow
7133 2017-08-04 Tom de Vries <tom@codesourcery.com>
7135 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
7137 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7139 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7140 Don't start diagnostic messages with a capital letter.
7141 * config/rs6000/rs6000.c (rs6000_option_override_internal):
7143 (rs6000_invalid_builtin): Likewise.
7144 (rs6000_trampoline_init): Likewise.
7146 2017-08-03 Jakub Jelinek <jakub@redhat.com>
7149 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
7150 after setting changeable df flags.
7152 2017-08-03 Richard Biener <rguenther@suse.de>
7154 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
7155 up if the use is in USE - X.
7157 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
7159 * toplev.c (dumpfile.h): New include.
7160 (internal_error_reentered): New static function. Use it...
7161 (internal_error_function): ...here to handle reentered internal_error.
7163 2017-08-03 Richard Biener <rguenther@suse.de>
7166 * fold-const.c (split_tree): Add minus_var and minus_con
7167 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
7168 here but always use minus_*.
7169 (associate_trees): Assert we never associate with MINUS_EXPR
7170 and NULL first operand. Do not recurse for PLUS_EXPR operands
7171 when associating as MINUS_EXPR either.
7172 (fold_binary_loc): Track minus_var and minus_con.
7174 2017-08-03 Tom de Vries <tom@codesourcery.com>
7177 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
7178 ACCEL_COMPILER, apply finish_options on
7179 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
7181 2017-08-03 Tom de Vries <tom@codesourcery.com>
7184 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
7185 function_entry_patch_area_size > 0.
7187 2017-08-03 Jakub Jelinek <jakub@redhat.com>
7190 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
7191 instead of 10??LU, perform unit multiplication in wide_int,
7192 don't change alloc_object_size_limit if the limit is larger
7195 PR tree-optimization/81655
7196 PR tree-optimization/81588
7197 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
7198 the case when ranges[i].low and high are 1 for unsigned type with
7202 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
7203 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
7205 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
7207 * tree-vrp.h: Add include guard.
7209 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
7212 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
7213 (ud2): New insn pattern.
7214 * config/i386/i386.c (ix86_expand_epilogue):
7215 For naked functions, generate ud2 instead of trap insn.
7217 2017-08-02 Marek Polacek <polacek@redhat.com>
7220 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
7222 2017-08-02 Tom de Vries <tom@codesourcery.com>
7223 Cesar Philippidis <cesar@codesourcery.com>
7225 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
7226 Add missing edge probabilities.
7228 2017-08-02 Tamar Christina <tamar.christina@arm.com>
7230 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
7233 2017-08-02 Jakub Jelinek <jakub@redhat.com>
7236 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
7237 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
7238 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
7239 prologue_seq sequences - if any.
7241 2017-08-02 Richard Biener <rguenther@suse.de>
7243 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
7244 via vectors if supported, integer extracts via punning if supported
7245 or otherwise vector extracts.
7247 2017-08-02 Richard Biener <rguenther@suse.de>
7249 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
7251 (bitmap_insert_into_set): ... this.
7253 2017-08-02 Richard Biener <rguenther@suse.de>
7255 PR tree-optimization/81633
7257 2015-08-17 Alan Hayward <alan.hayward@arm.com>
7259 PR tree-optimization/71752
7260 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
7262 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
7264 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
7265 (machine_function::call_ms2sysv_pad_out): Remove field.
7266 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
7267 (ix86_compute_frame_layout): Likewise.
7269 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
7272 * config/i386/i386.c (ix86_set_func_type): Disallow naked
7273 attribute with interrupt attribute.
7275 2017-08-01 Andrew Pinski <apinski@cavium.com>
7277 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
7278 BIT_INSERT_EXPR's operand 1
7279 to see if the types precision matches.
7281 2017-08-01 Martin Liska <mliska@suse.cz>
7284 * builtins.c (expand_builtin_memcpy_args): Remove.
7285 (expand_builtin_memcpy): Call newly added function
7286 expand_builtin_memory_copy_args.
7287 (expand_builtin_memcpy_with_bounds): Likewise.
7288 (expand_builtin_mempcpy): Remove last argument.
7289 (expand_builtin_mempcpy_with_bounds): Likewise.
7290 (expand_builtin_memory_copy_args): New function created from
7291 expand_builtin_mempcpy_args with small modifications.
7292 (expand_builtin_mempcpy_args): Remove.
7293 (expand_builtin_stpcpy): Remove unused argument.
7294 (expand_builtin): Likewise.
7295 (expand_builtin_with_bounds): Likewise.
7297 2017-08-01 Martin Liska <mliska@suse.cz>
7300 Make mempcpy more optimal (PR middle-end/70140).
7302 2017-08-01 Jakub Jelinek <jakub@redhat.com>
7305 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
7306 __builtin_vec_cmpne verify both arguments are compatible vectors
7307 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
7308 verify arg1_type is a pointer or array type. For __builtin_vec_st,
7309 move computation of aligned to after checking the argument types.
7313 * config/rs6000/vsx.md (vextract_fp_from_shorth,
7314 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
7317 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
7318 Doug Rupp <rupp@adacore.com>
7319 Olivier Hainque <hainque@adacore.com>
7321 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
7322 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
7324 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
7325 for TARGET_OS_CPP_BUILTIN.
7326 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
7327 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
7329 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
7330 passing required abi options to the assembler for EABI configurations.
7331 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
7332 of .text.hot and .text.unlikely sections for kernel modules when
7333 using ARM style exceptions.
7334 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
7335 options. Add EXTRA_CC1_SPEC.
7336 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
7338 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
7340 (ARM_TARGET2_DWARF_FORMAT): Define.
7341 * config/arm/t-vxworks: Adjust multilib control to removal of the
7342 Diab command line options.
7344 2017-08-01 Martin Liska <mliska@suse.cz>
7346 PR gcov-profile/81561
7347 * gcov.c (unblock): Make unblocking safe as we need to preserve
7348 index correspondence of blocks and block_lists.
7350 2017-08-01 Richard Biener <rguenther@suse.de>
7352 PR tree-optimization/81181
7353 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
7354 (compute_antic): ... end of iteration here.
7356 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
7358 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
7359 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
7360 (ftree-slp-vectorize): Likewise.
7361 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
7362 can no longer be set independent of flag_tree_loop_vectorize.
7363 * omp-general.c (emp_max_vf): Likewise.
7364 * opts.c (enable_fdo_optimizations): Remove references to
7365 flag_tree_vectorize, these are now implicit.
7366 (common_handle_option): Remove handling for OPT_ftree_vectorize,
7367 and leave it for the options machinery.
7369 2017-08-01 Martin Liska <mliska@suse.cz>
7372 * builtins.c (expand_builtin_memcpy_args): Remove.
7373 (expand_builtin_memcpy): Call newly added function
7374 expand_builtin_memory_copy_args.
7375 (expand_builtin_memcpy_with_bounds): Likewise.
7376 (expand_builtin_mempcpy): Remove last argument.
7377 (expand_builtin_mempcpy_with_bounds): Likewise.
7378 (expand_builtin_memory_copy_args): New function created from
7379 expand_builtin_mempcpy_args with small modifications.
7380 (expand_builtin_mempcpy_args): Remove.
7381 (expand_builtin_stpcpy): Remove unused argument.
7382 (expand_builtin): Likewise.
7383 (expand_builtin_with_bounds): Likewise.
7385 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
7388 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
7389 print "ds:" only for immediates in generic address space.
7391 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
7394 * config/i386/i386.c (ix86_funciton_naked): New prototype.
7395 (ix86_function_ok_for_sibcall): Return false for naked functions.
7397 2017-08-01 Richard Biener <rguenther@suse.de>
7399 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
7400 (compute_antic): Seed worklist with exit block predecessors.
7401 * cfganal.c (dfs_find_deadend): For a cycle return the source
7402 of the edge closing it.
7404 2017-08-01 Tamar Christina <tamar.christina@arm.com>
7406 * config/aarch64/aarch64.c
7407 (aarch64_can_const_movi_rtx_p): Move 0 check.
7409 2017-08-01 Bin Cheng <bin.cheng@arm.com>
7411 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
7412 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
7414 * match.pd: Ditto in address comparison pattern.
7416 2017-08-01 Bin Cheng <bin.cheng@arm.com>
7418 PR tree-optimization/81627
7419 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
7420 closed ssa form for store-store chain.
7422 2017-08-01 Bin Cheng <bin.cheng@arm.com>
7424 PR tree-optimization/81620
7425 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
7426 for store-store chain.
7428 2017-08-01 Jakub Jelinek <jakub@redhat.com>
7430 PR tree-optimization/81588
7431 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
7432 ranges[i].in_p, invert comparison code ccode. For >/>=,
7433 swap rhs1 and rhs2 and comparison code unconditionally,
7434 for </<= don't do that. Don't swap rhs1/rhs2 again if
7435 ranges[i].in_p, instead invert comparison code ccode if
7436 opcode or oe->rank is BIT_IOR_EXPR.
7439 * optabs.def (vec_extract_optab, vec_init_optab): Change from
7440 a direct optab to conversion optab.
7441 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
7442 with GET_MODE_INNER as last argument instead of optab_handler.
7443 * expmed.c (extract_bit_field_1): Likewise. Use vector from
7444 vector extraction if possible and optab is available.
7445 * expr.c (store_constructor): Use convert_optab_handler instead
7446 of optab_handler. Use vector initialization from smaller
7447 vectors if possible and optab is available.
7448 * tree-vect-stmts.c (vectorizable_load): Likewise.
7449 * doc/md.texi (vec_extract, vec_init): Document that the optabs
7451 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
7452 of vec_init from half-sized vectors with the same element mode.
7453 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
7454 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
7455 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
7456 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
7457 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
7458 after mode in gen_vec_extract* calls.
7459 (vec_extract<mode>): Renamed to ...
7460 (vec_extract<mode><ssescalarmodelower>): ... this.
7461 (vec_extract<mode><ssehalfvecmodelower>): New expander.
7462 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
7463 element mode after mode in gen_vec_init* calls.
7464 (VEC_INIT_HALF_MODE): New mode iterator.
7465 (vec_init<mode>): Renamed to ...
7466 (vec_init<mode><ssescalarmodelower>): ... this.
7467 (vec_init<mode><ssehalfvecmodelower>): New expander.
7468 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
7469 (vec_extractv2sfsf): ... this.
7470 (vec_initv2sf): Renamed to ...
7471 (vec_initv2sfsf): ... this.
7472 (vec_extractv2si): Renamed to ...
7473 (vec_extractv2sisi): ... this.
7474 (vec_initv2si): Renamed to ...
7475 (vec_initv2sisi): ... this.
7476 (vec_extractv4hi): Renamed to ...
7477 (vec_extractv4hihi): ... this.
7478 (vec_initv4hi): Renamed to ...
7479 (vec_initv4hihi): ... this.
7480 (vec_extractv8qi): Renamed to ...
7481 (vec_extractv8qiqi): ... this.
7482 (vec_initv8qi): Renamed to ...
7483 (vec_initv8qiqi): ... this.
7484 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
7485 (vec_init<mode>): Renamed to ...
7486 (vec_init<mode><VEC_base_l>): ... this.
7487 (vec_extract<mode>): Renamed to ...
7488 (vec_extract<mode><VEC_base_l>): ... this.
7489 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
7490 (vec_initv2sfsf): ... this.
7491 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
7492 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
7493 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
7494 element mode after mode in gen_vec_init* calls.
7495 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
7496 (vec_init<mode><Vel>): ... this.
7497 (vec_extract<mode>): Renamed to ...
7498 (vec_extract<mode><Vel>): ... this.
7499 * config/aarch64/iterators.md (Vel): New mode attribute.
7500 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
7501 Add element mode after mode in gen_vec_extract* calls.
7502 * config/s390/vector.md (non_vec_l): New mode attribute.
7503 (vec_extract<mode>): Renamed to ...
7504 (vec_extract<mode><non_vec_l>): ... this.
7505 (vec_init<mode>): Renamed to ...
7506 (vec_init<mode><non_vec_l>): ... this.
7507 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
7508 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
7510 * config/arm/iterators.md (V_elem_l): New mode attribute.
7511 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
7512 (vec_extract<mode><V_elem_l>): ... this.
7513 (vec_extractv2di): Renamed to ...
7514 (vec_extractv2didi): ... this.
7515 (vec_init<mode>): Renamed to ...
7516 (vec_init<mode><V_elem_l>): ... this.
7517 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
7518 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
7519 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
7520 Add element mode after gen_vec_extract* calls.
7521 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
7522 (vec_init<mode><unitmode>): ... this.
7523 (vec_extract<mode>): Renamed to ...
7524 (vec_extract<mode><unitmode>): ... this.
7525 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
7526 (vec_init<mode><unitmode>): ... this.
7527 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
7528 (vec_initv2sfsf): ... this.
7529 (vec_extractv2sf): Renamed to ...
7530 (vec_extractv2sfsf): ... this.
7531 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
7532 Add element mode after gen_vec_extract* calls.
7533 * config/mips/mips.md (unitmode): New mode iterator.
7534 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
7535 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
7536 * config/spu/spu.md (inner_l): New mode attribute.
7537 (vec_init<mode>): Renamed to ...
7538 (vec_init<mode><inner_l>): ... this.
7539 (vec_extract<mode>): Renamed to ...
7540 (vec_extract<mode><inner_l>): ... this.
7541 * config/sparc/sparc.md (veltmode): New mode iterator.
7542 (vec_init<VMALL:mode>): Renamed to ...
7543 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
7544 * config/ia64/vect.md (vec_initv2si): Renamed to ...
7545 (vec_initv2sisi): ... this.
7546 (vec_initv2sf): Renamed to ...
7547 (vec_initv2sfsf): ... this.
7548 (vec_extractv2sf): Renamed to ...
7549 (vec_extractv2sfsf): ... this.
7550 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
7551 (vec_init<mode>): Renamed to ...
7552 (vec_init<mode><VEC_base_l>): ... this.
7553 (vec_extract<mode>): Renamed to ...
7554 (vec_extract<mode><VEC_base_l>): ... this.
7555 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
7556 (vec_initv2sfsf): ... this.
7557 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
7558 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
7559 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
7560 gen_vec_init* calls.
7562 2017-08-01 Richard Biener <rguenther@suse.de>
7564 PR tree-optimization/81297
7565 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
7566 TREE_OVERFLOW from INTEGER_CSTs.
7568 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
7570 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
7572 2017-07-31 Carl Love <cel@us.ibm.com>
7574 * config/rs6000/rs6000-c: Add support for built-in functions
7575 vector signed char vec_xl_be (signed long long, signed char *);
7576 vector unsigned char vec_xl_be (signed long long, unsigned char *);
7577 vector signed int vec_xl_be (signed long long, signed int *);
7578 vector unsigned int vec_xl_be (signed long long, unsigned int *);
7579 vector signed long long vec_xl_be (signed long long, signed long long *);
7580 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
7581 vector signed short vec_xl_be (signed long long, signed short *);
7582 vector unsigned short vec_xl_be (signed long long, unsigned short *);
7583 vector double vec_xl_be (signed long long, double *);
7584 vector float vec_xl_be (signed long long, float *);
7585 * config/rs6000/altivec.h (vec_xl_be): Add #define.
7586 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
7587 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
7589 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
7590 (altivec_expand_builtin): Add switch statement to call
7591 altivec_expand_xl_be for each builtin.
7592 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
7593 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
7594 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
7595 __builtin_vsx_le_be_v16qi.
7596 * doc/extend.texi: Update the built-in documentation file for the
7597 new built-in functions.
7599 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
7602 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
7604 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
7606 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7608 * config.gcc: Add z14.
7609 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
7610 CPU model numbers for z13s and z14.
7611 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
7613 * config/s390/s390-opts.h (enum processor_type): Rename
7614 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
7615 * config/s390/s390.c (processor_table): Add field for CPU name to
7616 be passed to Binutils.
7617 (s390_asm_output_machine_for_arch): Use the new field in
7618 processor_table for Binutils.
7619 (s390_expand_builtin): Replace arch12 with z14.
7620 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
7621 (s390_get_sched_attrmask): Likewise.
7622 (s390_get_unit_mask): Likewise.
7623 * config/s390/s390.opt: Add z14 to processor_type enum.
7625 2017-07-31 Martin Jambor <mjambor@suse.cz>
7628 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
7629 regardless of optimization level.
7631 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
7632 Martin Liska <mliska@suse.cz>
7634 * predict.def: Remove old comment and adjust probability.
7635 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
7638 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
7641 * config/i386/i386.c (ix86_function_naked): New function.
7642 (ix86_can_use_return_insn_p): Return false for naked functions.
7643 (ix86_expand_prologue): Skip prologue for naked functions.
7644 (ix86_expand_epilogue): Skip epilogue for naked functions
7645 and emit trap instruction.
7646 (ix86_warn_func_return): New function.
7647 (ix86_attribute_table): Add "naked" attribute specification.
7648 (TARGET_WARN_FUNC_RETURN): Define.
7649 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
7651 2017-07-31 Martin Liska <mliska@suse.cz>
7653 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
7654 (dump_gimple_bb_header): Always dump BB info.
7655 (pp_cfg_jump): Do not append info about BB when dumping a jump.
7657 2017-07-31 Martin Liska <mliska@suse.cz>
7660 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
7661 also with current_function_decl non-null equality.
7663 2017-07-31 Jakub Jelinek <jakub@redhat.com>
7666 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
7667 change type to the element type, instead add eltype variable and
7668 use it where we are interested in the element type.
7670 PR tree-optimization/81603
7671 * ipa-polymorphic-call.c
7672 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
7673 offset arithmetic in offset_int, bail out if the resulting bit offset
7674 doesn't fit into shwi.
7676 2017-07-31 Martin Liska <mliska@suse.cz>
7678 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
7679 (gimplify_save_expr): Fix comment.
7681 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
7684 * config/i386/i386.c (ix86_function_arg): Update arguments for
7686 (ix86_compute_frame_layout): Set the initial stack offset to
7687 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
7688 INCOMING_FRAME_SP_OFFSET.
7689 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
7690 stack before exception handler returns.
7691 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
7692 the 'ERROR_CODE' for exception handler.
7694 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
7696 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
7697 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
7698 (ASM_OUTPUT_REG_POP): Ditto.
7699 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
7700 instead of asm_fprintf to output pure string.
7702 2017-07-29 Jakub Jelinek <jakub@redhat.com>
7704 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
7705 to imported_module_or_decl hook.
7706 (debug_nothing_tree_tree_tree_bool): Remove.
7707 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
7708 * debug.c (do_nothing_debug_hooks): Use
7709 debug_nothing_tree_tree_tree_bool_bool instead of
7710 debug_nothing_tree_tree_tree_bool.
7711 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
7712 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
7713 * sdbout.c (sdb_debug_hooks): Likewise.
7714 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
7715 (gen_namespace_die): Add DW_AT_export_symbols attribute if
7717 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
7718 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
7719 attribute, don't add anything.
7721 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7723 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
7724 (fold_build2_stat_loc): Likewise.
7725 (fold_build3_stat_loc): Likewise.
7726 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
7727 (fold_build1_loc): Remove macro.
7728 (fold_build2_loc): Likewise.
7729 (fold_build3_loc): Likewise.
7731 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7733 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
7734 (gimple_build_debug_bind_source_stat): Likewise.
7735 * gimple.h (gimple_build_debug_bind): Remove macro.
7736 (gimple_build_debug_bind_source): Likewise.
7738 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7740 * bitmap.c (bitmap_alloc): Adjust.
7741 (bitmap_gc_alloc): Likewise.
7742 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
7744 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7746 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
7747 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
7748 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
7749 (bitmap_gc_alloc_stat): Likewise.
7750 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
7752 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7754 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
7755 * rtl.h (shallow_copy_rtx): Remove macro.
7757 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7759 * emit-rtl.c (gen_raw_REG): Adjust.
7760 * gengenrtl.c (gendef): Likewise.
7761 * rtl.c (rtx_alloc_stat): Remove _stat from name.
7762 * rtl.h (rtx_alloc): Remove macro.
7764 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7766 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
7767 (build_tree_list_stat): Likewise.
7768 * tree.h (build_tree_list): Remove macro.
7769 (build_tree_list_vec): Likewise.
7771 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7773 * tree.c (make_vector_stat): Remove _stat from name.
7774 (build_vector_stat): Likewise.
7775 * tree.h (make_vector_stat): Remove macro.
7776 (build_vector_stat): Likewise.
7778 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7780 * tree.h (build_var_debug_value): Remove prototype.
7782 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7784 * tree.c (tree_cons_stat): Remove _stat from name.
7785 * tree.h (tree_cons): Remove macro.
7787 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7789 * tree.c (build_vl_exp_stat): Remove _stat from name.
7790 * tree.h (build_vl_exp): Remove macro.
7792 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7794 * tree.c (build_decl_stat): Remove _stat from name.
7795 * tree.h (build_decl): Remove macro.
7797 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7799 * gimple.c (gimple_build_with_ops_stat): Adjust.
7800 (gimple_alloc_stat): Remove _stat from name.
7801 * gimple.h (gimple_alloc): Remove macro.
7803 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7805 * tree.c (make_tree_vec_stat): Remove _stat from name.
7806 (grow_tree_vec_stat): Likewise.
7807 * tree.h (make_tree_vec_stat): Adjust prototype.
7808 (grow_tree_vec_stat): Likewise.
7809 (make_tree_vec): Remove macro.
7810 (grow_tree_vec): Likewise.
7812 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7814 * fold-const.c (fold_build1_stat_loc): Adjust.
7815 (fold_build2_stat_loc): Likewise.
7816 (fold_build3_stat_loc): Likewise.
7817 * tree.c (build0_stat): Remove _stat from name.
7818 (build1_stat): Likewise.
7819 (build2_stat): Likewise.
7820 (build3_stat): Likewise.
7821 (build4_stat): Likewise.
7822 (build5_stat): Likewise.
7823 * tree.h (build1_loc): Remove macro, and rename _stat function
7825 (build2_loc): Likewise.
7826 (build3_loc): Likewise.
7827 (build4_loc): Likewise.
7828 (build5_loc): Likewise.
7830 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7832 * tree.c (make_int_cst_stat): Remove _stat from name.
7833 * tree.h (make_int_cst_stat): Adjust prototype.
7834 (make_int_cst): Remove macro.
7836 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7838 * tree.c (make_tre_binfo_stat): Remove _stat from name.
7839 * tree.h (make_tree_binfo_stat): Adjust prototype.
7840 (make_tree_binfo): Remove.
7842 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7844 * tree.c (copy_node_stat): Rename to copy_node.
7845 (build_distinct_type_copy): Adjust.
7846 * tree.h (copy_node_stat): Adjust prototype.
7847 (copy_node): Remove macro.
7849 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7851 * tree.c (make_node_stat): rename to make_node.
7852 (build_tree_list_stat): Adjust.
7853 (build0_stat): Likewise.
7854 (build2_stat): Likewise.
7855 (build3_stat): Likewise.
7856 (build4_stat): Likewise.
7857 (build5_stat): Likewise.
7858 (build_decl_stat): Likewise.
7859 * tree.h (make_node_stat): Adjust prototype.
7860 (make_node): remove macro.
7862 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
7864 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
7865 (PPC_FEATURE2_SCV): Likewise.
7866 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
7868 2017-07-28 Tamar Christina <tamar.christina@arm.com>
7870 * config/aarch64/aarch64.c
7871 (aarch64_internal_mov_immediate): Add new special pattern.
7872 * config/aarch64/aarch64.md (*movdi_aarch64):
7873 Add reg/32bit const mov case.
7875 2017-07-28 Tamar Christina <tamar.christina@arm.com>
7876 Richard Sandiford <richard.sandiford@linaro.org>
7878 * config/aarch64/aarch64.md (mov<mode>): Generalize.
7879 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
7880 Add integer and movi cases.
7881 (movi-split-hf-df-sf split, fp16): New.
7882 (enabled): Added TARGET_FP_F16INST.
7883 * config/aarch64/iterators.md (GPF_HF): New.
7884 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
7886 2017-07-28 Tamar Christina <tamar.christina@arm.com>
7888 * config/aarch64/aarch64.c
7889 (aarch64_simd_container_mode): Add prototype.
7890 (aarch64_expand_mov_immediate): Add HI support.
7891 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
7892 (aarch64_can_const_movi_rtx_p): New.
7893 (aarch64_preferred_reload_class):
7894 Remove restrictions of using FP registers for certain SIMD operations.
7895 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
7896 (aarch64_valid_floating_const): Add integer move validation.
7897 (aarch64_simd_imm_scalar_p): Remove.
7898 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
7899 (aarch64_legitimate_constant_p): Expand list of supported cases.
7900 * config/aarch64/aarch64-protos.h
7901 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
7902 (aarch64_reinterpret_float_as_int): New.
7903 (aarch64_simd_imm_scalar_p): Remove.
7904 * config/aarch64/constraints.md (Uvi): New.
7905 (Dd): Split into Ds and new Dd.
7906 * config/aarch64/aarch64.md (*movsi_aarch64):
7908 (*movdi_aarch64): Add SIMD mov case.
7910 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7912 * tree-predcom.c: (struct chain): Handle store-store chain in which
7913 stores for elimination only store loop invariant values.
7914 (execute_pred_commoning_chain): Ditto.
7915 (prepare_initializers_chain_store_elim): Ditto.
7916 (prepare_finalizers): Ditto.
7917 (is_inv_store_elimination_chain): New function.
7918 (initialize_root_vars_store_elim_1): New function.
7920 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7922 * tree-predcom.c: Revise general description of the pass.
7923 (enum chain_type): New enum type for store elimination.
7924 (struct chain): New field supporting store elimination.
7925 (struct component): Ditto.
7926 (dump_chain): Dump store-stores chain.
7927 (release_chain): Release resources.
7928 (split_data_refs_to_components): Compute and create component
7929 contains only stores for elimination.
7930 (get_chain_last_ref_at): New function.
7931 (make_invariant_chain): Initialization.
7932 (make_rooted_chain): Specify chain type in parameter and record it.
7933 (add_looparound_copies): Skip for store-stores chain.
7934 (determine_roots_comp): Compute type of chain and pass it to
7936 (initialize_root_vars_store_elim_2): New function.
7937 (finalize_eliminated_stores): New function.
7938 (remove_stmt): Handle store for elimination.
7939 (execute_pred_commoning_chain): Execute predictive commoning on
7941 (determine_unroll_factor): Skip unroll for store-stores chain.
7942 (prepare_initializers_chain_store_elim): New function.
7943 (prepare_initializers_chain): Hanlde store-store chain.
7944 (prepare_finalizers_chain, prepare_finalizers): New function.
7945 (tree_predictive_commoning_loop): Return integer value indicating
7946 if loop is unrolled or lcssa form is corrupted.
7947 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
7949 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7951 * tree-predcom.c (initialize_root): Delete.
7952 (execute_pred_commoning_chain): Initialize root vars and replace
7953 reference of non-combined chain directly, rather than call above
7956 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7958 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
7959 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
7961 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7963 * tree-predcom.c (struct chain): New field init_seq.
7964 (release_chain): Release init_seq.
7965 (prepare_initializers_chain): Record intialization stmts in above
7967 (insert_init_seqs): New function.
7968 (tree_predictive_commoning_loop): Call insert_init_seqs.
7970 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7972 * tree-predcom.c (determine_roots_comp): Skip trivial components.
7974 2017-07-28 Richard Biener <rguenther@suse.de>
7976 * match.pd: Remove superfluous :c.
7977 * genmatch.c (simplify::id): Add member.
7978 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
7980 (current_id): New global.
7981 (dt_node::parent): Move from ...
7982 (dt_operand::parent): ... here. Add for_id member.
7983 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
7984 (decision_tree::find_node): Relax order requirement when
7985 merging DT_TRUE nodes to ones inbetween the current simplify
7986 and the one we try to merge with. Add diagnostic whenever
7987 we need to enforce pattern order by not merging.
7988 (decision_tree::insert): Set current_id.
7989 (decision_tree::print_node): Dump parent node and for_id.
7990 (parser::last_id): Add member.
7991 (parser::push_simplify): Assign unique id.
7992 (parser::parser): Initialize last_id.
7994 2017-07-28 Martin Liska <mliska@suse.cz>
7997 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
7998 gimple_build_debug_bind.
8000 2017-07-28 Richard Biener <rguenther@suse.de>
8002 PR tree-optimization/81502
8003 * match.pd: Add pattern combining BIT_INSERT_EXPR with
8005 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
8007 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
8008 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
8009 for BIT_FIELD_REF args.
8010 * fold-const.c (make_bit_field_ref): Likewise.
8011 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
8013 2017-07-28 Jakub Jelinek <jakub@redhat.com>
8016 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
8017 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
8018 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
8019 Or it into SANITIZER_UNDEFINED.
8020 * ubsan.c: Include gimple-fold.h and varasm.h.
8021 (ubsan_expand_ptr_ifn): New function.
8022 (instrument_pointer_overflow): New function.
8023 (maybe_instrument_pointer_overflow): New function.
8024 (instrument_object_size): Formatting fix.
8025 (pass_ubsan::execute): Call instrument_pointer_overflow
8026 and maybe_instrument_pointer_overflow.
8027 * internal-fn.c (expand_UBSAN_PTR): New function.
8028 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
8029 * sanitizer.def (__ubsan_handle_pointer_overflow,
8030 __ubsan_handle_pointer_overflow_abort): New builtins.
8031 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
8032 * internal-fn.def (UBSAN_PTR): New internal function.
8033 * opts.c (sanitizer_opts): Add pointer-overflow.
8034 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
8035 * fold-const.c (build_range_check): Compute pointer range check in
8036 integral type if pointer arithmetics would be needed. Formatting
8039 2017-07-28 Martin Liska <mliska@suse.cz>
8042 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
8043 parameters that are of a variable-length.
8045 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
8047 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
8049 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
8050 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
8051 (CRT_CALL_STATIC_FUNCTION): Likewise.
8052 (ASM_DEFAULT_SPEC): New define.
8053 (ASM_SPEC32): Likewise.
8054 (ASM_SPEC64): Likewise.
8055 (ASM_SPEC_COMMON): Likewise.
8056 (ASM_SPEC): Likewise.
8057 (INVALID_64BIT): Likewise.
8058 (LINK_OS_DEFAULT_SPEC): Likewise.
8059 (LINK_OS_SPEC32): Likewise.
8060 (LINK_OS_SPEC64): Likewise.
8061 (POWERPC_LINUX): Likewise.
8062 (PTRDIFF_TYPE): Likewise.
8063 (RESTORE_FP_PREFIX): Likewise.
8064 (RESTORE_FP_SUFFIX): Likewise.
8065 (SAVE_FP_PREFIX): Likewise.
8066 (SAVE_FP_SUFFIX): Likewise.
8067 (SIZE_TYPE): Likewise.
8068 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
8069 (TARGET_64BIT): Likewise.
8070 (TARGET_64BIT): Likewise.
8071 (TARGET_AIX): Likewise.
8072 (WCHAR_TYPE_SIZE): Likewise.
8073 (WCHAR_TYPE): Undefine.
8074 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
8075 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
8076 (CPP_OS_RTEMS_SPEC): Delete.
8077 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
8078 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
8080 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
8082 2017-07-28 Jakub Jelinek <jakub@redhat.com>
8084 PR tree-optimization/81578
8085 * tree-parloops.c (build_new_reduction): Bail out if
8086 reduction_code isn't one of the standard OpenMP reductions.
8087 Move the details printing after that decision.
8089 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
8091 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
8092 related to reload_in_progress.
8093 (splat_input_operand): Likewise.
8094 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
8096 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
8098 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
8099 (TARGET_INSTANTIATE_DECLS): Likewise.
8100 (legitimate_indexed_address_p): Delete reload_in_progress code.
8101 (rs6000_debug_legitimate_address_p): Likewise.
8102 (rs6000_eliminate_indexed_memrefs): Likewise.
8103 (rs6000_emit_le_vsx_store): Likewise.
8104 (rs6000_emit_move_si_sf_subreg): Likewise.
8105 (rs6000_emit_move): Likewise.
8106 (register_to_reg_type): Likewise.
8107 (rs6000_pre_atomic_barrier): Likewise.
8108 (rs6000_machopic_legitimize_pic_address): Likewise.
8109 (rs6000_allocate_stack_temp): Likewise.
8110 (rs6000_address_for_fpconvert): Likewise.
8111 (rs6000_address_for_altivec): Likewise.
8112 (rs6000_secondary_memory_needed_rtx): Delete function.
8113 (rs6000_check_sdmode): Likewise.
8114 (rs6000_alloc_sdmode_stack_slot): Likewise.
8115 (rs6000_instantiate_decls): Likewise.
8116 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
8117 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
8118 Delete reload_in_progress.
8119 (*vec_reload_and_plus_<mptrsize>): Likewise.
8120 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
8121 (vsx_div_v2di): Likewise.
8122 (vsx_udiv_v2di): Likewise.
8124 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
8126 * config/rs6000/rs6000.opt (mlra): Replace with stub.
8127 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
8128 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
8129 (rs6000_debug_reg_global): Delete print of LRA status.
8130 (rs6000_option_override_internal): Delete dead LRA related code.
8131 (rs6000_lra_p): Delete function.
8132 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
8134 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
8136 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
8137 * config/riscv/rtems.h: New file.
8139 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8140 Sudakshina Das <sudi.das@arm.com>
8142 * config/aarch64/aarch64.md
8143 (define_split for and<mode>3nr_compare): Move
8144 non aarch64_logical_operand to a register.
8145 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
8146 register immediate operand to a register.
8147 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
8149 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
8152 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
8154 2017-07-27 Richard Biener <rguenther@suse.de>
8156 PR tree-optimization/81573
8157 PR tree-optimization/81494
8158 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
8159 multi defuse cycle case.
8161 2017-07-27 Richard Biener <rguenther@suse.de>
8163 PR tree-optimization/81571
8164 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
8167 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
8169 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
8170 earlier and only if MASK_FPU is set. Adjust formatting.
8172 2017-07-27 Martin Liska <mliska@suse.cz>
8174 * opt-functions.awk: Add validation of value of Init.
8175 * optc-gen.awk: Pass new argument.
8177 2017-07-27 Martin Liska <mliska@suse.cz>
8179 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
8180 Fix wrong condition.
8182 2017-07-27 Martin Liska <mliska@suse.cz>
8184 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
8185 BBs and edges seen by autoFDO.
8187 2017-07-27 Richard Biener <rguenther@suse.de>
8189 PR tree-optimization/81502
8190 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
8191 with incompatible but same sized type.
8192 (execute_update_addresses_taken): Likewise.
8194 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
8196 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
8197 flag_tree_loop_vectorize rather than flag_tree_vectorize.
8199 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8202 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
8203 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
8204 Change s_operand to memory_operand.
8206 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
8208 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
8209 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
8210 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
8211 Emit instructions rather than returning an expression. Handle TFmode
8212 and KFmode by casting to TImode.
8213 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
8214 (rs6000_emit_le_vsx_store): Likewise.
8215 * config/rs6000/vsx.md (VSX_TI): New iterator.
8216 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
8217 (*vsx_le_undo_permute_<mode>): Likewise.
8218 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
8219 emit the split sequence.
8220 (*vsx_le_perm_store_<mode>): Likewise.
8222 2017-07-27 Jakub Jelinek <jakub@redhat.com>
8224 PR tree-optimization/81555
8225 PR tree-optimization/81556
8226 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
8227 if true, force CHANGED for the recursive invocation.
8228 (reassociate_bb): Remember original length of ops array, pass
8229 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
8231 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
8232 attributes for noipa attribute. For naked attribute use
8233 lookup_attribute first before lookup_attribute_spec.
8234 * final.c (rest_of_handle_final): Disable IPA RA for functions with
8236 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
8237 for functions with noipa attribute.
8238 (cgraph_externally_visible_p): Return true for functions with noipa
8240 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
8241 for functions with noipa attribute.
8242 * doc/extend.texi: Document noipa function attribute.
8243 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
8244 also for functions with noipa attribute.
8245 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
8247 2017-07-26 Andrew Pinski <apinski@cavium.com>
8249 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
8250 vec_unalign_load_cost and vec_unalign_store_cost.
8252 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
8254 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
8255 -mvsx-small-integer option.
8256 (ISA_3_0_MASKS_IEEE): Likewise.
8257 (OTHER_VSX_VECTOR_MASKS): Likewise.
8258 (POWERPC_MASKS): Likewise.
8259 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
8260 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
8261 code, only testing for DImode being allowed in non-VSX floating
8263 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
8264 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
8266 (rs6000_option_override_internal): Delete -mvsx-small-integer.
8267 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
8268 TARGET_P8_VECTOR test.
8269 (rs6000_secondary_reload_simple_move): Likewise.
8270 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
8271 since TARGET_P9_VECTOR was already tested.
8272 (rs6000_opt_masks): Remove -mvsx-small-integer.
8273 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
8274 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
8276 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
8277 test for TARGET_VEXTRACTUB was used, and that uses
8279 (p9 extract splitter): Likewise.
8280 (vsx_extract_<mode>_di_p9): Likewise.
8281 (vsx_extract_<mode>_store_p9): Likewise.
8282 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
8283 for TARGET_P9_VECTOR was used. Delete code that is now dead with
8284 the elimination of TARGET_VSX_SMALL_INTEGER.
8285 (vsx_extract_<mode>_p8): Likewise.
8286 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
8287 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
8288 (vsx_set_<mode>_p9): Likewise.
8289 (vsx_set_v4sf_p9): Likewise.
8290 (vsx_set_v4sf_p9_zero): Likewise.
8291 (vsx_insert_extract_v4sf_p9): Likewise.
8292 (vsx_insert_extract_v4sf_p9_2): Likewise.
8293 * config/rs6000/rs6000.md (sign extend splitter): Change
8294 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
8295 (floatsi<mode>2_lfiwax_mem): Likewise.
8296 (floatunssi<mode>2_lfiwzx_mem): Likewise.
8297 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
8298 since a test for TARGET_P9_VECTOR was used.
8299 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
8300 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
8301 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
8302 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
8303 TARGET_P8_VECTOR test.
8304 (fix_trunc<mode>si2_stfiwx): Likewise.
8305 (fix_trunc<mode>si2_internal): Likewise.
8306 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
8307 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
8309 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
8310 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
8311 TARGET_P8_VECTOR test.
8312 (fixuns_trunc<mode>si2_stfiwx): Likewise.
8313 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
8314 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
8316 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
8317 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
8318 since a test for TARGET_P9_VECTOR was used.
8319 (splitter for loading small constants): Likewise.
8321 2017-07-26 Andrew Pinski <apinski@cavium.com>
8323 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
8326 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
8329 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
8330 (fp_valid_at): Likewise.
8332 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
8334 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
8335 (qdf24xx_addrcost_table): Likewise.
8336 (cortexa57_tunings): Update to use generic_branch_cost.
8337 (cortexa72_tunings): Likewise.
8338 (cortexa73_tunings): Likewise.
8339 (qdf24xx_tunings): Likewise.
8341 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
8343 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
8344 (thunderx2t99_branch_cost): Likewise.
8345 (cortexa35_tunings): Update to use generic_branch_cost.
8346 (cortexa53_tunings): Likewise.
8347 (cortexa57_tunings): Likewise.
8348 (cortexa72_tunings): Likewise.
8349 (cortexa73_tunings): Likewise.
8350 (thunderx2t99_tunings): Likewise.
8352 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
8354 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
8355 (sparc_option_override): Honour MASK_FSMULD.
8356 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
8357 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
8358 * config/sparc/sparc.opt (mfsmuld): New option.
8359 * doc/invoke.texi (mfsmuld): Document option.
8361 2017-07-26 Marek Polacek <polacek@redhat.com>
8364 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
8366 2017-07-26 Richard Biener <rguenther@suse.de>
8368 * gimple-match-head.c (do_valueize): Return OP if valueize
8370 (get_def): New helper to get at the def stmt of a SSA name
8372 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
8373 do_valueize to get at the def stmt.
8374 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
8376 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
8379 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
8381 2017-07-26 Martin Liska <mliska@suse.cz>
8384 * function.c (expand_function_start): Make expansion of
8385 nonlocal_goto_save_area after parm_birth_insn.
8387 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
8389 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
8390 from all CPU target flags enable members.
8392 2017-07-26 Richard Biener <rguenther@suse.de>
8394 * genmatch.c (dt_simplify::gen): Make iterator vars const.
8395 (decision_tree::gen): Make 'type' const.
8396 (write_predicate): Likewise.
8398 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
8400 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
8401 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
8402 (rs6000_option_override_internal): Likewise.
8403 (rs6000_expand_vector_set): Likewise.
8404 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
8405 (TARGET_UPPER_REGS_SF): Likewise.
8406 (TARGET_UPPER_REGS_DI): Likewise.
8407 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
8408 (TARGET_DIRECT_MOVE_64BIT): Likewise.
8409 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
8410 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
8411 (Splitters for DI constants in Altivec registers): Likewise.
8412 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
8413 (vsx_set_v4sf_p9): Likewise.
8414 (vsx_set_v4sf_p9_zero): Likewise.
8415 (vsx_insert_extract_v4sf_p9): Likewise.
8416 (vsx_insert_extract_v4sf_p9_2): Likewise.
8418 2017-07-25 Carl Love <cel@us.ibm.com>
8420 * doc/extend.texi: Update the built-in documentation file for the
8421 existing built-in functions
8422 vector signed char vec_cnttz (vector signed char);
8423 vector unsigned char vec_cnttz (vector unsigned char);
8424 vector signed short vec_cnttz (vector signed short);
8425 vector unsigned short vec_cnttz (vector unsigned short);
8426 vector signed int vec_cnttz (vector signed int);
8427 vector unsigned int vec_cnttz (vector unsigned int);
8428 vector signed long long vec_cnttz (vector signed long long);
8429 vector unsigned long long vec_cnttz (vector unsigned long long);
8431 2017-07-25 Andrew Pinski <apinski@cavium.com>
8433 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
8434 accesses where the use is for the first operand of a BIT_INSERT.
8436 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
8439 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
8440 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
8442 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
8444 * config/i386/gstabs.h: Delete.
8445 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
8447 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
8449 * config/i386/i386.c (ix86_decompose_address): Do not check for
8450 register RTX when looking at index_reg or base_reg.
8451 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
8453 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
8455 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
8456 to update EH info here.
8458 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
8460 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
8462 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
8464 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
8466 2017-07-25 Torsten Duwe <duwe@suse.de>
8468 * common.opt: Introduce -fpatchable-function-entry
8469 command line option, and its variables function_entry_patch_area_size
8470 and function_entry_patch_area_start.
8471 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
8472 including a two-value parser.
8473 * target.def (print_patchable_function_entry): New target hook.
8474 * targhooks.h (default_print_patchable_function_entry): New function.
8475 * targhooks.c (default_print_patchable_function_entry): Likewise.
8476 * toplev.c (process_options): Switch off IPA-RA if
8477 patchable function entries are being generated.
8478 * varasm.c (assemble_start_function): Look at the
8479 patchable-function-entry command line switch and current
8480 function attributes and maybe generate NOP instructions by
8481 calling the print_patchable_function_entry hook.
8482 * doc/extend.texi: Document patchable_function_entry attribute.
8483 * doc/invoke.texi: Document -fpatchable_function_entry
8484 command line option.
8485 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
8487 * doc/tm.texi: Re-generate.
8489 2017-07-25 Jakub Jelinek <jakub@redhat.com>
8492 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
8493 TARGET_AVX512DQ rather than TARGET_AVX512BW.
8495 2017-07-25 Tamar Christina <tamar.christina@arm.com>
8497 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
8499 2017-07-25 Richard Biener <rguenther@suse.de>
8501 PR tree-optimization/81455
8502 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
8503 not walk in cycles when looking for guards.
8505 2017-07-25 Richard Biener <rguenther@suse.de>
8507 PR tree-optimization/81529
8508 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
8509 when optimizing backedge uses.
8511 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
8513 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
8515 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
8516 to dl_section_ref. On AIX, append an expression to subtract
8517 the size of the section length to dl_section_ref.
8519 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
8521 * configure.ac: If any of the config.* scripts fail, exit 1.
8522 * configure: Regenerate.
8524 2017-07-25 Richard Biener <rguenther@suse.de>
8527 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
8528 of immediate uses, be more verbose on errors.
8530 2017-07-25 Richard Biener <rguenther@suse.de>
8532 PR tree-optimization/81510
8533 * tree-vect-loop.c (vect_is_simple_reduction): When the
8534 reduction stmt is not inside the loop bail out.
8536 2017-07-25 Richard Biener <rguenther@suse.de>
8538 PR tree-optimization/81303
8539 * tree-vect-loop-manip.c (vect_loop_versioning): Build
8540 profitability check against LOOP_VINFO_NITERSM1.
8542 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
8544 * domwalk.c (cmp_bb_postorder): Simplify.
8545 (sort_bbs_postorder): New function. Use it...
8546 (dom_walker::walk): ...here to optimize common cases.
8548 2017-07-25 Martin Liska <mliska@suse.cz>
8551 * ipa-visibility.c (function_and_variable_visibility): Make the
8552 redirection just on target that supports aliasing.
8553 Fix GNU coding style.
8555 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
8558 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
8560 * config/arm/rtems.h: Likewise.
8561 * config/bfin/rtems.h: Likewise.
8562 * config/i386/rtemself.h: Likewise.
8563 * config/lm32/rtems.h: Likewise.
8564 * config/m32c/rtems.h: Likewise.
8565 * config/m68k/rtemself.h: Likewise.
8566 * config/microblaze/rtems.h: Likewise.
8567 * config/mips/rtems.h: Likewise.
8568 * config/moxie/rtems.h: Likewise.
8569 * config/nios2/rtems.h: Likewise.
8570 * config/powerpcspe/rtems.h: Likewise.
8571 * config/rs6000/rtems.h: Likewise.
8572 * config/rtems.h: Likewise.
8573 * config/sh/rtems.h: Likewise.
8574 * config/sh/rtemself.h: Likewise.
8575 * config/sparc/rtemself.h: Likewise.
8577 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
8580 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
8581 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
8582 * tree-ssa-structalias.c (alias_get_name): Same.
8584 2017-07-25 Bin Cheng <bin.cheng@arm.com>
8587 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
8588 instructions if no du chain is found.
8590 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
8592 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
8594 2017-07-25 Richard Biener <rguenther@suse.de>
8597 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
8600 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
8602 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
8604 (ISA_2_7_MASKS_SERVER): Likewise.
8605 (ISA_3_0_MASKS_IEEE): Likewise.
8606 (OTHER_P8_VECTOR_MASKS): Likewise.
8607 (OTHER_VSX_VECTOR_MASKS): Likewise.
8608 (POWERPC_MASKS): Likewise.
8609 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
8610 duplicate list of options.
8611 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
8612 explicit -mupper-regs options.
8613 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
8614 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
8615 alias for -mupper-regs-df.
8616 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
8617 (rs6000_init_hard_regno_mode_ok): Likewise.
8618 (rs6000_option_override_internal): Likewise.
8619 (rs6000_opt_masks): Likewise.
8620 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
8621 options in terms of whether -mvsx or -mpower8-vector was used.
8622 (TARGET_UPPER_REGS_DI): Likewise.
8623 (TARGET_UPPER_REGS_SF): Likewise.
8624 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
8625 -mupper-regs-* options.
8627 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
8629 * passes.c (emergency_dump_function): Print some empty lines and a
8630 header before the RTL dump.
8632 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
8634 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
8636 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
8639 * config/aarch64/aarch64.c (aarch64_classify_symbol):
8640 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
8642 2017-07-24 Carl Love <cel@us.ibm.com>
8644 * config/rs6000/rs6000-c.c: Add support for built-in functions
8645 vector float vec_extract_fp32_from_shorth (vector unsigned short);
8646 vector float vec_extract_fp32_from_shortl (vector unsigned short);
8647 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
8648 vec_extract_fp_from_shortl): Add defines for the two builtins.
8649 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
8650 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
8652 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
8653 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
8654 * doc/extend.texi: Update the built-in documentation file for the
8655 new built-in function.
8657 2017-07-24 Jakub Jelinek <jakub@redhat.com>
8660 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
8662 * doc/generic.texi: Likewise.
8663 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
8664 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
8666 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
8668 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
8669 (aarch64_mls_elt_merge<mode>): Likewise.
8671 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
8673 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
8674 having __cxa_atexit.
8676 2017-07-23 Michael Collison <michael.collison@arm.com>
8678 * config/arm/arm.c (arm_option_override): Deprecate
8679 use of -mstructure-size-boundary.
8680 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
8681 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
8683 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8686 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
8687 Reduce cost estimate for direct moves.
8689 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
8692 * config/i386/i386.c (ix86_option_override_internal): Disable
8693 BMI, BMI2 and TBM instructions for -m16.
8695 2017-07-21 Carl Love <cel@us.ibm.com>
8697 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8698 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
8699 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
8700 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
8701 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
8702 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
8703 VMULOSW): New enum "unspec" values.
8704 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
8705 altivec_vmulosw): New patterns.
8706 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
8707 VMULOSW): Add definitions.
8709 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
8711 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
8712 (qdf24xx): Likewise.
8713 * config/aarch64/aarch64-options-extensions.def (rdma); New.
8714 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
8715 (AARCH64_FL_V8_1): Renumber.
8716 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
8717 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
8718 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
8719 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
8720 rdma to feature modifiers list.
8722 2017-07-21 Yury Gribov <tetra2005@gmail.com>
8725 * ipa-visibility (function_and_variable_visibility): Convert
8726 recursive PLT call to direct call if appropriate.
8728 2017-07-21 Andrew Pinski <apinski@cavium.com>
8730 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
8731 operand 1 to see if the types precision matches.
8732 * fold-const.c (operand_equal_p): Likewise.
8734 2017-07-21 Richard Biener <rguenther@suse.de>
8736 PR tree-optimization/81303
8737 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
8738 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
8739 (vect_peeling_hash_get_lowest_cost): Adjust.
8740 (vect_enhance_data_refs_alignment): Likewise. Use
8741 vect_get_peeling_costs_all_drs to compute the penalty for no
8742 peeling to match up costs.
8744 2017-07-21 Richard Biener <rguenther@suse.de>
8746 PR tree-optimization/81500
8747 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
8748 we didn't identify a reduction path.
8750 2017-07-21 Tom de Vries <tom@codesourcery.com>
8751 Cesar Philippidis <cesar@codesourcery.com>
8753 PR gcov-profile/81442
8754 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
8757 2017-07-21 Tom de Vries <tom@codesourcery.com>
8760 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
8762 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
8763 nvptx_override_options_after_change.
8765 2017-07-21 Ulrich Drepper <drepper@redhat.com>
8767 * dwarf2out.c (output_file_names): Avoid double testing for
8770 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
8772 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
8774 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
8776 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
8778 (try_crossjump_to_edge): Do not punt on partitioned functions.
8780 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
8782 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
8783 Put all BBs reachable only via paths crossing cold region to cold
8785 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
8787 2016-07-21 Richard Biener <rguenther@suse.de>
8789 PR tree-optimization/81303
8790 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
8791 into account prologue and epilogue iterations when raising
8792 min_profitable_iters to sth at least covering one vector iteration.
8794 2017-07-21 Tamar Christina <tamar.christina@arm.com>
8796 * config/arm/arm.c (arm_test_cpu_arch_dat):
8799 2017-07-20 Nathan Sidwell <nathan@acm.org>
8801 Remove TYPE_METHODS.
8802 * tree.h (TYPE_METHODS): Delete.
8803 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
8804 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
8805 (dbxout_type_methods): Scan TYPE_FIELDS.
8806 (dbxout_type): Don't check TYPE_METHODS here.
8807 * function.c (use_register_for_decl): Always ignore register for
8808 class types when not optimizing.
8809 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
8810 * tree.c (free_lang_data_in_type): Stitch out member functions and
8811 templates from TYPE_FIELDS.
8812 (build_distinct_type_copy, verify_type_variant,
8813 verify_type): Member fns are on TYPE_FIELDS.
8814 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
8815 * tree-pretty-print.c (dump_generic_node): Likewise.
8817 2017-07-20 Jakub Jelinek <jakub@redhat.com>
8820 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
8821 V2TImode and V4TImode.
8822 (ix86_expand_vector_extract): Likewise.
8823 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
8824 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
8825 (ssescalarmode): Handle V4TImode and V2TImode.
8826 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
8827 (*vec_extractv2ti, *vec_extractv4ti): New insns.
8828 (VEXTRACTI128_MODE): New mode iterator.
8829 (splitter for *vec_extractv?ti first element): New.
8830 (VEC_INIT_MODE): New mode iterator.
8831 (vec_init<mode>): Consolidate 3 expanders into one using
8832 VEC_INIT_MODE mode iterator.
8834 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
8836 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
8837 non_spilled_static_chain_regno_p.
8839 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
8841 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
8843 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
8845 * bb-reorder.c (connect_traces): Allow copying of blocks within
8848 2017-07-20 Richard Biener <rguenther@suse.de>
8850 * gimple.h (gimple_phi_result): Add gphi * overload.
8851 (gimple_phi_result_ptr): Likewise.
8852 (gimple_phi_arg): Likewise. Adjust index assert to only
8853 allow actual argument accesses rather than all slots available
8855 (gimple_phi_arg_def): Add gphi * overload.
8856 * tree-phinodes.c (make_phi_node): Initialize only actual
8858 (resize_phi_node): Clear memory not covered by old node,
8859 do not initialize excess argument slots.
8860 (reserve_phi_args_for_new_edge): Initialize new argument slot
8863 2017-07-20 Bin Cheng <bin.cheng@arm.com>
8865 PR tree-optimization/81388
8867 2016-07-21 Bin Cheng <bin.cheng@arm.com>
8869 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
8870 by removing computation of may_be_zero.
8872 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
8873 Tom de Vries <tom@codesourcery.com>
8876 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
8877 when gimple level profile disagrees with what RTL expander did.
8879 2017-07-20 Richard Biener <rguenther@suse.de>
8881 PR tree-optimization/61171
8882 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
8883 (vect_analyze_stmt): Add slp instance parameter.
8884 (vectorizable_reduction): Likewise.
8885 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
8886 (vect_is_simple_reduction): Deal with chains not detected
8887 as SLP reduction chain, specifically not properly associated
8888 chains containing a mix of plus/minus.
8889 (get_reduction_op): Remove.
8890 (get_initial_defs_for_reduction): Simplify, pass in whether
8891 this is a reduction chain, pass in the SLP node for the PHIs.
8892 (vect_create_epilog_for_reduction): Get the SLP instance as
8894 (vectorizable_reduction): Get the SLP instance as arg.
8895 During analysis remember the SLP node with the PHIs in the
8896 instance. Simplify getting at the vectorized reduction PHIs.
8897 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
8898 through SLP instance.
8899 (vect_slp_analyze_operations): Likewise.
8900 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
8901 (vect_transform_stmt): Likewise.
8903 2017-07-20 Tom de Vries <tom@codesourcery.com>
8905 PR tree-optimization/81489
8906 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
8907 read of phi arg location to before loop that modifies phi.
8909 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
8911 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
8914 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
8917 * except.c (execute): Fix ordering issue.
8919 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
8921 PR rtl-optimization/81423
8922 * combine.c (make_compound_operation_int): Don't try to optimize
8923 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
8925 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
8927 PR rtl-optimization/81423
8928 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
8929 with a constant that is -1 in the truncated to mode.
8931 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
8933 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
8934 (determine_unlikely_bbs): ... here.
8935 * predict.h (propagate_unlikely_bbs_forward): Declare.
8936 * cfgexpand.c (pass_expand::execute): Use it.
8937 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
8939 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
8940 propagate_unlikely_bbs_forward.
8942 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
8945 * except.c (maybe_add_nop_after_section_switch): New function.
8948 2017-07-19 Tom de Vries <tom@codesourcery.com>
8950 * gimple.h (gimple_phi_set_arg): Make assert more strict.
8952 2017-07-19 Tom de Vries <tom@codesourcery.com>
8954 * gimple.h (gimple_phi_arg): Make assert more strict.
8956 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
8958 * config.gcc (powerpc*-*-*): Add mmintrin.h.
8959 * config/rs6000/mmintrin.h: New file.
8960 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
8962 2017-07-19 Jakub Jelinek <jakub@redhat.com>
8964 PR tree-optimization/81346
8965 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
8967 2017-07-19 Tom de Vries <tom@codesourcery.com>
8969 * config/nvptx/nvptx.md (VECIM): Add V2DI.
8971 2017-07-19 Tom de Vries <tom@codesourcery.com>
8973 * config/nvptx/nvptx-modes.def: Add V2DImode.
8974 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
8975 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
8976 (nvptx_output_mov_insn): Handle lack of mov.b128.
8977 (nvptx_print_operand): Handle 'H' and 'L' codes.
8978 (nvptx_vector_mode_supported): Allow V2DImode.
8979 (nvptx_preferred_simd_mode): New function.
8980 (nvptx_data_alignment): New function.
8981 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
8982 nvptx_preferred_simd_mode.
8983 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
8985 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
8987 2017-07-19 Tom de Vries <tom@codesourcery.com>
8989 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
8990 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
8991 (nvptx_vector_mode_supported): New function. Allow V2SImode.
8992 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
8993 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
8994 (mov<VECIM>_insn): New define_insn.
8995 (define_expand "mov<VECIM>): New define_expand.
8997 2017-07-19 Tom de Vries <tom@codesourcery.com>
8999 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
9001 2017-07-19 Jakub Jelinek <jakub@redhat.com>
9003 PR tree-optimization/81346
9004 * fold-const.h (fold_div_compare, range_check_type): Declare.
9005 * fold-const.c (range_check_type): New function.
9006 (build_range_check): Use range_check_type.
9007 (fold_div_compare): No longer static, rewritten into
9008 a match.pd helper function.
9009 (fold_comparison): Don't call fold_div_compare here.
9010 * match.pd (X / C1 op C2): New optimization using fold_div_compare
9013 2017-07-19 Nathan Sidwell <nathan@acm.org>
9015 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
9016 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
9017 * tree.c (find_decls_types_r, verify_type): Use
9018 TYPE_{MIN,MAX}_VALUE_RAW.
9019 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
9020 (hash_tree): Likewise.
9021 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
9023 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
9026 2017-07-18 Tom de Vries <tom@codesourcery.com>
9029 * omp-expand.c (expand_omp_for_static_chunk): Handle
9030 equal-argument loop exit phi.
9032 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
9035 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
9036 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
9037 operand 2 predicate.
9038 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
9039 operand 2 predicate.
9040 (ror,rol -> rorx splitters): Use const_int_operand as
9041 operand 2 predicate.
9043 2017-06-18 Richard Biener <rguenther@suse.de>
9045 PR tree-optimization/81410
9046 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
9047 the gap in the ! slp_perm SLP case after each group.
9049 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
9052 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
9055 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
9058 * predict.c (set_even_probabilities): Cleanup; do not affect
9059 probabilities that are already known.
9060 (combine_predictions_for_bb): Call even when count is set.
9062 2017-07-18 Nathan Sidwell <nathan@acm.org>
9064 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
9067 2017-07-18 Bin Cheng <bin.cheng@arm.com>
9070 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
9071 optimization for loop niter analysis.
9073 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
9076 * config/avr/avr.c (avr_optimize_casesi): Don't use
9077 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
9079 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
9081 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
9082 body_cost_vec from _vect_peel_extended_info.
9083 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
9084 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
9087 2017-07-18 Bin Cheng <bin.cheng@arm.com>
9089 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
9091 2017-07-18 Richard Biener <rguenther@suse.de>
9093 PR tree-optimization/80620
9094 PR tree-optimization/81403
9095 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
9096 info when re-using a VN table entry.
9098 2017-07-18 Richard Biener <rguenther@suse.de>
9100 PR tree-optimization/81418
9101 * tree-vect-loop.c (vectorizable_reduction): Properly compute
9102 vectype_in. Verify that with lane-reducing reduction operations
9103 we have a single def-use cycle.
9105 2017-07-17 Carl Love <cel@us.ibm.com>
9107 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
9109 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9110 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
9111 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
9112 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
9113 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
9114 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
9115 VMULOSW): New enum "unspec" values.
9116 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
9117 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
9118 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
9119 altivec_vmulosw): New patterns.
9120 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
9121 VMULOSW): Add definitions.
9122 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
9124 * config/alpha/alpha.c: Include predict.h.
9126 2017-07-17 Yury Gribov <tetra2005@gmail.com>
9128 * tree-vrp.c (compare_assert_loc): Fix comparison function
9129 to return predictable results.
9131 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
9133 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
9136 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
9137 * doc/invoke.texi (mexpand-adddi): Update text.
9139 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
9141 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
9142 that also clobbers the CC register. The old expand code is moved
9144 (*arc_clzsi2): ... here.
9145 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
9146 the CC register. The old expand code is moved to ...
9147 (arc_ctzsi2): ... here.
9149 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
9151 * config/arc/arc.opt (mindexed-loads): Use initial value
9152 TARGET_INDEXED_LOADS_DEFAULT.
9153 (mauto-modify-reg): Use initial value
9154 TARGET_AUTO_MODIFY_REG_DEFAULT.
9155 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
9156 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
9157 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
9158 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
9160 2017-07-17 Martin Liska <mliska@suse.cz>
9163 * opts.c (finish_options): Do not allow -fgnu-tm
9164 w/ -fsanitize={kernel-,}address. Say sorry.
9166 2017-07-17 Bin Cheng <bin.cheng@arm.com>
9169 * tree-loop-distribution.c (classify_partition): Only assert on
9170 numer of iterations.
9171 (merge_dep_scc_partitions): Delete prameter. Update function call.
9172 (distribute_loop): Remove code handling loop with unknown niters.
9173 (pass_loop_distribution::execute): Skip loop with unknown niters.
9175 2017-07-17 Bin Cheng <bin.cheng@arm.com>
9178 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
9179 function sort_partitions_by_post_order.
9181 2017-07-17 Bin Cheng <bin.cheng@arm.com>
9183 PR tree-optimization/81374
9184 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
9185 the max index of basic blocks, rather than number of basic blocks.
9187 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
9189 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
9191 (arc_legitimate_pic_operand_p): Likewise.
9192 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
9194 (arc_needs_pcl_p): Likewise.
9195 (arc_legitimate_pc_offset_p): Likewise.
9196 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
9197 function is also used in constrains.md.
9198 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
9199 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
9200 PLUS. Only return true/false in known cases, otherwise assert.
9201 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
9202 is already called in arc_legitimate_constant_p.
9203 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
9205 (LEGITIMATE_PIC_OPERAND_P): Use
9206 arc_raw_symbolic_reference_mentioned_p function.
9207 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
9212 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
9213 Andrew Burgess <andrew.burgess@embecosm.com>
9215 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
9216 (arc_return_address_register): New function.
9217 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
9218 (arc_handle_fndecl_attribute): Add naked attribute.
9219 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
9220 (TARGET_WARN_FUNC_RETURN): Likewise.
9221 (arc_allocate_stack_slots_for_args): New function.
9222 (arc_warn_func_return): Likewise.
9223 (machine_function): Change type fn_type.
9224 (arc_compute_function_type): Consider new naked function type,
9225 change function return type.
9226 (arc_must_save_register): Adapt to handle new
9227 arc_compute_function_type's return type.
9228 (arc_expand_prologue): Likewise.
9229 (arc_expand_epilogue): Likewise.
9230 (arc_return_address_regs): Delete.
9231 (arc_return_address_register): New function.
9232 (arc_epilogue_uses): Use above function.
9233 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
9234 (arc_function_type): Change encoding, add naked type.
9235 (ARC_INTERRUPT_P): Change to handle the new encoding.
9236 (ARC_FAST_INTERRUPT_P): Likewise.
9237 (ARC_NORMAL_P): Define.
9238 (ARC_NAKED_P): Likewise.
9239 (arc_compute_function_type): Delete prototype.
9240 * config/arc/arc.md (in_ret_delay_slot): Use
9241 arc_return_address_register function.
9242 (simple_return): Likewise.
9243 (p_return_i): Likewise.
9245 2017-07-17 Jakub Jelinek <jakub@redhat.com>
9247 PR tree-optimization/81428
9248 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
9249 can't be built for those types.
9251 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
9253 Remove stuff dead since r239246.
9255 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
9256 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
9257 (avr_inform_devices): Remove dead stuff.
9259 2017-07-17 Tamar Christina <tamar.christina@arm.com>
9261 * config/arm/arm_neon.h: Fix softp typo.
9263 2017-07-17 Jakub Jelinek <jakub@redhat.com>
9265 PR tree-optimization/81365
9266 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
9267 aggregate moves onto bb predecessor edges, make sure there are no
9268 loads that could alias the lhs in between the start of bb and the
9271 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
9274 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
9275 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
9276 [LSHIFTRT, outer_code = TRUNCATE]: Same.
9278 2017-07-17 Jakub Jelinek <jakub@redhat.com>
9280 PR tree-optimization/81396
9281 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
9282 (init_symbolic_number): Initialize it to 1.
9283 (perform_symbolic_merge): Add n_ops from both operands into the new
9285 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
9286 without base_addr as useless if they need more than one operation.
9287 (bswap_replace): Handle !bswap case for NULL base_addr.
9289 2017-07-17 Tom de Vries <tom@codesourcery.com>
9292 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
9295 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
9297 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
9298 conditional builtin define __FIX_LEON3FT_B2BST.
9300 2017-07-17 Daniel Cederman <cederman@gaisler.com>
9302 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
9303 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
9306 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
9308 PR rtl-optimization/81424
9309 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
9310 to remove potential trapping from operands if -fnon-call-exceptions.
9312 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9314 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
9315 profile_proability for scalling.
9316 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
9318 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9320 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
9322 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9324 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
9325 fixpoint arithmetics.
9327 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9329 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
9330 fixpoint arithmetics.
9332 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9334 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
9335 fixpoint arithmetics.
9337 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9339 * profile-count.h (profile_probability::from_reg_br_prob_note,
9340 profile_probability::to_reg_br_prob_note): New functions.
9341 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
9342 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
9343 * predict.c (probability_reliable_p): Update.
9344 (edge_probability_reliable_p): Update.
9345 (br_prob_note_reliable_p): Update.
9346 (invert_br_probabilities): Update.
9347 (add_reg_br_prob_note): New function.
9348 (combine_predictions_for_insn): Update.
9349 * asan.c (asan_clear_shadow): Update.
9350 * cfgbuild.c (compute_outgoing_frequencies): Update.
9351 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
9352 (update_br_prob_note): Update.
9353 (rtl_verify_edges): Update.
9354 (purge_dead_edges): Update.
9355 (fixup_reorder_chain): Update.
9356 * emit-rtl.c (try_split): Update.
9357 * ifcvt.c (cond_exec_process_insns): Update.
9358 (cond_exec_process_if_block): Update.
9359 (dead_or_predicable): Update.
9360 * internal-fn.c (expand_addsub_overflow): Update.
9361 (expand_neg_overflow): Update.
9362 (expand_mul_overflow): Update.
9363 * loop-doloop.c (doloop_modify): Update.
9364 * loop-unroll.c (compare_and_jump_seq): Update.
9365 * optabs.c (emit_cmp_and_jump_insn_1): Update.
9366 * predict.h: Update.
9367 * reorg.c (mostly_true_jump): Update.
9369 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
9370 * config/alpha/alpha.c (emit_unlikely_jump): Update.
9371 * config/arc/arc.c: (emit_unlikely_jump): Update.
9372 * config/arm/arm.c: (emit_unlikely_jump): Update.
9373 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
9374 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
9375 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
9376 (ix86_print_operand): Update.
9377 (ix86_split_fp_branch): Update.
9378 (predict_jump): Update.
9379 * config/ia64/ia64.c (ia64_print_operand): Update.
9380 * config/mmix/mmix.c (mmix_print_operand): Update.
9381 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
9382 (rs6000_expand_split_stack_prologue): Update.
9383 * config/rs6000/rs6000.c: Update.
9384 * config/s390/s390.c (s390_expand_vec_strlen): Update.
9385 (s390_expand_vec_movstr): Update.
9386 (s390_expand_cs_tdsi): Update.
9387 (s390_expand_split_stack_prologue): Update.
9388 * config/sh/sh.c (sh_print_operand): Update.
9389 (expand_cbranchsi4): Update.
9390 (expand_cbranchdi4): Update.
9391 * config/sparc/sparc.c (output_v9branch): Update.
9392 * config/spu/spu.c (get_branch_target): Update.
9393 (ea_load_store_inline): Update.
9394 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
9395 * config/tilepro/tilepro.c: Update.
9397 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
9399 * gimplify.c (mostly_copy_tree_r): Revert latest change.
9400 (gimplify_save_expr): Likewise.
9402 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
9404 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
9406 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
9408 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
9410 * timevar.def (TV_IPA_FNSUMMARY): Define.
9412 2017-07-16 Daniel Cederman <cederman@gaisler.com>
9414 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
9415 to back store errata sensitive sequence from being generated.
9416 (sqrtdf2_fix): Likewise.
9418 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
9420 * tree-ssa-threadupdate.c (compute_path_counts,
9421 update_joiner_offpath_counts): Use profile_probability.
9423 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
9426 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
9428 * config/arm/arm-c.c (arm_cpu_builtins): Define
9429 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
9431 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
9433 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9434 array entries to represent __ieee128 versions of the
9435 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
9436 scalar_extract_sig, and scalar_insert_exp built-in functions.
9437 (altivec_resolve_overloaded_builtin): Add special case handling
9438 for the __builtin_scalar_insert_exp function, as represented by
9439 the P9V_BUILTIN_VEC_VSIEDP constant.
9440 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
9441 exponent support for __ieee128 argument.
9442 (VSESQP): Add scalar extract signature support for __ieee128
9444 (VSTDCNQP): Add scalar test negative support for __ieee128
9446 (VSIEQP): Add scalar insert exponent support for __int128 argument
9447 with __ieee128 result.
9448 (VSIEQPF): Add scalar insert exponent support for __ieee128
9449 argument with __ieee128 result.
9450 (VSTDCQP): Add scalar test data class support for __ieee128
9452 (VSTDCNQP): Add overload support for scalar test negative with
9454 (VSTDCQP): Add overload support for scalar test data class
9456 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
9458 (UNSPEC_VSX_SIEXPQP): New constant.
9459 (xsxexpqp): New insn for VSX scalar extract exponent quad
9461 (xsxsigqp): New insn for VSX scalar extract significand quad
9463 (xsiexpqpf): New insn for VSX scalar insert exponent quad
9464 precision with floating point argument.
9465 (xststdcqp): New expand for VSX scalar test data class quad
9467 (xststdcnegqp): New expand for VSX scalar test negative quad
9469 (xststdcqp): New insn to match expansions for VSX scalar test data
9470 class quad precision and VSX scalar test negative quad precision.
9471 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
9472 special case operand checking to enforce that second operand of
9473 VSX scalar test data class with quad precision argument is a 7-bit
9475 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
9476 prototypes and descriptions of __ieee128 versions of
9477 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
9478 scalar_test_data_class, and scalar_test_neg built-in functions.
9480 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9482 PR tree-optimization/81162
9483 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
9484 replace a negate with an add.
9486 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
9488 * doc/invoke.texi (arm/-mcpu): Document +crypto.
9490 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
9492 * config/arm/arm-c.c (arm_cpu_builtins): Define
9493 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
9495 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
9497 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
9498 (armv8-r): Set ARM Cortex-R52 as default CPU.
9499 * config/arm/arm-tables.opt: Regenerate.
9500 * config/arm/arm-tune.md: Regenerate.
9501 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
9503 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
9504 extension for -mcpu=cortex-r52.
9506 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
9508 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
9509 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
9510 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
9511 (fp-armv8): Define it as FP_ARMv8 only.
9512 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
9513 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
9515 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
9516 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
9517 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
9518 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
9519 than TARGET_FPU_ARMV8.
9520 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
9521 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
9522 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
9524 * config/arm/neon.md (neon_vrint): Likewise.
9525 (neon_vcvt): Likewise.
9526 (neon_<fmaxmin_op><mode>): Likewise.
9527 (<fmaxmin><mode>3): Likewise.
9528 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
9529 * config/arm/predicates.md (arm_cond_move_operator): Check against
9530 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
9532 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
9534 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
9537 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9539 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
9540 loop in comment with memset.
9542 2017-07-14 Martin Liska <mliska@suse.cz>
9544 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
9545 * dwarf2out.c (is_java): Remove the function.
9546 (output_pubname): Remove usage of the function.
9547 (lower_bound_default): Remove usage of DW_LANG_Java.
9548 (gen_compile_unit_die): Likewise.
9549 * gcc.c: Remove compiler defaults for .java and .zip files.
9550 * gimple-expr.c (remove_suffix): Change as there's no longer
9551 extension than 4-letter one.
9552 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
9553 (gimplify_save_expr): Likewise.
9554 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
9555 as it's possible even for other languages than Java.
9556 * langhooks.h (struct lang_hooks): Remove Java from a comment.
9557 * lto-opts.c (lto_write_options): Remove reference to Java.
9558 * opts.c (strip_off_ending): Update file extension handling.
9559 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
9560 * tree-eh.c (lower_resx): Likewise.
9561 * tree.c (free_lang_data_in_type): Remove dead code.
9562 (find_decls_types_r): Likewise.
9563 (build_common_builtin_nodes): Remove Java from a comment.
9564 (verify_type): Remove dead code.
9565 * varasm.c (assemble_external): Remove Java from a comment.
9567 2017-07-14 Martin Liska <mliska@suse.cz>
9569 * opts.c (finish_options): Add quotes.
9570 (common_handle_option): Likewise.
9572 2017-07-14 Martin Liska <mliska@suse.cz>
9574 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
9575 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
9577 * dwarf2out.c (lower_bound_default): Do not handle
9579 (gen_compile_unit_die): Likewise.
9580 * gcc.c: Remove default extension binding for GNU Pascal.
9581 * stmt.c: Remove Pascal language from a comment.
9582 * xcoffout.c: Likewise.
9584 2017-07-13 David Malcolm <dmalcolm@redhat.com>
9587 * diagnostic-show-locus.c (fixit_cmp): New function.
9588 (layout::layout): Sort m_fixit_hints.
9589 (column_range::column_range): Assert that the values are valid.
9590 (struct char_span): New struct.
9591 (correction::overwrite): New method.
9592 (struct source_line): New struct.
9593 (line_corrections::add_hint): Add assertions. Reimplement memcpy
9594 calls in terms of classes source_line and char_span, and
9595 correction::overwrite.
9596 (selftest::test_overlapped_fixit_printing_2): New function.
9597 (selftest::diagnostic_show_locus_c_tests): Call it.
9599 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
9601 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
9602 early if there is no lhs.
9604 2017-07-13 Martin Liska <mliska@suse.cz>
9606 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
9607 (gen_reference_type_die): Likewise.
9608 * stor-layout.c: Remove Pascal-related comment.
9610 2017-07-13 Martin Liska <mliska@suse.cz>
9612 * opts.c (finish_options): Add quotes to error messages.
9613 (parse_sanitizer_options): Likewise.
9615 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
9617 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
9619 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
9621 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
9623 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
9625 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
9627 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
9629 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
9632 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
9633 provides the hardware capability bits, define the macro
9634 __BUILTIN_CPU_SUPPORTS__.
9635 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
9636 if GLIBC does not provide the hardware capability bits. Add a
9637 gcc_unreachable call if the built-in cpu function is neither
9638 __builtin_cpu_is nor __builtin_cpu_supports.
9639 (rs6000_get_function_versions_dispatcher): Change the warning
9640 that an old GLIBC is used which does not export the capability
9641 bits to be an error.
9642 * doc/extend.texi (target_clones attribute): Document the
9643 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
9644 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
9645 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
9646 the macros defined by GCC if the newer GLIBC is available.
9648 2017-07-12 Jeff Law <law@redhat.com>
9650 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
9651 remaining includes slightly.
9652 * config/riscv/riscv-builtins.c: Include profile-count.h.
9654 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
9657 * config/avr/avr.c (avr_set_current_function): In diagnostic
9658 messages: Quote keywords and (parts of) identifiers.
9659 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
9662 2017-07-12 Carl Love <cel@us.ibm.com>
9664 * config/rs6000/rs6000-c.c: Add support for built-in functions
9665 vector bool char vec_revb (vector bool char);
9666 vector bool short vec_revb (vector short char);
9667 vector bool int vec_revb (vector bool int);
9668 vector bool long long vec_revb (vector bool long long);
9669 * doc/extend.texi: Update the built-in documentation file for the
9670 new built-in functions.
9672 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9674 * config/s390/s390.md: Remove movcc splitter.
9676 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9678 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
9679 load/store on condition.
9681 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
9684 * config/avr/avr.c (avr_encode_section_info)
9685 [progmem && !TREE_READONLY]: Error if progmem object needs
9688 2017-07-11 Michael Collison <michael.collison@arm.com>
9690 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
9693 2017-07-11 Carl Love <cel@us.ibm.com>
9695 * config/rs6000/rs6000-c.c: Add support for builtins
9696 vector unsigned int vec_parity_lsbb (vector signed int);
9697 vector unsigned int vec_parity_lsbb (vector unsigned int);
9698 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
9699 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
9700 vector unsigned long long vec_parity_lsbb (vector signed long long);
9701 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
9702 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
9703 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
9704 * doc/extend.texi: Update the built-in documentation file for the
9705 new built-in functions.
9707 2017-07-11 David Malcolm <dmalcolm@redhat.com>
9709 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
9710 (layout::m_primary_loc): New field.
9711 (layout::layout): Initialize new field. Move location filtering
9712 logic from here to...
9713 (layout::maybe_add_location_range): ...this new method. Add
9714 support for filtering to just the lines already specified by other
9716 (layout::will_show_line_p): New method.
9717 (gcc_rich_location::add_location_if_nearby): New method.
9718 (selftest::test_add_location_if_nearby): New test function.
9719 (selftest::diagnostic_show_locus_c_tests): Call it.
9720 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
9723 2017-07-11 Tom de Vries <tom@codesourcery.com>
9725 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
9726 (bb_first_real_insn): New function.
9727 (nvptx_single): Add extra initialization of broadcasted condition
9730 2017-07-11 Nathan Sidwell <nathan@acm.org>
9732 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
9734 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
9736 * doc/extend.texi (AVR Function Attributes): Remove weblink to
9737 Binutils doc as TEXI will mess them up.
9738 * doc/invoke.texi (AVR Options): Same here.
9740 2017-07-11 Daniel Cederman <cederman@gaisler.com>
9742 * config/sparc/sparc.opt (mfix-ut700): New option.
9743 (mfix-gr712rc): Likewise.
9744 (sparc_fix_b2bst): New variable.
9745 * doc/invoke.texi (SPARC options): Document them.
9746 (ARM options): Fix warnings.
9747 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
9748 instructions to prevent sequences that can trigger the store-store
9749 errata for certain LEON3FT processors.
9750 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
9751 (sparc_option_override): Set sparc_fix_b2bst appropriately.
9752 * config/sparc/sparc.md (fix_b2bst): New attribute.
9753 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
9755 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
9758 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
9760 (*rsqrtsf2_sse): Ditto.
9762 (div<mode>3): Macroize insn from divdf3 and divsf3
9763 using MODEF mode iterator.
9765 2017-07-10 Martin Sebor <msebor@redhat.com>
9767 PR tree-optimization/80397
9768 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
9769 instead of testing for equality to INTEGER_TYPE.
9771 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
9773 * config.gcc: Remove uclibc from arc target spec.
9775 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
9777 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
9779 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
9782 * lto-wrapper.c (remove_option): New function.
9783 (merge_and_complain): Merge PIC/PIE options more realistically.
9785 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
9787 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
9791 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
9792 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
9793 * config.in: Regenerate.
9794 * configure: Regenerate.
9795 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
9796 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
9797 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
9798 (TARGET_GASISR_PROLOGUES): ...target mask.
9799 * common/config/avr/avr-common.c
9800 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
9801 Set -mgas-isr-prologues.
9802 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
9803 INSERT_PASS_BEFORE for it.
9804 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
9805 * config/avr/avr.c (avr_option_override)
9806 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
9807 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
9808 (avr_attribute_table) <no_gccisr>: Add new function attribute.
9809 (avr_set_current_function) <is_no_gccisr>: Init machine field.
9810 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
9812 (make_avr_pass_pre_proep): New function.
9813 (emit_push_sfr) <treg>: Add argument to function and use it
9815 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
9816 and set machine->gasisr.yes.
9817 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
9818 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
9819 __gcc_isr.n_pushed to .L__stack_usage.
9820 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
9821 (avr_asm_final_postscan_insn): ...this new static function.
9822 * config/avr/avr.h (machine_function)
9823 <is_no_gccisr, use_L__stack_usage>: New fields.
9824 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
9825 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
9826 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
9827 (gasisr, *gasisr): New expander and insn.
9828 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
9829 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
9830 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
9832 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
9834 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
9837 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
9839 Move jump-tables out of .text again.
9842 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
9843 (ASM_OUTPUT_ADDR_VEC): New function.
9844 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
9845 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
9846 INSN_ADDRESSes as asm comment.
9847 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
9848 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
9849 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
9850 * config/avr/avr.md (*tablejump): Adjust comment.
9851 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
9852 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
9854 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
9855 (avr_output_addr_vec): New proto.
9856 (avr_log_t) <insn_addresses>: New field.
9858 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
9861 * config/i386/i386.c (ix86_function_arg_advance): Set
9862 outgoing_args_on_stack to true if there are outgoing arguments
9864 (ix86_function_arg): Likewise.
9865 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
9866 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
9867 * config/i386/i386.h (machine_function): Add
9868 outgoing_args_on_stack.
9870 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
9872 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
9874 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
9876 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
9878 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
9879 (REAL_H): Remove $(MACHMODE_H).
9880 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
9882 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
9883 $(MACHMODE_H) and double-int.h.
9884 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
9886 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
9889 2017-07-07 Andrew Pinski <apinski@cavium.com>
9891 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
9892 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
9894 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
9896 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
9897 Add warning if GCC was not configured to link against a GLIBC that
9898 exports the hardware capability bits.
9899 (make_resolver_func): Make resolver function private and not a
9900 COMDAT function. Create the name with clone_function_name instead
9901 of make_unique_name.
9904 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
9905 correct operand in doing the split.
9907 2017-07-07 Carl Love <cel@us.ibm.com>
9909 * config/rs6000/rs6000-c: Add support for built-in function
9910 vector unsigned short vec_pack_to_short_fp32 (vector float,
9912 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
9913 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
9914 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
9915 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
9916 (convert_4f32_8i16): Add define_expand.
9917 * doc/extend.texi: Update the built-in documentation file for the
9918 new built-in function.
9920 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9922 * config/sparc/m8.md: New file.
9923 * config/sparc/sparc.md: Include m8.md.
9925 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9927 * config/sparc/sparc.opt: New option -mvis4b.
9928 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
9929 (sparc_option_override): Handle VIS4B.
9930 (enum sparc_builtins): Define
9931 SPARC_BUILTIN_DICTUNPACK{8,16,32},
9932 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
9933 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
9934 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
9935 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
9936 (check_constant_argument): New function.
9937 (sparc_vis_init_builtins): Define builtins
9938 __builtin_vis_dictunpack{8,16,32},
9939 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
9940 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
9941 __builtin_vis_fpcmpde{8,16,32}shl and
9942 __builtin_vis_fpcmpur{8,16,32}shl.
9943 (sparc_expand_builtin): Check that the constant operands to
9944 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
9945 constant and in range.
9946 * config/sparc/sparc-c.c (sparc_target_macros): Handle
9948 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
9949 (SPARC_IMM5_P): Likewise.
9950 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
9951 (enabled): Handle vis4b.
9952 (UNSPEC_DICTUNPACK): New unspec.
9953 (UNSPEC_FPCMPSHL): Likewise.
9954 (UNSPEC_FPUCMPSHL): Likewise.
9955 (UNSPEC_FPCMPDESHL): Likewise.
9956 (UNSPEC_FPCMPURSHL): Likewise.
9957 (cpu_feature): New CPU feature `vis4b'.
9958 (dictunpack{8,16,32}): New insns.
9959 (FPCSMODE): New mode iterator.
9960 (fpcscond): New code iterator.
9961 (fpcsucond): Likewise.
9962 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
9963 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
9964 (fpcmpde{8,16,32}{si,di}shl): Likewise.
9965 (fpcmpur{8,16,32}{si,di}shl): Likewise.
9966 * config/sparc/constraints.md: Define constraints `q' for unsigned
9967 2-bit integer constants and `t' for unsigned 5-bit integer
9969 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
9971 (imm5_operand_dictunpack16): Likewise.
9972 (imm5_operand_dictunpack32): Likewise.
9973 (imm2_operand): Likewise.
9974 * doc/invoke.texi (SPARC Options): Document -mvis4b.
9975 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
9976 ditunpack* and fpcmp*shl builtins.
9978 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9980 * config.gcc: Handle m8 in --with-{cpu,tune} options.
9981 * config.in: Add HAVE_AS_SPARC6 define.
9982 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
9984 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
9986 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
9987 (CPP_CPU_SPEC): Handle m8.
9988 (ASM_CPU_SPEC): Likewise.
9989 * config/sparc/sparc-opts.h (enum processor_type): Add
9991 * config/sparc/sparc.c (m8_costs): New struct.
9992 (sparc_option_override): Handle TARGET_CPU_m8.
9993 (sparc32_initialize_trampoline): Likewise.
9994 (sparc64_initialize_trampoline): Likewise.
9995 (sparc_issue_rate): Likewise.
9996 (sparc_register_move_cost): Likewise.
9997 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
9998 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
9999 (ASM_CPU64_DEFAULT_SPEC): Likewise.
10000 (CPP_CPU_SPEC): Handle M8.
10001 (ASM_CPU_SPEC): Likewise.
10002 (AS_M8_FLAG): Define.
10003 * config/sparc/sparc.md: Add m8 to the cpu attribute.
10004 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
10005 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
10007 * configure: Regenerate.
10008 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
10011 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10013 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
10015 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
10016 ("*movdi_insn_sp32"): Do not set v3pipe.
10017 ("*movsi_insn"): Likewise.
10018 ("*movdi_insn_sp64"): Likewise.
10019 ("*movsf_insn"): Likewise.
10020 ("*movdf_insn_sp32"): Likewise.
10021 ("*movdf_insn_sp64"): Likewise.
10022 ("*zero_extendsidi2_insn_sp64"): Likewise.
10023 ("*sign_extendsidi2_insn"): Likewise.
10024 ("*mov<VM32:mode>_insn"): Likewise.
10025 ("*mov<VM64:mode>_insn_sp64"): Likewise.
10026 ("*mov<VM64:mode>_insn_sp32"): Likewise.
10027 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
10028 ("<vlop:code><VL:mode>3"): Likewise.
10029 ("*not_<vlop:code><VL:mode>3"): Likewise.
10030 ("*nand<VL:mode>_vis"): Likewise.
10031 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
10032 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
10033 ("one_cmpl<VL:mode>2"): Likewise.
10034 ("faligndata<VM64:mode>_vis"): Likewise.
10035 ("alignaddrsi_vis"): Likewise.
10036 ("alignaddrdi_vis"): Likweise.
10037 ("alignaddrlsi_vis"): Likewise.
10038 ("alignaddrldi_vis"): Likewise.
10039 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
10040 ("bmaskdi_vis"): Likewise.
10041 ("bmasksi_vis"): Likewise.
10042 ("bshuffle<VM64:mode>_vis"): Likewise.
10043 ("cmask8<P:mode>_vis"): Likewise.
10044 ("cmask16<P:mode>_vis"): Likewise.
10045 ("cmask32<P:mode>_vis"): Likewise.
10046 ("pdistn<P:mode>_vis"): Likewise.
10047 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
10049 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10051 * config/sparc/sparc.md ("subtype"): New insn attribute.
10052 ("*wrgsr_sp64"): Set insn subtype.
10053 ("*rdgsr_sp64"): Likewise.
10054 ("alignaddrsi_vis"): Likewise.
10055 ("alignaddrdi_vis"): Likewise.
10056 ("alignaddrlsi_vis"): Likewise.
10057 ("alignaddrldi_vis"): Likewise.
10058 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
10059 ("fexpand_vis"): Likewise.
10060 ("fpmerge_vis"): Likewise.
10061 ("faligndata<VM64:mode>_vis"): Likewise.
10062 ("bshuffle<VM64:mode>_vis"): Likewise.
10063 ("cmask8<P:mode>_vis"): Likewise.
10064 ("cmask16<P:mode>_vis"): Likewise.
10065 ("cmask32<P:mode>_vis"): Likewise.
10066 ("fchksm16_vis"): Likewise.
10067 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
10068 ("fmean16_vis"): Likewise.
10069 ("fp<plusminus_insn>64_vis"): Likewise.
10070 ("<plusminus_insn>v8qi3"): Likewise.
10071 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
10072 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
10073 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
10074 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
10075 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
10076 ("*movqi_insn"): Likewise.
10077 ("*movhi_insn"): Likewise.
10078 ("*movsi_insn"): Likewise.
10079 ("movsi_pic_gotdata_op"): Likewise.
10080 ("*movdi_insn_sp32"): Likewise.
10081 ("*movdi_insn_sp64"): Likewise.
10082 ("movdi_pic_gotdata_op"): Likewise.
10083 ("*movsf_insn"): Likewise.
10084 ("*movdf_insn_sp32"): Likewise.
10085 ("*movdf_insn_sp64"): Likewise.
10086 ("*zero_extendhisi2_insn"): Likewise.
10087 ("*zero_extendqihi2_insn"): Likewise.
10088 ("*zero_extendqisi2_insn"): Likewise.
10089 ("*zero_extendqidi2_insn"): Likewise.
10090 ("*zero_extendhidi2_insn"): Likewise.
10091 ("*zero_extendsidi2_insn_sp64"): Likewise.
10092 ("ldfsr"): Likewise.
10093 ("prefetch_64"): Likewise.
10094 ("prefetch_32"): Likewise.
10095 ("tie_ld32"): Likewise.
10096 ("tie_ld64"): Likewise.
10097 ("*tldo_ldub_sp32"): Likewise.
10098 ("*tldo_ldub1_sp32"): Likewise.
10099 ("*tldo_ldub2_sp32"): Likewise.
10100 ("*tldo_ldub_sp64"): Likewise.
10101 ("*tldo_ldub1_sp64"): Likewise.
10102 ("*tldo_ldub2_sp64"): Likewise.
10103 ("*tldo_ldub3_sp64"): Likewise.
10104 ("*tldo_lduh_sp32"): Likewise.
10105 ("*tldo_lduh1_sp32"): Likewise.
10106 ("*tldo_lduh_sp64"): Likewise.
10107 ("*tldo_lduh1_sp64"): Likewise.
10108 ("*tldo_lduh2_sp64"): Likewise.
10109 ("*tldo_lduw_sp32"): Likewise.
10110 ("*tldo_lduw_sp64"): Likewise.
10111 ("*tldo_lduw1_sp64"): Likewise.
10112 ("*tldo_ldx_sp64"): Likewise.
10113 ("*mov<VM32:mode>_insn"): Likewise.
10114 ("*mov<VM64:mode>_insn_sp64"): Likewise.
10115 ("*mov<VM64:mode>_insn_sp32"): Likewise.
10117 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10119 * config/sparc/sparc.md ("type"): New insn type viscmp.
10120 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
10122 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
10123 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
10124 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
10125 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
10127 ("n7_vis_logical_11cycle"): Likewise.
10128 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
10129 * config/sparc/niagara2.md ("niag3_vis": Likewise.
10130 * config/sparc/niagara.md ("niag_vis"): Likewise.
10131 * config/sparc/ultra3.md ("us3_fga"): Likewise.
10132 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
10134 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10136 * config/sparc/sparc.md: New instruction type `bmask'.
10137 (bmaskdi_vis): Use the `bmask' type.
10138 (bmasksi_vis): Likewise.
10139 * config/sparc/ultra3.md (us3_array): Likewise.
10140 * config/sparc/niagara7.md (n7_array): Likewise.
10141 * config/sparc/niagara4.md (n4_array): Likewise.
10142 * config/sparc/niagara2.md (niag2_vis): Likewise.
10143 (niag3_vis): Likewise.
10144 * config/sparc/niagara.md (niag_vis): Likewise.
10146 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
10148 * ipa-comdats.c: Remove optimize check from gate.
10149 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
10150 for functions not optimized.
10151 (ipa_fn_summary_read): Skip optimize check.
10152 (ipa_fn_summary_write): Likewise.
10153 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
10155 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
10157 (can_inline_edge_p): Check flag_pcc_struct_return for match.
10158 (check_callers): Give up on caller which is not optimized.
10159 (inline_small_functions): Likewise.
10160 (ipa_inline): Do not give up when not optimizing.
10161 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
10162 away unoptimizes cdtors.
10163 (whole_program_function_and_variable_visibility): Do
10164 ipa_discover_readonly_nonaddressable_vars in LTO mode.
10165 * ipa.c (process_references): Do not check optimize.
10166 (symbol_table::remove_unreachable_nodes): Update optimize check.
10167 (set_writeonly_bit): Update optimize check.
10168 (pass_ipa_cdtor_merge::gate): Do not check optimize.
10169 (pass_ipa_single_use::gate): Remove.
10171 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
10173 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
10174 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
10175 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
10176 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
10177 permute_load, permute_store, adjust_extract, adjust_splat,
10178 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
10179 replace_swap_with_copy, dump_swap_insn_table,
10180 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
10181 recombine_lvx_pattern, recombine_stvx_pattern,
10182 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
10183 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
10184 to file rs6000-p8swap.c.
10185 * config/rs6000/rs6000-p8swap.c: New file.
10186 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
10187 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
10188 and rs6000*-*-* targets.
10190 2017-07-06 David Malcolm <dmalcolm@redhat.com>
10192 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
10194 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
10196 * lto-wrapper.c (merge_and_complain): Do not merge
10197 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
10198 fsigned_zeros, ftrapping_math, fwrapv.
10199 (append_compiler_options): Do not track these options.
10200 (append_linker_options): Likewie
10202 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
10204 * cgraphunit.c (cgraph_node::finalize_function): When
10205 !flag_toplevel_reorde set no_reorder flag.
10206 (varpool_node::finalize_decl): Likewise.
10207 (symbol_table::compile): Drop no toplevel reorder path.
10209 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
10211 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
10212 edges; zero probability is not better than uninitialized.
10214 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
10216 * asan.h (asan_sanitize_allocas_p): Declare.
10217 * asan.c (asan_sanitize_allocas_p): New function.
10218 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
10219 (handle_builtin_alloca): Likewise.
10220 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
10221 if !asan_sanitize_allocas_p.
10222 * params.def (asan-instrument-allocas): Add new option.
10223 * params.h (ASAN_PROTECT_ALLOCAS): Define.
10224 * opts.c (common_handle_option): Disable allocas sanitization for
10227 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
10229 * asan.c: Include gimple-fold.h.
10230 (get_last_alloca_addr): New function.
10231 (handle_builtin_stackrestore): Likewise.
10232 (handle_builtin_alloca): Likewise.
10233 (asan_emit_allocas_unpoison): Likewise.
10234 (get_mem_refs_of_builtin_call): Add new parameter, remove const
10235 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
10236 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
10237 (instrument_builtin_call): Pass gimple iterator to
10238 get_mem_refs_of_builtin_call.
10239 (last_alloca_addr): New global.
10240 * asan.h (asan_emit_allocas_unpoison): Declare.
10241 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
10242 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
10243 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
10244 if function calls alloca.
10245 * gimple-fold.c (replace_call_with_value): Remove static keyword.
10246 * gimple-fold.h (replace_call_with_value): Declare.
10247 * internal-fn.c: Include asan.h.
10248 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
10249 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
10251 2017-07-06 David Malcolm <dmalcolm@redhat.com>
10253 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
10254 (C_SELFTEST_FLAGS): New.
10255 (CPP_SELFTEST_FLAGS): New.
10256 (SELFTEST_DEPS): New, from deps of s-selftest.
10257 (C_SELFTEST_DEPS): New, from deps of s-selftest.
10258 (CPP_SELFTEST_DEPS): New.
10259 (selftest): Add dependency on s-selftest-c++.
10260 (s-selftest): Rename to...
10261 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
10262 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
10263 than SELFTEST_FLAGS.
10264 (selftest-gdb): Rename to...
10265 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
10267 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
10268 (selftest-valgrind): Rename to...
10269 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
10271 (selftest-valgrind): Reintroduce as an alias for
10272 selftest-c-valgrind.
10273 (s-selftest-c++): New.
10274 (selftest-c++-gdb): New.
10275 (selftest-c++-valgrind): New.
10277 2017-07-06 Olivier Hainque <hainque@adacore.com>
10279 * gcc.c (process_command): When deciding if undefined variables
10280 should be ignored when processing specs, accept "gcc -v" as well.
10282 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
10284 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
10285 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
10287 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
10289 * config/arm/arm-cpus.in (armv8-r): Add new entry.
10290 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
10291 * config/arm/arm-tables.opt: Regenerate.
10292 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
10294 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
10296 2017-07-06 Carl Love <cel@us.ibm.com>
10298 * ChangeLog: Clean up from mid air collision
10300 2017-07-06 Carl Love <cel@us.ibm.com>
10302 * config/rs6000/rs6000-c.c: Add support for built-in functions
10303 vector signed int vec_subc (vector signed int, vector signed int);
10304 vector signed __int128 vec_subc (vector signed __int128,
10305 vector signed __int128);
10306 vector unsigned __int128 vec_subc (vector unsigned __int128,
10307 vector unsigned __int128);
10308 vector signed int vec_sube (vector signed int, vector signed int,
10309 vector signed int);
10310 vector unsigned int vec_sube (vector unsigned int,
10311 vector unsigned int,
10312 vector unsigned int);
10313 vector signed __int128 vec_sube (vector signed __int128,
10314 vector signed __int128,
10315 vector signed__int128);
10316 vector unsigned __int128 vec_sube (vector unsigned __int128,
10317 vector unsigned __int128,
10318 vector unsigned __int128);
10319 vector signed int vec_subec (vector signed int, vector signed int,
10320 vector signed int);
10321 vector unsigned int vec_subec (vector unsigned int,
10322 vector unsigned int,
10323 vector unsigned int);
10324 vector signed __int128 vec_subec (vector signed __int128,
10325 vector signed __int128,
10326 vector signed__int128);
10327 vector unsigned __int128 vec_subec (vector unsigned __int128,
10328 vector unsigned __int128,
10329 vector unsigned __int128);
10330 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
10331 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
10332 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
10333 BU_ALTIVEC_OVERLOAD_X definitions.
10334 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
10335 * doc/extend.texi: Update the built-in documentation file for the new
10336 built-in functions.
10338 2017-07-06 David Malcolm <dmalcolm@redhat.com>
10341 * diagnostic-show-locus.c (layout::layout): Use start and finish
10342 spelling location for the start and finish of each range.
10343 * genmatch.c (linemap_client_expand_location_to_spelling_point):
10344 Add unused aspect param.
10345 * input.c (expand_location_1): Add "aspect" param, and use it
10346 to access the correct part of the location.
10347 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
10349 (expand_location_to_spelling_point): Likewise.
10350 (linemap_client_expand_location_to_spelling_point): Add "aspect"
10351 param, and pass it to expand_location_1.
10353 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
10355 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
10356 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
10357 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
10358 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
10359 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
10360 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
10361 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
10362 _mm_maskz_getmant_ss): New intrinsics.
10363 (__builtin_ia32_getexpss128_mask): Changed to ...
10364 __builtin_ia32_getexpss128_round ... this.
10365 (__builtin_ia32_getexpsd128_mask): Changed to ...
10366 __builtin_ia32_getexpsd128_round ... this.
10367 * config/i386/i386-builtin-types.def
10368 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
10369 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
10370 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
10371 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
10372 __builtin_ia32_getmantss_mask_round): New builtins.
10373 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
10374 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
10375 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
10376 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
10377 * config/i386/sse.md
10378 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
10379 avx512f_sgetexp<mode><mask_scalar_name>
10380 <round_saeonly_scalar_name> ... this.
10381 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
10382 %0, %1, %2<round_saeonly_op3>}): Changed to ...
10383 vgetexp<ssescalarmodesuffix>
10384 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
10385 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
10386 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
10387 avx512f_vgetmant<mode><mask_scalar_name>
10388 <round_saeonly_scalar_name> ... this.
10389 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
10390 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
10391 vgetmant<ssescalarmodesuffix>
10392 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
10393 %0<mask_scalar_operand4>, %1, %2
10394 <round_saeonly_scalar_mask_op4>, %3} ... this.
10395 * config/i386/subst.md (mask_scalar_operand4,
10396 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
10397 round_saeonly_scalar_nimm_predicate): New subst attributes.
10399 2017-07-06 Julia Koval <julia.koval@intel.com>
10401 * config/i386/i386.c (ix86_erase_embedded_rounding):
10402 Remove code for old rounding pattern.
10404 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
10406 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
10408 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
10410 * doc/sourcebuild.texi (Test Directives, Variants of
10411 dg-require-support): Add documentation for dg-require-stack-check.
10413 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
10415 * config/i386/subst.md (mask_scalar, round_scalar,
10416 round_saeonly_scalar): New meta-templates.
10417 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
10418 round_scalar_mask_operand3, round_scalar_mask_op3,
10419 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
10420 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
10421 round_saeonly_scalar_constraint,
10422 round_saeonly_scalar_prefix): New subst attribute.
10423 * config/i386/sse.md
10424 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
10425 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
10426 <round_scalar_name> ... this.
10427 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
10428 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
10429 <round_scalar_name> ... this.
10430 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
10431 <sse>_vm<code><mode>3<mask_scalar_name>
10432 <round_saeonly_scalar_name> ... this.
10433 (v<plusminus_mnemonic><ssescalarmodesuffix>
10434 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
10435 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
10436 v<plusminus_mnemonic><ssescalarmodesuffix>
10437 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
10438 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
10439 (v<multdiv_mnemonic><ssescalarmodesuffix>
10440 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
10441 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
10442 v<multdiv_mnemonic><ssescalarmodesuffix>
10443 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
10444 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
10445 (v<maxmin_float><ssescalarmodesuffix>
10446 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
10447 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
10448 v<maxmin_float><ssescalarmodesuffix>
10449 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
10450 %0<mask_scalar_operand3>, %1, %<iptr>2
10451 <round_saeonly_scalar_mask_op3>} ... this.
10453 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
10455 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
10456 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
10458 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
10459 Alan Hayward <alan.hayward@arm.com>
10460 David Sherwood <david.sherwood@arm.com>
10462 * combine.c (simplify_if_then_else): Remove "enum" before
10464 * compare-elim.c (can_eliminate_compare): Likewise.
10465 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
10467 (aarch64_lookup_simd_builtin_type): Likewise.
10468 (aarch64_simd_builtin_type): Likewise.
10469 (aarch64_init_simd_builtin_types): Likewise.
10470 (aarch64_simd_expand_args): Likewise.
10471 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
10473 (aarch64_reverse_mask): Likewise.
10474 (aarch64_simd_emit_reg_reg_move): Likewise.
10475 (aarch64_gen_adjusted_ldpstp): Likewise.
10476 (aarch64_ccmp_mode_to_code): Likewise.
10477 (aarch64_operands_ok_for_ldpstp): Likewise.
10478 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10479 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
10481 (aarch64_min_divisions_for_recip_mul): Likewise.
10482 (aarch64_reassociation_width): Likewise.
10483 (aarch64_get_condition_code_1): Likewise.
10484 (aarch64_simd_emit_reg_reg_move): Likewise.
10485 (aarch64_simd_attr_length_rglist): Likewise.
10486 (aarch64_reverse_mask): Likewise.
10487 (aarch64_operands_ok_for_ldpstp): Likewise.
10488 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10489 (aarch64_gen_adjusted_ldpstp): Likewise.
10490 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
10492 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
10493 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
10494 (arm_lookup_simd_builtin_type): Likewise.
10495 (arm_simd_builtin_type): Likewise.
10496 (arm_init_simd_builtin_types): Likewise.
10497 (arm_expand_builtin_args): Likewise.
10498 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
10499 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
10500 (ft32_setup_incoming_varargs): Likewise.
10501 (ft32_function_arg): Likewise.
10502 (ft32_function_arg_advance): Likewise.
10503 (ft32_pass_by_reference): Likewise.
10504 (ft32_arg_partial_bytes): Likewise.
10505 (ft32_valid_pointer_mode): Likewise.
10506 (ft32_addr_space_pointer_mode): Likewise.
10507 (ft32_addr_space_legitimate_address_p): Likewise.
10508 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
10510 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
10511 (ix86_emit_outlined_ms2sysv_restore): Likewise.
10512 (iamcu_alignment): Likewise.
10513 (canonicalize_vector_int_perm): Likewise.
10514 (ix86_noce_conversion_profitable_p): Likewise.
10515 (ix86_mpx_bound_mode): Likewise.
10516 (ix86_operands_ok_for_move_multiple): Likewise.
10517 * config/microblaze/microblaze-protos.h
10518 (microblaze_expand_conditional_branch_reg): Likewise.
10519 * config/microblaze/microblaze.c
10520 (microblaze_expand_conditional_branch_reg): Likewise.
10521 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
10523 (rs6000_reassociation_width): Likewise.
10524 (rs6000_invalid_binary_op): Likewise.
10525 (fusion_p9_p): Likewise.
10526 (emit_fusion_p9_load): Likewise.
10527 (emit_fusion_p9_store): Likewise.
10528 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
10530 (riscv_hard_regno_mode_ok_p): Likewise.
10531 (riscv_address_insns): Likewise.
10532 (riscv_split_symbol): Likewise.
10533 (riscv_legitimize_move): Likewise.
10534 (riscv_function_value): Likewise.
10535 (riscv_hard_regno_nregs): Likewise.
10536 (riscv_expand_builtin): Likewise.
10537 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
10538 (riscv_build_integer): Likewise.
10539 (riscv_split_integer): Likewise.
10540 (riscv_legitimate_constant_p): Likewise.
10541 (riscv_cannot_force_const_mem): Likewise.
10542 (riscv_regno_mode_ok_for_base_p): Likewise.
10543 (riscv_valid_base_register_p): Likewise.
10544 (riscv_valid_offset_p): Likewise.
10545 (riscv_valid_lo_sum_p): Likewise.
10546 (riscv_classify_address): Likewise.
10547 (riscv_legitimate_address_p): Likewise.
10548 (riscv_address_insns): Likewise.
10549 (riscv_load_store_insns): Likewise.
10550 (riscv_force_binary): Likewise.
10551 (riscv_split_symbol): Likewise.
10552 (riscv_force_address): Likewise.
10553 (riscv_legitimize_address): Likewise.
10554 (riscv_move_integer): Likewise.
10555 (riscv_legitimize_const_move): Likewise.
10556 (riscv_legitimize_move): Likewise.
10557 (riscv_address_cost): Likewise.
10558 (riscv_subword): Likewise.
10559 (riscv_output_move): Likewise.
10560 (riscv_canonicalize_int_order_test): Likewise.
10561 (riscv_emit_int_order_test): Likewise.
10562 (riscv_function_arg_boundary): Likewise.
10563 (riscv_pass_mode_in_fpr_p): Likewise.
10564 (riscv_pass_fpr_single): Likewise.
10565 (riscv_pass_fpr_pair): Likewise.
10566 (riscv_get_arg_info): Likewise.
10567 (riscv_function_arg): Likewise.
10568 (riscv_function_arg_advance): Likewise.
10569 (riscv_arg_partial_bytes): Likewise.
10570 (riscv_function_value): Likewise.
10571 (riscv_pass_by_reference): Likewise.
10572 (riscv_setup_incoming_varargs): Likewise.
10573 (riscv_print_operand): Likewise.
10574 (riscv_elf_select_rtx_section): Likewise.
10575 (riscv_save_restore_reg): Likewise.
10576 (riscv_for_each_saved_reg): Likewise.
10577 (riscv_register_move_cost): Likewise.
10578 (riscv_hard_regno_mode_ok_p): Likewise.
10579 (riscv_hard_regno_nregs): Likewise.
10580 (riscv_class_max_nregs): Likewise.
10581 (riscv_memory_move_cost): Likewise.
10582 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
10583 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
10584 (rl78_addr_space_address_mode): Likewise.
10585 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10587 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
10588 (rs6000_reassociation_width): Likewise.
10589 (rs6000_invalid_binary_op): Likewise.
10590 (fusion_p9_p): Likewise.
10591 (emit_fusion_p9_load): Likewise.
10592 (emit_fusion_p9_store): Likewise.
10593 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
10594 (ok_for_simple_move_operands): Likewise.
10595 (ok_for_simple_move_strict_operands): Likewise.
10596 (ok_for_simple_arith_logic_operands): Likewise.
10597 (visium_legitimize_reload_address): Likewise.
10598 (visium_select_cc_mode): Likewise.
10599 (output_cbranch): Likewise.
10600 (visium_split_double_move): Likewise.
10601 (visium_expand_copysign): Likewise.
10602 (visium_expand_int_cstore): Likewise.
10603 (visium_expand_fp_cstore): Likewise.
10604 * config/visium/visium.c (visium_pass_by_reference): Likewise.
10605 (visium_function_arg): Likewise.
10606 (visium_function_arg_advance): Likewise.
10607 (visium_libcall_value): Likewise.
10608 (visium_setup_incoming_varargs): Likewise.
10609 (visium_legitimate_constant_p): Likewise.
10610 (visium_legitimate_address_p): Likewise.
10611 (visium_legitimize_address): Likewise.
10612 (visium_secondary_reload): Likewise.
10613 (visium_register_move_cost): Likewise.
10614 (visium_memory_move_cost): Likewise.
10615 (prepare_move_operands): Likewise.
10616 (ok_for_simple_move_operands): Likewise.
10617 (ok_for_simple_move_strict_operands): Likewise.
10618 (ok_for_simple_arith_logic_operands): Likewise.
10619 (visium_function_value_1): Likewise.
10620 (rtx_ok_for_offset_p): Likewise.
10621 (visium_legitimize_reload_address): Likewise.
10622 (visium_split_double_move): Likewise.
10623 (visium_expand_copysign): Likewise.
10624 (visium_expand_int_cstore): Likewise.
10625 (visium_expand_fp_cstore): Likewise.
10626 (visium_split_cstore): Likewise.
10627 (visium_select_cc_mode): Likewise.
10628 (visium_split_cbranch): Likewise.
10629 (output_cbranch): Likewise.
10630 (visium_print_operand_address): Likewise.
10631 * expmed.c (flip_storage_order): Likewise.
10632 * expmed.h (emit_cstore): Likewise.
10633 (flip_storage_order): Likewise.
10634 * genrecog.c (validate_pattern): Likewise.
10635 * hsa-gen.c (gen_hsa_addr): Likewise.
10636 * internal-fn.c (expand_arith_overflow): Likewise.
10637 * ira-color.c (allocno_copy_cost_saving): Likewise.
10638 * lra-assigns.c (find_hard_regno_for_1): Likewise.
10639 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
10640 (process_invariant_for_inheritance): Likewise.
10641 * lra-eliminations.c (move_plus_up): Likewise.
10642 * omp-low.c (lower_oacc_reductions): Likewise.
10643 * simplify-rtx.c (simplify_subreg): Likewise.
10644 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
10645 (TARGET_CHKP_BOUND_MODE): Likewise..
10646 * targhooks.c (default_chkp_bound_mode): Likewise.
10647 (default_setup_incoming_vararg_bounds): Likewise.
10648 * targhooks.h (default_chkp_bound_mode): Likewise.
10649 (default_setup_incoming_vararg_bounds): Likewise.
10650 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
10651 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
10652 (have_whole_vector_shift): Likewise.
10653 * tree-vect-stmts.c (vectorizable_load): Likewise.
10654 * doc/tm.texi: Regenerate.
10656 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
10658 Graceful degrade if Binutils PR21472 is not available.
10661 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
10662 .rodata in flash test fails.
10663 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
10664 * confgure: Regenerate.
10665 * config.in: Regenerate.
10666 * config/avr/avr.c (avr_asm_named_section)
10667 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
10668 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
10669 (avr_asm_init_sections): Same.
10671 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
10673 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
10674 (fma<VH:mode>4_intrinsic): Likewise.
10675 (*fmsub<VCVTF:mode>4): Likewise.
10676 (*fmsub<VH:mode>4_intrinsic): Likewise.
10678 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
10681 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
10682 Don't depend on "optimize > 0".
10683 (out_movhi_r_mr, out_movqi_mr_r): Same.
10684 (out_movhi_mr_r, out_movqi_r_mr): Same.
10685 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
10686 io_address_operand on "optimize > 0".
10688 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10690 * tree-loop-distribution.c: Add general explanantion on the pass.
10691 (generate_loops_for_partition): Mark distributed loop.
10692 (pg_add_dependence_edges): New parameter. Handle alias data
10693 dependence specially and record it in the parameter if asked.
10694 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
10695 (init_partition_graph_vertices, add_partition_graph_edge): New.
10696 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
10697 (free_partition_graph_vdata, build_partition_graph): New.
10698 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
10699 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
10700 (data_ref_segment_size, latch_dominated_by_data_ref): New.
10701 (compute_alias_check_pairs, version_loop_by_alias_check): New.
10702 (version_for_distribution_p, finalize_partitions): New.
10703 (distribute_loop): Handle alias data dependence specially. Factor
10704 out loop fusion code as functions and call these functions.
10706 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10708 * tree-loop-distribution.c (classify_partition): New parameter and
10709 better handle reduction statement.
10710 (rdg_build_partitions): Revise comment.
10711 (distribute_loop): Compute statements in all partitions and pass it
10712 to classify_partition.
10714 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10716 * tree-loop-distribution.c (enum partition_type): New.
10717 (struct partition): New field type.
10718 (partition_merge_into): Add parameter. Update partition type.
10719 (data_dep_in_cycle_p, update_type_for_merge): New functions.
10720 (build_rdg_partition_for_vertex): Compute partition type.
10721 (rdg_build_partitions): Dump partition type.
10722 (distribute_loop): Update calls to partition_merge_into.
10724 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10726 * tree-loop-distribution.c (struct ddr_hasher): New.
10727 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
10729 (classify_partition): Call get_data_dependence.
10730 (pg_add_dependence_edges): Ditto.
10731 (distribute_loop): Release data dependence hash table.
10733 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10735 * tree-loop-distribution.c (ref_base_address): Delete.
10736 (similar_memory_accesses): Rename ...
10737 (share_memory_accesses): ... to this. Check if partitions access
10738 the same memory reference.
10739 (distribute_loop): Call share_memory_accesses.
10741 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10743 * tree-loop-distribution.c (struct partition): New field recording
10744 its data reference.
10745 (partition_alloc, partition_free): Init and release data refs.
10746 (partition_merge_into): Merge data refs.
10747 (build_rdg_partition_for_vertex): Collect data refs for partition.
10748 (pg_add_dependence_edges): Change parameters from vector to bitmap.
10750 (distribute_loop): Remve data refs from vertice data of partition
10753 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10755 * tree-loop-distribution.c (params.h): Include header file.
10756 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
10757 (datarefs_vec): New global var.
10758 (create_rdg_vertices): Use datarefs_vec directly.
10759 (free_rdg): Don't free data references.
10760 (build_rdg): Update use. Don't free data references.
10761 (distribute_loop): Compute global variable for data references.
10762 Bail out if there are too many data references.
10764 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10766 * tree-loop-distribution.c (loop_nest): New global var.
10767 (build_rdg): Use loop directly, rather than loop nest.
10768 (pg_add_dependence_edges): Remove loop nest parameter. Use global
10770 (distribute_loop): Compute global variable loop nest. Update use.
10772 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10774 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
10775 (partition_merge_into): New parameter. Dump reason for fusion.
10776 (distribute_loop): Update use of partition_merge_into.
10778 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10780 * tree-loop-distribution.c (bb_top_order_index): New.
10781 (bb_top_order_index_size, bb_top_order_cmp): New.
10782 (stmts_from_loop): Use topological order.
10783 (pass_loop_distribution::execute): Compute and release topological
10784 order for basic blocks.
10786 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10788 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
10791 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10793 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
10794 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
10795 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
10796 * internal-fn.def (LOOP_DIST_ALIAS): New.
10797 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
10798 (fold_loop_internal_call): ... this.
10799 (vect_loop_dist_alias_call): New function.
10800 (set_uid_loop_bbs): Call fold_loop_internal_call.
10801 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
10804 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
10807 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
10808 Require dead FLAGS_REG at the beginning of a peephole.
10810 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
10813 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
10814 arguments in the call to __builtin_ia32_sbb_u32.
10815 (_subborrow_u64): Swap _X and _Y arguments in the call to
10816 __builtin_ia32_sbb_u64.
10818 2017-07-04 Jakub Jelinek <jakub@redhat.com>
10821 * tree-vrp.c (compare_assert_loc): Turn into a function template
10822 with stable template parameter. Only test if a->e is NULL,
10823 !a->e == !b->e has been verified already. Use e == NULL or
10824 e != NULL instead of e or ! e tests. If stable is true, don't use
10825 iterative_hash_expr, on the other side allow a or b or both NULL
10826 and sort the NULLs last.
10827 (process_assert_insertions): Sort using compare_assert_loc<false>
10828 instead of compare_assert_loc, later sort using
10829 compare_assert_loc<true> before calling process_assert_insertions_for
10830 in a loop. Use break instead of continue once seen NULL pointer.
10832 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
10834 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
10835 Cortex-R7 and Cortex-R8 processors.
10837 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
10839 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
10840 uninitialized while src is not.
10842 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
10844 * common/config/arm/arm-common.c: Adjust include path for
10846 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
10847 (arm-cpu.h): Create in build directory. Adjust dependency rules.
10848 (arm-cpu-data.h): Likewise.
10849 (arm-cpu-cdata.h): Likewise.
10850 * config/arm/arm-cpu.h: Delete.
10851 * config/arm/arm-cpu-cdata.h: Delete.
10852 * config/arm/arm-cpu-data.h: Delete.
10854 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
10856 * config/arm/arm-cpus.in (cortex-a55): New.
10857 (cortex-a75): Likewise.
10858 (cortex-a75.cortex-a55): Likewise.
10859 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
10861 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
10862 * config/arm/arm-cpu-cdata.h: Regenerate.
10863 * config/arm/arm-cpu-data.h: Regenerate.
10864 * config/arm/arm-cpu.h: Regenerate.
10865 * config/arm/arm-tables.opt: Regenerate.
10866 * config/arm/arm-tune.md: Regenerate.
10868 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
10870 * haifa-sched.c (sched_create_recovery_edges): Update profile.
10872 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
10874 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
10877 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
10879 PR tree-optimization/81292
10880 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
10881 full_string_p, also call adjust_related_strinfos if the adjustment
10882 is simple, otherwise invalidate related strinfos.
10884 2017-07-04 Martin Liska <mliska@suse.cz>
10887 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
10888 newly created variable as DECL_IGNORED_P.
10890 2017-07-04 Martin Liska <mliska@suse.cz>
10893 * ipa-inline.c (inline_small_functions):
10894 Use xstrdup_for_dump.
10896 2017-07-04 Tom de Vries <tom@codesourcery.com>
10898 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
10900 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
10903 * config/darwin.c (darwin_function_switched_text_sections):
10906 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
10908 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
10910 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
10912 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
10914 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10916 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
10917 min_profitable_iters, and th as inclusive lower bounds.
10918 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
10919 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
10920 for min_profitable_iters and min_profitable_estimate.
10921 (vect_transform_loop): Treat th as an inclusive lower bound.
10922 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
10924 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
10927 * config/darwin.c (darwin_function_switched_text_sections):
10928 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
10929 in two pieces, and suppress the use of buf.
10931 2017-07-03 Nathan Sidwell <nathan@acm.org>
10933 * hash-table.h (hash_table_mod1): Fix indentation.
10935 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10937 PR middle-end/81290
10938 * predict.c (force_edge_cold): Be more careful about propagation
10940 * profile-count.h (profile_probability::guessed,
10941 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
10943 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
10945 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
10947 * doc/invoke.texi (rcpc architecture extension): Document it.
10949 2017-07-03 Richard Biener <rguenther@suse.de>
10951 PR tree-optimization/60510
10952 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
10953 the scalar reduction PHI and use it.
10954 (vectorizable_reduction): Properly guard the single_defuse_cycle
10955 path for non-SLP reduction chains where we cannot use it.
10956 Rework reduc_def/index and vector type deduction. Rework
10957 vector operand gathering during reduction op code-gen.
10958 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
10959 chains dissolve the chain and leave it to non-SLP reduction
10962 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10964 * tree-data-ref.h (dr_alignment): Declare.
10965 * tree-data-ref.c (dr_alignment): New function.
10966 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
10967 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
10969 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
10971 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10973 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
10974 and base_misalignment fields.
10975 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
10976 * tree-data-ref.c: Include builtins.h.
10977 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
10978 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
10979 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
10980 * tree-vect-data-refs.c: Include tree-cfg.h.
10981 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
10982 fields instead of calculating an alignment here.
10983 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
10984 innermost_loop_behavior fields.
10986 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10988 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
10990 (DR_STEP_ALIGNMENT): New macro.
10991 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
10992 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
10993 (create_data_ref): Print it.
10994 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
10995 to tell whether the step preserves vector (mis)alignment.
10996 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
10997 Move the check for an integer step and generalise to all INTEGER_CST.
10998 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
10999 Print the outer step alignment.
11001 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11003 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
11004 with offset_alignment.
11005 (DR_ALIGNED_TO): Delete.
11006 (DR_OFFSET_ALIGNMENT): New macro.
11007 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
11008 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
11009 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
11010 (create_data_ref): Likewise.
11011 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
11012 (vect_analyze_data_refs): Likewise.
11013 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
11014 creating dummy innermost behavior.
11016 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11018 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
11019 with a "innermost_loop_behavior *" and refeence tree.
11020 * tree-data-ref.c (dr_analyze_innermost): Likewise.
11021 (create_data_ref): Update call accordingly.
11022 * tree-predcom.c (find_looparound_phi): Likewise.
11024 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11026 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
11027 fields with dr_wrt_vec_loop.
11028 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
11029 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
11030 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
11031 (vect_dr_behavior): New function.
11032 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
11033 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
11034 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
11035 track whether the step preserves the misalignment.
11036 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
11037 Use vect_dr_behavior.
11038 (vect_setup_realignment): Update call accordingly.
11039 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
11040 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
11041 call to vect_create_addr_base_for_vector_ref.
11042 (vect_create_cond_for_align_checks): Likewise.
11043 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
11044 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
11045 (vect_recog_mask_conversion_pattern): Likewise.
11046 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
11047 (new_stmt_vec_info): Remove redundant zeroing.
11049 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
11051 * common/config/arm/arm-common.c (arm_be8_option): New function.
11052 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
11053 (ISA_ARMv6): Add isa_bit_be8.
11054 * config/arm/arm.h (arm_be8_option): Add prototype.
11055 (BE8_SPEC_FUNCTION): New define.
11056 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
11057 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
11058 (mlittle-endian): Similarly.
11059 (mbe8, mbe32): New options.
11060 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
11061 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
11063 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11065 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
11067 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11069 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
11070 (cleanup_tree_cfg_bb): Use it.
11071 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
11073 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
11075 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11078 * loop-doloop.c (add_test): Update profile.
11080 2017-07-03 Martin Liska <mliska@suse.cz>
11083 * sanopt.c (rewrite_usage_of_param): New function.
11084 (sanitize_rewrite_addressable_params): Likewise.
11085 (pass_sanopt::execute): Call rewrite_usage_of_param.
11087 2017-07-03 Richard Biener <rguenther@suse.de>
11089 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
11090 back to using VIEW_CONVERT_EXPR.
11092 2017-07-03 Martin Liska <mliska@suse.cz>
11095 * doc/extend.texi: Document when a resolver function is
11096 generated for target_clones.
11098 2017-07-03 Martin Liska <mliska@suse.cz>
11100 * asan.c (asan_emit_stack_protection): Unpoison just red zones
11101 and shadow memory of auto variables which are subject of
11102 use-after-scope sanitization.
11103 (asan_expand_mark_ifn): Add do set only when is_poison.
11105 2016-07-03 Richard Biener <rguenther@suse.de>
11107 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
11109 (vect_force_simple_reduction): Record reduction def -> phi mapping.
11110 (vectorizable_reduction): Perform reduction PHI creation when
11111 visiting a reduction PHI and adjust and simplify code generation
11112 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
11113 (vect_transform_loop): Visit reduction PHIs.
11114 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
11115 defs into the SLP tree.
11116 (vect_build_slp_tree): Reduction defs terminate the recursion.
11117 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
11119 (vect_get_vec_defs_for_stmt_copy): Export.
11120 (vect_get_vec_defs): Likewise.
11121 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
11123 (vect_get_vec_defs_for_stmt_copy): Declare.
11124 (vect_get_vec_defs): Likewise.
11126 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11128 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
11129 parameter with a "loop" parameter and use it instead of the
11130 loop containing DR_STMT. Don't check simple_iv when doing
11131 BB analysis. Describe the two analysis modes in the comment.
11133 2017-07-03 Tom de Vries <tom@codesourcery.com>
11135 PR tree-optimization/69468
11136 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
11137 (find_same_succ_bb): Handle ignore_edge_flags.
11139 2017-07-03 Tom de Vries <tom@codesourcery.com>
11141 PR tree-optimization/81192
11142 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
11144 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
11146 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
11148 2017-07-03 Tom de Vries <tom@codesourcery.com>
11150 PR tree-optimization/81192
11151 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
11152 BB_SAME_SUCC (bb) == NULL.
11154 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11156 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
11159 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11161 * dumpfile.c: Include profile-count.h
11162 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
11164 (insert_cond_bb): Update profile.
11165 * tree-cfg.h (insert_cond_bb): Update prototype.
11166 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
11167 * tree-dump.c: Do not include tree-cfg.
11169 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11171 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
11173 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11175 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
11178 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11180 * tree-complex.c (expand_complex_div_wide): update profile.
11182 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
11183 Alan Hayward <alan.hayward@arm.com>
11184 David Sherwood <david.sherwood@arm.com>
11186 * Makefile.in (MACHMODE_H): Remove insn-modes.h
11187 (CORETYPES_H): New define.
11188 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
11189 (insn-modes-inline.h, s-modes-inline-h): New rules.
11190 (generated_files): Add insn-modes-inline.h.
11191 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
11192 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
11193 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
11194 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
11195 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
11196 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
11197 (build/gencodes.o, build/genconditions.o): Likewise.
11198 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
11199 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
11200 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
11201 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
11202 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
11203 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
11204 * coretypes.h: Include everything up to real.h for generators.
11205 Include insn-modes.h first. Include wide-int-print.h after
11206 wide-int.h. Include insn-modes-inline.h and then machmode.h.
11207 * machmode.h: Don't include insn-modes.h here.
11208 * function-tests.c: Remove includes of signop.h, machmode.h,
11209 double-int.h and wide-int.h.
11211 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
11213 * optc-save-gen.awk: Likewise.
11214 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
11215 * godump.c: Remove include of wide-int-print.h.
11216 * pretty-print.h: Likewise.
11217 * wide-int-print.cc: Likewise.
11218 * wide-int.cc: Likewise.
11219 * hash-map-tests.c: Remove include of signop.h.
11220 * hash-set-tests.c: Likewise.
11221 * rtl-tests.c: Likewise.
11222 * mkconfig.sh: Remove include of machmode.h.
11223 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
11225 (emit_insn_modes_inline_h): ...this new function. Emit the code
11226 into an insn-modes-inline.h header file, adding appropriate
11227 include guards and end comments.
11228 (emit_insn_modes_c_header): Remove include of machmode.h.
11229 (emit_min_insn_modes_c_header): Include coretypes.h rather than
11231 (main): Handle -i flag and call emit_insn_modes_inline_h when
11234 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
11236 * tree-ssa-strlen.c (strinfo): Rename the length field to
11237 nonzero_chars. Add a full_string_p field.
11238 (compare_nonzero_chars, zero_length_string_p): New functions.
11239 (get_addr_stridx): Add an offset_out parameter.
11240 Use compare_nonzero_chars.
11241 (get_stridx): Update accordingly. Use compare_nonzero_chars.
11242 (new_strinfo): Update after above changes to strinfo.
11243 (set_endptr_and_length): Set full_string_p.
11244 (get_string_length): Update after above changes to strinfo.
11245 (unshare_strinfo): Update call to new_strinfo.
11246 (maybe_invalidate): Likewise.
11247 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
11248 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
11249 as a uhwi instead of an shwi. Update after above changes to
11250 strinfo and new_strinfo.
11251 (zero_length_string): Assert that chainsi contains full strings.
11252 Use zero_length_string_p. Update call to new_strinfo.
11253 (adjust_related_strinfos): Update after above changes to strinfo.
11254 Copy full_string_p from origsi.
11255 (adjust_last_stmt): Use zero_length_string_p.
11256 (handle_builtin_strlen): Update after above changes to strinfo and
11257 new_strinfo. Install the lhs as the string length if the previous
11258 entry didn't describe a full string.
11259 (handle_builtin_strchr): Update after above changes to strinfo
11261 (handle_builtin_strcpy): Likewise.
11262 (handle_builtin_strcat): Likewise.
11263 (handle_builtin_malloc): Likewise.
11264 (handle_pointer_plus): Likewise.
11265 (handle_builtin_memcpy): Likewise. Track nonzero characters
11266 that aren't necessarily followed by a nul terminator.
11267 (handle_char_store): Likewise.
11269 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
11271 PR tree-optimization/80769
11272 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
11273 for malloc and calloc. Document the new invariant that all related
11274 strinfos have delayed lengths or none do.
11275 (verify_related_strinfos): Move earlier in file.
11276 (set_endptr_and_length): New function, split out from...
11277 (get_string_length): ...here. Also set the lengths of related
11279 (zero_length_string): Assert that chainsi has known (rather than
11281 (adjust_related_strinfos): Likewise.
11283 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
11285 PR tree-optimization/81136
11286 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
11287 assert that two references with the same misalignment have the same
11288 compile-time misalignment if those compile-time misalignments
11291 2017-07-01 Andi Kleen <ak@linux.intel.com>
11293 * print-tree.c (print_node): Print all attributes.
11295 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11297 * cfg.c (scale_bbs_frequencies): New function.
11298 * cfg.h (scale_bbs_frequencies): Declare it.
11299 * cfgloopanal.c (single_likely_exit): Cleanup.
11300 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
11302 (scale_loop_profile): Likewise.
11303 (loop_version): Likewise.
11304 (create_empty_loop_on_edge): Update.
11305 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
11306 scale_loop_frequencies, scale_loop_profile, loopify,
11307 loop_version): Update prototypes.
11308 * modulo-sched.c (sms_schedule): Update.
11309 * predict.c (unlikely_executed_edge_p): Also check probability.
11310 (probably_never_executed_edge_p): Fix typo.
11311 * tree-if-conv.c (version_loop_for_if_conversion): Update.
11312 * tree-parloops.c (gen_parallel_loop): Update.
11313 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
11314 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
11315 * tree-ssa-loop-split.c (split_loop): Update.
11316 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
11317 * tree-vect-loop-manip.c (vect_do_peeling): Update.
11318 (vect_loop_versioning): Update.
11319 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
11321 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11323 * trans-mem.c (split_bb_make_tm_edge): Update profile.
11325 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11327 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
11328 to keep profile consistent.
11330 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11332 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
11333 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
11334 * profile-count.h (max_safe_multiplier): Make unsigned.
11335 (profile_count::guessed_zero): New.
11337 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11339 * bb-reorder.c (fix_up_crossing_landing_pad,
11340 fix_crossing_conditional_branches): Use make_single_succ_edge
11341 to keep profile consistent.
11343 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11345 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
11348 2017-07-01 Jakub Jelinek <jakub@redhat.com>
11351 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
11352 the right scopes, make sure cond_jump isn't preserved between multiple
11353 iterations. Search for fallthru edge whenever there are 3+ edges and
11354 use find_fallthru_edge for it.
11356 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
11358 Patch by Alexander Monakov <amonakov@ispras.ru>
11359 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
11360 probabilities consistently.
11362 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
11364 * pa.c (pa_expand_compare_and_swap_loop): Update call of
11365 emit_cmp_and_jump_insns.
11367 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
11370 * tree-inline.c (expand_call_inline): Combine profile statuses.
11372 2017-06-30 Andrew Pinski <apinski@cavium.com>
11374 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
11375 fold_stmt returned true.
11377 2017-06-30 Nathan Sidwell <nathan@acm.org>
11379 * ggc.h (empty_string): Delete.
11380 * cfgexpand.c (expand_asm_stmt): Use plain "".
11381 * optabs.c (expand_asm_memory_barrier): Likewise.
11382 * stringpool.c (empty_string): Delete.
11383 (digit_vector, digit_string): Delete.
11384 (ggc_alloc_string): Use plain "", don't optimize single digit
11385 strings. Use ggc_alloc_atomic.
11387 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
11389 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
11390 comparison set and one other set, use the cost of the non-comparison
11393 2017-06-30 Nathan Sidwell <nathan@acm.org>
11395 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
11398 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
11400 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
11401 loops. Remove now unneeded calls to gimple_switch_set_label() that
11402 just set removed labels to NULL_TREE.
11404 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
11406 * tree-ssanames.c (set_range_info_raw): Abstract from ...
11407 (set_range_info): ...here. Only call set_range_info_raw if domain
11409 (set_nonzero_bits): Call set_range_info_raw.
11410 * tree-ssanames.h (set_range_info_raw): New.
11412 2017-06-30 Jakub Jelinek <jakub@redhat.com>
11415 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
11416 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
11417 of nonimmediate_operand and <store_mask_constraint> instead of m
11418 for the input operand. For V8FI iterator, always split if input
11419 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
11420 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
11421 <store_mask_predicate> instead of register_operand and
11422 <store_mask_constraint> instead of v for the input operand. Make
11423 sure both operands aren't MEMs for if not <mask_applied>.
11425 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
11427 * lto-wrapper.c (copy_file) Close both file descriptors before
11430 2017-06-30 Martin Liska <mliska@suse.cz>
11433 * multiple_target.c (create_dispatcher_calls): Make ifunc
11434 also for function that don't have calls or are not referenced.
11436 2017-06-30 Richard Biener <rguenther@suse.de>
11438 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
11439 analyze the first scalar stmt. Move vector type computation
11440 for the BB case here from ...
11441 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
11442 live operation processing in the SLP case properly.
11444 2017-06-30 Richard Biener <rguenther@suse.de>
11446 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
11448 2017-06-30 Martin Liska <mliska@suse.cz>
11451 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
11452 before BUILT_IN_UNWIND_RESUME when ASAN is used.
11454 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
11456 * doc/invoke.texi (AArch64): Add missing options and remove redundant
11459 2017-06-30 Richard Biener <rguenther@suse.de>
11461 PR tree-optimization/81249
11462 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
11463 condition reduction result to original scalar type.
11465 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11467 * profile-count.h (enum profile_quality): Fix typos and whitespace
11470 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11472 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
11473 type for branch probabilities.
11475 2017-06-29 Julian Brown <julian@codesourcery.com>
11476 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11478 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
11479 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
11480 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
11481 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
11483 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11485 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
11486 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
11487 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
11488 CC usage from generic code to here.
11489 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
11490 CC usage into the target macros.
11492 2017-06-29 Maya Rashish <coypu@sdf.org>
11494 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
11497 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
11499 * arm/arm-builtins.c: Include profile-count.h
11500 * except.c (sjlj_emit_function_enter): Use
11501 profile_probability::unlikely.
11503 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
11505 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
11506 and tocrel_offset be pointer args rather than implicitly using
11508 (legitimate_constant_pool_address_p, rs6000_emit_move,
11509 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
11510 tocrel_offset and use in toc_relative_expr_p call.
11511 (print_operand, print_operand_address): Use static tocrel_base_oac
11512 and tocrel_offset_oac.
11513 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
11516 2017-06-29 Maya Rashish <coypu@sdf.org>
11518 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
11520 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
11522 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
11523 objects, take into account only the alignment of 'op0' and 'mode1' if
11526 2017-06-29 Steve Ellcey <sellcey@cavium.com>
11528 * ccmp.c (ccmp_tree_comparison_p): New function.
11529 (ccmp_candidate_p): Update to use above function.
11530 (get_compare_parts): New function.
11531 (expand_ccmp_next): Update to use new functions.
11532 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
11534 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
11535 take mode as argument.
11536 * ccmp.h (expand_ccmp_expr): Add mode as argument.
11537 * expr.c (expand_expr_real_1): Pass mode as argument.
11539 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
11541 * combine.c (combine_instructions): Print insns to dump_file, together
11544 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
11546 * asan.c (asan_emit_stack_protection): Update.
11547 (create_cond_insert_point): Update.
11548 * auto-profile.c (afdo_propagate_circuit): Update.
11549 * basic-block.h (struct edge_def): Turn probability to
11550 profile_probability.
11551 (EDGE_FREQUENCY): Update.
11552 * bb-reorder.c (find_traces_1_round): Update.
11553 (better_edge_p): Update.
11554 (sanitize_hot_paths): Update.
11555 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
11556 (make_single_succ_edge): Update.
11557 (check_bb_profile): Update.
11558 (dump_edge_info): Update.
11559 (update_bb_profile_for_threading): Update.
11560 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
11561 probabilitycount to 0.
11562 * cfgbuild.c (compute_outgoing_frequencies): Update.
11563 * cfgcleanup.c (try_forward_edges): Update.
11564 (outgoing_edges_match): Update.
11565 (try_crossjump_to_edge): Update.
11566 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
11567 (expand_gimple_tailcall): Update.
11568 (construct_init_block): Use make_single_succ_edge.
11569 (construct_exit_block): Use make_single_succ_edge.
11570 * cfghooks.c (verify_flow_info): Update.
11571 (redirect_edge_succ_nodup): Update.
11572 (split_edge): Update.
11573 (account_profile_record): Update.
11574 * cfgloopanal.c (single_likely_exit): Update.
11575 * cfgloopmanip.c (scale_loop_profile): Update.
11576 (set_zero_probability): Remove.
11577 (duplicate_loop_to_header_edge): Update.
11578 * cfgloopmanip.h (loop_version): Update prototype.
11579 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
11580 (force_nonfallthru_and_redirect): Update.
11581 (update_br_prob_note): Update.
11582 (rtl_verify_edges): Update.
11583 (purge_dead_edges): Update.
11584 (rtl_lv_add_condition_to_bb): Update.
11585 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
11586 * cgraphunit.c (init_lowered_empty_function): Update.
11587 (cgraph_node::expand_thunk): Update.
11588 * cilk-common.c: Include profile-count.h
11589 * dojump.c (inv): Remove.
11590 (jumpifnot): Update.
11591 (jumpifnot_1): Update.
11592 (do_jump_1): Update.
11594 (do_jump_by_parts_greater_rtx): Update.
11595 (do_compare_rtx_and_jump): Update.
11596 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
11597 do_jump_1. do_compare_rtx_and_jump): Update prototype.
11598 * dwarf2cfi.c: Include profile-count.h
11599 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
11600 (sjlj_emit_dispatch_table): Likewise.
11601 * explow.c: Include profile-count.h
11602 * expmed.c (emit_store_flag_force): Update.
11603 (do_cmp_and_jump): Update.
11604 * expr.c (compare_by_pieces_d::generate): Update.
11605 (compare_by_pieces_d::finish_mode): Update.
11606 (emit_block_move_via_loop): Update.
11607 (store_expr_with_bounds): Update.
11608 (store_constructor): Update.
11609 (expand_expr_real_2): Update.
11610 (expand_expr_real_1): Update.
11611 * expr.h (try_casesi, try_tablejump): Update prototypes.
11612 * gimple-pretty-print.c (dump_probability): Update.
11613 (dump_profile): New.
11614 (dump_gimple_label): Update.
11615 (dump_gimple_bb_header): Update.
11616 * graph.c (draw_cfg_node_succ_edges): Update.
11617 * hsa-gen.c (convert_switch_statements): Update.
11618 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
11619 (find_if_case_1): Update.
11620 (find_if_case_2): Update.
11621 * internal-fn.c (expand_arith_overflow_result_store): Update.
11622 (expand_addsub_overflow): Update.
11623 (expand_neg_overflow): Update.
11624 (expand_mul_overflow): Update.
11625 (expand_vector_ubsan_overflow): Update.
11626 * ipa-cp.c (good_cloning_opportunity_p): Update.
11627 * ipa-split.c (split_function): Use make_single_succ_edge.
11628 * ipa-utils.c (ipa_merge_profiles): Update.
11629 * loop-doloop.c (add_test): Update.
11630 (doloop_modify): Update.
11631 * loop-unroll.c (compare_and_jump_seq): Update.
11632 (unroll_loop_runtime_iterations): Update.
11633 * lra-constraints.c (lra_inheritance): Update.
11634 * lto-streamer-in.c (input_cfg): Update.
11635 * lto-streamer-out.c (output_cfg): Update.
11636 * mcf.c (adjust_cfg_counts): Update.
11637 * modulo-sched.c (sms_schedule): Update.
11638 * omp-expand.c (expand_omp_for_init_counts): Update.
11639 (extract_omp_for_update_vars): Update.
11640 (expand_omp_ordered_sink): Update.
11641 (expand_omp_for_ordered_loops): Update.
11642 (expand_omp_for_generic): Update.
11643 (expand_omp_for_static_nochunk): Update.
11644 (expand_omp_for_static_chunk): Update.
11645 (expand_cilk_for): Update.
11646 (expand_omp_simd): Update.
11647 (expand_omp_taskloop_for_outer): Update.
11648 (expand_omp_taskloop_for_inner): Update.
11649 * omp-simd-clone.c (simd_clone_adjust): Update.
11650 * optabs.c (expand_doubleword_shift): Update.
11651 (expand_abs): Update.
11652 (emit_cmp_and_jump_insn_1): Update.
11653 (expand_compare_and_swap_loop): Update.
11654 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
11655 * predict.c (predictable_edge_p): Update.
11656 (edge_probability_reliable_p): Update.
11657 (set_even_probabilities): Update.
11658 (combine_predictions_for_insn): Update.
11659 (combine_predictions_for_bb): Update.
11660 (propagate_freq): Update.
11661 (estimate_bb_frequencies): Update.
11662 (force_edge_cold): Update.
11663 * profile-count.c (profile_count::dump): Add missing space into dump.
11664 (profile_count::debug): Add newline.
11665 (profile_count::differs_from_p): Explicitly convert to unsigned.
11666 (profile_count::stream_in): Update.
11667 (profile_probability::dump): New member function.
11668 (profile_probability::debug): New member function.
11669 (profile_probability::differs_from_p): New member function.
11670 (profile_probability::differs_lot_from_p): New member function.
11671 (profile_probability::stream_in): New member function.
11672 (profile_probability::stream_out): New member function.
11673 * profile-count.h (profile_count_quality): Rename to ...
11674 (profile_quality): ... this one.
11675 (profile_probability): New.
11676 (profile_count): Update.
11677 * profile.c (compute_branch_probabilities): Update.
11678 * recog.c (peep2_attempt): Update.
11679 * sched-ebb.c (schedule_ebbs): Update.
11680 * sched-rgn.c (find_single_block_region): Update.
11681 (compute_dom_prob_ps): Update.
11682 (schedule_region): Update.
11683 * sel-sched-ir.c (compute_succs_info): Update.
11684 * stmt.c (struct case_node): Update.
11685 (do_jump_if_equal): Update.
11686 (get_outgoing_edge_probs): Update.
11687 (conditional_probability): Update.
11688 (emit_case_dispatch_table): Update.
11689 (expand_case): Update.
11690 (expand_sjlj_dispatch_table): Update.
11691 (emit_case_nodes): Update.
11692 * targhooks.c: Update.
11693 * tracer.c (better_p): Update.
11694 (find_best_successor): Update.
11695 * trans-mem.c (expand_transaction): Update.
11696 * tree-call-cdce.c: Update.
11697 * tree-cfg.c (gimple_split_edge): Upate.
11698 (move_sese_region_to_fn): Upate.
11699 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
11700 * tree-eh.c (lower_resx): Upate.
11701 (cleanup_empty_eh_move_lp): Upate.
11702 * tree-if-conv.c (version_loop_for_if_conversion): Update.
11703 * tree-inline.c (copy_edges_for_bb): Update.
11704 (copy_cfg_body): Update.
11705 * tree-parloops.c (gen_parallel_loop): Update.
11706 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
11707 (gimple_gen_time_profiler): Update.
11708 * tree-ssa-dce.c (remove_dead_stmt): Update.
11709 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
11710 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
11711 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
11712 (unloop_loops): Update.
11713 (try_peel_loop): Update.
11714 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
11715 * tree-ssa-loop-split.c (connect_loops): Update.
11716 (split_loop): Update.
11717 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
11718 (hoist_guard): Update.
11719 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
11720 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
11721 (value_replacement): Update.
11722 * tree-ssa-reassoc.c (branch_fixup): Update.
11723 * tree-ssa-tail-merge.c (replace_block_by): Update.
11724 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
11725 (create_edge_and_update_destination_phis): Update.
11726 (compute_path_counts): Update.
11727 (recompute_probabilities): Update.
11728 (update_joiner_offpath_counts): Update.
11729 (freqs_to_counts_path): Update.
11730 (duplicate_thread_path): Update.
11731 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
11732 (struct switch_conv_info): Update.
11733 (gen_inbound_check): Update.
11734 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
11735 (vect_do_peeling): Update.
11736 (vect_loop_versioning): Update.
11737 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
11738 (optimize_mask_stores): Update.
11739 * ubsan.c (ubsan_expand_null_ifn): Update.
11740 * value-prof.c (gimple_divmod_fixed_value): Update.
11741 (gimple_divmod_fixed_value_transform): Update.
11742 (gimple_mod_pow2): Update.
11743 (gimple_mod_pow2_value_transform): Update.
11744 (gimple_mod_subtract): Update.
11745 (gimple_mod_subtract_transform): Update.
11746 (gimple_ic): Update.
11747 (gimple_stringop_fixed_value): Update.
11748 (gimple_stringops_transform): Update.
11749 * value-prof.h: Update.
11751 2017-06-29 Carl Love <cel@us.ibm.com>
11753 * config/rs6000/rs6000-c.c: Add support for built-in functions
11754 vector signed int vec_signed (vector float);
11755 vector signed long long vec_signed (vector double);
11756 vector signed int vec_signed2 (vector double, vector double);
11757 vector signed int vec_signede (vector double);
11758 vector signed int vec_signedo (vector double);
11759 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
11760 instruction generator.
11761 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
11762 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
11763 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
11765 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
11766 vunsignede_v2df): Add define_expands.
11767 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
11768 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
11769 VEC_UNSIGNEDO): Add definitions.
11770 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
11771 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
11772 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
11773 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
11774 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
11775 * config/rs6000/altivec.h (vec_signed, vec_signed2,
11776 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
11777 vec_unsignede, vec_unsignedo): Add builtin defines.
11778 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
11780 * doc/extend.texi: Update the built-in documentation file for the
11781 new built-in functions.
11783 2017-06-29 Richard Biener <rguenther@suse.de>
11785 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
11786 reduction chains to LOOP_VINFO_REDUCTIONS.
11787 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
11788 SLP reductions after processing reduction chains.
11790 2017-06-29 Nathan Sidwell <nathan@acm.org>
11792 * builtins.c (fold_builtin_FUNCTION): Use
11793 lang_hooks.decl_printable_name.
11795 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
11797 PR middle-end/81194
11798 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
11799 with only one label.
11800 * stmt.c (expand_case): Assert NCASES is greater than one.
11802 2017-06-29 Richard Biener <rguenther@suse.de>
11804 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
11806 (group_case_labels): Likewise.
11807 (find_taken_edge): Push sanity checking on val to workers...
11808 (find_taken_edge_cond_expr): ... here
11809 (find_taken_edge_switch_expr): ... and here, handle cases
11810 with just a default label.
11811 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
11812 (group_case_labels): Likewise.
11813 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
11814 group_case_labels does anything cleanup the CFG again.
11816 2017-06-29 Bin Cheng <bin.cheng@arm.com>
11818 PR tree-optimization/81196
11819 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
11820 exit condition comparing two IVs.
11822 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
11824 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
11825 profile to the dummy entry at the end of the list of architectures.
11826 * config/arm/arm-cpu-cdata.h: Regenerated.
11828 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11829 Michael Collison <michael.collison@arm.com>
11832 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
11834 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
11835 (*aarch64_reg_<mode>3_minus_mask): New pattern.
11836 (*aarch64_<optab>_reg_di3_mask2): New pattern.
11837 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
11838 of shift when the shift amount is masked with constant equal to
11839 the size of the mode.
11840 * config/aarch64/predicates.md (subreg_lowpart_operator): New
11843 2017-06-29 Martin Liska <mliska@suse.cz>
11845 * config/i386/i386.opt: Change range from [1,5] to [0,5].
11847 2017-06-29 Yury Gribov <tetra2005@gmail.com>
11850 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
11853 (edge_growth_cache_entry::edge_growth_cache_entry): New
11855 (reset_edge_growth_cache): Update to use constructor.
11857 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
11859 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
11860 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
11861 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
11863 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
11865 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
11866 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
11868 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
11870 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
11871 (*-linux-uclibc*): Add t-uclibc tmake_file.
11872 * config/t-musl: New.
11873 * config/t-uclibc: New.
11875 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
11877 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
11879 (gen_comm_data): Emit architectural setting of arch_prof.
11880 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
11882 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
11883 (armv8-m.base, armv8-m.main): Likewise.
11884 * arm-protos.h (arm_build_target): Add profile field.
11885 (arch_option): Likewise.
11886 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
11888 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
11889 arm_active_target.profile.
11891 2017-06-28 Richard Biener <rguenther@suse.de>
11893 PR middle-end/81227
11894 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
11895 TYPE_OVERFLOW_WRAPS.
11896 * match.pd (negate_expr_p): Likewise.
11897 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
11898 fold_build2, not fold_binary.
11900 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
11902 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
11903 Convert memory address to Pmode.
11904 (aarch64_print_operand): Assert MEM operands are always Pmode.
11906 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
11909 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
11910 Remove redundant if.
11911 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
11912 * config/arm/aarch-common-protos.h
11913 (aarch_forward_to_shift_is_not_shifted_re): Remove.
11914 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
11916 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
11919 * multiple_target.c (create_dispatcher_calls): Set the default
11920 clone to be static, not public.
11922 2017-06-28 Richard Biener <rguenther@suse.de>
11924 * tree-vect-loop.c (vectorizable_reduction): Move special
11925 cond reduction IV var creation ...
11926 (vect_create_epilog_for_reduction): ... here. Remove induction_index
11927 parameter. Use STMT_VINFO_VECTYPE.
11928 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
11931 2017-06-28 Martin Liska <mliska@suse.cz>
11934 * ipa-visibility.c (non_local_p): Handle visibility.
11936 2017-06-28 Martin Liska <mliska@suse.cz>
11939 * common.opt: Add IntegerRange to various options.
11940 * opt-functions.awk (integer_range_info): New function.
11941 * optc-gen.awk: Add integer_range_info to cl_options struct.
11942 * opts-common.c (decode_cmdline_option): Handle
11943 CL_ERR_INT_RANGE_ARG.
11944 (cmdline_handle_error): Likewise.
11945 * opts.c (print_filtered_help): Show valid interval in
11946 when --help is provided.
11947 * opts.h (struct cl_option): Add range_min and range_max fields.
11948 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
11950 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
11952 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
11953 (x * C EQ/NE y * C): New transformation.
11955 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
11957 * genmultilib (combination_space): Accept '+' in option names.
11959 2017-06-28 Martin Liska <mliska@suse.cz>
11962 * asan.c (instrument_derefs): Bail out inner references
11963 that are hard register variables.
11965 2017-06-28 Jakub Jelinek <jakub@redhat.com>
11968 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
11969 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
11971 2017-06-28 Richard Biener <rguenther@suse.de>
11973 * tree-vectorizer.h (vect_get_vec_defs): Remove.
11974 (vect_get_slp_defs): Adjust.
11975 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
11977 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
11979 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
11980 get_initial_defs_for_reduction instead of vect_get_vec_defs.
11981 (vectorizable_reduction): Adjust.
11982 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
11984 (vect_get_slp_defs): Likewise.
11985 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
11986 (vectorizable_bswap): Adjust.
11987 (vectorizable_call): Likewise.
11988 (vectorizable_conversion): Likewise.
11989 (vectorizable_assignment): Likewise.
11990 (vectorizable_shift): Likewise.
11991 (vectorizable_operation): Likewise.
11992 (vectorizable_store): Likewise.
11993 (vectorizable_condition): Likewise.
11994 (vectorizable_comparison): Likewise.
11996 2017-06-28 Michael Collison <michael.collison@arm.com>
11999 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
12001 (arm_gen_movmemqi): Removed unused variable 'i'.
12002 Convert 'for' loop into 'while' loop.
12003 (arm_expand_prologue): Remove last unnecessary set of insn.
12004 (thumb_pop): Remove unused variable 'pushed_words'.
12005 (thumb_exit): Remove last unnecessary set of regs_to_pop.
12007 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12009 * config/s390/predicates.md: Use s390_rel_address_ok_p.
12010 * config/s390/s390-protos.h: Add prototype of
12011 s390_rel_address_ok_p.
12012 * config/s390/s390.c (s390_got_symbol): New function.
12013 (s390_rel_address_ok_p): New function.
12014 (legitimize_pic_address): Use s390_rel_address_ok_p.
12015 (s390_load_got): Use s390_got_symbol.
12016 (s390_option_override): Issue error if
12017 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
12018 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
12020 * config/s390/s390.opt: New option mpic-data-is-text-relative.
12022 2017-06-27 Andrew Pinski <apinski@cavium.com>
12024 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
12025 (X * copysign (1.0, X)): New pattern.
12026 (X * copysign (1.0, -X)): New pattern.
12027 (copysign (-1.0, CST)): New pattern.
12029 2017-06-27 Joseph Myers <joseph@codesourcery.com>
12031 * genmultilib (combination_space): Remove variable.
12032 Validate reuse rules against regular expression for any sequence
12033 of multilib options in any order.
12035 2017-06-27 Michael Collison <michael.collison@arm.com>
12037 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
12038 call aarch64_split_simd_combine.
12039 * (aarch64_combine_internal<mode>): Delete pattern.
12040 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
12041 Allow register and subreg operands.
12043 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
12045 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
12046 specific need, just fallback on defaults.
12047 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
12049 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
12050 Olivier Hainque <hainque@adacore.com>
12052 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
12054 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
12055 targets. Pick a default if no particular attempt applied.
12056 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
12059 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
12061 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
12062 (x86_64-wrs-vxworks7): Likewise.
12064 2017-06-27 Marek Polacek <polacek@redhat.com>
12067 * ubsan.c (instrument_null): Check get_base_address's result for null.
12069 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
12071 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
12073 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
12075 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
12076 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
12077 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
12078 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
12079 New function types.
12080 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
12081 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
12082 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
12083 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
12084 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
12085 BUILT_IN_FEUPDATEENV): New builtins.
12086 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
12087 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
12088 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
12089 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
12091 (builtin_structptr_types): Adjust size.
12092 * tree.c (builtin_structptr_types): Add four entries.
12094 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
12095 Olivier Hainque <hainque@adacore.com>
12097 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
12098 (TLS_SYM): New local macro, forcing reference to __tls__ on
12099 link command lines for VxWorks 7 RTPs, triggering initialization
12101 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
12102 OS features TLS support, true for RTPs on VxWorks 7.
12103 * config/vxworks.c (vxworks_override_options): Setup emutls
12106 2017-06-27 Jakub Jelinek <jakub@redhat.com>
12108 * predict.c (test_prediction_value_range): Use -1U instead of -1
12109 to avoid narrowing conversion warning.
12110 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
12111 to avoid narrowing conversion warning.
12112 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
12114 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
12116 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
12118 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
12119 64bit configurations.
12120 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
12121 (SIZE_TYPE): Likewise.
12122 * config/vxworks.c (vxworks_emutls_var_fields): Use
12123 long_unsigned_type_node instead of unsigned_type_node as the offset
12124 field type, which is "pointer" mode in emutls.c.
12126 2017-06-27 Jakub Jelinek <jakub@redhat.com>
12129 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
12131 PR middle-end/81207
12132 * gimple-fold.c (replace_call_with_call_and_fold): Handle
12133 gimple_vuse copying separately from gimple_vdef copying.
12135 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
12137 * value-prof.c (free_hist): Remove call to memset and the enclosing if
12140 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
12141 Olivier Hainque <hainque@adacore.com>
12143 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
12144 for all vxworks7 targets.
12145 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
12146 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
12147 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
12148 variations for VX6/VX7 and 32/64bits later on in ...
12149 (VXWORKS_LIB_SPEC): Leverage new macros.
12150 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
12151 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
12153 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
12155 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
12156 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
12158 2017-06-26 Carl Love <cel@us.ibm.com>
12160 * config/rs6000/rs6000-c.c: Add support for built-in functions
12161 vector bool char vec_reve (vector bool char);
12162 vector signed char vec_reve (vector signed char);
12163 vector unsigned char vec_reve (vector unsigned char);
12164 vector bool int vec_reve (vector bool int);
12165 vector signed int vec_reve (vector signed int);
12166 vector unsigned int vec_reve (vector unsigned int);
12167 vector bool long long vec_reve (vector bool long long);
12168 vector signed long long vec_reve (vector signed long long);
12169 vector unsigned long long vec_reve (vector unsigned long long);
12170 vector bool short vec_reve (vector bool short);
12171 vector signed short vec_reve (vector signed short);
12172 vector double vec_reve (vector double);
12173 vector float vec_reve (vector float);
12174 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
12175 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
12176 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
12177 (altivec_vreve): New pattern.
12178 * config/rs6000/altivec.h (vec_reve): New define.
12179 * doc/extend.texi (vec_rev): Update the built-in documentation file
12180 for the new built-in functions.
12182 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12184 PR tree-optimization/71815
12185 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
12187 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
12189 (slsr_process_phi): Likewise.
12190 (replace_uncond_cands_and_profitable_phis): Don't replace a
12191 multiply candidate with a stride of 1 (copy or cast).
12192 (phi_incr_cost): Call uses_consumed_by_stmt rather than
12194 (lowest_cost_path): Likewise.
12195 (total_savings): Likewise.
12197 2017-06-26 Richard Biener <rguenther@suse.de>
12200 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
12201 Use def_builtin_pure for all gather builtins.
12203 2017-06-26 Richard Biener <rguenther@suse.de>
12205 PR tree-optimization/81203
12206 * tree-tailcall.c (find_tail_calls): Do not move stmts into
12207 non-dominating BBs.
12209 2017-06-26 Marek Polacek <polacek@redhat.com>
12212 * doc/invoke.texi: Document -Wmultistatement-macros.
12214 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
12216 * doc/sourcebuild.texi (ARM-specific attributes): Document new
12217 arm_neon_ok_no_float_abi effective target.
12219 2017-06-26 Richard Biener <rguenther@suse.de>
12221 PR tree-optimization/80928
12222 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
12223 (copy_bbs): Set BB_DUPLICATED flag early.
12224 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
12226 (execute_on_shrinking_pred): Likewise.
12227 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
12228 BB_DUPLICATED blocks.
12229 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
12230 iterate over all PHIs considering removal of *gsi.
12232 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
12234 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
12237 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
12239 * config/rs6000/rs6000-string.c: (expand_block_clear,
12240 do_load_for_compare, select_block_compare_mode,
12241 compute_current_alignment, expand_block_compare,
12242 expand_strncmp_align_check, expand_strn_compare,
12243 expand_block_move, rs6000_output_load_multiple)
12244 Move functions related to string/block move/compare
12245 to a separate file.
12246 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
12247 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
12248 for this function which is now used in two files.
12249 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
12250 * config.gcc: Add rs6000-string.o to extra_objs for
12251 targets powerpc*-*-* and rs6000*-*-*.
12253 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
12256 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
12257 32-bit, since indexed is not valid for DImode.
12258 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
12259 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
12260 (define_peephole2 for Altivec d-form load): Add 32-bit support.
12261 (define_peephole2 for Altivec d-form store): Likewise.
12264 * multiple_target.c (create_dispatcher_calls): Only create the
12265 dispatcher call if the function is the default clone of a
12266 versioned function.
12268 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
12270 PR middle-end/80902
12271 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
12272 a call, force the call to not be a tail call.
12274 2017-06-23 Jeff Law <law@redhat.com>
12276 * doc/contrib.texi: Add entry for Steven Pemberton's work on
12279 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
12281 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
12282 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
12283 handling for early expansion of vector shifts (sl,sr,sra,rl).
12284 (builtin_function_type): Add vector shift right instructions
12285 to the unsigned argument list.
12287 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
12289 rtl-optimizatoin/79286
12290 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
12291 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
12292 trap. PIC register plus a const unspec without offset can never trap.
12294 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
12296 * tree.h (builtin_structptr_type): New type.
12297 (builtin_structptr_types): Declare new array.
12298 * tree.c (builtin_structptr_types): New array.
12299 (free_lang_data, build_common_tree_nodes): Use it.
12301 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
12304 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
12307 2017-06-22 Matt Turner <mattst88@gmail.com>
12309 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
12310 Lake models to skylake case. Assume skylake for unknown
12311 models with clflushopt.
12313 2017-06-22 Jeff Law <law@redhat.com>
12315 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
12316 frame sizes that do not satisfy aarch64_uimm12_shift.
12318 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
12320 * profile-count.h (apply_probability,
12321 apply_scale, probability_in): Fix checks for zero.
12323 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12325 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
12326 * doc/cppdiropts.texi (-I @var{dir}): Document it.
12328 2016-06-22 Richard Biener <rguenther@suse.de>
12330 * tree-vect-loop.c (vect_model_reduction_cost): Handle
12331 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
12332 REDUC_MAX_EXPR support.
12333 (vectorizable_reduction): Likewise.
12334 (vect_create_epilog_for_reduction): Likewise.
12336 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
12338 * match.pd (A / (1 << B) -> A >> B): New.
12339 * generic-match-head.c: Include optabs-tree.h.
12340 * gimple-match-head.c: Likewise.
12341 * optabs-tree.h (target_supports_op_p): New.
12342 * optabs-tree.c (target_supports_op_p): New.
12344 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12346 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
12347 $gcc_cv_ld --help output.
12348 (gcc_cv_ld_demangle): Likewise.
12349 (gcc_cv_ld_eh_frame_hdr): Likewise.
12350 (gcc_cv_ld_pie): Likewise.
12351 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
12352 (gcc_cv_ld_buildid): Likewise.
12353 (gcc_cv_ld_sysroot): Likewise.
12354 (ld_bndplt_support): Likewise.
12355 (ld_pushpopstate_support): Likewise.
12356 * configure: Regenerate.
12357 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
12359 2017-06-21 Jakub Jelinek <jakub@redhat.com>
12362 * config/i386/sse.md (round<mode>2): Renumber match_dup and
12363 operands indexes to avoid gap between operands and match_dups.
12365 2017-06-21 Andrew Pinski <apinski@cavium.com>
12367 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
12368 Increment Arith_shift and Arith_shift_reg by 1.
12369 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
12371 * config/aarch64/aarch64.c (thunderx_tunings): Enable
12372 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
12373 (aarch64_strip_extend): Add new argument and test for it.
12374 (aarch64_cheap_mult_shift_p): New function.
12375 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
12376 add a cost if it is true.
12377 Update calls to aarch64_strip_extend.
12378 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
12380 2017-06-21 Andrew Pinski <apinski@cavium.com>
12382 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
12384 (thunderxt88): Likewise.
12385 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
12386 (thunderx_prefetch_tune): New variable.
12387 (thunderx2t99_prefetch_tune): Update for the correct values.
12388 (thunderxt88_tunings): New variable.
12389 (thunderx_tunings): Use thunderx_prefetch_tune instead of
12390 generic_prefetch_tune.
12391 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
12393 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12395 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
12396 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
12397 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
12398 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
12399 (aarch64_atomic_cas<mode>, GPI): Likewise.
12401 2017-06-21 Martin Liska <mliska@suse.cz>
12403 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
12404 statements on cold and hot labels.
12405 * predict.c (tree_estimate_probability_bb): Remove the
12406 prediction from this place.
12408 2017-06-21 Martin Liska <mliska@suse.cz>
12410 PR tree-optimization/79489
12411 * gimplify.c (maybe_add_early_return_predict_stmt): New
12413 (gimplify_return_expr): Call the function.
12414 * predict.c (tree_estimate_probability_bb): Remove handling
12416 * predict.def: Update comment about early return predictor.
12417 * gimple-predict.h (is_gimple_predict): New function.
12418 * predict.def: Change default value of early return to 66.
12419 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
12421 * passes.def: Put pass_strip_predict_hints to the beginning of
12424 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
12426 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
12427 FUNCTION_DECL declarations.
12428 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
12430 (dwaf2out_decl): Likewise.
12431 * godump.c (go_early_global_decl): Skip call to the real debug hook
12432 for FUNCTION_DECL declarations.
12433 * passes.c (rest_of_decl_compilation): Skip call to the
12434 early_global_decl debug hook for FUNCTION_DECL declarations, unless
12435 -fdump-go-spec is passed.
12437 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
12439 * config/i386/i386.c (struct builtin_isa): New field pure_p.
12440 Reorder for compactness.
12441 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
12442 (def_builtin_pure, def_builtin_pure2): New functions.
12443 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
12445 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
12447 * match.pd (nop_convert): New predicate.
12448 ((A +- CST1) +- CST2): Allow some NOP conversions.
12450 2017-06-21 Jakub Jelinek <jakub@redhat.com>
12453 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
12454 with ctors/dtors if GOVD_SHARED is set.
12456 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
12458 * config/aarch64/aarch64.md (movti_aarch64):
12459 Emit mov rather than orr.
12460 (movtf_aarch64): Likewise.
12461 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
12462 Emit mov rather than orr.
12464 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
12466 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
12467 Swap alternatives, make integer dup more expensive.
12469 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
12471 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
12472 Return true for non-tls symbols.
12474 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
12476 * config/aarch64/aarch64-cores.def (cortex-a55): New.
12477 (cortex-a75): Likewise.
12478 (cortex-a75.cortex-a55): Likewise.
12479 * config/aarch64/aarch64-tune.md: Regenerate.
12480 * doc/invoke.texi (-mtune): Document new values for -mtune.
12482 2017-06-21 Tom de Vries <tom@codesourcery.com>
12484 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
12485 stack_size feature.
12486 (Effective-Target Keywords, Other attributes): Suggest using
12487 dg-add-options stack_size feature to get stack limit in stack_size
12488 effective target documentation.
12490 2017-06-21 Julian Brown <julian@codesourcery.com>
12491 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
12493 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
12494 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
12495 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
12497 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
12498 attribute type list for neon_multiply.
12499 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
12500 attribute type list for neon_multiply.
12501 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
12502 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
12503 attribute type list for neon_multiply.
12504 * config/arm/types.md (crypto_pmull): Add.
12505 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
12506 attribute type list.
12508 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
12510 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
12513 2017-06-20 Jakub Jelinek <jakub@redhat.com>
12515 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
12516 to make sure not to dereference a NULL cost_classes_ptr pointer.
12518 2017-06-20 Carl Love <cel@us.ibm.com>
12520 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12521 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
12522 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
12523 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
12524 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
12525 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
12526 VMULOSW): New enum "unspec" values.
12527 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
12528 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
12529 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
12530 altivec_vmulosw): New patterns.
12531 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
12532 VMULOSW): Add definitions.
12534 2017-06-20 Julia Koval <julia.koval@intel.com>
12536 * config/i386/i386.c: Fix rounding expand for new pattern.
12537 * config/i386/subst.md: Fix pattern (parallel -> unspec).
12539 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
12541 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
12542 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
12544 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
12546 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
12549 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
12551 * config/aarch64/aarch64-cores.def: Rearrange to sort by
12552 architecture, then by implementer ID.
12553 * config/aarch64/aarch64-tune.md: Regenerate.
12555 2017-06-20 Richard Biener <rguenther@suse.de>
12557 PR middle-end/81097
12558 * fold-const.c (split_tree): Fold to type before negating.
12560 2017-06-20 David Malcolm <dmalcolm@redhat.com>
12562 * diagnostic-show-locus.c
12563 (selftest::test_fixit_deletion_affecting_newline): New function.
12564 (selftest::diagnostic_show_locus_c_tests): Call it.
12566 2017-06-20 Andreas Schwab <schwab@suse.de>
12569 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
12572 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
12574 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
12575 __ARM_FEATURE_COPROC according to support.
12577 2017-06-20 Jakub Jelinek <jakub@redhat.com>
12579 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
12580 Rewritten to avoid overflow for > 32-bit pointers.
12583 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
12584 by removing enum keyword.
12585 (ubsan_type_descriptor): Likewise. Formatting fix.
12588 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
12589 splitter): Require TARGET_SSE2 in the condition.
12591 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
12594 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
12595 for doing vector set of SFmode on ISA 3.0.
12596 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
12597 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
12599 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
12600 SFmode value into a V4SF variable that was extracted from another
12601 V4SF variable without converting the element to double precision
12602 and back to single precision vector format.
12603 (vsx_insert_extract_v4sf_p9_2): Likewise.
12605 2017-06-19 Jakub Jelinek <jakub@redhat.com>
12607 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
12608 in UWHI to avoid undefined overflow.
12611 * ubsan.h (enum ubsan_encode_value_phase): New.
12612 (ubsan_encode_value): Change second argument to
12613 enum ubsan_encode_value_phase with default value of
12614 UBSAN_ENCODE_VALUE_GENERIC.
12615 * ubsan.c (ubsan_encode_value): Change second argument to
12616 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
12617 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
12618 create_tmp_var_raw instead of create_tmp_var and use a
12620 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
12621 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
12622 ubsan_encode_value callers.
12625 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
12626 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
12627 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
12629 2017-06-19 Richard Biener <rguenther@suse.de>
12631 PR middle-end/81118
12632 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
12633 estimates if we changed anything.
12635 2017-06-19 Richard Biener <rguenther@suse.de>
12637 PR tree-optimization/80887
12638 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
12639 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
12640 simplified lookups, then reset mprts_hook.
12641 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
12643 (try_to_simplify): Likewise.
12645 2017-06-19 Martin Liska <mliska@suse.cz>
12648 * gimplify.c (gimplify_switch_expr):
12649 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
12651 2017-06-19 Martin Liska <mliska@suse.cz>
12653 * doc/install.texi: Document that PGO runs in 4 stages.
12655 2017-06-19 Martin Liska <mliska@suse.cz>
12658 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
12659 to dispatcher function name.
12660 * multiple_target.c (replace_function_decl): New function.
12661 (create_dispatcher_calls): Redirect both edges and references.
12663 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
12665 * profile-count.c (profile_count::dump): Dump quality.
12666 (profile_count::differs_from_p): Update for unsigned val.
12667 * profile-count.h (profile_count_quality): New enum.
12668 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
12670 2017-06-19 Richard Biener <rguenther@suse.de>
12672 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
12673 struct function as arg.
12674 (estimate_numbers_of_iterations): Export overload with loop arg.
12675 (free_numbers_of_iterations_estimates_loop): Use an overload of
12676 free_numbers_of_iterations_estimates instead.
12677 * tree-cfg.c (remove_bb): Adjust.
12678 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
12679 * tree-parloops.c (gen_parallel_loop): Likewise.
12680 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
12682 (tree_unroll_loops_completely): Likewise.
12683 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
12684 Use an overload instead and export.
12685 (estimated_loop_iterations): Adjust.
12686 (max_loop_iterations): Likewise.
12687 (likely_max_loop_iterations): Likewise.
12688 (estimate_numbers_of_iterations): Take struct function as arg
12690 (loop_exits_before_overflow): Adjust.
12691 (free_numbers_of_iterations_estimates_loop): Use an overload.
12692 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
12693 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
12695 2017-06-19 Richard Biener <rguenther@suse.de>
12698 * ipa-prop.c (find_constructor_constant_at_offset): Handle
12699 RANGE_EXPR conservatively.
12701 2017-06-16 Carl Love <cel@us.ibm.com>
12703 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12704 definitions for vec_float, vec_float2, vec_floato,
12705 vec_floate built-ins.
12706 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
12707 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
12709 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
12710 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
12711 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
12712 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
12713 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
12714 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
12715 vec_floato): Add builtin defines.
12716 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
12717 Update the built-in documentation file for the new built-in
12720 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12722 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
12723 (mthumb): Mark as the negative of -marm.
12725 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12727 * doc/invoke.texi (ARM Options, -mcpu): Document supported
12729 (ARM Options, -mtune): Document that this accepts the same
12730 extension options as -mcpu.
12731 (ARM Options, -mfpu): Document addition of -mfpu=auto.
12733 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12735 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
12736 permitted extensions.
12738 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12740 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
12741 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
12742 (armv8-m.main): Add option +nodsp.
12743 * config/arm/arm-cpu-cdata.h: Regenerated.
12745 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12747 * config/arm/t-fuchsia: New file.
12748 * config.gcc (arm*-*-fuchsia*): Use it.
12750 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12752 * config/arm/t-symbian: Rewrite for new option infrastructure.
12754 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12756 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
12757 (MULTILIB_REQUIRED): Likewise.
12759 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12761 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
12762 (MULTILIB_RESUE): Likewise.
12763 (MULTILIB_MATCHES): Likewise.
12764 (MULTLIB_REQUIRED): Likewise.
12766 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12768 * config/arm/t-rtems: Rewrite for new option framework.
12770 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12772 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
12773 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
12774 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
12775 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
12776 * config/arm/t-multilib: ... here.
12777 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
12778 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
12779 armv7-a and armv8*-a when A-profile libraries have not been built.
12780 * config/arm/t-rmprofile: Rewrite.
12782 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12784 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
12785 with a backslash. Remove the backslash after substituting unescaped
12787 * doc/fragments.texi (MULTILIB_REUSE): Document it.
12789 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12791 * config.gcc: (arm*-*-*): When building a-profile libraries, force
12792 the driver to pass through the default setting of -mfloat-abi.
12793 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
12795 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
12796 (all_feat_combs): New rule.
12797 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
12799 * config/arm/t-aprofile: Rewrite.
12801 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12803 * config/arm/arm.h (FPUTYPE_AUTO): Define.
12804 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
12805 fpu is not specified by the user/command-line.
12806 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
12807 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
12808 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
12809 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
12810 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
12811 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
12813 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12815 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
12816 * config/arm/t-arm-elf: Rewritten.
12818 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12820 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
12821 have some floating-point instructions.
12822 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
12823 (TARGET_MAYBE_HARD_FLOAT): New macro.
12824 * config/arm/arm-builtins.c (arm_init_builtins): Use
12825 TARGET_MAYBE_HARD_FLOAT.
12826 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
12828 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12830 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
12831 (configargs.h): Include it.
12832 (arm_print_hint_for_fpu_option): New function.
12833 (arm_parse_fpu_option): New function.
12834 (candidate_extension): New class.
12835 (arm_canon_for_multilib): New function.
12836 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
12837 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
12838 (ARCH_CANONICAL_SPECS): New macro.
12839 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
12841 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12843 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
12844 are set after handling multilib fragments. Set target_cpu_default2
12847 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12849 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
12851 (arm*-*-*): Set target_cpu_default2 to a quoted string.
12852 * config/arm/parsecpu.awk (check_cpu): Validate any extension
12854 (check_arch): Likewise.
12855 * config/arm/arm.c (arm_configure_build_target): Handle
12856 TARGET_CPU_DEFAULT being a string constant. Scan any feature
12857 options in the default.
12859 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12861 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
12862 when an option is an alias of another.
12863 * config/arm/parsecpu.awk (optalias): New parser token.
12864 (gen_comm_data): Mark non-alias options as such. Emit entries
12865 for extension aliases.
12866 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
12867 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
12868 (armv6kz, armv6zk, armv6t2): Likewise.
12869 (armv7): Make vfpv3-d16 an alias.
12870 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
12872 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
12873 Sort in canonical order.
12874 (armv8-a): Sort in canonical order.
12875 (armv8.1-a, armv8.2-a): Likewise.
12876 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
12878 (cortex-a9): Sort in canonical order.
12879 * config/arm/arm.c (selftests.h): Include it.
12880 (arm_test_cpu_arch_data): New function.
12881 (arm_run_self_tests): New function.
12882 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
12883 (targetm): Move declaration to the end of the file.
12884 * arm-cpu-cdata.h: Regenerated.
12886 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12888 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
12889 call to target_mode_check describing the type of option passed.
12890 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
12891 (arm_target_thumb_only): Use arm_parse_arch_option_name or
12892 arm_parse_cpu_option_name to match parameters against list of
12894 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
12895 arm_arch_core_flags data structure.
12896 * config/arm/arm-cpu_cdata.h: Regenerated.
12898 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12900 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
12902 (arm_print_hint_for_cpu_option): Likewise.
12903 (arm_print_hint_for_arch_option): Likewise.
12904 (arm_parse_cpu_option_name): Likewise.
12905 (arm_parse_arch_option_name): Likewise.
12906 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
12907 of entries in the all_fpus list.
12908 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
12909 (arm_parse_cpu_option_name): Declare.
12910 (arm_parse_arch_option_name): Declare.
12911 (arm_parse_option_features): Declare.
12912 (arm_intialize_isa): Declare.
12913 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
12915 (gen_comm_data): ... here. Make definitions non-static.
12916 * config/arm/arm-cpu-data.h: Regenerated.
12917 * config/arm/arm-cpu-cdata.h: Regenerated.
12919 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12921 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
12922 (cpu_arch_extension): New structure.
12923 (cpu_arch_option, arch_option, cpu_option): New structures.
12924 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
12925 architecture types.
12926 (gen_data): Generate new format data tables.
12927 * config/arm/arm.c (cpu_tune): New structure.
12928 (cpu_option, processors): Delete.
12929 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
12930 (arm_print_hint_for_cpu_option): ... this and ...
12931 (arm_print_hint_for_arch_option): ... this.
12932 (arm_parse_arch_cpu_name): Delete. Replace with ...
12933 (arm_parse_cpu_option_name): ... this and ...
12934 (arm_parse_arch_option_name): ... this.
12935 (arm_unrecognized_feature): Change type of target parameter to
12937 (arm_parse_arch_cpu_features): Delete. Replace with ...
12938 (arm_parse_option_features): ... this.
12939 (arm_configure_build_target): Rework to use new configuration data
12941 (arm_print_tune_info): Rework for new configuration data tables.
12942 * config/arm/arm-cpu-data.h: Regenerated.
12943 * config/arm/arm-cpu.h: Regenerated.
12945 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12947 * Makefile.in (OBJS): Move sbitmap.o from here ...
12948 (OBJS-libcommon): ... to here.
12950 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12952 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
12953 (ISA_ALL_CRYPTO): New macro.
12954 (ISA_ALL_SIMD): New macro
12955 (ISA_ALL_FP): New macro.
12956 * config/arm/arm.c (fpu_bitlist): Update initializer.
12957 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
12959 (arm9e): Add fpu. Add option for nofp
12960 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
12961 (arm926ej-s, arm1026ej-s): Likewise.
12962 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
12963 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
12964 neon-fp16, neon-vfpv4, nofp and nosimd.
12965 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
12966 (cortex-a8): Add fpu. Add option for nofp.
12967 (cortex-a9): Add fpu. Add options for nosimd and nofp.
12968 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
12969 (cortex-r4f): Add fpu.
12970 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
12971 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
12973 (cortex-r8): Likewise.
12974 (cortex-m4): Add fpu. Add option for nofp.
12975 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
12976 (cortex-a17.cortex-a7): Likewise.
12977 (cortex-a32): Add fpu. Add options for crypto and nofp.
12978 (cortex-a35, cortex-a53): Likewise.
12979 (cortex-a57): Add fpu. Add option for crypto.
12980 (cortex-a72, cortex-a73): Likewise.
12981 (exynos-m1): Likewise.
12982 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
12983 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
12984 (cortex-m33): Add fpu. Add option for nofp.
12985 * config/arm/arm-cpu-cdata.h: Regenerated
12986 * config/arm/arm-cpu-data.h: Regenerated.
12988 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12990 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
12991 (armv5te, armv5tej): Likewise.
12992 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
12993 (armv7): Add options fp and vfpv3-d16.
12994 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
12995 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
12997 (armv7ve): Likewise.
12998 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
12999 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
13000 (armv8-a): Add nocrypto option.
13001 (armv8.1-a, armv8.2-a): Likewise.
13002 (armv8-m.main): add options fp, fp.dp and nofp.
13004 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13006 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
13008 (armv8-a+crc): Delete.
13009 (armv8.1-a): Add options simd, crypto and nofp.
13010 (armv8.2-a): Add options fp16, simd, crypto and nofp.
13011 (armv8.2-a+fp16): Delete.
13012 (armv8-m.main): Add option dsp.
13013 (armv8-m.main+dsp): Delete.
13014 (cortex-a8): Add fpu. Add nofp option.
13015 (cortex-a9): Add fpu. Add nofp and nosimd options.
13016 * config/arm/parsecpu.awk (gen_data): Generate option tables and
13017 link to main cpu and architecture data structures.
13018 (gen_comm_data): Only put isa attributes from the main architecture
13020 (option): New statement for architecture and CPU entries.
13021 * arm.c (struct cpu_option): New structure.
13022 (struct processors): Add entry for options.
13023 (arm_unrecognized_feature): New function.
13024 (arm_parse_arch_cpu_name): Ignore any characters after the first
13026 (arm_parse_arch_cpu_feature): New function.
13027 (arm_configure_build_target): Separate out any CPU and architecture
13028 features and parse separately. Don't error out if -mfpu=auto is
13029 used with only an architecture string.
13030 (arm_print_asm_arch_directives): New function.
13031 (arm_file_start): Call it.
13032 * config/arm/arm-cpu-cdata.h: Regenerated.
13033 * config/arm/arm-cpu-data.h: Likewise.
13034 * config/arm/arm-tables.opt: Likewise.
13036 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13038 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
13039 assembler when it is not -mfpu=auto.
13041 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13043 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
13044 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
13045 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
13046 (ASM_CPU_SPEC): Rewrite.
13047 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
13048 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
13049 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
13050 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
13051 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
13052 copied string is NUL-terminated. Also strip any characters prefixed
13054 (arm_rewrite_selected_arch): New function.
13055 (arm_rewrite_march): New function.
13057 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13059 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
13060 (x_arm_cpu_string, x_arm_tune_string): Likewise.
13061 (march, mcpu, mtune): Convert to string-based options.
13062 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
13063 (arm_parse_arch_cpu_name): New function.
13064 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
13065 identify selected architecture or CPU.
13066 (arm_option_save): New function.
13067 (TARGET_OPTION_SAVE): Redefine.
13068 (arm_option_restore): Restore string options.
13069 (arm_option_print): Print string options.
13071 2017-06-16 Martin Sebor <msebor@redhat.com>
13073 PR tree-optimization/80933
13074 PR tree-optimization/80934
13075 * builtins.c (fold_builtin_3): Do not handle bcmp here.
13076 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
13077 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
13078 (gimple_fold_builtin): Call them.
13080 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13082 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
13083 as unlikely; update profile.
13085 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13087 * predict.c (force_edge_cold): Handle declaring edges impossible
13090 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13092 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
13094 (try_unroll_loop_completely): Fix reporting.
13096 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13098 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
13100 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
13103 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
13104 if given a non-constant argument for an intrinsic which requires a
13107 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13109 * profile.c (compare_freqs): New function.
13110 (branch_prob): Sort edge list.
13111 (find_spanning_tree): Assume that the list is priority sorted.
13113 2017-06-16 Richard Biener <rguenther@suse.de>
13115 PR tree-optimization/81090
13116 * passes.def (pass_record_bounds): Remove.
13117 * tree-pass.h (make_pass_record_bounds): Likewise.
13118 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
13119 make_pass_record_bounds): Likewise.
13120 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
13121 not free niter estimates at the beginning but at the end.
13122 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
13124 2017-06-16 Richard Biener <rguenther@suse.de>
13126 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
13127 initializer to workaround ICE in host GCC 4.8.
13129 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13131 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
13133 (clone_inlined_nodes): Update.
13135 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
13137 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
13138 prefetch settings, and enable prefetching by default at -O3.
13140 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
13142 * config/aarch64/aarch64.c (aarch64_override_options_internal):
13143 Set flag_prefetch_loop_arrays according to tuning data.
13145 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
13147 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
13148 New tune structure.
13149 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
13150 [Unrelated to main purpose of the patch] Place the pointer field last
13151 to enable type checking errors when tune structure are wrongly merged.
13152 * config/aarch64/aarch64.c (generic_prefetch_tune,)
13153 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
13154 (thunderx2t99_prefetch_tune): New tune constants.
13155 (tune_params *_tunings): Update all tunings (no functional change).
13156 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
13157 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
13158 from tunings structures.
13160 2017-06-16 Jakub Jelinek <jakub@redhat.com>
13163 * ubsan.c (instrument_null): Add T argument, use it instead
13164 of computing it based on IS_LHS.
13165 (instrument_object_size): Likewise.
13166 (pass_ubsan::execute): Adjust instrument_null and
13167 instrument_object_size callers to pass gimple_get_lhs or
13168 gimple_assign_rhs1 result to it. Use instrument_null instead of
13169 calling get_base_address and instrument_mem_ref. Handle
13170 aggregate call arguments for object-size sanitization.
13172 2017-06-16 Yury Gribov <tetra2005@gmail.com>
13174 PR tree-optimization/81089
13175 * tree-vrp.c (is_masked_range_test): Validate operands of
13178 2017-06-15 Martin Sebor <msebor@redhat.com>
13181 * dumpfile.c (dump_register): Avoid calling memset to initialize
13182 a class with a default ctor.
13183 * gcc.c (struct compiler): Remove const qualification.
13184 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
13185 * hash-table.h: Ditto.
13186 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
13188 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
13189 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
13191 * params.h (struct param_info): Make struct members non-const.
13192 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
13193 with default initialization.
13194 * vec.h (vec_copy_construct, vec_default_construct): New helper
13196 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
13197 with vec_copy_construct.
13198 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
13199 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
13200 * doc/invoke.texi (-Wclass-memaccess): Document.
13202 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
13204 * emit-rtl.h (is_leaf): Update comment about local
13205 register allocator.
13207 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13210 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
13211 for a variable to have a section before checking if the section has a
13213 Set section to.persistent if persistent attribute is set.
13214 Warn if .persistent attribute is used on an automatic variable.
13216 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
13218 PR rtl-optimization/80474
13219 * reorg.c (update_block): Do not ignore instructions in a delay slot.
13221 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
13223 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
13226 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
13228 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
13229 (casesi): Emit bounds checking as RTL.
13230 (casesi_internal_mips16_<mode>): Remove bounds checking.
13232 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
13234 * config/xtensa/xtensa.c (xtensa_option_override): Append
13235 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
13236 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
13237 xtensa_doloop_hooks): Define unconditionally.
13238 (xtensa_reorg_loops): Only call reorg_loops in the presence of
13240 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
13241 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
13242 for it in xtensa_option_override.
13243 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
13244 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
13246 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
13248 * doc/cppopts.texi: Document '-' special value to -MF.
13250 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
13252 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
13253 (cortex_a53_fconst): Likewise.
13254 (cortex_a53_fpmul): Likewise.
13255 (cortex_a53_f_load_64): Likewise.
13256 (cortex_a53_f_load_many): Likewise.
13257 (cortex_a53_advsimd_alu): Likewise.
13258 (cortex_a53_advsimd_alu_q): Likewise.
13259 (cortex_a53_advsimd_mul): Likewise.
13260 (cortex_a53_advsimd_mul_q): Likewise.
13261 (fpmac bypass): Add new bypass for fpmac-fpmac case.
13262 Add missing fmul, r2f_cvt and fconst cases.
13264 2017-06-14 Richard Biener <rguenther@suse.de>
13266 PR middle-end/81088
13267 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
13269 (fold_binary_loc): When associating do not treat pre-existing
13270 TREE_OVERFLOW on literal constants as a reason to allow
13271 TREE_OVERFLOW on associated literal constants.
13273 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
13275 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
13276 (MASK_FEATURES): New macro.
13277 * config/sparc/sparc.c (sparc_option_override): Remove the special
13278 handling of -mfpu and generalize it to all MASK_FEATURES switches.
13280 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
13282 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
13283 a division of 0 if non-call exceptions are enabled.
13285 2017-06-14 Andrew Pinski <apinski@cavium.com>
13286 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
13289 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
13290 Improve vector initialization code gen for only variable case.
13292 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
13294 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
13296 2017-06-14 Richard Biener <rguenther@suse.de>
13298 PR tree-optimization/81083
13299 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
13302 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13304 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
13305 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
13306 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
13307 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
13308 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
13309 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
13311 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13313 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
13314 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
13316 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13318 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
13320 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13322 * config/rs6000/t-rtems: Don't handle SPE.
13324 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13326 * config/rs6000/t-linux: Don't handle SPE.
13328 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13330 * config/rs6000/eabispe.h: Delete file.
13332 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13334 * config/rs6000/t-spe: Delete file.
13336 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13338 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
13339 (rs6000_legitimate_offset_address_p): Return false for anything in
13340 V2SImode or V2SFmode.
13342 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13344 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
13345 except V2SF and V2SI. Rearrange the vector modes, and add comments.
13346 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
13348 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
13349 (rs6000_legitimate_offset_address_p): Ditto.
13350 (rs6000_emit_move): Ditto.
13351 (rs6000_init_builtins): Remove V4HI_type_node.
13353 2017-06-13 Martin Liska <mliska@suse.cz>
13356 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
13357 (gate_asan): Likewise.
13358 * asan.h (asan_no_sanitize_address_p): Remove the function.
13359 (sanitize_flags_p): New function.
13360 * builtins.def: Fix coding style.
13361 * common.opt: Use renamed enum value.
13362 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
13363 * doc/extend.texi: Document no_sanitize attribute.
13364 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
13365 to SANITIZE_UNDEFINED_NONDEFAULT.
13366 * gcc.c (sanitize_spec_function): Use the renamed enum value.
13367 * gimple-fold.c (optimize_atomic_compare_exchange_p):
13368 Use sanitize_flags_p.
13369 * gimplify.c (gimplify_function_tree): Likewise.
13370 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
13371 * opts.c (parse_no_sanitize_attribute): New function.
13372 (common_handle_option): Use renamed enum value.
13373 * opts.h (parse_no_sanitize_attribute): Declare.
13374 * tree.c (sanitize_flags_p): New function.
13375 * tree.h: Declared here.
13376 * tsan.c: Use sanitize_flags_p.
13377 * ubsan.c (ubsan_expand_null_ifn): Likewise.
13378 (instrument_mem_ref): Likewise.
13379 (instrument_bool_enum_load): Likewise.
13380 (do_ubsan_in_current_function): Remove the function.
13381 (pass_ubsan::execute): Use sanitize_flags_p.
13382 * ubsan.h: Remove do_ubsan_in_current_function
13383 * tree-cfg.c (print_no_sanitize_attr_value): New function.
13384 (dump_function_to_file): Use it here.
13386 2017-06-13 Martin Jambor <mjambor@suse.cz>
13388 PR tree-optimization/80803
13389 PR tree-optimization/81063
13390 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
13391 (propagate_subaccesses_across_link): Enqueue subtree whenever
13392 necessary instead of relying on the caller.
13394 2017-06-13 Martin Jambor <mjambor@suse.cz>
13396 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
13397 that have a first_link.
13398 (sort_and_splice_var_accesses): Do not check first_link before
13400 (subtree_mark_written_and_enqueue): Likewise.
13401 (propagate_all_subaccesses): Likewise and do not stop at first
13402 parent with a first_link.
13404 2017-06-13 Martin Jambor <mjambor@suse.cz>
13406 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
13409 2017-06-13 Yury Gribov <tetra2005@gmail.com>
13411 * match.pd: New pattern.
13413 2017-06-13 Yury Gribov <tetra2005@gmail.com>
13415 * tree-vrp.c (is_masked_range_test): New function.
13416 (register_edge_assert_for): Determine ranges for
13419 2017-06-13 Yury Gribov <tetra2005@gmail.com>
13421 PR tree-optimization/67328
13422 * fold-const.c (maskable_range_p): New function.
13423 (build_range_check): Generate bittests if possible.
13425 2017-06-13 Martin Liska <mliska@suse.cz>
13427 * gimple-pretty-print.c (dump_probability): Add new argument.
13428 (dump_edge_probability): Dump both probability and count.
13429 (dump_gimple_label): Likewise.
13430 (dump_gimple_bb_header): Likewise.
13432 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
13435 * config/avr/avr-devices.c: Fix indentation.
13436 * config/avr/gen-avr-mmcu-specs.c: Dito.
13438 2017-06-13 Richard Biener <rguenther@suse.de>
13440 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
13441 instead get vector type from stmt_info.
13442 (vectorizable_reduction): Adjust. Remove dead code.
13444 2017-06-13 Richard Biener <rguenther@suse.de>
13446 PR middle-end/81065
13447 * fold-const.c (extract_muldiv_1): Remove bogus distribution
13448 case of C * (x * C2 + C3).
13449 (fold_addr_of_array_ref_difference): Properly fold index difference.
13451 2017-06-12 David S. Miller <davem@davemloft.net>
13454 * config/sparc/sparc.md (return expander): Emit frame blockage if
13455 function uses alloca.
13457 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
13459 * combine.c (make_field_assignment): Check len rather than the mode
13460 precision when calling force_to_mode.
13462 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
13464 Support multilibs and devices that see flash in RAM address range.
13467 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
13468 (avr_mcu_t) <flash_pm_offset>: New field.
13469 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
13470 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
13471 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
13472 (AVR_TINY_PM_OFFSET): Remove macro.
13473 * config/avr/avr.opt (-mshort-calls): New option.
13474 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
13475 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
13476 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
13477 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
13478 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
13479 instead of avr_arch->have_jmp_call.
13480 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
13481 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
13482 avr_arch->flash_pm_offset to define.
13483 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
13484 new field flash_pm_offset. Add entry for avrxmega3.
13485 (avr_texinfo): Add entry for avrxmega3.
13486 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
13487 attiny212, attiny214,
13488 attiny412, attiny414, attiny416, attiny417,
13489 attiny814, attiny816, attiny817,
13490 attiny1614, attiny1616, attiny1617,
13491 attiny3214, attiny3216, attiny3217.
13492 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
13493 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
13494 (avr_print_operand_address) [AVR_TINY]: Same.
13495 (avr_asm_init_sections) <readonly_data_section>: Only patch
13496 callback if avr_arch->flash_pm_offset = 0.
13497 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
13498 for rodata if avr_arch->flash_pm_offset != 0.
13499 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
13500 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
13501 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
13502 (m_options): Append opt_rcall.
13503 (m_dirnames): Append dir_rcall.
13504 * config/avr/t-multilib: Regenerate.
13506 * configure.ac [target=avr]: Check whether avrxmega3 default
13507 linker description file works as needed.
13508 * configure: Regenerate.
13509 * doc/avr-mmcu.texi: Regenerate.
13510 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
13511 <__AVR_ARCH__>: Document avrxmega3 and 103.
13512 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
13513 <__AVR_SHORT_CALLS__>: Document it.
13514 <__AVR_PM_BASE_ADDRESS__>: Document it.
13515 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
13516 (AVR Variable Attributes) <progmem>: Document this is
13517 not needed for avrxmega3.
13518 (AVR Named Address Spaces) <__flash>: Dito.
13520 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
13522 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
13524 2017-06-12 Doug Rupp <rupp@adacore.com>
13526 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
13527 Append vxworks-stdint.h to the tm_file list.
13528 * config/vxworks-stdint.h: New file.
13530 2017-06-12 Martin Liska <mliska@suse.cz>
13532 PR tree-optimization/81041
13533 * tree-profile.c (gimple_gen_ic_func_profiler):
13534 Create an extra BB in profile-generate
13535 (gimple_gen_time_profiler): Likewise.
13537 2017-06-12 Jakub Jelinek <jakub@redhat.com>
13539 PR tree-optimization/81003
13540 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
13541 (update_range_test): Use it instead of force_gimple_operand_gsi.
13543 2017-06-12 Richard Biener <rguenther@suse.de>
13545 PR tree-optimization/81053
13546 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
13547 with backedge value not defined in loop. Simplify def stmt
13550 2017-06-11 Tom de Vries <tom@codesourcery.com>
13553 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
13555 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
13556 nvptx_cannot_force_const_mem.
13558 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
13560 * opts.c (finish_options): Move test for flag_split_stack after
13561 it has been initialized.
13563 2017-06-11 Jason Merrill <jason@redhat.com>
13565 * tree.h (id_equal): New.
13566 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
13567 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
13568 instead of strcmp of IDENTIFIER_POINTER.
13570 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
13572 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
13573 (mark_all_inlined_calls_cdtor): Fix formating.
13574 (inline_transform): Rescale profile before inlining.
13576 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
13578 * cgraph.h (cgraph_edge::clone): Update prototype.
13579 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
13580 (cgraph_node::create_clone): Update.
13581 (cgraph_node::create_version_clone): Update.
13582 * tree-inline.c (copy_bb): Update.
13583 (expand_call_inline): Update.
13585 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
13587 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
13588 factored out from ...
13589 (rs6000_emit_prologue): ... here.
13591 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
13593 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
13594 factored out from ...
13595 (rs6000_emit_prologue): ... here.
13597 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
13599 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
13601 (handle_missing_profiles): Fix computation of tp_first_run.
13602 (counts_to_freqs): Do not touch freqs when count is 0.
13604 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
13606 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
13609 2017-06-10 Tom de Vries <tom@codesourcery.com>
13611 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
13612 attributes): Document signal effective target.
13614 2017-06-10 Tom de Vries <tom@codesourcery.com>
13616 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
13617 Document effective target stack_size.
13619 2017-06-09 David Malcolm <dmalcolm@redhat.com>
13621 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
13622 to the edit_context if they can be auto-applied.
13624 2017-06-9 Ian Lance Taylor <iant@golang.org>
13626 * opts.c (finish_options): If -fsplit-stack, disable implicit
13627 -forder-blocks-and-partition.
13628 * doc/invoke.texi (Optimize Options): Document that when using
13629 -fsplit-stack -forder-blocks-and-partition is not implicitly
13632 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
13634 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
13635 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
13636 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
13637 * builtins.def (abort, trap, unreachable): Declare cold.
13638 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
13639 * tree-core.h (ECF_COLD): New.
13640 * tree.c (set_call_expr_flags): Handle ECF_COLD.
13641 (build_common_builtin_nodes): Mark unreachable and abort as cold.
13643 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
13645 * predict.c (unlikely_executed_stmt_p): Cleanup.
13647 2017-06-09 Richard Biener <rguenther@suse.de>
13649 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
13650 model if the ref is always written to.
13652 2017-06-09 Tamar Christina <tamar.christina@arm.com>
13654 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
13656 2017-06-09 Tamar Christina <tamar.christina@arm.com>
13658 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
13661 2017-06-09 Tom de Vries <tom@codesourcery.com>
13664 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
13665 "target cannot support label values" when encountering LABEL_REF.
13667 2017-06-09 Martin Liska <mliska@suse.cz>
13669 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
13670 (gimple_gen_ic_func_profiler): Emit direct comparison
13671 of __gcov_indirect_call_callee with NULL.
13672 (gimple_gen_time_profiler): Change probability from
13673 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
13675 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
13677 * profile.c (edge_gcov_counts): Turn to pointer.
13678 (compute_branch_probabilities, compute_branch_probabilities): Update.
13679 (branch_prob): Do not clear edge_gcov_count.
13680 * profile.h (edge_gcov_counts): Turn to pointer.
13681 (edge_gcov_count): Update.
13683 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
13685 * gimple.h (gimple_check_failed): Mark cold.
13687 2017-06-09 Richard Biener <rguenther@suse.de>
13689 PR tree-optimization/66623
13690 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
13691 refactor check_reduction into two parts, properly computing
13692 whether we have to check reduction validity for outer loop
13695 2017-06-09 Richard Biener <rguenther@suse.de>
13697 PR tree-optimization/79483
13698 * graphite-scop-detection.c (order): New global.
13699 (get_order): Compute bb to order mapping that satisfies code
13700 generation constraints.
13701 (cmp_pbbs): New helper.
13702 (build_scops): Start domwalk at entry block, sort generated
13705 2017-06-09 Richard Biener <rguenther@suse.de>
13707 PR middle-end/81007
13708 * ipa-polymorphic-call.c
13709 (ipa_polymorphic_call_context::restrict_to_inner_class):
13710 Skip FIELD_DECLs with error_mark_node type.
13711 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
13714 2017-06-09 Martin Liska <mliska@suse.cz>
13716 * predict.c (struct branch_predictor): New struct.
13717 (test_prediction_value_range): New test.
13718 (predict_c_tests): New function.
13719 * selftest-run-tests.c (selftest::run_tests): Run the function.
13720 * selftest.h: Declare new tests.
13722 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
13725 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
13726 gen_add3_insn did not fail.
13727 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
13728 r0, construct that number in a temporary reg and add that reg to r0.
13729 If asked to put the result in r0 as well, fail.
13731 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
13733 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13734 for early expansion of vec_eqv.
13736 2017-06-08 Jakub Jelinek <jakub@redhat.com>
13738 PR middle-end/81005
13739 * ubsan.c (instrument_null): Avoid pointless code temporary.
13740 (pass_ubsan::execute): Instrument aggregate arguments of calls.
13742 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
13746 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
13749 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
13752 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13754 * predict.c (unlikely_executed_edge_p): Move ahead.
13755 (probably_never_executed_edge_p): Use it.
13757 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
13759 PR middle-end/79988
13760 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
13761 gimple_call_builtin_p call.
13763 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13765 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
13766 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
13767 rtl_check_failed_type2, rtl_check_failed_code1,
13768 rtl_check_failed_code2, rtl_check_failed_code_mode,
13769 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
13770 rtvec_check_failed_bounds, rtl_check_failed_flag,
13771 _fatal_insn_not_found, _fatal_insn): Likewise.
13772 * tree.h (tree_contains_struct_check_failed,
13773 tree_check_failed, tree_not_check_failed,
13774 tree_class_check_failed, tree_range_check_failed,
13775 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
13776 tree_vec_elt_check_failed, phi_node_elt_check_failed,
13777 tree_operand_check_failed, omp_clause_check_failed,
13778 omp_clause_operand_check_failed, omp_clause_range_check_failed):
13781 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13783 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
13784 flag_branch_probabilities.
13785 * ipa-inline.c (edge_badness): Likewise.
13786 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
13787 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
13788 * predict.c (maybe_hot_frequency_p): Likewise.
13789 (probably_never_executed): Likewise.
13790 * sched-ebb.c (schedule_ebbs): Likewise.
13791 * sched-rgn.c (find_single_block_region): Likewise.
13792 * tracer.c (tail_duplicate): Likewise.
13794 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13796 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
13797 longer requires x_flag_profile_use.
13799 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13801 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
13802 instead of flag_reorder_blocks_and_partition.
13803 * dbxout.c (dbxout_function_end): Likewise.
13804 * dwarf2out.c (gen_subprogram_die): Likewise.
13805 * haifa-sched.c (sched_create_recovery_edges): Likewise.
13806 * hw-doloop.c (reorg_loops): Likewise.
13807 * varasm.c (assemble_start_function,
13808 assemble_end_function): Likewise.
13809 (decide_function_section): Do not check for
13810 flag_reorder_blocks_and_partition.
13812 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
13814 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
13816 (chkp_get_hard_register_fake_addr_expr): Ditto.
13817 (chkp_build_addr_expr): Add check for hard reg case.
13818 (chkp_parse_array_and_component_ref): Ditto.
13819 (chkp_find_bounds_1): Ditto.
13820 (chkp_process_stmt): Don't generate bounds store for
13823 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13825 * predict.c (maybe_hot_bb_p): Do not check profile status.
13826 (maybe_hot_edge_p): Likewise.
13827 (probably_never_executed): Check for zero counts even if profile
13829 (unlikely_executed_edge_p): New function.
13830 (unlikely_executed_stmt_p): New function.
13831 (unlikely_executed_bb_p): New function.
13832 (set_even_probabilities): Use unlikely predicates.
13833 (combine_predictions_for_bb): Likewise.
13834 (predict_paths_for_bb): Likewise.
13835 (predict_paths_leading_to_edge): Likewise.
13836 (determine_unlikely_bbs): New function.
13837 (estimate_bb_frequencies): Use it.
13838 (compute_function_frequency): Use zero counts even if profile is
13840 * profile-count.h: Fix typo.
13842 2017-08-08 Julia Koval <julia.koval@intel.com>
13844 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
13845 _mm512_mask_cvtsepi16_storeu_epi8,
13846 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
13847 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
13848 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
13849 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
13850 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
13851 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
13852 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
13853 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
13854 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
13855 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
13856 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
13857 __builtin_ia32_pmovuswb256mem_mask,
13858 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
13859 __builtin_ia32_pmovwb512mem_mask): New builtins.
13861 2017-08-08 Julia Koval <julia.koval@intel.com>
13863 PR target/73350,80862
13864 * config/i386/subst.md (round): Fix round pattern.
13865 * config/i386/i386.c (ix86_erase_embedded_rounding):
13866 Fix erasing rounding for the fixed pattern.
13868 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13870 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
13872 2017-06-08 Martin Liska <mliska@suse.cz>
13874 PR gcov-profile/80911
13875 * gcov.c (block_info::block_info): New constructor.
13877 2017-06-07 Carl Love <cel@us.ibm.com>
13879 * config/rs6000/rs6000-c: The return type of the following
13880 built-in functions was implemented as int not long long. Fix sign
13881 of return value for the unsigned version of vec_mulo and vec_mule.
13882 vector unsigned long long vec_bperm (vector unsigned long long,
13883 vector unsigned char)
13884 vector signed long long vec_mule (vector signed int,
13886 vector unsigned long long vec_mule (vector unsigned int,
13887 vector unsigned int)
13888 vector signed long long vec_mulo (vector signed int,
13890 vector unsigned long long vec_mulo (vector unsigned int,
13891 vector unsigned int)
13892 * doc/extend.texi: Fix the documentation for the built-in
13895 2017-06-07 Carl Love <cel@us.ibm.com>
13898 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
13901 2017-06-07 Carl Love <cel@us.ibm.com>
13903 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
13904 support, Generate doublehv for signed int/float for BE case only.
13906 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
13908 * doc/invoke.texi (mcx16): Rewrite.
13910 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13912 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
13913 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
13914 *mov<mode>_softfloat, and an anonymous splitter): Use
13915 nonimmediate_operand instead of rs6000_nonimmediate_operand.
13917 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13919 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
13921 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
13922 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
13923 (rs6000_debug_reg_global): Adjust.
13924 (rs6000_init_hard_regno_mode_ok): Adjust.
13925 (rs6000_dbx_register_number): Adjust.
13926 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
13927 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
13928 Remove SPE_ACC and SPEFSCR.
13929 (REG_ALLOC_ORDER): Ditto.
13930 (FRAME_POINTER_REGNUM): Change to 111.
13931 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
13932 (REG_CLASS_NAMES): Ditto.
13933 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
13934 (REGISTER_NAMES): Ditto.
13935 (ADDITIONAL_REG_NAMES): Ditto.
13936 (rs6000_reg_names): Ditto.
13937 * config/rs6000/rs6000.md: Renumber some register number
13940 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13942 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
13944 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
13945 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
13947 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
13948 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
13949 Delete the SPE high registers.
13950 (REG_ALLOC_ORDER): Ditto.
13951 (enum reg_class): Remove SPE_HIGH_REGS.
13952 (REG_CLASS_NAMES): Ditto.
13953 (REG_CLASS_CONTENTS): Delete the SPE high registers.
13954 (REGISTER_NAMES): Ditto.
13955 (rs6000_reg_names): Ditto.
13956 * doc/tm.texi.in: Remove SPE as example.
13957 * doc/tm.texi: Regenerate.
13959 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13961 * config/rs6000/8540.md (ppc8540_brinc): Delete.
13962 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
13963 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
13964 * config/rs6000/rs6000.md (type): Remove "brinc".
13966 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13968 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
13969 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
13970 * config/rs6000/linuxspe.h: Delete file.
13971 * config/rs6000/rs6000.md: Don't include spe.md.
13972 * config/rs6000/spe.h: Delete file.
13973 * config/rs6000/spe.md: Delete file.
13974 * config/rs6000/t-rs6000: Remove spe.md.
13976 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13978 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
13979 (reg_or_none500mem_operand): Delete.
13980 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
13981 instead of reg_or_none500mem_operand.
13983 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13985 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
13986 handling of SPE flags.
13987 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
13989 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13991 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
13993 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
13994 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
13995 paired_divv2sf3): Similar.
13996 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
13997 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
13998 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
14000 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
14001 Rename the paired_* instruction patterns.
14002 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
14004 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
14005 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
14006 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
14007 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
14008 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
14009 PAIRED_VECTOR_MODE.
14010 (struct machine_function): Delete field spe_insn_chain_scanned_p.
14011 (spe_func_has_64bit_regs_p): Delete.
14012 (spe_expand_predicate_builtin): Delete.
14013 (spe_expand_evsel_builtin): Delete.
14014 (TARGET_DWARF_REGISTER_SPAN): Do not define.
14015 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
14016 (invalid_e500_subreg): Delete.
14017 (rs6000_legitimize_address): Always force_reg op2 as well, for
14018 paired single memory accesses.
14019 (rs6000_member_type_forces_blk): Delete.
14020 (rs6000_spe_function_arg): Delete.
14021 (rs6000_expand_unop_builtin): Delete SPE handling.
14022 (rs6000_expand_binop_builtin): Ditto.
14023 (spe_expand_stv_builtin): Delete.
14024 (bdesc_2arg_spe): Delete.
14025 (spe_expand_builtin): Delete.
14026 (spe_expand_predicate_builtin): Delete.
14027 (spe_expand_evsel_builtin): Delete.
14028 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
14029 (spe_init_builtins): Delete.
14030 (spe_func_has_64bit_regs_p): Delete.
14031 (savres_routine_name): Delete "info" parameter. Adjust callers.
14032 (rs6000_emit_stack_reset): Ditto.
14033 (rs6000_dwarf_register_span): Delete.
14034 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
14035 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
14036 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
14038 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
14040 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
14041 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
14042 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
14043 mulv2sf3, divv2sf3): Delete expanders.
14045 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14047 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
14049 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14051 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
14052 * config/rs6000/rs6000.c: Ditto.
14054 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14056 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
14057 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
14058 comparison_operator.
14060 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14062 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
14063 * config/rs6000/rs6000.opt: Ditto.
14064 * config/rs6000/t-rtems: Ditto.
14066 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14068 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
14069 TARGET_E500_SINGLE by 0, simplify.
14070 * config/rs6000/rs6000.c: Ditto.
14071 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
14072 (spe_build_register_parallel): Delete.
14073 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
14074 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
14075 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
14076 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
14077 (E500_CONVERT): Delete.
14078 * config/rs6000/spe.md: Remove many patterns and all define_constants.
14080 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14082 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
14083 * config/rs6000/dfp.md: Ditto.
14084 (negdd2, *negdd2_fpr): Merge.
14085 (absdd2, *absdd2_fpr): Merge.
14086 (negtd2, *negtd2_fpr): Merge.
14087 (abstd2, *abstd2_fpr): Merge.
14088 * config/rs6000/e500.h: Delete file.
14089 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
14090 TARGET_FPRS by 1 and simplify.
14091 * config/rs6000/rs6000-c.c: Ditto.
14092 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
14093 TARGET_DF_SPE by 0.
14094 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
14096 * config/rs6000/rs6000.md: Ditto.
14097 (floatdidf2, *floatdidf2_fpr): Merge.
14098 (move_from_CR_gt_bit): Delete.
14099 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
14100 (E500_CR_IOR_COMPARE): Delete.
14101 (All patterns that require !TARGET_FPRS): Delete.
14102 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
14104 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14106 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
14108 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14110 * graphds.c (add_edge): Intitialize edge's attached data.
14111 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
14112 pointer parameter. Call pointed function on each edge during
14113 graph traversing. Skip traversing the edge when the function
14115 (graphds_dfs, graphds_scc): Ditto.
14116 (for_each_edge): New parameter. Pass the new parameter to callback
14118 * graphds.h (skip_edge_callback): New function pointer type.
14119 (graphds_dfs, graphds_scc): New function pointer parameter.
14120 (graphds_edge_callback, for_each_edge): New parameter.
14122 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14124 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
14125 out code checking if runtime alias check is possible to below ...
14126 Call the new function.
14127 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
14128 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
14130 2017-06-07 Marek Polacek <polacek@redhat.com>
14133 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
14134 TYPE_OVERFLOW_WRAPS check.
14136 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14138 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
14139 if versioning is required.
14140 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
14141 peeling with the check for versioning.
14143 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14145 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
14146 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
14147 Set true to new parameter if new ssa variable is defined.
14148 (vect_gen_vector_loop_niters): Refactor. Set range information
14149 for the new vector loop bound variable.
14150 (vect_do_peeling): Ditto.
14152 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14154 * tree-affine.c (ssa.h): Include header file.
14155 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
14156 has wrapping overflow behavior.
14158 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14160 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
14162 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14164 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
14165 (tree_to_aff_combination): ... here.
14167 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14169 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
14170 reg_pressure model function.
14171 (ivopts_global_cost_for_size): Delete.
14172 (determine_set_costs, iv_ca_recount_cost): Call new model function
14173 ivopts_estimate_reg_pressure.
14175 2017-06-07 Tamar Christina <tamar.christina@arm.com>
14177 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
14178 expensive than udiv. Remove floating point cases from mod.
14180 2017-06-07 Tamar Christina <tamar.christina@arm.com>
14182 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
14183 Increase idiv cost.
14185 2017-06-07 Tamar Christina <tamar.christina@arm.com>
14187 * config/aarch64/aarch64.md
14188 (copysignsf3): Fix mask generation.
14190 2017-06-07 Jakub Jelinek <jakub@redhat.com>
14192 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
14194 (class dump_manager): Add register_dumps method.
14195 * dumpfile.c: Include langhooks.h.
14196 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
14197 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
14198 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
14199 (dump_manager::dump_register): Start with 512 entries instead of 32.
14200 (dump_manager::register_dumps): New method.
14201 * toplev.c (general_init): Instead of invoking register_dumps
14202 langhook, invoke register_dumps method on the dump manager.
14203 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
14206 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
14208 * doc/md.texi: Clarify the restrictions on a define_insn condition.
14209 Say that # requires an associated define_split to exist, and that
14210 the define_split must be suitable for use after register allocation.
14212 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14214 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
14215 (compute_outgoing_frequencies): Also initialize zero counts.
14216 (find_many_sub_basic_blocks): Do not produce uninitialized profile
14217 around loops; preserve more of profile when nothing changes.
14219 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
14221 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
14223 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
14224 * config/arm/arm-cpu-cdata.h: Regenerate.
14225 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
14226 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
14227 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
14228 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
14229 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
14231 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
14233 * config/arm/t-rmprofile: Likewise.
14234 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
14236 2017-06-06 David S. Miller <davem@davemloft.net>
14239 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
14240 blockage if function uses alloca.
14242 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
14244 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
14245 New "uid" fields to hold pretty-print IDs of group and ref.
14246 Memory references are now identified as <group_id>:<ref_id>
14247 instead of using [random] addresses.
14248 (dump_mem_details): Simplify, no functional change.
14249 (dump_mem_ref): Simplify and make output more concise.
14250 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
14251 (find_or_create_group): Initialize group uid.
14252 (record_ref): Initialize ref uid. Improve debug output.
14253 (prune_group_by_reuse, should_issue_prefetch_p,)
14254 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
14255 (mark_nontemporal_store, determine_loop_nest_reuse):
14256 Improve debug output.
14258 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
14260 * dbgcnt.def (prefetch): New debug counter.
14261 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
14262 (schedule_prefetches): Stop issueing prefetches if debug counter
14265 2017-06-06 Tom de Vries <tom@codesourcery.com>
14267 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
14268 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
14270 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14272 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
14273 Use aarch64_reg_or_zero predicate for operand 4.
14274 (aarch64_compare_and_swap<mode> define_insn_and_split):
14275 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
14276 (aarch64_store_exclusive<mode>): Likewise for operand 2.
14278 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
14280 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
14281 (arm_compute_save_core_reg_mask): This.
14282 (thumb1_compute_save_reg_mask): Rename into ...
14283 (thumb1_compute_save_core_reg_mask): This.
14284 (arm_compute_save_reg0_reg12_mask): Adapt comment.
14285 (arm_compute_frame_layout): Likewise.
14287 2017-06-06 Richard Biener <rguenther@suse.de>
14289 PR tree-optimization/80974
14290 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
14291 keep or clear leaders SSA info.
14293 2017-06-06 Tom de Vries <tom@codesourcery.com>
14295 * config/nvptx/nvptx.c (split_mode_p): New function.
14296 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
14298 2017-06-06 Tom de Vries <tom@codesourcery.com>
14300 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
14302 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14305 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
14308 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14310 * shrink-wrap.c (handle_simple_exit): Update profile.
14311 (try_shrink_wrapping): Upate profile.
14313 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14315 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
14316 (tree_guess_outgoing_edge_probabilities): New.
14317 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
14318 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
14320 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14322 * ipa-split.c (split_function): Initialize return bb profile.
14324 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14326 * profile.c (compute_branch_probabilities): Also initialize
14327 EXIT_BLOCK profile.
14329 2017-06-06 Richard Biener <rguenther@suse.de>
14331 PR tree-optimization/80928
14332 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
14333 (vect_analyze_loop_operations): Properly guard analysis for
14335 (vect_transform_loop): Likewise.
14336 (vect_analyze_loop_2): Also reset SLP type on PHIs.
14337 (vect_model_induction_cost): Do not cost for pure SLP.
14338 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
14339 of induction in inner loop vectorization.
14340 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
14341 (vect_get_and_check_slp_defs): Handle vect_induction_def.
14342 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
14344 (vect_analyze_slp_cost_1): Cost induction.
14345 (vect_detect_hybrid_slp_stmts): Handle PHIs.
14346 (vect_get_slp_vect_defs): Likewise.
14347 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
14348 (vect_transform_stmt): Handle SLP reductions.
14349 * tree-vectorizer.h (vectorizable_induction): Adjust.
14351 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
14353 * config/rs6000/rs6000.c (make_resolver_func): Update
14354 init_lowered_empty_function call.
14356 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
14358 * doc/invoke.texi: Document the -fprofile-abs-path option.
14359 * common.opt (fprofile-abs-path): New option.
14360 * gcov-io.h (gcov_write_filename): Declare.
14361 * gcov-io.c (gcov_write_filename): New function.
14362 * coverage.c (coverage_begin_function): Use gcov_write_filename.
14363 * profile.c (output_location): Likewise.
14365 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
14367 * shring-wrap.c: Revert accidental commit.
14369 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
14371 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
14373 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
14375 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
14377 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
14378 profile in callgraph edge.
14379 * profile-count.h (apply_probability): If THIS is 0, then result is 0
14380 (apply_scale): Likewise.
14381 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
14382 Also scale profile when inlining function with zero profile.
14383 (initialize_cfun): Update exit block profile even when it is zero.
14384 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
14385 when profile is read.
14387 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
14389 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
14390 (CLONE_*): New constants to define the processors we can generate
14391 code for with the target_clone attribute.
14392 (rs6000_clone_map): New array to identify which clone processors
14393 the current program is running on.
14394 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
14395 target_clone attribute.
14396 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
14397 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
14398 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
14399 (cpu_expand_builtin): Add support for target_clone attribute.
14400 (rs6000_valid_attribute_p): Allow "default" attribute.
14401 (get_decl_name): New debug function to simplify printing the
14402 current function name in debugging statements.
14403 (rs6000_clone_priority): New functions to support the target_clone
14404 attribute, and be able to generate code to switch between ISA 2.05
14405 through ISA 3.0 (power6 through power9).
14406 (rs6000_compare_version_priority): Likewise.
14407 (rs6000_get_function_versions_dispatcher): Likewise.
14408 (make_resolver_func): Likewise.
14409 (add_condition_to_bb): Likewise.
14410 (dispatch_function_versions): Likewise.
14411 (rs6000_generate_version_dispatcher_body): Likewise.
14412 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
14413 (fusion_gpr_load_p): Fix a spacing issue.
14414 * doc/extend.texi (Common Function Attributes): Document that the
14415 PowerPC supports the target_clone attribute.
14417 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
14419 * config/arm/arm.h: explain F symbol found in description of ARM
14420 register allocation in its legend.
14422 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
14424 * config/mips/frame-header-opt.c: Include profile-count.h.
14425 * config/riscv/riscv.c: Include profile-count.h
14427 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
14429 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
14431 (sm_set_flag_if_changed): Add bbs field.
14432 (execute_sm_if_changed_flag_set): Pass BBS.
14433 (execute_sm): Update.
14435 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14437 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
14440 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14442 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
14443 (peephole2): New peephole2 to emit the above.
14444 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
14446 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14448 * config/aarch64/aarch64.c (define_peephole2 above
14449 *sub_<shift>_<mode>): New peephole.
14451 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
14453 * config/i386/i386.c (make_resolver_func): Update.
14454 * Makefile.in: Add profile-count.h and profile-count.o
14455 * auto-profile.c (afdo_indirect_call): Update to new API.
14456 (afdo_set_bb_count): Update.
14457 (afdo_propagate_edge): Update.
14458 (afdo_propagate_circuit): Update.
14459 (afdo_calculate_branch_prob): Update.
14460 (afdo_annotate_cfg): Update.
14461 * basic-block.h: Include profile-count.h
14462 (struct edge_def): Turn count to profile_count.
14463 (struct basic_block_def): Likewie.
14464 (REG_BR_PROB_BASE): Move to profile-count.h
14465 (RDIV): Move to profile-count.h
14466 * bb-reorder.c (max_entry_count): Turn to profile_count.
14467 (find_traces): Update.
14468 (rotate_loop):Update.
14469 (connect_traces):Update.
14470 (sanitize_hot_paths):Update.
14471 * bt-load.c (migrate_btr_defs): Update.
14472 * cfg.c (RDIV): Remove.
14473 (init_flow): Use alloc_block.
14474 (alloc_block): Uninitialize count.
14475 (unchecked_make_edge): Uninitialize count.
14476 (check_bb_profile): Update.
14477 (dump_edge_info): Update.
14478 (dump_bb_info): Update.
14479 (update_bb_profile_for_threading): Update.
14480 (scale_bbs_frequencies_int): Update.
14481 (scale_bbs_frequencies_gcov_type): Update.
14482 (scale_bbs_frequencies_profile_count): New.
14483 * cfg.h (update_bb_profile_for_threading): Update.
14484 (scale_bbs_frequencies_profile_count): Declare.
14485 * cfgbuild.c (compute_outgoing_frequencies): Update.
14486 (find_many_sub_basic_blocks): Update.
14487 * cfgcleanup.c (try_forward_edges): Update.
14488 (try_crossjump_to_edge): Update.
14489 * cfgexpand.c (expand_gimple_tailcall): Update.
14490 (construct_exit_block): Update.
14491 * cfghooks.c (verify_flow_info): Update.
14492 (dump_bb_for_graph): Update.
14493 (split_edge): Update.
14494 (make_forwarder_block): Update.
14495 (duplicate_block): Update.
14496 (account_profile_record): Update.
14497 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
14498 (get_estimated_loop_iterations): Update.
14499 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
14500 (single_likely_exit): Update.
14501 * cfgloopmanip.c (scale_loop_profile): Update.
14503 (set_zero_probability): Update.
14504 (lv_adjust_loop_entry_edge): Update.
14505 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
14506 (purge_dead_edges): Update.
14507 (rtl_account_profile_record): Update.
14508 * cgraph.c (cgraph_node::create): Uninitialize count.
14509 (symbol_table::create_edge): Uninitialize count.
14510 (cgraph_update_edges_for_call_stmt_node): Update.
14511 (cgraph_edge::dump_edge_flags): Update.
14512 (cgraph_node::dump): Update.
14513 (cgraph_edge::maybe_hot_p): Update.
14514 * cgraph.h: Include profile-count.h
14515 (create_clone), create_edge, create_indirect_edge): Update.
14516 (cgraph_node): Turn count to profile_count.
14517 (cgraph_edge0: Likewise.
14518 (make_speculative, clone): Update.
14519 (create_edge): Update.
14520 (init_lowered_empty_function): Update.
14521 * cgraphclones.c (cgraph_edge::clone): Update.
14522 (duplicate_thunk_for_node): Update.
14523 (cgraph_node::create_clone): Update.
14524 * cgraphunit.c (cgraph_node::analyze): Update.
14525 (cgraph_node::expand_thunk): Update.
14526 * final.c (dump_basic_block_info): Update.
14527 * gimple-streamer-in.c (input_bb): Update.
14528 * gimple-streamer-out.c (output_bb): Update.
14529 * graphite.c (print_global_statistics): Update.
14530 (print_graphite_scop_statistics): Update.
14531 * hsa-brig.c: Include basic-block.h.
14532 * hsa-dump.c: Include basic-block.h.
14533 * hsa-gen.c (T sum_slice): Update.
14534 (convert_switch_statements):Update.
14535 * hsa-regalloc.c: Include basic-block.h.
14536 * ipa-chkp.c (chkp_produce_thunks): Update.
14537 * ipa-cp.c (struct caller_statistics): Update.
14538 (init_caller_stats): Update.
14539 (gather_caller_stats): Update.
14540 (ipcp_cloning_candidate_p): Update.
14541 (good_cloning_opportunity_p): Update.
14542 (get_info_about_necessary_edges): Update.
14543 (dump_profile_updates): Update.
14544 (update_profiling_info): Update.
14545 (update_specialized_profile): Update.
14546 (perhaps_add_new_callers): Update.
14547 (decide_about_value): Update.
14548 (ipa_cp_c_finalize): Update.
14549 * ipa-devirt.c (struct odr_type_warn_count): Update.
14550 (struct decl_warn_count): Update.
14551 (struct final_warning_record): Update.
14552 (possible_polymorphic_call_targets): Update.
14553 (ipa_devirt): Update.
14554 * ipa-fnsummary.c (redirect_to_unreachable): Update.
14555 * ipa-icf.c (sem_function::merge): Update.
14556 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
14557 * ipa-inline.c (compute_uninlined_call_time): Update.
14558 (compute_inlined_call_time): Update.
14559 (want_inline_small_function_p): Update.
14560 (want_inline_self_recursive_call_p): Update.
14561 (edge_badness): Update.
14562 (lookup_recursive_calls): Update.
14563 (recursive_inlining): Update.
14564 (inline_small_functions): Update.
14565 (dump_overall_stats): Update.
14566 (dump_inline_stats): Update.
14567 * ipa-profile.c (ipa_profile_generate_summary): Update.
14568 (ipa_propagate_frequency): Update.
14569 (ipa_profile): Update.
14570 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
14571 * ipa-utils.c (ipa_merge_profiles): Update.
14572 * loop-doloop.c (doloop_modify): Update.
14573 * loop-unroll.c (report_unroll): Update.
14574 (unroll_loop_runtime_iterations): Update.
14575 * lto-cgraph.c (lto_output_edge): Update.
14576 (lto_output_node): Update.
14577 (input_node): Update.
14578 (input_edge): Update.
14579 (merge_profile_summaries): Update.
14580 * lto-streamer-in.c (input_cfg): Update.
14581 * lto-streamer-out.c (output_cfg): Update.
14582 * mcf.c (create_fixup_graph): Update.
14583 (adjust_cfg_counts): Update.
14584 (sum_edge_counts): Update.
14585 * modulo-sched.c (sms_schedule): Update.
14586 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
14587 * predict.c (maybe_hot_count_p): Update.
14588 (probably_never_executed): Update.
14589 (dump_prediction): Update.
14590 (combine_predictions_for_bb): Update.
14591 (propagate_freq): Update.
14592 (handle_missing_profiles): Update.
14593 (counts_to_freqs): Update.
14594 (rebuild_frequencies): Update.
14595 (force_edge_cold): Update.
14596 * predict.h: Include profile-count.h
14597 (maybe_hot_count_p, counts_to_freqs): UPdate.
14598 * print-rtl-function.c: Do not include cfg.h
14599 * print-rtl.c: Include basic-block.h
14600 * profile-count.c: New file.
14601 * profile-count.h: New file.
14602 * profile.c (is_edge_inconsistent): Update.
14603 (correct_negative_edge_counts): Update.
14604 (is_inconsistent): Update.
14605 (set_bb_counts): Update.
14606 (read_profile_edge_counts): Update.
14607 (compute_frequency_overlap): Update.
14608 (compute_branch_probabilities): Update; Initialize and deinitialize
14610 (branch_prob): Update.
14611 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
14612 (edge_gcov_count): New.
14613 (bb_gcov_count): New.
14614 * shrink-wrap.c (try_shrink_wrapping): Update.
14615 * tracer.c (better_p): Update.
14616 * trans-mem.c (expand_transaction): Update.
14617 (ipa_tm_insert_irr_call): Update.
14618 (ipa_tm_insert_gettmclone_call): Update.
14619 * tree-call-cdce.c: Update.
14620 * tree-cfg.c (gimple_duplicate_sese_region): Update.
14621 (gimple_duplicate_sese_tail): Update.
14622 (gimple_account_profile_record): Update.
14623 (execute_fixup_cfg): Update.
14624 * tree-inline.c (copy_bb): Update.
14625 (copy_edges_for_bb): Update.
14626 (initialize_cfun): Update.
14627 (freqs_to_counts): Update.
14628 (copy_cfg_body): Update.
14629 (expand_call_inline): Update.
14630 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
14631 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
14632 (try_unroll_loop_completely): Update.
14633 (try_peel_loop): Update.
14634 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
14635 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
14636 * tree-ssa-loop-split.c (connect_loops): Update.
14637 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
14638 * tree-ssa-reassoc.c (branch_fixup): Update.
14639 * tree-ssa-tail-merge.c (replace_block_by): Update.
14640 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
14641 (compute_path_counts): Update.
14642 (update_profile): Update.
14643 (recompute_probabilities): Update.
14644 (update_joiner_offpath_counts): Update.
14645 (estimated_freqs_path): Update.
14646 (freqs_to_counts_path): Update.
14647 (clear_counts_path): Update.
14648 (ssa_fix_duplicate_block_edges): Update.
14649 (duplicate_thread_path): Update.
14650 * tree-switch-conversion.c (case_bit_test_cmp): Update.
14651 (struct switch_conv_info): Update.
14652 * tree-tailcall.c (decrease_profile): Update.
14653 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
14654 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
14655 * value-prof.c (check_counter): Update.
14656 (gimple_divmod_fixed_value): Update.
14657 (gimple_mod_pow2): Update.
14658 (gimple_mod_subtract): Update.
14659 (gimple_ic_transform): Update.
14660 (gimple_stringop_fixed_value): Update.
14661 * value-prof.h (gimple_ic): Update.
14663 2017-06-02 Carl Love <cel@us.ibm.com>
14665 * config/rs6000/rs6000-c: Add support for built-in functions
14666 vector double vec_doublee (vector signed int);
14667 vector double vec_doublee (vector unsigned int);
14668 vector double vec_doublee (vector float);
14669 vector double vec_doubleh (vector signed int);
14670 vector double vec_doubleh (vector unsigned int);
14671 vector double vec_doubleh (vector float);
14672 vector double vec_doublel (vector signed int);
14673 vector double vec_doublel (vector unsigned int);
14674 vector double vec_doublel (vector float);
14675 vector double vec_doubleo (vector signed int);
14676 vector double vec_doubleo (vector unsigned int);
14677 vector double vec_doubleo (vector float);.
14678 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
14679 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
14681 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
14682 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
14683 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
14685 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
14686 vec_doublel, vec_doubleh.
14687 * doc/extend.texi: Update the built-in documentation file for the
14688 new built-in functions.
14690 2017-06-02 David Malcolm <dmalcolm@redhat.com>
14693 * ipa-inline-analysis.c (free_growth_caches): Set
14694 edge_removal_hook_holder to NULL after removing it.
14696 2017-06-02 Sudakshina Das <sudi.das@arm.com>
14698 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
14699 comparision with zero.
14701 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
14702 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14703 for early expansion of vec_min and vec_max builtins.
14704 (builtin_function_type): Add min/max unsigned variants to those
14705 identified as having unsigned arguments.
14707 2017-06-02 Olivier Hainque <hainque@adacore.com>
14709 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
14711 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14713 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
14714 Use VALL_F16 iterator rather than VALL.
14716 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14718 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
14719 Emit CBNZ inside loop when doing a strong exchange and comparing
14720 against zero. Generate the CC flags after the loop.
14722 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
14724 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
14725 (dl_section_ref): New.
14726 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
14727 On AIX, append an expression to subtract the size of the
14728 section length to dl_section_ref.
14730 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
14732 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14733 for early expansion of vector absolute builtins.
14735 2017-06-02 Richard Biener <rguenther@suse.de>
14737 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
14738 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
14740 2017-06-02 Richard Biener <rguenther@suse.de>
14742 PR tree-optimization/80948
14743 * tree-tailcall.c (find_tail_calls): Track stmts to move in
14744 stmt order as well.
14746 2017-06-02 Richard Biener <rguenther@suse.de>
14748 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
14750 * tree-vect-stmts.c (process_use): Do not mark backedge defs
14751 for inductions as relevant.
14753 2017-06-02 Richard Biener <rguenther@suse.de>
14755 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
14756 (vectorizable_induction): ... this. Remove dead code.
14758 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
14760 * builtins. (expand_builtin_alloca): Remove second parameter and
14761 infer its value from the first parameter instead.
14762 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
14764 2017-06-02 Jakub Jelinek <jakub@redhat.com>
14766 PR rtl-optimization/80903
14767 * loop-doloop.c (add_test): Unshare sequence.
14769 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
14771 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
14773 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
14775 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
14777 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
14778 xlogue_layout::get_instance, logue_layout::xlogue_layout,
14779 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
14780 (xlogue_layout::get_stub_rtx): Make static.
14781 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
14782 (xlogue_layout::compute_stub_managed_regs): Rename to...
14783 (xlogue_layout::count_stub_managed_regs): ...this.
14784 (xlogue_layout::is_stub_managed_reg): New function.
14785 (xlogue_layout::m_stub_names): Rename to...
14786 (xlogue_layout::s_stub_names): ...this, make static.
14787 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
14788 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
14789 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
14790 xlogue_layout::s_stub_names): Instantiate statics.
14791 (stub_managed_regs): Remove.
14792 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
14793 (disable_call_ms2sysv_xlogues): Rename to...
14794 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
14795 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
14797 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
14798 change after reload_completed.
14799 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
14801 (ix86_expand_prologue): Likewise.
14802 (ix86_expand_epilogue): Likewise.
14803 (ix86_expand_split_stack_prologue): Likewise.
14804 (ix86_compute_frame_layout): Remove frame parameter ...
14805 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
14806 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
14808 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
14809 (ix86_frame): Move from here ...
14810 * config/i386/i386.h (ix86_frame): ... to here.
14811 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
14812 complete ix86_frame data structure instead. Remove some_ld_name.
14814 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
14816 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
14817 symbols that hold a DECL_VALUE_EXPR.
14819 2017-06-01 Martin Jambor <mjambor@suse.cz>
14821 PR tree-optimization/80898
14822 * tree-sra.c (process_subtree_disqualification): Removed.
14823 (disqualify_candidate): Do not acll
14824 process_subtree_disqualification.
14825 (subtree_mark_written_and_enqueue): New function.
14826 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
14827 RHS has been disqualified and re-queue LHS if necessary. Apart
14828 from that, ignore disqualified RHS.
14830 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14832 * config/s390/s390.c (s390_emit_epilogue): Disable early return
14833 address fetch for z10 or later.
14835 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14837 * config/arc/arc.md (tst_movb): Add guard when splitting.
14839 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14841 * config/arc/arc.c (arc_can_eliminate): Test against
14842 arc_frame_pointer_needed.
14844 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14846 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
14847 to prevent store reordering.
14848 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
14849 (type): Add block type.
14850 (stack_tie): Define special instruction to be used in
14853 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14855 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
14856 constraint. It is not valid for the pattern.
14857 (noncommutative_binary_comparison): Likewise.
14859 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14861 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
14864 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14866 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
14867 be used by the reg-alloc.
14869 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14871 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
14872 reg-alloc when having mul64 or mul32x16 instructions.
14873 (mulsidi3): Likewise.
14874 (umulsidi3): Likewise.
14875 (mulsi32x16): New pattern.
14876 (mulsi64): Likewise.
14877 (mulsidi64): Likewise.
14878 (umulsidi64): Likewise.
14879 (MUL32x16_REG): Define.
14880 (mul64_600): Use MUL32x16_REG.
14881 (mac64_600): Likewise.
14882 (umul64_600): Likewise.
14883 (umac64_600): Likewise.
14885 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14887 * config/arc/arc.md (mulsi3_700): Make it commutative.
14889 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
14891 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
14893 (*sign_extendsidi2_insn): Likewise for movstosw.
14895 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
14897 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
14898 the type of the input discriminant value. Convert the
14899 discriminant value of signedness vary.
14901 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
14903 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
14904 Add to -Wall section.
14906 2017-06-01 Richard Biener <rguenther@suse.de>
14908 PR middle-end/66313
14909 * fold-const.c (fold_plusminus_mult_expr): If the factored
14910 factor may be zero use a wrapping type for the inner operation.
14911 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
14912 and handle moved defs.
14913 (process_assignment): Properly guard the unary op case. Return a
14914 tri-state indicating that moving the stmt before the call may allow
14915 to continue. Pass through to_move.
14916 (find_tail_calls): Handle moving unrelated defs before
14919 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
14922 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
14923 splitter result in the canonical way.
14925 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
14927 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
14928 also for 32bit target. Update insn attributes.
14929 (zero-extendsidi2 splitter): Allow all registers for operand 1.
14931 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
14933 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
14934 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
14935 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
14936 (_mm_maskz_min_ss): New intrinsics.
14938 2017-05-31 Martin Liska <mliska@suse.cz>
14940 * tree-vect-loop.c (vect_create_epilog_for_reduction):
14941 Change comment style to one we normally use.
14942 (vectorizable_reduction): Likewise.
14943 (vectorizable_induction): Likewise.
14944 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
14945 (vectorizable_call): Likewise.
14946 (vectorizable_simd_clone_call): Likewise.
14947 (vectorizable_conversion): Likewise.
14948 (vectorizable_assignment): Likewise.
14949 (vectorizable_shift): Likewise.
14950 (vectorizable_operation): Likewise.
14951 (vectorizable_store): Likewise.
14952 (vectorizable_load): Likewise.
14953 * tree-vectorizer.h: Likewise.
14955 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
14957 * passes.c (emergency_dump_function): New.
14958 * tree-pass.h (emergency_dump_function): Declare.
14959 * plugin.c (plugins_internal_error_function): Remove.
14960 * plugin.h (plugins_internal_error_function): Remove declaration.
14961 * toplev.c (internal_error_function): New static function. Use it...
14962 (general_init): ...here.
14964 2017-05-31 Graham Markall <graham.markall@embecosm.com>
14966 * config/arc/arc.c (arc_print_operand): Handle constant operands.
14967 (arc_rtx_costs): Add costs for new patterns.
14968 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
14969 * config/arc/predicates.md: Add _1_2_3_operand predicate.
14971 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
14973 * tree-ssa-strlen.c (get_next_strinfo): New function.
14974 (get_stridx_plus_constant): Use it.
14975 (zero_length_string): Likewise.
14976 (adjust_related_strinfos): Likewise.
14977 (adjust_last_stmt): Likewise.
14979 2017-05-31 Richard Biener <rguenther@suse.de>
14982 * config/i386/i386.c (ix86_expand_builtin): Remove assert
14983 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
14985 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
14987 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
14988 loop_vinfo argument and use of dependence distance vectors.
14989 Check instead whether the two references differ only in their
14990 initial value and assume that they have the same alignment if the
14991 difference is a multiple of the vector alignment.
14992 (vect_analyze_data_refs_alignment): Update call accordingly.
14994 2017-05-31 Martin Liska <mliska@suse.cz>
14997 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
14999 2017-05-31 Bin Cheng <bin.cheng@arm.com>
15001 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
15002 (create_intersect_range_checks): Move from ...
15003 * tree-data-ref.c (create_intersect_range_checks_index)
15004 (create_intersect_range_checks): ... to here.
15005 (create_runtime_alias_checks): New function factored from ...
15006 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
15007 here. Call above function.
15008 * tree-data-ref.h (create_runtime_alias_checks): New function.
15010 2017-05-31 Bin Cheng <bin.cheng@arm.com>
15012 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
15013 segment length for dr_b and compute it in wide_int.
15015 2017-05-31 Richard Biener <rguenther@suse.de>
15017 PR tree-optimization/80906
15018 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
15019 and pass through iv_map.
15020 (copy_bb_and_scalar_dependences): Adjust.
15021 (translate_pending_phi_nodes): Likewise.
15022 (copy_loop_close_phi_args): Handle code-generating IVs instead
15025 2017-05-30 David Malcolm <dmalcolm@redhat.com>
15027 * diagnostic-color.c (color_dict): Add "type-diff".
15028 (parse_gcc_colors): Update comment.
15029 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
15030 -fdiagnostics-show-template-tree and -fno-elide-type.
15031 (GCC_COLORS): Add type-diff to example.
15033 (-fdiagnostics-show-template-tree): New.
15034 (-fno-elide-type): New.
15035 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
15036 the pp_format_decoder callback. Call any m_format_postprocessor's
15038 (pretty_printer::pretty_printer): Initialize
15039 m_format_postprocessor.
15040 (pretty_printer::~pretty_printer): Delete any
15041 m_format_postprocessor.
15042 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
15043 (class format_postprocessor): New class.
15044 (struct pretty_printer::format_decoder): Document the new parameters.
15045 (struct pretty_printer::m_format_postprocessor): New field.
15046 * tree-diagnostic.c (default_tree_printer): Update for new
15047 bool and const char ** params.
15048 * tree-diagnostic.h (default_tree_printer): Likewise.
15050 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
15052 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
15053 (lwa_operand): Delete rs6000_gen_cell_microcode test.
15054 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
15055 rs6000_gen_cell_microcode code.
15056 (rs6000_final_prescan_insn): Delete.
15057 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
15058 "warn-cell-microcode" entries.
15059 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
15060 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
15061 throughout. Change cc_reg_not_micro_cr0_operand to
15062 cc_reg_not_cr0_operand throughout.
15063 (*extendhi<mode>2_noload): Delete.
15064 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
15065 (mwarn-cell-microcode): Delete.
15066 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
15067 -mgen-cell-microcode and -mwarn-cell-microcode.
15069 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
15072 * config/i386/constraints.md (Yd): New constraint.
15074 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
15075 and (?Yd, r) alternatives. Update insn attributes.
15076 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
15077 and (?*Yd, r) alternatives. Update insn attributes.
15078 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
15080 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
15082 * gimplify.c (gimplify_modify_expr): Don't create a
15083 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
15086 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
15088 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
15090 2017-05-30 Richard Biener <rguenther@suse.de>
15092 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
15093 and reduc_def fields.
15094 (STMT_VINFO_REDUC_TYPE): New define.
15095 (STMT_VINFO_REDUC_DEF): Likewise.
15096 (vect_force_simple_reduction): Adjust prototype.
15097 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
15098 (vect_is_simple_reduction): Remove check_reduction argument.
15099 (vect_force_simple_reduction): Adjust and set
15100 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
15101 (vectorizable_reduction): Do not re-do reduction analysis
15102 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
15103 * tree-parloops.c (gather_scalar_reductions): Adjust.
15105 2017-05-30 Richard Biener <rguenther@suse.de>
15107 PR middle-end/80901
15108 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
15111 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
15113 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
15114 Introduce unknown_misalignment parameter and remove vf.
15115 (vect_peeling_hash_get_lowest_cost):
15116 Pass unknown_misalignment parameter.
15117 (vect_enhance_data_refs_alignment):
15118 Fix unsupportable data ref treatment.
15120 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
15122 * tree-vect-data-refs.c (vect_get_data_access_cost):
15123 Workaround for SLP handling.
15124 (vect_enhance_data_refs_alignment):
15125 Compute costs for doing no peeling at all, compare to the best
15126 peeling costs so far and avoid peeling if cheaper.
15128 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
15130 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
15131 Return peeling info and set costs to zero for unlimited cost
15133 (vect_enhance_data_refs_alignment): Also inspect all datarefs
15134 with unknown misalignment. Compute and costs for unknown
15135 misalignment, compare them to the costs for known misalignment
15136 and choose the cheapest for peeling.
15138 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
15140 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
15141 (vect_get_peeling_costs_all_drs): Create function.
15142 (vect_peeling_hash_get_lowest_cost):
15143 Use vect_get_peeling_costs_all_drs.
15144 (vect_peeling_supportable): Create function.
15145 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
15147 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
15149 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
15150 DR_HAS_NEGATIVE_STEP.
15151 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
15152 (vect_enhance_data_refs_alignment): Use.
15153 (vect_duplicate_ssa_name_ptr_info): Use.
15154 * tree-vectorizer.h (dr_misalignment): Use.
15155 (known_alignment_for_access_p): Use.
15157 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
15160 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
15162 (has_section_name): New function.
15164 2017-05-30 Martin Liska <mliska@suse.cz>
15167 * auto-profile.c (get_function_decl_from_block): Fix
15170 2017-05-30 Richard Biener <rguenther@suse.de>
15172 PR middle-end/80876
15173 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
15175 2017-05-30 Martin Liska <mliska@suse.cz>
15177 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
15178 * dumpfile.h (struct dump_file_info): Remove ctors.
15180 2017-05-30 Martin Liska <mliska@suse.cz>
15182 * predict.def: Fix GNU coding style.
15184 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
15186 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
15187 Mark 'to' argument with ATTRIBUTE_UNUSED.
15189 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
15191 * config/xtensa/xtensa.c (xtensa_emit_call): Use
15192 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
15193 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
15196 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
15198 * doc/install.texi (Options specification): Restore entry of
15199 --enable-sjlj-exceptions.
15201 2017-05-27 Michael Eager <eager@eagercon.com>
15204 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
15206 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
15208 * config/microblaze/microblaze.h
15209 (FIXED_REGISTERS): Update in macro.
15210 (CALL_USED_REGISTERS): Update in macro.
15212 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
15214 * doc/install.texi: Add links to macOS binary distributions.
15216 2017-05-27 Jakub Jelinek <jakub@redhat.com>
15220 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
15222 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
15224 2017-05-26 Martin Liska <mliska@suse.cz>
15226 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
15228 2017-05-26 Martin Liska <mliska@suse.cz>
15230 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
15231 always leading ';; '.
15232 (dump_bb_info): Likewise.
15233 (brief_dump_cfg): Likewise.
15234 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
15235 * dumpfile.c: Remove usage of TDF_VERBOSE.
15236 * dumpfile.h (enum dump_kind): Likewise.
15237 (dump_gimple_bb_header): Do not use TDF_COMMENT.
15238 * print-tree.c (debug_verbose): Remove.
15239 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
15240 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
15241 * tree-diagnostic.c (default_tree_printer): Replace
15242 TDF_DIAGNOSTIC with TDF_SLIM.
15244 2017-05-26 Bin Cheng <bin.cheng@arm.com>
15246 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
15247 in parameter loop, rather than loop_vinfo.
15248 (create_intersect_range_checks): Ditto.
15249 (vect_create_cond_for_alias_checks): Update call to above functions.
15251 2017-05-26 Bin Cheng <bin.cheng@arm.com>
15253 PR tree-optimization/80815
15254 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
15255 for merging runtime alias checks. Handle negative DR_STEPs.
15257 2017-05-26 Bin Cheng <bin.cheng@arm.com>
15259 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
15261 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
15262 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
15263 out code pruning runtime alias checks.
15264 * tree-data-ref.c (prune_runtime_alias_test_list): New function
15265 factored out from above.
15266 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
15268 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
15270 (prune_runtime_alias_test_list): New decalaration.
15272 2017-05-26 Bin Cheng <bin.cheng@arm.com>
15274 * tree-vect-data-refs.c (compare_tree): Rename and move ...
15275 * tree-data-ref.c (data_ref_compare_tree): ... to here.
15276 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
15277 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
15278 (operator==, comp_dr_with_seg_len_pair): Ditto.
15279 (vect_prune_runtime_alias_test_list): Ditto.
15281 2017-05-26 Martin Liska <mliska@suse.cz>
15284 * params.def: Bound partial-inlining-entry-probability param.
15286 2017-05-26 Marek Polacek <polacek@redhat.com>
15289 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
15292 2017-05-26 Richard Biener <rguenther@suse.de>
15294 PR tree-optimization/80842
15295 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
15298 2017-05-26 Richard Biener <rguenther@suse.de>
15300 PR tree-optimization/80844
15301 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
15303 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
15305 * doc/md.texi (Machine Constraints): Update x86 family
15306 machine constraints section to match 'config/i386/constraints.md'.
15308 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
15310 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
15312 2017-05-25 Nathan Sidwell <nathan@acm.org>
15314 * doc/invoke.texi (--enable-languages): Update documentation.
15316 2017-05-25 Martin Liska <mliska@suse.cz>
15318 * dumpfile.c: Add TDF_FOLDING.
15319 * dumpfile.h (enum dump_kind): Likewise.
15320 * genmatch.c (dt_simplify::gen_1): Use it.
15322 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
15324 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
15326 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
15328 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
15329 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
15331 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
15333 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
15334 * match.pd (X == C): Rewrite it here.
15335 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
15336 with_certain_nonzero_bits2): New predicates.
15337 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
15339 2017-05-24 Nathan Sidwell <nathan@acm.org>
15341 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
15344 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
15347 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
15349 * config/powerpcspe: New port. Files are copied from the rs6000
15350 port, with "rs6000" in filenames replaced by "powerpcspe".
15352 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
15354 PR rtl-optimization/80754
15355 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
15357 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
15359 * config/sparc/sparc.md (length): Return the correct value for -mflat
15360 sibcalls to match output_sibcall.
15362 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
15366 * config/rs6000/rs6000.c (struct machine_function): Add new field
15368 (rs6000_get_separate_components): Init that field, use it.
15369 (rs6000_components_for_bb): Use the field.
15371 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
15373 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
15375 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
15377 PR middle-end/80823
15378 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
15380 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15383 * config/s390/s390.c (s390_check_qrst_address): Check incoming
15384 address against address_operand predicate.
15385 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
15387 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
15389 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
15390 parameters passed indirectly.
15392 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
15394 * config/i386/i386.md (*movdi_internal): Remove SSE4
15395 alternative 18 (?r, *v). Update insn attributes.
15396 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
15397 Update insn attributes.
15398 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
15399 Update insn attributes.
15400 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
15401 alternative 1 (r, v). Remove isa attribute.
15402 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
15403 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
15404 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
15406 2017-05-23 Tom de Vries <tom@codesourcery.com>
15408 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
15411 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
15413 * cgraphunit.c (symbol_table::process_new_functions): Update.
15414 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
15415 (inline_generate_summary): Rename to ...
15416 (ipa_fn_summary_generate): ... this one.
15417 (inline_read_summary): Rename to ...
15418 (ipa_fn_summary_read): ... this one.
15419 (inline_write_summary): Rename to ...
15420 (ipa_fn_summary_write): ... this one.
15421 (inline_free_summary): Rename to ...
15422 (ipa_free_fn_summary): ... this one.
15423 (pass_data_local_fn_summary, pass_local_fn_summary,
15424 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
15425 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
15426 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
15427 make_pass_ipa_fn_summary): New.
15428 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
15429 inline_write_summary, inline_free_summary): Remove.
15430 (ipa_free_fn_summary) : New.
15431 * ipa-inline.c (ipa_inline): Update.
15432 (pass_ipa_inline): Do not generate summaries.
15433 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
15435 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
15436 and add pass_ipa_fn_summary.
15437 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
15439 (make_pass_inline_parameters): Remove.
15441 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
15443 * omp-low.c (struct omp_context): Remove "default_kind" member.
15446 * omp-offload.c (execute_oacc_device_lower): Remove the
15447 parallelism dimensions function attributes for unparallelized
15448 OpenACC kernels constructs.
15450 2017-05-23 Martin Liska <mliska@suse.cz>
15452 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
15454 (cgraph_edge::make_speculative): Likewise.
15455 (cgraph_edge::resolve_speculation): Likewise.
15456 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
15457 (cgraph_node::dump): Likewise.
15458 * cgraph.h: Likewise.
15459 * cgraphunit.c (analyze_functions): Likewise.
15460 (symbol_table::compile): Likewise.
15461 * ipa-cp.c (print_all_lattices): Likewise.
15462 (determine_versionability): Likewise.
15463 (initialize_node_lattices): Likewise.
15464 (ipcp_verify_propagated_values): Likewise.
15465 (estimate_local_effects): Likewise.
15466 (update_profiling_info): Likewise.
15467 (create_specialized_node): Likewise.
15468 (perhaps_add_new_callers): Likewise.
15469 (decide_about_value): Likewise.
15470 (decide_whether_version_node): Likewise.
15471 (identify_dead_nodes): Likewise.
15472 (ipcp_store_bits_results): Likewise.
15473 * ipa-devirt.c (dump_targets): Likewise.
15474 (ipa_devirt): Likewise.
15475 * ipa-icf.c (sem_item::dump): Likewise.
15476 (sem_function::equals): Likewise.
15477 (sem_variable::equals): Likewise.
15478 (sem_item_optimizer::read_section): Likewise.
15479 (sem_item_optimizer::execute): Likewise.
15480 (congruence_class::dump): Likewise.
15481 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
15482 (dump_inline_summary): Likewise.
15483 (estimate_node_size_and_time): Likewise.
15484 (inline_analyze_function): Likewise.
15485 * ipa-inline-transform.c (inline_call): Likewise.
15486 * ipa-inline.c (report_inline_failed_reason): Likewise.
15487 (want_early_inline_function_p): Likewise.
15488 (edge_badness): Likewise.
15489 (update_edge_key): Likewise.
15490 (inline_small_functions): Likewise.
15491 * ipa-profile.c (ipa_profile): Likewise.
15492 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
15493 (ipa_make_edge_direct_to_target): Likewise.
15494 (remove_described_reference): Likewise.
15495 (ipa_impossible_devirt_target): Likewise.
15496 (propagate_controlled_uses): Likewise.
15497 (ipa_print_node_params): Likewise.
15498 (ipcp_transform_function): Likewise.
15499 * ipa-pure-const.c (pure_const_read_summary): Likewise.
15500 (propagate_pure_const): Likewise.
15501 * ipa-reference.c (generate_summary): Likewise.
15502 (read_write_all_from_decl): Likewise.
15503 (propagate): Likewise.
15504 (ipa_reference_read_optimization_summary): Likewise.
15505 * ipa-utils.c (ipa_merge_profiles): Likewise.
15506 * ipa.c (walk_polymorphic_call_targets): Likewise.
15507 (symbol_table::remove_unreachable_nodes): Likewise.
15508 (ipa_single_use): Likewise.
15509 * passes.c (execute_todo): Likewise.
15510 * predict.c (drop_profile): Likewise.
15511 * symtab.c (symtab_node::get_dump_name): New function.
15512 (symtab_node::dump_name): Likewise.
15513 (symtab_node::dump_asm_name): Likewise.
15514 (symtab_node::dump_references): Likewise.
15515 (symtab_node::dump_referring): Likewise.
15516 (symtab_node::dump_base): Likewise.
15517 (symtab_node::debug_symtab): Likewise.
15518 * tree-sra.c (convert_callers_for_node): Likewise.
15519 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
15520 * value-prof.c (init_node_map): Likewise.
15522 2017-05-23 Martin Liska <mliska@suse.cz>
15524 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
15525 and symtab_node::debug_symtab to symbol_table::debug.
15526 * cgraphunit.c (analyze_functions): Use the renamed function.
15527 (symbol_table::compile): Likewise.
15528 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
15529 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
15530 * passes.c (execute_todo): Likewise.
15531 * symtab.c (symbol_table::dump): New function.
15532 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
15534 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
15536 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
15537 that nonconst implies exec.
15539 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
15541 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
15542 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
15543 (inline_edge_summary_vec): Turn into ...
15544 (ipa_call_summaries): ... this one.
15545 (redirect_to_unreachable, edge_set_predicate,
15546 evaluate_properties_for_edge, inline_summary_alloc,
15547 reset_ipa_call_summary, reset_inline_summary,
15548 inline_summary_t::duplicate): Update.
15549 (inline_edge_duplication_hook): Turn to ...
15550 (ipa_call_summary_t::duplicate): ... this one.
15551 (inline_edge_removal_hook): Turn to ...
15552 (ipa_call_summary_t::remove): ... this one.
15553 (dump_inline_edge_summary): Turn to ...
15554 (dump_ipa_call_summary): ... this one.
15555 (estimate_function_body_sizes): Update.
15556 (inline_update_callee_summaries): Update.
15557 (remap_edge_change_prob): Update.
15558 (remap_edge_summaries): Update.
15559 (inline_merge_summary): Update.
15560 (do_estimate_edge_time): Update.
15561 (inline_generate_summary): Update.
15562 (inline_read_section): Update.
15563 (inline_read_summary): Update.
15564 (inline_free_summary): Update.
15565 * ipa-inline.c (can_inline_edge_p): Update.
15566 (compute_inlined_call_time): Update.
15567 (want_inline_small_function_p): Update.
15568 (edge_badness): Update.
15569 (early_inliner): Update.
15570 * ipa-inline.h (inline_edge_summary): Turn to ...
15571 (ipa_call_summary): ... this one.
15572 (ipa_call_summary_t): New class.
15573 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
15574 (ipa_call_summaries): New.
15575 (inline_edge_summary): Remove.
15576 (estimate_edge_growth): Update.
15577 * ipa-profile.c (ipa_propagate_frequency_1): Update.
15578 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
15579 * ipa-split.c (execute_split_functions): Update.
15580 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
15582 2017-05-23 Tom de Vries <tom@codesourcery.com>
15584 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
15585 attributes): Document rdrand effective target.
15587 2017-05-23 Tom de Vries <tom@codesourcery.com>
15589 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
15590 attributes): Sort alphabetically.
15592 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
15594 * config/avr/genmultilib.awk: Use gsub instead of gensub.
15596 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
15599 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
15600 V2DF/V2DI splat into two separate patterns, one that handles
15601 registers, and the other that only handles memory. Drop support
15602 for splatting from a GPR on ISA 2.07 and then splitting the
15603 splat into direct move and splat.
15604 (vsx_splat_<mode>_reg): Likewise.
15605 (vsx_splat_<mode>_mem): Likewise.
15607 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
15609 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
15611 2017-05-22 Jakub Jelinek <jakub@redhat.com>
15613 PR middle-end/80809
15614 * omp-low.c (finish_taskreg_remap): New function.
15615 (finish_taskreg_scan): If unit size of ctx->record_type
15616 is non-constant, unshare the size expression and replace
15617 decls in it with possible outer var refs.
15619 PR middle-end/80809
15620 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
15621 GOVD_SHARED rather than GOVD_PRIVATE with it.
15622 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
15623 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
15625 PR middle-end/80853
15626 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
15627 as last argument to build_outer_var_ref for pointer bases of array
15628 section reductions.
15630 2017-05-19 Martin Sebor <msebor@redhat.com>
15632 * print-tree.c (print_node): Print DECL_READ_P flag.
15634 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15636 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
15637 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
15638 * cgraph.c: Likewise.
15639 * cgraphunit.c: Likewise.
15640 * gengtype.c: Likewise.
15641 * ipa-cp.c: Likewise.
15642 * ipa-devirt.c: Likewise.
15643 * ipa-icf.c: Likewise.
15644 * ipa-predicate.c: Likewise.
15645 * ipa-profile.c: Likewise.
15646 * ipa-prop.c: Likewise.
15647 * ipa-split.c: Likewise.
15649 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
15650 edge_predicate_pool, dump_inline_hints,
15651 inline_summary::account_size_time, redirect_to_unreachable,
15652 edge_set_predicate, set_hint_predicate,
15653 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
15654 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
15655 inline_summary_t::remove, remap_hint_predicate_after_duplication,
15656 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
15657 ipa_call_summary_t::remove, initialize_growth_caches,
15658 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
15659 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
15660 mark_modified, unmodified_parm_1, unmodified_parm,
15661 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
15662 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
15663 compute_bb_predicates, will_be_nonconstant_expr_predicate,
15664 will_be_nonconstant_predicate, record_modified_bb_info,
15665 get_minimal_bb, record_modified, param_change_prob,
15666 phi_result_unknown_predicate, predicate_for_phi_result,
15667 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
15668 estimate_function_body_sizes, compute_inline_parameters,
15669 compute_inline_parameters_for_curren, pass_data_inline_parameters,
15670 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
15671 inline_update_callee_summaries, remap_edge_change_prob,
15672 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
15673 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
15674 inline_analyze_function, inline_summary_t::insert,
15675 inline_generate_summary, read_ipa_call_summary, inline_read_section,
15676 inline_read_summary, write_ipa_call_summary, inline_write_summary,
15677 inline_free_summary): Move to ipa-fnsummary.h
15678 (predicate_t): Remove.
15679 * ipa-fnsummary.c: New file.
15680 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
15681 (enum inline_hints_vals, inline_hints, agg_position_info,
15682 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
15683 inline_summaries, ipa_call_summary, ipa_call_summary_t,
15684 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
15685 dump_inline_summary, dump_inline_hints, inline_generate_summary,
15686 inline_read_summary, inline_write_summary, inline_free_summary,
15687 inline_analyze_function, initialize_inline_failed,
15688 inline_merge_summary, inline_update_overall_summary,
15689 compute_inline_parameters): Move to ipa-fnsummary.h
15690 * ipa-fnsummary.h: New file.
15691 * ipa-inline-transform.h: Include ipa-inline.h.
15692 * ipa-inline.c: LIkewise.
15694 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15696 * ipa-inline.c (edge_badness): Use inlined_time instead of
15697 inline_summaries->get.
15699 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15701 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
15703 2017-05-22 Nathan Sidwell <nathan@acm.org>
15705 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
15706 (fdump-lang): Document 'raw' option.
15707 * dumpfile.h (TDI_tu): Delete.
15708 * dumpfile.c (dump_files): Remove translation-unit.
15709 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
15711 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
15713 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
15714 command option from $(AWK) call.
15715 * config/avr/genmultilib.awk: Simplify and rewrite so that it
15716 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
15717 [FORMAT]: Remove handling of variable.
15718 * config/avr/t-multilib: Regenerate.
15720 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15722 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
15724 (dump_inline_summary): Do not print self_time.
15725 (estimate_function_body_sizes): Do not set self_time.
15726 (compute_inline_parameters): Likewise.
15727 (inline_read_section, inline_write_summary): Do not stream self_time.
15728 * ipa-inline.h (inline_summary): Drop self_time.
15730 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15732 * ipa-inline-analysis.c (account_size_time): Rename to ...
15733 (inline_summary::account_size_time): ... this one.
15734 (reset_ipa_call_summary): Turn to ...
15735 (ipa_call_summary::reset): ... this one.
15736 (reset_inline_summary): Turn to ...
15737 (inline_summary::reset): ... this one.
15738 (inline_summary_t::remove): Update.
15739 (inline_summary_t::duplicate): Update.
15740 (ipa_call_summary_t::remove): Update.
15741 (dump_inline_summary): Update.
15742 (estimate_function_body_sizes): Update.
15743 (compute_inline_parameters): Update.
15744 (estimate_node_size_and_time): Update.
15745 (inline_merge_summary): Update.
15746 (inline_update_overall_summary): Update.
15747 (inline_read_section): Update.
15748 (inline_write_summary): Update.
15749 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
15750 add account_size_time and reset member functions.
15751 (ipa_call_summary): Add reset function.
15752 * ipa-predicate.h (predicate::operator &): Constify.
15754 2017-05-22 Richard Biener <rguenther@suse.de>
15756 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
15758 2017-05-19 Jason Merrill <jason@redhat.com>
15760 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
15762 2017-05-19 Marek Polacek <polacek@redhat.com>
15765 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
15766 TYPE_OVERFLOW_WRAPS checks.
15768 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
15770 * tree-core.h (enum omp_clause_default_kind): Add
15771 "OMP_CLAUSE_DEFAULT_PRESENT".
15772 * tree-pretty-print.c (dump_omp_clause): Handle it.
15773 * gimplify.c (enum gimplify_omp_var_data): Add
15774 "GOVD_MAP_FORCE_PRESENT".
15775 (gimplify_adjust_omp_clauses_1): Map it to
15776 "GOMP_MAP_FORCE_PRESENT".
15777 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
15779 * gimplify.c (oacc_default_clause): Clarify.
15781 2017-05-19 Nathan Sidwell <nathan@acm.org>
15783 LANG_HOOK_REGISTER_DUMPS
15784 * toplev.c (general_init): Call register dump lang hook.
15785 * doc/invoke.texi: Document -fdump-lang option family.
15786 * dumpfile.c (dump_files): Remove class dump here.
15787 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
15788 * dumpfile.h (tree_dump_index): Remove TDI_class.
15789 * langhooks-def.h (lhd_register_dumps): Declare.
15790 (LANG_HOOKS_REGISTER_DUMPS): Define.
15791 (LANG_HOOKS_INITIALIZER): Add it.
15792 * langhooks.c (lhd_register_dumps): Define.
15793 * langhooks.h (struct lang_hooks): Add register_dumps.
15795 2017-05-19 Nathan Sidwell <nathan@acm.org>
15797 * context.h (context::set_passes): New.
15798 * context.c (context::context): Do not create pass manager.
15799 * toplev.c (general_init): Create pass manager here.
15801 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
15803 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
15804 use this splitter if two add or or instructions would also work for
15805 the constant we want to generate.
15807 2017-05-19 Richard Biener <rguenther@suse.de>
15810 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
15811 predicate evaluation.
15813 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
15815 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
15817 * ipa-inline.c (want_inline_small_function_p): Do not cast to
15820 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
15822 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
15823 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
15824 (inline_edge_summary_vec): Turn into ...
15825 (ipa_call_summaries): ... this one.
15826 (redirect_to_unreachable, edge_set_predicate,
15827 evaluate_properties_for_edge, inline_summary_alloc,
15828 reset_ipa_call_summary, reset_inline_summary,
15829 inline_summary_t::duplicate): Update.
15830 (inline_edge_duplication_hook): Turn to ...
15831 (ipa_call_summary_t::duplicate): ... this one.
15832 (inline_edge_removal_hook): Turn to ...
15833 (ipa_call_summary_t::remove): ... this one.
15834 (dump_inline_edge_summary): Turn to ...
15835 (dump_ipa_call_summary): ... this one.
15836 (estimate_function_body_sizes): Update.
15837 (inline_update_callee_summaries): Update.
15838 (remap_edge_change_prob): Update.
15839 (remap_edge_summaries): Update.
15840 (inline_merge_summary): Update.
15841 (do_estimate_edge_time): Update.
15842 (inline_generate_summary): Update.
15843 (inline_read_section): Update.
15844 (inline_read_summary): Update.
15845 (inline_free_summary): Update.
15846 * ipa-inline.c (can_inline_edge_p): Update.
15847 (compute_inlined_call_time): Update.
15848 (want_inline_small_function_p): Update.
15849 (edge_badness): Update.
15850 (early_inliner): Update.
15851 * ipa-inline.h (inline_edge_summary): Turn to ...
15852 (ipa_call_summary): ... this one.
15853 (ipa_call_summary_t): New class.
15854 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
15855 (ipa_call_summaries): New.
15856 (inline_edge_summary): Remove.
15857 (estimate_edge_growth): Update.
15858 * ipa-profile.c (ipa_propagate_frequency_1): Update.
15859 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
15860 * ipa-split.c (execute_split_functions): Update.
15861 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
15863 2017-05-19 Richard Biener <rguenther@suse.de>
15865 PR middle-end/80764
15866 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
15868 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
15870 * config/rs6000/rs6000.c (struct machine_function): Add field
15871 fpr_is_wrapped_separately.
15872 (rs6000_get_separate_components): Use 64 components. Handle the
15873 new FPR components.
15874 (rs6000_components_for_bb): Handle the FPR components.
15875 (rs6000_emit_prologue_components): Handle the FPR components.
15876 (rs6000_emit_epilogue_components): Handle the FPR components.
15877 (rs6000_set_handled_components): Handle the FPR components.
15878 (rs6000_emit_prologue): Don't output prologue code for those FPRs
15879 that are already separately shrink-wrapped.
15880 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
15881 that are already separately shrink-wrapped.
15883 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
15886 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
15889 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
15890 (define_peephole2 for Altivec d-form load): Add peepholes to catch
15891 cases where the register allocator uses a move and an offsettable
15892 memory operation to/from a FPR register on ISA 2.06/2.07.
15893 (define_peephole2 for Altivec d-form store): Likewise.
15895 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
15898 * config/i386/mmx.md (*mov<mode>_internal): Enable
15899 alternatives 11, 12, 13 and 14 also for 32bit targets.
15900 Remove alternatives 15, 16, 17 and 18.
15901 * config/i386/sse.md (vec_concatv2di): Change
15902 alternative (!x, *y) to (x, ?!*Yn).
15904 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
15906 * dumpfile.h (enum dump_kind): Remove stray comma.
15908 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
15910 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
15911 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
15912 predicate::num_conditions
15913 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
15914 (CHANGED): turn into predicate::changed.
15915 (agg_position_info): Move to ipa-predicate.h
15916 (add_condition, predicate::add_clause, predicate::operator &=,
15917 predicate::or_with, predicate::evaluate, predicate::probability,
15918 dump_condition, dump_clause, predicate::dump,
15919 predicate::remap_after_duplication, predicate::remap_after_inlining,
15920 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
15921 (evaluate_conditions_for_known_args): Update.
15922 (set_cond_stmt_execution_predicate): Update.
15923 * ipa-inline.h: Include ipa-predicate.h
15924 (condition, inline_param_summary, conditions, agg_position_info,
15925 predicate): Move to ipa-predicate.h
15926 * ipa-predicate.c: New file.
15927 * ipa-predicate.h: New file.
15929 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
15931 * final.c (leaf_function_p): Check we are not in a sequence.
15933 2017-05-18 Martin Liska <mliska@suse.cz>
15935 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
15936 * dumpfile.c (dump_register): Use new enum dump_kind.
15937 (get_dump_file_name): Likewise.
15938 (dump_enable_all): Likewise.
15939 (dump_switch_p_1): Likewise.
15940 (enable_rtl_dump_file): Remove usage of TDF_RTL.
15941 * dumpfile.h (enum dump_kind): New enum type.
15942 (struct dump_file_info): Create constructor and
15943 format fields and comments.
15944 * passes.c (pass_manager::register_one_dump_file):
15946 * statistics.c (statistics_early_init): Likewise.
15947 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
15948 TDF_TREE with TDF_SLIM.
15949 (gather_memory_references_ref): Likewise.
15951 2017-05-18 Martin Liska <mliska@suse.cz>
15953 * vec.h (struct vnull): Use it.
15955 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
15957 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
15958 (true_predicate, false_predicate, true_predicate_p,
15959 false_predicate_p): Remove.
15960 (single_cond_predicate, not_inlined_predicate): Turn to member function
15962 (add_condition): Update.
15963 (add_clause): Turn to...
15964 (predicate::add_clause): ... this one; update; allow passing NULL
15966 (and_predicates): Turn to ...
15967 (predicate::operator &=): ... this one.
15968 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
15969 (or_predicates): Turn to ...
15970 (predicate::or_with): ... this one.
15971 (evaluate_predicate): Turn to ...
15972 (predicate::evaluate): ... this one.
15973 (predicate_probability): Turn to ...
15974 (predicate::probability): ... this one.
15975 (dump_condition): Update.
15976 (dump_predicate): Turn to ...
15977 (predicate::dump): ... this one.
15978 (account_size_time): Update.
15979 (edge_set_predicate): Update.
15980 (set_hint_predicate): UPdate.
15981 (evaluate_conditions_for_known_args): Update.
15982 (evaluate_properties_for_edge): Update.
15983 (remap_predicate_after_duplication): Turn to...
15984 (predicate::remap_after_duplication): ... this one.
15985 (remap_hint_predicate_after_duplication): Update.
15986 (inline_summary_t::duplicate): UPdate.
15987 (dump_inline_edge_summary): Update.
15988 (dump_inline_summary): Update.
15989 (set_cond_stmt_execution_predicate): Update.
15990 (set_switch_stmt_execution_predicate): Update.
15991 (compute_bb_predicates): Update.
15992 (will_be_nonconstant_expr_predicate): Update.
15993 (will_be_nonconstant_predicate): Update.
15994 (phi_result_unknown_predicate): Update.
15995 (predicate_for_phi_result): Update.
15996 (array_index_predicate): Update.
15997 (estimate_function_body_sizes): Update.
15998 (estimate_node_size_and_time): Update.
15999 (estimate_ipcp_clone_size_and_time): Update.
16000 (remap_predicate): Rename to ...
16001 (predicate::remap_after_inlining): ... this one.
16002 (remap_hint_predicate): Update.
16003 (inline_merge_summary): Update.
16004 (inline_update_overall_summary): Update.
16005 (estimate_size_after_inlining): Update.
16006 (read_predicate): Rename to ...
16007 (predicate::stream_in): ... this one.
16008 (read_inline_edge_summary): Update.
16009 (write_predicate): Rename to ...
16010 (predicate::stream_out): ... this one.
16011 (write_inline_edge_summary): Update.
16012 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
16013 (clause_t): Turn to uint32_t
16014 (predicate): Turn to class; implement constructor and operators
16016 (size_time_entry): Update.
16017 (inline_summary): Update.
16018 (inline_edge_summary): Update.
16020 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
16022 * fold-const.c (fold_binary_loc): Move transformation...
16023 * match.pd (C - X CMP X): ... here.
16025 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
16027 * config/sparc/sparc.c (sparc_option_override): Set function
16028 alignment for -mcpu=niagara7 to 64 to match the I$ line.
16029 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
16031 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
16033 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
16035 2017-05-18 Marek Polacek <polacek@redhat.com>
16038 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
16039 (pass_ubsan::execute): Call gimple_assign_single_p instead of
16040 gimple_assign_load_p.
16042 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
16044 PR middle-end/80692
16045 * real.c (do_compare): Give decimal_do_compare preference over
16046 comparing just the signs.
16048 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
16050 * doc/md.texi (Canonicalization of Instructions): Describe the
16051 canonical form of instructions that inherently set a condition
16054 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
16056 PR middle-end/80775
16057 * tree-cfg.c: Move deletion of unreachable case statements to after
16058 the merging of consecutive case labels.
16060 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
16062 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
16063 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
16064 restoring of callee-saved registers.
16066 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
16068 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
16069 * config/visium/visium.c (single_set_and_flags): Likewise.
16070 * config/visium/visium.md (Substitutions): Likewise.
16072 2017-05-17 Martin Liska <mliska@suse.cz>
16074 * cfg.c: Introduce dump_flags_t type and
16075 use it instead of int type.
16077 * cfghooks.c: Likewise.
16078 * cfghooks.h (struct cfg_hooks): Likewise.
16079 * cfgrtl.c: Likewise.
16080 * cfgrtl.h: Likewise.
16081 * cgraph.c (cgraph_node::get_body): Likewise.
16082 * coretypes.h: Likewise.
16083 * domwalk.c: Likewise.
16084 * domwalk.h: Likewise.
16085 * dumpfile.c (struct dump_option_value_info): Likewise.
16086 (dump_enable_all): Likewise.
16087 (dump_switch_p_1): Likewise.
16088 (opt_info_switch_p): Likewise.
16089 * dumpfile.h (enum tree_dump_index): Likewise.
16090 (struct dump_file_info): Likewise.
16091 * genemit.c: Likewise.
16092 * generic-match-head.c: Likewise.
16093 * gengtype.c (open_base_files): Likewise.
16094 * gimple-pretty-print.c: Likewise.
16095 * gimple-pretty-print.h: Likewise.
16096 * graph.c (print_graph_cfg): Likewise.
16097 * graphite-scop-detection.c (dot_all_sese): Likewise.
16098 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
16099 * loop-unroll.c (report_unroll): Likewise.
16100 * passes.c (pass_manager::register_one_dump_file): Likewise.
16101 * print-tree.c: Likewise.
16102 * statistics.c: Likewise.
16103 * tree-cfg.c: Likewise.
16104 * tree-cfg.h: Likewise.
16105 * tree-dfa.c: Likewise.
16106 * tree-dfa.h: Likewise.
16107 * tree-dump.c (dump_function): Likewise.
16108 * tree-dump.h (struct dump_info): Likewise.
16109 * tree-pretty-print.c: Likewise.
16110 * tree-pretty-print.h: Likewise.
16111 * tree-ssa-live.c: Likewise.
16112 * tree-ssa-live.h: Likewise.
16113 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
16114 * tree-vect-loop.c: Likewise.
16115 * tree-vect-slp.c: Likewise.
16117 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
16118 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16120 PR tree-optimization/80457
16121 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
16122 of all arguments to a statement as scalar_to_vec operations.
16123 (vectorizable_call): Adjust call to vect_model_simple_cost for
16125 (vectorizable_conversion): Likewise.
16126 (vectorizable_assignment): Likewise.
16127 (vectorizable_shift): Likewise.
16128 (vectorizable_operation): Likewise.
16129 (vectorizable_comparison): Likewise.
16130 (vect_is_simple_cond): Record the def types for operands.
16131 (vectorizable_condition): Likewise, call vect_model_simple_cost.
16132 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
16133 for statement argument count.
16135 2017-05-16 Carl Love <cel@us.ibm.com>
16137 * config/rs6000/rs6000-c: Add support for built-in functions
16138 vector unsigned long long vec_bperm (vector unsigned long long,
16139 vector unsigned char)
16140 vector signed long long vec_mule (vector signed int,
16142 vector unsigned long long vec_mule (vector unsigned int,
16143 vector unsigned int)
16144 vector signed long long vec_mulo (vector signed int,
16146 vector unsigned long long vec_mulo (vector unsigned int,
16147 vector unsigned int)
16148 vector signed char vec_sldw (vector signed char,
16149 vector signed char,
16151 vector unsigned char vec_sldw (vector unsigned char,
16152 vector unsigned char,
16154 vector signed short vec_sldw (vector signed short,
16155 vector signed short,
16157 vector unsigned short vec_sldw (vector unsigned short,
16158 vector unsigned short,
16160 vector signed int vec_sldw (vector signed int,
16163 vector unsigned int vec_sldw (vector unsigned int,
16164 vector unsigned int,
16166 vector signed long long vec_sldw (vector signed long long,
16167 vector signed long long,
16169 vector unsigned long long vec_sldw (vector unsigned long long,
16170 vector unsigned long long,
16172 * config/rs6000/rs6000-c: Add support for built-in functions
16173 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
16174 * config/rs6000/altivec.h: Add defintion for vec_sldw.
16175 * doc/extend.texi: Update the built-in documentation for the
16176 new built-in functions.
16178 2017-05-16 Marek Polacek <polacek@redhat.com>
16182 * tree.c (save_expr): Don't fold the expression.
16184 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
16186 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
16187 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
16188 and (?*y,m). Update insn attributes.
16190 2017-05-16 Martin Liska <mliska@suse.cz>
16192 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
16193 flags argument of print_gimple_stmt, print_gimple_expr,
16194 print_generic_stmt and print_generic_expr.
16195 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
16196 * coretypes.h: Likewise.
16197 * except.c (dump_eh_tree): Likewise.
16198 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
16199 * gimple-pretty-print.h: Likewise.
16200 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
16201 (backprop::push_to_worklist): Likewise.
16202 (backprop::pop_from_worklist): Likewise.
16203 (backprop::process_use): Likewise.
16204 (backprop::intersect_uses): Likewise.
16205 (note_replacement): Likewise.
16206 * gimple-ssa-store-merging.c
16207 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
16208 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
16209 (pass_store_merging::execute): Likewise.
16210 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
16211 (ssa_base_cand_dump_callback): Likewise.
16212 (dump_incr_vec): Likewise.
16213 (replace_refs): Likewise.
16214 (replace_mult_candidate): Likewise.
16215 (create_add_on_incoming_edge): Likewise.
16216 (create_phi_basis): Likewise.
16217 (insert_initializers): Likewise.
16218 (all_phi_incrs_profitable): Likewise.
16219 (introduce_cast_before_cand): Likewise.
16220 (replace_one_candidate): Likewise.
16221 * gimplify.c (gimplify_expr): Likewise.
16222 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
16223 (set_rename): Likewise.
16224 (rename_uses): Likewise.
16225 (copy_loop_phi_nodes): Likewise.
16226 (add_close_phis_to_merge_points): Likewise.
16227 (copy_loop_close_phi_args): Likewise.
16228 (copy_cond_phi_args): Likewise.
16229 (graphite_copy_stmts_from_block): Likewise.
16230 (translate_pending_phi_nodes): Likewise.
16231 * graphite-poly.c (print_pdr): Likewise.
16232 (dump_gbb_cases): Likewise.
16233 (dump_gbb_conditions): Likewise.
16234 (print_scop_params): Likewise.
16235 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
16236 (build_cross_bb_scalars_use): Likewise.
16237 (gather_bbs::before_dom_children): Likewise.
16238 * hsa-dump.c (dump_hsa_immed): Likewise.
16239 * ipa-cp.c (print_ipcp_constant_value): Likewise.
16240 (get_replacement_map): Likewise.
16241 * ipa-inline-analysis.c (dump_condition): Likewise.
16242 (estimate_function_body_sizes): Likewise.
16243 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
16244 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
16245 * ipa-prop.c (ipa_dump_param): Likewise.
16246 (ipa_print_node_jump_functions_for_edge): Likewise.
16247 (ipa_modify_call_arguments): Likewise.
16248 (ipa_modify_expr): Likewise.
16249 (ipa_dump_param_adjustments): Likewise.
16250 (ipa_dump_agg_replacement_values): Likewise.
16251 (ipcp_modif_dom_walker::before_dom_children): Likewise.
16252 * ipa-pure-const.c (check_stmt): Likewise.
16253 (pass_nothrow::execute): Likewise.
16254 * ipa-split.c (execute_split_functions): Likewise.
16255 * omp-offload.c (dump_oacc_loop_part): Likewise.
16256 (dump_oacc_loop): Likewise.
16257 * trans-mem.c (tm_log_emit): Likewise.
16258 (tm_memopt_accumulate_memops): Likewise.
16259 (dump_tm_memopt_set): Likewise.
16260 (dump_tm_memopt_transform): Likewise.
16261 * tree-cfg.c (gimple_verify_flow_info): Likewise.
16262 (print_loop): Likewise.
16263 * tree-chkp-opt.c (chkp_print_addr): Likewise.
16264 (chkp_gather_checks_info): Likewise.
16265 (chkp_get_check_result): Likewise.
16266 (chkp_remove_check_if_pass): Likewise.
16267 (chkp_use_outer_bounds_if_possible): Likewise.
16268 (chkp_reduce_bounds_lifetime): Likewise.
16269 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
16270 (chkp_mark_completed_bounds): Likewise.
16271 (chkp_register_incomplete_bounds): Likewise.
16272 (chkp_mark_invalid_bounds): Likewise.
16273 (chkp_maybe_copy_and_register_bounds): Likewise.
16274 (chkp_build_returned_bound): Likewise.
16275 (chkp_get_bound_for_parm): Likewise.
16276 (chkp_build_bndldx): Likewise.
16277 (chkp_get_bounds_by_definition): Likewise.
16278 (chkp_generate_extern_var_bounds): Likewise.
16279 (chkp_get_bounds_for_decl_addr): Likewise.
16280 * tree-chrec.c (chrec_apply): Likewise.
16281 * tree-data-ref.c (dump_data_reference): Likewise.
16282 (dump_subscript): Likewise.
16283 (dump_data_dependence_relation): Likewise.
16284 (analyze_overlapping_iterations): Likewise.
16285 * tree-inline.c (expand_call_inline): Likewise.
16286 (tree_function_versioning): Likewise.
16287 * tree-into-ssa.c (dump_defs_stack): Likewise.
16288 (dump_currdefs): Likewise.
16289 (dump_names_replaced_by): Likewise.
16290 (dump_update_ssa): Likewise.
16291 (update_ssa): Likewise.
16292 * tree-object-size.c (pass_object_sizes::execute): Likewise.
16293 * tree-parloops.c (build_new_reduction): Likewise.
16294 (try_create_reduction_list): Likewise.
16295 (ref_conflicts_with_region): Likewise.
16296 (oacc_entry_exit_ok_1): Likewise.
16297 (oacc_entry_exit_single_gang): Likewise.
16298 * tree-pretty-print.h: Likewise.
16299 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
16300 (get_scalar_evolution): Likewise.
16301 (add_to_evolution): Likewise.
16302 (get_loop_exit_condition): Likewise.
16303 (analyze_evolution_in_loop): Likewise.
16304 (analyze_initial_condition): Likewise.
16305 (analyze_scalar_evolution): Likewise.
16306 (instantiate_scev): Likewise.
16307 (number_of_latch_executions): Likewise.
16308 (gather_chrec_stats): Likewise.
16309 (final_value_replacement_loop): Likewise.
16310 (scev_const_prop): Likewise.
16311 * tree-sra.c (dump_access): Likewise.
16312 (disqualify_candidate): Likewise.
16313 (create_access): Likewise.
16314 (reject): Likewise.
16315 (maybe_add_sra_candidate): Likewise.
16316 (create_access_replacement): Likewise.
16317 (analyze_access_subtree): Likewise.
16318 (analyze_all_variable_accesses): Likewise.
16319 (sra_modify_assign): Likewise.
16320 (initialize_constant_pool_replacements): Likewise.
16321 (find_param_candidates): Likewise.
16322 (decide_one_param_reduction): Likewise.
16323 (replace_removed_params_ssa_names): Likewise.
16324 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
16325 * tree-ssa-copy.c (dump_copy_of): Likewise.
16326 (copy_prop_visit_cond_stmt): Likewise.
16327 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
16328 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
16329 (record_equivalences_from_stmt): Likewise.
16330 * tree-ssa-dse.c (compute_trims): Likewise.
16331 (delete_dead_call): Likewise.
16332 (delete_dead_assignment): Likewise.
16333 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
16334 (forward_propagate_into_cond): Likewise.
16335 (pass_forwprop::execute): Likewise.
16336 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
16337 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
16339 (move_computations_worker): Likewise.
16340 (execute_sm): Likewise.
16341 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
16342 (remove_exits_and_undefined_stmts): Likewise.
16343 (remove_redundant_iv_tests): Likewise.
16344 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
16345 (adjust_iv_update_pos): Likewise.
16346 * tree-ssa-math-opts.c (bswap_replace): Likewise.
16347 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
16348 (value_replacement): Likewise.
16349 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
16350 * tree-ssa-pre.c (print_pre_expr): Likewise.
16351 (get_representative_for): Likewise.
16352 (create_expression_by_pieces): Likewise.
16353 (insert_into_preds_of_block): Likewise.
16354 (eliminate_insert): Likewise.
16355 (eliminate_dom_walker::before_dom_children): Likewise.
16356 (eliminate): Likewise.
16357 (remove_dead_inserted_code): Likewise.
16358 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
16359 * tree-ssa-reassoc.c (get_rank): Likewise.
16360 (eliminate_duplicate_pair): Likewise.
16361 (eliminate_plus_minus_pair): Likewise.
16362 (eliminate_not_pairs): Likewise.
16363 (undistribute_ops_list): Likewise.
16364 (eliminate_redundant_comparison): Likewise.
16365 (update_range_test): Likewise.
16366 (optimize_range_tests_var_bound): Likewise.
16367 (optimize_vec_cond_expr): Likewise.
16368 (rewrite_expr_tree): Likewise.
16369 (rewrite_expr_tree_parallel): Likewise.
16370 (linearize_expr): Likewise.
16371 (break_up_subtract): Likewise.
16372 (linearize_expr_tree): Likewise.
16373 (attempt_builtin_powi): Likewise.
16374 (attempt_builtin_copysign): Likewise.
16375 (transform_stmt_to_copy): Likewise.
16376 (transform_stmt_to_multiply): Likewise.
16377 (dump_ops_vector): Likewise.
16378 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
16379 (print_scc): Likewise.
16380 (set_ssa_val_to): Likewise.
16381 (visit_reference_op_store): Likewise.
16382 (visit_use): Likewise.
16383 (sccvn_dom_walker::before_dom_children): Likewise.
16384 (run_scc_vn): Likewise.
16385 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
16387 (expr_hash_elt::print): Likewise.
16388 (const_and_copies::pop_to_marker): Likewise.
16389 (const_and_copies::record_const_or_copy_raw): Likewise.
16390 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
16391 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
16392 (dump_predicates): Likewise.
16393 (find_uninit_use): Likewise.
16394 (warn_uninitialized_phi): Likewise.
16395 (pass_late_warn_uninitialized::execute): Likewise.
16396 * tree-ssa.c (verify_vssa): Likewise.
16397 (verify_ssa): Likewise.
16398 (maybe_optimize_var): Likewise.
16399 * tree-vrp.c (dump_value_range): Likewise.
16400 (dump_all_value_ranges): Likewise.
16401 (dump_asserts_for): Likewise.
16402 (register_edge_assert_for_2): Likewise.
16403 (vrp_visit_cond_stmt): Likewise.
16404 (vrp_visit_switch_stmt): Likewise.
16405 (vrp_visit_stmt): Likewise.
16406 (vrp_visit_phi_node): Likewise.
16407 (simplify_cond_using_ranges_1): Likewise.
16408 (fold_predicate_in): Likewise.
16409 (evrp_dom_walker::before_dom_children): Likewise.
16410 (evrp_dom_walker::push_value_range): Likewise.
16411 (evrp_dom_walker::pop_value_range): Likewise.
16412 (execute_early_vrp): Likewise.
16414 2017-05-16 Richard Biener <rguenther@suse.de>
16416 * dwarf2out.c (loc_list_from_tree_1): Do not create
16417 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
16419 2017-05-16 Richard Biener <rguenther@suse.de>
16421 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
16423 (note_variable_value_in_expr): If we resolved the decl ref
16424 do not push to the stack.
16426 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
16428 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
16429 operations in fast-math mode.
16430 (vaddq_f16): Likewise.
16431 (vmul_f16): Likewise.
16432 (vmulq_f16): Likewise.
16433 (vsub_f16): Likewise.
16434 (vsubq_f16): Likewise.
16435 * config/arm/neon.md (add<mode>3): New.
16437 (fma:<VH:mode>3): New. Also remove outdated comment.
16440 2017-05-16 Martin Liska <mliska@suse.cz>
16444 * ipa-devirt.c (warn_types_mismatch): Fix typo.
16445 (odr_types_equivalent_p): Likewise.
16447 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
16449 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
16451 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
16454 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
16455 non-interunit SSE move alternatives with '?'.
16456 (zero-extendsidi peephole2): New peephole to skip intermediate
16457 general register in SSE zero-extend sequence.
16459 2017-05-15 Jeff Law <law@redhat.com>
16461 * reorg.c (relax_delay_slots): Create a new variable to hold
16462 the temporary target rather than clobbering TARGET_LABEL.
16464 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
16465 missing argument to extract_bit_field call.
16466 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
16468 2017-05-15 Martin Liska <mliska@suse.cz>
16471 * gcc.c (process_command): Do not allow empty argument of -o option.
16473 2017-05-15 Renlin Li <renlin.li@arm.com>
16475 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
16476 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
16477 * config/aarch64/constraints.md (Usf): Add long call check.
16478 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
16479 (call_value): Likewise.
16480 (sibcall): Likewise.
16481 (sibcall_value): Likewise.
16483 (call_value_insn): New.
16484 (sibcall_insn): Update rtx pattern.
16485 (sibcall_value_insn): Likewise.
16486 (call_internal): Remove.
16487 (call_value_internal): Likewise.
16488 (sibcall_internal): Likewise.
16489 (sibcall_value_internal): Likewise.
16490 (call_reg): Likewise.
16491 (call_symbol): Likewise.
16492 (call_value_reg): Likewise.
16493 (call_value_symbol): Likewise.
16495 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
16498 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
16500 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
16502 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
16503 compatible with CCGOCmode and with CCZmode.
16505 2017-05-14 Martin Sebor <msebor@redhat.com>
16507 PR middle-end/77671
16508 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
16509 (gimple_fold_builtin_snprintf): Same.
16510 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
16511 (gimple_fold_builtin_snprintf): Same.
16512 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
16513 of character types.
16514 (is_call_safe): New function.
16515 (try_substitute_return_value): Call it.
16516 (try_simplify_call): New function.
16517 (pass_sprintf_length::handle_gimple_call): Call it.
16519 2017-05-14 Martin Sebor <msebor@redhat.com>
16521 PR middle-end/80669
16522 * builtins.c (expand_builtin_stpncpy): Simplify.
16524 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
16526 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
16527 * config/i386/i386.h
16528 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
16529 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
16530 (struct machine_function): Add new members call_ms2sysv,
16531 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
16532 (struct machine_frame_state): New fields sp_realigned and
16533 sp_realigned_offset.
16534 * config/i386/i386.c
16535 (enum xlogue_stub): New enum.
16536 (enum xlogue_stub_sets): New enum.
16537 (class xlogue_layout): New class.
16538 (struct ix86_frame): New fields stack_realign_allocate_offset,
16539 stack_realign_offset and outlined_save_offset. Modify comments to
16540 detail stack layout when using out-of-line stubs.
16541 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
16542 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
16543 -mcall-ms2sysv-xlogues.
16544 (stub_managed_regs): New static variable.
16545 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
16546 registers managed by out-of-line stub.
16547 (disable_call_ms2sysv_xlogues): New function.
16548 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
16549 m->call_ms2sysv when appropriate and compute frame layout for
16551 (sp_valid_at, fp_valid_at): New inline functions.
16552 (choose_basereg): New function.
16553 (choose_baseaddr): Add align parameter, use choose_basereg and modify
16555 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
16556 Use align parameter of choose_baseaddr to generated aligned SSE movs
16558 (pro_epilogue_adjust_stack): Modify to track
16559 machine_frame_state::sp_realigned.
16560 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
16561 (ix86_nsaved_sseregs): Likewise.
16562 (ix86_emit_save_regs): Likewise.
16563 (ix86_emit_save_regs_using_mov): Likewise.
16564 (ix86_emit_save_sse_regs_using_mov): Likewise.
16565 (get_scratch_register_on_entry): Likewise.
16566 (gen_frame_set): New function.
16567 (gen_frame_load): Likewise.
16568 (gen_frame_store): Likewise.
16569 (emit_outlined_ms2sysv_save): Likewise.
16570 (emit_outlined_ms2sysv_restore): Likewise.
16571 (ix86_expand_prologue): Modify stack re-alignment code and call
16572 emit_outlined_ms2sysv_save when appropriate.
16573 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
16574 parameter rtx_insn *insn, which allows the function to be used to only
16575 generate the notes.
16576 (ix86_expand_epilogue): Modify validity checks of frame and stack
16577 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
16578 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
16579 * config/i386/predicates.md
16580 (save_multiple): New predicate.
16581 (restore_multiple): Likewise.
16582 * config/i386/sse.md
16583 (save_multiple<mode>): New pattern.
16584 (save_multiple_realign<mode>): Likewise.
16585 (restore_multiple<mode>): Likewise.
16586 (restore_multiple_and_return<mode>): Likewise.
16587 (restore_multiple_leave_return<mode>): Likewise.
16588 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
16590 2017-05-14 Julia Koval <julia.koval@intel.com>
16592 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
16593 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
16594 (__builtin_ia32_xsetbv): New builtins.
16595 * config/i386/i386.c (ix86_expand_special_args_builtin):
16597 (ix86_expand_builtin): Special expand for new intrinsics.
16598 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
16599 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
16600 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
16602 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16604 * cfganal.c (inverted_post_order_compute): Change argument type
16606 * cfganal.h (inverted_post_order_compute): Adjust prototype.
16607 * df-core.c (rest_of_handle_df_initialize): Adjust.
16608 (rest_of_handle_df_finish): Likewise.
16609 (df_analyze_1): Likewise.
16610 (df_analyze): Likewise.
16611 (loop_inverted_post_order_compute): Change argument to be a vec *.
16612 (df_analyze_loop): Adjust.
16613 (df_get_n_blocks): Likewise.
16614 (df_get_postorder): Likewise.
16615 * df.h (struct df_d): Change field to be a vec.
16616 * lcm.c (compute_laterin): Adjust.
16617 (compute_available): Likewise.
16618 * lra-lives.c (lra_create_live_ranges_1): Likewise.
16619 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
16620 * tree-ssa-pre.c (compute_antic): Likewise.
16622 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16624 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
16625 (depth_first_search::depth_first_search): Change structure init
16626 function to this constructor.
16627 (depth_first_search::add_bb): Rename function to this member.
16628 (depth_first_search::execute): Likewise.
16629 (flow_dfs_compute_reverse_finish): Adjust.
16631 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16633 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
16634 (longest_simple_path): Likewise.
16635 * shrink-wrap.c (spread_components): Likewise.
16636 (disqualify_problematic_components): Likewise.
16637 (emit_common_heads_for_components): Likewise.
16638 (emit_common_tails_for_components): Likewise.
16639 (insert_prologue_epilogue_for_components): Likewise.
16641 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16643 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
16646 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16648 * df-core.c (df_set_blocks): Start using auto_bitmap.
16649 (df_compact_blocks): Likewise.
16650 * df-problems.c (df_rd_confluence_n): Likewise.
16651 * df-scan.c (df_insn_rescan_all): Likewise.
16652 (df_process_deferred_rescans): Likewise.
16653 (df_update_entry_block_defs): Likewise.
16654 (df_update_exit_block_uses): Likewise.
16655 (df_entry_block_bitmap_verify): Likewise.
16656 (df_exit_block_bitmap_verify): Likewise.
16657 (df_scan_verify): Likewise.
16658 * lra-constraints.c (lra_constraints): Likewise.
16659 (undo_optional_reloads): Likewise.
16660 (lra_undo_inheritance): Likewise.
16661 * lra-remat.c (calculate_gen_cands): Likewise.
16662 (do_remat): Likewise.
16663 * lra-spills.c (assign_spill_hard_regs): Likewise.
16664 (spill_pseudos): Likewise.
16665 * tree-ssa-pre.c (bitmap_set_and): Likewise.
16666 (bitmap_set_subtract_values): Likewise.
16668 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16670 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
16671 management with auto_bitmap.
16672 (fix_inter_tick): Likewise.
16673 (fix_recovery_deps): Likewise.
16674 * ira.c (add_store_equivs): Likewise.
16675 (find_moveable_pseudos): Likewise.
16676 (split_live_ranges_for_shrink_wrap): Likewise.
16677 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
16678 (rtx_reuse_manager::seen_def_p): Likewise.
16679 (rtx_reuse_manager::set_seen_def): Likewise.
16680 * print-rtl.h (class rtx_reuse_manager): Likewise.
16682 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16684 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
16686 (migrate_btr_def): Likewise.
16687 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
16688 * df-core.c (loop_post_order_compute): Likewise.
16689 (loop_inverted_post_order_compute): Likewise.
16690 * hsa-common.h: Likewise.
16691 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
16692 * init-regs.c (initialize_uninitialized_regs): Likewise.
16693 * ipa-inline.c (resolve_noninline_speculation): Likewise.
16694 (inline_small_functions): Likewise.
16695 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
16696 * ira.c (combine_and_move_insns): Likewise.
16697 (build_insn_chain): Likewise.
16698 * loop-invariant.c (find_invariants): Likewise.
16699 * lower-subreg.c (propagate_pseudo_copies): Likewise.
16700 * predict.c (tree_predict_by_opcode): Likewise.
16701 (predict_paths_leading_to): Likewise.
16702 (predict_paths_leading_to_edge): Likewise.
16703 (estimate_loops_at_level): Likewise.
16704 (estimate_loops): Likewise.
16705 * shrink-wrap.c (try_shrink_wrapping): Likewise.
16706 (spread_components): Likewise.
16707 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
16708 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
16709 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
16710 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
16711 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
16712 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
16713 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
16714 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
16715 (mark_threaded_blocks): Likewise.
16716 (thread_through_all_blocks): Likewise.
16717 * tree-ssa.c (verify_ssa): Likewise.
16718 (execute_update_addresses_taken): Likewise.
16719 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
16721 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16723 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
16725 (post_order_compute): Likewise.
16726 (inverted_post_order_compute): Likewise.
16727 (pre_and_rev_post_order_compute_fn): Likewise.
16729 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16731 * genrecog.c (int_set::int_set): Explicitly construct our
16732 auto_vec base class.
16733 * vec.h (auto_vec::auto_vec): New constructor.
16735 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16737 * bitmap.h (class auto_bitmap): New constructor taking
16738 bitmap_obstack * argument.
16740 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16742 * bitmap.h (class auto_bitmap): Change type of m_bits to
16743 bitmap_head, and adjust ctor / dtor and member operators.
16745 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
16747 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
16748 when returned register mode doesn't match original mode.
16750 2017-05-12 Jeff Law <law@redhat.com>
16751 Jakub Jelinek <jakub@redhat.com>
16753 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
16754 we look for cc setter after the compare-elim changes.
16755 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
16756 within the vector to match what compare-elim now expects.
16757 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
16758 (xorsi3_flags, one_cmplsi2_flags): Likewise.
16760 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
16761 after the compare-elim changes.
16762 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
16763 the vector to match what compare-elim now expects.
16764 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
16765 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
16766 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
16767 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
16768 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
16770 * config/visium/visium.c (single_set_and_flags): Fix where
16771 we look for cc setter after the compare-elim changes.
16772 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
16773 with the vector to match what compare-elim now expects.
16774 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
16775 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
16776 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
16777 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
16778 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
16779 (neg<mode>2_insn_set_overflow): Likewise.
16781 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
16783 PR middle-end/79794
16784 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
16785 maybe_expand_insn call, set ops[0].target. If still set after call,
16786 set alt_rtl. Add extra arg to recursive calls.
16787 (extract_bit_field): Add alt_rtl argument. Pass to
16789 * expmed.h (extract_bit_field): Fix prototype.
16790 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
16791 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
16792 to extract_bit_field_calls.
16793 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
16794 Pass alt_rtl to extract_bit_field calls.
16795 * calls.c (store_unaligned_arguments_into_psuedos)
16796 load_register_parameters): Pass extra NULL to extract_bit_field calls.
16797 * optabs.c (maybe_legitimize_operand): Clear op->target when call
16799 * optabs.h (struct expand_operand): Add target bitfield.
16801 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
16803 * compare-elim.c (try_eliminate_compare): Canonicalize
16804 operation with embedded compare to
16805 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
16806 (set (reg) (operation)].
16808 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
16810 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
16813 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
16814 cost of adding a carry flag for ADC instruction.
16815 [case MINUS]: Ignore the cost of subtracting a carry flag
16816 for SBB instruction.
16818 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
16820 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
16822 * config/rs6000/bmiintrin.h: New file.
16823 * config/rs6000/bmi2intrin.h: New file.
16824 * config/rs6000/x86intrin.h: New file.
16826 2017-05-12 Jeff Law <law@redhat.com>
16828 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
16831 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
16833 PR middle-end/80707
16834 * tree-cfg.c: Remove cfg edges of unreachable case statements.
16836 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
16838 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
16839 early expansion of vector divide builtins.
16840 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
16841 builtins identified as having unsigned arguments.
16843 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
16845 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
16846 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
16847 expansion of vector logical operations (and, andc, or, xor,
16850 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
16852 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
16853 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
16855 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
16857 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
16858 early GIMPLE expansion of vector multiplies.
16860 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
16862 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
16863 TARGET_HAVE_MOVT conditional.
16864 (movt splitter): Likewise.
16866 2017-05-12 Richard Biener <rguenther@suse.de>
16868 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
16869 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
16870 Fold all stmts not inplace.
16872 2017-05-12 Richard Biener <rguenther@suse.de>
16874 PR tree-optimization/80713
16875 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
16876 inserted_exprs bit for not removed stmts.
16878 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
16880 PR middle-end/69921
16881 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
16882 parallelized" attribute for parallelized OpenACC kernels.
16883 * omp-offload.c (execute_oacc_device_lower): Use it.
16885 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
16886 Set "oacc kernels" attribute.
16887 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
16888 parameter. Adjust all users.
16889 (oacc_fn_attrib_kernels_p): Remove function.
16890 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
16891 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
16892 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
16893 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
16894 assert "oacc kernels" attribute is set.
16896 2017-05-11 Carl Love <cel@us.ibm.com>
16898 * config/rs6000/rs6000-c: Add support for built-in functions
16899 vector unsigned char vec_popcnt (vector signed char)
16900 vector unsigned char vec_popcnt (vector unsigned char)
16901 vector unsigned short vec_popcnt (vector signed short)
16902 vector unsigned short vec_popcnt (vector unsigned short)
16903 vector unsigned int vec_popcnt (vector signed int)
16904 vector unsigned int vec_popcnt (vector unsigned int)
16905 vector unsigned long long vec_popcnt (vector signed long long)
16906 vector unsigned long long vec_popcnt (vector unsigned long long)
16907 vector signed long long vec_slo (vector signed long long,
16908 vector signed char)
16909 vector signed long long vec_slo (vector signed long long,
16910 vector unsigned char)
16911 vector unsigned long long vec_slo (vector unsigned long long,
16912 vector signed char)
16913 vector unsigned long long vec_slo (vector unsigned long long,
16914 vector unsigned char)
16915 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
16916 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
16917 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
16918 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
16919 * doc/extend.texi: Update the built-in documentation file for the
16920 new built-in functions.
16922 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
16924 * attribs.h (sorted_attr_string): Move machine independent
16925 functions for target clone support from the i386 port to common
16926 code. Rename ix86_function_versions to common_function_versions.
16927 Rename make_name to make_unique_name.
16928 (common_function_versions): Likewise.
16929 (make_unique_name): Likewise.
16930 (make_dispatcher_decl): Likewise.
16931 (is_function_default_version): Likewise.
16932 * attribs.c (attr_strcmp): Likewise.
16933 (sorted_attr_string): Likewise.
16934 (common_function_versions): Likewise.
16935 (make_unique_name): Likewise.
16936 (make_dispatcher_decl): Likewise.
16937 (is_function_default_version): Likewise.
16938 * config/i386/i386.c (attr_strcmp): Likewise.
16939 (sorted_attr_string): Likewise.
16940 (ix86_function_versions): Likewise.
16941 (make_name): Likewise.
16942 (make_dispatcher_decl): Likewise.
16943 (is_function_default_version): Likewise.
16944 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
16946 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16949 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
16950 Account for direct move costs for vec_construct of integer
16953 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
16956 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
16957 (UNSPEC_STX_ATOMIC): Ditto.
16958 (loaddi_via_sse): New insn.
16959 (storedi_via_sse): Ditto.
16960 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
16961 Update corresponding peephole2 patterns.
16962 (atomic_storedi_fpu): Ditto.
16964 2017-05-11 Julia Koval <julia.koval@intel.com>
16966 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
16967 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
16969 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
16970 (__builtin_ia32_rsqrt14ss_mask): New builtins.
16971 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
16973 2017-05-11 Nathan Sidwell <nathan@acm.org>
16975 * graphite-poly.c: Include dumpfile.h.
16977 * dumpfle.h (dump_function): Declare here ...
16978 * tree-dump.h (dump_function): ... not here.
16979 * dumpfile.c: #include tree-cfg.h.
16980 (dump_function): Move here from ...
16981 * tree-dump.c (dump_function): ... here.
16982 * gimplify.c: #include splay-tree.h, not tree-dump.h.
16983 * graphite-poly.c: Don't include tree-dump.h.
16984 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
16985 * print-tree.c: Likewise.
16986 * stor-layout.c: Likewise.
16987 * tree-nested.c: Likewise.
16989 * dumpfile.c (dump_start): Use TDF_FLAGS.
16990 (dump_enable_all): Fix TDF_KIND check thinko.
16992 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
16994 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16995 array entries to represent two legal parameterizations of the
16996 overloaded __builtin_cmpb function, as represented by the
16997 P6_OV_BUILTIN_CMPB constant.
16998 (altivec_resolve_overloaded_builtin): Add special case handling
16999 for the __builtin_cmpb function, as represented by the
17000 P6_OV_BUILTIN_CMPB constant.
17001 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
17002 (BU_P6_64BIT_2): New macro.
17003 (BU_P6_OVERLOAD_2): New macro
17004 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
17005 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
17006 (CMPB): Add overload support to represent both 32-bit and 64-bit
17007 compare-bytes function.
17008 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
17009 support for TARGET_CMPB.
17010 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
17011 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
17012 documentation of the __builtin_cmpb overloaded built-in function.
17014 2017-05-11 Richard Biener <rguenther@suse.de>
17016 PR tree-optimization/80705
17017 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
17018 bases are not vectorizable.
17020 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17022 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
17023 when counting register pressure.
17025 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17027 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
17028 (iv_ca_more_deps): Renamed to ...
17029 (iv_ca_compare_deps): ... this.
17030 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
17032 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17034 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
17036 (determine_group_iv_costs): ... here.
17037 (find_inv_vars_cb): Record inv var if it's not recorded before.
17039 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17041 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
17042 (get_shiftadd_cost): Ditto.
17044 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17046 * tree-ssa-address.c: Include header file.
17047 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
17049 (add_to_parts): Refactor.
17050 (addr_to_parts): New parameter. Update use of move_hint_to_base.
17051 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
17054 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17056 PR tree-optimization/53090
17057 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
17059 (extract_cond_operands): Detect condition with IV on both sides
17060 and return COMP_IV_EXPR_2.
17061 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
17062 (rewrite_use_compare): Simplify by removing call to function
17063 extract_cond_operands.
17065 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17067 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
17068 (extract_cond_operands): Detect condition comparing against non-
17069 invariant bound and return appropriate enum value.
17070 (find_interesting_uses_cond): Update use of extract_cond_operands.
17071 Handle its return value accordingly.
17072 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
17074 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17076 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
17077 nonlinear iv_use computation in loop invariant sensitive way.
17079 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17081 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
17082 (find_iv_candidates): Call relate_compare_use_with_all_cands.
17084 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17086 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
17087 (dump_cand): Support iv_cand.inv_exprs.
17088 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
17090 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
17093 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17095 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
17097 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
17098 as local function. Include necessary header files.
17099 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
17101 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17103 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
17105 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17107 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
17108 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
17109 RSHIFT_EXPR and BIT_NOT_EXPR.
17111 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17113 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
17114 (adjust_setup_cost): New parameter supporting round up adjustment.
17115 (struct address_cost_data): Delete.
17116 (force_expr_to_var_cost): Don't bound cost with spill_cost.
17117 (split_address_cost, ptr_difference_cost): Delete.
17118 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
17119 (struct ainc_cost_data): New struct.
17120 (get_address_cost_ainc): New function.
17121 (get_address_cost, get_computation_cost): Reimplement.
17122 (determine_group_iv_cost_address): Record inv_expr for all uses of
17124 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
17125 (iv_ca_has_deps): Reimplemented to ...
17126 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
17128 (iv_ca_extend): Call iv_ca_more_deps.
17130 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17132 * tree-ssa-address.c (struct mem_address): Move to header file.
17133 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
17134 * tree-ssa-address.h (struct mem_address): Move from C file.
17135 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
17137 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17139 * tree-affine.h (aff_combination_type): New interface.
17140 (aff_combination_zero_p): Remove static.
17141 (aff_combination_const_p): New interface.
17142 (aff_combination_singleton_var_p): New interfaces.
17144 2017-05-11 Richard Biener <rguenther@suse.de>
17146 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
17147 Skip unreachable blocks and destinations.
17148 (eliminate): Move stmt removal and fixup ...
17149 (fini_eliminate): ... here. Skip inserted exprs.
17150 (pass_pre::execute): Move fini_pre after fini_eliminate.
17151 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
17152 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
17153 PRE to get rid of dead code that has invalid SSA form and
17154 split critical edges again.
17156 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17158 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
17160 2017-05-11 Richard Biener <rguenther@suse.de>
17162 * passes.c (execute_function_todo): Verify loops if they are
17163 said to be up-to-date.
17164 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
17165 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
17167 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
17170 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
17171 handle calling assemble_external ourself.
17174 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
17175 modes with zero size. Enhance comment.
17177 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17179 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
17180 built-ins for vec_xl and vec_xst with short and char pointer
17183 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
17185 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
17186 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
17187 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
17188 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
17189 (_mm_maskz_min_round_ss): New intrinsics.
17190 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
17191 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
17192 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
17193 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
17194 (__builtin_ia32_minss_mask_round): New builtins.
17195 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
17196 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
17197 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
17199 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
17200 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
17202 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
17205 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
17207 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
17208 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
17209 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
17210 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
17211 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
17212 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
17213 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
17214 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
17215 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
17216 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
17217 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
17218 (__builtin_ia32_mulss_mask_round): New builtins.
17219 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
17220 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
17221 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
17223 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
17224 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
17226 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
17229 2017-05-10 Julia Koval <julia.koval@intel.com>
17231 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
17232 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
17233 (_mm256_setr_m128i): New intrinsics.
17235 2017-05-10 Julia Koval <julia.koval@intel.com>
17237 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
17238 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
17239 (_mm_maskz_rcp14_ss): New intrinsics.
17240 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
17241 (__builtin_ia32_rcp14ss_mask): New builtins.
17242 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
17244 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
17246 PR tree-optimization/51513
17247 * tree-cfg.c (gimple_seq_unreachable_p): New function.
17248 (assert_unreachable_fallthru_edge_p): Use it.
17249 (group_case_labels_stmt): Likewise.
17250 * tree-cfg.h: Prototype it.
17251 * stmt.c: Include cfghooks.h and tree-cfg.h.
17252 (emit_case_dispatch_table) <gap_label>: New local variable.
17253 Use it to fill dispatch table gaps.
17254 Test for default_label before updating probabilities.
17255 (expand_case) <default_label>: Remove unneeded initialization.
17256 Test for unreachable default case statement and remove its edge.
17257 Set default_label accordingly.
17258 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
17260 2017-05-10 Carl Love <cel@us.ibm.com>
17262 * config/rs6000/rs6000-c: Add support for built-in functions
17263 vector signed char vec_neg (vector signed char)
17264 vector signed short int vec_neg (vector short int)
17265 vector signed int vec_neg (vector signed int)
17266 vector signed long long vec_neg (vector signed long long)
17267 vector float vec_neg (vector float)
17268 vector double vec_neg (vector double)
17269 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
17271 * config/rs6000/altivec.h: Add define for vec_neg
17272 * doc/extend.texi: Update the built-in documentation for the
17273 new built-in functions.
17275 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17277 PR tree-optimization/77644
17278 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
17280 2017-05-10 Nathan Sidwell <nathan@acm.org>
17282 * dumpfile.h (TDI_lang_all): New.
17283 (TDF_KIND): New. Renumber others
17284 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
17286 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
17288 (get_dump_file_name): Adjust suffix generation.
17289 (dump_enable_all): Use TDF_KIND.
17290 * doc/invoke.texi (-fdump-lang-all): Document.
17292 * dumpfile.h: Tabify.
17294 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
17297 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
17298 Move member access before delete.
17300 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
17302 * tree-inline.c (expand_call_inline): Split block at stmt
17305 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
17308 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
17309 are now unused after splitting mov{sf,sd}_hardfloat.
17310 (f32_lr2): Likewise.
17311 (f32_lm): Likewise.
17312 (f32_lm2): Likewise.
17313 (f32_li): Likewise.
17314 (f32_li2): Likewise.
17315 (f32_lv): Likewise.
17316 (f32_sr): Likewise.
17317 (f32_sr2): Likewise.
17318 (f32_sm): Likewise.
17319 (f32_sm2): Likewise.
17320 (f32_si): Likewise.
17321 (f32_si2): Likewise.
17322 (f32_sv): Likewise.
17323 (f32_dm): Likewise.
17324 (f32_vsx): Likewise.
17325 (f32_av): Likewise.
17326 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
17327 For movsf, order stores so the VSX stores occur before the GPR
17328 store which encourages the register allocator to use a traditional
17329 FPR instead of a GPR. For movsd, order the stores so that the GPR
17330 store comes before the VSX stores to allow the power6 to work.
17331 This is due to the power6 not having a 32-bit integer store
17332 instruction from a FPR.
17333 (movsf_hardfloat): Likewise.
17334 (movsd_hardfloat): Likewise.
17336 2017-05-09 Martin Sebor <msebor@redhat.com>
17338 PR translation/80280
17339 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
17342 PR translation/80280
17343 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
17344 data member added in r247778.
17345 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
17347 2017-05-09 Nathan Sidwell <nathan@acm.org>
17349 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
17351 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
17354 2017-05-09 Marek Polacek <polacek@redhat.com>
17356 * doc/invoke.texi: Fix typo.
17358 2017-05-09 Richard Biener <rguenther@suse.de>
17360 * tree-vrp.c (vrp_val_is_max): Adjust comment.
17361 (vrp_val_is_min): Likewise.
17362 (set_value_range_to_value): Likewise.
17363 (set_value_range_to_nonnegative): Likewise.
17364 (gimple_assign_nonzero_p): Likewise.
17365 (gimple_stmt_nonzero_p): Likewise.
17366 (vrp_int_const_binop): Likewise. Remove unreachable case.
17367 (adjust_range_with_scev): Adjust comments.
17368 (compare_range_with_value): Likewise.
17369 (extract_range_from_phi_node): Likewise.
17370 (test_for_singularity): Likewise.
17372 2017-05-09 Richard Biener <rguenther@suse.de>
17374 * tree-vrp.c (get_single_symbol): Add assert that we don't
17375 get overflowed constants as invariant part.
17376 (compare_values_warnv): Add comment before the TREE_NO_WARNING
17377 checks. Use wi::cmp instead of recursing for integer constants.
17378 (compare_values): Just ignore whether we assumed undefined
17379 overflow instead of failing the compare.
17380 (extract_range_for_var_from_comparison_expr): Add comment before the
17381 TREE_NO_WARNING sets.
17382 (test_for_singularity): Likewise.
17383 (extract_range_from_comparison): Do not disable optimization
17384 when we assumed undefined overflow.
17385 (extract_range_basic): Remove init of unused var.
17387 2017-05-09 Richard Biener <rguenther@suse.de>
17389 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
17390 (extract_range_from_multiplicative_op_1): Adjust.
17391 (extract_range_from_binary_expr_1): Use int_const_binop.
17393 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
17396 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
17397 rs6000_store_data_bypass_p in seven define_bypass directives and
17398 in several comments.
17399 * config/rs6000/rs6000-protos.h: Add prototype for
17400 rs6000_store_data_bypass_p function.
17401 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
17402 function implements slightly different (rs6000-specific) semantics
17403 than store_data_bypass_p, returning false rather than aborting
17404 with assertion error when arguments do not satisfy the
17405 requirements of store data bypass.
17406 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
17407 rs6000_store_data_bypass_p.
17409 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
17411 * config/xtensa/xtensa-protos.h
17412 (xtensa_initial_elimination_offset): New declaration.
17413 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
17414 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
17415 macro definition, add case for FRAME_POINTER_REGNUM when
17416 FRAME_GROWS_DOWNWARD.
17417 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
17418 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
17419 xtensa_initial_elimination_offset.
17421 2017-05-08 Nathan Sidwell <nathan@acm.org>
17423 * doc/invoke.texi: Alphabetize -fdump options.
17425 2017-05-08 Martin Sebor <msebor@redhat.com>
17427 PR translation/80280
17428 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
17430 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
17432 * target.def (compute_frame_layout): New optional target hook.
17433 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
17434 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
17435 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
17437 * reload1.c (verify_initial_elim_offsets): Likewise.
17438 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
17439 (use_simple_return_p): Call arm_compute_frame_layout if needed.
17440 (arm_get_frame_offsets): Split up into this ...
17441 (arm_compute_frame_layout): ... and this function.
17443 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
17445 * config/aarch64/constraints.md (Usa): New constraint.
17446 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
17448 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
17450 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
17451 with_multilib_list after it has been checked.
17453 2017-05-08 Richard Biener <rguenther@suse.de>
17455 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
17456 (bitmap_set_subtract_values): Likewise.
17458 2017-05-08 Richard Biener <rguenther@suse.de>
17460 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
17461 (gimple_assign_nonzero): ... this and remove strict_overflow_p
17463 (gimple_stmt_nonzero_warnv_p): Rename to ...
17464 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
17466 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
17467 (extract_range_basic): Adjust, do not disable propagation on
17468 strict overflow sensitive simplification.
17469 (vrp_visit_cond_stmt): Likewise.
17471 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
17473 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
17474 body size unconditionally.
17476 2017-05-07 Jeff Law <law@redhat.com>
17479 2017-05-06 Jeff Law <law@redhat.com>
17480 PR tree-optimization/78496
17481 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
17484 PR tree-optimization/78496
17485 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
17486 (simplify_stmt_using_ranges): Call it.
17487 (vrp_dom_walker::before_dom_children): Extract equivalences
17488 from an ASSERT_EXPR with an equality comparison against a
17491 2017-05-06 Jeff Law <law@redhat.com>
17493 PR tree-optimization/78496
17494 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
17497 PR tree-optimization/78496
17498 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
17499 (simplify_stmt_using_ranges): Call it.
17500 (vrp_dom_walker::before_dom_children): Extract equivalences
17501 from an ASSERT_EXPR with an equality comparison against a
17504 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
17506 * lra-constraints.c (lra_copy_reg_equiv): New function.
17507 (split_reg): Use it to copy equivalence information from the
17508 original register to the spill register.
17510 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
17512 PR rtl-optimization/75964
17513 * simplify-rtx.c (simplify_const_relational_operation): Remove
17514 invalid handling of comparisons of integer ABS.
17516 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
17518 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
17519 initialize to zero.
17520 (init_regs): Remove declaration.
17521 (function_arg_advance_32): Initialize error_p as boolean variable.
17523 2017-05-05 Nathan Sidwell <nathan@acm.org>
17525 * store-motion.c (remove_reachable_equiv_notes): Reformat long
17526 lines. Use for (;;).
17528 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17530 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
17531 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
17532 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
17533 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
17534 VF=2 that require versioning.
17536 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17538 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
17541 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17543 * diagnostic.h (diagnostic_override_option_index): Convert from
17544 macro to inline function.
17546 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17548 * diagnostic.c (last_module_changed_p): New function.
17549 (set_last_module): New function.
17550 (diagnostic_report_current_module): Convert macro usage to
17551 the above functions.
17552 * diagnostic.h (diagnostic_context::last_module): Strengthen
17553 from const line_map * to const line_map_ordinary *.
17554 (diagnostic_last_module_changed): Delete macro.
17555 (diagnostic_set_last_module): Delete macro.
17557 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17559 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
17560 with diagnostic_report_diagnostic.
17561 (diagnostic_n_impl_richloc): Likewise.
17562 * diagnostic.h (report_diagnostic): Delete macro.
17563 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
17564 with diagnostic_report_diagnostic.
17565 * substring-locations.c (format_warning_va): Likewise.
17567 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17569 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
17570 save/restor of format_spec. Move option-printing code to...
17571 (print_option_information): ...this new function, and
17572 reimplement by simply printing to the pretty_printer,
17573 rather than appending to the format string.
17575 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17577 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
17578 handling logic into...
17579 (update_effective_level_from_pragmas): ...this new function.
17581 2017-05-04 Andrew Waterman <andrew@sifive.com>
17583 * config/riscv/riscv.opt (mstrict-align): New option.
17584 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
17585 (SLOW_UNALIGNED_ACCESS): Define.
17586 (riscv_slow_unaligned_access): Declare.
17587 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
17589 (riscv_slow_unaligned_access): New variable.
17590 (rocket_tune_info): Set slow_unaligned_access to true.
17591 (optimize_size_tune_info): Set slow_unaligned_access to false.
17592 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
17593 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
17594 (riscv_option_override): Set riscv_slow_unaligned_access.
17595 * doc/invoke.texi: Add -mstrict-align to RISC-V.
17597 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
17599 * config/riscv/riscv.md: Unify indentation.
17601 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
17606 * config/rs6000/rs6000.md (u code attribute): Add FIX and
17608 (extendsi<mode>2): Add support for doing sign extension via
17609 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
17610 don't have ISA 3.0 instructions.
17611 (extendsi<mode>2 splitter): Likewise.
17612 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
17613 generate the normal insns since SImode can now go in vector
17614 registers. Disallow the special UNSPECs needed for previous
17615 machines to hide SImode being used. Add new insns
17616 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
17617 (fix_trunc<mode>si2_stfiwx): Likewise.
17618 (fix_trunc<mode>si2_internal): Likewise.
17619 (fixuns_trunc<mode>si2): Likewise.
17620 (fixuns_trunc<mode>si2_stfiwx): Likewise.
17621 (fctiw<u>z_<mode>_smallint): Likewise.
17622 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
17623 of floating point to 32-bit integer from doing a direct move to
17624 the GPR registers to do a store.
17625 (fctiwz_<mode>): Break long line.
17627 2017-05-05 Bin Cheng <bin.cheng@arm.com>
17629 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
17630 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
17631 (addr_list, addr_offset_valid_p): New.
17632 (split_address_groups): Check offset validity with above function.
17633 (gt-tree-ssa-loop-ivopts.h): Include header file.
17635 2017-05-05 Nathan Sidwell <nathan@acm.org>
17637 * config.gcc (arm*-*-*): Add missing 'fi'.
17639 2017-05-05 Steve Ellcey <sellcey@cavium.com>
17641 * doc/invoke.texi (-fopt-info): Explicitly say order of options
17642 included in -fopt-info does not matter.
17643 * doc/optinfo.texi (-fopt-info): Fix description of default
17644 behavour. Explicitly say order of options included in -fopt-info
17647 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
17649 * config.gcc: Allow combinations of aprofile and rmprofile values for
17650 --with-multilib-list.
17651 * config/arm/t-multilib: New file.
17652 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
17653 variables. Remove setting of ISA and floating-point ABI in
17654 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
17655 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
17656 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
17657 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
17659 * config/arm/t-rmprofile: Likewise except for the matches changes.
17660 * doc/install.texi (--with-multilib-list): Document the combination of
17661 aprofile and rmprofile values and warn about pitfalls in doing that.
17663 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
17665 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
17666 (movdi_aarch64): Likewise.
17668 2017-05-05 Jakub Jelinek <jakub@redhat.com>
17670 PR tree-optimization/80632
17671 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
17673 (build_arrays): Initialize it for virtual phis.
17674 (fix_phi_nodes): Use it for virtual phis.
17676 PR tree-optimization/80558
17677 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
17678 [x, y] op z into [x op, y op z] for op & or | if conditions
17681 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
17682 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
17685 * config/arm/arm.md (use_literal_pool): Remove.
17686 (64-bit immediate split): No longer takes cost into consideration
17687 if arm_disable_literal_pool is enabled.
17688 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
17689 used when arm_disable_literal_pool is enabled.
17690 (arm_max_const_double_inline_cost): Remove use of
17691 arm_disable_literal_pool.
17692 (push_minipool_fix): Add assert.
17693 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
17694 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
17695 (no_literal_pool_sf_immediate): New.
17697 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17699 PR tree-optimization/80613
17700 * tree-ssa-dce.c (propagate_necessity): Remove cases for
17701 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
17703 2017-05-05 Richard Biener <rguenther@suse.de>
17705 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
17707 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
17709 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
17710 of this flag from insn conditions due to removal from r247495.
17712 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
17714 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
17716 (arm_early_store_addr_dep_ptr): Likewise.
17717 * config/arm/aarch-common-protos.h
17718 (arm_early_load_addr_dep_ptr): Add prototype.
17719 (arm_early_store_addr_dep_ptr): Likewise.
17720 * config/arm/cortex-a53.md: Add new bypasses.
17722 2017-05-05 Jakub Jelinek <jakub@redhat.com>
17724 * tree.c (next_type_uid): Change type to unsigned.
17725 (type_hash_canon): Decrement back next_type_uid if
17726 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
17727 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
17730 2017-05-04 Martin Sebor <msebor@redhat.com>
17732 * builtins.c: Fix a trivial typo in a comment.
17734 PR middle-end/79234
17735 * builtins.c (check_sizes): Adjust to handle reading past the end.
17736 Avoid printing excessive upper bound of ranges. Use %E to print
17737 tree nodes instead of converting them to %wu.
17738 (expand_builtin_memchr): New function.
17739 (compute_dest_size): Rename...
17740 (compute_objsize): ...to this.
17741 (expand_builtin_memcpy): Adjust.
17742 (expand_builtin_mempcpy): Adjust.
17743 (expand_builtin_strcat): Adjust.
17744 (expand_builtin_strcpy): Adjust.
17745 (check_strncat_sizes): Adjust.
17746 (expand_builtin_strncat): Adjust.
17747 (expand_builtin_strncpy): Adjust and simplify.
17748 (expand_builtin_memset): Adjust.
17749 (expand_builtin_bzero): Adjust.
17750 (expand_builtin_memcmp): Adjust.
17751 (expand_builtin): Handle memcmp.
17752 (maybe_emit_chk_warning): Check strncat just once.
17754 2017-05-04 Martin Sebor <msebor@redhat.com>
17756 PR preprocessor/79214
17757 PR middle-end/79222
17758 PR middle-end/79223
17759 * builtins.c (check_sizes): Add inlining context and issue
17760 warnings even when -Wno-system-headers is set.
17761 (check_strncat_sizes): Same.
17762 (expand_builtin_strncat): Same.
17763 (expand_builtin_memmove): New function.
17764 (expand_builtin_stpncpy): Same.
17765 (expand_builtin): Handle memmove and stpncpy.
17767 2017-05-04 Bin Cheng <bin.cheng@arm.com>
17769 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
17770 which is not used any more.
17772 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
17774 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
17776 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
17778 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
17779 (cortexa53_tunings): Likewise.
17780 (cortexa57_tunings): Likewise.
17781 (cortexa72_tunings): Likewise.
17782 (cortexa73_tunings): Likewise.
17784 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
17786 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
17787 Set loop alignment to 8.
17789 2017-05-04 Martin Sebor <msebor@redhat.com>
17791 PR translation/80280
17792 * builtins.c (expand_builtin_object_size): Add missing quoting to
17793 %D and like directives.
17794 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
17795 (hsa_type_for_tree_type): Same.
17796 (verify_function_arguments): Same.
17797 * symtab.c (symbol_table::change_decl_assembler_name): Same.
17798 * varasm.c (get_section): Same.
17801 2017-05-04 Martin Sebor <msebor@redhat.com>
17803 PR translation/80280
17804 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
17806 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
17808 * config/aarch64/aarch64.c (generic_addrcost_table):
17809 Change HI/TI mode setting.
17811 2017-05-04 Martin Jambor <mjambor@suse.cz>
17813 PR tree-optimization/80622
17814 * tree-sra.c (comes_initialized_p): New function.
17815 (build_accesses_from_assign): Only set write lazily when
17816 comes_initialized_p is false.
17817 (analyze_access_subtree): Use comes_initialized_p.
17818 (propagate_subaccesses_across_link): Assert !comes_initialized_p
17819 instead of testing for PARM_DECL.
17821 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17823 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
17824 constraint on operand 0 to allow more general addressing modes.
17825 Adjust output template.
17826 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
17828 * config/aarch64/aarch64-protos.h
17829 (aarch64_address_valid_for_prefetch_p): Declare prototype.
17830 * config/aarch64/constraints.md (Dp): New address constraint.
17831 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
17834 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
17836 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
17837 update use of estimate_ipcp_clone_size_and_time.
17838 (estimate_local_effects): Update use of
17839 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
17840 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
17841 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
17842 Return nonspecialized time.
17844 2017-05-04 Richard Biener <rguenther@suse.de>
17846 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
17847 for the last VUSE which def dominates the PHI. Directly call
17849 (get_continuation_for_phi_1): Remove.
17851 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
17853 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
17854 to explain the use of truncating division. Cap the number of
17855 iterations to the maximum given by nb_iterations_upper_bound,
17858 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
17860 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
17861 * configure: Regenerate.
17862 * config.in: Regenerate.
17863 * config/i386/driver-mingw32.c: new file.
17864 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
17865 * config.host: Link driver-mingw32.o on MinGW host.
17866 * doc/install.texi: Document new --enable-mingw-wildcard configure
17869 2017-05-04 Marek Polacek <polacek@redhat.com>
17871 PR tree-optimization/80612
17872 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
17874 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
17875 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
17877 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
17878 (movt splitter): Likewise.
17879 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
17880 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
17881 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
17882 block for Thumb-1 with MOVT.
17883 (thumb2_legitimate_address_p): Move code block ...
17884 (can_avoid_literal_pool_for_label_p): ... into this new function.
17885 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
17887 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
17888 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
17889 "M-profile targets with the MOVT instruction".
17891 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
17893 * config/arm/arm-builtins.c (arm_init_builtins): Rename
17894 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
17895 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
17897 2017-05-04 Martin Liska <mliska@suse.cz>
17899 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
17900 variable cond_code.
17902 2017-05-04 Richard Biener <rguenther@suse.de>
17904 * tree.c (array_at_struct_end_p): Handle arrays at struct
17905 end with flexarrays more conservatively. Refactor and treat
17906 arrays of arrays or aggregates more strict. Fix
17907 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
17908 * tree.c (array_at_struct_end_p): Adjust prototype.
17909 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
17910 * gimple-fold.c (get_range_strlen): Likewise.
17911 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
17913 2017-05-04 Richard Biener <rguenther@suse.de>
17915 PR tree-optimization/31130
17916 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
17918 (supports_overflow_infinity): Likewise.
17919 (is_negative_overflow_infinity): Likewise.
17920 (is_positive_overflow_infinity): Likewise.
17921 (is_overflow_infinity): Likewise.
17922 (stmt_overflow_infinity): Likewise.
17923 (overflow_infinity_range_p): Likewise.
17924 (usable_range_p): Remove as always returning true.
17925 (make_overflow_infinity): Remove.
17926 (negative_overflow_infinity): Likewise.
17927 (positive_overflow_infinity): Likewise.
17928 (avoid_overflow_infinity): Likewise.
17929 (set_value_range): Adjust accordingly.
17930 (set_value_range_to_nonnegative): Likewise, remove now unused
17931 overflow_infinity arg.
17932 (vrp_operand_equal_p): Adjust.
17933 (update_value_range): Likewise.
17934 (range_int_cst_singleton_p): Likewise.
17935 (operand_less_p): Likewise.
17936 (compare_values_warnv): Likewise.
17937 (extract_range_for_var_from_comparison_expr): Likewise.
17938 (vrp_int_const_binop): Likewise.
17939 (zero_nonzero_bits_from_vr): Likewise.
17940 (extract_range_from_multiplicative_op_1): Likewise.
17941 (extract_range_from_binary_expr_1): Likewise.
17942 (extract_range_from_unary_expr): Likewise.
17943 (extract_range_from_comparison): Likewise.
17944 (extract_range_basic): Likewise.
17945 (adjust_range_with_scev): Likewise.
17946 (compare_ranges): Likewise.
17947 (compare_range_with_value): Likewise.
17948 (dump_value_range): Likewise.
17949 (test_for_singularity): Likewise, remove strict_overflow_p parameter
17951 (simplify_cond_using_ranges): Adjust.
17953 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
17955 * brig-builtins.def: Added a builtin for class_f64.
17956 * builtin-types.def: Added a builtin type needed by class_f64.
17958 2017-05-03 Jason Merrill <jason@redhat.com>
17960 * timevar.def: Add TV_CONSTEXPR.
17962 2017-05-03 David Malcolm <dmalcolm@redhat.com>
17964 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
17966 2017-05-03 Martin Jambor <mjambor@suse.cz>
17968 * ipa-prop.c (ipa_update_after_lto_read): Removed.
17969 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
17970 * ipa-cp.c (ipcp_propagate_stage): Do not call
17971 ipa_update_after_lto_read.
17972 * ipa-inline.c (ipa_inline): Likewise.
17974 2017-05-03 Martin Jambor <mjambor@suse.cz>
17976 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
17977 tag. Added a default constructor and a destructor.
17978 (ipa_edge_args_sum_t): New class;
17979 (ipa_edge_args_sum): Declare.
17980 (ipa_edge_args_vector): Remove declaration.
17981 (IPA_EDGE_REF): Use ipa_edge_args_sum.
17982 (ipa_free_edge_args_substructures): Remove declaration.
17983 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
17984 (ipa_edge_args_info_available_for_edge_p): Likewise.
17985 * ipa-prop.c (ipa_edge_args_vector): Removed.
17986 (edge_removal_hook_holder): Likewise.
17987 (edge_duplication_hook_holder): Likewise.
17988 (ipa_edge_args_sum): New variable.
17989 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
17990 ipa_edge_args_vector.
17991 (ipa_free_edge_args_substructures): Likewise.
17992 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
17993 ipa_edge_args_vector.
17994 (ipa_edge_removal_hook): Turned into method
17995 ipa_edge_args_sum_t::remove.
17996 (ipa_edge_duplication_hook): Turned into method
17997 ipa_edge_args_sum_t::duplicate.
17998 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
17999 registering edge hooks.
18000 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
18001 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
18002 ipa_edge_args_sum instead of ipa_edge_args_vector.
18003 * ipa-profile.c (ipa_profile): Likewise.
18005 2017-05-03 Martin Jambor <mjambor@suse.cz>
18007 * symbol-summary.h (function_summary): New method exists.
18008 (function_summary::symtab_removal): Deallocate through release.
18009 (call_summary): New class.
18010 (gt_ggc_mx): New overload.
18011 (gt_pch_nx): Likewise.
18012 (gt_pch_nx): Likewise.
18014 2017-05-03 Jeff Law <law@redhat.com>
18016 PR tree-optimization/78496
18017 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
18018 from simplify_cond_using_ranges. Split off code to walk
18019 backwards through casts into ...
18020 (simplify_cond_using_ranges_2): New function.
18021 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
18022 (execute_vrp): After identifying jump threads, call
18023 simplify_cond_using_ranges_2.
18025 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
18028 * ipa-inline.h (inline_summary): Add ctor.
18029 (create_ggc): Do not use ggc_cleared_alloc.
18031 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
18032 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18034 * gcc.c (handle_braces): Support escaping in switch matching
18036 * doc/invoke.texi (Spec Files): Document it.
18037 Remove superfluous @code markup in items.
18039 2017-05-03 David Malcolm <dmalcolm@redhat.com>
18041 * diagnostic-show-locus.c (struct column_range): New struct.
18042 (get_affected_columns): New function.
18043 (get_printed_columns): New function.
18044 (struct correction): New struct.
18045 (correction::ensure_capacity): New function.
18046 (correction::ensure_terminated): New function.
18047 (struct line_corrections): New struct.
18048 (line_corrections::~line_corrections): New dtor.
18049 (line_corrections::add_hint): New function.
18050 (layout::print_trailing_fixits): Reimplement in terms of the new
18052 (selftest::test_overlapped_fixit_printing): New function.
18053 (selftest::diagnostic_show_locus_c_tests): Call it.
18055 2017-05-03 Nathan Sidwell <nathan@acm.org>
18057 Canonicalize canonical type hashing
18058 * tree.h (type_hash_canon_hash): Declare.
18059 * tree.c (type_hash_list, attribute_hash_list): Move into
18060 type_hash_canon_hash.
18061 (build_type_attribute_qual_variant): Break out hash code calc into
18062 type_hash_canon_hash.
18063 (type_hash_canon_hash): New. Generic type hash computation.
18064 (build_range_type_1, build_array_type_1, build_function_type,
18065 build_method_type_directly, build_offset_type, build_complex_type,
18066 make_vector_type): Call it.
18068 2017-05-03 Richard Biener <rguenther@suse.de>
18070 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
18071 When all DRs have unknown misaligned do not always peel
18072 when there is a store but apply the same costing model as if
18073 there were only loads.
18075 2017-05-03 Richard Biener <rguenther@suse.de>
18078 PR tree-optimization/80492
18079 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
18080 compare_base_decls returning dont-know properly.
18082 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
18084 * config/arm/iterators.md (CCSI): New mode iterator.
18085 (arch): New mode attribute.
18086 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
18087 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
18088 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
18089 code iterator for success result mode.
18090 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
18091 the corresponding new insn generators.
18093 2017-05-03 Bin Cheng <bin.cheng@arm.com>
18096 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18097 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
18099 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
18101 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
18102 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
18103 (DDR_A): Wrap DDR argument in brackets.
18104 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
18105 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
18106 (DDR_REVERSED_P): Likewise.
18108 2017-05-03 Jakub Jelinek <jakub@redhat.com>
18110 PR tree-optimization/79472
18111 * tree-switch-conversion.c (struct switch_conv_info): Add
18112 contiguous_range and default_case_nonstandard fields.
18113 (collect_switch_conv_info): Compute contiguous_range and
18114 default_case_nonstandard fields, don't clear final_bb if
18115 contiguous_range and only the default case doesn't have the required
18117 (check_all_empty_except_final): Set default_case_nonstandard instead
18118 of failing if contiguous_range and the default case doesn't have empty
18120 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
18121 and only the default case doesn't have the required constants. Skip
18123 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
18124 if default_case_nonstandard.
18125 (build_constructors): Build constant 1 just once. Assert that default
18126 values aren't inserted in between cases if contiguous_range. Skip
18128 (build_arrays): Skip virtual phis.
18129 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
18130 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
18131 Handle virtual phis.
18132 (gen_inbound_check): Handle default_case_nonstandard case.
18133 (process_switch): Adjust check_final_bb caller. Call
18134 gather_default_values with the first non-default case instead of
18135 default case if default_case_nonstandard.
18137 2017-05-02 Nathan Sidwell <nathan@acm.org>
18139 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
18140 check. Fix formatting.
18142 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
18144 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
18145 errors when comparing specialized and unspecialized times.
18147 2017-05-02 David Malcolm <dmalcolm@redhat.com>
18149 * diagnostic-show-locus.c
18150 (layout::should_print_annotation_line_p): Make private.
18151 (layout::print_annotation_line): Make private.
18152 (layout::annotation_line_showed_range_p): Make private.
18153 (layout::show_ruler): Make private.
18154 (layout::print_source_line): Make private. Pass in line and
18155 line_width, rather than calling location_get_source_line. Drop
18157 (layout::print_leading_fixits): New method.
18158 (layout::print_any_fixits): Rename to...
18159 (layout::print_trailing_fixits): ...this, and make private.
18160 Don't print newline fixits.
18161 (diagnostic_show_locus): Move logic for printing one row into...
18162 (layout::print_line): ...this new function. Move the
18163 location_get_source_line call and error-handling from
18164 print_source_line to here. Call print_leading_fixits, and rename
18165 print_any_fixits to print_trailing_fixits.
18166 (selftest::test_fixit_insert_containing_newline): Update now that
18167 newlines are partially supported.
18168 (selftest::test_fixit_insert_containing_newline_2): New test.
18169 (selftest::test_fixit_replace_containing_newline): Update comments.
18170 (selftest::diagnostic_show_locus_c_tests): Call the new test.
18171 * edit-context.c (class added_line): New class.
18172 (class edited_line): Describe newline handling in comment.
18173 (edited_line::actually_edited_p): New method.
18174 (edited_line::print_content): Delete redundant decl.
18175 (edited_line::m_predecessors): New field.
18176 (edited_file::print_content): Call edited_line::print_content.
18177 (edited_file::print_diff): Update to support newlines.
18178 (edited_file::print_diff_hunk): Likewise.
18179 (edited_file::print_run_of_changed_lines): New function.
18180 (edited_file::print_diff_line): Convert to...
18181 (print_diff_line): ...this.
18182 (edited_file::get_effective_line_count): New function.
18183 (edited_line::edited_line): Initialize new field m_predecessors.
18184 (edited_line::~edited_line): Clean up m_predecessors.
18185 (edited_line::apply_fixit): Handle newlines.
18186 (edited_line::get_effective_line_count): New function.
18187 (edited_line::print_content): New function.
18188 (edited_line::print_diff_lines): New function.
18189 (selftest::test_applying_fixits_insert_containing_newline): New
18191 (selftest::test_applying_fixits_replace_containing_newline): New
18193 (selftest::insert_line): New function.
18194 (selftest::test_applying_fixits_multiple_lines): Add example of
18196 (selftest::edit_context_c_tests): Call the new tests.
18198 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18200 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
18201 parameter cand. Update dump information.
18202 (get_computation_cost): Update uses.
18204 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18206 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
18207 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
18208 (get_computation_at, rewrite_use_address): Update use of
18209 get_computation_aff.
18211 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18213 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
18214 (get_computation): Delete.
18215 (get_computation_cost): Implement like get_computation_cost_at.
18216 Use get_computation_at.
18217 (get_computation_cost_at): Delete.
18218 (rewrite_use_nonlinear_expr): Use get_computation_at.
18219 (rewrite_use_compare, remove_unused_ivs): Ditto.
18221 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18223 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
18225 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18227 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
18228 (ivopts_global_cost_for_size): Rename parameter and update uses.
18229 (iv_ca_recount_cost): Update uses.
18230 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
18231 candidates seperately in n_invs and n_cands.
18232 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
18234 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18236 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
18237 (find_inv_vars_cb): New.
18238 (find_depends): Renamed to ...
18239 (find_inv_vars): ... this.
18240 (add_candidate_1, force_var_cost): Call find_inv_vars.
18241 (split_address_cost, determine_group_iv_cost_cond): Ditto.
18243 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18245 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
18246 inv_vars. Add inv_exprs.
18247 (struct iv_cand): Rename depends_on to inv_vars.
18248 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
18249 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
18250 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
18251 (dump_cand): Dump inv_vars.
18252 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
18253 (record_invariant, find_depends, add_candidate_1): Ditto.
18254 (set_group_iv_cost, force_var_cost): Ditto.
18255 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
18256 (get_computation_cost_at, get_computation_cost): Ditto.
18257 (determine_group_iv_cost_generic): Ditto.
18258 (determine_group_iv_cost_address): Ditto.
18259 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
18260 (determine_group_iv_costs): Ditto.
18261 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
18262 (iv_ca_set_remove_invariants): Renamed to ...
18263 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
18264 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
18265 (iv_ca_set_add_invariants): Renamed to ...
18266 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
18267 (iv_ca_set_cp): Use iv_ca_set_add_invs.
18268 (iv_ca_has_deps): Support inv_vars and inv_exprs.
18269 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
18270 (create_new_ivs): Remove useless dump.
18272 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18274 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
18276 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
18277 (iv_ca_set_no_cp, create_new_iv): Ditto.
18279 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18281 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
18283 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
18285 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
18286 function tree_check2.
18288 2017-05-02 Martin Liska <mliska@suse.cz>
18290 * doc/gcov.texi: Add missing preposition.
18291 * gcov.c (function_info::function_info): Properly fill up
18292 all member variables.
18294 2017-05-02 Tamar Christina <tamar.christina@arm.com>
18296 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
18298 2017-05-02 Tamar Christina <tamar.christina@arm.com>
18300 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
18302 2017-05-02 Martin Liska <mliska@suse.cz>
18305 * lto-streamer-in.c (lto_read_tree_1): Remove
18306 LTO_STREAMER_DEBUG.
18307 * lto-streamer.c (struct tree_hash_entry): Likewise.
18308 (struct tree_entry_hasher): Likewise.
18309 (tree_entry_hasher::hash): Likewise.
18310 (tree_entry_hasher::equal): Likewise.
18311 (lto_streamer_init): Likewise.
18312 (lto_orig_address_map): Likewise.
18313 (lto_orig_address_get): Likewise.
18314 (lto_orig_address_remove): Likewise.
18315 * lto-streamer.h: Likewise.
18316 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
18317 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
18319 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
18321 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
18322 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
18323 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
18324 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
18325 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
18326 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
18327 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
18328 (mm_maskz_sub_ss): New intrinsics.
18329 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
18330 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
18331 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
18332 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
18333 (__builtin_ia32_subss_mask_round): New builtins.
18334 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
18335 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
18336 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
18338 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
18339 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
18341 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
18344 2017-05-02 Martin Jambor <mjambor@suse.cz>
18346 PR tree-optimization/78687
18347 * tree-sra.c (access): New field parent.
18348 (process_subtree_disqualification): New function.
18349 (disqualify_candidate): Call it.
18350 (build_accesses_from_assign): Reset write flag if creating an
18352 (build_access_subtree): Fill in parent field and also prpagate
18353 down grp_write flag.
18354 (create_artificial_child_access): New parameter set_grp_write, set
18355 grp_write to its value.
18356 (propagate_subaccesses_across_link): Also propagate grp_write flag
18358 (propagate_all_subaccesses): Push the closest parent back to work
18359 queue if add_access_to_work_queue returned true.
18361 2017-05-02 Richard Biener <rguenther@suse.de>
18363 * common.opt (fstrict-overflow): Alias negative to fwrapv.
18364 * doc/invoke.texi (fstrict-overflow): Remove all traces of
18365 -fstrict-overflow documentation.
18366 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
18367 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
18368 flag_strict_overflow.
18369 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
18370 * lto-opts.c (lto_write_options): Do not stream it.
18371 * lto-wrapper.c (merge_and_complain): Do not handle it.
18372 * opts.c (default_options_table): Do not set -fstrict-overflow.
18373 (finish_options): Likewise do not clear it when sanitizing.
18374 * simplify-rtx.c (simplify_const_relational_operation): Do not
18375 test flag_strict_overflow.
18377 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
18379 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
18380 using enabled attribute.
18381 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
18382 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
18383 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
18384 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
18385 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
18386 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
18387 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
18388 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
18389 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
18390 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
18392 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
18394 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
18396 2017-05-02 Richard Biener <rguenther@suse.de>
18398 PR tree-optimization/80591
18400 2017-04-10 Richard Biener <rguenther@suse.de>
18402 * tree-ssa-structalias.c (find_func_aliases): Properly handle
18405 2017-05-02 Richard Biener <rguenther@suse.de>
18407 PR tree-optimization/80549
18408 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
18409 (cleanup_tree_cfg_noloop): Create forwarders to known loop
18410 headers if they do not have a preheader.
18412 2017-05-02 Martin Liska <mliska@suse.cz>
18415 * common.opt: Fix typo.
18416 * doc/invoke.texi: Likewise.
18418 2017-05-01 Jan Beulich <jbeulich@suse.com>
18420 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
18421 swapping, add (x,x,m,x,n) alternative.
18423 2017-05-01 Nathan Sidwell <nathan@acm.org>
18425 * calls.c (combine_pending_stack_adjustment_and_call): Remove
18426 unnecessary unadjusted_alignment check.
18428 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
18431 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
18433 * gimplify.c (gimplify_cilk_detach): New function.
18434 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
18435 * tree-core.h: Document EXPR_CILK_SPAWN.
18436 * tree.h (EXPR_CILK_SPAWN): Define.
18438 2017-05-01 David Malcolm <dmalcolm@redhat.com>
18440 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
18441 to use new fixit_hint representation, using the "replace" logic.
18442 (get_line_span_for_fixit_hint): Likewise.
18443 (layout::print_any_fixits): Likewise.
18444 (selftest::test_one_liner_many_fixits): Rename to...
18445 (selftest::test_one_liner_many_fixits_1): ...this, and update
18446 comment and expected output to reflect that the multiple fix-it
18447 hints are now consolidated into one insertion.
18448 (selftest::test_one_liner_many_fixits_2): New test.
18449 (selftest::test_diagnostic_show_locus_one_liner): Update for
18451 (selftest::test_fixit_consolidation): Update for fix-it API
18453 * diagnostic.c (print_parseable_fixits): Likewise.
18454 * edit-context.c (edited_line::m_line_events): Convert from
18455 auto_vec <line_event *> to auto_vec <line_event>.
18456 (class line_event): Convert from abstract base class to a concrete
18457 class, taking over the role of replace_event.
18458 (class insert_event): Delete.
18459 (class replace_event): Rename to class line_event. Convert to
18461 (edit_context::add_fixits): Reimplement.
18462 (edit_context::apply_insert): Delete.
18463 (edit_context::apply_replace): Rename to...
18464 (edit_context::apply_fixit): ...this. Convert to half-open range.
18465 (edited_file::apply_insert): Delete.
18466 (edited_file::apply_replace): Rename to...
18467 (edited_file::apply_fixit): ...this.
18468 (edited_line::~edited_line): Drop deletion of events.
18469 (edited_line::apply_insert): Delete.
18470 (edited_line::apply_replace): Rename to...
18471 (edited_line::apply_fixit): ...this. Convert to half-open range.
18472 Update for change to type of m_line_events.
18473 * edit-context.h (edit_context::apply_insert): Delete.
18474 (edit_context::apply_replace): Rename to...
18475 (edit_context::apply_fixit): ...this.
18477 2017-05-01 Martin Sebor <msebor@redhat.com>
18479 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
18482 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
18485 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
18486 __get_cpuid_max returns 0.
18487 (__get_cpuid_count): Ditto.
18489 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
18491 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
18492 replacement expression is another instance of one of its arguments.
18494 2017-05-01 Jakub Jelinek <jakub@redhat.com>
18497 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
18498 check for stack push/pop autoinc.
18499 * config/i386/i386.c (ix86_agi_dependent): Return false
18500 if the only reason why modified_in_p returned true is that
18501 addr is SP based and set_insn is a push or pop.
18503 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
18505 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
18508 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
18511 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
18512 (account_size_time): Use two predicates - exec_pred and
18514 (evaluate_conditions_for_known_args): Compute both clause and
18516 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
18517 (inline_summary_t::duplicate): Update.
18518 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
18520 (compute_inline_parameters): Likewise.
18521 (estimate_edge_size_and_time): Update caluclation of time.
18522 (estimate_node_size_and_time): Compute both time and nonspecialized
18524 (estimate_ipcp_clone_size_and_time): Update.
18525 (inline_merge_summary): Update.
18526 (do_estimate_edge_time): Update.
18527 (do_estimate_edge_size): Update.
18528 (do_estimate_edge_hints): Update.
18529 (inline_read_section, inline_write_summary): Stream both new predicates.
18530 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
18532 (compute_inlined_call_time): Cleanup.
18533 (big_speedup_p): Update.
18534 (edge_badness): Update.
18535 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
18536 (size_time_entry): Replace predicate by exec_predicate and
18537 nonconst_predicate.
18538 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
18539 (estimate_edge_time): Return also nonspec_time.
18540 (reset_edge_growth_cache): Update.
18542 2017-04-29 Jakub Jelinek <jakub@redhat.com>
18544 PR rtl-optimization/80491
18545 * ifcvt.c (noce_process_if_block): When looking for x setter
18546 with missing else_bb, don't check only the insn right before
18547 cond_earliest, but look for the last insn that x is modified in
18548 within the same bb.
18550 PR rtl-optimization/80491
18551 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
18553 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
18555 PR tree-optimization/80487
18556 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
18558 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18560 PR tree-optimization/79697
18561 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
18562 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
18563 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
18565 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
18566 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
18568 2017-04-28 Martin Sebor <msebor@redhat.com>
18570 PR tree-optimization/80523
18571 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
18572 (init_target_to_host_charmap, target_to_host, target_strtol10): New
18574 (maybe_warn, format_directive, parse_directive): Use new functions.
18575 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
18577 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
18579 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
18581 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
18583 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
18584 target_header_dir): Set correctly.
18585 * configure: Regenerated.
18586 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
18587 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
18588 instead of SYSTEM_HEADER_DIR.
18590 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
18592 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
18593 (estimate_local_effects): Likewise.
18594 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
18595 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
18596 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
18597 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
18598 do_estimate_edge_time, estimate_edge_time): Likewise.
18599 * ipa-inline-analysis.c (estimate_node_size_and_time,
18600 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
18601 (estimate_time_after_inlining): Remove.
18603 2017-04-28 Martin Liska <mliska@suse.cz>
18605 * doc/gcov.texi: Enhance documentation of gcov.
18607 2017-04-28 Martin Liska <mliska@suse.cz>
18609 * doc/gcov.texi: Sort options in alphabetic order.
18610 * doc/gcov-dump.texi: Likewise.
18611 * doc/gcov-tool.texi: Likewise.
18612 * gcov.c (print_usage): Likewise.
18613 * gcov-dump.c (print_usage): Likewise.
18614 * gcov-tool.c (print_merge_usage_message): Likewise.
18615 (print_rewrite_usage_message): Likewise.
18616 (print_overlap_usage_message): Likewise.
18618 2017-04-28 Martin Liska <mliska@suse.cz>
18620 PR gcov-profile/53915
18621 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
18623 2017-04-28 Martin Liska <mliska@suse.cz>
18625 PR gcov-profile/79891
18626 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
18627 is marked by compiler as living on a line.
18628 (get_cycles_count): Remove usage of the union.
18629 (output_intermediate_file): Likewise.
18630 (find_source): Fix GNU coding style.
18631 (accumulate_line_counts): Remove old non-all block mode.
18632 (output_lines): Remove usage of the union.
18633 * profile.c (output_location): Include all BBs, even if
18634 belonging to a same line (and file) as a previous BB.
18636 2017-04-28 Martin Liska <mliska@suse.cz>
18638 * gcov.c (process_args): Handle new argument 'w'.
18639 (read_graph_file): Assign ID to BBs.
18640 (output_branch_count): Display BB # if verbose flag is set.
18641 (output_lines): Likewise for arcs.
18642 (print_usage): Add '--verbose' option help.
18643 * doc/gcov.texi: Document --verbose (-w) option.
18645 2017-04-28 Martin Liska <mliska@suse.cz>
18647 * gcov.c (struct block_location_info): New struct.
18648 (process_file): Fill up the new structure.
18649 (read_graph_file): Replace usage of encoding by the newly added
18651 (add_line_counts): Likewise.
18652 (accumulate_line_counts): Remove usage of the union.
18653 (function_info::function_info): New function.
18654 (function_info::~function_info): Likewise.
18655 (process_file): Call delete instead of release_function.
18656 (release_function): Release the function.
18657 (release_structures): Call delete instead of release_function.
18658 (solve_flow_graph): Replace usage of num_blocks.
18659 (find_exception_blocks): Likewise.
18660 (output_lines): Fix GNU coding style.
18662 2017-04-28 Martin Liska <mliska@suse.cz>
18665 * coverage.c (coverage_remove_note_file): New function.
18666 * coverage.h: Declare the function.
18667 * toplev.c (finalize): Clean if an error has been seen.
18669 2017-04-28 Martin Liska <mliska@suse.cz>
18671 PR gcov-profile/80031
18672 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
18673 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
18674 * gcov.c (read_graph_file): Read just number of blocks.
18675 * profile.c (branch_prob): Do not stream 0 flags per a basic
18678 2017-04-28 Martin Liska <mliska@suse.cz>
18680 * gcov-dump.c (tag_*): Add new argument to declarations.
18681 (dump_gcov_file): Likewise.
18682 (tag_blocks): Add and use new argument depth.
18683 (tag_arcs): Likewise.
18684 (tag_lines): Likewise.
18685 (tag_counters): Likewise.
18686 (tag_summary): Likewise.
18687 (dump_working_sets): Use depth to do a proper indentation.
18689 2017-04-28 Jakub Jelinek <jakub@redhat.com>
18692 * cgraph.h (symtab_node::debug_symtab): No longer inline.
18693 * symtab.c (symtab_node::debug_symtab): Move definition here.
18695 2017-04-28 Richard Biener <rguenther@suse.de>
18697 * lto-streamer.h (LTO_major_version): Bump to 7.
18699 2017-04-28 Richard Biener <rguenther@suse.de>
18701 * tree-vrp.c (assert_info): New struct.
18702 (add_assert_info): New helper.
18703 (register_edge_assert_for_2): Refactor to add asserts to a vector
18705 (register_edge_assert_for_1): Likewise.
18706 (register_edge_assert_for): Likewise.
18707 (finish_register_edge_assert_for): New helper actually registering
18708 asserts where live on edge.
18709 (find_conditional_asserts): Adjust.
18710 (find_switch_asserts): Likewise.
18711 (evrp_dom_walker::try_find_new_range): Generalize.
18712 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
18714 2017-04-27 Marek Polacek <polacek@redhat.com>
18717 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
18718 arg10 and arg11 to itype.
18720 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
18722 * doc/extend.texi (Object Size Checking): Improve grammar.
18724 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
18727 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
18728 that the logic for permitting reciprocal estimates matches that
18731 2017-04-27 Jakub Jelinek <jakub@redhat.com>
18734 * tree.c (type_cache_hasher::equal): Only compare
18735 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
18736 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
18737 non-aggregate element types.
18738 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
18739 about the flag on ARRAY_TYPEs in the comment, formatting fix.
18741 2017-04-27 Richard Biener <rguenther@suse.de>
18743 PR middle-end/80533
18744 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
18745 stripping ARRAY_REFs from MEM_EXPR make sure we're not
18746 keeping a reference to a trailing array.
18748 2017-04-27 Richard Biener <rguenther@suse.de>
18750 PR middle-end/80539
18751 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
18752 being in loop-closed SSA form conservatively.
18753 (chrec_fold_multiply_poly_poly): Likewise.
18755 2017-04-27 Tamar Christina <tamar.christina@arm.com>
18757 PR middle-end/79665
18758 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
18759 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
18761 2017-04-27 Jakub Jelinek <jakub@redhat.com>
18764 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
18765 (aarch64_function_arg_alignment): Return unsigned int again, but still
18766 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
18767 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
18768 Don't emit -Wpsabi note.
18769 (aarch64_function_arg_boundary): Likewise.
18770 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
18773 2017-04-26 Nathan Sidwell <nathan@acm.org>
18775 * tree.h (crc32_unsigned_n): Declare.
18776 (crc32_unsigned, crc32_unsigned): Make inline.
18777 * tree.c (crc32_unsigned_bits): Replace with ...
18778 (crc32_unsigned_n): ... this.
18779 (crc32_unsigned, crc32_byte): Remove.
18780 (crc32_string): Remove unnecessary braces.
18782 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
18784 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
18785 * ipa-inline-analysis.c (MAX_TIME): Remove.
18786 (account_size_time): Use sreal for time.
18787 (dump_inline_summary): Update.
18788 (estimate_function_body_sizes): Update.
18789 (estimate_edge_size_and_time): Update.
18790 (estimate_calls_size_and_time): Update.
18791 (estimate_node_size_and_time): Update.
18792 (inline_merge_summary): Update.
18793 (inline_update_overall_summary): Update.
18794 (estimate_time_after_inlining): Update.
18795 (inline_read_section): Update.
18796 (inline_write_summary): Update.
18797 * ipa-inline.c (compute_uninlined_call_time): Update.
18798 (compute_inlined_call_time): Update.
18799 (recursive_inlining): Update.
18800 (inline_small_functions): Update.
18801 (dump_overall_stats): Update.
18802 * ipa-inline.h: Include sreal.h.
18803 (size_time_entry): Turn time to sreal.
18804 (inline_summary): Turn self_time nad time to sreal.
18806 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
18808 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
18810 (sreal::stream_out, sreal::stream_in): New.
18811 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
18813 2017-04-25 Jakub Jelinek <jakub@redhat.com>
18815 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
18818 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
18821 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
18823 (dimode_scalar_chain::compute_convert_gain): Ditto.
18824 (dimode_scalar_chain::make_vector_copies): Ditto.
18825 (dimode_scalar_chain::convert_reg): Ditto.
18826 (dimode_scalar_chain::convert_insn): Ditto.
18827 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
18828 (VI248_AVX512BW_1): New mode iterator.
18829 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
18830 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
18833 2017-04-25 Martin Sebor <msebor@redhat.com>
18835 PR tree-optimization/80497
18836 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
18837 constants are representable in HOST_WIDE_INT.
18838 (parse_directive): Ditto.
18840 2017-04-25 Martin Sebor <msebor@redhat.com>
18843 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
18844 (new_zero_array): Adjust signature.
18845 (dom_info::dom_init): Used unsigned rather that size_t.
18846 (dom_info::dom_info): Same.
18848 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18849 Jakub Jelinek <jakub@redhat.com>
18852 * config/arm/arm.c: Include gimple.h.
18853 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
18854 returns negative, increment ncrn only if it returned positive.
18855 (arm_needs_doubleword_align): Return int instead of bool,
18856 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
18857 members, but if there is any such non-FIELD_DECL
18858 > PARM_BOUNDARY aligned decl, return -1 instead of false.
18859 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
18860 returns negative, increment nregs only if it returned positive.
18861 (arm_setup_incoming_varargs): Likewise.
18862 (arm_function_arg_boundary): Emit -Wpsabi note if
18863 arm_needs_doubleword_align returns negative, return
18864 DOUBLEWORD_ALIGNMENT only if it returned positive.
18866 2017-04-25 Marek Polacek <polacek@redhat.com>
18869 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
18870 first argument to type.
18872 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
18875 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
18876 type checks to test for compatibility instead of equality.
18878 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18879 Jakub Jelinek <jakub@redhat.com>
18882 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
18884 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
18885 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
18886 the alignment computation, but return their maximum in warn_alignment.
18887 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
18888 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
18890 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
18891 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
18894 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18896 * config/arc/simdext.md (dmpyh): Fix typo.
18898 2017-04-25 Richard Biener <rguenther@suse.de>
18900 PR tree-optimization/80492
18901 * alias.c (compare_base_decls): Handle registers with asm
18902 specification conservatively.
18903 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
18904 compare_base_decls returning dont-know properly.
18906 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18908 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
18909 (legitimate_offset_address_p): New function.
18910 (arc_legitimate_address_p): Use above function.
18912 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18914 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
18916 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18918 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
18919 ACCH registers whenever they are available.
18921 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18923 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
18924 double regs fix when not used.
18926 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18928 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
18930 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
18931 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
18933 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18935 * config/arc/arc.c (arc_output_addsi): Check for h-register class
18936 when emitting short ADD instructions.
18938 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18940 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
18942 (cmpsi_cc_c_insn): Likewise.
18943 (cbranchsi4_scratch): Compute proper instruction length using
18944 compact_hreg_operand.
18945 * config/arc/predicates.md (compact_hreg_operand): New predicate.
18947 2017-04-25 Richard Biener <rguenther@suse.de>
18949 PR middle-end/80509
18950 * passes.c (pass_manager::pass_manager): Initialize
18951 m_name_to_pass_map.
18953 2017-04-25 Richard Biener <rguenther@suse.de>
18955 PR tree-optimization/79201
18956 * tree-ssa-sink.c (statement_sink_location): Handle calls.
18958 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18961 * config/s390/vector.md: Split MEM->GPR vector moves for
18962 non-s_operand addresses.
18964 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18967 * config/s390/predicates.md (reload_const_wide_int_operand): New
18969 * config/s390/s390.md ("movti"): Remove d/P alternative.
18970 ("movti_bigconst"): New pattern definition.
18972 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
18975 * s390-protos.h (s390_expand_cs_hqi): Removed.
18976 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
18977 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
18978 modes as well as CCZ1mode and CCZmode.
18979 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
18980 signature of s390_emit_compare_and_swap.
18981 (s390_expand_cs_hqi): Likewise, make static.
18982 (s390_expand_cs_tdsi): Generate an explicit compare before trying
18983 compare-and-swap, in some cases.
18984 (s390_expand_cs): Wrapper function.
18985 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
18987 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
18988 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
18989 patterns for small and large integers. Forbid symref memory operands.
18990 Move expander to s390.c. Require cc register.
18991 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
18992 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
18993 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
18994 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
18995 symref memory operands. Remove CC mode and call s390_match_ccmode
18997 ("atomic_exchange<mode>"): Allow and implement all integer modes.
18999 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
19001 * config/s390/s390.md (define_peephole2): New peephole to help
19002 combining the load-and-test pattern with volatile memory.
19004 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
19006 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
19007 with CCZmode for TARGET_Z196.
19009 2017-04-25 Jakub Jelinek <jakub@redhat.com>
19011 PR rtl-optimization/80501
19012 * combine.c (make_compound_operation_int): Set subreg_code to SET
19013 even for AND with mask of the sign bit of mode.
19015 PR rtl-optimization/80500
19016 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
19017 sum's initial value.
19019 2017-04-25 Julian Brown <julian@codesourcery.com>
19020 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19022 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
19024 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
19026 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
19028 2017-04-25 Julian Brown <julian@codesourcery.com>
19029 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19031 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
19032 (thunderx2t99_sha): New Reservation.
19034 2017-04-25 Julian Brown <julian@codesourcery.com>
19035 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19037 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
19038 type for 1-element load.
19040 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
19042 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
19044 2017-04-24 Martin Jambor <mjambor@suse.cz>
19046 PR tree-optimization/80293
19047 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
19048 char arrays not totally scalarizable if it is false.
19049 (analyze_all_variable_accesses): Pass correct value in the new
19050 parameter. Add a statistics counter.
19052 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
19054 PR middle-end/79931
19055 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
19057 2017-04-24 Richard Biener <rguenther@suse.de>
19059 PR tree-optimization/80494
19060 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
19061 out for complex types.
19063 2017-04-24 Richard Biener <rguenther@suse.de>
19065 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
19066 * tree-ssa-sccvn.c (print_scc): Print SCC size.
19067 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
19068 (DFS): Adjust and never fail.
19069 (sccvn_dom_walker::fail): Remove.
19070 (sccvn_dom_walker::before_dom_children): Adjust.
19071 (run_scc_vn): Likewise and never fail.
19072 * tree-ssa-pre.c (pass_pre::execute): Adjust.
19073 (pass_fre::execute): Likewise.
19075 2017-04-24 Richard Biener <rguenther@suse.de>
19077 PR tree-optimization/79725
19078 * tree-ssa-sink.c (statement_sink_location): Return whether
19079 failure reason was zero uses. Move that check later.
19080 (sink_code_in_bb): Deal with zero uses by removing the stmt
19083 2017-04-24 Richard Biener <rguenther@suse.de>
19086 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
19087 pointer-based references.
19089 2017-04-24 Richard Biener <rguenther@suse.de>
19092 * pass_manager.h (pass_manager::operator new): Remove.
19093 (pass_manager::operator delete): Likewise.
19094 * passes.c (pass_manager::operator new): Remove.
19095 (pass_manager::operator delete): Likewise.
19096 (pass_manager::pass_manager): Zero individual pass members.
19098 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
19101 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
19102 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
19103 Check "XEXP (src, 1)" operand here.
19104 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
19105 Check "XEXP (src, 1)" operand here.
19106 (dimode_scalar_chain::make_vector_copies): Detect count register
19107 of a shift instruction. Zero extend count register from QImode
19108 to DImode to satisfy vector shift pattern count operand predicate.
19109 Substitute vector shift count operand with a DImode copy.
19110 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
19113 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
19115 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
19116 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
19117 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
19118 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
19119 (UNSPEC_NOREX_MEM): Remove definition.
19121 2017-04-21 Richard Biener <rguenther@suse.de>
19123 PR tree-optimization/79547
19124 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
19125 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
19126 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
19127 without any constraints.
19129 2017-04-21 Richard Biener <rguenther@suse.de>
19131 PR tree-optimization/78847
19132 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
19134 2017-04-21 Richard Biener <rguenther@suse.de>
19136 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
19137 (build_distinct_type_copy): Likewise.
19138 (build_variant_type_copy): Likewise.
19139 * tree.c (build_qualified_type): Pass down mem-stat info.
19140 (build_distinct_type_copy): Likewise.
19141 (build_variant_type_copy): Likewise.
19143 2017-04-21 Richard Biener <rguenther@suse.de>
19145 PR tree-optimization/80237
19146 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
19148 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
19149 for a simplified result.
19151 2016-04-21 Richard Biener <rguenther@suse.de>
19153 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
19154 sth as strict as a simple_iv but a chrec without symbols and an
19155 operand defined in the loop we are peeling (and not some subloop).
19156 (propagate_constants_for_unrolling): Propagate all constants.
19158 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
19161 * config/i386/i386.c (print_reg): Remove assert for disalowed
19162 regno values, call output_operand_lossage instead.
19164 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
19167 * config/i386/constraints.md (Yc): New register constraint.
19168 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
19169 Use Yc constraint for alternative 2 of operand 0. Remove
19170 preferred_for_speed attribute.
19172 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
19174 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
19175 lastprivate clauses in SIMT case.
19177 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
19179 * doc/invoke.texi (-Wextra-semi): Document new warning option.
19181 2017-04-20 Richard Biener <rguenther@suse.de>
19183 PR tree-optimization/57796
19184 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
19185 as N scalar stores.
19186 (vect_model_load_cost): Cost gathers as N scalar loads.
19188 2017-04-20 Richard Biener <rguenther@suse.de>
19190 * ggc-page.c (ggc_allocated_p): Rename to ...
19191 (safe_lookup_page_table_entry): ... this and return the lookup
19193 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
19195 2017-04-20 Richard Biener <rguenther@suse.de>
19197 PR tree-optimization/80453
19198 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
19199 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
19200 from the conditions.
19201 (vn_phi_eq): Pass them down.
19202 (vn_phi_lookup): Record them.
19203 (vn_phi_insert): Likewise.
19205 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
19207 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
19208 uninitialized variable warning to avoid buffer overrun.
19210 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
19213 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
19214 is suppressed for '{ 0 }' in C.
19216 2017-04-20 Jakub Jelinek <jakub@redhat.com>
19218 * BASE-VER: Set to 8.0.0.
19220 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
19222 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
19223 priority .init_array and .fini_array section with SECTION_NOTYPE
19226 2017-04-20 Jakub Jelinek <jakub@redhat.com>
19228 PR middle-end/80423
19229 * tree.h (build_array_type): Add typeless_storage default argument.
19230 * tree.c (type_cache_hasher::equal): Also compare
19231 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
19232 (build_array_type): Add typeless_storage argument, set
19233 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
19235 (build_nonshared_array_type): Adjust build_array_type_1 caller.
19236 (build_array_type): Likewise. Add typeless_storage argument.
19238 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
19239 Jakub Jelinek <jakub@redhat.com>
19241 PR tree-optimization/80426
19242 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
19243 operation on symbolic operands, also compute the overflow for the
19244 invariant part when the operation degenerates into a negation.
19246 2017-04-19 Jakub Jelinek <jakub@redhat.com>
19249 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
19250 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
19253 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
19255 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
19258 * config/avr/avr.c (tree.h): Include it.
19259 (cgraph.h): Include it.
19260 (avr_encode_section_info): Don't warn for uninitialized progmem
19261 variable if it's just an alias.
19263 2017-04-19 Richard Biener <rguenther@suse.de>
19266 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
19267 when needed by AutoPGO.
19269 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
19272 * doc/lto.texi: Remove an extra 'that'.
19274 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
19276 PR rtl-optimization/80429
19277 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
19278 are only used in debug insns.
19280 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
19281 Vladimir Makarov <vmakarov@redhat.com>
19283 * config/sparc/predicates.md (input_operand): Add comment. Return
19284 true for any memory operand when LRA is in progress.
19285 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
19287 2017-04-18 Jeff Law <law@redhat.com>
19290 * mips.md ({return,simple_return}_internal): Do not overwrite
19293 2017-04-18 Jakub Jelinek <jakub@redhat.com>
19295 PR tree-optimization/80443
19296 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
19297 instead of adding 1, subtract -1 and similarly instead of subtracting
19300 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
19302 PR rtl-optimization/80357
19303 * haifa-sched.c (tmp_bitmap): New variable.
19304 (model_recompute): Handle duplicate use records.
19305 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
19306 (free_global_sched_pressure_data): Free it.
19308 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
19311 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
19312 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
19313 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
19314 instead of SYSTEM_HEADER_DIR.
19316 2017-04-18 Jeff Law <law@redhat.com>
19318 PR middle-end/80422
19319 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
19320 predecessors after walking up the insn chain.
19322 2017-04-18 Jakub Jelinek <jakub@redhat.com>
19325 * dwarf2out.c (modified_type_die): Try harder not to emit internal
19326 sizetype type into debug info.
19328 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
19331 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
19332 unneeded test for TARGET_UPPER_REGS_SF.
19333 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
19335 2017-04-18 Jakub Jelinek <jakub@redhat.com>
19338 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
19339 instead of gsi_after_labels.
19341 2017-04-18 Jeff Law <law@redhat.com>
19343 * regcprop.c (maybe_mode_change): Avoid creating copies of the
19347 2017-04-13 Jeff Law <law@redhat.com>
19348 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
19349 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
19351 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
19354 * config/avr/avr.c (intl.h): Include it.
19355 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
19357 2017-04-18 Martin Liska <mliska@suse.cz>
19359 PR gcov-profile/78783
19360 * gcov-tool.c (gcov_output_files): Validate that destination
19361 file is either removed by the tool or by a user.
19363 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
19364 Guy Benyei <guybe@mellanox.com>
19366 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
19367 block, and do not negate it, the stored id is already negative.
19369 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
19371 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
19373 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
19376 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
19377 masks of options that should be turned off if the VSX vector
19378 options are turned off.
19379 (OTHER_P8_VECTOR_MASKS): Likewise.
19380 (OTHER_VSX_VECTOR_MASKS): Likewise.
19381 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
19382 rs6000_disable_incompatible_switches to validate no type switches
19384 (rs6000_incompatible_switch): New function to disallow turning on
19385 other vector options if -mno-vsx, -mno-power8-vector, or
19386 -mno-power9-vector are specified.
19388 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
19390 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
19392 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
19394 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
19395 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
19396 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
19397 (ARG_POINTER_CFA_OFFSET): Likewise.
19399 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
19401 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
19402 conditions to take advantage of various optimizations.
19404 2017-04-13 Jeff Law <law@redhat.com>
19406 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
19407 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
19408 (zero_extendsidi2_dext): Likewise.
19410 2017-04-13 Jakub Jelinek <jakub@redhat.com>
19413 * fold-const.c (fold_ternary_loc): Revert
19414 use op0 instead of fold_convert_loc (loc, type, arg0) part of
19417 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
19419 PR rtl-optimization/80343
19420 * lra-remat.c (update_scratch_ops): Assign original hard reg to
19421 new scratch pseudo.
19423 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
19426 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
19427 to ubsan_encode_value.
19429 2017-04-13 Jeff Law <law@redhat.com>
19431 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
19432 appearing in DEBUG_INSNs.
19434 2017-04-13 Martin Liska <mliska@suse.cz>
19436 PR gcov-profile/80413
19437 * gcov-io.c (gcov_write_string): Copy to buffer just when
19438 allocated size is greater than zero.
19440 2017-04-13 Jakub Jelinek <jakub@redhat.com>
19443 * dwarf2out.c (decls_for_scope): Ignore declarations of
19444 current_function_decl in BLOCK_NONLOCALIZED_VARS.
19446 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
19449 * ipa-visibility.c (non_local_p): Fix typos.
19450 (localize_node): When localizing symbol in same comdat group,
19451 dissolve the group only when we know external symbols are going
19453 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
19455 2017-04-12 Jakub Jelinek <jakub@redhat.com>
19457 PR tree-optimization/79390
19458 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
19459 order does not result in usable sequence, retry with reversed operand
19465 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
19466 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
19467 op0 instead of fold_convert_loc (loc, type, arg0).
19469 2017-04-12 Jeff Law <law@redhat.com>
19471 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
19472 has a delay slot in the generated code.
19474 * config/cris/cris.md (cris_preferred_reload_class): Return
19475 GENNONACR_REGS rather than GENERAL_REGS.
19477 2017-04-12 Jakub Jelinek <jakub@redhat.com>
19480 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
19481 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
19482 signedness of the result type.
19484 2017-04-12 Richard Biener <rguenther@suse.de>
19485 Jeff Law <law@redhat.com>
19487 PR tree-optimization/80359
19488 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
19489 trim stores to TARGET_MEM_REFs.
19491 2017-04-12 Richard Biener <rguenther@suse.de>
19493 PR tree-optimization/79390
19494 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
19495 threading case even more.
19497 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
19500 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
19501 for quad_address_p for TImode, instead of just not indexed_address.
19503 2017-04-12 Richard Biener <rguenther@suse.de>
19504 Bernd Edlinger <bernd.edlinger@hotmail.de>
19506 PR middle-end/79671
19507 * alias.c (component_uses_parent_alias_set_from): Handle
19508 TYPE_TYPELESS_STORAGE.
19509 (get_alias_set): Likewise.
19510 * tree-core.h (tree_type_common): Add typeless_storage flag.
19511 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
19512 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
19513 for types containing members with TYPE_TYPELESS_STORAGE.
19514 (place_field): Likewise.
19515 (layout_type): Likewise for ARRAY_TYPE.
19516 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
19517 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
19518 TYPE_TYPELESS_STORAGE.
19519 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
19521 2017-04-12 Jakub Jelinek <jakub@redhat.com>
19524 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
19525 first argument to type.
19527 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19531 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
19532 CONST0_RTX (mode) rather than const0_rtx where appropriate.
19533 (rs6000_expand_binop_builtin): Likewise.
19534 (rs6000_expand_ternop_builtin): Likewise; also add missing
19535 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
19536 vshasigma built-ins.
19537 * doc/extend.texi: Document that vec_xxpermdi's third argument
19538 must be a constant.
19540 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
19542 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
19543 Use shift_const cost parameter when calculating gain of STV shifts.
19545 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
19547 PR rtl-optimization/70478
19548 * lra-constraints.c (process_alt_operands): Check memory for
19549 disfavoring memory insn operand.
19551 2017-04-11 Jakub Jelinek <jakub@redhat.com>
19553 PR middle-end/80100
19554 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
19555 left shift in unsigned HOST_WIDE_INT type.
19557 PR rtl-optimization/80385
19558 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
19559 (not (neg X)) into (plus X -1) for complex or non-integral modes.
19562 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
19563 if they have any depend clauses.
19565 2017-04-11 Martin Liska <mliska@suse.cz>
19568 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
19569 * ipa-split.c (split_function): Create a local comdat symbol
19570 if caller is in a comdat group.
19572 2017-04-11 Martin Liska <mliska@suse.cz>
19575 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
19578 2017-04-11 Martin Sebor <msebor@redhat.com>
19580 PR middle-end/80364
19581 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
19582 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
19584 (directive::set_width, directive::set_precision, format_character):
19586 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
19589 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
19592 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
19593 conflict, set target->arch_name instead of target->cpu_name.
19595 2017-04-11 Richard Biener <rguenther@suse.de>
19597 PR tree-optimization/80374
19598 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
19599 build_zero_cst, remove fold_convertible_p check again.
19601 2017-04-11 Martin Liska <mliska@suse.cz>
19604 * ubsan.c (instrument_object_size): Do not instrument register
19607 2017-04-11 Jakub Jelinek <jakub@redhat.com>
19610 * config/i386/i386-builtin-types.def
19611 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
19612 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
19613 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
19614 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
19615 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
19616 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
19617 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
19618 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
19619 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
19620 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
19621 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
19622 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
19623 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
19624 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
19625 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
19626 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
19627 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
19628 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
19629 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
19630 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
19631 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
19632 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
19633 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
19634 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
19635 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
19636 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
19637 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
19638 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
19639 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
19640 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
19641 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
19642 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
19643 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
19644 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
19645 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
19646 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
19647 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
19648 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
19649 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
19650 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
19651 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
19652 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
19653 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
19654 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
19655 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
19656 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
19658 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
19659 flag to second_arg_count, handle 4 argument function type _COUNT
19660 aliases, handle second_arg_count on second argument rather than last.
19662 2017-04-10 Jeff Law <law@redhat.com>
19664 PR tree-optimization/80374
19665 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
19666 record anything if we can not convert integer_zero_node to the
19669 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
19672 * config/rs6000/rs6000.c (rs6000_option_override_internal):
19673 Enhance special handling given to the TARGET_P9_MINMAX option in
19674 relation to certain other options.
19676 2017-04-10 Bin Cheng <bin.cheng@arm.com>
19678 PR tree-optimization/80153
19679 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
19680 remove POINTER_PLUS_EXPR's base part directly, rather than through
19683 2017-04-10 Richard Biener <rguenther@suse.de>
19684 Bin Cheng <bin.cheng@arm.com>
19686 PR tree-optimization/80153
19687 * tree-affine.c (aff_combination_to_tree): Get base pointer from
19688 the first element of pointer type aff_tree. Build result expr in
19690 (add_elt_to_tree): Convert to type unconditionally. Remove other
19691 fold_convert calls.
19692 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
19693 (rewrite_use_nonlinear_expr): Check invariant using iv information.
19695 2017-04-10 Richard Biener <rguenther@suse.de>
19697 * tree-ssa-structalias.c (find_func_aliases): Properly handle
19700 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
19702 PR rtl-optimization/70478
19703 * lra-constraints.c (curr_small_class_check): New.
19704 (update_and_check_small_class_inputs): New.
19705 (process_alt_operands): Update curr_small_class_check. Disfavor
19706 alternative insn memory operands. Check available regs for small
19709 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
19712 * config/mips/mips.opt (-mvirt): Update description.
19713 * doc/invoke.texi (-mvirt): Likewise.
19715 2017-04-10 Richard Biener <rguenther@suse.de>
19717 PR middle-end/80362
19718 * fold-const.c (fold_binary_loc): Look at unstripped ops when
19719 looking for NEGATE_EXPR in -A / -B to A / B folding.
19721 2017-04-10 Martin Liska <mliska@suse.cz>
19723 PR gcov-profile/80224
19724 * gcov.c (print_usage): Fix usage string.
19725 (get_gcov_intermediate_filename): Remove.
19726 (output_gcov_file): Use both for normal and intermediate format.
19727 (generate_results): Do not initialize special file for
19728 intermediate format.
19730 2017-04-10 Richard Biener <rguenther@suse.de>
19732 PR tree-optimization/80304
19733 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
19736 2017-04-10 Nathan Sidwell <nathan@acm.org>
19739 * config/rs6000/rs6000.c (rs6000_vector_type): New.
19740 (rs6000_init_builtins): Use it.
19742 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19744 * config/arm/arm.md (<mrc>): Add mode to SET source.
19745 (<mrrc>): Likewise.
19747 2017-04-10 Richard Biener <rguenther@suse.de>
19749 PR middle-end/80344
19750 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
19752 2017-04-10 Jakub Jelinek <jakub@redhat.com>
19755 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
19756 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
19757 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
19758 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
19759 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
19760 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
19761 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
19762 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
19763 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
19764 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
19765 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
19766 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
19767 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
19768 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
19769 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
19770 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
19771 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
19772 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
19773 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
19774 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
19775 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
19776 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
19777 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
19779 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
19781 PR rtl-optimization/70478
19782 * lra-constraints.c: Reverse the last patch.
19784 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
19786 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
19787 Add comment for WCHAR_T.
19789 2017-04-08 Martin Liska <mliska@suse.cz>
19792 2017-04-07 Martin Liska <mliska@suse.cz>
19795 * ipa-split.c (split_function): Add function part to a same comdat
19798 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19801 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
19803 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
19805 * rs6000/rs6000.c (vec_load_pendulum): Rename...
19806 (vec_pairing): ...to this.
19807 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
19808 (rs6000_sched_init): Adjust for name change.
19809 (struct rs6000_sched_context): Likewise.
19810 (rs6000_init_sched_context): Likewise.
19811 (rs6000_set_sched_context): Likewise.
19813 2017-04-07 Jakub Jelinek <jakub@redhat.com>
19819 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
19821 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
19822 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
19823 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
19825 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
19827 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
19829 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
19831 PR rtl-optimization/70703
19832 * ira-color.c (update_conflict_hard_regno_costs): Use
19833 int64_t instead of HOST_WIDE_INT.
19835 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
19837 PR rtl-optimization/70478
19838 * lra-constraints.c (process_alt_operands): Disfavor alternative
19839 insn memory operands.
19841 2017-04-07 Jeff Law <law@redhat.com>
19843 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
19844 CALL and NOTE_INSN_CALL_ARG_LOCATION.
19846 2017-04-07 Martin Liska <mliska@suse.cz>
19849 * config/aarch64/aarch64.c (aarch64_process_target_attr):
19850 Show error message instead of an ICE.
19852 2017-04-07 Martin Liska <mliska@suse.cz>
19855 * ipa-split.c (split_function): Add function part to a same comdat
19858 2017-04-07 Richard Biener <rguenther@suse.de>
19860 PR middle-end/80341
19861 * tree.c (get_unwidened): Also handle ! for_type case for
19863 * convert.c (do_narrow): Split out from ...
19864 (convert_to_integer_1): ... here. Do not pass final truncation
19865 type to get_unwidened for TRUNC_DIV_EXPR.
19867 2017-04-07 Richard Biener <rguenther@suse.de>
19869 * tree-affine.c (wide_int_ext_for_comb): Take type rather
19871 (aff_combination_const): Adjust.
19872 (aff_combination_scale): Likewise.
19873 (aff_combination_add_elt): Likewise.
19874 (aff_combination_add_cst): Likewise.
19875 (aff_combination_convert): Likewise.
19876 (add_elt_to_tree): Likewise. Remove unused argument.
19877 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
19879 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
19881 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
19883 * config/arm/arm.c (arm_default_short_enums): Use
19884 ARM_DEFAULT_SHORT_ENUMS.
19885 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
19887 2017-04-06 Jakub Jelinek <jakub@redhat.com>
19890 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
19891 members with redundant out-of-class redeclaration.
19893 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
19896 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
19897 * config/i386/i386.md (*zero_extendsidi2):
19898 Add (?*x,*x) and (?*v,*v) alternatives.
19900 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
19903 * config/i386/i386.c (ix86_expand_builtin)
19904 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
19905 mode from insn data. Convert operands to insn operand mode.
19906 Copy operands that don't satisfy insn predicate to a register.
19908 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
19910 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
19913 2017-04-06 Richard Biener <rguenther@suse.de>
19915 PR tree-optimization/80334
19916 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
19917 preserve alignment of accesses.
19919 2017-04-06 Richard Biener <rguenther@suse.de>
19921 PR tree-optimization/80262
19922 * tree-sra.c (build_ref_for_offset): Preserve address-space
19924 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
19925 Drop useless address-space information on MEM_REF offsets.
19927 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
19929 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
19931 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
19933 PR rtl-optimization/70703
19934 * ira-color.c (update_conflict_hard_regno_costs): Use
19935 HOST_WIDE_INT instead of long.
19937 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
19940 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
19941 not defined for x86_64 target. Add -mmmx target option when __SSE2__
19943 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
19944 for x86_64 target. Handle -m3dnowa option.
19946 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
19948 PR rtl-optimization/70703
19949 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
19950 (update_conflict_hard_regno_costs): Use long instead of unsigned
19951 arithmetic for cost calculation.
19953 2017-04-05 Jakub Jelinek <jakub@redhat.com>
19954 Bernd Edlinger <bernd.edlinger@hotmail.de>
19957 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
19960 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
19963 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
19964 ptr_mode with Pmode throughout.
19965 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
19966 into probe_stack_range and use DImode.
19968 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
19971 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
19972 call_eh_return is true.
19974 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19976 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
19977 Initialize last_match_fntype_index.
19979 2017-04-05 Jakub Jelinek <jakub@redhat.com>
19982 * tree-nvr.c: Include internal-fn.h.
19983 (pass_return_slot::execute): Ignore internal calls without
19986 2017-04-04 Jakub Jelinek <jakub@redhat.com>
19987 Richard Biener <rguenther@suse.de>
19990 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
19991 captures used multiple times, except for the last use.
19992 * generic-match-head.c: Include gimplify.h.
19994 2017-04-04 Jakub Jelinek <jakub@redhat.com>
19996 PR tree-optimization/79390
19997 * target.h (struct noce_if_info): Declare.
19998 * targhooks.h (default_noce_conversion_profitable_p): Declare.
19999 * target.def (noce_conversion_profitable_p): New target hook.
20000 * ifcvt.h (struct noce_if_info): New type, moved from ...
20001 * ifcvt.c (struct noce_if_info): ... here.
20002 (noce_conversion_profitable_p): Renamed to ...
20003 (default_noce_conversion_profitable_p): ... this. No longer
20005 (noce_try_store_flag_constants, noce_try_addcc,
20006 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
20007 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
20008 instead of noce_conversion_profitable_p.
20009 * config/i386/i386.c: Include ifcvt.h.
20010 (ix86_option_override_internal): Don't override
20011 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
20012 (ix86_noce_conversion_profitable_p): New function.
20013 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
20014 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
20015 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
20016 * doc/tm.texi: Regenerated.
20018 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20020 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
20023 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
20026 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
20027 instructions for small multiply cores.
20029 2017-04-04 Jeff Law <law@redhat.com>
20031 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
20033 (mips_expand_vec_perm_const): Initialize elements in orig_perm
20034 that are not set by the loop over the elements.
20036 2017-04-04 Jakub Jelinek <jakub@redhat.com>
20039 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
20040 int mode, convert_modes it to mode as unsigned, otherwise use
20041 lowpart_subreg to mode rather than SImode.
20042 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
20043 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
20044 Use DImode instead of SImode for the shift count operand.
20045 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
20048 2017-04-04 Richard Biener <rguenther@suse.de>
20050 PR middle-end/80281
20051 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
20052 arithmetic done for the negate or the plus. Simplify.
20053 (A - (-B) -> A + B): Likewise.
20054 * fold-const.c (split_tree): Make sure to not negate pointers.
20056 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
20058 PR rtl-optimization/60818
20059 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
20060 a compare of comparisons with the thing compared if this results
20061 in a different machine mode.
20063 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
20065 * alias.c (base_alias_check): Fix typo in comment.
20066 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
20067 * cgraphunit.c (symbol_table::compile): Likewise.
20068 * collect2.c (maybe_run_lto_and_relink): Likewise.
20069 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
20070 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
20071 * config/avr/avr.c (avr_map_op_t): Likewise.
20072 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
20073 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
20074 * config/epiphany/epiphany.md (movcc): Likewise.
20075 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
20076 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
20078 * config/mips/mips.c (mips_save_restore_reg): Likewise.
20079 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
20080 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
20081 * config/sh/sh.c (sh_rtx_costs): Likewise.
20082 * fold-const.c (fold_truth_andor): Likewise.
20083 * genautomata.c (collapse_flag): Likewise.
20084 * gengtype.h (struct type::u::s): Likewise.
20085 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
20086 * input.c (FORMAT_AMOUNT): Likewise.
20087 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
20088 (known_aggs_to_agg_replacement_list): Likewise.
20089 * ipa-inline-analysis.c: Likewise.
20090 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
20091 * ipa-polymorphic-call.c
20092 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
20093 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
20094 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
20096 * modulo-sched.c (apply_reg_moves): Likewise.
20097 * omp-expand.c (build_omp_regions_1): Likewise.
20098 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
20099 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
20100 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
20101 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
20102 * value-prof.c: Likewise.
20103 * var-tracking.c (val_reset): Likewise.
20105 2017-04-03 Richard Biener <rguenther@suse.de>
20107 PR tree-optimization/80275
20108 * fold-const.c (split_address_to_core_and_offset): Handle
20111 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
20113 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
20114 descriptors is at least equal to that of functions.
20116 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
20118 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
20120 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
20123 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
20124 (mov<IMOD4:mode>): New expander.
20125 (*mov<IMOD4:mode>_internal): New insn and split pattern.
20127 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
20129 PR rtl-optimization/79405
20130 * fwprop.c (propagations_left): New variable.
20131 (forward_propagate_into): Decrement it.
20132 (fwprop_init): Initialize it.
20133 (fw_prop): If the variable has reached zero, stop propagating.
20134 (fwprop_addr): Ditto.
20136 2017-03-31 Jakub Jelinek <jakub@redhat.com>
20139 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
20140 a FUNCTION_DECL, pass it as decl instead of origin to
20143 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
20145 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
20148 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
20151 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
20152 TARGET_VSX_SMALL_INTEGER.
20154 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20156 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
20157 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
20159 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
20161 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
20162 extraction from odd-numbered MSA register.
20164 2017-03-31 Jakub Jelinek <jakub@redhat.com>
20166 PR middle-end/80173
20167 * expmed.c (store_bit_field_1): Don't attempt to create
20168 a word subreg out of hard registers wider than word if they
20169 have HARD_REGNO_NREGS of 1 for their mode.
20171 PR middle-end/80163
20172 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
20173 conversions to integer types wider than word and pointer.
20176 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
20177 (rtx_equal_for_cselib_p): Pass 0 to it.
20178 * cselib.c (cselib_hasher::equal): Likewise.
20179 (rtx_equal_for_cselib_1): Add depth argument. If depth
20180 is 128, don't look up VALUE locs and punt. Increment
20181 depth in recursive calls when walking VALUE locs.
20183 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
20185 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
20186 (make_gcov_file_name): Use the canonical path name for generating
20188 (read_line): Fix handling of files with ascii null bytes.
20190 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
20192 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
20193 to initialise a vector register instead
20194 of using a const_int.
20196 2017-03-30 Jakub Jelinek <jakub@redhat.com>
20198 PR translation/80189
20199 * gimplify.c (omp_default_clause): Use %qs instead of %s in
20200 diagnostic messages.
20202 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
20205 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
20206 (dfp_diex_<mode>): Update mode of operand 1.
20207 * doc/extend.texi (dxex, dxexq): Document change to return type.
20208 (diex, diexq): Document change to argument type.
20210 2017-03-30 Martin Jambor <mjambor@suse.cz>
20213 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
20214 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
20215 it reflects the signature changes performed at the callee side.
20216 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
20217 to cgraph_build_function_type_skip_args.
20218 (build_function_decl_skip_args): Adjust call to the above function.
20220 2017-03-30 Jakub Jelinek <jakub@redhat.com>
20223 * config/i386/sse.md
20224 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
20225 register as dest whenever it is a MEM not rtx_equal_p to the
20226 corresponding dup operand, and when forcing into reg move the
20227 reg into the memory afterwards.
20228 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
20229 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
20230 for the force_reg mode.
20231 (avx512vl_vextractf128<mode>): Use register as dest either
20232 always when a MEM, or when it is a MEM not rtx_equal_p to the
20233 corresponding dup operand, or even not when it is a CONST_VECTOR
20234 depending on the mode and lo vs. hi.
20235 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
20237 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
20238 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
20239 Likewise. Require that operands[2] is even.
20240 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
20241 Remove extraneous parens. Require that operands[2] is a multiple
20243 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
20244 operands[0] is a MEM if <mask_applied>, the predicates/constraints
20245 disallow memory then.
20247 2017-03-30 Richard Biener <rguenther@suse.de>
20249 PR tree-optimization/77498
20250 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
20251 to non-constants over backedges.
20253 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
20255 PR rtl-optimization/80233
20256 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
20257 as last_combined_insn. Do not test for BARRIER_P separately.
20259 2017-03-29 Andreas Schwab <schwab@suse.de>
20262 * calls.c (prepare_call_address): Convert funexp to Pmode before
20263 copying to temp reg.
20265 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20267 PR tree-optimization/80158
20268 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
20269 Handle possible future case of more than one alternate
20271 (replace_rhs_if_not_dup): Likewise.
20272 (replace_one_candidate): Likewise.
20274 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
20276 PR rtl-optimization/80193
20277 * ira.c (ira): Do not check allocation for LRA.
20279 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
20281 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
20282 (nvptx_output_simt_exit): Declare.
20283 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
20284 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
20285 (init_softstack_frame): Move initialization of crtl->is_leaf to...
20286 (nvptx_declare_function_name): ...here. Emit declaration of local
20287 memory space buffer for omp_simt_enter insn.
20288 (nvptx_output_unisimt_switch): New.
20289 (nvptx_output_softstack_switch): New.
20290 (nvptx_output_simt_enter): New.
20291 (nvptx_output_simt_exit): New.
20292 * config/nvptx/nvptx.h (struct machine_function): New fields
20293 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
20294 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
20295 (UNSPECV_SIMT_EXIT): Ditto.
20296 (omp_simt_enter_insn): New insn.
20297 (omp_simt_enter): New expansion.
20298 (omp_simt_exit): New insn.
20299 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
20301 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
20302 (expand_GOMP_SIMT_ENTER_ALLOC): New.
20303 (expand_GOMP_SIMT_EXIT): New.
20304 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
20305 (GOMP_SIMT_ENTER_ALLOC): Ditto.
20306 (GOMP_SIMT_EXIT): Ditto.
20307 * target-insns.def (omp_simt_enter): New insn.
20308 (omp_simt_exit): Ditto.
20309 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
20311 (lower_rec_simd_input_clauses): Implement SIMT privatization.
20312 (lower_rec_input_clauses): Likewise.
20313 (lower_lastprivate_clauses): Handle SIMT privatization.
20315 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
20316 (ompdevlow_adjust_simt_enter): New.
20317 (find_simtpriv_var_op): New.
20318 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
20319 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
20321 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
20322 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
20323 (copy_decl_for_dup_finish): Ditto.
20325 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
20327 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
20330 * config/i386/i386.c (ix86_option_override_internal): Always
20331 allow -mpreferred-stack-boundary=3 for 64-bit targets.
20333 2017-03-28 Bin Cheng <bin.cheng@arm.com>
20335 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
20337 2017-03-28 Bin Cheng <bin.cheng@arm.com>
20339 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
20340 mark new edge's irreducible flag accordign to it.
20341 (vect_do_peeling): Check loop preheader edge's irreducible flag
20342 and pass it to function slpeel_add_loop_guard.
20344 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
20346 PR tree-optimization/80218
20347 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
20348 Update block frequencies and counts.
20350 2017-03-28 Richard Biener <rguenther@suse.de>
20352 PR tree-optimization/78644
20353 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
20354 of a simplification result we may not use it at all.
20356 2017-03-28 Richard Biener <rguenther@suse.de>
20359 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
20360 without arguments, generate default definition of a SSA name.
20362 2017-03-28 Richard Biener <rguenther@suse.de>
20364 PR middle-end/80222
20365 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
20366 TYPE_REF_CAN_ALIAS_ALL references.
20367 * fold-const.c (fold_indirect_ref_1): Likewise.
20369 2017-03-28 Martin Liska <mliska@suse.cz>
20372 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
20373 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
20375 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
20376 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
20378 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
20379 (EXTRA_SPECS): Define.
20380 (SUBTARGET_EXTRA_SPECS): Likewise.
20381 (SUBTARGET_CPP_SPEC): Likewise.
20382 * config/arc/elf.h (EXTRA_SPECS): Renamed to
20383 SUBTARGET_EXTRA_SPECS.
20384 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
20386 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
20388 * config/arc/simdext.md (vst64_insn): Update pattern.
20389 (vld32wh_insn): Likewise.
20390 (vld32wl_insn): Likewise.
20391 (vld64_insn): Likewise.
20392 (vld32_insn): Likewise.
20394 2017-03-28 Marek Polacek <polacek@redhat.com>
20397 * fold-const.c (fold_comparison): Use protected_set_expr_location
20398 instead of SET_EXPR_LOCATION.
20400 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
20402 * tree.c (add_expr): Avoid name lookup warning.
20404 2017-03-27 Jeff Law <law@redhat.com>
20406 PR tree-optimization/80216
20407 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
20408 function name. Limit recursion depth.
20409 (record_temporary_equivalences): Corresponding changes.
20411 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
20413 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
20416 2017-03-27 Jakub Jelinek <jakub@redhat.com>
20419 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
20421 * cfgcleanup.c (reg_note_cfa_p): New array.
20422 (insns_have_identical_cfa_notes): New function.
20423 (old_insns_match_p): Don't cross-jump in between /f
20424 and non-/f instructions. If both i1 and i2 are frame related,
20425 verify all CFA notes, their order and content.
20427 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
20430 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
20431 HImode and SImode with zero extend to DImode to one insn.
20432 (bswap<mode>2_extenddi): Likewise.
20433 (bswapsi2_extenddi): Likewise.
20434 (bswaphi2_extendsi): Likewise.
20435 (bswaphi2): Combine bswap HImode and SImode into one insn.
20436 Separate memory insns from swapping register.
20437 (bswapsi2): Likewise.
20438 (bswap<mode>2): Likewise.
20439 (bswaphi2_internal): Delete, no longer used.
20440 (bswapsi2_internal): Likewise.
20441 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
20442 store, and gpr<-gpr swap insns.
20443 (bswap<mode>2_store): Likewise.
20444 (bswaphi2_reg): Register only splitter, combine with the splitter.
20445 (bswaphi2 splitter): Likewise.
20446 (bswapsi2_reg): Likewise.
20447 (bswapsi2 splitter): Likewise.
20448 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
20449 the insns into load, store, and register/register insns.
20450 (bswapdi2_ldbrx): Likewise.
20451 (bswapdi2_load): Likewise.
20452 (bswapdi2_store): Likewise.
20453 (bswapdi2_reg): Likewise.
20455 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
20457 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
20458 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
20460 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
20463 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
20465 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
20466 special handling for target option conflicts between dform
20467 options (-mpower9-dform, -mpower9-dform-vector,
20468 -mpower9-dform-scalar) and -mno-direct-move.
20470 2017-03-27 Richard Biener <rguenther@suse.de>
20472 PR tree-optimization/80181
20473 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
20475 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
20477 * config/arc/predicates.md (move_double_src_operand): Replace the
20478 call to move_double_src_operand with a call to address_operand.
20480 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
20482 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
20483 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
20484 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
20486 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
20488 * config/arc/predicates.md (long_immediate_loadstore_operand):
20489 Consider scaled addresses cases.
20491 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
20493 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
20494 restored when in interrupt.
20495 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
20496 doesn't have delay slot.
20498 2017-03-27 Richard Biener <rguenther@suse.de>
20501 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
20502 inlined thunk clones.
20504 2017-03-27 Jakub Jelinek <jakub@redhat.com>
20507 * asan.c (instrument_derefs): Copy over last operand from
20508 original COMPONENT_REF to the new COMPONENT_REF with
20509 DECL_BIT_FIELD_REPRESENTATIVE.
20510 * ubsan.c (instrument_object_size): Likewise.
20512 2017-03-27 Richard Biener <rguenther@suse.de>
20514 PR tree-optimization/80170
20515 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
20516 sure DR/SCEV didnt fold in constants we do not see when looking
20517 at the reference base alignment.
20519 2017-03-27 Richard Biener <rguenther@suse.de>
20521 PR middle-end/80171
20522 * gimple-fold.c (fold_ctor_reference): Properly guard against
20523 NULL return value from canonicalize_constructor_val.
20525 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
20528 * config/i386/i386.c (ix86_expand_builtin)
20529 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
20530 flags reg setting and flags reg using instructions.
20531 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
20532 clobbering instructions to zero extend op2.
20534 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
20536 * doc/install.texi (Configuration) <--with-aix-soname>:
20537 Update link to AIX ld.
20539 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
20541 PR rtl-optimization/80160
20542 PR rtl-optimization/80159
20543 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
20544 reg_alternate_class into account.
20546 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
20549 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
20550 to consider in curr_insn_transform.
20552 2017-03-24 Jakub Jelinek <jakub@redhat.com>
20554 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
20555 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
20556 and emit_mode_inner.
20558 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20560 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
20561 argument to the overloaded builtin variants. Use the new flag to
20562 deprecate certain builtin variants.
20563 * config/s390/s390-builtin-types.def: Add new builtin types.
20564 * config/s390/s390-builtins.h: Support new flags field for
20565 overloaded builtins.
20566 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
20567 (s390_macro_to_expand): Enable vector float data type.
20568 (s390_cpu_cpp_builtins_internal): Indicate support of the new
20569 builtins by incrementing the __VEC__ version number.
20570 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
20572 (s390_resolve_overloaded_builtin): Emit error messages depending
20573 on the builtin flags.
20574 * config/s390/s390.c (s390_expand_builtin): Support additional
20575 flags argument. Change error message to match the messages
20576 emitted in s390-c.c.
20577 * config/s390/s390.md: New UNSPEC_* constants.
20578 (op_type): Add new instruction types.
20579 * config/s390/vecintrin.h: Add new builtins and test data class
20581 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
20582 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
20583 (VEC_INEXACT, VEC_NOINEXACT): New constants.
20584 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
20585 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
20586 ("vec_mergel<mode>"): V_HW -> VEC_HW.
20588 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
20589 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
20590 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
20591 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
20593 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
20594 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
20595 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
20596 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
20598 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
20599 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
20600 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
20601 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
20602 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
20603 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
20604 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
20606 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
20607 ("vec_scatter_element<V_HW_4:mode>_DI")
20608 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
20609 ("vec_fpint<mode>", "vflls")
20610 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
20611 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
20612 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
20613 ("*vec_cmphe<mode>_cc"): ... these.
20615 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
20616 mode constant instead of magic value.
20618 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20620 * config/s390/s390.c (s390_expand_vec_compare): Support other
20621 vector floating point modes than just V2DF.
20622 (s390_expand_vcond): Likewise.
20623 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
20624 (s390_cannot_change_mode_class): Prevent mode changes between TF
20625 and V1TF in vector registers.
20626 * config/s390/s390.md (DF, SF): New mode attributes.
20627 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
20628 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
20629 SFmode support for VRs.
20630 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
20632 (VFT, VF_HW): New mode iterators.
20633 (vw, sdx): New mode attributes.
20634 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
20635 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
20636 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
20637 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
20638 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
20639 also the new vector floating point modes. Renaming to ...
20641 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
20642 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
20643 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
20644 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
20645 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
20646 ("vec_unordered<mode>"): ... these.
20648 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
20649 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
20650 ("*vec_extendv2df"): New insn definitions.
20652 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20654 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
20655 ("mulditi3_2", "*muldi3_sign"): New patterns.
20656 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
20657 rename the pattern definition.
20659 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20661 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
20663 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
20665 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20667 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
20668 instruction if possible.
20669 * config/s390/vector.md (vec_halfnumelts): New mode
20671 ("*vec_vllezlf<mode>"): New pattern.
20673 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20675 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
20676 ("popcountv4si2", "popcountv2di2"): Rename to ...
20677 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
20678 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
20680 ("popcount<mode>2_vxe"): New pattern.
20682 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20684 * common/config/s390/s390-common.c (processor_flags_table): Add
20686 * config.gcc: Add arch12.
20687 * config/s390/driver-native.c (s390_host_detect_local_cpu):
20688 Default to arch12 for unknown CPU model numbers.
20689 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
20690 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
20691 PROCESSOR_max sanity check.
20692 * config/s390/s390-opts.h (enum processor_type): Add
20694 * config/s390/s390.c (processor_table): Add arch12.
20695 (s390_expand_builtin): Add check for B_VXE flag.
20696 (s390_issue_rate): Add PROCESSOR_ARCH12.
20697 (s390_get_sched_attrmask): Likewise.
20698 (s390_get_unit_mask): Likewise.
20699 (s390_sched_score): Enable z13 scheduling for arch12.
20700 (s390_sched_reorder): Likewise.
20701 (s390_sched_variable_issue): Likewise.
20702 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
20704 (s390_tune_attr): Use z13 scheduling also for arch12.
20705 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
20706 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
20707 (TARGET_VXE_P): New macros.
20708 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
20709 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
20710 * config/s390/s390.opt: Add arch12 as processor_type.
20712 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20714 * config/s390/s390.md
20715 ("fixuns_truncdddi2", "fixuns_trunctddi2")
20716 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
20717 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
20719 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
20720 Rename expanders to ...
20722 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
20723 ("fixuns_truncdddi2_emu"): ... these.
20725 ("fixuns_trunc<mode>si2_emu"): New expander.
20727 ("*fixuns_truncdfdi2_z13"): Rename to ...
20728 ("*fixuns_truncdfdi2_vx"): ... this.
20730 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20732 * config/s390/2964.md: Remove the single element vector compare
20733 instructions which are no longer used.
20734 * config/s390/s390.c (s390_select_ccmode): Remove handling of
20736 (s390_canonicalize_comparison): Remove handling of DFmode
20738 (s390_expand_vec_compare_scalar): Remove function.
20739 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
20740 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
20742 ("*cmp<mode>_ccs"): Add wfcdb instruction.
20744 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20746 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
20748 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
20749 will anyway by matched by mov<mode>_64dfp.
20751 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20753 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
20754 vlef/vstef. Add missing operand to vleif.
20756 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20758 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
20759 pair for all vector types with 64 bit elements.
20760 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
20761 * config/s390/vector.md (V_HW_64): ... here.
20762 (V_128_NOSINGLE): New mode iterator.
20763 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
20764 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
20765 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
20766 ("*vec_load_pairv2di"): Change to ...
20767 ("*vec_load_pair<mode>"): ... this one.
20769 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20771 * config/s390/constraints.md: Add comments.
20772 (jKK): Reject element sizes > 8 bytes.
20773 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
20775 * config/s390/s390.md: Add the s_operand checks formerly in
20776 s390_split_ok_p to various splitters where they are still
20778 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
20779 for 128 bit vectors. Plus two splitters.
20781 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20783 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
20786 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20789 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
20790 error if the boundary argument is not constant.
20792 2017-03-24 Jakub Jelinek <jakub@redhat.com>
20794 PR rtl-optimization/80112
20795 * loop-doloop.c (doloop_condition_get): Don't check condition
20796 if cmp isn't SET with IF_THEN_ELSE src.
20798 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20800 PR tree-optimization/80158
20801 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
20802 replacing a candidate statement, also replace it for the
20803 candidate's alternate interpretation.
20804 (replace_rhs_if_not_dup): Likewise.
20805 (replace_one_candidate): Likewise.
20807 2017-03-24 Richard Biener <rguenther@suse.de>
20809 PR tree-optimization/80167
20810 * graphite-isl-ast-to-gimple.c
20811 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
20813 (translate_isl_ast_to_gimple::get_rename): Likewise.
20815 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
20817 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
20818 handling of certain combinations of target options, including the
20819 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
20820 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
20822 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20825 * config/arm/arm.md (*load_multiple): Add reload_completed to
20826 matching condition.
20828 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20829 Richard Biener <rguenth@suse.de>
20831 PR tree-optimization/79908
20832 PR tree-optimization/80136
20833 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
20834 been cast away, gimplify_and_add suffices.
20836 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
20838 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
20840 2017-03-23 Richard Biener <rguenther@suse.de>
20842 PR tree-optimization/80032
20843 * gimplify.c (gimple_push_cleanup): Forced unconditional
20844 cleanups still have to go to the conditional_cleanups
20847 2017-03-22 Jakub Jelinek <jakub@redhat.com>
20849 PR tree-optimization/80072
20850 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
20852 (next_operand_entry_id): Change type to unsigned int.
20853 (sort_by_operand_rank): Make sure to return the right return value
20854 even if unsigned fields are bigger than INT_MAX.
20855 (struct oecount): Change cnt and id type to unsigned int.
20856 (oecount_hasher::equal): Formatting fix.
20857 (oecount_cmp): Make sure to return the right return value
20858 even if unsigned fields are bigger than INT_MAX.
20859 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
20862 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
20863 TREE_READONLY on result if writing it more than once.
20866 * doc/invoke.texi (-fsanitize=thread): Document that with
20867 -fnon-call-exceptions atomics are not able to throw
20871 * tsan.c: Include tree-eh.h.
20872 (instrument_builtin_call): Call maybe_clean_eh_stmt or
20873 maybe_clean_or_replace_eh_stmt where needed.
20874 (instrument_memory_accesses): Add cfg_changed argument.
20875 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
20876 if it returned true.
20877 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
20879 PR rtl-optimization/63191
20880 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
20881 wrapper function, moved the whole old content into ...
20882 (ix86_delegitimize_address_1): ... this. New inline function.
20883 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
20884 true as last argument instead of ix86_delegitimize_address.
20886 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
20888 * config/aarch64/aarch64.c (generic_branch_cost): Copy
20889 cortexa57_branch_cost.
20891 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
20893 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
20895 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20898 * doc/md.texi (Constraints): Document wA constraint.
20899 * config/rs6000/constraints.md (wA): New.
20900 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
20901 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
20902 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
20903 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
20905 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
20908 * gimplify.c (is_oacc_declared): New function.
20909 (oacc_default_clause): Use it to set default flags for acc declared
20910 variables inside parallel regions.
20911 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
20912 declared variables.
20913 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
20914 declare attribute to any decl as necessary.
20916 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
20919 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
20920 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
20921 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
20922 (arm_arch_lpae): This.
20923 * config/arm/arm.c (arm_arch7ve): Rename into ...
20924 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
20925 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
20928 2017-03-22 Martin Liska <mliska@suse.cz>
20931 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
20932 error message instead of an ICE.
20934 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20936 * doc/extend.texi (6.11 Additional Floating Types): Revise.
20938 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
20940 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
20942 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
20945 2017-03-21 Martin Sebor <msebor@redhat.com>
20947 * doc/extend.texi: Use "cannot" instead of "can't."
20948 * doc/hostconfig.texi: Same.
20949 * doc/install.texi: Same.
20950 * doc/invoke.texi: Same.
20951 * doc/loop.texi: Same.
20952 * doc/md.texi: Same.
20953 * doc/objc.texi: Same.
20954 * doc/rtl.texi: Same.
20955 * doc/tm.texi: Same.
20956 * doc/tm.texi.in: Same.
20957 * doc/trouble.texi: Same.
20959 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
20962 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
20963 (collect_checksum_attributes): Set it.
20964 (die_checksum_ordered): Use it.
20966 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20968 PR tree-optimization/79908
20969 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
20970 change: For a VA_ARG whose LHS has been cast away, use
20971 force_gimple_operand to construct the side effects.
20973 2017-03-21 David Malcolm <dmalcolm@redhat.com>
20975 PR translation/80001
20976 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
20977 more amenable to translation.
20978 (oacc_loop_auto_partitions): Likewise.
20980 2017-03-21 Marek Polacek <polacek@redhat.com>
20981 Martin Sebor <msebor@redhat.com>
20983 PR tree-optimization/80109
20984 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
20985 on INTEGRAL_TYPE_P.
20987 2017-03-21 Jakub Jelinek <jakub@redhat.com>
20988 Segher Boessenkool <segher@kernel.crashing.org>
20991 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
20992 check reg_used_between_p between insn and one of succ or succ2
20993 depending on if succ is artificial insn not inserted into insn
20996 2017-03-21 Martin Liska <mliska@suse.cz>
20998 PR gcov-profile/80081
20999 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
21000 * doc/gcc.texi: Include gcov-dump stuff.
21001 * doc/gcov-dump.texi: New file.
21003 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
21005 PR rtl-optimization/79150
21006 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
21007 conditional jump, if the jump is the last insn of the loop.
21009 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21010 Richard Biener <rguenth@suse.de>
21012 PR tree-optimization/79908
21013 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
21014 been cast away, use force_gimple_operand to construct the side
21017 2017-03-21 Martin Liska <mliska@suse.cz>
21019 PR libfortran/79956
21020 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
21023 2017-03-21 Brad Spengler <spender@grsecurity.net>
21026 * plugin.c (htab_hash_plugin): New function.
21027 (add_new_plugin): Use it and adjust.
21028 (parse_plugin_arg_opt): Adjust.
21029 (init_one_plugin): Likewise.
21031 2017-03-21 Richard Biener <rguenther@suse.de>
21033 PR tree-optimization/80032
21034 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
21035 if set force the cleanup to happen unconditionally.
21036 (gimplify_target_expr): Push inserted clobbers with force_uncond
21037 to avoid them being removed by control-dependent DCE.
21039 2017-03-21 Richard Biener <rguenther@suse.de>
21041 PR tree-optimization/80122
21042 * tree-inline.c (copy_bb): Do not expans va-arg packs or
21043 va_arg_pack_len when the inlined call stmt requires pack
21045 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
21047 2017-03-21 Jakub Jelinek <jakub@redhat.com>
21050 * tsan.c (instrument_builtin_call): If the memory model argument
21051 is not a constant, assume it is valid.
21054 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
21057 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
21059 PR rtl-optimization/79910
21060 * combine.c (can_combine_p): Do not allow combining an I0 or I1
21061 if its dest is used by an insn before I2 (other than the combined
21062 insns themselves, which are properly handled already).
21064 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
21067 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
21069 * combine.c (record_used_regs): New static function.
21070 (try_combine): Handle situations where there is an additional
21071 instruction between I2 and I3 which needs to have a LOG_LINK
21075 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
21077 * combine.c (try_combine): Delete redundant i1 test. Call
21078 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
21080 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21083 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
21084 alternatives 13/14.
21086 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21088 PR tree-optimization/80054
21089 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
21090 the optimization if a PHI or any of its arguments is not dominated
21091 by the candidate's basis. Use gphi* rather than gimple* as
21093 (replace_profitable_candidates): Clean up a gimple* variable that
21094 should be a gphi* variable.
21096 2017-03-20 Martin Sebor <msebor@redhat.com>
21099 * doc/extend.texi (attribute constructor): Document present limitation.
21101 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
21104 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
21105 __POWER9_VECTOR__ #ifdef control, change template definition to
21106 use Power9-specific built-in function.
21107 (vec_any_eq): Likewise.
21108 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
21109 to control outcomes from this test.
21110 (vector_ae_<mode>p): For VEC_F modes, likewise.
21112 2017-03-20 Ian Lance Taylor <iant@google.com>
21114 * config/i386/i386.c (ix86_function_regparm): Save an extra
21115 register for -fsplit-stack with DECL_STATIC_CHAIN.
21117 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
21120 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
21121 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
21123 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
21125 * config/riscv/riscv.c (riscv_print_operand): Use "fence
21127 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
21130 2017-03-20 Marek Polacek <polacek@redhat.com>
21133 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
21135 2017-03-20 Richard Biener <rguenther@suse.de>
21137 PR tree-optimization/80113
21138 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
21139 allocate extra SSA name for PHI def.
21140 (add_close_phis_to_outer_loops): Likewise.
21141 (add_close_phis_to_merge_points): Likewise.
21142 (copy_loop_close_phi_args): Likewise.
21143 (copy_cond_phi_nodes): Likewise.
21145 2017-03-20 Martin Liska <mliska@suse.cz>
21147 PR middle-end/79753
21148 * tree-chkp.c (chkp_build_returned_bound): Do not build
21149 returned bounds for a LHS that's not a BOUNDED_P type.
21151 2017-03-20 Martin Liska <mliska@suse.cz>
21155 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
21156 COMPLEX_CST and VECTOR_CST.
21158 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21161 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
21162 target operand. A new splitter adds the clobber statement in case
21163 the target operand is dead anyway.
21165 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
21167 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
21168 to age-old versions of binutils and glibc.
21170 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
21172 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
21174 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
21176 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
21178 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
21180 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
21181 requirement for binutils 2.13.
21183 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
21185 * combine.c (try_combine): Delete redundant i1 test. Call
21186 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
21188 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
21190 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
21191 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
21193 <riscv64-*-elf>: Re-arrange section
21194 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
21195 <riscv32-*-linux>: Likewise.
21196 <riscv64-*-elf>: Likewise
21197 <riscv64-*-linux>: Likewise.
21199 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
21202 * aarch64.opt(verbose-cost-dump): Fix typo.
21204 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
21207 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
21208 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
21210 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
21212 * reload.c (find_reloads): When reloading a nonoffsettable address,
21213 use RELOAD_OTHER for it and its address reloads.
21215 PR rtl-optimization/79910
21216 * combine.c (record_used_regs): New static function.
21217 (try_combine): Handle situations where there is an additional
21218 instruction between I2 and I3 which needs to have a LOG_LINK
21221 2017-03-17 Jeff Law <law@redhat.com>
21223 PR tree-optimization/71437
21224 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
21225 conditional in the hash table first.
21226 (vrp_dom_walker::before_dom_children): Extract condition from
21227 ASSERT_EXPR. Record condition, its inverion and any implied
21228 conditions as well.
21230 2017-03-17 Marek Polacek <polacek@redhat.com>
21231 Markus Trippelsdorf <markus@trippelsdorf.de>
21233 PR tree-optimization/80079
21234 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
21237 2017-03-17 Richard Biener <rguenther@suse.de>
21239 PR middle-end/80075
21240 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
21241 Properly verify the LHS before the RHS possibly claims to be
21243 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
21246 2017-03-17 Martin Jambor <mjambor@suse.cz>
21248 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
21249 (List of -O2 options): Likewise.
21250 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
21253 2017-03-17 Tom de Vries <tom@codesourcery.com>
21255 * gcov-dump.c (print_usage): Print bug_report_url.
21257 2017-03-17 Richard Biener <rguenther@suse.de>
21259 PR middle-end/80050
21260 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
21261 (parser::peek): Likewise.
21263 2017-03-17 Richard Biener <rguenther@suse.de>
21265 PR tree-optimization/80048
21266 * sese.c (free_sese_info): Properly release rename_map and
21267 copied_bb_map elements.
21269 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
21271 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
21272 Add linked-list forward and backlinks. Insert on
21273 construction, remove on destruction.
21274 (class pass_store_merging): Add m_stores_head field.
21275 (pass_store_merging::terminate_and_process_all_chains):
21276 Iterate over m_stores_head list.
21277 (pass_store_merging::terminate_all_aliasing_chains):
21279 (pass_store_merging::execute): Check for debug stmts first.
21280 Push new chains onto the m_stores_head stack.
21282 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
21285 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
21286 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
21287 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
21289 2017-03-16 Jeff Law <law@redhat.com>
21291 PR tree-optimization/71437
21292 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
21293 member function. Implementation moved into after_dom_children
21294 member function and into the threader's thread_outgoing_edges
21296 (dom_opt_dom_walker::after_dom_children): Simplify by moving
21297 some code into new thread_outgoing_edges.
21298 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
21299 definition. Simplify marker handling (do it here). Assume we always
21300 have the available expression and the const/copies tables.
21301 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
21303 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
21304 * tree-vrp.c (equiv_stack): No longer file scoped.
21305 (vrp_dom_walker): New class.
21306 (vrp_dom_walker::before_dom_children): New member function.
21307 (vrp_dom_walker::after_dom_children): Likewise.
21308 (identify_jump_threads): Setup domwalker. Use it rather than
21309 walking edges in a random order by hand. Simplify setup/finalization.
21310 (finalize_jump_threads): Remove.
21311 (vrp_finalize): Do not call identify_jump_threads here.
21312 (execute_vrp): Do it here instead and call thread_through_all_blocks
21315 PR tree-optimization/71437
21316 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
21318 (simplify_stmt_for_jump_threading): Add basic_block argument. All
21320 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
21321 (dom_opt_dom_walker::thread_across_edge): Remove
21322 handle_dominating_asserts argument. All callers changed.
21323 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
21324 changes. Remove calls to lhs_of_dominating_assert. Other
21325 uses of handle_dominating_asserts turn into unconditional code
21326 (simplify_control_stmt_condition_1): Likewise.
21327 (simplify_control_stmt_condition): Likewise.
21328 (thread_through_normal_block, thread_across_edge): Likewise.
21329 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
21330 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
21331 object if it is not an SSA_NAME.
21332 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
21333 before calling into the VRP specific simplifiers.
21334 (identify_jump_threads): Remove handle_dominating_asserts
21337 2017-03-16 Jakub Jelinek <jakub@redhat.com>
21340 * tree-diagnostic.c (default_tree_printer): No longer static.
21341 * tree-diagnostic.h (default_tree_printer): New prototype.
21343 2017-03-16 Tamar Christina <tamar.christina@arm.com>
21345 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
21346 Change ins into fmov.
21348 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21350 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
21351 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
21352 Use h_con constraint for operand 1.
21353 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
21354 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
21356 2017-03-15 Jeff Law <law@redhat.com>
21358 PR tree-optimization/71437
21359 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
21360 (record_temporary_equivalences): Use it.
21362 PR tree-optimization/71437
21363 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
21364 tree-ssa-scopedtables.
21365 (lookup_avail_expr, build_and_record_new_cond): Likewise.
21366 (record_conditions, record_cond, vuse_eq): Likewise.
21367 (record_edge_info): Adjust to API tweak of record_conditions.
21368 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
21369 (record_temporary_equivalences, optimize_stmt): Likewise.
21370 (eliminate_redundant_computations): Likewise.
21371 (record_equivalences_from_stmt): Likewise.
21372 * tree-ssa-scopedtables.c: Include options.h and params.h.
21373 (vuse_eq): New function, moved from tree-ssa-dom.c
21374 (build_and_record_new_cond): Likewise.
21375 (record_conditions): Likewise. Accept vector of conditions rather
21376 than edge_equivalence structure for first argument.
21377 for the first argument.
21378 (avail_exprs_stack::lookup_avail_expr): New member function, moved
21379 from tree-ssa-dom.c.
21380 (avail_exprs_stack::record_cond): Likewise.
21381 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
21382 from tree-ssa-dom.c.
21383 (avail_exprs_stack): Add new member functions lookup_avail_expr
21385 (record_conditions): Declare.
21387 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
21390 * lra-constraints.c (process_alt_operands): Increase reject for
21391 reloading an input/output operand.
21393 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
21396 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
21397 insns to convert from signed/unsigned char/short to IEEE 128-bit
21399 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
21401 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
21404 * config/i386/i386.c (ix86_vector_duplicate_value): Create
21405 subreg of inner mode for values already in registers.
21407 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
21409 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
21410 iteration reg is used after the loop.
21412 2017-03-14 Martin Sebor <msebor@redhat.com>
21414 PR tree-optimization/79800
21415 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
21416 precision in negative-positive range.
21417 (format_floating): Call non-const overload with adjusted precision.
21419 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
21422 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
21425 2017-03-14 Martin Sebor <msebor@redhat.com>
21427 PR middle-end/80020
21428 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
21429 * builtins.def (aligned_alloc): Use it.
21432 * Makefile.in (GTFILES): Add calls.c.
21433 * calls.c: Include "gt-calls.h".
21435 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
21437 PR rtl-optimization/79728
21438 * regs.h (struct target_regs): New field
21439 x_contains_allocatable_regs_of_mode.
21440 (contains_allocatable_regs_of_mode): New macro.
21441 * reginfo.c (init_reg_sets_1): Initialize it, and change
21442 contains_reg_of_mode so it includes global regs as well.
21443 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
21444 rather than contains_regs_of_mode.
21446 2017-03-14 Martin Liska <mliska@suse.cz>
21448 * doc/invoke.texi: Document options that can't be combined with
21449 -fcheck-pointer-bounds.
21451 2017-03-14 Martin Liska <mliska@suse.cz>
21453 PR middle-end/79831
21454 * doc/invoke.texi (-Wchkp): Document the option.
21456 2017-03-14 Martin Liska <mliska@suse.cz>
21458 * Makefile.in: Install gcov-dump.
21460 2017-03-14 Martin Liska <mliska@suse.cz>
21462 * multiple_target.c (expand_target_clones): Bail out for
21463 an invalid attribute.
21465 2017-03-14 Richard Biener <rguenther@suse.de>
21467 * alias.c (struct alias_set_entry): Pack properly.
21468 * cfgloop.h (struct loop): Likewise.
21469 * cse.c (struct set): Likewise.
21470 * ipa-utils.c (struct searchc_env): Likewise.
21471 * loop-invariant.c (struct invariant): Likewise.
21472 * lra-remat.c (struct cand): Likewise.
21473 * recog.c (struct change_t): Likewise.
21474 * rtl.h (struct address_info): Likewise.
21475 * symbol-summary.h (function_summary): Likewise.
21476 * tree-loop-distribution.c (struct partition): Likewise.
21477 * tree-object-size.c (struct object_size_info): Likewise.
21478 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
21479 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
21480 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
21481 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
21482 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
21483 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
21484 (struct _stmt_vec_info): Likewise.
21486 2017-03-14 Martin Liska <mliska@suse.cz>
21489 * multiple_target.c (create_dispatcher_calls): Check that
21490 a target can create a function dispatcher.
21492 2017-03-14 Martin Liska <mliska@suse.cz>
21495 * multiple_target.c (expand_target_clones): Drop local.local
21496 flag for default implementation.
21498 2017-03-14 Richard Biener <rguenther@suse.de>
21500 PR tree-optimization/80030
21501 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
21503 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
21505 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
21506 gcc_fallthrough() instead of __attribute__((fallthrough));
21508 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
21510 * doc/gcc.texi: Remove "up" link to (DIR).
21511 * doc/gccint.texi: Ditto.
21513 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
21515 * doc/install.texi (Specific) <avr>: Remove reference to
21518 2017-03-13 Jeff Law <law@redhat.com>
21520 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
21521 attribute rather than comments.
21523 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
21524 match_scratch operand is highest.
21526 2017-03-13 Martin Liska <mliska@suse.cz>
21528 PR middle-end/78339
21529 * ipa-pure-const.c (warn_function_noreturn): If the declarations
21530 is a CHKP clone, use original declaration.
21532 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
21534 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
21535 (arc_conditional_register_usage): Use a different allocation order
21536 when optimizing for size.
21537 * common/config/arc/arc-common.c (arc_option_optimization_table):
21538 Section anchors default on when optimizing for size.
21540 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
21542 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
21544 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
21546 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
21547 * config/arc/arc.md (cpu_facility): Add cd variant.
21548 (*movqi_insn): Add code density variant.
21549 (*movhi_insn): Likewise.
21550 (*movqi_insn): Likewise.
21551 (*addsi3_mixed): Likewise.
21552 (subsi3_insn): Likewise.
21554 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
21556 * config/arc/arc.md (movsi_cond_exec): Update constraint.
21558 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
21560 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
21561 expressions with MINUS and UNARY ops.
21563 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21566 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
21568 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
21569 between vec_select and vector argument.
21570 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
21571 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
21572 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
21573 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
21574 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
21575 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
21577 2017-03-13 Richard Biener <rguenther@suse.de>
21580 * params.def (vect-max-peeling-for-alignment): Fix typo.
21582 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
21584 * doc/install.texi (Specific) <mips-*-*>: Remove description of
21585 issue that only occurred with binutils below 2.18.
21587 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
21589 * doc/install.texi (Specific) <cris-axis-elf>: No longer
21590 refer to binutils 2.11/2.12 minimum.
21592 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
21594 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
21595 ftp.kernel.org and simplify binutils requirement.
21597 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
21599 * doc/invoke.texi (Warning Options): Fix spelling of link-time
21601 (Optimize Options): Ditto. Also remove redundancy.
21603 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21605 PR translation/79848
21606 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
21608 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
21609 to G_ to avoid double translation.
21611 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21613 PR translation/79923
21614 * auto-profile.c (get_combined_location): Convert leading
21615 character of diagnostics to lower case and remove trailing period.
21616 (read_profile): Likewise for various diagnostics.
21617 * config/arm/arm.c (arm_option_override): Remove trailing period
21618 from various diagnostics.
21619 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
21620 (msp430_expand_delay_cycles): Likewise.
21622 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21625 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
21626 full command-line argument, rather than just "str".
21627 (aarch64_validate_march): Likewise.
21628 (aarch64_validate_mtune): Likewise.
21630 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
21632 PR rtl-optimization/78911
21633 * lra-assigns.c (must_not_spill_p): New function.
21634 (spill_for): Use it.
21636 2017-03-10 Jakub Jelinek <jakub@redhat.com>
21638 PR tree-optimization/79981
21639 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
21640 ATOMIC_COMPARE_EXCHANGE ifn result.
21641 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
21642 IFN_ATOMIC_COMPARE_EXCHANGE.
21644 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21647 * opts.c (parse_sanitizer_options): Add missing question mark to
21648 "did you mean" message.
21650 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21652 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
21654 (VMULEUH_UNS): Likewise.
21655 (VMULOUB_UNS): Likewise.
21656 (VMULOUH_UNS): Likewise.
21657 * config/rs6000/rs6000.c (builtin_function_type): Remove
21658 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
21660 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21663 * read-rtl-function.c (function_reader::read_rtx_operand): Update
21664 x with result of extra_parsing_for_operand_code_0.
21665 (function_reader::extra_parsing_for_operand_code_0): Convert
21666 return type from void to rtx, returning x. When reading
21667 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
21668 larger size containing struct block_symbol.
21670 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
21672 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
21673 -mfloat128-hardware without -m64.
21675 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
21678 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
21679 entries to the case statement that marks unsigned arguments to
21680 overloaded functions.
21682 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
21684 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
21685 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
21687 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
21690 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
21691 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
21693 2017-03-10 Martin Liska <mliska@suse.cz>
21697 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
21698 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
21701 2017-03-10 Olivier Hainque <hainque@adacore.com>
21703 * tree-switch-conversion (array_value_type): Start by resetting
21704 candidate type to it's main variant.
21706 2017-03-10 Jakub Jelinek <jakub@redhat.com>
21708 PR rtl-optimization/79909
21709 * combine.c (try_combine): Use simplify_replace_rtx on individual
21710 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
21711 of the whole CALL_INSN_FUNCTION_USAGE.
21713 PR tree-optimization/79972
21714 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
21715 get_range_info on SSA_NAMEs. Formatting fixes.
21717 2017-03-10 Richard Biener <rguenther@suse.de>
21718 Jakub Jelinek <jakub@redhat.com>
21720 PR tree-optimization/77975
21721 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
21722 edge to be constant.
21723 (get_val_for): For constant x return it. Formatting fix.
21724 (loop_niter_by_eval): Avoid pointless looping if the next iteration
21725 would use the same bases as the current one.
21727 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21729 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
21730 instead of vec_select for V1TImode.
21731 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
21733 (VSX_LE_128): Add V1TI to this mode iterator.
21734 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
21735 (*vsx_le_perm_store_<mode>): Likewise.
21736 (pre-reload splitter for VSX stores): Likewise.
21737 (post-reload splitter for VSX stores): Likewise.
21738 (*vsx_xxpermdi2_le_<mode>): Likewise.
21739 (*vsx_lxvd2x2_le_<mode>): Likewise.
21740 (*vsx_stxvd2x2_le_<mode>): Likewise.
21742 2017-03-09 Michael Eager <eager@eagercon.com>
21744 Correct failures with --enable-checking=yes,rtl.
21746 * config/microblaze/microblaze.c (microblaze_expand_shift):
21747 Replace GET_CODE test with CONST_INT_P and INTVAL test with
21748 test for const0_rtx.
21749 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
21750 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
21752 2017-03-09 Richard Biener <rguenther@suse.de>
21754 PR tree-optimization/79977
21755 * graphite-scop-detection.c (scop_detection::merge_sese):
21756 Handle the case of extra exits to blocks dominating the entry.
21758 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
21760 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
21763 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
21765 PR rtl-optimization/79949
21766 * lra-constraints.c (process_alt_operands): Check memory when
21767 trying to predict a cycle. Print about the overall increase.
21769 2017-03-09 Richard Biener <rguenther@suse.de>
21771 PR middle-end/79971
21772 * gimple-expr.c (useless_type_conversion_p): Preserve
21773 TYPE_SATURATING for fixed-point types.
21775 2017-03-09 Richard Biener <rguenther@suse.de>
21778 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
21779 alignment of BLKmode params.
21781 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21784 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
21785 (VALL_NO_V2Q): Likewise.
21787 * config/aarch64/aarch64-simd.md
21788 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
21790 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
21791 VALL_NO_V2Q mode iterator.
21792 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
21794 2017-03-09 Martin Liska <mliska@suse.cz>
21796 PR tree-optimization/79631
21797 * tree-chkp-opt.c (chkp_is_constant_addr): Call
21798 tree_int_cst_sign_bit just for INTEGER constants.
21800 2017-03-09 Martin Liska <mliska@suse.cz>
21804 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
21807 2017-03-09 Marek Polacek <polacek@redhat.com>
21810 * tree.c (inchash::add_expr): Handle TREE_VEC.
21812 2017-03-09 Martin Liska <mliska@suse.cz>
21815 (chkp_narrow_size_and_offset): New function.
21816 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
21817 (void chkp_parse_bit_field_ref): New function.
21818 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
21819 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
21821 2017-03-09 Martin Liska <mliska@suse.cz>
21824 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
21825 (chkp_find_bounds_1): Remove gcc_unreachable.
21827 2017-03-09 Jakub Jelinek <jakub@redhat.com>
21830 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
21831 BUILT_IN_SYNC*, determine the access type from the size suffix and
21832 always build a MEM_REF with that type. Handle forgotten
21833 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
21836 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
21837 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
21838 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
21839 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
21840 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
21841 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
21842 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
21843 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
21844 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
21845 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
21846 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
21847 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
21848 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
21849 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
21850 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
21851 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
21852 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
21853 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
21854 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
21855 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
21856 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
21857 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
21858 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
21859 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
21860 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
21861 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
21862 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
21863 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
21864 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
21865 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
21866 definitions outside of __OPTIMIZE__ guarded section.
21869 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
21870 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
21871 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
21872 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
21875 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21877 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
21878 ("vfenez<mode>"): Add missing constraints.
21880 2017-03-08 Martin Sebor <msebor@redhat.com>
21883 * config/nds32/nds32.c (nds32_option_override):
21884 Fix misspelled diagnostic.
21886 2017-03-08 Jakub Jelinek <jakub@redhat.com>
21889 * gimplify.c (gimplify_omp_for): Replace index var in outer
21890 taskloop statement with an artificial variable and add
21891 OMP_CLAUSE_PRIVATE clause for it.
21893 2017-03-08 Richard Biener <rguenther@suse.de>
21895 PR tree-optimization/79955
21896 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
21897 for accesses that are completely outside of the variable.
21899 2017-03-08 Andrew Haley <aph@redhat.com>
21901 PR tree-optimization/79943
21902 * tree-ssa-loop-split.c (compute_new_first_bound): When
21903 calculating the new upper bound, (END-BEG) should be added, not
21906 2017-03-08 Jakub Jelinek <jakub@redhat.com>
21908 * config/avr/avr.md (setmemhi): Make sure match_dup
21909 operand number comes before match_scratch.
21911 2017-03-08 Richard Biener <rguenther@suse.de>
21913 PR tree-optimization/79920
21914 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
21915 with ncopies == 1 to ...
21916 (vect_transform_slp_perm_load): ... here. Properly compute
21917 all element loads by iterating VF times over the group. Do
21918 not handle ncopies (computed in a broken way) in
21919 vect_create_mask_and_perm.
21921 2017-03-08 Jakub Jelinek <jakub@redhat.com>
21924 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
21925 is a uniform vector, use uniform_vector_p return value instead of
21926 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
21928 2017-03-07 Marek Polacek <polacek@redhat.com>
21930 PR middle-end/79809
21931 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
21932 (alloca_call_type): Likewise.
21934 2017-03-07 Martin Liska <mliska@suse.cz>
21936 * gcov.c (process_args): Put comment to correct location.
21938 2017-03-07 Martin Liska <mliska@suse.cz>
21940 PR middle-end/68270
21941 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
21942 Use array_at_struct_end_p instead of DECL_CHAIN (field).
21943 (chkp_narrow_bounds_for_field): Likewise.
21944 (chkp_parse_array_and_component_ref): Pass one more argument to
21947 2017-03-07 Richard Biener <rguenther@suse.de>
21949 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
21952 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
21954 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
21955 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
21957 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21960 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
21961 to end of description.
21962 (PARAM_MAX_STORES_TO_MERGE): Likewise.
21964 2017-03-07 Jakub Jelinek <jakub@redhat.com>
21966 PR rtl-optimization/79901
21967 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
21969 (*avx512f_<code><mode>3<mask_name>): ... this.
21970 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
21971 iterator instead of VI8_AVX2_AVX512BW.
21973 PR rtl-optimization/79901
21974 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
21975 min/max expander, expand it using expand_vec_cond_expr.
21978 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
21981 2017-03-06 Jakub Jelinek <jakub@redhat.com>
21984 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
21985 to void * for PCH reasons.
21986 * dwarf2out.c (output_loc_operands, output_die): Cast
21987 v.val_vec.array to unsigned char *.
21989 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
21992 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
21995 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
21997 PR rtl-optimization/79571
21998 * lra-constraints.c (process_alt_operands): Calculate static
21999 reject and subtract it from overall when only addresses will be
22002 2017-03-06 Julia Koval <julia.koval@intel.com>
22005 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
22006 incoming stack boundary to 128 for 64-bit targets.
22008 2017-03-06 Richard Biener <rguenther@suse.de>
22010 PR tree-optimization/79894
22011 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
22012 to NULL after folding it.
22014 2017-03-06 Richard Biener <rguenther@suse.de>
22016 PR tree-optimization/79824
22017 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
22018 check disabling peeling for gaps.
22020 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
22022 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
22023 attributes): Document gettimeofday.
22025 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
22027 * config/s390/s390.c (s390_option_override_internal): Set
22028 PARAM_MIN_VECT_LOOP_BOUND
22030 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
22032 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
22033 * config/s390/s390.md: Likewise.
22035 2017-03-06 Jakub Jelinek <jakub@redhat.com>
22038 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
22039 (<avx2_avx512>_perm<mode>): Rename to ...
22040 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
22042 (<avx512>_perm<mode>_mask): Rename to ...
22043 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
22045 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
22046 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
22047 instead of VI8F_256_512.
22048 (avx512f_perm<mode>): New define_expand.
22049 (avx512f_perm<mode>_mask): Likewise.
22050 (avx512f_perm<mode>_1<mask_name>): New define_insn.
22051 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
22053 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
22055 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
22056 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
22058 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
22060 2017-03-06 Martin Liska <mliska@suse.cz>
22063 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
22064 when having a SSA NAME w/o VAR_DECL assigned to it.
22066 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
22068 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
22069 msa_dpsub_<su>_d): Fix MODE for vec_select.
22071 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
22073 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
22075 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
22077 2017-03-06 Richard Biener <rguenther@suse.de>
22079 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
22080 * plugin.c (register_plugin_info): Likewise.
22081 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
22083 2017-03-05 Jakub Jelinek <jakub@redhat.com>
22085 * config/i386/sse.md (sse_storehps, sse_storelps,
22086 avx_<castmode><avxsizesuffix>_<castmode>,
22087 avx512f_<castmode><avxsizesuffix>_<castmode>,
22088 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
22089 in condition that at least one operand is not a MEM.
22091 2017-03-03 Jakub Jelinek <jakub@redhat.com>
22093 PR middle-end/79805
22094 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
22095 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
22097 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
22098 gimple_call_nothrow_p flag based on whether original builtin can throw.
22099 If it can, emit following stmts on the fallthrough edge.
22100 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
22101 don't create new bb if inserting just debug stmts on the edge, try to
22102 insert them on the fallthru bb or just reset debug stmts.
22104 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
22107 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
22108 restore recog_data (including the operand rtxes inside it) around
22109 the call to get_insn_template.
22111 2017-03-03 Martin Sebor <msebor@redhat.com>
22113 PR tree-optimization/79699
22114 * context.c (context::~context): Free MPFR caches to avoid
22115 a memory leak on program exit.
22117 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22119 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
22120 Use wide_int::ulow () instead of .elt (0).
22122 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
22124 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
22125 (*pushxf): Limit oF constraint to 32bit targets and add oC
22126 constraint for 64bit targets.
22127 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
22128 (*pushdf): Change rmF constraint to rmC.
22130 2017-03-03 Martin Liska <mliska@suse.cz>
22132 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
22133 Remove unused variable.
22135 2017-03-03 Jakub Jelinek <jakub@redhat.com>
22138 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
22139 is a memory operand, increase num_memory.
22140 (ix86_expand_args_builtin): Likewise.
22142 2017-03-03 Jan Hubicka <jh@suse.cz>
22145 * ipa-devirt.c (maybe_record_node): Properly handle
22146 __cxa_pure_virtual visibility.
22148 2017-03-03 Martin Liska <mliska@suse.cz>
22150 PR tree-optimization/79803
22151 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
22153 (pass_loop_prefetch::execute): Disabled optimization if an
22154 assumption about L1 cache size is not met.
22156 2017-03-03 Martin Liska <mliska@suse.cz>
22158 PR rtl-optimization/79574
22159 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
22160 (hash_scan_set): Likewise.
22161 (dump_hash_table): Likewise.
22162 (hoist_code): Likewise.
22164 2017-03-03 Richard Biener <rguenther@suse.de>
22166 * fixed-value.c (fixed_from_string): Restore use of elt (1)
22167 in place of uhigh ().
22168 (fixed_convert_from_real): Likewise.
22170 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
22173 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
22175 2017-03-03 Richard Biener <rguenther@suse.de>
22177 PR middle-end/79818
22178 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
22179 TYPE_OVERFLOW_UNDEFINED check.
22181 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22183 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
22185 (vector_ae_<mode>_p): Likewise.
22186 (vector_nez_<mode>_p): Likewise.
22187 (vector_ne_v2di_p): Likewise.
22188 (vector_ae_v2di_p): Likewise.
22189 (vector_ne_<mode>_p): Likewise.
22190 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
22192 (vsx_tsqrt<mode>2_fe): Likewise.
22194 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
22197 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
22199 2017-03-02 Jakub Jelinek <jakub@redhat.com>
22201 PR rtl-optimization/79780
22202 * cprop.c (one_cprop_pass): When second and further conditional trap
22203 in a single basic block is turned into an unconditional trap, turn it
22204 into a deleted note to avoid RTL verification failures.
22206 2017-03-02 Richard Biener <rguenther@suse.de>
22208 * fold-const.c (const_binop): Use ulow () instead of elt (0).
22210 2017-03-02 Richard Biener <rguenther@suse.de>
22212 PR tree-optimization/79345
22214 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
22215 param and abort the walk, returning -1 if it is hit.
22216 (walk_aliased_vdefs): Take a limit param and pass it on.
22217 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
22218 defaulting to 0 and return a signed int.
22219 * tree-ssa-uninit.c (struct check_defs_data): New struct.
22220 (check_defs): New helper.
22221 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
22222 about uninitialized memory.
22223 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
22224 bogus uninitialized warning.
22225 (fixed_convert_from_real): Likewise.
22227 2017-03-02 Bin Cheng <bin.cheng@arm.com>
22229 PR tree-optimization/66768
22230 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
22231 iv_use if base object can't be determined.
22233 2017-03-02 Jakub Jelinek <jakub@redhat.com>
22235 PR tree-optimization/79345
22236 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
22237 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
22238 (get_pattern_stats): Initialize it.
22239 * genemit.c (gen_expand): Verify match_scratch numbers come after
22240 match_operand/match_dup numbers.
22241 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
22242 match_scratch numbers.
22243 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
22245 * config/s390/s390.md (trunctdsd2): Likewise.
22247 2017-03-02 Richard Biener <rguenther@suse.de>
22249 * wide-int.h (wide_int_storage::operator=): Implement in terms
22252 2017-03-02 Richard Biener <rguenther@suse.de>
22254 PR tree-optimization/79777
22255 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
22256 the to insert expression to sth existing.
22258 2017-03-01 Martin Sebor <msebor@redhat.com>
22260 PR middle-end/79692
22261 * gimple-ssa-sprintf.c
22262 (directive::known_width_and_precision): New function.
22263 (format_integer): Use it.
22264 (get_mpfr_format_length): Consider the full range of precision
22265 when computing %g output with the # flag. Set the likely byte
22266 count to 3 rather than 1 when precision is indeterminate.
22267 (format_floating): Correct the lower bound of precision.
22269 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22271 * doc/invoke.texi: Document default code model for 64-bit Linux.
22273 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22276 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
22277 udiv rather than div since input pattern is unsigned.
22279 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
22281 * config/i386/i386.c (print_reg): Warn for values of
22282 unsupported size in integer register.
22284 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
22287 * config/rs6000/predicates.md (current_file_function_operand): Do
22288 not allow self calls to be local if the function is replaceable.
22290 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
22293 * config/rs6000/altivec.h (vec_ctz and others): Change the
22294 preprocessor macro that controls conditional compilation from
22295 _ARCH_PWR9 to __POWER9_VECTOR__.
22296 (vec_all_ne): Change parameterization of __altivec_scalar_pred
22297 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
22298 control (instead of _ARCH_PWR9 control) so that template
22299 definition uses power9-specific function.
22300 (vec_any_eq): Likewise.
22301 (vec_all_ne): Change macro definition to use a power9-specific
22302 expansion under #ifdef __POWER9_VECTOR__ control (instead of
22303 _ARCH_PWR9 control).
22304 (vec_any_eq) Likewise.
22305 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
22306 expansion for CMPNEF to remove support for xvcmpnesp instruction.
22307 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
22308 support for xvcmpnedp instruction.
22309 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
22310 macro expansion so that Power9 implementation of vec_all_ne does
22311 not use the AltiVec predicate framework.
22312 (VCMPNEH_P): Likewise.
22313 (VCMPNEW_P): Likewise.
22314 (VCMPNED_P): Likewise.
22315 (VCMPNEFP_P): Likewise.
22316 (VCMPNEDP_P): Likewise.
22317 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
22318 implementation of vec_any_eq to not use AltiVec predicate
22320 (VCMPAEH_P): Likewise.
22321 (VCMPAEW_P): Likewise.
22322 (VCMPAED_P): Likewise.
22323 (VCMPAEFP_P): Likewise.
22324 (VCMPAEDP_P): Likewise.
22325 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
22326 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
22327 not use the AltiVec predicate framework.
22328 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
22329 of vec_any_eq to not use AltiVec predicate framework.
22330 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
22331 support for predefined __POWER9_VECTOR__ macro to indicate that
22332 Power9 instruction selection is enabled.
22333 (altivec_overloaded_builtins): Remove extraneous
22334 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
22335 function argument types RS6000_BTI_bool_V16QI and
22336 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
22337 entry for overloaded function argument types RS6000_BTI_bool_V4SI
22338 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
22339 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
22340 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
22341 Power9 for implementations of vec_cmpne. Change the signature for
22342 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
22343 (representing vec_all_ne) to remove the previously described first
22344 argument of type RS6000_BTI_INTSI, as this was an artifact of
22345 reliance on the AltiVec predicate framework, which is no longer
22346 used in the implementation of these functions. Add
22347 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
22348 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
22349 since, unlike the AltiVec predicate framework implementation, we
22350 do not share function descriptors between vec_alle and vec_anyeq.
22351 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
22352 set of modes that receive special treatment even when
22353 TARGET_P9_VECTOR is true. The special treatment emits code that
22354 does not depend on Power9 instructions.
22355 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
22356 define_expand to not rely on AltiVec predicate framework.
22357 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
22359 (vector_ne_v2di_p): Change this define_expand to not rely on
22360 AltiVec predicate framework.
22361 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
22363 (vector_ne_<mode>_p): Change this define_expand to not rely on
22364 AltiVec predicate framework.
22365 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
22367 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
22368 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
22369 define_insn pattern.
22370 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
22371 define_insn pattern because the xvcmpne<VSs>. instruction is not
22373 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
22374 instruction is not supported.
22376 2017-03-01 Jakub Jelinek <jakub@redhat.com>
22378 * config/nvptx/nvptx.c: Include intl.h.
22380 2017-03-01 Martin Jambor <mjambor@suse.cz>
22383 * ipa-prop.h (ipa_bits): Removed field known.
22384 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
22385 to pointers. Adjusted their comments to warn about their sharing.
22386 (ipcp_transformation_summary): Change bits to a vector of pointers.
22387 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
22388 (ipa_get_ipa_bits_for_value): Declare.
22389 * tree-vrp.h (value_range): Mark as GTY((for_user)).
22390 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
22391 (ipa_bits_hash_table): Likewise.
22392 (ipa_vr_ggc_hash_traits): Likewise.
22393 (ipa_vr_hash_table): Likewise.
22394 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
22395 being pointers and vr_known being removed.
22396 (ipa_set_jf_unknown): Likewise.
22397 (ipa_get_ipa_bits_for_value): New function.
22398 (ipa_set_jfunc_bits): Likewise.
22399 (ipa_get_value_range): New overloaded functions.
22400 (ipa_set_jfunc_vr): Likewise.
22401 (ipa_compute_jump_functions_for_edge): Use the above functions to
22402 construct bits and vr parts of jump functions.
22403 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
22404 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
22406 (ipcp_grow_transformations_if_necessary): Also allocate
22407 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
22409 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
22410 them. Fix too long lines.
22411 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
22412 vr_known being removed.
22413 (ipa_read_jump_function): Use new setter functions to construct bits
22414 and vr parts of jump functions or set them to NULL.
22415 (write_ipcp_transformation_info): Adjust for bits being pointers.
22416 (read_ipcp_transformation_info): Likewise.
22417 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
22419 Include gt-ipa-prop.h.
22420 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
22422 (ipcp_store_bits_results): Likewise.
22423 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
22424 Do not write to existing jump functions but use a temporary instead.
22426 2017-03-01 Jakub Jelinek <jakub@redhat.com>
22429 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
22430 attempt to use its first operand as BIT_FIELD_REF base.
22432 2017-03-01 Richard Biener <rguenther@suse.de>
22434 PR middle-end/79721
22435 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
22436 interpolating formula in wrapping arithmetic.
22437 (chrec_apply): Convert chrec_evaluate return value to wanted type.
22439 2017-03-01 Jakub Jelinek <jakub@redhat.com>
22441 PR tree-optimization/79734
22442 * tree-vect-generic.c (expand_vector_condition): Optimize
22443 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
22444 Handle VEC_COND_EXPR where comparison has different inner width from
22445 type's inner width.
22447 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
22449 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
22450 markup, and similar issues. Remove @opindex entries for things
22451 that aren't options. Add missing -mmpy-option entries.
22453 2017-02-28 Jakub Jelinek <jakub@redhat.com>
22455 PR tree-optimization/79737
22456 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
22457 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
22458 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
22459 instead of byte_size. Formatting fix.
22460 (shift_bytes_in_array_right): Formatting fix.
22462 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
22465 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
22466 condition on optimize for the leaf function test.
22468 2017-02-28 Martin Liska <mliska@suse.cz>
22471 * read-rtl-function.c (function_reader::handle_unknown_directive):
22472 Bail out when one uses -flto.
22474 2017-02-28 Martin Liska <mliska@suse.cz>
22476 * common.opt: Replace space with tabular for options of <number>
22478 * config/i386/i386.opt: Show <number> value for
22479 -mlarge-data-threshold.
22480 * opts.c (print_filtered_help): Do not display number in hexadecimal
22483 2017-02-28 Martin Liska <mliska@suse.cz>
22485 * common.opt: Fix --help=option -Q for options which are of
22488 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
22490 * config/i386/i386.c (print_reg): Error out for values
22491 of 8-bit size in invalid integer register.
22493 2017-02-28 Martin Sebor <msebor@redhat.com>
22495 PR tree-optimization/79691
22496 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
22498 2017-02-28 Jakub Jelinek <jakub@redhat.com>
22501 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
22502 gcc_unreachable with output_operand_lossage.
22504 2017-02-28 Richard Biener <rguenther@suse.de>
22506 PR tree-optimization/79740
22507 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
22509 (visit_nary_op): Insert the nary into the hashtable if we
22510 pattern-matched sth.
22511 * tree-ssa-pre.c (eliminate_insert): Robustify.
22513 2017-02-28 Richard Biener <rguenther@suse.de>
22515 PR middle-end/79731
22516 * fold-const.c (decode_field_reference): Reject out-of-bound
22519 2017-02-28 Jakub Jelinek <jakub@redhat.com>
22521 * config/i386/i386.c: Include intl.h.
22522 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
22523 instead of just cond ? "..." : "...".
22524 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
22525 * coverage.c (read_counts_file): Likewise.
22526 * omp-offload.c: Include intl.h.
22527 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
22528 of just cond ? "..." : "...".
22529 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
22530 of just cond ? "..." : "...".
22532 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
22535 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
22537 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
22538 'tune for' CPU name.
22539 * config/arm/arm-cpu-data.h: Regenerated.
22541 2017-02-28 Richard Biener <rguenther@suse.de>
22543 PR tree-optimization/79732
22544 * tree-inline.c (expand_call_inline): Do not shadow var.
22546 2017-02-28 Richard Biener <rguenther@suse.de>
22548 PR tree-optimization/79723
22549 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
22550 address-space properly.
22552 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
22554 * doc/optinfo.texi (Optimization groups): Fix option used for
22556 * doc/invoke.texi (-fopt-info): Document "omp".
22557 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
22558 (OPTGROUP_ALL): Add OPTGROUP_OMP.
22559 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
22560 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
22561 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
22563 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
22565 * dumpfile.c (optgroup_options): Instead of "openmp", associate
22566 OPTGROUP_OMP with "omp".
22568 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
22571 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
22572 for arithmetic shift of unsigned V2DI.
22574 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
22576 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
22577 arc/linux.h headers.
22578 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
22579 (LINK_SPEC): Likewise.
22580 (ARC_TLS_EXTRA_START_SPEC): Likewise.
22581 (EXTRA_SPECS): Likewise.
22582 (STARTFILE_SPEC): Likewise.
22583 (ENDFILE_SPEC): Likewise.
22584 (LIB_SPEC): Likewise.
22585 (TARGET_SDATA_DEFAULT): Likewise.
22586 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
22587 (MULTILIB_DEFAULTS): Likewise.
22588 (DWARF2_UNWIND_INFO): Likewise.
22589 * config/arc/big.h: New file.
22590 * config/arc/elf.h: Likewise.
22591 * config/arc/linux.h: Likewise.
22592 * config/arc/t-uClibc: Remove.
22594 2017-02-27 Bin Cheng <bin.cheng@arm.com>
22596 PR tree-optimization/77536
22597 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
22598 (tree_transform_and_unroll_loop): Use above function to compute the
22599 estimated niter of unrolled loop and use it when scaling profile.
22600 Also use count info rather than frequency if it's non-zero.
22601 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
22602 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
22603 (vect_transform_loop): Call above function.
22605 2017-02-27 Richard Biener <rguenther@suse.de>
22607 PR tree-optimization/45397
22608 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
22609 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
22610 (visit_nary_op): Add pattern matching for CSEing sign-changed
22611 or truncated operations with wider ones.
22613 2017-02-27 Richard Biener <rguenther@suse.de>
22615 PR tree-optimization/79690
22616 * tree-vect-stmts.c (vectorizable_store): Use vector type
22617 built from the DR with address-space.
22619 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
22621 * doc/invoke.texi (Optimize Options): Refine the description
22622 of asan-use-after-return.
22624 2017-02-25 Alan Modra <amodra@gmail.com>
22626 PR rtl-optimization/79584
22627 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
22628 base, not ad->base_term, the reg within base. Remove assertion
22629 that ad->base == ad->base_term. Replace gen_int_mode using
22630 bogus mode with const0_rtx.
22632 2017-02-25 Jakub Jelinek <jakub@redhat.com>
22634 PR middle-end/79396
22635 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
22636 FMA_EXPR like tcc_binary or tcc_unary.
22638 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
22641 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
22643 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
22644 (output_loc_operands): Handle DW_OP_call_ref and
22645 DW_OP_GNU_variable_value.
22646 (struct variable_value_struct): New type.
22647 (struct variable_value_hasher): Likewise.
22648 (variable_value_hash): New variable.
22649 (string_types): Remove.
22650 (copy_loc_descr): New function.
22651 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
22652 (prepend_loc_descr_to_each): New function.
22653 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
22654 instead of add_loc_descr_to_each if the first argument is single
22655 location list and the second has multiple.
22656 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
22657 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
22658 when looking for variable value which doesn't have other location info.
22659 (loc_list_from_tree): Formatting fix.
22660 (gen_array_type_die): Simplify DW_AT_string_length handling.
22661 (adjust_string_types): Remove.
22662 (gen_subprogram_die): Don't call adjust_string_types nor test/set
22663 string_types. Call resolve_variable_values.
22664 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
22665 (resolve_addr_in_expr): Likewise. Add A argument.
22666 (copy_deref_exprloc): Remove deref argument. Adjust for the
22667 original expression being DW_OP_GNU_variable_value with optionally
22668 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
22669 optionally after it.
22670 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
22671 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
22672 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
22673 (variable_value_hasher::hash, variable_value_hasher::equal): New
22675 (resolve_variable_value_in_expr, resolve_variable_value,
22676 resolve_variable_values, note_variable_value_in_expr,
22677 note_variable_value): New functions.
22678 (dwarf2out_early_finish): Call note_variable_value on all toplevel
22681 2017-02-24 Jakub Jelinek <jakub@redhat.com>
22684 * opts.h (handle_generated_option): Add GENERATED_P argument.
22685 * opts-common.c (handle_option): Adjust function comment.
22686 (handle_generated_option): Add GENERATED_P argument, pass it to
22688 (control_warning_option): Pass false to handle_generated_option
22690 * opts.c (maybe_default_option): Pass true to handle_generated_option
22692 * optc-gen.awk: Likewise.
22694 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22696 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
22697 a REG, look at the REG it is a SUBREG of.
22698 (splitter for cmpeqsi_t): Ditto.
22700 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22702 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
22703 the special USEs with the pattern of the insn, not the insn itself.
22705 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
22708 * doc/invoke.texi: Document -mload-store-pairs.
22710 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22711 Sandra Loosemore <sandra@codesourcery.com>
22713 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
22714 argument isn't a CONST_INT.
22715 (nios2_alternate_compare_const): Assert op is a CONST_INT.
22716 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
22717 (nios2_validate_compare): Bypass alternate compare logic if *op2
22718 is not a CONST_INT.
22719 (ldstwm_operation_p): Return false if first_base is not a REG or
22720 if first_offset is not a CONST_INT.
22722 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22724 * config/cris/cris.md: Use correct operand in a define_peephole2.
22726 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22728 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
22730 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22732 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
22733 this_insn if it is an INSN or JUMP_INSN.
22734 (force_offsettable): Look at base, not at addr.
22735 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
22736 on things that aren't necessarily CONST_INTs.
22738 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
22740 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
22741 -mfpmath=sse is the default also for x86-32 targets with SSE2
22742 instruction set when @option{-ffast-math} is enabled
22744 2017-02-24 Jeff Law <law@redhat.com>
22746 PR rtl-optimizatoin/79286
22747 * ira.c (update_equiv_regs): Drop may_trap_p exception to
22750 2017-02-24 Richard Biener <rguenther@suse.de>
22752 PR tree-optimization/79389
22753 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
22756 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
22758 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
22759 function comment to reflect reality.
22760 (loop_exits_before_overflow): Fix typo in function description.
22762 2017-02-24 Richard Biener <rguenther@suse.de>
22764 PR tree-optimization/79389
22765 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
22766 properly that a threading opportunity exists. Detect conditional
22767 copy/constant propagation opportunities.
22769 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
22771 * config/visium/visium.md (type): Add trap.
22772 (b): New mode attribute.
22773 (*btst): Rename into...
22774 (*btst<mode>): ...this and adjust.
22775 (*cbranchsi4_btst_insn): Rename into...
22776 (*cbranch<mode>4_btst_insn): ...this and adjust.
22777 (trap): New define_insn.
22779 2017-02-23 Jakub Jelinek <jakub@redhat.com>
22781 PR tree-optimization/79389
22782 * ifcvt.c (struct noce_if_info): Add rev_cond field.
22783 (noce_reversed_cond_code): New function.
22784 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
22785 reversed_comparison_code. Formatting fix.
22786 (noce_try_store_flag): Test rev_cond != NULL in addition to
22787 reversed_comparison_code.
22788 (noce_try_store_flag_constants): Likewise.
22789 (noce_try_store_flag_mask): Likewise.
22790 (noce_try_addcc): Use rev_cond if non-NULL instead of
22791 reversed_comparison_code.
22792 (noce_try_cmove_arith): Likewise. Formatting fixes.
22793 (noce_try_minmax, noce_try_abs): Clear rev_cond.
22794 (noce_find_if_block): Initialize rev_cond.
22795 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
22796 instead of false as last argument never attempt to reverse it
22799 2017-02-23 Bin Cheng <bin.cheng@arm.com>
22801 PR tree-optimization/79663
22802 * tree-predcom.c (combine_chains): Process refs in reverse order
22803 only for ZERO length chains, and add explaining comment.
22805 2017-02-23 Jeff Law <law@redhat.com>
22807 PR tree-optimization/79578
22808 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
22809 in call to operand_equal_p.
22811 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
22814 * config/i386/cpuid.h: Fix another undefined behavior.
22816 2017-02-23 Richard Biener <rguenther@suse.de>
22818 PR tree-optimization/79683
22819 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
22820 vector types for data-refs.
22822 2017-02-23 Martin Liska <mliska@suse.cz>
22824 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
22826 2017-02-23 Jakub Jelinek <jakub@redhat.com>
22828 PR middle-end/79665
22829 * internal-fn.c (get_range_pos_neg): Moved to ...
22830 * tree.c (get_range_pos_neg): ... here. No longer static.
22831 * tree.h (get_range_pos_neg): New prototype.
22832 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
22833 are known to be in between 0 and signed maximum inclusive, try to
22834 expand both unsigned and signed divmod and use the cheaper one from
22837 2017-02-22 Jeff Law <law@redhat.com>
22839 PR tree-optimization/79578
22840 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
22841 to compare base operands.
22843 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
22846 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
22847 gpc_reg_operand instead of fpr_reg_operand.
22849 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
22851 * config/mips/mips.c (mips_return_in_memory): Force FP
22852 vector types to be returned in memory for o32 ABI.
22854 2017-02-22 Jakub Jelinek <jakub@redhat.com>
22856 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
22857 instead of DW_TAG_member for static data member declarations and don't
22858 set no_linkage_name for static inline data members.
22859 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
22862 2017-02-22 Martin Liska <mliska@suse.cz>
22864 * doc/invoke.texi: Replace inequality signs with square brackets
22867 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22869 PR tree-optimization/68644
22870 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
22872 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
22875 * lra-constraints.c (simplify_operand_subreg): Handle
22876 WORD_REGISTER_OPERATIONS targets.
22878 2017-02-22 Jakub Jelinek <jakub@redhat.com>
22881 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
22882 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
22883 elimination by swapping fld*.
22885 2017-02-22 Richard Biener <rguenther@suse.de>
22887 PR tree-optimization/79673
22888 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
22889 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
22890 irrelevant address-space qualifiers and avoiding a
22891 ADDR_SPACE_CONVERT_EXPR from fold_convert.
22893 2017-02-22 Richard Biener <rguenther@suse.de>
22895 PR tree-optimization/79666
22896 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
22897 to not symbolically negate if that may introduce undefined
22900 2017-02-22 Martin Liska <mliska@suse.cz>
22903 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
22904 * data-streamer-out.c (streamer_write_gcov_count_stream):
22906 * value-prof.c (stream_out_histogram_value): Make assert more
22907 precise based on type of counter.
22909 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
22912 * config/i386/i386.md (standard_x87sse_constant_load splitter):
22913 Use nonimmediate_operand instead of memory_operand for operand 1.
22914 (float-extend standard_x87sse_constant_load splitter): Ditto.
22916 2017-02-21 Jeff Law <law@redhat.com>
22918 PR tree-optimization/79621
22919 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
22920 blocks with edges to themselves.
22922 2017-02-21 Jakub Jelinek <jakub@redhat.com>
22925 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
22926 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
22927 Use gimple_call_builtin_p.
22930 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
22931 on temporarily removed DEBUG_INSNs.
22933 PR tree-optimization/79649
22934 * tree-loop-distribution.c (classify_partition): Give up on
22935 non-generic address space loads/stores.
22937 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
22939 * doc/loop.texi (Loop manipulation): Remove nonexistent
22940 tree_ssa_loop_version from the documentation.
22941 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
22943 2017-02-21 Jakub Jelinek <jakub@redhat.com>
22946 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
22947 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
22948 * config/rs6000/rs6000.c: Include except.h.
22949 (rs6000_expand_split_stack_prologue): Call
22950 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
22952 2017-02-21 Martin Jambor <mjambor@suse.cz>
22955 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
22956 have been analyzed.
22958 2017-02-21 Martin Jambor <mjambor@suse.cz>
22960 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
22961 for backward compatibility only.
22962 * doc/invoke.texi (Option Summary): Remove all references to
22963 -fipa-cp-alignment.
22965 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
22969 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
22971 * lra-constraints.c (curr_insn_transform): Handle
22972 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
22974 2017-02-21 Martin Liska <mliska@suse.cz>
22976 * config/i386/i386.opt: Replace -masm-dialect with -masm.
22978 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
22980 PR translation/79638
22981 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
22983 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
22986 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
22987 (arm_function_ok_for_sibcall): Return false for an indirect call by
22988 descriptor if all the argument registers are used.
22989 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
22990 alignment of the function.
22992 2017-02-21 Jakub Jelinek <jakub@redhat.com>
22994 PR tree-optimization/61441
22995 * simplify-rtx.c (simplify_const_unary_operation): For
22996 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
22997 the sNaN unmodified.
22999 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
23001 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
23002 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
23003 instead of SYSTEM_HEADER_DIR.
23005 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
23006 Martin Liška <mliska@suse.cz>
23008 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
23009 Fix typos and grammar, use active voice, and clarify.
23011 2017-02-20 Marek Polacek <polacek@redhat.com>
23013 PR middle-end/79537
23014 * gimplify.c (gimplify_expr): Handle unused *&&L;.
23017 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
23019 2017-02-20 Jakub Jelinek <jakub@redhat.com>
23022 * config/i386/i386.c (ix86_expand_builtin): Handle
23023 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
23024 ix86_builtins_isa[fcode].isa as a requirement of those
23025 flags and any other flag in the bitmask.
23026 (ix86_init_mmx_sse_builtins): Use 0 instead of
23027 ~OPTION_MASK_ISA_64BIT as mask.
23028 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
23029 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
23030 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
23031 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
23033 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
23036 * lra-constraints.c (split_reg): Check requested split mode
23037 is supported by the register.
23039 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
23041 * lra-constraints.c (simplify_operand_subreg): Remove early
23044 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
23047 * lra-constraints.c (curr_insn_transform): Tighten condition
23048 for converting SUBREG reloads from OP_OUT to OP_INOUT.
23050 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
23053 * lra-constraints.c (curr_insn_transform): Handle
23054 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
23056 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
23059 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
23061 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
23063 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
23066 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
23069 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23071 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
23074 2017-02-18 Jakub Jelinek <jakub@redhat.com>
23076 * final.c (last_columnnum, override_columnnum): New variables.
23077 (final_start_function): Set last_columnnum, pass it to begin_prologue
23078 hook and pass 0 to dwarf2out_begin_prologue.
23079 (final_scan_insn): Update override_columnnum. Pass last_columnnum
23080 to source_line debug hook.
23081 (notice_source_line): Compute last_columnnum and for debug_column_info
23082 return true on column changes.
23083 * debug.h (struct gcc_debug_hooks): Add column argument to
23084 source_line and begin_prologue hooks.
23085 (debug_nothing_int_charstar_int_bool): Remove prototype.
23086 (debug_nothing_int_int_charstar,
23087 debug_nothing_int_int_charstar_int_bool): New prototypes.
23088 (dwarf2out_begin_prologue): Add column argument.
23089 * debug.c (do_nothing_debug_hooks): Adjust source_line and
23090 begin_prologue hooks.
23091 (debug_nothing_int_charstar_int_bool): Remove.
23092 (debug_nothing_int_int_charstar,
23093 debug_nothing_int_int_charstar_int_bool): New functions.
23094 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
23095 through to dwarf2out_source_line.
23096 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
23097 (dwarf2out_source_line): Add column argument, emit it if requested.
23098 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
23100 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
23101 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
23102 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
23103 through to dwarf2out_begin_prologue.
23104 (vmsdbgout_source_line): Add column argument, pass it through to
23105 dwarf2out_source_line.
23106 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
23107 dbxout_source_line caller.
23108 (dbxout_source_line): Add column argument.
23110 * common.opt (gno-column-info, gcolumn-info): New options.
23111 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
23112 (check_die): Also test for multiple DW_AT_decl_column attributes.
23113 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
23114 DW_AT_decl_column if requested.
23115 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
23117 (gen_variable_die): Likewise.
23118 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
23119 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
23122 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
23123 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
23124 (ix86_handle_option): Handle OPT_m3dnowa.
23125 * doc/invoke.texi (-m3dnowa): Document.
23126 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
23127 -m3dnowa instead of -m3dnow -march=athlon.
23130 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
23131 instead of gcc_assert for K, r and R code checks. Formatting fixes.
23133 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23136 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
23137 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
23138 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
23139 generator for vsx_xxpermdi_<mode>_be.
23140 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
23141 force big-endian semantics.
23142 (vsx_xxpermdi_<mode>_be): New define_expand with same
23143 implementation as previous version of vsx_xxpermdi_<mode>.
23145 2017-02-17 Jakub Jelinek <jakub@redhat.com>
23147 PR tree-optimization/79327
23148 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
23149 variable, its initialization and use.
23151 2017-02-17 Julia Koval <julia.koval@intel.com>
23153 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
23154 (OPTION_MASK_ISA_PKU_UNSET): New.
23155 (ix86_handle_option): Handle -mrdpid.
23156 * config/i386/cpuid.h (bit_RDPID): New.
23157 * config/i386/driver-i386.c (host_detect_local_cpu):
23158 Detect RDPID feature.
23159 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
23160 * config/i386/i386-c.c (ix86_target_macros_internal):
23162 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
23163 (ix86_valid_target_attribute_inner_p): Add "rdpid".
23164 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
23165 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
23166 * config/i386/i386.md (define_insn "rdpid"): New.
23167 * config/i386/i386.opt Add -mrdpid.
23168 * config/i386/immintrin.h (_rdpid_u32): New.
23170 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
23172 PR rtl-optimization/79541
23173 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
23174 instead of transforming it into USE.
23176 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
23178 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
23179 If HONOR_SNANS (SFmode) force the input to a register.
23180 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
23181 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
23182 an frsp or similar insn.
23184 2017-02-17 Martin Liska <mliska@suse.cz>
23186 PR rtl-optimization/79577
23187 * params.def (selsched-max-sched-times): Increase minimum to 1.
23189 2017-02-17 Martin Liska <mliska@suse.cz>
23191 PR rtl-optimization/79574
23192 * gcse.c (want_to_gcse_p): Prevent integer overflow.
23194 2017-02-17 Martin Liska <mliska@suse.cz>
23196 PR tree-optimization/79529
23197 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
23198 ssa_defined_default_def_p to handle cases which are implicitly
23200 * tree-ssa.c (ssa_defined_default_def_p): New function.
23201 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
23202 which are implicitly defined.
23203 * tree-ssa.h (ssa_defined_default_def_p): Declare.
23205 2017-02-17 Richard Biener <rguenther@suse.de>
23207 PR middle-end/79576
23208 * params.def (max-ssa-name-query-depth): Limit to 10.
23210 2017-02-17 Richard Biener <rguenther@suse.de>
23212 PR tree-optimization/79552
23213 * tree-ssa-structalias.c (visit_loadstore): Properly verify
23216 2017-02-17 Richard Biener <rguenther@suse.de>
23219 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
23221 2017-02-17 Marek Polacek <polacek@redhat.com>
23223 PR middle-end/79536
23224 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
23225 (fold_negate_expr): New wrapper.
23227 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
23229 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
23230 Correct terminology and de-emphasize pre-standard behavior.
23232 2017-02-16 Alan Modra <amodra@gmail.com>
23234 PR rtl-optimization/79286
23235 * ira.c (def_dominates_uses): New function.
23236 (update_equiv_regs): Don't create an equivalence for insns that
23237 may trap where the register def does not dominate the use.
23239 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
23241 PR rtl-optimization/78127
23242 * lra.c (lra): Call lra_eliminate before finish the loop after
23245 2017-02-16 Richard Biener <rguenther@suse.de>
23247 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
23249 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
23250 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
23251 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
23252 (isl_val_int_from_wi): New function.
23253 (extract_affine_gmp): Rename to ...
23254 (extract_affine_wi): ... this, take a widest_int.
23255 (extract_affine_int): Just wrap extract_affine_wi.
23256 (add_param_constraints): Use isl_val_int_from_wi.
23257 (add_loop_constraints): Likewise, and extract_affine_wi.
23259 2017-02-15 Jeff Law <law@redhat.com>
23261 PR middle-end/79521
23262 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
23263 ira_init_register_move_cost_if_necessary.
23265 2017-02-15 Martin Sebor <msebor@redhat.com>
23267 PR middle-end/32003
23268 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
23269 removed in a prior commit.
23271 2017-02-15 Bin Cheng <bin.cheng@arm.com>
23273 PR tree-optimization/79347
23274 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
23275 counters during peeling.
23277 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
23279 * Makefile.in (site.exp): Remove "set ISLVER".
23281 2017-02-15 Jakub Jelinek <jakub@redhat.com>
23284 * real.c (real_from_integer): Call real_convert even for decimal.
23286 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
23289 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
23291 2017-02-14 Andrew Pinski <apinski@cavium.com>
23293 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
23294 cores and change the partno/implementer to be correct.
23295 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
23296 the 'B" as the implementer.
23297 * config/aarch64/aarch64-tune.md: Regenerate.
23299 2017-02-14 Carl Love <cel@us.ibm.com>
23301 * config/rs6000/rs6000.c: Add case statement entry to make the
23302 xvcvuxdsp built-in argument unsigned.
23303 * config/rs6000/vsx.md: Fix the source and return operand types so they
23304 match the instruction definitions from the ISA document. Fix typo
23305 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
23308 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
23311 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
23312 member early_clobber_alts.
23313 * lra-lives.c (reg_early_clobber_p): New.
23314 (process_bb_lives): Use it.
23315 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
23316 (debug_operand_data): Initialize early_clobber_alts.
23317 (setup_operand_alternative): Set up early_clobber_alts.
23318 (collect_non_operand_hard_regs): Ditto. Pass early clobber
23319 alternatives to new_insn_reg.
23320 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
23322 (lra_update_insn_regno_info): Pass the new arg.
23324 2017-02-14 Jakub Jelinek <jakub@redhat.com>
23326 PR middle-end/79505
23327 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
23328 (new_oacc_loop_raw): Don't clear already cleared fields.
23331 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
23332 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
23333 _mm512_prefetch_i64gather_ps): New inline functions and macros.
23335 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
23338 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
23340 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
23343 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
23344 the extra instruction to the right place to store 128-bit constant
23347 2017-02-14 Martin Sebor <msebor@redhat.com>
23349 PR middle-end/79448
23350 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
23351 warning for strings of unknown length.
23353 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
23355 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
23357 2017-02-14 Jeff Law <law@redhat.com>
23360 * ira-costs.c (scan_one_insn): Initialize register move costs
23361 for pseudos seen in USE/CLOBBER insns.
23363 PR tree-optimization/79095
23364 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
23365 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
23366 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
23367 if the operands are known to be not equal, then the resulting range
23369 (intersect_ranges): If the new range is ~[0,0] and the old range is
23370 wide, then prefer ~[0,0].
23371 * tree-vrp.c (overflow_comparison_p_1): New function.
23372 (overflow_comparison_p): New function.
23373 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
23374 if NAME is used in an overflow test.
23375 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
23376 overflow check that can be expressed as an equality test, then adjust
23377 ops to be that equality test.
23379 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23381 * config/s390/s390-builtin-types.def: Remove flags argument.
23382 * config/s390/s390.c (s390_init_builtins): Likewise.
23384 2017-02-14 Martin Liska <mliska@suse.cz>
23386 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
23387 vector. Fix trailing white spaces.
23389 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
23391 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
23394 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23396 PR rtl-optimization/68664
23397 * config/arm/arm.c (arm_sched_can_speculate_insn):
23398 New function. Declare prototype.
23399 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
23401 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23403 PR rtl-optimization/68664
23404 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
23406 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
23408 2017-02-14 Amit Pawar <amit.pawar@amd.com>
23410 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
23411 max skip bytes for function, loop and jump.
23413 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23415 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
23416 ABS_EXPR for gimple dump.
23418 2017-02-14 Jakub Jelinek <jakub@redhat.com>
23421 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
23423 PR tree-optimization/79408
23424 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
23425 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
23426 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
23427 also if rhs1 is INTEGER_CST.
23429 2017-02-14 Richard Biener <rguenther@suse.de>
23431 PR middle-end/79432
23432 * tree-into-ssa.c (insert_phi_nodes): When the function can
23433 have abnormal edges rewrite SSA names with broken use-def
23434 dominance out of SSA and register them for PHI insertion.
23436 2017-02-13 Martin Sebor <msebor@redhat.com>
23438 PR middle-end/79496
23439 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
23440 clearing info.nowrite flag when snprintf size argument is a range.
23442 2017-02-13 Jakub Jelinek <jakub@redhat.com>
23444 * cprop.c (cprop_jump): Add missing space in string literal.
23445 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
23446 (get_constraint_for_component_ref): Likewise.
23447 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
23448 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
23449 * lra-constraints.c (process_alt_operands): Likewise.
23450 * ipa-inline.c (inline_small_functions): Likewise.
23451 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
23452 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
23453 * trans-mem.c (diagnose_tm_1_op): Likewise.
23454 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
23455 (grid_parallel_clauses_gridifiable): Likewise.
23457 * config/nvptx/mkoffload.c (process): Add space in between
23460 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
23461 "MOD4_SSE_REGS" and "ALL_REGS".
23463 * spellcheck.c (test_data): Add , in between "foo" and "food".
23465 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23468 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
23469 boundary crossing check and subsequent code generation agree.
23471 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23473 * config/aarch64/aarch64.c (has_memory_op): Delete.
23474 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
23477 2017-02-13 Jakub Jelinek <jakub@redhat.com>
23479 PR rtl-optimization/79388
23480 PR rtl-optimization/79450
23481 * combine.c (distribute_notes): When removing TEM_INSN for which
23482 corresponding dest has last value recorded, invalidate that last
23485 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23487 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
23488 of explicit '@'. Add missing assembly comment marker on branch costs
23491 2017-02-13 Nathan Sidwell <nathan@acm.org>
23493 * gengtype-lex.l (<in_struct>): Add '/'.
23495 2017-02-13 Martin Liska <mliska@suse.cz>
23498 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
23500 2017-02-13 Richard Biener <rguenther@suse.de>
23502 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
23504 * configure: Re-generate.
23505 * config.in: Likewise.
23506 * graphite-dependences.c: Simplify as if
23507 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
23508 * graphite-isl-ast-to-gimple.c: Likewise.
23509 * graphite-optimize-isl.c: Likewise.
23510 * graphite-poly.c: Likewise.
23511 * graphite-sese-to-poly.c: Likewise.
23512 * graphite.h: Likewise.
23513 * toplev.c: Include isl/version.h and use isl_version () for
23514 printing the ISL version.
23515 * doc/install.texi: Update ISL requirement.
23517 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
23519 * doc/standards.texi (Standards): Update reference to
23522 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
23524 * doc/extend.texi (Named Address Spaces): sourceware.org now
23526 * doc/install.texi (Binaries): Ditto.
23529 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
23531 * doc/cpp.texi: Replace "stringify"/"stringification" with C
23532 standard terminology "stringize"/"stringizing" throughout.
23533 * doc/cppinternals.texi: Likewise.
23535 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
23537 * doc/extend.texi: Fix some spelling mistakes and typos.
23538 * doc/invoke.texi: Likewise.
23540 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
23543 * params.def (inline-min-speedup) Change from 10 to 8.
23545 2017-02-11 Jakub Jelinek <jakub@redhat.com>
23547 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
23550 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
23553 * ipa-inline-analysis.c (get_minimal_bb): New function.
23554 (record_modified): Use it.
23555 (remap_edge_change_prob): Handle also ancestor functions.
23557 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
23559 * doc/contrib.texi (Contributors): Remove broken link into
23560 the Mauve CVS repository.
23562 2017-02-11 Jakub Jelinek <jakub@redhat.com>
23564 PR middle-end/79454
23565 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
23566 result computation whenever lhs doesn't have vector mode, not
23567 just when it has BLKmode.
23569 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
23571 * doc/makefile.texi (profiledbootstrap): Refer to the
23572 installation instructions only in textual form.
23574 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23577 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
23579 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
23581 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
23582 (Specific): Update mingw-w64 reference.
23584 (Specific): Remove broken link to Renesas RX processor.
23586 2017-02-10 Richard Biener <rguenther@suse.de>
23588 * toplev.c (process_options): Do not mention obsolete graphite
23589 options when printing sorry message about missing graphite support.
23590 Mention -floop-nest-optimize.
23592 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
23594 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
23595 (vtst_p16): Likewise.
23596 (vtstq_p8): Likewise.
23597 (vtstq_p16): Likewise.
23599 (vtstq_p64): Likewise.
23600 * config/arm/arm_neon.h (vgetq_lane_p64): New.
23601 (vset_lane_p64): New.
23602 (vsetq_lane_p64): New.
23604 2017-02-10 Jakub Jelinek <jakub@redhat.com>
23606 PR tree-optimization/79411
23607 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
23608 stmt operands are SSA_NAMEs used in abnormal phis.
23609 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
23612 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
23615 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
23618 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
23620 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
23622 2017-02-09 Jakub Jelinek <jakub@redhat.com>
23624 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
23628 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
23629 not arbitrary TREE_CONSTANT.
23632 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
23633 "omp declare target link" attribute unless is_global_var.
23634 * omp-offload.c (find_link_var_op): Likewise.
23636 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
23637 Chung-Lin Tang <cltang@codesourcery.com>
23639 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
23641 (gimplify_adjust_omp_clauses): Don't delete TILE.
23642 (gimplify_omp_for): Deal with TILE.
23643 * internal-fn.c (expand_GOACC_TILE): New function.
23644 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
23646 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
23647 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
23649 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
23650 avoid DIV for outermost collapse var.
23651 (expand_oacc_for): Insert tile element loop as needed. Adjust.
23652 Remove out of date comments, fix whitespace.
23653 * omp-general.c (omp_extract_for_data): Deal with tiling.
23654 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
23655 adjust OLF_DIM_BASE value.
23656 (struct omp_for_data): Add tiling field.
23657 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
23658 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
23659 for auto loops. Remove default auto determining, moved to
23660 oacc_loop_fixed_partitions.
23661 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
23662 stmts, add e_mask field.
23663 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
23664 (oacc_thread_numbers): Use oacc_dim_call.
23665 (oacc_xform_tile): New.
23666 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
23667 (finish_oacc_loop): Adjust for ifns vector.
23668 (oacc_loop_discover_walk): Append loop abstraction sites to list,
23669 add case for GOACC_TILE fns.
23670 (oacc_loop_xform_loop): Delete.
23671 (oacc_loop_process): Iterate over call list directly, and add
23672 handling for GOACC_TILE fns.
23673 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
23675 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
23676 vector partitioning to outer loops. Assign 2 partitions to loops
23677 when available. Add TILE handling.
23678 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
23679 (execite_oacc_device_lower): Process GOACC_TILE fns,
23680 ignore unknown specs.
23681 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
23682 * tree.c (omp_clause_num_ops): Adjust TILE ops.
23683 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
23685 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
23687 * configure.ac (ACX_BUGURL): Update.
23688 * configure: Regenerate.
23690 2017-02-09 Richard Biener <rguenther@suse.de>
23692 PR tree-optimization/69823
23693 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
23694 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
23696 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
23698 * config/arc/arc-c.def: Add __NPS400__ definition.
23699 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
23700 (TARGET_NPS400): Define.
23702 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
23704 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
23706 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
23707 pointer, arch_info.
23708 (arc_cpu_types): Fill the arch_info field with a pointer into the
23709 arc_arch_types table.
23710 (arc_selected_cpu): Declare.
23711 * config/arc/arc.c (arc_selected_cpu): Make global.
23712 (arc_selected_arch): Delete.
23713 (arc_base_cpu): Delete.
23714 (arc_override_options): Remove references to deleted variables,
23715 update access to arch information.
23716 (ARC_OPT): Update access to arch information.
23717 (ARC_OPTX): Likewise.
23718 * config/arc/arc.h (arc_base_cpu): Remove declaration.
23719 (TARGET_ARC600): Update access to arch information.
23720 (TARGET_ARC601): Likewise.
23721 (TARGET_ARC700): Likewise.
23722 (TARGET_EM): Likewise.
23723 (TARGET_HS): Likewise.
23724 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
23727 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
23730 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
23731 condition/operands for integer GE/LE/GEU/LEU operations.
23733 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
23735 PR translation/79397
23736 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
23739 2017-02-08 Martin Jambor <mjambor@suse.cz>
23742 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
23743 whether allocation happened.
23744 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
23745 nothing was allocated.
23747 2017-02-08 Jakub Jelinek <jakub@redhat.com>
23749 PR tree-optimization/79408
23750 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
23751 constant, but SSA_NAME with a known integer range, use the minimum
23752 of that range instead of op1 to determine if modulo can be replaced
23753 with its first operand.
23755 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23757 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
23759 2017-02-08 Richard Biener <rguenther@suse.de>
23761 PR tree-optimization/71824
23762 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
23763 Check all loops contained in the merged region.
23765 2017-02-07 Andrew Pinski <apinski@cavium.com>
23767 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
23769 2017-02-07 Andrew Pinski <apinski@cavium.com>
23771 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
23772 (thunderxt88): Likewise.
23773 (thunderxt81): Disable LSE and change v8.1 to v8.
23774 (thunderxt83): Likewise.
23776 2017-02-07 Jakub Jelinek <jakub@redhat.com>
23777 Richard Biener <rguenther@suse.de>
23779 PR middle-end/79399
23780 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
23781 type from int to size_t.
23782 * ira-costs.c (struct_costs_size): Change type from int to size_t.
23784 2017-02-07 Jakub Jelinek <jakub@redhat.com>
23786 PR rtl-optimization/79386
23787 * cprop.c (bypass_conditional_jumps): Initialize
23788 bypass_last_basic_block already before splitting bbs after
23789 unconditional traps...
23790 (bypass_conditional_jumps): ... rather than here.
23793 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
23794 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
23795 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
23796 fix -masm=intel patterns.
23798 2017-02-07 Richard Biener <rguenther@suse.de>
23800 PR tree-optimization/79256
23801 PR middle-end/79278
23802 * builtins.c (get_object_alignment_2): Use min_align_of_type
23803 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
23804 and ADJUST_FIELD_ALIGN.
23806 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
23808 * doc/tm.texi: Regenerate.
23809 * stor-layout.c (layout_decl): Adjust.
23810 (update_alignment_for_field): Likewise.
23811 (place_field): Likewise.
23812 (min_align_of_type): Likewise.
23813 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
23814 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
23815 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
23816 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
23817 * config/frv/frv.c (frv_adjust_field_align): Likewise.
23818 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
23819 * config/i386/i386.c (x86_field_alignment): Likewise.
23820 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
23821 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
23822 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
23823 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
23824 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
23825 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
23829 2017-01-30 Richard Biener <rguenther@suse.de>
23831 PR tree-optimization/79256
23832 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
23833 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
23836 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
23838 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
23839 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
23840 builtins to SImode and emit a zero-extend, if necessary.
23842 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
23844 * docs/invoke.texi (RISC-V Options): Alphabetize.
23846 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
23848 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
23851 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
23853 * config/riscv/riscv.c: New file.
23854 * common/config/riscv/riscv-common.c: Likewise.
23855 * config.gcc: Likewise.
23856 * config/riscv/constraints.md: Likewise.
23857 * config/riscv/elf.h: Likewise.
23858 * config/riscv/generic.md: Likewise.
23859 * config/riscv/linux.h: Likewise.
23860 * config/riscv/multilib-generator: Likewise.
23861 * config/riscv/peephole.md: Likewise.
23862 * config/riscv/pic.md: Likewise.
23863 * config/riscv/predicates.md: Likewise.
23864 * config/riscv/riscv-builtins.c: Likewise.
23865 * config/riscv/riscv-c.c: Likewise.
23866 * config/riscv/riscv-ftypes.def: Likewise.
23867 * config/riscv/riscv-modes.def: Likewise.
23868 * config/riscv/riscv-opts.h: Likewise.
23869 * config/riscv/riscv-protos.h: Likewise.
23870 * config/riscv/riscv.h: Likewise.
23871 * config/riscv/riscv.md: Likewise.
23872 * config/riscv/riscv.opt: Likewise.
23873 * config/riscv/sync.md: Likewise.
23874 * config/riscv/t-elf-multilib: Likewise.
23875 * config/riscv/t-linux: Likewise.
23876 * config/riscv/t-linux-multilib: Likewise.
23877 * config/riscv/t-riscv: Likewise.
23878 * configure.ac: Likewise.
23879 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
23880 Waterman as RISC-V maintainers.
23881 * doc/install.texi: Add RISC-V entries.
23882 * doc/invoke.texi: Add RISC-V options section.
23883 * doc/md.texi: Add RISC-V constraints section.
23884 * configure: Regenerated.
23886 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
23889 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
23890 false values to be constant vectors with all 0 or all 1 bits set.
23891 (vcondu<mode><mode>): Likewise.
23892 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
23894 (fpmask_comparison_operator): Update comment.
23895 (vecint_comparison_operator): New predicate.
23896 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
23897 vector conditionals when the true and false values are constant
23898 vectors with all 0 bits or all 1 bits set.
23900 2017-02-06 Martin Sebor <msebor@redhat.com>
23902 PR tree-optimization/79376
23903 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
23905 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
23907 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
23908 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
23909 to simplify split condition.
23911 2017-02-06 Jakub Jelinek <jakub@redhat.com>
23913 * omp-expand.c (oxpand_omp_atomic_fetch_op,
23914 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
23917 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
23919 PR rtl-optimization/68664
23920 * target.def (can_speculate_insn): New hook.
23921 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
23922 * doc/tm.texi: Regenerate.
23923 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
23924 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
23925 (rs6000_sched_can_speculate_insn): New function.
23927 2017-02-06 Jakub Jelinek <jakub@redhat.com>
23929 PR tree-optimization/79284
23930 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
23931 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
23932 vectorizable_mask_load_store, vectorizable_operation,
23933 vect_is_simple_cond, get_same_sized_vectype): Use it instead
23934 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
23935 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
23936 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
23937 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
23938 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
23939 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
23940 is_gimple_assign (stmt). Replace another such test with
23941 is_gimple_assign (stmt).
23943 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
23946 * config/avr/avr.c (rtl-iter.h): Include it.
23947 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
23948 (avr_legitimate_combined_insn): ...and implementation.
23950 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
23952 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
23953 * config/s390/s390.c (s390_const_operand_ok)
23954 (s390_canonicalize_comparison, s390_extract_part)
23955 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
23956 (s390_contiguous_bitmask_p, s390_rtx_costs)
23957 (legitimize_pic_address): Likewise.
23958 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
23959 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
23960 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
23961 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
23962 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
23964 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
23966 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
23967 REGNO($0) == REGNO($1).
23969 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23971 * config/s390/linux.h(SIZE_TYPE): Add comment.
23973 2017-02-06 Julian Brown <julian@codesourcery.com>
23974 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23975 Virendra Pathak <virendra.pathak@broadcom.com>
23977 * config/aarch64/aarch64-cores.def: Change the scheduler
23979 * config/aarch64/aarch64.md: Include thunderx2t99.md.
23980 * config/aarch64/thunderx2t99.md: New file.
23982 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
23984 * doc/standards.texi (Go Language): Update link to language
23987 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
23989 * tree-eh.c (lower_resx): Sanitize profile.
23990 (cleanup_empty_eh_move_lp): Likewise.
23992 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
23995 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
23997 * cfgloopmanip.h (loop_version): Update prototype.
23998 * modulo-sched.c (sms_schedule): Update call of loop_version.
23999 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
24000 * tree-parloops.c (gen_parallel_loop): Likewise.
24001 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
24002 * tree-ssa-loop-split.c (split_loop): Likewise.
24003 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
24004 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
24006 2017-02-05 Martin Liska <mliska@suse.cz>
24009 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
24011 (print_operand_address): Initialize a struct to zero.
24013 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
24015 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
24016 garbage collector only in textual form.
24018 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
24020 * doc/extend.texi (x86 specific memory model extensions for
24021 transactional memory): Simplify a phrase.
24023 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
24026 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
24027 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
24028 (atomic_storedi_1): Likewise.
24030 2017-02-04 Jakub Jelinek <jakub@redhat.com>
24032 PR tree-optimization/79338
24033 * tree-parloops.c (gather_scalar_reductions): Don't call
24034 vect_analyze_loop_form for loop->inner before destroying loop's
24037 2017-02-03 Martin Sebor <msebor@redhat.com>
24039 PR tree-optimization/79327
24040 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
24041 when precision has resulted in leading zeros.
24042 (format_integer): Adjust the likely counter to assume an unknown
24043 argument that may be zero is non-zero.
24045 2017-02-03 Jason Merrill <jason@redhat.com>
24048 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
24049 avoid copying non-taken branch.
24051 2017-02-03 Jakub Jelinek <jakub@redhat.com>
24053 PR tree-optimization/79340
24054 * tree-vect-loop.c (vectorizable_reduction): Release
24055 vec_defs elements after safe_splicing them into other vectors.
24058 PR tree-optimization/79327
24059 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
24060 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
24062 (format_integer): Use wide_int_to_tree instead of build_int_cst
24063 + to_?hwi. If argmin is NULL, just set argmin and argmax to
24064 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
24065 of shortest and longest sequence.
24067 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
24069 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
24070 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
24072 2017-02-03 Walter Lee <walt@tilera.com>
24075 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
24076 after initial stackframe link reg save.
24077 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
24079 2017-02-03 Jakub Jelinek <jakub@redhat.com>
24082 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
24083 wu for stxssp alternative.
24085 2017-02-03 Martin Sebor <msebor@redhat.com>
24087 PR tree-optimization/79352
24088 * gimple-fold.c (get_range_strlen): Add argument.
24089 (get_range_strlen): Change return type to bool.
24090 (get_maxval_strlen): Pass in a dummy argument.
24091 * gimple-fold.h (get_range_strlen): Change return type to bool.
24092 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
24093 * tree.h (array_at_struct_end_p): Add argument.
24094 * tree.c (array_at_struct_end_p): Handle it.
24096 2017-02-03 Martin Liska <mliska@suse.cz>
24099 * multiple_target.c (create_dispatcher_calls): Redirect edge
24100 from a caller of a dispatcher.
24101 (expand_target_clones): Make the clones local.
24102 (ipa_target_clone): Do both target clones and resolvers.
24103 (ipa_dispatcher_calls): Remove the pass.
24104 (pass_dispatcher_calls::gate): Likewise.
24105 (make_pass_dispatcher_calls): Likewise.
24106 * passes.def (pass_target_clone): Put as very first IPA early
24109 2017-02-03 Martin Liska <mliska@suse.cz>
24111 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
24112 in case of a function with ifunc attribute.
24114 2017-02-03 Martin Liska <mliska@suse.cz>
24116 * cgraph.c (cgraph_node::dump): Dump function version info.
24117 * symtab.c (symtab_node::dump_base): Add missing new line.
24119 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
24121 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
24122 (ifcombine_ifandif): Use it.
24124 2017-02-03 Martin Liska <mliska@suse.cz>
24126 * doc/invoke.texi: Document default value for
24127 use-after-scope-direct-emission-threshold.
24129 2017-02-03 Martin Liska <mliska@suse.cz>
24131 PR tree-optimization/79339
24132 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
24133 (format_floating): Likewise.
24135 2017-02-03 Martin Liska <mliska@suse.cz>
24138 * ipa-prop.c (ipa_node_params_t::insert): Remove current
24140 (ipa_node_params_t::remove): Likewise.
24141 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
24142 initialization from removed ipa_node_params_t::insert.
24143 (ipa_node_params::~ipa_node_params): Move from removed
24144 ipa_node_params_t::release.
24145 * symbol-summary.h (symbol_summary::m_released): New member.
24146 Do not release a summary twice. Do not allow to call finalizer
24147 for types of a summary that live in GGC memory.
24149 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24151 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
24154 2017-02-02 Martin Sebor <msebor@redhat.com>
24156 PR middle-end/79275
24157 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
24158 (format_string): Tighten up the range of output for non-constant
24159 strings and correct the expected range for wide non-constant strings.
24161 2017-02-02 Martin Sebor <msebor@redhat.com>
24163 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
24165 PR middle-end/32003
24166 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
24168 (-fdump-tree-@var): Add to index and document how to come up
24169 with pass-specific option and dump file names.
24170 (-fdump-passes): Clarify where to look for output.
24172 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
24174 PR middle-end/77445
24175 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
24176 statistics of the analyzed path; allow threading for speed when
24177 any of BBs along the path are optimized for speed.
24179 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
24181 PR middle-end/78468
24182 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
24183 settings of the virtual registers.
24186 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
24188 * explow.c (get_dynamic_stack_size): Take known alignment of stack
24189 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
24192 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24194 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
24195 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
24197 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24199 * config/s390/s390.md: Add missing comments with the expanded
24201 * config/s390/vector.md: Likewise.
24202 * config/s390/vx-builtins.md: Likewise.
24204 2017-02-02 Jakub Jelinek <jakub@redhat.com>
24207 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
24208 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
24209 conditions on a single line.
24211 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24213 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
24214 __S390_VX__ to __VX__.
24216 2017-02-01 Andrew Pinski <apinski@cavium.com>
24218 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
24219 stmt_info to record_stmt_cost.
24220 (vect_get_known_peeling_cost): Pass stmt_info if known to
24222 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
24223 cpu_vector_cost field into
24224 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
24225 field into vec_int_stmt_cost and vec_fp_stmt_cost.
24226 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
24227 splitting of scalar_stmt_cost and vec_stmt_cost.
24228 (thunderx_vector_cost): Likewise.
24229 (cortexa57_vector_cost): LIkewise.
24230 (exynosm1_vector_cost): Likewise.
24231 (xgene1_vector_cost): Likewise.
24232 (thunderx2t99_vector_cost): Improve after the splitting of the two
24234 (aarch64_builtin_vectorization_cost): Update for the splitting of
24235 scalar_stmt_cost and vec_stmt_cost.
24237 2017-02-01 Torvald Riegel <triegel@redhat.com>
24238 Richard Henderson <rth@redhat.com>
24240 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
24241 conditional on existance of a fast atomic load.
24242 * optabs-query.c (can_atomic_load_p): New function.
24243 * optabs-query.h (can_atomic_load_p): Declare it.
24244 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
24245 no fast atomic load is available for the particular size of access.
24246 (expand_atomic_compare_and_swap): Likewise.
24247 (expand_atomic_load): Likewise.
24248 (expand_atomic_store): Likewise.
24249 (expand_atomic_fetch_op): Likewise.
24250 * testsuite/lib/target-supports.exp
24251 (check_effective_target_sync_int_128): Remove x86 because it provides
24252 no fast atomic load.
24253 (check_effective_target_sync_int_128_runtime): Likewise.
24255 2017-02-01 Richard Biener <rguenther@suse.de>
24257 * graphite.c: Include tree-vectorizer.h for find_loop_location.
24258 (graphite_transform_loops): Provide opt-info for optimized nests.
24259 * tree-parloop.c (parallelize_loops): Provide opt-info for
24260 parallelized loops.
24262 2017-02-01 Richard Biener <rguenther@suse.de>
24264 PR middle-end/79315
24265 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
24266 was not set before.
24268 2017-02-01 Richard Biener <rguenther@suse.de>
24270 PR tree-optimization/71824
24271 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
24272 Verify the loops are valid in the merged SESE region.
24273 (scop_detection::can_represent_loop_1): Check analyzing the
24274 evolution of the number of iterations in the region succeeds.
24276 2017-01-31 Ian Lance Taylor <iant@golang.org>
24278 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
24279 REG_ARGS_SIZE note to 32-bit push insns and call insn.
24281 2017-01-31 David Malcolm <dmalcolm@redhat.com>
24283 PR preprocessor/79210
24284 * input.c (get_substring_ranges_for_loc): Replace line_width
24285 assertion with error-handling.
24287 2017-01-31 Richard Biener <rguenther@suse.de>
24289 PR tree-optimization/77318
24290 * graphite-sese-to-poly.c (extract_affine): Fix assert.
24291 (create_pw_aff_from_tree): Take loop parameter.
24292 (add_condition_to_pbb): Pass loop of the condition to
24293 create_pw_aff_from_tree.
24295 2017-01-31 Jakub Jelinek <jakub@redhat.com>
24297 * config/s390/s390.c (s390_asan_shadow_offset): New function.
24298 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
24300 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
24304 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
24306 (convert_int_to_float128): Likewise.
24307 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
24308 (convert_int_to_float128): Likewise.
24309 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
24310 (UNSPEC_IEEE128_CONVERT): Likewise.
24311 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
24312 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
24313 Use local variables for IBM extended format.
24314 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
24315 (fix_trunc<mode>si2_fprs): Likewise.
24316 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
24317 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
24318 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
24319 to know that we can now have integers of all sizes in vector
24321 (fix<uns>_<mode>di2_hw): Likewise.
24322 (float<uns>_<mode>si2_hw): Likewise.
24323 (fix_<mode>si2_hw): Likewise.
24324 (fixuns_<mode>si2_hw): Likewise.
24325 (float<uns>_<mode>di2_hw): Likewise.
24326 (float_<mode>di2_hw): Likewise.
24327 (float_<mode>si2_hw): Likewise.
24328 (floatuns_<mode>di2_hw): Likewise.
24329 (floatuns_<mode>si2_hw): Likewise.
24330 (xscvqp<su>wz_<mode>): Delete, no longer used.
24331 (xscvqp<su>dz_<mode>): Likewise.
24332 (xscv<su>dqp_<mode>): Likewise.
24333 (ieee128_mfvsrd_64bit): Likewise.
24334 (ieee128_mfvsrd_32bit): Likewise.
24335 (ieee128_mfvsrwz): Likewise.
24336 (ieee128_mtvsrw): Likewise.
24337 (ieee128_mtvsrd_64bit): Likewise.
24338 (ieee128_mtvsrd_32bit): Likewise.
24340 2017-01-31 Martin Liska <mliska@suse.cz>
24343 * ipa-prop.c (ipa_free_all_node_params): Call release method
24344 instead of ~sumbol_summary to not to trigger double times
24347 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
24349 PR tree-optimization/71691
24350 * bitmap.h (class auto_bitmap): New.
24351 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
24352 is_maybe_undefined instead of ssa_undefined_value_p.
24354 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24356 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
24357 __S390_ARCH_LEVEL__ to __ARCH__.
24359 2017-01-31 Jakub Jelinek <jakub@redhat.com>
24361 PR tree-optimization/79267
24362 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
24363 if should_remove_lhs_p is true.
24365 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
24368 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
24369 (add_alignment_attribute): New.
24370 (base_type_die): Add alignment attribute.
24371 (subrange_type_die): Likewise.
24372 (modified_type_die): Likewise.
24373 (gen_array_type_die): Likewise.
24374 (gen_descr_array_type_die: Likewise.
24375 (gen_enumeration_type_die): Likewise.
24376 (gen_subprogram_die): Likewise.
24377 (gen_variable_die): Likewise.
24378 (gen_field_die): Likewise.
24379 (gen_ptr_to_mbr_type_die): Likewise.
24380 (gen_struct_or_union_type_die): Likewise.
24381 (gen_subroutine_type_die): Likewise.
24382 (gen_typedef_die): Likewise.
24383 (base_type_cmp): Compare alignment attribute.
24385 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24388 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
24389 (setb_unsigned) New pattern for setb with CCUNS.
24390 * config/rs6000/rs6000.c (expand_block_compare): Use a different
24391 subfc./subfe sequence to avoid overflow problems. Generate a
24392 shorter sequence with cmpld/setb for power9.
24393 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
24394 for generating subfc. instruction.
24395 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
24396 now uses this instruction.
24398 2017-01-30 Ian Lance Taylor <iant@google.com>
24401 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
24402 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
24404 2017-01-30 Martin Sebor <msebor@redhat.com>
24406 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
24407 Move constant to the right of a relational operator.
24408 (get_mpfr_format_length, format_character, format_string): Ditto.
24409 (should_warn_p, maybe_warn): Same.
24411 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
24413 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
24416 * asan.c (get_translation_unit_decl): Remove function.
24417 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
24419 2017-01-30 Martin Liska <mliska@suse.cz>
24421 PR gcov-profile/79259
24422 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
24423 -fprofile-generate.
24425 2017-01-30 Martin Liska <mliska@suse.cz>
24428 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
24429 Initialize variables with NULL value.
24431 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
24434 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
24436 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
24438 2017-01-30 Richard Biener <rguenther@suse.de>
24440 PR tree-optimization/79276
24441 * tree-vrp.c (process_assert_insertions): Properly adjust common
24442 when removing a duplicate.
24444 * gcc.dg/torture/pr79276.c: New testcase.
24446 2017-01-30 Richard Biener <rguenther@suse.de>
24448 PR tree-optimization/79256
24449 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
24450 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
24452 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
24454 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
24457 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
24458 ("*r<noxa>sbg_<mode>_sll_bitmask")
24459 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
24460 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
24461 Use contiguous_bitmask_nowrap_operand.
24463 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24466 * config/rs6000/altivec.h (vec_xl): Revise #define.
24467 (vec_xst): Likewise.
24469 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
24471 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
24473 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
24475 PR rtl-optimization/79194
24476 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
24477 traps before call to bypass_conditional_jumps.
24479 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
24481 PR tree-optimization/71374
24482 * lra-constraints.c (check_conflict_input_operands): New.
24483 (match_reload): Use it.
24485 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
24488 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
24489 account to calculate conflict_set.
24491 2017-01-27 Bin Cheng <bin.cheng@arm.com>
24493 PR rtl-optimization/78559
24494 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
24495 other_insn in combine.
24497 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
24499 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
24500 uint16_type_node for BT_UINT16.
24502 2017-01-27 David Malcolm <dmalcolm@redhat.com>
24504 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
24505 "RTL Tests" to menu.
24506 (GIMPLE Tests): New node.
24507 (RTL Tests): New node.
24509 2017-01-27 Richard Biener <rguenther@suse.de>
24511 PR tree-optimization/79245
24512 * tree-loop-distribution.c (distribute_loop): Apply cost
24513 modeling also to detected patterns.
24515 2017-01-27 Richard Biener <rguenther@suse.de>
24517 PR tree-optimization/71433
24518 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
24519 (compare_assert_loc): New function.
24520 (process_assert_insertions): Sort and optimize assert locations
24521 to remove duplicates and push down identical assertions on
24522 edges to their destination block.
24524 2017-01-27 Richard Biener <rguenther@suse.de>
24526 PR tree-optimization/79244
24527 * tree-vrp.c (remove_range_assertions): Forcefully propagate
24528 out SSA names even if abnormal.
24530 2017-01-27 Jakub Jelinek <jakub@redhat.com>
24532 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
24533 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
24534 instead of MPFR_RNDN.
24536 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
24539 * arm.c (arm_option_override): Don't call build_target_option_node
24540 until after doing all option overrides.
24541 (arm_valid_target_attribute_tree): Likewise.
24543 2017-01-27 Martin Liska <mliska@suse.cz>
24545 * doc/invoke.texi (-fprofile-arcs): Document profiling support
24546 for {cd}tors and C++ {cd}tors.
24548 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
24550 * config/s390/s390.md ("*setmem_long_and")
24551 ("*setmem_long_and_31z"): Use zero_extend instead of and.
24553 2017-01-26 Martin Sebor <msebor@redhat.com>
24555 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
24558 2017-01-26 Martin Sebor <msebor@redhat.com>
24560 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
24561 HAVE_DFmode before using XFmode or DFmode.
24562 (parse_directive): Avoid using the z length modifier to avoid
24563 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
24565 PR middle-end/78703
24566 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
24567 to accept adjustment as an array.
24568 (get_int_range): New function.
24569 (struct directive): Make width and prec arrays.
24570 (directive::set_width, directive::set_precision): Call get_int_range.
24571 (format_integer, format_floating): Handle width and precision ranges.
24572 (format_string, parse_directive): Same.
24574 2017-01-26 Jakub Jelinek <jakub@redhat.com>
24577 * dwarf2out.c (generate_skeleton_bottom_up): For children with
24578 comdat_type_p set, just clone them, but keep the children in the
24582 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
24583 which have direct callers with -fvar-tracking-assignments enabled
24585 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
24586 inside of type units.
24588 2017-01-26 Martin Sebor <msebor@redhat.com>
24590 PR middle-end/78703
24591 * gimple-ssa-sprintf.c (struct result_range): Add likely and
24593 (struct format_result): Replace number_chars, number_chars_min,
24594 and number_chars_max with a single member of struct result_range.
24596 (format_result::operator+=): Adjust.
24597 (struct fmtresult): Remove bounded. Handle likely and unlikely
24599 (fmtresult::adjust_for_width_or_precision): New function.
24600 (fmtresult:type_max_digits): New function.
24601 (bytes_remaining): Handle likely and unlikely counters.
24602 (min_bytes_remaining): Remove.
24603 (format_percent): Simplify.
24604 (format_integer, format_floating): Set likely and unlikely counters.
24605 (get_string_length, format_character, format_string): Same.
24606 (format_plain, should_warn_p): New function.
24607 (maybe_warn): Call should_warn_p. Update diagnostic messages
24608 and handle those for all directives, including plain strings.
24609 (format_directive): Handle likely and unlikely counters.
24610 Remove unnecessary quoting from diagnostics. Add an informational
24612 (add_bytes): Remove.
24613 (pass_sprintf_length::compute_format_length): Simplify.
24614 (try_substitute_return_value): Handle likely and unlikely counters.
24616 2017-01-26 Carl Love <cel@us.ibm.com>
24618 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
24619 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
24621 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
24624 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
24625 endianess for subregs into account.
24626 * lra-constraints.c (lra_constraints): Do risky transformations
24627 always on the first iteration.
24628 * lra-lives.c (check_pseudos_live_through_calls): Add arg
24629 last_call_used_reg_set.
24630 (process_bb_lives): Define and use last_call_used_reg_set.
24631 * lra.c (lra): Always continue after lra_constraints on the first
24634 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
24636 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
24638 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
24640 2017-01-26 Jakub Jelinek <jakub@redhat.com>
24642 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
24643 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
24644 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
24645 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
24646 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
24647 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
24648 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
24649 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
24650 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
24652 2017-01-26 Marek Polacek <polacek@redhat.com>
24655 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
24656 for the third operand.
24658 2017-01-26 Jakub Jelinek <jakub@redhat.com>
24660 PR middle-end/79236
24661 * omp-low.c (struct omp_context): Add simt_stmt field.
24662 (scan_omp_for): Return omp_context *.
24663 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
24664 context to the _simt_ SIMD stmt.
24665 (lower_omp_for): For combined SIMD with sibling _simt_
24666 SIMD, make sure to use the same decls in _looptemp_
24667 clauses as in the sibling.
24669 2017-01-26 David Sherwood <david.sherwood@arm.com>
24671 PR middle-end/79212
24672 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
24675 2017-01-26 Jakub Jelinek <jakub@redhat.com>
24678 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
24679 emit fld b; fld a; if possible.
24681 * brig-builtins.def: Update copyright years.
24682 * config/arm/arm_acle_builtins.def: Update copyright years.
24684 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
24687 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
24688 constraint instead of o for the stxsd instruction.
24690 2017-01-25 Carl Love <cel@us.ibm.com>
24692 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
24693 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
24695 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
24697 * doc/invoke.texi (C++ Dialect Options): Fix typo.
24699 2017-01-25 Richard Biener <rguenther@suse.de>
24701 PR tree-optimization/69264
24702 * target.def (vector_alignment_reachable): Improve documentation.
24703 * doc/tm.texi: Regenerate.
24704 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
24706 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
24707 earlier changes with respect to TYPE_USER_ALIGN.
24708 (vector_alignment_reachable_p): Likewise. Improve dumping.
24710 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24713 * config/arm/arm.md (xordi3): Force constant operand into a register
24716 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24718 * doc/invoke.texi (-fstore-merging): Correct default optimization
24719 levels at which it is enabled.
24720 (-O): Move -fstore-merging from list to...
24723 2017-01-25 Richard Biener <rguenther@suse.de>
24726 * omp-expand.c: Include debug.h.
24727 (expand_omp_taskreg): Make sure to generate early debug before
24728 outlining anything from a function.
24729 (expand_omp_target): Likewise.
24730 (grid_expand_target_grid_body): Likewise.
24732 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
24735 * asan.c (get_translation_unit_decl): New function.
24736 (asan_add_global): Extract modules file name from globals
24737 TRANSLATION_UNIT_DECL name.
24739 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
24742 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
24743 for long calls with APCS frame and VFP.
24745 2017-01-24 David Malcolm <dmalcolm@redhat.com>
24747 * cfg.c (original_copy_tables_initialized_p): New function.
24748 * cfg.h (original_copy_tables_initialized_p): New decl.
24749 * cfgrtl.c (relink_block_chain): Guard the call to
24750 free_original_copy_tables with a call to
24751 original_copy_tables_initialized_p.
24752 * cgraph.h (symtab_node::native_rtl_p): New decl.
24753 * cgraphunit.c (symtab_node::native_rtl_p): New function.
24754 (symtab_node::needed_p): Don't assert for early assembly output
24755 for __RTL functions.
24756 (cgraph_node::finalize_function): Set "force_output" for __RTL
24758 (cgraph_node::analyze): Bail out early for __RTL functions.
24759 (analyze_functions): Update assertion to support __RTL functions.
24760 (cgraph_node::expand): Bail out early for __RTL functions.
24761 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
24763 * function.h (struct function): Update comment for field
24765 * gimple-expr.c: Include "tree-pass.h".
24766 (gimple_has_body_p): Return false for __RTL functions.
24767 * Makefile.in (OBJS): Add run-rtl-passes.o.
24768 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
24770 (gcc::pass_manager::get_clean_slate): New accessor.
24771 * passes.c: Include "insn-addr.h".
24772 (should_skip_pass_p): Add logging. Update logic for running
24773 "expand" to be compatible with both __GIMPLE and __RTL. Guard
24774 property-provider override so it is only done for gimple passes.
24776 (skip_pass): New function.
24777 (execute_one_pass): Call skip_pass when skipping passes.
24778 * read-md.c (md_reader::read_char): Support filtering
24779 the input to a subset of line numbers.
24780 (md_reader::md_reader): Initialize fields
24781 m_first_line and m_last_line.
24782 (md_reader::read_file_fragment): New function.
24783 * read-md.h (md_reader::read_file_fragment): New decl.
24784 (md_reader::m_first_line): New field.
24785 (md_reader::m_last_line): New field.
24786 * read-rtl-function.c (function_reader::create_function): Only
24787 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
24788 curr_properties. Set DECL_INITIAL to a dummy block.
24789 (read_rtl_function_body_from_file_range): New function.
24790 * read-rtl-function.h (read_rtl_function_body_from_file_range):
24792 * run-rtl-passes.c: New file.
24793 * run-rtl-passes.h: New file.
24795 2017-01-24 Jeff Law <law@redhat.com>
24797 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
24800 2017-01-24 Bin Cheng <bin.cheng@arm.com>
24802 PR tree-optimization/79159
24803 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
24804 (record_nonwrapping_iv): Improve boundary using above function if no
24805 value range information.
24807 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
24808 Martin Jambor <mjambor@suse.cz>
24810 * brig-builtins.def: New file.
24811 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
24812 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
24813 (DEF_HSAIL_SAT_BUILTIN): Likewise.
24814 (DEF_HSAIL_INTR_BUILTIN): Likewise.
24815 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
24816 * builtin-types.def (BT_INT8): New.
24817 (BT_INT16): Likewise.
24818 (BT_UINT8): Likewise.
24819 (BT_UINT16): Likewise.
24820 (BT_FN_ULONG): Likewise.
24821 (BT_FN_UINT_INT): Likewise.
24822 (BT_FN_UINT_ULONG): Likewise.
24823 (BT_FN_UINT_LONG): Likewise.
24824 (BT_FN_UINT_PTR): Likewise.
24825 (BT_FN_ULONG_PTR): Likewise.
24826 (BT_FN_INT8_FLOAT): Likewise.
24827 (BT_FN_INT16_FLOAT): Likewise.
24828 (BT_FN_UINT32_FLOAT): Likewise.
24829 (BT_FN_UINT16_FLOAT): Likewise.
24830 (BT_FN_UINT8_FLOAT): Likewise.
24831 (BT_FN_UINT64_FLOAT): Likewise.
24832 (BT_FN_UINT16_UINT32): Likewise.
24833 (BT_FN_UINT32_UINT16): Likewise.
24834 (BT_FN_UINT16_UINT16_UINT16): Likewise.
24835 (BT_FN_INT_PTR_INT): Likewise.
24836 (BT_FN_UINT_PTR_UINT): Likewise.
24837 (BT_FN_LONG_PTR_LONG): Likewise.
24838 (BT_FN_ULONG_PTR_ULONG): Likewise.
24839 (BT_FN_VOID_UINT64_UINT64): Likewise.
24840 (BT_FN_UINT8_UINT8_UINT8): Likewise.
24841 (BT_FN_INT8_INT8_INT8): Likewise.
24842 (BT_FN_INT16_INT16_INT16): Likewise.
24843 (BT_FN_INT_INT_INT): Likewise.
24844 (BT_FN_UINT_FLOAT_UINT): Likewise.
24845 (BT_FN_FLOAT_UINT_UINT): Likewise.
24846 (BT_FN_ULONG_UINT_UINT): Likewise.
24847 (BT_FN_ULONG_UINT_PTR): Likewise.
24848 (BT_FN_ULONG_ULONG_ULONG): Likewise.
24849 (BT_FN_UINT_UINT_UINT): Likewise.
24850 (BT_FN_VOID_UINT_PTR): Likewise.
24851 (BT_FN_UINT_UINT_PTR: Likewise.
24852 (BT_FN_UINT32_UINT64_PTR): Likewise.
24853 (BT_FN_INT_INT_UINT_UINT): Likewise.
24854 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
24855 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
24856 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
24857 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
24858 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
24859 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
24860 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
24861 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
24862 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
24863 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
24864 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
24865 * doc/frontends.texi: List BRIG FE.
24866 * doc/install.texi (Testing): Add BRIG tesring requirements.
24867 * doc/invoke.texi (Overall Options): Mention BRIG.
24868 * doc/standards.texi (Standards): Doucment BRIG HSA version.
24870 2017-01-24 Richard Biener <rguenther@suse.de>
24872 PR translation/79208
24873 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
24875 2017-01-24 Martin Jambor <mjambor@suse.cz>
24878 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
24879 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
24880 and known_contexts.
24882 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
24884 PR middle-end/79123
24885 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
24886 casts from signed to unsigned really don't have a range.
24888 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
24890 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
24891 GMP_RNDx for compatiblity.
24893 2017-01-24 Martin Liska <mliska@suse.cz>
24896 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
24897 that would prevent us to call alloca with -1 as argument.
24899 2017-01-24 Jakub Jelinek <jakub@redhat.com>
24901 * dwarf2out.c (output_compilation_unit_header, output_file_names):
24902 Avoid -Wformat-security warning.
24904 2017-01-23 Andrew Pinski <apinski@cavium.com>
24906 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
24909 2017-01-23 Martin Sebor <msebor@redhat.com>
24911 PR middle-end/78703
24912 * gimple-ssa-sprintf.c (warn_level): New global.
24913 (format_integer): Use it here and throughout the rest of the file.
24914 Use the same switch to compute sign as base.
24915 (maybe_warn): New function.
24916 (format_directive): Factor out warnings into maybe_warn.
24917 Add debugging output. Use warn_level.
24918 (add_bytes): Use warn_level.
24919 (pass_sprintf_length::compute_format_length): Add debugging output.
24920 (try_substitute_return_value): Same.
24921 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
24923 PR middle-end/78703
24924 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
24925 (struct fmtresult, format_integer, format_floating): Adjust.
24926 (fmtresult::fmtresult): Set max correctly in two argument ctor.
24927 (get_string_length, format_string,format_directive): Same.
24928 (pass_sprintf_length::compute_format_length): Same.
24929 (try_substitute_return_value): Simplify slightly.
24931 PR middle-end/78703
24932 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
24933 (fmtresult::operator+=): Outlined.
24934 (struct fmtresult): Add ctors.
24935 (struct conversion_spec): Rename...
24936 (struct directive): ...to this. Add and remove data members.
24937 (directive::set_width, directive::set_precision): New functions.
24938 (format_percent): Use fmtresult ctor.
24939 (get_width_and_precision): Remove.
24940 (format_integer): Make naming changes. Avoid computing width and
24942 (format_floating): Same. Adjust indentation.
24943 (format_character, format_none): New functions.
24944 (format_string): Moved character handling to format_character.
24945 (format_directive): Remove arguments, change return type.
24946 (parse_directive): New function.
24947 (pass_sprintf_length::compute_format_length): Move directive
24948 parsing to parse_directive.
24950 2017-01-23 Jakub Jelinek <jakub@redhat.com>
24952 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
24953 (assign_assembler_name_if_needed): ... this.
24954 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
24955 (assign_assembler_name_if_needed): ... this.
24956 (free_lang_data_in_cgraph): Adjust callers.
24957 * cgraphunit.c (cgraph_node::analyze): Likewise.
24958 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
24961 2017-01-23 Richard Biener <rguenther@suse.de>
24963 PR tree-optimization/79088
24964 PR tree-optimization/79188
24965 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
24966 resetting loop bounds after last path deletion. Reset loop
24967 bounds of the target loop, make code match the comments.
24968 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
24969 Make sure loops need no fixups.
24971 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
24973 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
24974 exponent support with double type for first argument.
24975 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
24976 type returned by __builtin_vec_extract_sig,
24977 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
24978 functions from "vector int" to "vector unsigned int" or from
24979 "vector long long int" to "vector unsigned long long int".
24980 Changed type returned by __builtin_vec_extract_exp,
24981 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
24982 functions from "vector int" to "vector unsigned int" or from
24983 "vector long long int" to "vector unsigned long long int".
24984 Changed return type of __builtin_vec_test_data_class,
24985 __builtin_vec_test_data_class_sp, and
24986 __builtin_vec_test_data_class_dp from "vector int" to
24987 "vector bool int" or from "vector long long int" to "vector bool
24988 long long int" and changed second argument type from "unsigned
24989 int" to "int". Added new overloaded function forms "vector float
24990 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
24991 "vector float __builtin_vec_insert_exp_sp (vector float, vector
24992 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
24993 double, vector unsigned long long int)" and "vector double
24994 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
24995 long int)". Changed return type of
24996 __builtin_scalar_test_data_class and
24997 __builtin_scalar_test_data_class_sp and
24998 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
24999 int" and changed second argument from "unsigned int" to "int".
25000 Changed type returned by __builtin_scalar_test_neg,
25001 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
25002 from "int" to "bool int". Added new overloaded function form
25003 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
25004 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
25005 exponent double-precision with floating point first argument.
25006 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
25007 documentation of scalar_test_data_class, scalar_test_neg,
25008 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
25009 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
25010 vec_test_data_class built-in functions to reflect refinements in
25011 their type signatures.
25013 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
25015 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
25017 (aarch64_elf_asm_destructor): Likewise.
25019 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
25021 PR rtl-optimization/78634
25022 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
25023 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
25024 * ifcvt.c (noce_try_cmove): Add missing cost check.
25026 PR rtl-optimization/71724
25027 * combine.c (if_then_else_cond): Look for situations where it is
25028 beneficial to undo the work of one of the recursive calls.
25030 2017-01-23 Bin Cheng <bin.cheng@arm.com>
25032 PR tree-optimization/70754
25033 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
25034 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
25035 combined stmt before it if not NULL.
25036 (combine_chains): Process refs reversely and compute dominance point
25039 2017-01-23 Martin Liska <mliska@suse.cz>
25041 PR tree-optimization/79196
25042 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
25043 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
25045 (strlen_optimize_stmt): Call the renamed function.
25047 2017-01-23 Michael Matz <matz@suse.de>
25049 PR tree-optimization/78384
25050 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
25052 2017-01-23 Richard Biener <rguenther@suse.de>
25054 PR tree-optimization/79186
25055 * tree-vrp.c (register_new_assert_for): Make sure we've seen
25056 both incoming edges before moving an assert.
25058 2017-01-23 Martin Jambor <mjambor@suse.cz>
25060 * ipa-prop.c (load_from_param_1): Removed.
25061 (load_from_unmodified_param): Bits from load_from_param_1 put back
25063 (load_from_param): Removed.
25064 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
25065 with stmt. Reverted back to use of load_from_unmodified_param.
25067 2017-01-23 Martin Jambor <mjambor@suse.cz>
25070 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
25071 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
25072 field a pointer to garbage collected vector, mark lattices and
25073 ipcp_orig_node with GTY((skip)).
25074 (ipa_get_param_count): Adjust to descriptors being a pointer.
25075 (ipa_get_param): Likewise.
25076 (ipa_get_type): Likewise.
25077 (ipa_get_param_move_cost): Likewise.
25078 (ipa_set_param_used): Likewise.
25079 (ipa_get_controlled_uses): Likewise.
25080 (ipa_set_controlled_uses): Likewise.
25081 (ipa_is_param_used): Likewise.
25082 (ipa_node_params_t): Move into garbage collector. New methods insert
25084 (ipa_node_params_sum): Annotate wth GTY(()).
25085 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
25087 (ipa_load_from_parm_agg): Adjust declaration.
25088 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
25089 * ipa-profile.c (ipa_profile): Likewise.
25090 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
25091 (ipa_populate_param_decls): Make descriptors parameter garbage
25093 (ipa_dump_param): Adjust to descriptors being a pointer.
25094 (ipa_alloc_node_params): Likewise.
25095 (ipa_initialize_node_params): Likewise.
25096 (load_from_param_1): Make descriptors parameter garbage collected.
25097 (load_from_unmodified_param): Likewise.
25098 (load_from_param): Likewise.
25099 (ipa_load_from_parm_agg): Likewise.
25100 (ipa_node_params::~ipa_node_params): Removed.
25101 (ipa_free_all_node_params): Remove call to delete operator.
25102 (ipa_node_params_t::insert): New.
25103 (ipa_node_params_t::remove): Likewise.
25104 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
25105 copy known_csts and known_contexts vectors.
25106 (ipa_read_node_info): Adjust to descriptors being a pointer.
25107 (ipcp_modif_dom_walker): Make m_descriptors field garbage
25109 (ipcp_transform_function): Make descriptors variable garbage
25112 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
25114 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
25115 * config/i386/avx512dqintrin.h: Ditto.
25116 * config/i386/avx512fintrin.h: Ditto.
25117 * config/i386/i386.c: Handle new builtins.
25118 * config/i386/i386-builtin.def: Add new builtins.
25119 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
25120 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
25122 2017-01-23 Jakub Jelinek <jakub@redhat.com>
25123 Martin Liska <mliska@suse.cz>
25125 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
25126 * asan.c (asan_expand_poison_ifn): Support stores and use
25127 appropriate ASAN report function.
25128 * internal-fn.c (expand_ASAN_POISON_USE): New function.
25129 * internal-fn.def (ASAN_POISON_USE): Declare.
25130 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
25131 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
25132 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
25133 ASAN_POISON calls w/o LHS.
25134 * tree-ssa.c (execute_update_addresses_taken): Create clobber
25135 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
25136 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
25137 * gimplify.c (asan_poison_variables): Add attribute
25138 use_after_scope_memory to variables that really needs to live
25140 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
25141 having the attribute.
25143 2017-01-23 Martin Liska <mliska@suse.cz>
25145 * asan.c (create_asan_shadow_var): New function.
25146 (asan_expand_poison_ifn): Likewise.
25147 * asan.h (asan_expand_poison_ifn): New declaration.
25148 * internal-fn.c (expand_ASAN_POISON): Likewise.
25149 * internal-fn.def (ASAN_POISON): New builtin.
25150 * sanopt.c (pass_sanopt::execute): Expand
25151 asan_expand_poison_ifn.
25152 * tree-inline.c (copy_decl_for_dup_finish): Make function
25154 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
25155 * tree-ssa.c (is_asan_mark_p): New function.
25156 (execute_update_addresses_taken): Rewrite local variables
25157 (identified just by use-after-scope as addressable) into SSA.
25159 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
25161 * doc/install.texi (Specific): opensource.apple.com uses https
25162 now. Remove trailing slash.
25164 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
25166 * README.Portability: Remove note on an Irix compatibility issue.
25168 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
25170 * gcov.c (INCLUDE_ALGORITHM): Define.
25171 (INCLUDE_VECTOR): Define.
25172 No longer include <vector> and <algorithm> directly.
25174 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
25176 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
25178 * doc/invoke.texi (Code Gen Options): Ditto.
25180 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
25183 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
25185 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
25187 rtl-optimization/79125
25188 * cprop.c (local_cprop_pass): Handle cases where we make an
25189 unconditional trap.
25191 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
25195 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
25196 read from, for big endian.
25198 2017-01-20 Jiong Wang <jiong.wang@arm.com>
25200 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
25201 register pauth builtins for LP64 only.
25203 2017-01-20 Marek Polacek <polacek@redhat.com>
25206 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
25209 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
25211 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
25213 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
25214 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
25215 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
25217 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25220 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
25221 in big-endian mode when they are not a single duplicated value.
25223 2017-01-20 Richard Biener <rguenther@suse.de>
25225 * BASE-VER: Bump to 7.0.1.
25227 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
25229 * omp-low.c (omplow_simd_context): New struct. Use it...
25230 (lower_rec_simd_input_clauses): ...here and...
25231 (lower_rec_input_clauses): ...here to hold common data. Adjust all
25232 references to idx, lane, max_vf, is_simt.
25234 2017-01-20 Graham Markall <graham.markall@embecosm.com>
25236 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
25239 2017-01-20 Martin Jambor <mjambor@suse.cz>
25241 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
25242 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
25244 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
25245 (GTFILES): Rename hsa.c to hsa-common.c.
25246 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
25247 * hsa-dump.c: Likewise.
25248 * hsa-gen.c: Likewise.
25249 * hsa-regalloc.c: Likewise.
25250 * ipa-hsa.c: Likewise.
25251 * omp-expand.c: Likewise.
25252 * omp-low.c: Likewise.
25253 * toplev.c: Likewise.
25255 2017-01-20 Marek Polacek <polacek@redhat.com>
25258 * doc/invoke.texi: Document -Wduplicated-branches.
25259 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
25260 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
25261 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
25262 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
25263 return 0 only when not OEP_LEXICOGRAPHIC.
25264 (fold_build_cleanup_point_expr): Use the expression
25265 location when building CLEANUP_POINT_EXPR.
25266 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
25267 * tree.c (add_expr): Handle error_mark_node.
25269 2017-01-20 Martin Liska <mliska@suse.cz>
25272 * tree-profile.c (init_ic_make_global_vars): Do not call
25274 (gimple_init_gcov_profiler): Likewise.
25276 2017-01-20 Martin Liska <mliska@suse.cz>
25279 * cgraph.h (maybe_create_reference): Remove argument and
25281 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
25283 * ipa-cp.c (create_specialized_node): Likewise.
25284 * symtab.c (symtab_node::maybe_create_reference): Handle
25285 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
25287 2017-01-20 Martin Liska <mliska@suse.cz>
25289 * read-rtl-function.c (function_reader::create_function): Use
25290 build_decl instread of build_decl_stat.
25292 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
25294 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
25295 * config/i386/avx512dqintrin.h: Ditto.
25296 * config/i386/avx512fintrin.h: Ditto.
25297 * config/i386/i386-builtin-types.def: Add new types.
25298 * config/i386/i386.c: Handle new types.
25299 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
25300 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
25301 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
25302 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
25303 (__builtin_ia32_kshiftridi): New.
25304 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
25306 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
25310 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
25311 define to rs6000_init_stack_protect_guard.
25312 (rs6000_init_stack_protect_guard): New function.
25314 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
25315 Yunqiang Su <yunqiang.su@imgtec.com>
25317 * config.gcc (supported_defaults): Add madd4.
25318 (with_madd4): Add validation.
25319 (all_defaults): Add madd4.
25320 * config/mips/mips.opt (mmadd4): New option.
25321 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
25323 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
25325 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
25326 (ISA_HAS_FUSED_MADD4): Likewise.
25327 * doc/invoke.texi (-mmadd4): Document the new option.
25328 * doc/install.texi (--with-madd4): Document the new option.
25330 2017-01-19 Jiong Wang <jiong.wang@arm.com>
25332 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
25333 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
25334 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
25335 (aarch64_init_pauth_hint_builtins): New.
25336 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
25337 (aarch64_expand_builtin): Expand new builtins.
25339 2017-01-19 Jiong Wang <jiong.wang@arm.com>
25341 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
25342 * combine-stack-adj.c (no_unhandled_cfa): Handle
25343 REG_CFA_TOGGLE_RA_MANGLE.
25344 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
25345 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
25346 info for return address signing.
25347 (aarch64_expand_epilogue): Likewise.
25349 2017-01-19 Jiong Wang <jiong.wang@arm.com>
25351 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
25352 * config/aarch64/aarch64-protos.h
25353 (aarch64_return_address_signing_enabled): New declaration.
25354 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
25356 (aarch64_expand_prologue): Sign return address before it's pushed onto
25358 (aarch64_expand_epilogue): Authenticate return address fetched from
25360 (aarch64_override_options): Sanity check for ILP32 and ISA level.
25361 (aarch64_attributes): New function attributes for "sign-return-address".
25362 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
25363 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
25364 ("*do_return"): Generate combined instructions according to key index.
25365 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
25366 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
25368 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
25369 * config/aarch64/aarch64.opt (msign-return-address=): New.
25370 * doc/extend.texi (AArch64 Function Attributes): Documents
25371 "sign-return-address=".
25372 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
25374 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
25376 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
25377 overall option summary.
25379 2017-01-19 Jiong Wang <jiong.wang@arm.com>
25381 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
25382 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
25383 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
25384 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
25386 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
25388 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
25389 -mpower9-minmax by default for -mcpu=power9.
25390 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
25391 128-bit floating point.
25393 2017-01-20 Alan Modra <amodra@gmail.com>
25395 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
25396 optimizing for size.
25398 2017-01-20 Alan Modra <amodra@gmail.com>
25401 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
25402 for strcmp and strncmp from corresponding builtin decl.
25404 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
25406 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
25407 instead of i386/rtems-64.h.
25408 * config/i386/rtems-64.h: Remove.
25410 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
25414 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
25416 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
25418 2017-01-19 Tamar Christina <tamar.christina@arm.com>
25420 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
25421 Change int to HOST_WIDE_INT.
25422 * config/aarch64/aarch64-protos.h
25423 (aarch64_simd_gen_const_vector_dup): Likewise.
25424 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
25426 2017-01-19 David Malcolm <dmalcolm@redhat.com>
25428 * langhooks-def.h (lhd_type_for_size): New decl.
25429 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
25430 * langhooks.c (lhd_type_for_size): New function, taken from
25433 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
25435 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
25436 define_bypass for CR latency.
25437 (power9-cracked-alu): Update bypass latency and remove power9-branch.
25438 (power9-alu2): Add define_bypass for CR latency.
25440 (power9-mul): Update insn latency.
25441 (power9-mul-compare): Update insn latency, bypass latency and remove
25444 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25446 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
25448 * config/aarch64/aarch64.md
25449 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
25450 aarch64_nopcrelative_literal_loads.
25451 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
25453 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
25455 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
25456 TARGET_LOONGSON_3A.
25457 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
25459 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
25462 * config.gcc (supported_defaults): Add lxc1-sxc1.
25463 (with_lxc1_sxc1): Add validation.
25464 (all_defaults): Add lxc1-sxc1.
25465 * config/mips/mips.opt (mlxc1-sxc1): New option.
25466 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
25468 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
25469 __mips_no_lxc1_sxc1.
25470 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
25471 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
25472 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
25474 2017-01-19 Richard Biener <rguenther@suse.de>
25476 PR tree-optimization/72488
25477 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
25478 sure to restore SSA info.
25479 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
25481 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
25483 PR rtl-optimization/79121
25484 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
25485 of the inner type when shifting an extended value.
25487 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
25490 * symtab.c (symtab_node::equal_address_to): Fix comparing of
25491 interposable aliases.
25493 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
25496 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
25497 Use the evmergelohi instruction.
25498 (mov_si<mode>_e500_subreg4_2_le): Likewise.
25499 (mov_sitf_e500_subreg8_2_be): Likewise.
25500 (mov_sitf_e500_subreg12_2_le): Likewise.
25501 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
25502 (mov_si<mode>_e500_subreg4_2_be): Likewise.
25503 (mov_sitf_e500_subreg8_2_le): Likewise.
25504 (mov_sitf_e500_subreg12_2_be): Likewise.
25506 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25508 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
25509 attribute from vecsimple to vecperm.
25510 (altivec_vbpermq2): Likewise.
25512 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25515 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
25517 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25518 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
25519 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
25520 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
25521 case where N arg is SIZE_MAX.
25522 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
25523 (cmpstrsi): Add pattern.
25525 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
25527 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
25528 __builtin_vec_revb builtins.
25529 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
25530 built-in functions to support generation of the ISA 3.0 XXBR<x>
25531 vector byte reverse instructions.
25532 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
25533 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
25534 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
25535 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
25536 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
25537 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
25538 (P9V_BUILTIN_VEC_REVB): Likewise.
25539 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
25540 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
25541 (p9_xxbrq_v16qi): Likewise.
25542 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
25543 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
25544 (p9_xxbrh_v8hi): Likewise.
25545 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
25546 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
25547 vec_revb built-in functions.
25549 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
25551 PR rtl-optimization/78952
25552 * config/i386/i386.md (any_extract): New code iterator.
25553 (*insvqi_2): Use any_extract for source operand.
25554 (*insvqi_3): Use any_shiftrt for source operand.
25556 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
25558 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
25560 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
25562 2017-01-18 Matthias Klose <doko@ubuntu.com>
25564 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
25566 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25568 * config/rs6000/altivec.h (vec_bperm): Change #define.
25569 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
25570 (altivec_vbpermq2): New define_insn.
25571 (altivec_vbpermd): Likewise.
25572 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
25573 function interface.
25574 (VBPERMD): Likewise.
25575 (VBPERM): New polymorphic function interface.
25576 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
25577 Add entries for P9V_BUILTIN_VEC_VBPERM.
25578 * doc/extend.texi: Add interfaces for vec_bperm.
25580 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25582 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
25583 first letter of error messages.
25584 (s390_resolve_overloaded_builtin): Likewise.
25585 * config/s390/s390.c (s390_expand_builtin): Likewise.
25586 (s390_invalid_arg_for_unprototyped_fn): Likewise.
25587 (s390_valid_target_attribute_inner_p): Likewise.
25588 * config/s390/s390.md ("tabort"): Likewise.
25590 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
25592 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
25593 (ISA_AVOID_DIV_HILO): New macro.
25594 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
25595 (ISA_HAS_DDIV): Likewise.
25597 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
25599 * doc/invoke.texi (fabi-version): Correct number of occurrences.
25601 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
25603 * doc/invoke.texi (fabi-version): Spelling fix.
25605 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
25608 * doc/invoke.texi (fabi-version): Mention mangling fix for
25611 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
25614 * doc/invoke.texi (fabi-version): Document discriminator mangling.
25616 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
25619 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
25620 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
25622 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
25624 (stack_protect_test): Ditto.
25625 * config/rs6000/rs6000.opt (mstack-protector-guard=,
25626 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
25628 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
25629 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
25630 -mstack-protector-guard-offset=.
25631 (RS/6000 and PowerPC Options): Ditto.
25633 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
25635 * config/i386/i386.h (MASK_CLASS_P): New define.
25636 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
25637 there are no registers from different register sets also when
25638 mask registers are used. Update function comment.
25639 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
25640 to (*k/*r) and (*k/*km) alternatives.
25642 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
25644 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
25645 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
25646 (EH_RETURN_HANDLER_RTX): New define.
25647 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
25648 Force frame pointer in EH return functions.
25649 (aarch64_expand_epilogue): Add barrier for eh_return.
25650 (aarch64_final_eh_return_addr): Remove.
25651 (aarch64_eh_return_handler_rtx): New function.
25652 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
25654 (aarch64_eh_return_handler_rtx): New prototype.
25656 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25658 * config/rs6000/altivec.h (vec_rlmi): New #define.
25659 (vec_vrlnm): Likewise.
25660 (vec_rlnm): Likewise.
25661 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
25662 (UNSPEC_VRLNM): Likewise.
25663 (VIlong): New mode iterator.
25664 (altivec_vrl<VI_char>mi): New define_insn.
25665 (altivec_vrl<VI_char>nm): Likewise.
25666 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
25668 (VRLDNM): Likewise.
25669 (RLNM): New polymorphic function entry.
25670 (VRLWMI): New monomorphic function entry.
25671 (VRLDMI): Likewise.
25672 (RLMI): New polymorphic function entry.
25673 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
25674 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
25675 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
25678 2017-01-17 Jakub Jelinek <jakub@redhat.com>
25681 * dwarf2out.c (field_byte_offset): Restore the
25682 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
25683 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
25684 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
25687 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
25690 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
25692 2017-01-17 Jakub Jelinek <jakub@redhat.com>
25695 * dwarf2out.c (add_data_member_location_attribute): For constant
25696 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
25697 instead of DW_AT_data_member_location, DW_AT_bit_offset and
25698 DW_AT_byte_size attributes.
25700 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
25702 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
25703 after forcing to constant memory when the code model is medium.
25705 2017-01-17 Julia Koval <julia.koval@intel.com>
25708 * config/i386/avx512fintrin.h
25709 (_mm512_i32gather_ps): Change __addr type to void const*.
25710 (_mm512_mask_i32gather_ps): Ditto.
25711 (_mm512_i32gather_pd): Ditto.
25712 (_mm512_mask_i32gather_pd): Ditto.
25713 (_mm512_i64gather_ps): Ditto.
25714 (_mm512_mask_i64gather_ps): Ditto.
25715 (_mm512_i64gather_pd): Ditto.
25716 (_mm512_mask_i64gather_pd): Ditto.
25717 (_mm512_i32gather_epi32): Ditto.
25718 (_mm512_mask_i32gather_epi32): Ditto.
25719 (_mm512_i32gather_epi64): Ditto.
25720 (_mm512_mask_i32gather_epi64): Ditto.
25721 (_mm512_i64gather_epi32): Ditto.
25722 (_mm512_mask_i64gather_epi32): Ditto.
25723 (_mm512_i64gather_epi64): Ditto.
25724 (_mm512_mask_i64gather_epi64): Ditto.
25725 (_mm512_i32scatter_ps): Change __addr type to void*.
25726 (_mm512_mask_i32scatter_ps): Ditto.
25727 (_mm512_i32scatter_pd): Ditto.
25728 (_mm512_mask_i32scatter_pd): Ditto.
25729 (_mm512_i64scatter_ps): Ditto.
25730 (_mm512_mask_i64scatter_ps): Ditto.
25731 (_mm512_i64scatter_pd): Ditto.
25732 (_mm512_mask_i64scatter_pd): Ditto.
25733 (_mm512_i32scatter_epi32): Ditto.
25734 (_mm512_mask_i32scatter_epi32): Ditto.
25735 (_mm512_i32scatter_epi64): Ditto.
25736 (_mm512_mask_i32scatter_epi64): Ditto.
25737 (_mm512_i64scatter_epi32): Ditto.
25738 (_mm512_mask_i64scatter_epi32): Ditto.
25739 (_mm512_i64scatter_epi64): Ditto.
25740 (_mm512_mask_i64scatter_epi64): Ditto.
25741 * config/i386/avx512pfintrin.h
25742 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
25743 (_mm512_mask_prefetch_i32gather_ps): Ditto.
25744 (_mm512_mask_prefetch_i64gather_pd): Ditto.
25745 (_mm512_mask_prefetch_i64gather_ps): Ditto.
25746 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
25747 (_mm512_prefetch_i32scatter_ps): Ditto.
25748 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
25749 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
25750 (_mm512_prefetch_i64scatter_pd): Ditto.
25751 (_mm512_prefetch_i64scatter_ps): Ditto.
25752 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
25753 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
25754 * config/i386/avx512vlintrin.h
25755 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
25756 (_mm_mmask_i32gather_ps): Ditto.
25757 (_mm256_mmask_i32gather_pd): Ditto.
25758 (_mm_mmask_i32gather_pd): Ditto.
25759 (_mm256_mmask_i64gather_ps): Ditto.
25760 (_mm_mmask_i64gather_ps): Ditto.
25761 (_mm256_mmask_i64gather_pd): Ditto.
25762 (_mm_mmask_i64gather_pd): Ditto.
25763 (_mm256_mmask_i32gather_epi32): Ditto.
25764 (_mm_mmask_i32gather_epi32): Ditto.
25765 (_mm256_mmask_i32gather_epi64): Ditto.
25766 (_mm_mmask_i32gather_epi64): Ditto.
25767 (_mm256_mmask_i64gather_epi32): Ditto.
25768 (_mm_mmask_i64gather_epi32): Ditto.
25769 (_mm256_mmask_i64gather_epi64): Ditto.
25770 (_mm_mmask_i64gather_epi64): Ditto.
25771 (_mm256_i32scatter_ps): Change __addr type to void*.
25772 (_mm256_mask_i32scatter_ps): Ditto.
25773 (_mm_i32scatter_ps): Ditto.
25774 (_mm_mask_i32scatter_ps): Ditto.
25775 (_mm256_i32scatter_pd): Ditto.
25776 (_mm256_mask_i32scatter_pd): Ditto.
25777 (_mm_i32scatter_pd): Ditto.
25778 (_mm_mask_i32scatter_pd): Ditto.
25779 (_mm256_i64scatter_ps): Ditto.
25780 (_mm256_mask_i64scatter_ps): Ditto.
25781 (_mm_i64scatter_ps): Ditto.
25782 (_mm_mask_i64scatter_ps): Ditto.
25783 (_mm256_i64scatter_pd): Ditto.
25784 (_mm256_mask_i64scatter_pd): Ditto.
25785 (_mm_i64scatter_pd): Ditto.
25786 (_mm_mask_i64scatter_pd): Ditto.
25787 (_mm256_i32scatter_epi32): Ditto.
25788 (_mm256_mask_i32scatter_epi32): Ditto.
25789 (_mm_i32scatter_epi32): Ditto.
25790 (_mm_mask_i32scatter_epi32): Ditto.
25791 (_mm256_i32scatter_epi64): Ditto.
25792 (_mm256_mask_i32scatter_epi64): Ditto.
25793 (_mm_i32scatter_epi64): Ditto.
25794 (_mm_mask_i32scatter_epi64): Ditto.
25795 (_mm256_i64scatter_epi32): Ditto.
25796 (_mm256_mask_i64scatter_epi32): Ditto.
25797 (_mm_i64scatter_epi32): Ditto.
25798 (_mm_mask_i64scatter_epi32): Ditto.
25799 (_mm256_i64scatter_epi64): Ditto.
25800 (_mm256_mask_i64scatter_epi64): Ditto.
25801 (_mm_i64scatter_epi64): Ditto.
25802 (_mm_mask_i64scatter_epi64): Ditto.
25803 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
25804 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
25805 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
25806 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
25807 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
25808 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
25809 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
25810 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
25811 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
25812 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
25813 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
25814 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
25815 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
25816 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
25817 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
25818 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
25819 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
25820 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
25821 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
25822 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
25823 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
25824 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
25825 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
25826 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
25827 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
25828 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
25829 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
25830 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
25831 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
25832 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
25833 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
25834 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
25835 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
25836 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
25837 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
25838 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
25839 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
25840 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
25841 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
25842 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
25843 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
25844 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
25845 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
25846 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
25847 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
25848 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
25849 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
25850 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
25851 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
25852 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
25853 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
25854 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
25855 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
25856 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
25857 definitions accordingly.
25859 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
25860 Kuan-Lin Chen <kuanlinchentw@gmail.com>
25863 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
25866 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
25869 * ira-conflicts.c (ira_build_conflicts): Update total conflict
25870 hard regs for inner regno.
25872 2017-01-17 Martin Liska <mliska@suse.cz>
25875 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
25876 assumption and add comment.
25878 2017-01-17 Nathan Sidwell <nathan@acm.org>
25880 * ipa-visibility.c (localize_node): New function, broken out of ...
25881 (function_and_variable_visibility): ... here. Call it.
25883 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
25885 PR middle-end/77445
25886 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
25887 correctly set frequency of oudgoing edge.
25888 (duplicate_thread_path): Fix profile updating.
25890 2017-01-17 Jakub Jelinek <jakub@redhat.com>
25893 * configure.ac: Add GCC_BASE_VER.
25894 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
25895 version from BASE-VER file.
25896 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
25897 (gcc.o): Depend on $(BASEVER).
25898 * common.opt (dumpfullversion): New option.
25899 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
25900 * doc/invoke.texi: Document -dumpfullversion.
25901 * doc/install.texi: Document --with-gcc-major-version-only.
25902 * configure: Regenerated.
25904 2017-01-17 Richard Biener <rguenther@suse.de>
25906 PR tree-optimization/71433
25907 * tree-vrp.c (register_new_assert_for): Merge same asserts
25908 on all incoming edges.
25909 (process_assert_insertions_for): Handle insertions at the
25912 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
25914 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
25915 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
25917 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
25920 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
25923 2017-01-17 Alan Modra <amodra@gmail.com>
25926 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
25927 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
25928 symbolic stack limit when pic.
25930 2017-01-16 Martin Sebor <msebor@redhat.com>
25932 PR tree-optimization/78608
25933 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
25935 2017-01-16 Jeff Law <law@redhat.com>
25938 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
25939 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
25940 for several include directories that may be relative to sysroot.
25941 * config/i386/x-mingw32 (gplus_includedir): Define.
25942 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
25943 (native_system_includedir): Likewise.
25944 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
25945 override if TARGET_SYSTEM_ROOT is defined.
25946 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
25948 PR tree-optimization/79090
25949 PR tree-optimization/33562
25950 PR tree-optimization/61912
25951 PR tree-optimization/77485
25952 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
25953 and computed trims into the dump file.
25955 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
25957 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
25959 2017-01-16 Jakub Jelinek <jakub@redhat.com>
25962 * gimplify.c (gimplify_init_constructor): If want_value and
25963 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
25967 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
25968 sequence. Formatting fixes.
25969 (doloop_optimize): Formatting fixes.
25972 * gcc.c (debug_level_greater_than_spec_func): New function.
25973 (static_spec_functions): Add debug-level-gt spec function.
25974 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
25976 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
25977 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
25978 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
25979 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
25980 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
25981 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
25983 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
25985 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
25986 QImode fixups to general and mask registers only.
25988 2017-01-16 Carl Love <cel@us.ibm.com>
25990 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
25991 for built-in functions
25992 vector signed char vec_nabs (vector signed char)
25993 vector signed short vec_nabs (vector signed short)
25994 vector signed int vec_nabs (vector signed int)
25995 vector signed long long vec_nabs (vector signed long long)
25996 vector float vec_nabs (vector float)
25997 vector double vec_nabs (vector double)
25998 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
26000 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
26001 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
26002 * doc/extend.texi: Update the documentation file for the new built-in
26005 2017-01-16 Martin Sebor <msebor@redhat.com>
26007 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
26010 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26012 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
26013 UNSPEC_VSX__XXSPLTD to require special splat handling.
26015 2017-01-16 David Malcolm <dmalcolm@redhat.com>
26018 * system.h: Poison strndup.
26020 2017-01-16 Alan Modra <amodra@gmail.com>
26023 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
26026 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
26028 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
26030 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
26032 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
26033 call recog here. Assert that INSN_CODE (insn) is non-negative.
26035 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
26038 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
26040 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
26041 in the currently scheduled RTL fragment.
26043 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
26045 PR rtl-optimization/78751
26046 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
26049 2017-01-14 Jeff Law <law@redhat.com>
26051 PR tree-optimization/79090
26052 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
26053 variable length stores.
26054 (compute_trims): Delete dead assignment to *trim_tail.
26055 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
26058 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
26060 PR rtl-optimization/78626
26061 PR rtl-optimization/78727
26062 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
26063 of a block, and split such blocks after everything else is finished.
26065 2017-01-14 Alan Modra <amodra@gmail.com>
26068 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
26069 target legitimate_combined_insn.
26070 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
26071 (rs6000_legitimate_combined_insn): New function.
26072 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
26074 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
26075 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
26076 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
26078 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
26080 * doc/frontends.texi (G++ and GCC): Remove references to Java.
26082 2017-01-13 Jeff Law <law@redhat.com>
26084 PR tree-optimization/33562
26085 PR tree-optimization/61912
26086 PR tree-optimization/77485
26087 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
26089 (delete_dead_assignment): Likewise.
26090 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
26091 statement to delete_dead_call and delete_dead_assignment.
26093 2017-01-13 David Malcolm <dmalcolm@redhat.com>
26096 * substring-locations.c (format_warning_va): Strengthen case 1 so
26097 that both endpoints of the substring must be within the format
26098 range for just the substring to be printed.
26100 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
26102 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
26103 * config/i386/i386.c (ix86_target_string): Add missing options
26104 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
26105 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
26106 flags_other and ix86_target_other to flags2_other. Display unknown
26108 (ix86_valid_target_attribute_inner_p): Add missing options and
26109 reorder options by implied ISAs, as in ix86_target_string.
26111 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
26113 * hash-table.h (hash_table::too_empty_p): New function.
26114 (hash_table::expand): Use it.
26115 (hash_table::traverse): Likewise.
26116 (hash_table::empty_slot): Use sizeof (value_type) instead of
26117 sizeof (PTR) to convert bytes to elements. Shrink the table
26118 if the current size is excessive for the current number of
26121 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
26123 * ira-costs.c (record_reg_classes): Break from the inner loop
26124 early once alt_fail is known to be true. Update outer loop
26125 handling accordingly.
26127 2017-01-13 Jeff Law <law@redhat.com>
26129 * tree-ssa-dse.c (decrement_count): New function.
26130 (increment_start_addr, maybe_trim_memstar_call): Likewise.
26131 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
26132 when we know the partially dead statement is a mem* function.
26134 PR tree-optimization/61912
26135 PR tree-optimization/77485
26136 * tree-ssa-dse.c: Include expr.h.
26137 (maybe_trim_constructor_store): New function.
26138 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
26140 PR tree-optimization/33562
26141 PR tree-optimization/61912
26142 PR tree-optimization/77485
26143 * doc/invoke.texi: Document new dse-max-object-size param.
26144 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
26145 * tree-ssa-dse.c: Include params.h.
26146 (dse_store_status): New enum.
26147 (initialize_ao_ref_for_dse): New, partially extracted from
26149 (valid_ao_ref_for_dse, normalize_ref): New.
26150 (setup_live_bytes_from_ref, compute_trims): Likewise.
26151 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
26152 (maybe_trim_partially_dead_store): Likewise.
26153 (maybe_trim_complex_store): Likewise.
26154 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
26155 Track what bytes live from the original store. Return tri-state
26156 for dead, partially dead or live.
26157 (dse_dom_walker): Add constructor, destructor and new private members.
26158 (delete_dead_call, delete_dead_assignment): New extracted from
26160 (dse_optimize_stmt): Make a member of dse_dom_walker.
26161 Use initialize_ao_ref_for_dse.
26163 PR tree-optimization/33562
26164 PR tree-optimization/61912
26165 PR tree-optimization/77485
26166 * sbitmap.h (bitmap_count_bits): Prototype.
26167 (bitmap_clear_range, bitmap_set_range): Likewise.
26168 * sbitmap.c (bitmap_clear_range): New function.
26169 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
26171 2017-01-13 Martin Liska <mliska@suse.cz>
26174 * function.c (set_cfun): Add new argument force.
26175 * function.h (set_cfun): Likewise.
26176 * ipa-inline-transform.c (inline_call): Use the function when
26177 strict alising from is dropped for function we inline to.
26179 2017-01-13 Richard Biener <rguenther@suse.de>
26181 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
26182 for dumping GIMPLE INTEGER_CSTs.
26184 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26186 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
26187 to 201112L since C++17.
26189 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
26192 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
26193 if -fsanitize=kernel-address is present.
26195 2017-01-13 Richard Biener <rguenther@suse.de>
26197 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
26198 as _Literal ( type ) number in case usual suffixes do not
26199 preserve all information.
26201 2017-01-13 Richard Biener <rguenther@suse.de>
26203 PR tree-optimization/77283
26204 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
26205 and ssa-iterators.h.
26206 (is_feasible_trace): Implement a cost model based on joiner
26209 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
26212 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
26213 char or short to __float128/_Float128 directly.
26215 2017-01-12 Martin Sebor <msebor@redhat.com>
26217 to -Wformat-overflow.
26218 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
26219 (min_bytes_remaining): Same.
26220 (get_string_length): Same.
26221 (format_string): Same.
26222 (format_directive): Same.
26224 (pass_sprintf_length::handle_gimple_call): Same.
26226 2017-01-12 Jakub Jelinek <jakub@redhat.com>
26228 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
26229 info.nowrite calls with no lhs that can't throw. Return bool
26230 whether gsi_remove has been called or not.
26231 (pass_sprintf_length::handle_gimple_call): Return bool whether
26232 try_substitute_return_value called gsi_remove. Formatting fix.
26233 (pass_sprintf_length::execute): Don't use gsi_remove if
26234 handle_gimple_call returned true.
26237 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
26238 be removed due to side-effects, don't remove following barrier nor
26239 turn the successor edge into fallthru edge.
26241 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26244 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
26245 element-reversing loads and stores as not swappable.
26247 2017-01-12 Nathan Sidwell <nathan@acm.org>
26248 Nicolai Stange <nicstange@gmail.com>
26250 * combine.c (try_combine): Don't ignore result of overlap checking
26251 loop. Combine overlap & asm check into single loop.
26253 2017-01-12 Richard Biener <rguenther@suse.de>
26255 * tree-pretty-print.c (dump_generic_node): Provide -gimple
26256 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
26258 2017-01-12 Richard Biener <rguenther@suse.de>
26260 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
26261 and TS_TARGET_OPTION directly derive from TS_BASE.
26262 * tree-core.h (tree_optimization_option): Derive from tree_base.
26263 (tree_target_option): Likewise.
26265 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
26267 * config/i386/i386.c (memory_address_length): Increase len
26268 only when rip_relative_addr_p returns false.
26270 2017-01-11 Julia Koval <julia.koval@intel.com>
26272 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
26273 (OPTION_MASK_ISA_SGX_SET): New.
26274 (ix86_handle_option): Handle OPT_msgx.
26275 * config.gcc: Added sgxintrin.h.
26276 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
26277 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
26278 * config/i386/i386.c (ix86_target_string): Add -msgx.
26280 (ix86_option_override_internal): Handle new options.
26281 (ix86_valid_target_attribute_inner_p): Add sgx.
26282 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
26283 * config/i386/i386.opt: Add msgx.
26284 * config/i386/sgxintrin.h: New file.
26285 * config/i386/x86intrin.h: Add sgxintrin.h.
26287 2017-01-11 Jakub Jelinek <jakub@redhat.com>
26290 * fold-const.c (maybe_nonzero_address): Return 1 for function
26292 (tree_single_nonzero_warnv_p): Don't handle function local objects
26296 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
26299 2017-01-11 David Malcolm <dmalcolm@redhat.com>
26302 * opts.c: Include "spellcheck.h"
26303 (struct string_fragment): New struct.
26304 (struct edit_distance_traits<const string_fragment &>): New
26306 (get_closest_sanitizer_option): New function.
26307 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
26309 2017-01-11 Jakub Jelinek <jakub@redhat.com>
26311 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
26313 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
26314 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
26315 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
26316 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
26317 for initial die_offset if dwarf_split_debug_info.
26318 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
26319 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
26321 (output_skeleton_debug_sections): Formatting fix. Use
26322 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
26323 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
26325 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
26327 * config/arm/cortex-a53.md: Add bypasses for
26328 cortex_a53_r2f_cvt.
26329 (cortex_a53_r2f): Only use for transfers.
26330 (cortex_a53_f2r): Likewise.
26331 (cortex_a53_r2f_cvt): Add reservation for conversions.
26332 (cortex_a53_f2r_cvt): Likewise.
26334 2017-01-11 Tamar Christina <tamar.christina@arm.com>
26336 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
26337 to all inlined functions, change static to extern.
26339 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
26342 * config/arm/arm.c (legitimize_pic_address): Handle reference to
26344 (arm_assemble_integer): Likewise.
26346 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
26348 * config.gcc: Use new awk script to check CPU, FPU and architecture
26349 parameters for --with-... options.
26350 * config/arm/parsecpu.awk: New file
26351 * config/arm/arm-cpus.in: New file.
26352 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
26354 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
26356 * config/arm/t-arm: Update dependency rules.
26357 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
26358 of processing .def files.
26359 * config/arm/genopt.sh: Deleted.
26360 * config/arm/gentune.sh: Deleted.
26361 * config/arm/arm-cores.def: Deleted.
26362 * config/arm/arm-arches.def: Deleted.
26363 * config/arm/arm-fpus.def: Deleted.
26364 * config/arm/arm-tune.md: Regenerated.
26365 * config/arm/arm-tables.opt: Regenerated.
26366 * config/arm/arm-cpu.h: New generated file.
26367 * config/arm/arm-cpu-data.h: New generated file.
26368 * config/arm/arm-cpu-cdata.h: New generated file.
26370 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
26373 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
26375 (input_varpool_node): Unpack dynamically_initialized bit.
26377 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
26379 PR rtl-optimization/79032
26380 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
26381 the alignment of the adjusted memory reference against that of MODE,
26382 instead of the alignment of the original memory reference.
26384 2017-01-11 Martin Jambor <mjambor@suse.cz>
26386 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
26388 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
26389 decorated functions.
26391 2017-01-11 Richard Biener <rguenther@suse.de>
26393 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
26394 set range/nonnull info for PHI results. Do not set it on
26395 stmts marked for removal.
26397 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
26399 * expr.c (store_field): In the bitfield case, fetch the return value
26400 from the registers before applying a single big-endian adjustment.
26401 Always do a final load for a BLKmode value not larger than a word.
26403 2017-01-10 David Malcolm <dmalcolm@redhat.com>
26406 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
26407 that we correctly handle column numbers greater than
26408 LINE_MAP_MAX_COLUMN_NUMBER.
26410 2017-01-10 Martin Sebor <msebor@redhat.com>
26412 PR middle-end/78245
26413 * gimple-ssa-sprintf.c (get_destination_size): Call
26414 {init,fini}object_sizes.
26415 * tree-object-size.c (addr_object_size): Adjust.
26416 (pass_through_call): Adjust.
26417 (pass_object_sizes::execute): Adjust.
26418 * tree-object-size.h (fini_object_sizes): Declare.
26420 2017-01-10 Martin Sebor <msebor@redhat.com>
26422 PR tree-optimization/78775
26423 * builtins.c (get_size_range): Move...
26424 * calls.c: ...to here.
26425 (alloc_max_size): Accept zero argument.
26426 (operand_signed_p): Remove.
26427 (maybe_warn_alloc_args_overflow): Call get_size_range.
26428 * calls.h (get_size_range): Declare.
26430 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
26432 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
26433 from TI's devices.csv file as of September 2016.
26434 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
26436 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
26438 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
26439 * doc/invoke.texi: Likewise.
26440 * doc/md.texi: Likewise.
26441 * doc/objc.texi: Likewise.
26443 2017-01-10 Joshua Conner <joshconner@google.com>
26445 * config/arm/fuchsia-elf.h: New file.
26446 * config/fuchsia.h: New file.
26447 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
26448 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
26450 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
26452 2016-01-10 Richard Biener <rguenther@suse.de>
26454 PR tree-optimization/79034
26455 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
26456 Propagate out degenerate PHIs in the joiner.
26458 2017-01-10 Martin Liska <mliska@suse.cz>
26460 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
26461 (sort_congruence_classes_by_decl_uid): Likewise.
26462 (sort_congruence_class_groups_by_decl_uid): Likewise.
26463 (sem_item_optimizer::merge_classes): Sort class, groups in these
26464 classes and members in the groups by DECL_UID of declarations.
26465 This would make merge operations stable.
26467 2017-01-10 Martin Liska <mliska@suse.cz>
26469 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
26470 usage of m_classes_vec.
26471 (sem_item_optimizer::~sem_item_optimizer): Likewise.
26472 (sem_item_optimizer::get_group_by_hash): Likewise.
26473 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
26474 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
26475 (sem_item_optimizer::verify_classes): Likewise.
26476 (sem_item_optimizer::process_cong_reduction): Likewise.
26477 (sem_item_optimizer::dump_cong_classes): Likewise.
26478 (sem_item_optimizer::merge_classes): Likewise.
26479 * ipa-icf.h (congruence_class_hash): Rename from
26480 congruence_class_group_hash. Remove declaration of m_classes_vec.
26482 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
26484 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
26485 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
26486 * config.gcc: Add avx512vpopcntdqintrin.h.
26487 * config/i386/avx512vpopcntdqintrin.h: New.
26488 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
26489 * config/i386/i386-builtin-types.def: Add new types.
26490 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
26491 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
26492 __builtin_ia32_vpopcountq_v8di_mask): New.
26493 * config/i386/i386-c.c (ix86_target_macros_internal): Define
26494 __AVX512VPOPCNTDQ__.
26495 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
26496 (PTA_AVX512VPOPCNTDQ): Define.
26497 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
26498 TARGET_AVX512VPOPCNTDQ_P): Define.
26499 * config/i386/i386.opt: Add mavx512vpopcntdq.
26500 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
26501 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
26503 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
26505 PR middle-end/77484
26506 * predict.def (PRED_CALL): Set to 67.
26508 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
26510 * expr.c (store_field): In the bitfield case, if the value comes from
26511 a function call and is of an aggregate type returned in registers, do
26512 not modify the field mode; extract the value in all cases if the mode
26513 is BLKmode and the size is not larger than a word.
26515 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
26518 * config/i386/cpuid.h: Fix undefined behavior.
26520 2017-01-04 Jeff Law <law@redhat.com>
26522 PR tree-optimization/79007
26523 PR tree-optimization/67955
26524 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
26525 conservative for pt.null when flag_non_call_exceptions is on.
26527 2017-01-09 Jakub Jelinek <jakub@redhat.com>
26529 PR translation/79019
26530 PR translation/79020
26531 * params.def (PARAM_INLINE_MIN_SPEEDUP,
26532 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
26533 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
26535 * config/avr/avr.opt (maccumulate-args): Likewise.
26536 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
26537 * common.opt (freport-bug): Likewise.
26538 * cif-code.def (CIF_FINAL_ERROR): Likewise.
26539 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
26540 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
26541 translatable string.
26542 * config/i386/i386.c (function_value_32): Likewise.
26543 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
26544 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
26546 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
26547 * common/config/msp430/msp430-common.c (msp430_handle_option):
26549 * symtab.c (symtab_node::verify_base): Likewise.
26550 * opts.c (set_debug_level): Likewise.
26551 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
26552 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
26553 missing whitespace to translatable strings.
26554 * config/avr/avr.md (bswapsi2): Fix typo in comment.
26555 * config/sh/superh.h: Likewise.
26556 * config/i386/xopintrin.h: Likewise.
26557 * config/i386/znver1.md: Likewise.
26558 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
26559 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
26560 * double-int.h (struct double_int): Likewise.
26561 * double-int.c (div_and_round_double): Likewise.
26562 * wide-int.cc: Likewise.
26563 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
26564 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
26565 * cfgcleanup.c (crossjumps_occured): Renamed to ...
26566 (crossjumps_occurred): ... this.
26567 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
26570 PR tree-optimization/78899
26571 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
26572 returning bool return struct loop *, NULL for failure and the new
26574 (versionable_outer_loop_p): Don't version outer loop if it has
26575 dont_vectorized bit set.
26576 (tree_if_conversion): When versioning outer loop, ensure
26577 tree_if_conversion is performed also on the inner loop of the
26578 non-vectorizable outer loop copy.
26579 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
26580 LOOP_VECTORIZED in inner loop of the scalar outer loop and
26581 prevent vectorization of it.
26582 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
26583 the outer loop vectorization of the non-scalar version is attempted
26584 before vectorization of the inner loop in scalar version. If
26585 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
26586 vectorization of its inner loop.
26587 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
26588 has 2 inner loops, rename also on edges from bb whose single pred
26589 is outer_loop->header. Fix typo in function comment.
26591 2017-01-09 Martin Sebor <msebor@redhat.com>
26594 * asan.c (asan_emit_stack_protection): Increase local buffer size
26595 to avoid snprintf truncation warning.
26597 2017-01-09 Andrew Pinski <apinski@cavium.com>
26599 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
26600 to reference thunderx2t99 for the tuning structure
26601 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
26603 (thunderx2t99_extra_costs): This.
26604 * config/aarch64/aarch64-tune.md: Regenerate.
26605 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
26606 (vulcan_addrcost_table): This.
26607 (vulcan_regmove_cost): Rename to ...
26608 (thunderx2t99_regmove_cost): This.
26609 (vulcan_vector_cost): Rename to ...
26610 (thunderx2t99_vector_cost): this.
26611 (vulcan_branch_cost): Rename to ...
26612 (thunderx2t99_branch_cost): This.
26613 (vulcan_tunings): Rename to ...
26614 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
26615 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
26617 2017-01-09 Martin Jambor <mjambor@suse.cz>
26621 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
26622 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
26623 (propagate_vr_accross_jump_function): Use the above function for all
26624 value range computations for pass-through jump functions and type
26625 converasion from explicit value range values.
26626 (ipcp_propagate_stage): Do not attempt to deduce types of formal
26627 parameters from TYPE_ARG_TYPES.
26628 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
26629 (ipa_write_node_info): Stream type of the actual argument.
26630 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
26632 2017-01-09 Martin Liska <mliska@suse.cz>
26635 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
26636 (lookup_compiler): Do not show error message with have_E.
26638 2017-01-09 Jakub Jelinek <jakub@redhat.com>
26640 PR tree-optimization/78938
26641 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
26642 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
26643 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
26644 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
26647 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26649 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
26652 2017-01-09 Richard Biener <rguenther@suse.de>
26654 PR tree-optimization/78997
26655 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
26656 name condition properly.
26658 2017-01-09 Richard Biener <rguenther@suse.de>
26661 * dwarf2out.c (is_cxx): New overload with context.
26662 (is_naming_typedef_decl): Use it.
26664 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
26666 * invoke.texi (Option Summary): Correct spacing in option lists
26667 and add line breaks to fix over-long lines.
26669 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
26671 PR middle-end/17660
26673 * extend.texi (Common Variable Attributes): Add xref to GCC
26674 Internals manual to explain mode attribute keywords.
26676 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
26679 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
26680 and Preprocessor Options.
26681 (Options for Linking): Document -pthread here....
26682 (RS/6000 and PowerPC Options): ...not here.
26683 (Solaris 2 Options): ...or here.
26684 * doc/cppopts.texi: Document -pthread.
26686 2017-01-08 Martin Sebor <msebor@redhat.com>
26688 PR middle-end/77708
26689 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
26690 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
26691 New member functions.
26692 (format_directive): Used them.
26694 (pass_sprintf_length::handle_gimple_call): Same.
26695 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
26696 to avoid truncation for any argument.
26697 (extract_affine_mul): Same.
26698 * tree.c (get_file_function_name): Same.
26700 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
26702 PR middle-end/77484
26703 * predict.def (PRED_INDIR_CALL): Set to 86.
26705 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
26707 PR preprocessor/54124
26708 * doc/cppopts.texi: Reformat -d subtable to list the full name
26709 of the options. Add cross-reference to the docs for the general
26710 compiler -d options.
26711 * doc/invoke.texi (Developer Options): Add cross-reference to the
26712 preprocessor-specific -d option documentation.
26714 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
26716 PR preprocessor/13498
26717 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
26718 redudant material, and reflect new command-line options.
26719 (System Headers): Likewise.
26721 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
26723 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
26724 -isystem, and -idirafter. Copy-edit.
26725 * doc/cppopts.texi: Copy-edit. Remove contradiction about
26726 default for -ftrack-macro-expansion. Delete obsolete and
26727 badly-formatted implementation details about -fdebug-cpp output.
26728 * doc/cppwarnopts.texi: Copy-edit.
26730 2017-01-07 David Malcolm <dmalcolm@redhat.com>
26733 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
26734 that the transition from a max line width >= 1<<10 to narrower
26735 lines works correctly.
26737 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
26739 * doc/options.texi (PerFunction): New.
26740 * opt-functions.awk (switch_flags): Map both Optimization and
26741 PerFunction to CL_OPTIMIZATION.
26742 * opth-gen.awk: Test for PerFunction flag along with
26744 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
26745 it only when the latter is present. Skip those that don't in
26746 the hash function generator.
26747 * common.opt (fvar-tracking): Mark as PerFunction instead of
26749 (fvar-tracking-assignments): Likewise.
26750 (fvar-tracking-assignments-toggle): Likewise.
26751 (fvar-tracking-uninit): Likewise.
26753 2017-01-07 Jakub Jelinek <jakub@redhat.com>
26755 PR translation/79018
26756 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
26759 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
26762 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
26763 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
26764 TARGET_LONG_JUMP_TABLE_OFFSETS.
26765 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
26766 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
26767 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
26768 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
26769 * config/m68k/m68k.md (tablejump expander): Likewise.
26770 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
26771 TARGET_LONG_JUMP_TABLE_OFFSETS.
26772 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
26773 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
26775 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
26776 David Holsgrove <david.holsgrove@xilinx.com>
26778 * common/config/microblaze/microblaze-common.c
26779 (TARGET_EXCEPT_UNWIND_INFO): Remove.
26780 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
26782 * config/microblaze/microblaze.c (microblaze_must_save_register)
26783 (microblaze_expand_epilogue, microblaze_return_addr): Handle
26785 (microblaze_eh_return): New function.
26786 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
26787 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
26788 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
26789 * config/microblaze/microblaze.md (eh_return): New pattern.
26791 2017-01-06 Jakub Jelinek <jakub@redhat.com>
26793 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
26794 GCC_DIAGNOSTIC_STRINGIFY): Define.
26796 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
26798 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26800 * config/arm/arm.md (<mcrr>): New.
26802 * config/arm/arm.c (arm_arch5te): New.
26803 (arm_option_override): Set arm_arch5te.
26804 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
26806 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
26807 (arm_mcrr_qualifiers): ... this. New.
26808 (MRRC_QUALIFIERS): Define to...
26809 (arm_mrrc_qualifiers): ... this. New.
26810 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
26812 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
26813 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
26814 (MRRCI, mrrc, MRRC): New.
26815 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
26816 VUNSPEC_MRRC2): New.
26818 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26820 * config/arm/arm.md (<mcr>): New.
26822 * config/arm/arm.c (arm_coproc_builtin_available): Add
26823 support for mcr, mrc, mcr2 and mrc2.
26824 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
26825 (arm_mcr_qualifiers): ... this. New.
26826 (MRC_QUALIFIERS): Define to ...
26827 (arm_mrc_qualifiers): ... this. New.
26828 (MCR_QUALIFIERS): Define to ...
26829 (arm_mcr_qualifiers): ... this. New.
26830 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
26832 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
26833 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
26834 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
26835 VUNSPEC_MRC2): New.
26837 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26839 * config/arm/arm.md (*ldc): New.
26843 * config/arm/arm.c (arm_coproc_builtin_available): Add
26844 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
26845 (arm_coproc_ldc_stc_legitimate_address): New.
26846 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
26847 'qualifier_const_pointer'.
26848 (LDC_QUALIFIERS): Define to...
26849 (arm_ldc_qualifiers): ... this. New.
26850 (STC_QUALIFIERS): Define to...
26851 (arm_stc_qualifiers): ... this. New.
26852 * config/arm/arm-protos.h
26853 (arm_coproc_ldc_stc_legitimate_address): New.
26854 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
26855 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
26856 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
26857 stc2, stcl, stc2l): New.
26858 * config/arm/constraints.md (Uz): New.
26859 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
26860 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
26861 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
26862 VUNSPEC_STC2L): New.
26864 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26866 * config/arm/arm.md (<cdp>): New.
26867 * config/arm/arm.c (neon_const_bounds): Rename this ...
26868 (arm_const_bounds): ... this.
26869 (arm_coproc_builtin_available): New.
26870 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
26871 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
26872 (CDP_QUALIFIERS): Define to...
26873 (arm_cdp_qualifiers): ... this. New.
26875 (arm_expand_builtin_args): Add case for 6 arguments.
26876 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
26877 (arm_const_bounds): ... this.
26878 (arm_coproc_builtin_available): New.
26879 * config/arm/arm_acle.h (__arm_cdp): New.
26881 * config/arm/arm_acle_builtins.def (cdp): New.
26883 * config/arm/iterators.md (CDPI,CDP,cdp): New.
26884 * config/arm/neon.md: Rename all 'neon_const_bounds' to
26885 'arm_const_bounds'.
26886 * config/arm/types.md (coproc): New.
26887 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
26888 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
26889 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
26890 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
26892 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26894 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
26895 (UBINOP_QUALIFIERS): New.
26897 (acle_builtin_data): New. Change comment.
26898 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
26899 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
26900 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
26901 arm_acle_builtins.def.
26902 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
26903 (arm_init_acle_builtins): New.
26904 (CRC32_BUILTIN): Remove.
26905 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
26906 crc32cb, crc32ch and crc32cw.
26907 (arm_init_crc32_builtins): Remove.
26908 (arm_init_builtins): Use arm_init_acle_builtins rather
26909 than arm_init_crc32_builtins.
26910 (arm_expand_acle_builtin): New.
26911 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
26912 * config/arm/arm_acle_builtins.def: New.
26914 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26916 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
26917 (arm_builtin_datum): ... this.
26918 (arm_init_neon_builtin): Rename to ...
26919 (arm_init_builtin): ... this. Add a new parameters PREFIX
26920 and USE_SIG_IN_NAME.
26921 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
26922 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
26923 'arm_builtin_datum'.
26924 (arm_init_vfp_builtins): Likewise.
26925 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
26926 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
26927 (arm_expand_neon_args): Rename to ...
26928 (arm_expand_builtin_args): ... this. Rename builtin_arg
26929 enum values and differentiate between ARG_BUILTIN_MEMORY
26930 and ARG_BUILTIN_NEON_MEMORY.
26931 (arm_expand_neon_builtin_1): Rename to ...
26932 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
26933 values, arm_expand_builtin_args and add bool parameter NEON.
26934 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
26935 (arm_expand_vfp_builtin): Likewise.
26936 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
26938 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
26940 PR middle-end/77484
26941 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
26942 * predict.c (tree_estimate_probability_bb): Reverse direction of
26943 polymorphic call predictor.
26945 2017-01-06 David Malcolm <dmalcolm@redhat.com>
26947 * passes.c (execute_one_pass): Split out pass-skipping logic into...
26948 (determine_pass_name_match): ...this new function and...
26949 (should_skip_pass_p): ...this new function.
26951 2017-01-06 Nathan Sidwell <nathan@acm.org>
26953 * ipa-visibility.c (function_and_variable_visibility): Reformat
26954 comments and long lines. Remove extrneous if.
26955 * symtab.c (symtab_node::make_decl_local): Fix code format.
26956 (symtab_node::set_section_for_node): Fix comment typo.
26958 2017-01-06 Martin Liska <mliska@suse.cz>
26961 * lra-constraints.c: Rename invariant to lra_invariant.
26962 * predict.c (set_even_probabilities): Initialize e to NULL.
26964 2017-01-05 Martin Sebor <msebor@redhat.com>
26966 PR tree-optimization/78910
26967 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
26968 (format_integer): Correct off-by-one error in the handling
26969 of precision with negative numbers in signed conversions..
26971 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
26973 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
26975 2017-01-05 Jakub Jelinek <jakub@redhat.com>
26977 PR tree-optimization/71016
26978 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
26979 factor_out_conditional_conversion. Formatting fix.
26980 (factor_out_conditional_conversion): Add cond_stmt argument.
26981 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
26982 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
26985 2017-01-05 David Malcolm <dmalcolm@redhat.com>
26987 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
26988 read-rtl-function.o, and selftest-rtl.o.
26989 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
26990 (selftest::aarch64_test_loading_full_dump): New function.
26991 (selftest::aarch64_run_selftests): New function.
26992 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
26993 selftest::aarch64_run_selftests.
26994 * config/i386/i386.c
26995 (selftest::ix86_test_loading_dump_fragment_1): New function.
26996 (selftest::ix86_test_loading_call_insn): New function.
26997 (selftest::ix86_test_loading_full_dump): New function.
26998 (selftest::ix86_test_loading_unspec): New function.
26999 (selftest::ix86_run_selftests): Call the new functions.
27000 * emit-rtl.c (maybe_set_max_label_num): New function.
27001 * emit-rtl.h (maybe_set_max_label_num): New decl.
27002 * function.c (instantiate_decls): Guard call to
27003 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
27004 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
27006 * gensupport.c (gen_reader::gen_reader): Pass "false"
27007 for new "compact" param of rtx_reader.
27008 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
27009 rather than an empty string for NULL strings.
27010 * read-md.c: Potentially include config.h rather than bconfig.h.
27011 Wrap include of errors.h with #ifdef GENERATOR_FILE.
27012 (have_error): New global, copied from errors.c.
27013 (md_reader::read_name): Rename to...
27014 (md_reader::read_name_1): ...this, adding "out_loc" param,
27015 and converting "missing name or number" to returning false, rather
27017 (md_reader::read_name): Reimplement in terms of read_name_1.
27018 (md_reader::read_name_or_nil): New function.
27019 (md_reader::read_string): Handle "(nil)" by returning NULL.
27020 (md_reader::md_reader): Add new param "compact".
27021 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
27022 (md_reader::read_file): New method.
27023 * read-md.h (md_reader::md_reader): Add new param "compact".
27024 (md_reader::read_file): New method.
27025 (md_reader::is_compact): New accessor.
27026 (md_reader::read_name): Convert return type from void to file_location.
27027 (md_reader::read_name_or_nil): New decl.
27028 (md_reader::read_name_1): New decl.
27029 (md_reader::m_compact): New field.
27030 (noop_reader::noop_reader): Pass "false" for new "compact" param
27032 (rtx_reader::rtx_reader): Add new "compact" param.
27033 (rtx_reader::read_rtx_operand): Make virtual and convert return
27034 type from void to rtx.
27035 (rtx_reader::read_until): New decl.
27036 (rtx_reader::handle_any_trailing_information): New virtual function.
27037 (rtx_reader::postprocess): New virtual function.
27038 (rtx_reader::finalize_string): New virtual function.
27039 (rtx_reader::m_in_call_function_usage): New field.
27040 (rtx_reader::m_reuse_rtx_by_id): New field.
27041 * read-rtl-function.c: New file.
27042 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
27043 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
27044 (selftest::verify_three_block_rtl_cfg): New decl.
27045 * read-rtl-function.h: New file.
27046 * read-rtl.c: Potentially include config.h rather than bconfig.h.
27047 For host, include function.h, memmodel.h, and emit-rtl.h.
27048 (one_time_initialization): New function.
27049 (struct compact_insn_name): New struct.
27050 (compact_insn_names): New array.
27051 (find_code): Handle insn codes in compact dumps.
27052 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
27053 (bind_subst_iter_and_attr): Likewise.
27054 (add_condition_to_string): Likewise.
27055 (add_condition_to_rtx): Likewise.
27056 (apply_attribute_uses): Likewise.
27057 (add_current_iterators): Likewise.
27058 (apply_iterators): Likewise.
27059 (initialize_iterators): Guard usage of apply_subst_iterator with
27060 #ifdef GENERATOR_FILE.
27061 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
27062 (md_reader::read_mapping): Likewise.
27063 (add_define_attr_for_define_subst): Likewise.
27064 (add_define_subst_attr): Likewise.
27065 (read_subst_mapping): Likewise.
27066 (check_code_iterator): Likewise.
27067 (rtx_reader::read_rtx): Likewise. Move one-time initialization
27069 (one_time_initialization): New function.
27070 (rtx_reader::read_until): New method.
27071 (read_flags): New function.
27072 (parse_reg_note_name): New function.
27073 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
27074 Handle reuse_rtx ids.
27075 Wrap iterator lookup within #ifdef GENERATOR_FILE.
27076 Add parsing support for RTL dumps, mirroring the special-cases in
27077 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
27078 values, and calling handle_any_trailing_information.
27079 (rtx_reader::read_rtx_operand): Convert return type from void
27080 to rtx, returning return_rtx. Handle case 'e'. Call
27081 finalize_string on XSTR and XTMPL fields.
27082 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
27083 "(nil)" values were omitted. Call the postprocess vfunc on the
27085 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
27086 class ctor. Initialize m_in_call_function_usage. Call
27087 one_time_initialization.
27088 * rtl-tests.c (selftest::test_uncond_jump): Call
27089 set_new_first_and_last_insn.
27090 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
27091 * selftest-rtl.c: New file.
27092 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
27093 (selftest::get_insn_by_uid): New decl.
27094 * selftest-run-tests.c (selftest::run_tests): Call
27095 read_rtl_function_c_tests.
27096 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
27097 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
27100 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
27102 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
27103 operands in a special way. Assert that pos+len <= mode precision.
27105 2017-01-05 Jakub Jelinek <jakub@redhat.com>
27107 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
27108 3 argument Alias with unlimited for the negative form.
27109 (fno-vect-cost-model): Removed.
27111 2017-01-05 Martin Liska <mliska@suse.cz>
27113 * hsa-gen.c (gen_hsa_divmod): New function.
27114 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
27116 2017-01-05 Martin Liska <mliska@suse.cz>
27119 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
27122 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27124 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
27125 small constant length operands.
27127 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27129 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
27130 between loop iterations.
27132 2017-01-05 Martin Liska <mliska@suse.cz>
27135 * gimplify.c (gimplify_decl_expr): Compare to
27136 asan_poisoned_variables instread of checking flags.
27137 (gimplify_target_expr): Likewise.
27138 (gimplify_expr): Likewise.
27139 (gimplify_function_tree): Conditionally initialize
27140 asan_poisoned_variables.
27142 2017-01-04 Jeff Law <law@redhat.com>
27144 PR tree-optimizatin/78812
27145 * rtl.h (contains_mem_rtx_p): Prototype.
27146 * ifcvt.c (containts_mem_rtx_p): Move from here to...
27147 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
27148 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
27149 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
27150 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
27152 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
27154 * input.c (assert_char_at_range): Default-initialize actual_range.
27156 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
27158 * df-scan.c (df_ref_create_structure): Make regno unsigned,
27159 to match the caller.
27161 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
27163 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
27164 insns after final jump in test to emit dummy move.
27166 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
27168 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
27169 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
27171 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
27173 * multiple_target.c (create_dispatcher_calls): Init e_next.
27174 * tree-ssa-loop-split.c (split_loop): Init border.
27175 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
27178 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
27183 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
27184 (altivec_register_operand): Do not return true if the operand
27185 contains a SUBREG mixing SImode and SFmode.
27186 (vsx_register_operand): Likewise.
27187 (vsx_reg_sfsubreg_ok): New predicate.
27188 (vfloat_operand): Do not return true if the operand contains a
27189 SUBREG mixing SImode and SFmode.
27190 (vint_operand): Likewise.
27191 (vlogical_operand): Likewise.
27192 (gpc_reg_operand): Likewise.
27193 (int_reg_operand): Likewise.
27194 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
27195 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
27196 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
27198 (rs6000_emit_move_si_sf_subreg): New helper function.
27199 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
27200 fixup SUBREGs involving SImode and SFmode.
27201 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
27202 numbers for the new peephole2 optimization.
27203 (peephole2 for SFmode unions): New peephole2 to optimize cases in
27204 the GLIBC math library that do AND/IOR/XOR operations on single
27205 precision floating point.
27206 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
27207 target macros to say whether we need to avoid SUBREGs mixing
27209 (TARGET_ALLOW_SF_SUBREG): Likewise.
27210 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
27211 (UNSPEC_SI_FROM_SF): Likewise.
27212 (iorxor): Change spacing.
27213 (and_ior_xor): New iterator for AND, IOR, and XOR.
27214 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
27215 (movdi_from_sf_zero_ext): Likewise.
27216 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
27217 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
27218 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
27219 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
27220 (fms<mode>4): Likewise.
27221 (fnma<mode>4): Likewise.
27222 (fnms<mode>4): Likewise.
27223 (nfma<mode>4): Likewise.
27224 (nfms<mode>4): Likewise.
27226 2017-01-04 Marek Polacek <polacek@redhat.com>
27229 * doc/invoke.texi: Document -Wpointer-compare.
27231 2017-01-04 Jakub Jelinek <jakub@redhat.com>
27233 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
27236 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
27237 descriptions for -gdwarf-5 and emit them as uleb128 instead of
27240 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
27243 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
27244 documentation of the powerpc_popcntb_ok attribute.
27245 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
27246 code to issue warning messages if a requested CPU configuration is
27247 not supported by the binary (assembler and loader) toolchain.
27248 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
27249 made to define a built-in function that has been disabled.
27250 (paired_init_builtins): Add assertion to prevent ICE if attempt is
27251 made to define a built-in function that has been disabled.
27252 (altivec_init_builtins): Add comment explaining why definition
27253 of the DST built-in functions is not preceded by an assertion
27254 check. Add assertions to prevent ICE if attempts are made to
27255 define an altivec predicate or an abs* built-in function that has
27257 (htm_init_builtins): Add comment explaining why definition of the
27258 htm built-in functions is not preceded by an assertion check.
27260 2017-01-04 Jeff Law <law@redhat.com>
27262 PR tree-optimizatin/67955
27263 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
27264 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
27265 the points-to solution does not include pt_null. Use DECL_PT_UID
27268 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
27270 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
27271 Use gen_int_mode instead of gen_lopwart for const_int operands.
27273 2017-01-04 Jakub Jelinek <jakub@redhat.com>
27275 PR tree-optimization/71563
27276 * match.pd: Simplify X << Y into X if Y is known to be 0 or
27277 out of range value - has low bits known to be zero.
27279 2017-01-04 Alan Modra <amodra@gmail.com>
27281 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
27282 * configure: Regenerate.
27283 * config.in: Regenerate.
27285 2017-01-04 Jakub Jelinek <jakub@redhat.com>
27288 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
27289 a substring of the message, but strcmp with the whole message. Ifdef
27290 ENABLE_NLS, translate the message first using dgettext.
27292 2017-01-03 Jeff Law <law@redhat.com>
27294 PR tree-optimizatin/78856
27295 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
27296 (mark_threaded_blocks): Remove code to truncate thread paths that
27297 cross multiple loop headers. Instead invalidate the cached loop
27298 iteration information and handle case of a thread path walking
27299 into an irreducible region.
27301 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
27304 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
27305 assertions. Add support for doing the signbit if the IEEE 128-bit
27306 floating point value is in a GPR.
27307 * config/rs6000/rs6000.md (Fsignbit): Delete.
27308 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
27309 Update the length attribute if the value is in a GPR.
27310 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
27311 the sign or zero extension instruction, since the value is always 0/1.
27312 (signbit<mode>2_dm2): Delete using <Fsignbit>.
27315 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
27316 extracting SImode to a GPR register so that we can generate a
27317 store, limit the vector to be in a traditional Altivec register
27318 for the vextuwrx instruction.
27320 2017-01-03 Ian Lance Taylor <iant@google.com>
27322 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
27324 2017-01-03 Martin Sebor <msebor@redhat.com>
27326 PR tree-optimization/78696
27327 * gimple-ssa-sprintf.c (format_floating): Correct handling of
27328 precision. Use MPFR for %f for greater fidelity. Correct handling
27330 (pass_sprintf_length::compute_format_length): Set width and precision
27331 specified by asrerisk to void_node for vararg functions.
27332 (try_substitute_return_value): Adjust dump output.
27334 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
27336 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
27338 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
27340 * doc/invoke.texi (SPARC options): Document -mlra as the default.
27341 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
27342 -mlra/-mno-lra was passed to the compiler.
27344 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
27346 PR rtl-optimization/65618
27347 * emit-rtl.c (try_split): Move initialization of "before" and
27348 "after" to just before the call to emit_insn_after_setloc.
27350 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
27352 * doc/md.texi (Standard Names): Remove reference to Java frontend.
27354 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
27356 * dwarf2out.c (gen_enumeration_type_die): When
27357 -gno-strict-dwarf, add a DW_AT_encoding attribute.
27359 2017-01-03 Jakub Jelinek <jakub@redhat.com>
27361 PR tree-optimization/78965
27362 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
27363 Change first argument from const call_info & to call_info &. For %n
27364 set info.nowrite to false.
27366 PR middle-end/78901
27367 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
27368 possibly throwing calls.
27370 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
27371 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
27372 and fns handling, rather than in a separate case SSA_NAME.
27374 2017-01-02 Jeff Law <law@redhat.com>
27376 * config/darwin-driver.c (darwin_driver_init): Const-correctness
27377 fixes for first_period and second_period variables.
27379 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
27382 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
27383 (*insvqi_1): New insn pattern.
27384 (*insvqi_1_mem_rex64): Ditto.
27385 (*insvqi_2): Ditto.
27386 (*insvqi_3): Rename from *insvqi.
27388 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
27390 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
27392 * doc/cfg.texi (Edges): Remove reference to Java.
27393 (Maintaining the CFG): Ditto.
27395 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
27397 PR middle-end/77674
27398 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
27399 transparent aliases.
27401 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
27403 PR middle-end/77484
27404 * predict.def (PRED_CALL): Update hitrate.
27405 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
27406 * predict.c (tree_estimate_probability_bb): Split CALL predictor
27407 into direct/indirect/polymorphic variants.
27409 2017-01-01 Jakub Jelinek <jakub@redhat.com>
27411 Update copyright years.
27413 * gcc.c (process_command): Update copyright notice dates.
27414 * gcov-dump.c (print_version): Ditto.
27415 * gcov.c (print_version): Ditto.
27416 * gcov-tool.c (print_version): Ditto.
27417 * gengtype.c (create_file): Ditto.
27418 * doc/cpp.texi: Bump @copying's copyright year.
27419 * doc/cppinternals.texi: Ditto.
27420 * doc/gcc.texi: Ditto.
27421 * doc/gccint.texi: Ditto.
27422 * doc/gcov.texi: Ditto.
27423 * doc/install.texi: Ditto.
27424 * doc/invoke.texi: Ditto.
27426 Copyright (C) 2017 Free Software Foundation, Inc.
27428 Copying and distribution of this file, with or without modification,
27429 are permitted in any medium without royalty provided the copyright
27430 notice and this notice are preserved.