1 2017-09-19 Jeff Law <law@redhat.com>
3 * function.c (dump_stack_clash_frame_info): New function.
4 * function.h (dump_stack_clash_frame_info): Prototype.
5 (enum stack_clash_probes): New enum.
7 * config/alpha/alpha.c (alpha_expand_prologue): Also check
8 flag_stack_clash_protection.
9 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
10 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
11 (arm_frame_pointer_required): Likewise.
12 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
13 (ia64_expand_prologue): Likewise.
14 * config/mips/mips.c (mips_expand_prologue): Likewise.
15 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
16 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
17 (sparc_flat_expand_prologue): Likewise.
18 * config/spu/spu.c (spu_expand_prologue): Likewise.
20 * explow.c: Include "params.h".
21 (anti_adjust_stack_and_probe_stack_clash): New function.
22 (get_stack_check_protect): Likewise.
23 (compute_stack_clash_protection_loop_data): Likewise.
24 (emit_stack_clash_protection_loop_start): Likewise.
25 (emit_stack_clash_protection_loop_end): Likewise.
26 (allocate_dynamic_stack_space): Use get_stack_check_protect.
27 Use anti_adjust_stack_and_probe_stack_clash.
28 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
29 (emit_stack_clash_protection_loop_start): Likewise.
30 (emit_stack_clash_protection_loop_end): Likewise.
31 * rtl.h (get_stack_check_protect): Prototype.
32 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
33 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
34 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
36 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
38 * doc/tm.texi: Rebuilt.
39 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
40 get_stack_check_protect.
41 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
42 * config/arm/arm.c (arm_expand_prologue): Likewise.
43 (arm_frame_pointer_required): Likewise.
44 * config/i386/i386.c (ix86_expand_prologue): Likewise.
45 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
46 * config/mips/mips.c (mips_expand_prologue): Likewise.
47 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
48 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
49 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
50 (sparc_flat_expand_prologue): Likewise.
52 * common.opt (-fstack-clash-protection): New option.
53 * flag-types.h (enum stack_check_type): Note difference between
54 -fstack-check= and -fstack-clash-protection.
55 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
56 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
57 * toplev.c (process_options): Issue warnings/errors for cases
58 not handled with -fstack-clash-protection.
59 * doc/invoke.texi (-fstack-clash-protection): Document new option.
60 (-fstack-check): Note additional problem with -fstack-check=generic.
61 Note that -fstack-check is primarily for Ada and refer users
62 to -fstack-clash-protection for stack-clash-protection.
63 Document new params for stack clash protection.
65 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
67 * config/i386/i386.md (*scc_bt<mode>): New insn_and_split pattern.
68 (*scc_bt<mode>_1): Ditto.
69 (*scc_bt<mode>_mask): Ditto.
71 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
73 * config/i386/i386.c (ix86_split_long_move): Do not handle
74 address used for LEA in a special way.
76 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
78 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
80 2017-09-19 Martin Sebor <msebor@redhat.com>
83 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
84 of incompatible types.
86 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
88 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
89 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
90 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
91 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
93 2017-09-19 Richard Biener <rguenther@suse.de>
95 PR tree-optimization/82244
96 * tree-vrp.c (remove_range_assertions): Do not propagate
97 a constant to abnormals but replace the assert with a copy.
99 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
101 PR rtl-optimization/57878
102 PR rtl-optimization/68988
103 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
104 avoidance test involving non_reload_pseudos. Move frequency test
105 below the general fragmentation avoidance test.
107 2017-09-19 Richard Biener <rguenther@suse.de>
109 PR tree-optimization/69728
110 * graphite-sese-to-poly.c (schedule_error): New global.
111 (add_loop_schedule): Handle empty domain by failing the
113 (build_original_schedule): Handle schedule_error.
115 2017-09-19 Richard Biener <rguenther@suse.de>
117 * graphite-scop-detection.c (scop_detection::can_represent_loop):
118 Do not iterate to sibling loops but only to siblings of inner
121 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
124 * config/m68k/m68k.md (moveq feeding equality comparison): Check
125 that the registers are different.
127 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
129 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
130 to processor_model and "amdfam17h" to arch_names_table.
131 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
133 2017-09-18 Jakub Jelinek <jakub@redhat.com>
136 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
138 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
139 Alan Hayward <alan.hayward@arm.com>
140 David Sherwood <david.sherwood@arm.com>
142 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
144 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
146 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
147 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
148 vect_schedule_slp_instance.
149 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
150 Update call to vect_slp_analyze_node_operations. Simplify return
152 (vect_slp_analyze_bb_1): Update call accordingly.
153 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
154 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
155 (vect_schedule_slp): Update call accordingly.
157 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
158 Alan Hayward <alan.hayward@arm.com>
159 David Sherwood <david.sherwood@arm.com>
161 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
162 with types that aren't in fact scalar.
164 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
166 * tree-vect-slp.c (vect_record_max_nunits): New function,
168 (vect_build_slp_tree_1): ...here.
169 (vect_build_slp_tree_2): Call it for phis too.
171 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
173 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
174 to vect_get_vec_def_for_operand when getting the mask operand.
176 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
177 Alan Hayward <alan.hayward@arm.com>
178 David Sherwood <david.sherwood@arm.com>
180 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
183 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
184 Alan Hayward <alan.hayward@arm.com>
185 David Sherwood <david.sherwood@arm.com>
187 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
188 calculation for vector booleans.
190 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
191 Alan Hayward <alan.hayward@arm.com>
192 David Sherwood <david.sherwood@arm.com>
194 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
196 (vect_transform_stmt): ...here.
197 (vect_analyze_stmt): Use it instead of calling
198 vectorizable_live_operation directly.
200 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
202 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
203 non-SIMT targets in acc vector loops.
205 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
207 * configure.ac: Add arc and check if assembler supports gdwarf2.
208 * configure: Regenerate.
210 2017-09-18 Richard Biener <rguenther@suse.de>
212 PR tree-optimization/82220
213 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
214 epilogue niters from the min_profitable_iters compute.
216 2017-09-18 Jakub Jelinek <jakub@redhat.com>
219 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
220 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
221 (ix86_init_pic_reg): Revert 2017-09-01 changes.
223 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
226 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
227 switching to a new text section.
229 2017-09-18 Richard Biener <rguenther@suse.de>
231 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
233 (build_alias_set): Reject aliases with no access function.
235 2017-09-18 Richard Biener <rguenther@suse.de>
237 PR tree-optimization/79622
238 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
240 (build_cross_bb_scalars_use): Likewise.
242 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
244 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
246 2017-09-18 Alan Modra <amodra@gmail.com>
249 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
250 stack_pointer_rtx for count 0. Update comments. Break up
251 large rtl expression.
253 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
255 config/i386/i386.c: (xlogue_layout::STUB_NAME_MAX_LEN): Increase to 20
257 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
258 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
259 or avx version of the stub.
261 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
264 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
265 compute the minimum stack alignment. Also update preferred stack
266 boundary for leaf functions.
268 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
270 PR tree-optimization/82228
271 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
274 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
276 * common/config/nds32/nds32-common.c
277 (nds32_option_optimization_table): Refine formatting.
278 (nds32_option_optimization_table): Use -fsched-pressure and
279 -fomit-frame-pointer for specific optimization level.
281 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
283 * config/nds32/nds32.c: Refine formatting and comments.
284 * config/nds32/nds32.h: Likewise.
285 * config/nds32/nds32.md: Likewise.
286 * config/nds32/nds32-cost.c: Likewise.
287 * config/nds32/nds32-isr.c: Likewise.
288 * config/nds32/nds32-md-auxiliary.c: Likewise.
289 * config/nds32/nds32-multiple.md: Likewise.
290 * config/nds32/nds32-predicates.c: Likewise.
292 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
293 Jakub Jelinek <jakub@redhat.com>
295 Add support for -std=c++2a.
296 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
298 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
300 2017-09-15 Steve Ellcey <sellcey@cavium.com>
303 * doc/extend.texi (Common Function Attributes): Add
304 references to ARM, AArch64, and S/390 specific attributes.
305 (Function Specific Option Pragmas): Add AArch64 and S/390
306 to list of back ends that support the target pragma.
308 2017-09-15 Nathan Sidwell <nathan@acm.org>
310 * doc/standards.texi: Fix C++17 description. Update URLs for
313 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
315 * common.opt (Wcast-align=strict): New warning option.
316 * doc/invoke.texi: Document -Wcast-align=strict.
318 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
320 * cgraph.h (cgraph_thunk_info): Add comments.
321 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
322 assert for VIRTUAL_* arguments stricter.
324 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
326 PR tree-optimization/71026
327 * match.pd: Move RDIV patterns from fold-const.c
328 * fold-const.c (distribute_real_division): Removed.
329 (fold_binary_loc): Remove calls to distribute_real_divison.
331 2017-09-15 Jakub Jelinek <jakub@redhat.com>
333 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
334 c++1z and gnu++1z as deprecated. Change other references to
335 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
336 Change -Wc++1z-compat to -Wc++17-compat.
337 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
338 * dwarf2out.c (highest_c_language): Handle C++17.
339 (gen_compile_unit_die): Likewise.
341 2017-09-15 Jakub Jelinek <jakub@redhat.com>
343 PR rtl-optimization/82192
344 * combine.c (make_extraction): Don't look through non-paradoxical
345 SUBREGs or TRUNCATE if pos + len is or might be bigger than
348 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
349 Alan Hayward <alan.hayward@arm.com>
350 David Sherwood <david.sherwood@arm.com>
352 * target.def (function_arg_offset): New hook.
353 * targhooks.h (default_function_arg_offset): Declare.
354 * targhooks.c (default_function_arg_offset): New function.
355 * function.c (locate_and_pad_parm): Use
356 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
357 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
358 (TARGET_FUNCTION_ARG_OFFSET): ...this.
359 * doc/tm.texi: Regenerate.
360 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
361 * config/spu/spu.c (spu_function_arg_offset): New function.
362 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
363 * system.h (FUNCTION_ARG_OFFSET): Poison.
365 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
366 Alan Hayard <alan.hayward@arm.com>
367 David Sherwood <david.sherwood@arm.com>
369 * target.def (truly_noop_truncation): New hook.
370 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
371 than TRULY_NOOP_TRUNCATION.
372 * hooks.h (hook_bool_uint_uint_true): Declare.
373 * hooks.c (hook_bool_uint_uint_true): New function.
374 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
375 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
376 * doc/tm.texi: Regenerate.
377 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
378 rather than TRULY_NOOP_TRUNCATION in comments.
379 (simplify_comparison): Likewise.
380 (record_truncated_value): Likewise.
381 * expmed.c (extract_bit_field_1): Likewise.
382 (extract_split_bit_field): Likewise.
383 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
384 instead of TRULY_NOOP_TRUNCATION.
385 * function.c (assign_parm_setup_block): Likewise.
386 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
387 * rtlhooks.c: Include target.h.
388 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
389 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
390 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
391 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
392 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
393 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
394 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
395 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
396 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
397 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
398 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
399 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
400 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
401 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
402 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
403 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
404 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
405 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
406 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
407 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
408 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
409 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
410 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
411 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
412 * config/mips/mips.c (mips_truly_noop_truncation): New function.
413 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
414 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
415 rather than TRULY_NOOP_TRUNCATION in comments.
416 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
417 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
418 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
419 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
420 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
421 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
422 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
423 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
424 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
425 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
426 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
427 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
428 rather than TRULY_NOOP_TRUNCATION in comments.
429 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
430 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
431 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
432 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
433 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
434 TRULY_NOOP_TRUNCATION condition.
435 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
436 (TRULY_NOOP_TRUNCATION): Delete.
437 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
438 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
439 * config/spu/spu.c (spu_truly_noop_truncation): New function.
440 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
441 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
442 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
443 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
444 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
445 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
446 rather than TRULY_NOOP_TRUNCATION in comments.
447 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
448 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
449 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
450 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
451 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
452 * system.h (TRULY_NOOP_TRUNCATION): Poison.
454 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
457 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
458 (*cmp_ior): Likewise.
459 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
460 (*ior_scc_scc_cmp): Likewise.
461 (*and_scc_scc): Likewise.
462 (*and_scc_scc_cmp): Likewise.
464 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
465 Alan Hayard <alan.hayward@arm.com>
466 David Sherwood <david.sherwood@arm.com>
468 * target.def (can_change_mode_class): New hook.
469 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
470 (hard_regno_nregs): Likewise.
471 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
472 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
473 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
474 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
475 (LOAD_EXTEND_OP): Update accordingly.
476 * doc/tm.texi: Regenerate.
477 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
478 CANNOT_CHANGE_MODE_CLASS.
479 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
480 (REG_CAN_CHANGE_MODE_P): ...this new macro.
481 * combine.c (simplify_set): Update accordingly.
482 * emit-rtl.c (validate_subreg): Likewise.
483 * recog.c (general_operand): Likewise.
484 * regcprop.c (mode_change_ok): Likewise.
485 * reload1.c (choose_reload_regs): Likewise.
486 (inherit_piecemeal_p): Likewise.
487 * rtlanal.c (simplify_subreg_regno): Likewise.
488 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
489 instead of CANNOT_CHANGE_MODE_CLASS.
490 (reload_cse_simplify_operands): Likewise.
491 * reload.c (push_reload): Use targetm.can_change_mode_class
492 instead of CANNOT_CHANGE_MODE_CLASS.
493 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
494 REG_CANNOT_CHANGE_MODE_P.
495 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
496 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
497 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
498 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
499 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
500 (arm_can_change_mode_class): New function.
501 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
502 than CANNOT_CHANGE_MODE_CLASS in comments.
503 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
504 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
505 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
506 (ix86_can_change_mode_class): ...this new function, inverting the
507 sense of the return value.
508 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
509 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
510 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
511 (ia64_can_change_mode_class): New function.
512 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
513 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
514 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
515 (m32c_can_change_mode_class): ...this new function, inverting the
516 sense of the return value.
517 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
518 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
519 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
520 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
521 (mips_can_change_mode_class): ...this new function, inverting the
522 sense of the return value.
523 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
524 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
525 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
526 (msp430_can_change_mode_class): New function.
527 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
528 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
529 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
530 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
531 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
532 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
533 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
534 (pa_cannot_change_mode_class): Replace with...
535 (pa_can_change_mode_class): ...this new function, inverting the
536 sense of the return value.
537 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
538 than CANNOT_CHANGE_MODE_CLASS in comments.
539 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
540 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
541 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
542 (pdp11_cannot_change_mode_class): Replace with...
543 (pdp11_can_change_mode_class): ...this new function, inverting the
544 sense of the return value.
545 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
546 * config/powerpcspe/powerpcspe-protos.h
547 (rs6000_cannot_change_mode_class_ptr): Delete.
548 * config/powerpcspe/powerpcspe.c
549 (rs6000_cannot_change_mode_class_ptr): Delete.
550 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
551 (rs6000_option_override_internal): Assign to
552 targetm.can_change_mode_class instead of
553 rs6000_cannot_change_mode_class_ptr.
554 (rs6000_cannot_change_mode_class): Replace with...
555 (rs6000_can_change_mode_class): ...this new function, inverting the
556 sense of the return value.
557 (rs6000_debug_cannot_change_mode_class): Replace with...
558 (rs6000_debug_can_change_mode_class): ...this new function.
559 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
560 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
561 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
562 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
563 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
565 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
566 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
567 (rs6000_option_override_internal): Assign to
568 targetm.can_change_mode_class instead of
569 rs6000_cannot_change_mode_class_ptr.
570 (rs6000_cannot_change_mode_class): Replace with...
571 (rs6000_can_change_mode_class): ...this new function, inverting the
572 sense of the return value.
573 (rs6000_debug_cannot_change_mode_class): Replace with...
574 (rs6000_debug_can_change_mode_class): ...this new function.
575 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
576 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
577 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
578 (s390_can_change_mode_class): ...this new function, inverting the
579 sense of the return value.
580 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
581 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
582 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
583 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
584 (sh_cannot_change_mode_class): Replace with...
585 (sh_can_change_mode_class): ...this new function, inverting the
586 sense of the return value.
587 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
588 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
589 (sparc_can_change_mode_class): New function.
590 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
591 * config/spu/spu.c (spu_can_change_mode_class): New function.
592 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
593 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
594 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
595 (visium_can_change_mode_class): New function.
596 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
598 2017-09-15 Richard Biener <rguenther@suse.de>
600 PR tree-optimization/82217
601 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
602 but not undefined case.
604 2017-09-15 Jakub Jelinek <jakub@redhat.com>
607 * postreload.c (reload_cse_simplify_operands): Skip
608 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
610 2017-09-15 Richard Biener <rguenther@suse.de>
612 PR tree-optimization/68823
613 * graphite-scop-detection.c (build_alias_set): If we have a
614 possible dependence check whether we can handle them by just
615 looking at the DRs DR_ACCESS_FNs.
616 (build_scops): If build_alias_set fails, fail the SCOP.
618 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
620 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
621 to support float128 built-in functions that require the ISA 3.0
623 (BU_FLOAT128_3_HW): Likewise.
624 (SQRTF128): Add support for the IEEE 128-bit square root and fma
628 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
629 support for built-in functions that need the ISA 3.0 IEEE 128-bit
630 floating point instructions.
631 (rs6000_invalid_builtin): Likewise.
632 (rs6000_builtin_mask_names): Likewise.
633 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
634 (RS6000_BTM_FLOAT128_HW): Likewise.
635 (RS6000_BTM_COMMON): Likewise.
636 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
638 * doc/extend.texi (RS/6000 built-in functions): Document the
639 IEEE 128-bit floating point square root and fused multiply-add
642 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
644 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
645 reg (r2) isn't in the set of registers defined in the prologue.
647 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
648 Alan Hayward <alan.hayward@arm.com>
649 David Sherwood <david.sherwood@arm.com>
651 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
652 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
653 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
654 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
655 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
657 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
658 max_vectorization_factor.
659 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
661 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
662 Alan Hayward <alan.hayward@arm.com>
663 David Sherwood <david.sherwood@arm.com>
665 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
666 (vect_worthwhile_without_simd_p): Declare.
667 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
668 (vectorizable_reduction): Use it.
669 * tree-vect-stmts.c (vectorizable_shift): Likewise.
670 (vectorizable_operation): Likewise.
672 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
673 Alan Hayward <alan.hayward@arm.com>
674 David Sherwood <david.sherwood@arm.com>
676 * tree-vectorizer.h (vect_get_num_copies): New function.
677 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
678 * tree-vect-loop.c (vectorizable_reduction): Likewise.
679 (vectorizable_induction): Likewise.
680 (vectorizable_live_operation): Likewise.
681 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
682 (vectorizable_bswap): Likewise.
683 (vectorizable_call): Likewise.
684 (vectorizable_conversion): Likewise.
685 (vectorizable_assignment): Likewise.
686 (vectorizable_shift): Likewise.
687 (vectorizable_operation): Likewise.
688 (vectorizable_store): Likewise.
689 (vectorizable_load): Likewise.
690 (vectorizable_condition): Likewise.
691 (vectorizable_comparison): Likewise.
692 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
694 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
695 Alan Hayward <alan.hayward@arm.com>
696 David Sherwood <david.sherwood@arm.com>
698 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
701 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
702 Alan Hayward <alan.hayward@arm.com>
703 David Sherwood <david.sherwood@arm.com>
705 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
706 an inline wrapper that provides a location.
707 (gimple_build_vector): Likewise.
708 * gimple-fold.c (gimple_build_vector_from_val): New function.
709 (gimple_build_vector): Likewise.
710 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
711 functions to build the initial value. Always return a gimple value.
712 (get_initial_defs_for_reduction): Likewise. Only compute
714 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
715 vect_init_vector on the results from get_initial_def(s)_for_reduction.
716 (vectorizable_induction): Use gimple_build_vector rather than
719 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
720 Alan Hayward <alan.hayward@arm.com>
721 David Sherwood <david.sherwood@arm.com>
723 * target.h (vec_perm_indices): New typedef.
724 (auto_vec_perm_indices): Likewise.
725 * optabs-query.h: Include target.h
726 (can_vec_perm_p): Take a vec_perm_indices *.
727 * optabs-query.c (can_vec_perm_p): Likewise.
728 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
729 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
730 * tree-vect-generic.c (lower_vec_perm): Likewise.
731 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
732 (vect_grouped_load_supported): Likewise.
733 (vect_shift_permute_load_chain): Likewise.
734 (vect_permute_store_chain): Use auto_vec_perm_indices.
735 (vect_permute_load_chain): Likewise.
736 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
737 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
738 Update uses of can_vec_perm_p.
739 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
740 mode with a number of elements. Take a vec_perm_indices *.
741 (vect_create_epilog_for_reduction): Update accordingly.
742 Use auto_vec_perm_indices.
743 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
744 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
745 (vect_transform_slp_perm_load): Likewise.
746 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
747 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
748 (vect_gen_perm_mask_checked): Likewise.
749 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
750 (vect_gen_perm_mask_checked): Likewise.
751 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
752 (vectorizable_store): Likewise.
753 (vectorizable_load): Likewise.
754 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
755 (vectorizable_bswap): Likewise.
757 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
758 Alan Hayward <alan.hayward@arm.com>
759 David Sherwood <david.sherwood@arm.com>
761 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
762 * tree.c (build_vector): Likewise.
763 (build_vector_from_ctor): Update accordingly.
764 (build_vector_from_val): Likewise.
765 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
766 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
767 * tree-vect-generic.c (add_rshift): Likewise.
768 (expand_vector_divmod): Likewise.
769 (optimize_vector_constructor): Likewise.
770 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
771 (vect_transform_slp_perm_load): Likewise.
772 (vect_schedule_slp_instance): Likewise.
773 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
774 (vectorizable_call): Likewise.
775 (vect_gen_perm_mask_any): Likewise. Add elements in order.
776 * expmed.c (make_tree): Likewise.
777 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
778 a vector passed to build_vector.
779 (fold_convert_const): Likewise.
780 (exact_inverse): Likewise.
781 (fold_ternary_loc): Likewise.
782 (fold_relational_const): Likewise.
783 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
784 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
785 (const_unop): Likewise. Store the reduction accumulator in a
786 variable rather than an array.
787 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
788 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
789 the new vector, rather than constructing it after the input arrays.
790 (native_interpret_vector): Use auto_vec<tree> when building
791 a vector passed to build_vector. Add elements in order.
792 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
793 auto_vec<tree> when building a vector passed to build_vector.
794 (vect_create_epilog_for_reduction): Likewise.
795 (vectorizable_induction): Likewise.
796 (get_initial_def_for_reduction): Likewise. Fix indentation of
798 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
800 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
801 passed to build_vector.
803 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
804 Alan Hayward <alan.hayward@arm.com>
805 David Sherwood <david.sherwood@arm.com>
807 * tree-core.h (tree_base::u): Add an "nelts" field.
808 (tree_vector): Use VECTOR_CST_NELTS as the length.
809 * tree.c (tree_size): Likewise.
810 (make_vector): Initialize VECTOR_CST_NELTS.
811 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
812 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
813 TYPE_VECTOR_SUBPARTS.
814 * expr.c (const_vector_mask_from_tree): Consistently use "units"
815 as the number of units, setting it from VECTOR_CST_NELTS.
816 (const_vector_from_tree): Likewise.
817 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
818 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
819 (fold_negate_expr_1): Likewise.
820 (fold_convert_const): Likewise.
821 (const_binop): Likewise. Differentiate the number of output and
823 (const_unop): Likewise.
824 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
825 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
826 in cases that did the opposite.
828 2017-09-14 Richard Biener <rguenther@suse.de>
830 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
833 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
835 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
837 2017-09-14 Jakub Jelinek <jakub@redhat.com>
840 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
841 if and where to split a bb, except for splitting before debug insn
842 sequences followed by non-label real insn. Delete debug insns
843 in between basic blocks.
845 * combine.c (make_compound_operation_int): Formatting fixes.
847 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
848 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
849 * config/netbsd.h (LINK_EH_SPEC): Likewise.
850 * config/sol2.h (LINK_EH_SPEC): Likewise.
851 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
852 * config/s390/linux.h (LINK_SPEC): Likewise.
853 * config/freebsd.h (LINK_EH_SPEC): Likewise.
854 * config/openbsd.h (LINK_EH_SPEC): Likewise.
855 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
856 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
857 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
858 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
859 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
860 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
862 2017-09-13 Jakub Jelinek <jakub@redhat.com>
864 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
866 (ENDFILE_LINUX_SPEC): Likewise.
867 (LINK_EH_SPEC): Likewise.
868 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
869 (LINK_OS_LINUX_SPEC32): Likewise.
870 (LINK_OS_LINUX_SPEC64): Likewise.
871 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
872 (LINK_OS_LINUX_SPEC): Likewise.
874 2017-09-13 Martin Liska <mliska@suse.cz>
877 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
878 CASE_HIGH is NULL_TREE.
880 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
881 Alan Hayward <alan.hayward@arm.com>
882 David Sherwood <david.sherwood@arm.com>
884 * target.def (secondary_memory_needed): New hook.
885 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
886 instead of SECONDARY_MEMORY_NEEDED.
887 (secondary_memory_needed_mode): Likewise.
888 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
889 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
890 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
891 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
892 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
893 * doc/tm.texi: Regenerate.
894 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
895 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
896 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
897 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
898 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
899 * config/i386/i386.c (inline_secondary_memory_needed): Put the
900 mode argument first and change the reg_class arguments to reg_class_t.
901 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
902 Make static. Update the call to inline_secondary_memory_needed.
903 (ix86_register_move_cost): Update the call to
904 inline_secondary_memory_needed.
905 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
906 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
908 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
909 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
911 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
912 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
913 * config/mips/mips.c (mips_secondary_memory_needed): Make static
914 and match hook interface. Add comment from mips.h.
915 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
916 * config/mmix/mmix.md (truncdfsf2): Refer to
917 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
919 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
920 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
921 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
922 (pa_secondary_memory_needed): New function.
923 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
924 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
925 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
926 (pdp11_secondary_memory_needed): Make static and match hook interface.
927 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
928 * config/powerpcspe/powerpcspe-protos.h
929 (rs6000_secondary_memory_needed_ptr): Delete.
930 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
932 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
933 (rs6000_option_override_internal): Assign to
934 targetm.secondary_memory_needed rather than
935 rs6000_secondary_memory_needed_ptr.
936 (rs6000_secondary_memory_needed): Match hook interface.
937 (rs6000_debug_secondary_memory_needed): Likewise.
938 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
939 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
940 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
941 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
942 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
943 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
945 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
946 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
947 (rs6000_option_override_internal): Assign to
948 targetm.secondary_memory_needed rather than
949 rs6000_secondary_memory_needed_ptr.
950 (rs6000_secondary_memory_needed): Match hook interface.
951 (rs6000_debug_secondary_memory_needed): Likewise.
952 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
953 * config/s390/s390.c (s390_secondary_memory_needed): New function.
954 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
955 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
956 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
957 (sparc_secondary_memory_needed): New function.
958 * lra-constraints.c (check_and_process_move): Refer to
959 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
961 (curr_insn_transform): Likewise.
962 (process_alt_operands): Use targetm.secondary_memory_needed
963 instead of TARGET_SECONDARY_MEMORY_NEEDED.
964 (check_secondary_memory_needed_p): Likewise.
965 (choose_split_class): Likewise.
966 * reload.c: Unconditionally include code that was previously
967 conditional on SECONDARY_MEMORY_NEEDED.
968 (push_secondary_reload): Use targetm.secondary_memory_needed
969 instead of TARGET_SECONDARY_MEMORY_NEEDED.
970 (push_reload): Likewise.
971 * reload1.c: Unconditionally include code that was previously
972 conditional on SECONDARY_MEMORY_NEEDED.
973 (choose_reload_regs): Use targetm.secondary_memory_needed
974 instead of TARGET_SECONDARY_MEMORY_NEEDED.
975 (gen_reload): Likewise.
976 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
978 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
979 Alan Hayward <alan.hayward@arm.com>
980 David Sherwood <david.sherwood@arm.com>
982 * target.def (secondary_memory_needed_mode): New hook:
983 * targhooks.c (default_secondary_memory_needed_mode): Declare.
984 * targhooks.h (default_secondary_memory_needed_mode): New function.
985 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
986 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
987 * doc/tm.texi: Regenerate.
988 * lra-constraints.c (check_and_process_move): Use
989 targetm.secondary_memory_needed_mode instead of
990 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
991 (curr_insn_transform): Likewise.
992 * reload.c (get_secondary_mem): Likewise.
993 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
994 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
996 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
997 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
998 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
999 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
1000 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
1002 * config/powerpcspe/powerpcspe-protos.h
1003 (rs6000_secondary_memory_needed_mode): Delete.
1004 * config/powerpcspe/powerpcspe.c
1005 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
1006 (rs6000_secondary_memory_needed_mode): Make static.
1007 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
1008 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
1010 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
1012 (rs6000_secondary_memory_needed_mode): Make static.
1013 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
1014 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
1015 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
1016 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
1017 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
1019 (sparc_secondary_memory_needed_mode): New function.
1020 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
1022 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
1024 * config/aarch64/constraints.md (Umq): New constraint.
1025 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
1027 (mov<mode>): Update condition.
1029 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1031 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
1032 when bitposition is the same.
1034 2017-09-13 Richard Biener <rguenther@suse.de>
1036 * dwarf2out.c (output_die_symbol): Remove.
1037 (output_die): Do not output a DIEs symbol.
1039 2017-09-13 Richard Biener <rguenther@suse.de>
1042 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
1043 default-def to avoid breaking iterator update with the weird
1044 interaction with cgraph_update_edges_for_call_stmt_node.
1046 2017-09-13 Richard Biener <rguenther@suse.de>
1048 * tree-cfg.c (verify_gimple_assign_binary): Add verification
1049 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
1050 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
1051 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
1053 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
1055 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
1056 Disable pc relative literal load irrespective of
1057 TARGET_FIX_ERR_A53_84341 for default.
1059 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
1061 * config/sparc/sparc.c (output_return): Output the source location of
1062 the insn in the delay slot, if any.
1063 (output_sibcall): Likewise.
1065 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
1068 * common.opt (-static-pie): New alias.
1069 (shared): Negate static-pie.
1070 (-no-pie): Update help text.
1072 (static-pie): New option.
1073 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
1074 -static-pie support.
1075 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
1076 (LINK_EH_SPEC): Likewise.
1077 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
1078 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
1079 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
1080 * gcc.c (LINK_COMMAND_SPEC): Likewise.
1081 (init_gcc_specs): Likewise.
1082 (init_spec): Likewise.
1083 (display_help): Update help message for -pie.
1084 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
1087 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
1089 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
1090 (movdi_aarch64): Likewise.
1091 (movti_aarch64): Likewise.
1093 2017-09-12 Simon Wright <simon@pushface.org>
1096 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
1097 calculation of the minor version, always output as 0.
1099 2017-09-12 Jakub Jelinek <jakub@redhat.com>
1102 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
1103 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
1104 on it early, rather than manual conversion late. For
1105 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
1106 instead of performing manual conversion.
1108 2017-09-12 Carl Love <cel@us.ibm.com>
1110 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
1111 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
1113 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
1114 VMULOSW): Add definitions.
1115 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
1116 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
1117 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
1118 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
1119 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
1121 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
1123 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
1125 (movti_aarch64): Likewise.
1126 (movdf_aarch64): Likewise.
1127 (movtf_aarch64): Likewise.
1128 (load_pairdi): Likewise.
1129 (store_pairdi): Likewise.
1130 (load_pairdf): Likewise.
1131 (store_pairdf): Likewise.
1132 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
1133 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
1134 (ldr_got_small_<mode>): Likewise.
1135 (ldr_got_small_28k_<mode>): Likewise.
1136 (ldr_got_tiny): Likewise.
1137 * config/aarch64/iterators.md (ldst_sz): New.
1138 (ldpstp_sz): Likewise.
1139 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
1141 (thunderx_load): Split load_8 to load_16.
1142 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
1144 (thunderx2t99_storepair_basic): Split store_8 to store_16.
1145 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
1146 (xgene1_store_pair): Split store_8 to store_16.
1147 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
1148 (falkor_st_0_st_sd): Split store_8 to store_16.
1150 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
1152 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
1153 and store1/2/3/4 to store_4/8/12/16.
1154 * config/aarch64/aarch64.md: Update for rename.
1155 * config/arm/arm.md: Likewise.: Likewise.
1156 * config/arm/arm.c: Likewise.
1157 * config/arm/thumb1.md: Likewise.
1158 * config/arm/thumb2.md: Likewise.
1159 * config/arm/vfp.md: Likewise.
1160 * config/arm/arm-generic.md: Likewise.
1161 * config/arm/arm1020e.md: Likewise.
1162 * config/arm/arm1026ejs.md: Likewise.
1163 * config/arm/arm1136jfs.md: Likewise.
1164 * config/arm/arm926ejs.md: Likewise.
1165 * config/arm/cortex-a15.md: Likewise.
1166 * config/arm/cortex-a17.md: Likewise.
1167 * config/arm/cortex-a5.md: Likewise.
1168 * config/arm/cortex-a53.md: Likewise.
1169 * config/arm/cortex-a57.md: Likewise.
1170 * config/arm/cortex-a7.md: Likewise.
1171 * config/arm/cortex-a8.md: Likewise.
1172 * config/arm/cortex-a9.md: Likewise.
1173 * config/arm/cortex-m4.md: Likewise.
1174 * config/arm/cortex-m7.md: Likewise.
1175 * config/arm/cortex-r4.md: Likewise.
1176 * config/arm/exynos-m1.md: Likewise.
1177 * config/arm/fa526.md: Likewise.
1178 * config/arm/fa606te.md: Likewise.
1179 * config/arm/fa626te.md: Likewise.
1180 * config/arm/fa726te.md: Likewise.
1181 * config/arm/fmp626.md: Likewise.
1182 * config/arm/iwmmxt.md: Likewise.
1183 * config/arm/ldmstm.md: Likewise.
1184 * config/arm/marvell-pj4.md: Likewise.
1185 * config/arm/xgene1.md: Likewise.
1186 * config/aarch64/thunderx.md: Likewise.
1187 * config/aarch64/thunderx2t99.md: Likewise.
1188 * config/aarch64/falkor.md: Likewise.
1190 2017-09-12 Martin Liska <mliska@suse.cz>
1192 * attribs.c (private_lookup_attribute): New function.
1193 * attribs.h (private_lookup_attribute): Declared here.
1194 (lookup_attribute): Called from this place.
1196 2017-09-12 Richard Biener <rguenther@suse.de>
1198 PR tree-optimization/82157
1199 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
1200 stmts with side-effects.
1202 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1203 Alan Hayward <alan.hayward@arm.com>
1204 David Sherwood <david.sherwood@arm.com>
1206 * target.def (hard_regno_nregs): New hook.
1207 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
1208 * targhooks.h (default_hard_regno_nregs): Declare.
1209 * targhooks.c (default_hard_regno_nregs): New function.
1210 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
1211 (TARGET_HARD_REGNO_NREGS): ...this hook.
1212 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
1213 (CLASS_MAX_NREGS): Likewise.
1214 * doc/tm.texi: Regenerate.
1215 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
1216 instead of HARD_REGNO_NREGS.
1217 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
1218 HARD_REGNO_NREGS in the comment.
1219 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
1220 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
1221 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
1222 Return an unsigned int.
1223 (TARGET_HARD_REGNO_NREGS): Redefine.
1224 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
1225 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
1226 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
1227 (arc_hard_regno_nregs): New function.
1228 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
1229 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
1230 (arm_hard_regno_nregs): New function.
1231 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
1232 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
1233 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
1234 (TARGET_HARD_REGNO_NREGS): Redefine.
1235 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
1236 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
1237 (HARD_REGNO_NREGS): Delete.
1238 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
1239 (cr16_hard_regno_nregs): New function.
1240 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
1241 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
1242 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
1243 (cris_hard_regno_nregs): New function.
1244 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
1245 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
1246 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
1247 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
1248 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
1249 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
1250 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
1251 (frv_hard_regno_nregs): Make static. Take and return an
1253 (frv_class_max_nregs): Remove outdated copy of documentation.
1254 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
1255 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
1256 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
1257 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
1258 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
1259 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
1260 (TARGET_HARD_REGNO_NREGS): Redefine.
1261 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
1262 (CLASS_MAX_NREGS): Update comment.
1263 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
1264 (ia64_hard_regno_nregs): New function.
1265 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
1266 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
1267 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
1268 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
1269 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
1271 (m32c_hard_regno_nregs): Likewise. Make static.
1272 (TARGET_HARD_REGNO_NREGS): Redefine.
1273 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
1274 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
1275 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
1276 (m68k_hard_regno_nregs): New function.
1277 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
1278 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
1279 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
1280 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
1281 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
1282 Take and return an unsigned int.
1283 (TARGET_HARD_REGNO_NREGS): Redefine.
1284 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
1285 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
1286 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
1287 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
1288 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
1289 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
1290 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
1291 (msp430_hard_regno_nregs): Make static. Take and return an
1293 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
1294 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
1295 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
1296 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
1297 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
1298 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
1299 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
1300 (TARGET_HARD_REGNO_NREGS): Redefine.
1301 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
1302 (PA_HARD_REGNO_NREGS): ...this.
1303 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
1304 (PA_HARD_REGNO_NREGS): ...this.
1305 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
1306 (pa_hard_regno_nregs): New function.
1307 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
1308 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
1309 (pdp11_hard_regno_nregs): New function.
1310 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
1311 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
1312 (rs6000_hard_regno_nregs_hook): New function.
1313 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
1314 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
1315 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
1316 Take and return an unsigned int. Move earlier in file.
1317 (TARGET_HARD_REGNO_NREGS): Redefine.
1318 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
1319 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
1320 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
1321 (rl78_hard_regno_nregs): Make static. Take and return an
1323 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
1324 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
1325 (rs6000_hard_regno_nregs_hook): New function.
1326 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
1327 * config/rx/rx.c (rx_hard_regno_nregs): New function.
1328 (TARGET_HARD_REGNO_NREGS): Redefine.
1329 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
1330 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
1331 instead of HARD_REGNO_NREGS.
1332 (s390_hard_regno_nregs): New function.
1333 (s390_hard_regno_mode_ok): Add comment from s390.h.
1334 (TARGET_HARD_REGNO_NREGS): Redefine.
1335 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
1336 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
1337 (sh_hard_regno_nregs): New function.
1338 (sh_pass_in_reg_p): Use it.
1339 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
1340 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
1341 (sparc_hard_regno_nregs): New function.
1342 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
1343 * config/spu/spu.c (spu_hard_regno_nregs): New function.
1344 (spu_function_arg_advance): Use it, supplying a valid register number.
1345 (TARGET_HARD_REGNO_NREGS): Redefine.
1346 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
1347 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
1348 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
1349 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
1350 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
1351 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
1352 (CLASS_MAX_NREGS): Remove copy of old documentation.
1353 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
1354 (visium_hard_regno_nregs): New function.
1355 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
1356 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
1357 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
1358 xtensa_hard_regno_nregs): New function.
1359 * system.h (HARD_REGNO_NREGS): Poison.
1361 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1363 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
1364 hard_regno_nregs instead of HARD_REGNO_NREGS.
1365 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
1366 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
1367 (c6x_expand_epilogue): Likewise.
1368 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
1369 (frv_read_iacc_argument): Likewise.
1370 * config/sh/sh.c: Include regs.h.
1371 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
1372 (regs_used): Likewise.
1373 (output_stack_adjust): Likewise.
1374 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
1375 * expmed.c: Include regs.h.
1376 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
1377 * ree.c: Include regs.h.
1378 (combine_reaching_defs): Use hard_regno_nregs instead of
1380 (add_removable_extension): Likewise.
1382 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1384 * regs.h (hard_regno_nregs): Turn into a function.
1385 (end_hard_regno): Update accordingly.
1386 * caller-save.c (setup_save_areas): Likewise.
1387 (save_call_clobbered_regs): Likewise.
1388 (replace_reg_with_saved_mem): Likewise.
1389 (insert_restore): Likewise.
1390 (insert_save): Likewise.
1391 * combine.c (can_change_dest_mode): Likewise.
1392 (move_deaths): Likewise.
1393 (distribute_notes): Likewise.
1394 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
1395 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
1396 (rs6000_split_multireg_move): Likewise.
1397 (rs6000_register_move_cost): Likewise.
1398 (rs6000_memory_move_cost): Likewise.
1399 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
1400 (rs6000_split_multireg_move): Likewise.
1401 (rs6000_register_move_cost): Likewise.
1402 (rs6000_memory_move_cost): Likewise.
1403 * cselib.c (cselib_reset_table): Likewise.
1404 (cselib_lookup_1): Likewise.
1405 * emit-rtl.c (set_mode_and_regno): Likewise.
1406 * function.c (aggregate_value_p): Likewise.
1407 * ira-color.c (setup_profitable_hard_regs): Likewise.
1408 (check_hard_reg_p): Likewise.
1409 (calculate_saved_nregs): Likewise.
1410 (assign_hard_reg): Likewise.
1411 (improve_allocation): Likewise.
1412 (calculate_spill_cost): Likewise.
1413 * ira-emit.c (modify_move_list): Likewise.
1414 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
1415 (ira_hard_reg_in_set_p): Likewise.
1416 * ira.c (setup_reg_mode_hard_regset): Likewise.
1417 (clarify_prohibited_class_mode_regs): Likewise.
1418 (check_allocation): Likewise.
1419 * lra-assigns.c (find_hard_regno_for_1): Likewise.
1420 (lra_setup_reg_renumber): Likewise.
1421 (setup_try_hard_regno_pseudos): Likewise.
1422 (spill_for): Likewise.
1423 (assign_hard_regno): Likewise.
1424 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
1425 * lra-constraints.c (in_class_p): Likewise.
1426 (lra_constraint_offset): Likewise.
1427 (simplify_operand_subreg): Likewise.
1428 (lra_constraints): Likewise.
1429 (split_reg): Likewise.
1430 (split_if_necessary): Likewise.
1431 (invariant_p): Likewise.
1432 (inherit_in_ebb): Likewise.
1433 * lra-lives.c (process_bb_lives): Likewise.
1434 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
1435 (get_hard_regs): Likewise.
1436 (do_remat): Likewise.
1437 * lra-spills.c (assign_spill_hard_regs): Likewise.
1438 * mode-switching.c (create_pre_exit): Likewise.
1439 * postreload.c (reload_combine_recognize_pattern): Likewise.
1440 * recog.c (peep2_find_free_register): Likewise.
1441 * regcprop.c (kill_value_regno): Likewise.
1442 (set_value_regno): Likewise.
1443 (copy_value): Likewise.
1444 (maybe_mode_change): Likewise.
1445 (find_oldest_value_reg): Likewise.
1446 (copyprop_hardreg_forward_1): Likewise.
1447 * regrename.c (check_new_reg_p): Likewise.
1448 (regrename_do_replace): Likewise.
1449 * reload.c (push_reload): Likewise.
1450 (combine_reloads): Likewise.
1451 (find_dummy_reload): Likewise.
1452 (operands_match_p): Likewise.
1453 (find_reloads): Likewise.
1454 (find_equiv_reg): Likewise.
1455 (reload_adjust_reg_for_mode): Likewise.
1456 * reload1.c (count_pseudo): Likewise.
1457 (count_spilled_pseudo): Likewise.
1458 (find_reg): Likewise.
1459 (clear_reload_reg_in_use): Likewise.
1460 (free_for_value_p): Likewise.
1461 (allocate_reload_reg): Likewise.
1462 (choose_reload_regs): Likewise.
1463 (reload_adjust_reg_for_temp): Likewise.
1464 (emit_reload_insns): Likewise.
1465 (delete_output_reload): Likewise.
1466 * rtlanal.c (subreg_get_info): Likewise.
1467 * sched-deps.c (sched_analyze_reg): Likewise.
1468 * sel-sched.c (init_regs_for_mode): Likewise.
1469 (mark_unavailable_hard_regs): Likewise.
1470 (choose_best_reg_1): Likewise.
1471 (verify_target_availability): Likewise.
1472 * valtrack.c (dead_debug_insert_temp): Likewise.
1473 * var-tracking.c (track_loc_p): Likewise.
1474 (emit_note_insn_var_location): Likewise.
1475 * varasm.c (make_decl_rtl): Likewise.
1476 * reginfo.c (choose_hard_reg_mode): Likewise.
1477 (init_reg_modes_target): Refer directly to
1478 this_target_regs->x_hard_regno_nregs.
1480 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1482 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
1483 instead of hard_regno_nregs.
1485 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1487 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
1488 end_hard_regno instead of hard_regno_nregs.
1489 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
1490 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
1491 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
1492 * ira-color.c (improve_allocation): Likewise.
1493 * lra-assigns.c (find_hard_regno_for_1): Likewise.
1494 * lra-lives.c (mark_regno_live): Likewise.
1495 (mark_regno_dead): Likewise.
1496 * lra-remat.c (operand_to_remat): Likewise.
1497 * lra.c (collect_non_operand_hard_regs): Likewise.
1498 * postreload.c (reload_combine_note_store): Likewise.
1499 (move2add_valid_value_p): Likewise.
1500 * reload.c (regno_clobbered_p): Likewise.
1502 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1504 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
1506 * config/v850/v850.c (v850_reorg): Likewise.
1507 * reload.c (refers_to_regno_for_reload_p): Likewise.
1508 (find_equiv_reg): Likewise.
1509 * reload1.c (reload_reg_reaches_end_p): Likewise.
1511 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1513 * caller-save.c (add_used_regs): Use REG_NREGS instead of
1515 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
1516 * config/arm/arm.c (output_move_neon): Likewise.
1517 (arm_attr_length_move_neon): Likewise.
1518 (neon_split_vcombine): Likewise.
1519 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
1520 (c6x_mark_reg_written): Likewise.
1521 (c6x_dwarf_register_span): Likewise.
1522 * config/i386/i386.c (ix86_save_reg): Likewise.
1523 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
1524 (rws_access_reg): Likewise.
1525 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
1526 * mode-switching.c (create_pre_exit): Likewise.
1527 * ree.c (combine_reaching_defs): Likewise.
1528 (add_removable_extension): Likewise.
1529 * regcprop.c (find_oldest_value_reg): Likewise.
1530 (copyprop_hardreg_forward_1): Likewise.
1531 * reload.c (reload_inner_reg_of_subreg): Likewise.
1532 (push_reload): Likewise.
1533 (combine_reloads): Likewise.
1534 (find_dummy_reload): Likewise.
1535 (reload_adjust_reg_for_mode): Likewise.
1536 * reload1.c (find_reload_regs): Likewise.
1537 (forget_old_reloads_1): Likewise.
1538 (reload_reg_free_for_value_p): Likewise.
1539 (reload_adjust_reg_for_temp): Likewise.
1540 (emit_reload_insns): Likewise.
1541 (delete_output_reload): Likewise.
1542 * sel-sched.c (choose_best_reg_1): Likewise.
1543 (choose_best_pseudo_reg): Likewise.
1545 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1546 Alan Hayward <alan.hayward@arm.com>
1547 David Sherwood <david.sherwood@arm.com>
1549 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
1550 * target.def (slow_unaligned_access): New hook.
1551 * targhooks.h (default_slow_unaligned_access): Declare.
1552 * targhooks.c (default_slow_unaligned_access): New function.
1553 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
1554 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
1555 * doc/tm.texi: Regenerate.
1556 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
1557 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
1558 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
1560 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
1561 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
1563 (rs6000_slow_unaligned_access): New function.
1564 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
1565 (expand_block_compare): Likewise.
1566 (expand_strn_compare): Likewise.
1567 (rs6000_rtx_costs): Likewise.
1568 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
1569 (riscv_slow_unaligned_access): Likewise.
1570 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
1571 (riscv_slow_unaligned_access_p): ...this and make static.
1572 (riscv_option_override): Update accordingly.
1573 (riscv_slow_unaligned_access): New function.
1574 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
1575 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
1576 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
1577 (rs6000_slow_unaligned_access): New function.
1578 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
1579 (rs6000_rtx_costs): Likewise.
1580 * config/rs6000/rs6000-string.c (expand_block_compare)
1581 (expand_strn_compare): Use targetm.slow_unaligned_access instead
1582 of SLOW_UNALIGNED_ACCESS.
1583 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
1584 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
1585 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
1586 of SLOW_UNALIGNED_ACCESS.
1587 * expmed.c (simple_mem_bitfield_p): Likewise.
1588 * expr.c (alignment_for_piecewise_move): Likewise.
1589 (emit_group_load_1): Likewise.
1590 (emit_group_store): Likewise.
1591 (copy_blkmode_from_reg): Likewise.
1592 (emit_push_insn): Likewise.
1593 (expand_assignment): Likewise.
1594 (store_field): Likewise.
1595 (expand_expr_real_1): Likewise.
1596 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
1597 * lra-constraints.c (simplify_operand_subreg): Likewise.
1598 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
1599 * gimple-ssa-store-merging.c: Likewise in block comment at start
1601 * tree-ssa-strlen.c: Include target.h.
1602 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
1603 of SLOW_UNALIGNED_ACCESS.
1604 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
1606 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1608 PR rtl-optimization/82185
1609 * expmed.c (emit_store_flag_int): Only test tem if it has been
1612 2017-09-12 Richard Biener <rguenther@suse.de>
1615 * match.pd ((FTYPE) N CMP CST): Fix typo.
1617 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
1619 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
1621 (mips_near_type_p): Add 'short_call' attribute as a synonym
1623 * doc/extend.texi (short_call): Document new function attribute.
1625 2017-09-12 Jakub Jelinek <jakub@redhat.com>
1628 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
1629 assertion check that in the condition.
1630 (get_atomic_generic_size): Likewise. Before testing if parameter
1631 has pointer type, if it has array type, call for C++
1632 default_conversion to perform array-to-pointer conversion.
1634 2017-09-12 Richard Biener <rguenther@suse.de>
1636 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
1637 for operations we cannot scalarize.
1639 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
1641 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
1642 vectors heap vectors. Clean up comments.
1643 Make visited_bbs a reference.
1644 (profitable_jump_thread_path): Make GC
1645 vectors heap vectors. Clean up comments.
1647 (convert_and_register_jump_thread_path): Make GC vectors heap
1649 (check_subpath_and_update_thread_path): Same. Clean up comments.
1650 Make visited_bbs a reference.
1651 (handle_phi): Abstract common code to to
1652 register_jump_thread_path_if_profitable.
1653 Rename VAR_BB to DEF_BB.
1655 Make GC vectors heap vectors.
1656 Make visited_bbs a reference.
1657 (handle_assignment): Same.
1658 (register_jump_thread_path_if_profitable): New.
1659 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
1661 Make GC vectors heap vectors. Clean up comments.
1662 Make visited_bbs a reference.
1663 (find_jump_threads_backwards): Make visited_bbs live in the stack.
1664 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
1667 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
1670 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
1671 words of E_DImode object are reachable by xtensa_uimm8x4 access.
1673 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
1675 Revert r251800 and r251799.
1677 2017-09-11 Martin Jambor <mjambor@suse.cz>
1680 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
1681 arguments in advance.
1682 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
1683 use it to find predecessor edges.
1684 (naive_outof_ssa): Collect vector of predecessors.
1686 2017-09-08 Jason Merrill <jason@redhat.com>
1688 PR c++/70029 - ICE with ref-qualifier and -flto
1689 * langhooks.h (struct lang_hooks_for_types): Add
1690 copy_lang_qualifiers.
1691 * attribs.c (build_type_attribute_qual_variant): Use it.
1692 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
1694 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
1695 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
1697 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
1700 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
1701 (*mulsi3_sp64): New instruction.
1702 (mulsi3): New expander.
1704 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
1706 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
1708 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1710 * sancov.c: Include memmodel.h.
1712 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
1715 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
1718 2017-09-07 Carl Love <cel@us.ibm.com>
1720 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
1721 the sldi instruction.
1723 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
1725 * sancov.c: Include tm_p.h.
1727 2017-09-07 Jakub Jelinek <jakub@redhat.com>
1730 * output.h (switch_to_other_text_partition): New declaration.
1731 * varasm.c (switch_to_other_text_partition): New function.
1732 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
1733 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
1734 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
1735 to the other text partition before emitting LCL label and switch back
1736 after emitting the word after it.
1738 2017-09-07 Richard Biener <rguenther@suse.de>
1740 * passes.def (pass_split_crit_edges): Remove instance before PRE.
1741 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
1742 critical edges here, after loop init.
1743 (pass_data_pre): Remove PROP_no_crit_edges flags.
1744 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
1745 for valueization of call args to avoid leaking VN_TOP.
1746 (visit_use): Assert we do not visit default defs.
1747 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
1748 Use error_mark_node to more easily detect leaking VN_TOP.
1749 All default-defs are varying, not VN_TOP. Mark them visited.
1750 (run_scc_vn): Make code match comment.
1752 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
1754 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
1755 OPTION_MASK_FLOAT128_KEYWORD.
1756 (POWERPC_MASKS): Likewise.
1757 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
1758 support for the -mfloat128-type option, and make -mfloat128
1759 default on PowerPC Linux systems. Define or undefine
1760 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
1761 Define __float128 to be __ieee128 if IEEE 128-bit support is
1762 enabled, or undefine it.
1763 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
1764 Delete defining __FLOAT128_TYPE__.
1765 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
1766 -mfloat128-type option and make -mfloat128 default on PowerPC
1768 (TARGET_FLOAT128_TYPE): Likewise.
1769 (-mfloat128-type): Likewise.
1770 * config/rs6000/rs6000.c (rs6000_option_override_internal):
1771 Delete the -mfloat128-type option and make -mfloat128 default on
1772 PowerPC Linux systems. Always use __ieee128 to be the keyword for
1773 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
1774 128-bit floating point is enabled. Change tests from using
1775 -mfloat128-type to -mfloat128.
1776 (rs6000_mangle_type): Use the correct mangling for the __float128
1777 type even if normal long double is restricted to 64-bits.
1778 (floatn_mode): Enable the _Float128 type by default on VSX Linux
1780 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
1781 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
1782 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
1784 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
1785 documentation for -mfloat128.
1787 2017-09-06 Olivier Hainque <hainque@adacore.com>
1789 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
1791 2017-09-06 Wish Wu <wishwu007@gmail.com>
1792 Jakub Jelinek <jakub@redhat.com>
1794 * asan.c (initialize_sanitizer_builtins): Add
1795 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
1796 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
1797 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
1798 BT_FN_VOID_UINT64_PTR variables.
1799 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
1800 (BT_FN_VOID_UINT16_UINT16): Likewise.
1801 (BT_FN_VOID_UINT32_UINT32): Likewise.
1802 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
1803 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
1804 (BT_FN_VOID_UINT64_PTR): Likewise.
1805 * common.opt (flag_sanitize_coverage): New variable.
1806 (fsanitize-coverage=trace-pc): Remove.
1807 (fsanitize-coverage=): Add.
1808 * flag-types.h (enum sanitize_coverage_code): New enum.
1809 * fold-const.c (fold_range_test): Disable non-short-circuit
1810 optimization if flag_sanitize_coverage.
1811 (fold_truth_andor): Likewise.
1812 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
1813 * opts.c (COVERAGE_SANITIZER_OPT): Define.
1814 (coverage_sanitizer_opts): New array.
1815 (get_closest_sanitizer_option): Add OPTS argument, handle also
1816 OPT_fsanitize_coverage_.
1817 (parse_sanitizer_options): Adjusted to also handle
1818 OPT_fsanitize_coverage_.
1819 (common_handle_option): Add OPT_fsanitize_coverage_.
1820 * sancov.c (instrument_comparison, instrument_switch): New function.
1821 (sancov_pass): Add trace-cmp support.
1822 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
1823 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
1824 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
1825 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
1826 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
1827 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
1828 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
1829 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
1830 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
1831 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
1833 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
1835 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
1836 error. Only quit immediately if parsing is complete.
1837 (BEGIN): Initialize fatal_err and parse_done.
1838 (begin fpu, end fpu): Check number of arguments.
1839 (begin arch, end arch): Likewise.
1840 (begin cpu, end cpu): Likewise.
1841 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
1842 (optalias): Likewise.
1844 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
1846 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
1847 * config/arm/arm-isa.h: Delete. Move definitions to ...
1848 * arm-cpus.in: ... here. Use new feature and fgroup values.
1849 * config/arm/arm.c (arm_option_override): Use lower case for feature
1851 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
1852 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
1853 * config/arm/parsecpu.awk (END): Add new command 'isa'.
1855 (print_isa_bits_for): New function.
1856 (gen_isa): New function.
1857 (gen_comm_data): Use print_isa_bits_for.
1858 (define feature): New keyword.
1859 (define fgroup): New keyword.
1860 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
1861 (arm-isa.h): Add rule to generate file.
1862 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
1863 case for feature bit names.
1865 2017-09-06 Richard Biener <rguenther@suse.de>
1867 * tree-ssa-pre.c (NECESSARY): Remove.
1868 (create_expression_by_pieces): Do not touch pass-local flags.
1869 (insert_into_preds_of_block): Likewise.
1870 (do_pre_regular_insertion): Likewise.
1871 (eliminate_insert): Likewise.
1872 (eliminate_dom_walker::before_dom_children): Likewise.
1873 (fini_eliminate): Do not look at inserted_exprs.
1874 (mark_operand_necessary): Remove.
1875 (remove_dead_inserted_code): Replace with simple work-list
1876 algorithm based on inserted_exprs and SSA uses.
1877 (pass_pre::execute): Re-order fini_eliminate and
1878 remove_dead_inserted_code.
1880 2017-09-06 Olivier Hainque <hainque@adacore.com>
1882 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
1883 for VxWorks 7. Adjust surrounding comments.
1885 2017-09-06 Richard Biener <rguenther@suse.de>
1887 * gimple-ssa-strength-reduction.c
1888 (find_candidates_dom_walker::before_dom_children): Also allow
1891 2017-09-06 Richard Biener <rguenther@suse.de>
1893 PR tree-optimization/82108
1894 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
1895 for gap in the non-permutation SLP case.
1897 2017-09-06 Martin Jambor <mjambor@suse.cz>
1899 PR tree-optimization/82078
1900 * tree-sra.c (sort_and_splice_var_accesses): Move call to
1901 add_access_to_work_queue...
1902 (build_accesses_from_assign): ...here.
1903 (propagate_all_subaccesses): Make sure racc is the group
1904 representative, if there is one.
1906 2017-09-06 Jakub Jelinek <jakub@redhat.com>
1909 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
1912 2017-09-06 Richard Biener <rguenther@suse.de>
1914 * gimple-ssa-strength-reduction.c
1915 (find_candidates_dom_walker::before_doom_children): Use a
1916 type and not a mode check.
1918 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
1921 * config/arm/predicates.md (arm_general_adddi_operand): Create new
1923 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
1925 2017-09-05 Jeff Law <law@redhat.com>
1927 PR tree-optimization/64910
1928 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
1929 cases where we have 3 or more operands.
1931 2017-09-05 Jakub Jelinek <jakub@redhat.com>
1934 * omp-low.c (lower_omp_for): Recompute tree invariant if
1935 gimple_omp_for_initial/final is ADDR_EXPR.
1938 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
1939 into gimple val before gimplification fo the COND_EXPR.
1941 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
1943 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
1944 REGION_COPY argument.
1945 (thread_through_all_blocks): Remove unused argument to
1946 duplicate_thread_path.
1948 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1949 Alan Hayward <alan.hayward@arm.com>
1950 David Sherwood <david.sherwood@arm.com>
1952 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
1953 Take a scalar_mode rather than a machine_mode.
1954 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
1955 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
1956 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
1957 (aarch64_gen_adjusted_ldpstp): Likewise.
1958 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
1960 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1961 Alan Hayward <alan.hayward@arm.com>
1962 David Sherwood <david.sherwood@arm.com>
1964 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
1965 Take a scalar_int_mode instead of a machine_mode.
1966 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
1967 (aarch64_output_scalar_simd_mov_immediate): Likewise.
1968 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
1969 (aarch64_simd_attr_length_rglist): Delete.
1970 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
1971 a scalar_int_mode instead of a machine_mode.
1972 (aarch64_add_offset): Likewise.
1973 (aarch64_internal_mov_immediate): Likewise
1974 (aarch64_add_constant_internal): Likewise.
1975 (aarch64_add_constant): Likewise.
1976 (aarch64_movw_imm): Likewise.
1977 (aarch64_rtx_arith_op_extract_p): Likewise.
1978 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
1979 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
1980 Remove assert that the mode isn't a vector.
1981 (aarch64_output_scalar_simd_mov_immediate): Likewise.
1982 (aarch64_expand_mov_immediate): Update calls after above changes.
1983 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
1984 (aarch64_and_bitmask_imm): Check for scalar integer modes.
1985 (aarch64_move_imm): Likewise.
1986 (aarch64_can_const_movi_rtx_p): Likewise.
1987 (aarch64_strip_extend): Likewise.
1988 (aarch64_extr_rtx_p): Likewise.
1989 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
1990 a CONST_INT when the mode parameter is VOIDmode.
1991 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
1993 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1995 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
1996 * stor-layout.c (bitwise_mode_for_mode): Likewise.
1997 (bitwise_type_for_mode): Update accordingly.
1999 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2001 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
2002 * stor-layout.c (mode_for_size_tree): Likewise.
2003 (mode_for_array): Update accordingly.
2004 (layout_decl): Likewise.
2005 (compute_record_mode): Likewise. Only set the mode once.
2007 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2009 * target.def (get_mask_mode): Change return type to opt_mode.
2011 * doc/tm.texi: Regenerate.
2012 * targhooks.h (default_get_mask_mode): Return an opt_mode.
2013 * targhooks.c (default_get_mask_mode): Likewise.
2014 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
2015 * optabs-query.c (can_vec_mask_load_store_p): Update use of
2016 targetm.get_mask_mode.
2017 * tree.c (build_truth_vector_type): Likewise.
2019 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2021 * machmode.h (mode_for_vector): Return an opt_mode.
2022 * stor-layout.c (mode_for_vector): Likewise.
2023 (mode_for_int_vector): Update accordingly.
2024 (layout_type): Likewise.
2025 * config/i386/i386.c (emit_memmov): Likewise.
2026 (ix86_expand_set_or_movmem): Likewise.
2027 (ix86_expand_vector_init): Likewise.
2028 (ix86_get_mask_mode): Likewise.
2029 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
2031 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
2032 * expmed.c (extract_bit_field_1): Likewise.
2033 * expr.c (expand_expr_real_2): Likewise.
2034 * optabs-query.c (can_vec_perm_p): Likewise.
2035 (can_vec_mask_load_store_p): Likewise.
2036 * optabs.c (expand_vec_perm): Likewise.
2037 * targhooks.c (default_get_mask_mode): Likewise.
2038 * tree-vect-stmts.c (vectorizable_store): Likewise.
2039 (vectorizable_load): Likewise.
2040 (get_vectype_for_scalar_type_and_size): Likewise.
2042 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2044 * machmode.h (mode_for_int_vector): New function.
2045 * stor-layout.c (mode_for_int_vector): Likewise.
2046 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
2047 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
2048 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
2049 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
2050 (s390_expand_vcond): Likewise.
2052 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2054 * machmode.h (opt_machine_mode): New type.
2055 (opt_mode<T>): Allow construction from anything that can be
2057 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
2058 (mode_for_size): Return an opt_machine_mode.
2059 * stor-layout.c (mode_for_size): Likewise.
2060 (mode_for_size_tree): Update call accordingly.
2061 (bitwise_mode_for_mode): Likewise.
2062 (make_fract_type): Likewise.
2063 (make_accum_type): Likewise.
2064 * caller-save.c (replace_reg_with_saved_mem): Update call
2066 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
2067 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
2068 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
2069 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
2070 * expmed.c (extract_bit_field_1): Likewise.
2071 * reload.c (get_secondary_mem): Likewise.
2072 * varasm.c (assemble_integer): Likewise.
2073 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
2076 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2078 * machmode.h (decimal_float_mode_for_size): New function.
2079 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
2080 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
2081 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
2082 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
2083 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
2084 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
2086 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2088 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
2089 (get_builtin_sync_mode): Likewise.
2090 (expand_ifn_atomic_compare_exchange): Likewise.
2091 (expand_builtin_atomic_clear): Likewise.
2092 (expand_builtin_atomic_test_and_set): Likewise.
2093 (fold_builtin_atomic_always_lock_free): Likewise.
2094 * calls.c (compute_argument_addresses): Likewise.
2095 (emit_library_call_value_1): Likewise.
2096 (store_one_arg): Likewise.
2097 * combine.c (combine_instructions): Likewise.
2098 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
2099 * config/arm/arm.c (arm_function_value): Likewise.
2100 (aapcs_allocate_return_reg): Likewise.
2101 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
2102 * config/i386/i386.c (construct_container): Likewise.
2103 (ix86_gimplify_va_arg): Likewise.
2104 (ix86_expand_sse_cmp): Likewise.
2105 (emit_memmov): Likewise.
2106 (emit_memset): Likewise.
2107 (expand_small_movmem_or_setmem): Likewise.
2108 (ix86_expand_pextr): Likewise.
2109 (ix86_expand_pinsr): Likewise.
2110 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
2111 * config/microblaze/microblaze.c (microblaze_block_move_straight):
2113 * config/mips/mips.c (mips_function_value_1) Likewise.
2114 (mips_block_move_straight): Likewise.
2115 (mips_expand_ins_as_unaligned_store): Likewise.
2116 * config/powerpcspe/powerpcspe.c
2117 (rs6000_darwin64_record_arg_advance_flush): Likewise.
2118 (rs6000_darwin64_record_arg_flush): Likewise.
2119 * config/rs6000/rs6000.c
2120 (rs6000_darwin64_record_arg_advance_flush): Likewise.
2121 (rs6000_darwin64_record_arg_flush): Likewise.
2122 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
2123 (sparc_function_value_1): Likewise.
2124 * config/spu/spu.c (adjust_operand): Likewise.
2125 (spu_emit_branch_or_set): Likewise.
2126 (arith_immediate_p): Likewise.
2127 * emit-rtl.c (gen_lowpart_common): Likewise.
2128 * expr.c (expand_expr_real_1): Likewise.
2129 * function.c (assign_parm_setup_block): Likewise.
2130 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
2131 * reload1.c (alter_reg): Likewise.
2132 * stor-layout.c (mode_for_vector): Likewise.
2133 (layout_type): Likewise.
2135 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
2137 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
2138 (spu_convert_move): Likewise.
2139 * lower-subreg.c (resolve_simple_move): Likewise.
2141 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2144 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
2145 define_insn to a define_expand.
2146 (altivec_vsum2sws_direct): New define_insn.
2147 (altivec_vsumsws): Convert from a define_insn to a define_expand.
2149 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
2151 * config/arm/arm.c (arm_option_params_internal): Improve setting of
2154 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
2159 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
2160 realign stack if stack alignment needed is less than incoming
2163 2017-09-05 Marek Polacek <polacek@redhat.com>
2166 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
2169 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
2171 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
2173 2017-09-05 Richard Biener <rguenther@suse.de>
2175 PR tree-optimization/82084
2176 * fold-const.c (can_native_encode_string_p): Handle wide characters.
2178 2017-09-05 Richard Biener <rguenther@suse.de>
2180 PR tree-optimization/82102
2181 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
2183 2017-09-05 Martin Liska <mliska@suse.cz>
2185 PR tree-optimization/82032
2186 * tree-cfg.c (generate_range_test): New function.
2187 * tree-cfg.h (generate_range_test): Declared here.
2188 * tree-cfgcleanup.c (convert_single_case_switch): New function.
2189 (cleanup_control_expr_graph): Use it.
2190 * tree-switch-conversion.c (try_switch_expansion): Remove
2192 (emit_case_nodes): Use generate_range_test.
2194 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
2197 * config/i386/i386.md (*<btsc><mode>_mask): Add
2198 TARGET_USE_BT to insn constraint.
2199 (*btr<mode>_mask): Ditto.
2201 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
2203 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
2204 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
2206 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2209 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
2210 TARGET_NEON and TARGET_IWMMXT.
2211 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
2212 TARGET_NEON and TARGET_IWMMXT.
2213 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
2215 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
2217 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
2218 (ix86_rewrite_tls_address): Ditto.
2219 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
2220 (ix86_rewrite_tls_address_1): Ditto.
2221 (ix86_rewrite_tls_address): Ditto.
2222 * config/i386/predicates.md (tls_address_pattern): New predicate.
2223 * config/i386/i386.md (TLS address splitter): New splitter.
2225 2017-09-04 Richard Biener <rguenther@suse.de>
2227 PR tree-optimization/82084
2228 * fold-const.h (can_native_encode_string_p): Declare.
2229 * fold-const.c (can_native_encode_string_p): Factor out from ...
2230 (native_encode_string): ... here.
2231 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
2232 vectorizing stores from constants we later cannot handle.
2234 2017-09-04 Marek Polacek <polacek@redhat.com>
2237 * doc/invoke.texi: Update -Wtautological-compare documentation.
2239 2017-09-04 Jeff Law <law@redhat.com>
2241 PR tree-optimization/64910
2242 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
2243 swap the first and last operand if the last is a constant.
2245 2017-09-04 Marek Polacek <polacek@redhat.com>
2248 * convert.c (do_narrow): When sanitizing signed integer overflows,
2249 bail out for signed types.
2250 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
2252 2017-09-04 Richard Biener <rguenther@suse.de>
2254 PR tree-optimization/82060
2255 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
2256 Move devirtualization after stmt folding and before EH/AB/noreturn
2257 cleanup to get the stmt refs canonicalized. Use a bool instead
2258 of gimple_modified_p since that doesn't work for NOPs. Schedule
2259 NOPs generated by folding for removal.
2261 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2262 Alan Hayward <alan.hayward@arm.com>
2263 David Sherwood <david.sherwood@arm.com>
2265 * coretypes.h (pad_direction): New enum.
2266 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
2267 (FUNCTION_ARG_PADDING): Likewise.
2268 * target.def (function_arg_padding): New hook.
2269 * targhooks.h (default_function_arg_padding): Declare.
2270 * targhooks.c (default_function_arg_padding): New function.
2271 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
2272 (TARGET_FUNCTION_ARG_PADDING): ...this.
2273 * doc/tm.texi: Regenerate.
2274 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
2275 instead of direction.
2276 (compute_argument_addresses): Likewise.
2277 (load_register_parameters): Likewise.
2278 (emit_library_call_value_1): Likewise.
2279 (store_one_arg): Use targetm.calls.function_arg_padding instead
2280 of FUNCTION_ARG_PADDING.
2281 (must_pass_in_stack_var_size_or_pad): Likewise.
2282 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
2283 (emit_group_store): Likewise.
2284 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
2285 instead of FUNCTION_ARG_PADDING.
2286 (emit_push_insn): Likewise, and propagate enum change throughout
2288 * function.h (direction): Delete.
2289 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
2291 * function.c (assign_parm_find_stack_rtl): Likewise.
2292 (assign_parm_setup_block_p): Likewise.
2293 (assign_parm_setup_block): Likewise.
2294 (gimplify_parameters): Likewise.
2295 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
2296 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
2298 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
2299 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
2300 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
2301 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
2302 (aarch64_function_arg_padding): ...this new function.
2303 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
2304 (TARGET_FUNCTION_ARG_PADDING): Redefine.
2305 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
2306 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
2307 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
2308 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2309 (arm_pad_arg_upward): Replace with...
2310 (arm_function_arg_padding): ...this new function.
2311 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
2313 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
2314 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
2315 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2316 (ia64_hpux_function_arg_padding): Replace with...
2317 (ia64_function_arg_padding): ...this new function. Use pad_direction
2318 instead of direction. Check for TARGET_HPUX.
2319 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
2320 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2321 (iq2000_function_arg_padding): New function.
2322 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
2323 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
2324 (mips_function_arg_padding): ...this new function.
2325 (mips_pad_reg_upward): Update accordingly.
2326 (TARGET_FUNCTION_ARG_PADDING): Redefine.
2327 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
2328 targetm.calls.function_arg_padding.
2329 (FUNCTION_ARG_PADDING): Delete.
2330 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
2331 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
2332 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
2333 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
2334 (nios2_block_reg_padding): Return pad_direction instead of direction.
2335 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
2336 instead of direction.
2337 (nios2_function_arg_padding): Likewise. Make static.
2338 (TARGET_FUNCTION_ARG_PADDING): Redefine.
2339 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
2340 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
2341 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
2342 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2343 (pa_function_arg_padding): Make static. Return pad_direction instead
2345 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
2346 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
2347 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
2348 instead of direction. Use targetm.calls.function_arg_padding.
2349 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
2350 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
2351 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
2352 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
2353 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
2355 (function_arg_padding): Rename to...
2356 (rs6000_function_arg_padding): ...this. Make static. Return
2357 pad_direction instead of direction.
2358 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
2359 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
2360 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
2361 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
2362 instead of direction. Use targetm.calls.function_arg_padding.
2363 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
2364 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
2365 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
2366 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
2367 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2368 (function_arg_padding): Rename to...
2369 (rs6000_function_arg_padding): ...this. Make static. Return
2370 pad_direction instead of direction.
2371 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
2372 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
2373 * config/s390/s390.c (s390_function_arg_padding): New function.
2374 (TARGET_FUNCTION_ARG_PADDING): Redefine.
2375 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
2376 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
2377 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2378 (function_arg_padding): Rename to...
2379 (sparc_function_arg_padding): ...this. Make static. Return
2380 pad_direction instead of direction.
2381 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
2382 * config/spu/spu.c (spu_function_arg_padding): New function.
2383 (TARGET_FUNCTION_ARG_PADDING): Redefine.
2384 * system.h (FUNCTION_ARG_PADDING): Poison.
2386 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2387 Alan Hayward <alan.hayward@arm.com>
2388 David Sherwood <david.sherwood@arm.com>
2390 * target.def (modes_tieable_p): New hook.
2391 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
2392 (TARGET_MODES_TIEABLE_P): ...this.
2393 * doc/tm.texi.in: Regenerate.
2394 * hooks.h (hook_bool_mode_mode_true): Declare.
2395 * hooks.c (hook_bool_mode_mode_true): New function.
2396 * combine.c (subst): Use targetm.modes_tieable_p instead of
2398 * dse.c (find_shift_sequence): Likewise.
2399 * expmed.c (extract_low_bits): Likewise.
2400 * lower-subreg.c: Include target.h.
2401 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
2403 * rtlanal.c (rtx_cost): Likewise.
2404 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
2405 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
2406 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
2407 (TARGET_MODES_TIEABLE_P): Redefine.
2408 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
2409 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
2410 (TARGET_MODES_TIEABLE_P): Redefine.
2411 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
2412 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
2413 (arc_modes_tieable_p): New function.
2414 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
2415 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
2416 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
2417 (arm_modes_tieable_p): Make static.
2418 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
2419 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
2420 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
2421 (TARGET_MODES_TIEABLE_P): Redefine.
2422 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
2423 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
2424 (TARGET_MODES_TIEABLE_P): Redefine.
2425 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
2426 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
2427 (cr16_modes_tieable_p): New function.
2428 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
2429 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
2430 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
2431 (TRULY_NOOP_TRUNCATION): Update comment.
2432 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
2433 (TRULY_NOOP_TRUNCATION): Update comment.
2434 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
2435 (frv_modes_tieable_p): New function.
2436 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
2437 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
2438 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
2439 (TARGET_MODES_TIEABLE_P): Redefine.
2440 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
2441 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
2442 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
2443 (TARGET_MODES_TIEABLE_P): Redefine.
2444 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
2445 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
2446 (ia64_modes_tieable_p): New function.
2447 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
2448 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
2449 (iq2000_modes_tieable_p): New function.
2450 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
2451 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
2452 (lm32_modes_tieable_p): New function.
2453 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
2454 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
2455 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
2456 (TARGET_MODES_TIEABLE_P): Redefine.
2457 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
2458 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
2459 (m32r_modes_tieable_p): New function.
2460 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
2461 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
2462 (m68k_modes_tieable_p): New function.
2463 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
2464 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
2465 (mcore_modes_tieable_p): New function.
2466 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
2467 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
2469 (TARGET_MODES_TIEABLE_P): Redefine.
2470 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
2471 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
2472 * config/mips/mips.c (mips_modes_tieable_p): Make static.
2473 (TARGET_MODES_TIEABLE_P): Redefine.
2474 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
2475 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
2476 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
2477 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
2478 (mn10300_modes_tieable_p): ...this and make static.
2479 (TARGET_MODES_TIEABLE_P): Redefine.
2480 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
2481 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
2482 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
2483 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
2484 (msp430_modes_tieable_p): Make static.
2485 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
2486 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
2487 (TARGET_MODES_TIEABLE_P): Redefine.
2488 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
2489 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
2490 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
2491 (TARGET_MODES_TIEABLE_P): Redefine.
2492 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
2493 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
2494 * config/pa/pa.c (pa_modes_tieable_p): Make static.
2495 (TARGET_MODES_TIEABLE_P): Redefine.
2496 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
2497 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
2498 (pdp11_modes_tieable_p): New function.
2499 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
2500 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
2501 (rs6000_modes_tieable_p): New function.
2502 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
2503 * config/powerpcspe/powerpcspe.md: Update comment.
2504 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
2505 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
2506 (TARGET_MODES_TIEABLE_P): Redefine.
2507 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
2508 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
2509 (rl78_modes_tieable_p): New function.
2510 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
2511 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
2512 (rs6000_modes_tieable_p): New function.
2513 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
2514 * config/rs6000/rs6000.md: Update comment.
2515 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
2516 * config/rx/rx.c (rx_modes_tieable_p): New function.
2517 (TARGET_MODES_TIEABLE_P): Redefine.
2518 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
2519 * config/s390/s390.c (s390_modes_tieable_p): New function.
2520 (TARGET_MODES_TIEABLE_P): Redefine.
2521 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
2522 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
2523 (sh_modes_tieable_p): New function.
2524 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
2525 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
2526 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
2527 (sparc_modes_tieable_p): Make static.
2528 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
2529 * config/spu/spu.c (spu_modes_tieable_p): New function.
2530 (TARGET_MODES_TIEABLE_P): Redefine.
2531 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
2532 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
2533 (TARGET_MODES_TIEABLE_P): Redefine.
2534 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
2535 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
2536 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
2537 * config/v850/v850.c (v850_modes_tieable_p): New function.
2538 (TARGET_MODES_TIEABLE_P): Redefine.
2539 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
2540 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
2541 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
2542 (visium_modes_tieable_p): New function.
2543 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
2544 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
2545 (xtensa_modes_tieable_p): New function.
2546 * system.h (MODES_TIEABLE_P): Poison.
2548 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2549 Alan Hayward <alan.hayward@arm.com>
2550 David Sherwood <david.sherwood@arm.com>
2552 * target.def (hard_regno_mode_ok): New hook.
2553 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
2554 (TARGET_HARD_REGNO_MODE_OK): ...this.
2555 * doc/tm.texi.in: Regenerate.
2556 * hooks.h (hook_bool_uint_mode_true): Declare.
2557 * hooks.c (hook_bool_uint_mode_true): New function.
2558 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
2560 * genpreds.c (write_insn_preds_c): Add an include of target.h.
2561 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
2562 instead of HARD_REGNO_MODE_OK.
2563 * caller-save.c: Include target.h.
2564 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
2566 * combine.c (can_combine_p): Likewise.
2567 (combinable_i3pat): Likewise.
2568 (can_change_dest_mode): Likewise.
2569 * expr.c (init_expr_target): Likewise.
2570 (convert_move): Likewise.
2571 (convert_modes): Likewise.
2572 * ira.c (setup_prohibited_class_mode_regs): Likewise.
2573 (setup_prohibited_mode_move_regs): Likewise.
2574 * ira.h (target_ira): Likewise.
2575 * lra-assigns.c (find_hard_regno_for_1): Likewise.
2576 * lra-constraints.c (process_alt_operands): Likewise.
2577 (split_reg): Likewise.
2578 * recog.c (peep2_find_free_register): Likewise.
2579 * ree.c (combine_reaching_defs): Likewise.
2580 * regcprop.c (maybe_mode_change): Likewise.
2581 * reginfo.c (init_reg_sets_1): Likewise.
2582 (choose_hard_reg_mode): Likewise.
2583 (simplifiable_subregs): Likewise.
2584 * regrename.c (check_new_reg_p): Likewise.
2585 * reload.c (find_valid_class): Likewise.
2586 (find_valid_class_1): Likewise.
2587 (reload_inner_reg_of_subreg): Likewise.
2588 (push_reload): Likewise.
2589 (combine_reloads): Likewise.
2590 (find_dummy_reload): Likewise.
2591 (find_reloads): Likewise.
2592 * reload1.c (find_reg): Likewise.
2593 (set_reload_reg): Likewise.
2594 (allocate_reload_reg): Likewise.
2595 (choose_reload_regs): Likewise.
2596 (reload_adjust_reg_for_temp): Likewise.
2597 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
2598 (simplify_subreg_regno): Likewise.
2599 * sel-sched.c (init_regs_for_mode): Likewise.
2600 * varasm.c (make_decl_rtl): Likewise.
2601 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
2602 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
2604 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
2605 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
2606 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2607 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
2608 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
2609 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2610 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
2611 (arc_mode_class): Delete.
2612 (HARD_REGNO_MODE_OK): Delete.
2613 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2614 (arc_hard_regno_mode_ok): Rename old array to...
2615 (arc_hard_regno_mode_ok_modes): ...this.
2616 (arc_conditional_register_usage): Update accordingly.
2617 (arc_mode_class): Make static.
2618 (arc_hard_regno_mode_ok): New function.
2619 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
2620 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
2621 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2622 (arm_hard_regno_mode_ok): Make static.
2623 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
2625 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
2626 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
2627 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
2629 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2630 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
2631 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
2632 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
2633 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
2634 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2635 * config/bfin/predicates.md (valid_reg_operand): Use
2636 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
2637 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
2638 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
2639 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2640 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
2641 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
2642 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2643 (cr16_hard_regno_mode_ok): Make static and return a bool.
2644 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
2645 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2646 (cris_hard_regno_mode_ok): New function.
2647 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
2648 (epiphany_mode_class): Delete.
2649 (HARD_REGNO_MODE_OK): Delete.
2650 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
2651 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2652 (hard_regno_mode_ok): Rename to...
2653 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
2654 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
2655 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
2657 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
2658 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
2659 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2660 (frv_hard_regno_mode_ok): Make static and return a bool.
2661 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
2663 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
2664 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
2665 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
2666 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
2668 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2669 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
2670 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
2671 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
2673 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2674 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
2675 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2676 (ia64_hard_regno_mode_ok): New function.
2677 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
2678 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2679 (iq2000_hard_regno_mode_ok): New function.
2680 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
2681 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2682 (lm32_hard_regno_mode_ok): New function.
2683 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
2684 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
2685 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
2686 instead of HARD_REGNO_MODE_OK.
2687 (m32c_hard_regno_ok): Rename to...
2688 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
2689 (m32c_cannot_change_mode_class): Update accordingly.
2690 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2691 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
2692 (m32r_mode_class): Delete.
2693 (HARD_REGNO_MODE_OK): Delete.
2694 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2695 (m32r_hard_regno_mode_ok): Rename to...
2696 (m32r_hard_regno_modes): ...this.
2697 (m32r_mode_class): Make static.
2698 (m32r_hard_regno_mode_ok): New function.
2699 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
2700 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
2701 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2702 (m68k_hard_regno_mode_ok): Make static.
2703 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
2704 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2705 (mcore_hard_regno_mode_ok): New function.
2706 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
2707 (HARD_REGNO_MODE_OK): Delete.
2708 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
2710 (microblaze_hard_regno_mode_ok_p): ...this and make static.
2711 (microblaze_hard_regno_mode_ok): New function.
2712 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2713 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
2714 (mips_hard_regno_mode_ok): Delete.
2715 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
2716 (mips_hard_regno_mode_ok_p): ...this and make static.
2717 (mips_hard_regno_mode_ok_p): Rename to...
2718 (mips_hard_regno_mode_ok_uncached): ...this.
2719 (mips_hard_regno_mode_ok): New function.
2720 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
2721 of HARD_REGNO_MODE_OK.
2722 (mips_option_override): Update after above name changes.
2723 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2724 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
2725 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
2726 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
2727 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
2728 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2729 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
2730 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
2731 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
2732 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2733 (msp430_hard_regno_mode_ok): Make static and return a bool.
2734 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
2735 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
2736 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
2738 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2739 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
2740 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
2741 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
2742 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
2743 (PA_HARD_REGNO_MODE_OK): ...this
2744 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
2745 (PA_HARD_REGNO_MODE_OK): ...this.
2746 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2747 (pa_hard_regno_mode_ok): New function.
2748 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
2749 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2750 (pdp11_hard_regno_mode_ok): New function.
2751 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
2752 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
2754 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
2756 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2757 (rs6000_hard_regno_mode_ok): Rename to...
2758 (rs6000_hard_regno_mode_ok_uncached): ...this.
2759 (rs6000_init_hard_regno_mode_ok): Update accordingly.
2760 (rs6000_hard_regno_mode_ok): New function.
2761 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
2762 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
2763 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
2764 (riscv_hard_regno_mode_ok): ...this and make static.
2765 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2766 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
2767 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
2768 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2769 (rl78_hard_regno_mode_ok): Make static and return bool.
2770 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
2771 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
2773 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
2774 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2775 (rs6000_hard_regno_mode_ok): Rename to...
2776 (rs6000_hard_regno_mode_ok_uncached): ...this.
2777 (rs6000_init_hard_regno_mode_ok): Update accordingly.
2778 (rs6000_hard_regno_mode_ok): New function.
2779 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
2780 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
2781 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2782 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
2783 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
2784 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
2785 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2786 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
2787 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
2788 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2789 (sh_hard_regno_mode_ok): Make static.
2790 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
2791 instead of HARD_REGNO_MODE_OK.
2792 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
2793 (sparc_mode_class): Delete.
2794 (HARD_REGNO_MODE_OK): Delete.
2795 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2796 (hard_regno_mode_classes): Make static.
2797 (sparc_mode_class): Likewise.
2798 (sparc_hard_regno_mode_ok): New function.
2799 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
2800 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
2801 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
2803 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2804 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
2805 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
2806 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
2807 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
2808 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2809 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
2810 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
2811 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2812 (visium_hard_regno_mode_ok): New function.
2813 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
2814 instead of HARD_REGNO_MODE_OK.
2815 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
2816 (HARD_REGNO_MODE_OK): Delete.
2817 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
2818 (xtensa_hard_regno_mode_ok_p): ...this and make static.
2819 (xtensa_option_override): Update accordingly.
2820 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2821 (xtensa_hard_regno_mode_ok): New function.
2822 * system.h (HARD_REGNO_MODE_OK): Poison.
2824 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2825 Alan Hayward <alan.hayward@arm.com>
2826 David Sherwood <david.sherwood@arm.com>
2828 * target.def (hard_regno_call_part_clobbered): New hook.
2829 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
2830 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
2831 * doc/tm.texi: Regenerate.
2832 * hooks.h (hook_bool_uint_mode_false): Declare.
2833 * hooks.c (hook_bool_uint_mode_false): New function.
2834 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2835 * cselib.c (cselib_process_insn): Use
2836 targetm.hard_regno_call_part_clobbered instead of
2837 HARD_REGNO_CALL_PART_CLOBBERED.
2838 * ira-conflicts.c (ira_build_conflicts): Likewise.
2839 * ira-costs.c (ira_tune_allocno_costs): Likewise.
2840 * lra-constraints.c (need_for_call_save_p): Likewise.
2841 * lra-lives.c: Include target.h.
2842 (check_pseudos_live_through_calls): Use
2843 targetm.hard_regno_call_part_clobbered instead of
2844 HARD_REGNO_CALL_PART_CLOBBERED.
2845 * regcprop.c: Include target.h.
2846 (copyprop_hardreg_forward_1): Use
2847 targetm.hard_regno_call_part_clobbered instead of
2848 HARD_REGNO_CALL_PART_CLOBBERED.
2849 * reginfo.c (choose_hard_reg_mode): Likewise.
2850 * regrename.c (check_new_reg_p): Likewise.
2851 * reload.c (find_equiv_reg): Likewise.
2852 * reload1.c (emit_reload_insns): Likewise.
2853 * sched-deps.c (deps_analyze_insn): Likewise.
2854 * sel-sched.c (init_regs_for_mode): Likewise.
2855 (mark_unavailable_hard_regs): Likewise.
2856 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
2857 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2858 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
2860 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2861 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2862 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
2864 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
2866 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2867 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2868 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
2870 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2871 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2872 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
2874 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2875 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
2877 * config/powerpcspe/powerpcspe.c
2878 (rs6000_hard_regno_call_part_clobbered): New function.
2879 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2880 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2881 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
2883 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2884 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2885 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
2887 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2888 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2889 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
2891 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2892 Alan Hayward <alan.hayward@arm.com>
2893 David Sherwood <david.sherwood@arm.com>
2895 * rtl.h (subreg_memory_offset): Declare.
2896 * emit-rtl.c (subreg_memory_offset): New function.
2897 * expmed.c (store_bit_field_1): Use it.
2898 * expr.c (undefined_operand_subword_p): Likewise.
2899 * simplify-rtx.c (simplify_subreg): Likewise.
2901 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
2903 PR rtl-optimization/57448
2906 * optabs.c (expand_atomic_load): Place compiler memory barriers if
2907 using atomic_load pattern.
2908 (expand_atomic_store): Likewise.
2910 2017-09-04 Jakub Jelinek <jakub@redhat.com>
2913 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
2914 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
2915 handling. Use replace_call_with_value with NULL instead of
2916 gsi_replace, unlink_stmt_vdef and release_defs.
2918 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
2921 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
2923 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2926 * rtl.h (emit_library_call_value_1): Declare.
2927 (emit_library_call): Replace declaration with a series of overloads.
2928 Remove the parameter count argument.
2929 (emit_library_call_value): Likewise.
2930 * calls.c (emit_library_call_value_1): Make global. Replace varargs
2931 with an "rtx_mode_t *".
2932 (emit_library_call_value): Delete.
2933 (emit_library_call): Likewise.
2934 * asan.c (asan_emit_stack_protection): Update calls accordingly.
2935 (asan_emit_allocas_unpoison): Likewise.
2936 * builtins.c (expand_builtin_powi): Likewise.
2937 (expand_asan_emit_allocas_unpoison): Likewise.
2938 * cfgexpand.c (expand_main_function): Likewise.
2939 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
2940 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
2941 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
2942 * config/arm/arm.c (arm_trampoline_init): Likewise.
2943 (arm_call_tls_get_addr): Likewise.
2944 (arm_expand_divmod_libfunc): Likewise.
2945 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
2946 (smulsi3_highpart): Likewise.
2947 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
2948 (c6x_expand_compare): Likewise.
2949 (c6x_expand_movmem): Likewise.
2950 * config/frv/frv.c (frv_trampoline_init): Likewise.
2951 * config/i386/i386.c (ix86_trampoline_init): Likewise.
2952 (ix86_expand_divmod_libfunc): Likewise.
2953 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
2954 (ia64_expand_compare): Likewise.
2955 (ia64_profile_hook): Likewise.
2956 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
2957 (nonlocal_goto): Likewise.
2958 (restore_stack_nonlocal): Likewise.
2959 * config/m32r/m32r.c (block_move_call): Likewise.
2960 (m32r_trampoline_init): Likewise.
2961 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
2962 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
2963 (m68k_call_m68k_read_tp): Likewise.
2964 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
2965 (microblaze_expand_divide): Likewise.
2966 * config/mips/mips.h (mips_args): Likewise.
2967 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
2968 (MIPS_ICACHE_SYNC): Likewise.
2969 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
2970 (nios2_trampoline_init): Likewise.
2971 * config/pa/pa.c (hppa_tls_call): Likewise.
2972 (pa_trampoline_init): Likewise.
2973 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
2974 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
2975 (expand_strn_compare): Likewise.
2976 (rs6000_generate_compare): Likewise.
2977 (rs6000_expand_float128_convert): Likewise.
2978 (output_profile_hook): Likewise.
2979 (rs6000_trampoline_init): Likewise.
2980 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
2981 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
2982 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
2983 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
2984 (rs6000_generate_compare): Likewise.
2985 (rs6000_expand_float128_convert): Likewise.
2986 (output_profile_hook): Likewise.
2987 (rs6000_trampoline_init): Likewise.
2988 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
2989 * config/sh/sh.c (sh_trampoline_init): Likewise.
2990 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
2991 (sparc_emit_float_lib_cmp): Likewise.
2992 (sparc32_initialize_trampoline): Likewise.
2993 (sparc64_initialize_trampoline): Likewise.
2994 (sparc_profile_hook): Likewise.
2995 * config/spu/spu.c (ea_load_store): Likewise.
2996 * config/spu/spu.md (floatunssidf2): Likewise.
2997 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
2998 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
2999 * config/visium/visium.c (expand_block_move_4): Likewise.
3000 (expand_block_move_2): Likewise.
3001 (expand_block_move_1): Likewise.
3002 (expand_block_set_4): Likewise.
3003 (expand_block_set_2): Likewise.
3004 (expand_block_set_1): Likewise.
3005 (visium_trampoline_init): Likewise.
3006 (visium_profile_hook): Likewise.
3007 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
3008 (xtensa_setup_frame_addresses): Likewise.
3009 (xtensa_trampoline_init): Likewise.
3010 * except.c (sjlj_emit_function_enter): Likewise.
3011 (sjlj_emit_function_exit): Likewise.
3012 * explow.c (allocate_dynamic_stack_space): Likewise.
3013 (probe_stack_range): Likewise.
3014 * expr.c (convert_mode_scalar): Likewise.
3015 * optabs.c (expand_binop): Likewise.
3016 (expand_twoval_binop_libfunc): Likewise.
3017 (expand_unop): Likewise.
3018 (prepare_cmp_insn): Likewise.
3019 (prepare_float_lib_cmp): Likewise.
3020 (expand_float): Likewise.
3021 (expand_fix): Likewise.
3022 (expand_fixed_convert): Likewise.
3023 (maybe_emit_sync_lock_test_and_set): Likewise.
3024 (expand_atomic_compare_and_swap): Likewise.
3025 (expand_mem_thread_fence): Likewise.
3026 (expand_atomic_fetch_op): Likewise.
3028 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
3030 * doc/generic.texi (OpenACC): Adjust URL.
3031 * doc/invoke.texi (C Dialect Options): Ditto.
3033 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
3035 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
3036 predicate for operand 1. Add (m,<S>) constraint.
3037 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
3038 Prevent memory operand 1 with register operand 2.
3040 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
3042 PR rtl-optimization/82024
3043 * combine.c (try_combine): If the combination result is a PARALLEL,
3044 and we only need to retain the SET in there that would be placed
3045 at I2, check that we can place that at I3 instead, before doing so.
3047 2017-09-01 Jakub Jelinek <jakub@redhat.com>
3050 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
3052 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
3053 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
3056 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
3057 Jeff Law <law@redhat.com>
3059 * varasm.c (bss_initializer_p): Do not put constants into .bss
3060 (categorize_decl_for_section): Handle bss_initializer_p returning
3061 false when DECL_INITIAL is NULL.
3063 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3066 * config/s390/s390.c (s390_can_inline_p): New function.
3068 2017-09-01 Jeff Law <law@redhat.com>
3070 PR tree-optimization/82052
3071 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
3072 Always initialize the returned slot after a hash table miss
3073 when INSERT is true.
3075 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
3077 * config/s390/s390.md (mem_signal_fence): Remove.
3078 * doc/md.texi (mem_signal_fence): Remove.
3079 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
3081 * target-insns.def (mem_signal_fence): Remove.
3083 2017-09-01 Jakub Jelinek <jakub@redhat.com>
3086 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
3089 * asan.c (create_odr_indicator): Strip name encoding from assembler
3090 name before appending it after __odr_asan_.
3092 2017-09-01 Martin Liska <mliska@suse.cz>
3094 PR tree-optimization/82059
3095 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
3096 frequency only when an edge is redirected.
3098 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
3100 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
3101 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
3102 (arc_conditional_register_usage): Remove ARC600 lp_count
3104 (arc_file_start): Emit Tag_ARC_CPU_variation.
3105 (arc_can_use_doloop_p): New conditions to use ZOLs.
3106 (hwloop_fail): New function.
3107 (hwloop_optimize): Likewise.
3108 (hwloop_pattern_reg): Likewise.
3109 (arc_doloop_hooks): New struct, to be used with reorg_loops.
3110 (arc_reorg_loops): New function, calls reorg_loops.
3111 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
3112 (arc600_corereg_hazard): Remove ZOL checking, case handled by
3114 (arc_loop_hazard): Remove function, functionality moved into
3116 (arc_hazard): Remove arc_loop_hazard call.
3117 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
3118 into hwloop_optimize.
3119 (arc_label_align): Remove ZOL handling.
3120 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
3121 * config/arc/arc.md (doloop_begin): Remove pattern.
3122 (doloop_begin_i): Likewise.
3123 (doloop_end_i): Likewise.
3124 (doloop_fallback): Likewise.
3125 (doloop_fallback_m): Likewise.
3126 (doloop_end): Reimplement expand.
3127 (arc_lp): New pattern for LP instruction.
3128 (loop_end): New pattern.
3129 (loop_fail): Likewise.
3130 (decrement_and_branch_until_zero): Likewise.
3131 * config/arc/arc.opt (mlpc-width): New option.
3132 * doc/invoke.texi (mlpc-width): Document option.
3134 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
3136 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
3137 (arc_ccfsm_advance): Fix checking for delay slots.
3138 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
3140 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
3142 * config/arc/arc.md (movqi_insn): Add stores to save constant long
3144 (movhi_insn): Update store instruction constraint which are saving
3145 6-bit short immediates.
3146 (movsi_insn): Consider also short scaled load operations.
3147 (zero_extendhisi2_i): Use Usd constraint instead of T.
3148 (extendhisi2_i): Add q constraint.
3149 (arc_clzsi2): Add type and length attributes.
3150 (arc_ctzsi2): Likewise.
3151 * config/arc/constraints.md (Usc): Update constraint, the
3152 assembler can parse two relocations for a single instruction.
3154 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
3156 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
3157 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
3159 2017-08-31 Olivier Hainque <hainque@adacore.com>
3161 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
3162 match as powerpc-wrs-vxworks*.
3164 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
3166 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
3167 register constraint for by-element operand.
3168 (aarch64_mls_elt_merge<mode>): Likewise.
3170 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
3172 * config/arc/arc.c (arc_can_follow_jump): Check for short
3175 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
3177 * config.gcc: Use g.opt for arc.
3178 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
3179 functionality moved to ...
3180 (legitimate_scaled_address_p): New function, ...here.
3181 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
3182 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
3183 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
3185 (arc_override_options): Handle G option.
3186 (arc_output_pic_addr_const): Correct function definition.
3187 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
3188 (arc_decl_anon_ns_mem_p): Delete.
3189 (arc_in_small_data_p): Overhaul this function to take into
3190 consideration the value given via G option.
3191 (arc_rewrite_small_data_1): Renamed and corrected old
3192 arc_rewrite_small_data function.
3193 (arc_rewrite_small_data): New function.
3194 (small_data_pattern): Don't use pic_offset_table_rtx.
3195 * config/arc/arc.h (CC1_SPEC): Recognize G option.
3196 * config/arc/simdext.md (movmisalignv2hi): Use
3197 prepare_move_operands function.
3199 (movmisalign*): Likewise.
3200 * doc/invoke.texi (ARC options): Document -G option.
3202 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
3204 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
3206 * config/arc/arc.c (arc_print_operand): Output scalled address for
3207 sdata whenever is possible.
3208 (arc_in_small_data_p): Allow sdata for 64bit datum when double
3209 load/stores are available.
3210 (compact_sda_memory_operand): Check for the alignment required by
3211 code density instructions.
3212 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
3214 * config/arc/constraints.md (Usd): Update constraint.
3215 (Us0): New constraint.
3216 (Usc): Update constraint.
3218 2017-08-31 Richard Biener <rguenther@suse.de>
3221 * dwarf2out.c (dwarf2out_early_global_decl): Process each
3224 2017-08-31 Tamar Christina <tamar.christina@arm.com>
3226 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
3227 Resize type_signature.
3229 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
3230 Alan Hayward <alan.hayward@arm.com>
3231 David Sherwood <david.sherwood@arm.com>
3233 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
3234 subregs whose inner modes can be stored in GPRs.
3235 (aarch64_classify_index): Likewise.
3237 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
3238 Alan Hayward <alan.hayward@arm.com>
3239 David Sherwood <david.sherwood@arm.com>
3241 * config/aarch64/iterators.md (V_cmp_result): Rename to...
3242 (V_INT_EQUIV): ...this.
3243 (v_cmp_result): Rename to...
3244 (v_int_equiv): ...this.
3245 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
3246 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
3247 (copysign<mode>3): Likewise.
3248 (aarch64_simd_bsl<mode>_internal): Likewise.
3249 (aarch64_simd_bsl<mode>): Likewise.
3250 (vec_cmp<mode><mode>): Likewise.
3251 (vcond<mode><mode>): Likewise.
3252 (vcond<v_cmp_mixed><mode>): Likewise.
3253 (vcondu<mode><v_cmp_mixed>): Likewise.
3254 (aarch64_cm<optab><mode>): Likewise.
3255 (aarch64_cmtst<mode>): Likewise.
3256 (aarch64_fac<optab><mode>): Likewise.
3257 (vec_perm_const<mode>): Likewise.
3258 (vcond_mask_<mode><v_cmp_result>): Rename to...
3259 (vcond_mask_<mode><v_int_equiv>): ...this.
3260 (vec_cmp<mode><v_cmp_result>): Rename to...
3261 (vec_cmp<mode><v_int_equiv>): ...this.
3263 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
3264 Alan Hayward <alan.hayward@arm.com>
3265 David Sherwood <david.sherwood@arm.com>
3267 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
3269 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
3270 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
3271 (UNSPEC_LD4_DREG): New unspecs.
3272 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
3273 (aarch64_ld2<mode>_dreg_be): Replace with...
3274 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
3276 (aarch64_ld3<mode>_dreg_le)
3277 (aarch64_ld3<mode>_dreg_be): Replace with...
3278 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
3280 (aarch64_ld4<mode>_dreg_le)
3281 (aarch64_ld4<mode>_dreg_be): Replace with...
3282 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
3285 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3287 PR tree-optimization/81987
3288 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
3289 insert an initializer in a location not dominated by the stride
3292 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
3294 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
3295 on the entire header of the finally block in the fallthru case.
3297 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
3299 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
3301 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
3303 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
3304 rs6000_emit_move_from_cr and call renamed function.
3305 (rs6000_emit_prologue): Call renamed functions.
3306 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
3307 movesi_from_cr, remove volatile CRs.
3309 2017-08-30 Jon Beniston <jon@beniston.com>
3310 Richard Biener <rguenther@suse.de>
3312 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
3313 of VECTOR_MODE_P check.
3314 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
3315 element vector types.
3317 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3319 * df.h (df_read_modify_subreg_p): Remove in favor of...
3320 * rtl.h (read_modify_subreg_p): ...this new function. Take a
3321 const_rtx instead of an rtx.
3322 * cprop.c (local_cprop_find_used_regs): Update accordingly.
3323 * df-problems.c (df_word_lr_mark_ref): Likewise.
3324 * ira-lives.c (mark_pseudo_reg_live): Likewise.
3325 (mark_pseudo_reg_dead): Likewise.
3326 (mark_ref_dead): Likewise.
3327 * reginfo.c (init_subregs_of_mode): Likewise.
3328 * sched-deps.c (sched_analyze_1): Likewise.
3329 * df-scan.c (df_def_record_1): Likewise.
3330 (df_uses_record): Likewise.
3331 (df_read_modify_subreg_p): Remove in favor of...
3332 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
3333 const_rtx instead of an rtx.
3335 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3336 Alan Hayward <alan.hayward@arm.com>
3337 David Sherwood <david.sherwood@arm.com>
3339 * rtl.h (partial_subreg_p): New function.
3340 * caller-save.c (save_call_clobbered_regs): Use it.
3341 * calls.c (expand_call): Likewise.
3342 * combine.c (combinable_i3pat): Likewise.
3343 (simplify_set): Likewise.
3344 (make_extraction): Likewise.
3345 (make_compound_operation_int): Likewise.
3346 (gen_lowpart_or_truncate): Likewise.
3347 (force_to_mode): Likewise.
3348 (make_field_assignment): Likewise.
3349 (reg_truncated_to_mode): Likewise.
3350 (record_truncated_value): Likewise.
3351 (move_deaths): Likewise.
3352 * cse.c (record_jump_cond): Likewise.
3353 (cse_insn): Likewise.
3354 * cselib.c (cselib_lookup_1): Likewise.
3355 * expmed.c (extract_bit_field_using_extv): Likewise.
3356 * function.c (assign_parm_setup_reg): Likewise.
3357 * ifcvt.c (noce_convert_multiple_sets): Likewise.
3358 * ira-build.c (create_insn_allocnos): Likewise.
3359 * lra-coalesce.c (merge_pseudos): Likewise.
3360 * lra-constraints.c (match_reload): Likewise.
3361 (simplify_operand_subreg): Likewise.
3362 (curr_insn_transform): Likewise.
3363 * lra-lives.c (process_bb_lives): Likewise.
3364 * lra.c (new_insn_reg): Likewise.
3365 (lra_substitute_pseudo): Likewise.
3366 * regcprop.c (mode_change_ok): Likewise.
3367 (maybe_mode_change): Likewise.
3368 (copyprop_hardreg_forward_1): Likewise.
3369 * reload.c (push_reload): Likewise.
3370 (find_reloads): Likewise.
3371 (find_reloads_subreg_address): Likewise.
3372 * reload1.c (alter_reg): Likewise.
3373 (eliminate_regs_1): Likewise.
3374 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3376 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
3378 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
3379 back to if statements, including unpack.
3381 2017-08-30 Martin Liska <mliska@suse.cz>
3384 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
3386 (func_checker::compare_asm_inputs_outputs): ... this function.
3387 (func_checker::compare_gimple_asm): Use the function to compare
3388 also ASM constrains.
3389 * ipa-icf-gimple.h: Rename the function.
3391 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3392 Alan Hayward <alan.hayward@arm.com>
3393 David Sherwood <david.sherwood@arm.com>
3395 * coretypes.h (complex_mode): New type.
3396 * gdbhooks.py (build_pretty_printer): Handle it.
3397 * machmode.h (complex_mode): New class.
3398 (complex_mode::includes_p): New function.
3399 (is_complex_int_mode): Likewise.
3400 (is_complex_float_mode): Likewise.
3401 * genmodes.c (get_mode_class): Handle complex mode classes.
3402 * function.c (expand_function_end): Use is_complex_int_mode.
3404 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3405 Alan Hayward <alan.hayward@arm.com>
3406 David Sherwood <david.sherwood@arm.com>
3408 * coretypes.h (scalar_mode_pod): New typedef.
3409 * gdbhooks.py (build_pretty_printer): Handle it.
3410 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
3411 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
3412 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
3413 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
3416 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3417 Alan Hayward <alan.hayward@arm.com>
3418 David Sherwood <david.sherwood@arm.com>
3420 * machmode.h (mode_for_vector): Take a scalar_mode instead
3422 * stor-layout.c (mode_for_vector): Likewise.
3423 * explow.c (promote_mode): Use as_a <scalar_mode>.
3424 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
3426 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3427 Alan Hayward <alan.hayward@arm.com>
3428 David Sherwood <david.sherwood@arm.com>
3430 * target.def (preferred_simd_mode): Take a scalar_mode
3431 instead of a machine_mode.
3432 * targhooks.h (default_preferred_simd_mode): Likewise.
3433 * targhooks.c (default_preferred_simd_mode): Likewise.
3434 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
3435 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
3436 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
3437 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
3438 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
3439 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
3440 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
3441 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
3443 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
3444 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
3445 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
3446 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
3447 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
3448 * doc/tm.texi: Regenerate.
3449 * optabs-query.c (can_vec_mask_load_store_p): Return false for
3452 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3453 Alan Hayward <alan.hayward@arm.com>
3454 David Sherwood <david.sherwood@arm.com>
3456 * target.def (scalar_mode_supported_p): Take a scalar_mode
3457 instead of a machine_mode.
3458 * targhooks.h (default_scalar_mode_supported_p): Likewise.
3459 * targhooks.c (default_scalar_mode_supported_p): Likewise.
3460 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
3461 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
3462 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
3463 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
3464 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
3465 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
3466 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
3467 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
3468 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
3469 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
3470 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
3471 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
3473 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
3474 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
3475 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
3476 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
3477 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
3479 * doc/tm.texi: Regenerate.
3481 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3482 Alan Hayward <alan.hayward@arm.com>
3483 David Sherwood <david.sherwood@arm.com>
3485 * coretypes.h (opt_scalar_mode): New typedef.
3486 * gdbhooks.py (build_pretty_printers): Handle it.
3487 * machmode.h (mode_iterator::get_2xwider): Add overload for
3489 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
3491 * expr.c (convert_mode_scalar): Likewise.
3492 * omp-low.c (omp_clause_aligned_alignment): Likewise.
3493 * optabs.c (expand_float): Likewise.
3494 (expand_fix): Likewise.
3495 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
3497 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3498 Alan Hayward <alan.hayward@arm.com>
3499 David Sherwood <david.sherwood@arm.com>
3501 * optabs.c (expand_float): Explicitly check for scalars before
3502 using a branching expansion.
3503 (expand_fix): Likewise.
3505 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3506 Alan Hayward <alan.hayward@arm.com>
3507 David Sherwood <david.sherwood@arm.com>
3509 * expr.c (convert_mode): Split scalar handling out into...
3510 (convert_mode_scalar): ...this new function. Treat the modes
3513 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3514 Alan Hayward <alan.hayward@arm.com>
3515 David Sherwood <david.sherwood@arm.com>
3517 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
3519 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
3521 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3522 Alan Hayward <alan.hayward@arm.com>
3523 David Sherwood <david.sherwood@arm.com>
3525 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
3526 rather than a machine_mode.
3527 (fixed_from_string): Likewise.
3528 (fixed_convert): Likewise.
3529 (fixed_convert_from_int): Likewise.
3530 (fixed_convert_from_real): Likewise.
3531 (real_convert_from_fixed): Likewise.
3532 * fixed-value.c (fixed_from_double_int): Likewise.
3533 (fixed_from_string): Likewise.
3534 (fixed_convert): Likewise.
3535 (fixed_convert_from_int): Likewise.
3536 (fixed_convert_from_real): Likewise.
3537 (real_convert_from_fixed): Likewise.
3538 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
3540 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3541 Alan Hayward <alan.hayward@arm.com>
3542 David Sherwood <david.sherwood@arm.com>
3544 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
3545 of separate mode class checks. Do not allow vector modes here.
3546 (immed_wide_int_const): Use as_a <scalar_mode>.
3547 * explow.c (trunc_int_for_mode): Likewise.
3548 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
3549 (wi::shwi): Likewise.
3550 (wi::min_value): Likewise.
3551 (wi::max_value): Likewise.
3552 * dwarf2out.c (loc_descriptor): Likewise.
3553 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
3556 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3557 Alan Hayward <alan.hayward@arm.com>
3558 David Sherwood <david.sherwood@arm.com>
3560 * tree.h (SCALAR_TYPE_MODE): New macro.
3561 * expr.c (expand_expr_addr_expr_1): Use it.
3562 (expand_expr_real_2): Likewise.
3563 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
3564 (fold_convert_const_fixed_from_int): Likewise.
3565 (fold_convert_const_fixed_from_real): Likewise.
3566 (native_encode_fixed): Likewise
3567 (native_encode_complex): Likewise
3568 (native_encode_vector): Likewise.
3569 (native_interpret_fixed): Likewise.
3570 (native_interpret_real): Likewise.
3571 (native_interpret_complex): Likewise.
3572 (native_interpret_vector): Likewise.
3573 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
3574 (simd_clone_adjust_argument_types): Likewise.
3575 (simd_clone_init_simd_arrays): Likewise.
3576 (simd_clone_adjust): Likewise.
3577 * stor-layout.c (layout_type): Likewise.
3578 * tree.c (build_minus_one_cst): Likewise.
3579 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
3580 * tree-inline.c (estimate_move_cost): Likewise.
3581 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
3582 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
3583 (vectorizable_reduction): Likewise.
3584 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
3585 (vect_recog_mixed_size_cond_pattern): Likewise.
3586 (check_bool_pattern): Likewise.
3587 (adjust_bool_pattern): Likewise.
3588 (search_type_for_mask_1): Likewise.
3589 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
3590 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
3591 (vectorizable_load): Likewise.
3592 (vectorizable_store): Likewise.
3593 * ubsan.c (ubsan_encode_value): Likewise.
3594 * varasm.c (output_constant): Likewise.
3596 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3597 Alan Hayward <alan.hayward@arm.com>
3598 David Sherwood <david.sherwood@arm.com>
3600 * coretypes.h (scalar_mode): New class.
3601 * machmode.h (scalar_mode): Likewise.
3602 (scalar_mode::includes_p): New function.
3603 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
3604 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
3605 * genmodes.c (get_mode_class): Handle remaining scalar modes.
3606 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
3607 * expmed.c (store_bit_field_1): Likewise.
3608 (extract_bit_field_1): Likewise.
3609 * expr.c (write_complex_part): Likewise.
3610 (read_complex_part): Likewise.
3611 (emit_move_complex_push): Likewise.
3612 (expand_expr_real_2): Likewise.
3613 * function.c (assign_parm_setup_reg): Likewise.
3614 (assign_parms_unsplit_complex): Likewise.
3615 * optabs.c (expand_binop): Likewise.
3616 * rtlanal.c (subreg_get_info): Likewise.
3617 * simplify-rtx.c (simplify_immed_subreg): Likewise.
3618 * varasm.c (output_constant_pool_2): Likewise.
3620 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3621 Alan Hayward <alan.hayward@arm.com>
3622 David Sherwood <david.sherwood@arm.com>
3624 * expmed.c (extract_high_half): Use scalar_int_mode and remove
3626 (expmed_mult_highpart_optab): Likewise.
3627 (expmed_mult_highpart): Likewise.
3629 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3630 Alan Hayward <alan.hayward@arm.com>
3631 David Sherwood <david.sherwood@arm.com>
3633 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
3634 instead of a machine_mode.
3635 (builtin_memset_read_str): Likewise.
3636 * builtins.c (c_readstr): Likewise.
3637 (builtin_memcpy_read_str): Likewise.
3638 (builtin_strncpy_read_str): Likewise.
3639 (builtin_memset_read_str): Likewise.
3640 (builtin_memset_gen_str): Likewise.
3641 (expand_builtin_signbit): Use scalar_int_mode for local variables.
3642 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
3643 instead of a machine_mode.
3644 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
3646 (make_extraction): Likewise.
3647 (try_widen_shift_mode): Take and return scalar_int_modes instead
3649 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
3650 a scalar_int_mode instead of a machine_mode.
3651 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
3652 (avr_addr_space_pointer_mode): Likewise.
3653 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
3654 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
3655 (msp430_unwind_word_mode): Likewise.
3656 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
3657 (spu_addr_space_pointer_mode): Likewise.
3658 (spu_addr_space_address_mode): Likewise.
3659 (spu_libgcc_cmp_return_mode): Likewise.
3660 (spu_libgcc_shift_count_mode): Likewise.
3661 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
3662 (rl78_addr_space_pointer_mode): Likewise.
3663 (fl78_unwind_word_mode): Likewise.
3664 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
3666 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
3667 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
3668 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
3669 (mips_valid_pointer_mode): Likewise.
3670 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
3671 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
3672 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
3674 (ft32_addr_space_address_mode): Likewise.
3675 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
3676 scalar_int_mode instead of a machine_mode.
3677 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
3679 (m32c_addr_space_address_mode): Likewise.
3680 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
3681 (rs6000_eh_return_filter_mode): Likewise.
3682 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
3683 (rs6000_eh_return_filter_mode): Likewise.
3684 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
3685 (s390_libgcc_shift_count_mode): Likewise.
3686 (s390_unwind_word_mode): Likewise.
3687 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
3689 * target.def (mode_rep_extended): Likewise.
3690 (valid_pointer_mode): Likewise.
3691 (addr_space.valid_pointer_mode): Likewise.
3692 (eh_return_filter_mode): Return a scalar_int_mode rather than
3694 (libgcc_cmp_return_mode): Likewise.
3695 (libgcc_shift_count_mode): Likewise.
3696 (unwind_word_mode): Likewise.
3697 (addr_space.pointer_mode): Likewise.
3698 (addr_space.address_mode): Likewise.
3699 * doc/tm.texi: Regenerate.
3700 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
3702 (do_jump): Use scalar_int_mode for local variables.
3703 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
3704 rather than a machine_mode.
3705 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
3706 (scompare_loc_descriptor_wide): Likewise.
3707 (scompare_loc_descriptor_narrow): Likewise.
3708 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
3710 * except.c (sjlj_emit_dispatch_table): Likewise.
3711 (expand_builtin_eh_copy_values): Likewise.
3712 * explow.c (convert_memory_address_addr_space_1): Likewise.
3713 Take a scalar_int_mode rather than a machine_mode.
3714 (convert_memory_address_addr_space): Take a scalar_int_mode rather
3715 than a machine_mode.
3716 (memory_address_addr_space): Use scalar_int_mode for local variables.
3717 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
3718 rather than a machine_mode.
3719 * expmed.c (mask_rtx): Likewise.
3720 (init_expmed_one_conv): Likewise.
3721 (expand_mult_highpart_adjust): Likewise.
3722 (extract_high_half): Likewise.
3723 (expmed_mult_highpart_optab): Likewise.
3724 (expmed_mult_highpart): Likewise.
3725 (expand_smod_pow2): Likewise.
3726 (expand_sdiv_pow2): Likewise.
3727 (emit_store_flag_int): Likewise.
3728 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
3730 (extract_low_bits): Likewise.
3731 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
3733 * expr.c (pieces_addr::adjust): Likewise.
3734 (can_store_by_pieces): Likewise.
3735 (store_by_pieces): Likewise.
3736 (clear_by_pieces_1): Likewise.
3737 (expand_expr_addr_expr_1): Likewise.
3738 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
3739 (expand_expr_real_1): Likewise.
3740 (try_casesi): Likewise.
3741 * final.c (shorten_branches): Likewise.
3742 * fold-const.c (fold_convert_const_int_from_fixed): Change the
3743 type of "mode" to machine_mode.
3744 * internal-fn.c (expand_arith_overflow_result_store): Take a
3745 scalar_int_mode rather than a machine_mode.
3746 (expand_mul_overflow): Use scalar_int_mode for local variables.
3747 * loop-doloop.c (doloop_modify): Likewise.
3748 (doloop_optimize): Likewise.
3749 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
3750 than a machine_mode.
3751 (expand_doubleword_shift_condmove): Likewise.
3752 (expand_doubleword_shift): Likewise.
3753 (expand_doubleword_clz): Likewise.
3754 (expand_doubleword_popcount): Likewise.
3755 (expand_doubleword_parity): Likewise.
3756 (expand_absneg_bit): Use scalar_int_mode for local variables.
3757 (prepare_float_lib_cmp): Likewise.
3758 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
3759 rather than a machine_mode.
3760 (convert_memory_address_addr_space): Likewise.
3761 (get_mode_bounds): Likewise.
3762 (get_address_mode): Return a scalar_int_mode rather than a
3764 * rtlanal.c (get_address_mode): Likewise.
3765 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
3766 than a machine_mode.
3767 * targhooks.c (default_mode_rep_extended): Likewise.
3768 (default_valid_pointer_mode): Likewise.
3769 (default_addr_space_valid_pointer_mode): Likewise.
3770 (default_eh_return_filter_mode): Return a scalar_int_mode rather
3771 than a machine_mode.
3772 (default_libgcc_cmp_return_mode): Likewise.
3773 (default_libgcc_shift_count_mode): Likewise.
3774 (default_unwind_word_mode): Likewise.
3775 (default_addr_space_pointer_mode): Likewise.
3776 (default_addr_space_address_mode): Likewise.
3777 * targhooks.h (default_eh_return_filter_mode): Likewise.
3778 (default_libgcc_cmp_return_mode): Likewise.
3779 (default_libgcc_shift_count_mode): Likewise.
3780 (default_unwind_word_mode): Likewise.
3781 (default_addr_space_pointer_mode): Likewise.
3782 (default_addr_space_address_mode): Likewise.
3783 (default_mode_rep_extended): Take a scalar_int_mode rather than
3785 (default_valid_pointer_mode): Likewise.
3786 (default_addr_space_valid_pointer_mode): Likewise.
3787 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
3789 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
3790 rather than a machine_mode.
3791 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
3792 for local variables.
3793 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
3794 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
3795 than a machine_mode.
3797 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3798 Alan Hayward <alan.hayward@arm.com>
3799 David Sherwood <david.sherwood@arm.com>
3801 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
3802 the mode argument to scalar_int_mode.
3803 (do_jump_by_parts_zero_rtx): Likewise.
3804 (do_jump_by_parts_equality_rtx): Likewise.
3805 (do_jump_by_parts_greater): Take a mode argument.
3806 (do_jump_by_parts_equality): Likewise.
3807 (do_jump_1): Update calls accordingly.
3809 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3810 Alan Hayward <alan.hayward@arm.com>
3811 David Sherwood <david.sherwood@arm.com>
3813 * is-a.h (safe_dyn_cast): New function.
3814 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
3815 (jump_table_for_label): Likewise.
3816 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
3817 instead of an rtx_insn *.
3818 (shorten_branches): Use dyn_cast instead of LABEL_P and
3819 JUMP_TABLE_DATA_P. Use jump_table_for_label and
3820 rtx_jump_table_data::get_data_mode.
3821 (final_scan_insn): Likewise.
3823 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3824 Alan Hayward <alan.hayward@arm.com>
3825 David Sherwood <david.sherwood@arm.com>
3827 * combine.c (try_combine): Use is_a <scalar_int_mode> when
3828 trying to combine a full-register integer set with a subreg
3831 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3832 Alan Hayward <alan.hayward@arm.com>
3833 David Sherwood <david.sherwood@arm.com>
3835 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
3836 that is always either address_mode or pointer_mode.
3838 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3839 Alan Hayward <alan.hayward@arm.com>
3840 David Sherwood <david.sherwood@arm.com>
3842 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
3843 when the two are known to be equal.
3845 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3846 Alan Hayward <alan.hayward@arm.com>
3847 David Sherwood <david.sherwood@arm.com>
3849 * simplify-rtx.c (simplify_const_unary_operation): Use
3850 is_a <scalar_int_mode> instead of checking for a nonzero
3851 precision. Forcibly convert op_mode to a scalar_int_mode
3852 in that case. More clearly differentiate the operand and
3853 result modes and use the former when deciding what the value
3854 of a count-bits operation should be. Use is_int_mode instead
3855 of checking for a MODE_INT. Remove redundant check for whether
3856 this mode has a zero precision.
3858 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3859 Alan Hayward <alan.hayward@arm.com>
3860 David Sherwood <david.sherwood@arm.com>
3862 * optabs.c (widen_leading): Change the type of the mode argument
3863 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
3864 (widen_bswap): Likewise.
3865 (expand_parity): Likewise.
3866 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
3867 (expand_ffs): Likewise.
3868 (epand_unop): Check for scalar integer modes before calling the
3871 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3872 Alan Hayward <alan.hayward@arm.com>
3873 David Sherwood <david.sherwood@arm.com>
3875 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
3877 (expand_expr_real_1): Update call accordingly.
3879 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3880 Alan Hayward <alan.hayward@arm.com>
3881 David Sherwood <david.sherwood@arm.com>
3883 * expmed.c (store_bit_field_using_insv): Add op0_mode and
3884 value_mode arguments. Use scalar_int_mode internally.
3885 (store_bit_field_1): Rename the new integer mode from imode
3886 to op0_mode and use it instead of GET_MODE (op0). Update calls
3887 to store_split_bit_field, store_bit_field_using_insv and
3888 store_fixed_bit_field.
3889 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
3890 Use scalar_int_mode internally. Use a bit count rather than a mode
3891 when calculating the largest bit size for get_best_mode.
3892 Update calls to store_split_bit_field and store_fixed_bit_field_1.
3893 (store_fixed_bit_field_1): Add mode and value_mode arguments.
3894 Remove assertion that OP0 has a scalar integer mode.
3895 (store_split_bit_field): Add op0_mode and value_mode arguments.
3896 Update calls to extract_fixed_bit_field.
3897 (extract_bit_field_using_extv): Add an op0_mode argument.
3898 Use scalar_int_mode internally.
3899 (extract_bit_field_1): Rename the new integer mode from imode to
3900 op0_mode and use it instead of GET_MODE (op0). Update calls to
3901 extract_split_bit_field, extract_bit_field_using_extv and
3902 extract_fixed_bit_field.
3903 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
3904 to extract_split_bit_field and extract_fixed_bit_field_1.
3905 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
3906 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
3908 (extract_split_bit_field): Add an op0_mode argument. Update call
3909 to extract_fixed_bit_field.
3911 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3912 Alan Hayward <alan.hayward@arm.com>
3913 David Sherwood <david.sherwood@arm.com>
3915 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
3916 * explow.c (hard_function_value): Likewise.
3917 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
3918 convert_to_mode call outside the loop.
3919 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
3920 for the mode iterator. Require the mode specified by max_pieces
3922 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
3924 (copy_blkmode_to_reg): Likewise.
3925 (set_storage_via_setmem): Likewise.
3926 * optabs.c (prepare_cmp_insn): Likewise.
3927 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
3928 * stor-layout.c (finish_bitfield_representative): Likewise.
3930 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3931 Alan Hayward <alan.hayward@arm.com>
3932 David Sherwood <david.sherwood@arm.com>
3934 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
3935 * expr.c (convert_move): Use them.
3936 (convert_modes): Likewise.
3937 (store_expr_with_bounds): Likewise.
3939 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3940 Alan Hayward <alan.hayward@arm.com>
3941 David Sherwood <david.sherwood@arm.com>
3943 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
3944 parameter for the mode of "x". Remove the "known_x", "known_mode"
3945 and "known_ret" arguments. Change the type of the mode argument
3947 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
3948 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
3949 (reg_num_sign_bit_copies_for_combine): Likewise.
3950 * rtlanal.c (nonzero_bits1): Likewise.
3951 (num_sign_bit_copies1): Likewise.
3952 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
3953 (reg_num_sign_bit_copies_general): Likewise.
3954 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
3955 (reg_nonzero_bits_general): Likewise.
3957 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3958 Alan Hayward <alan.hayward@arm.com>
3959 David Sherwood <david.sherwood@arm.com>
3961 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
3962 than in subroutines. Return 1 for non-integer modes.
3963 (cached_num_sign_bit_copies): Change the type of the mode parameter
3965 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
3966 classes. Handle CONST_INT_P first and then check whether X also
3967 has a scalar integer mode. Check the same thing for inner registers
3968 of a SUBREG and for values that are being extended or truncated.
3970 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3971 Alan Hayward <alan.hayward@arm.com>
3972 David Sherwood <david.sherwood@arm.com>
3974 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
3975 in subroutines. Return the mode mask for non-integer modes.
3976 (cached_nonzero_bits): Change the type of the mode parameter
3978 (nonzero_bits1): Likewise. Remove early exit for other mode
3979 classes. Handle CONST_INT_P first and then check whether X
3980 also has a scalar integer mode.
3982 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3983 Alan Hayward <alan.hayward@arm.com>
3984 David Sherwood <david.sherwood@arm.com>
3986 * expr.c (widest_int_mode_for_size): Make the comment match the code.
3987 Return a scalar_int_mode and assert that the size is greater than
3989 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
3990 (op_by_pieces_d::op_by_pieces_d): Likewise.
3991 (op_by_pieces_d::run): Likewise.
3992 (can_store_by_pieces): Likewise.
3994 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3995 Alan Hayward <alan.hayward@arm.com>
3996 David Sherwood <david.sherwood@arm.com>
3998 * combine.c (extract_left_shift): Add a mode argument and update
4000 (make_compound_operation_int): Change the type of the mode parameter
4001 to scalar_int_mode and update the call to extract_left_shift.
4003 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4004 Alan Hayward <alan.hayward@arm.com>
4005 David Sherwood <david.sherwood@arm.com>
4007 * combine.c (simplify_and_const_int): Change the type of the mode
4008 parameter to scalar_int_mode.
4009 (simplify_and_const_int_1): Likewise. Update recursive call.
4011 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4012 Alan Hayward <alan.hayward@arm.com>
4013 David Sherwood <david.sherwood@arm.com>
4015 * combine.c (simplify_compare_const): Check that the mode is a
4016 scalar_int_mode (rather than VOIDmode) before testing its
4018 (simplify_comparison): Move COMPARISON_P handling out of the
4019 loop and restrict the latter part of the loop to scalar_int_modes.
4020 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
4021 and when considering SUBREG_REGs. Use is_int_mode instead of
4022 checking GET_MODE_CLASS against MODE_INT.
4024 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4025 Alan Hayward <alan.hayward@arm.com>
4026 David Sherwood <david.sherwood@arm.com>
4028 * combine.c (try_widen_shift_mode): Move check for equal modes to...
4029 (simplify_shift_const_1): ...here. Use scalar_int_mode for
4030 shift_unit_mode and for modes involved in scalar shifts.
4032 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4033 Alan Hayward <alan.hayward@arm.com>
4034 David Sherwood <david.sherwood@arm.com>
4036 * combine.c (force_int_to_mode): New function, split out from...
4037 (force_to_mode): ...here. Keep xmode up-to-date and use it
4038 instead of GET_MODE (x).
4040 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4041 Alan Hayward <alan.hayward@arm.com>
4042 David Sherwood <david.sherwood@arm.com>
4044 * optabs-query.h (extraction_insn::struct_mode): Change type to
4045 opt_scalar_int_mode and update comment.
4046 (extraction_insn::field_mode): Change type to scalar_int_mode.
4047 (extraction_insn::pos_mode): Likewise.
4048 * combine.c (make_extraction): Update accordingly.
4049 * optabs-query.c (get_traditional_extraction_insn): Likewise.
4050 (get_optab_extraction_insn): Likewise.
4051 * recog.c (simplify_while_replacing): Likewise.
4052 * expmed.c (narrow_bit_field_mem): Change the type of the mode
4053 parameter to opt_scalar_int_mode.
4055 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4056 Alan Hayward <alan.hayward@arm.com>
4057 David Sherwood <david.sherwood@arm.com>
4059 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
4060 to a scalar_int_mode instead of a machine_mode.
4061 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
4062 (get_best_mode): Return a boolean and use a pointer argument to store
4063 the selected mode. Replace the limit mode parameter with a bit limit.
4064 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
4065 for the values returned by bit_field_mode_iterator::next_mode.
4066 (store_bit_field): Update call to get_best_mode.
4067 (store_fixed_bit_field): Likewise.
4068 (extract_fixed_bit_field): Likewise.
4069 * expr.c (optimize_bitfield_assignment_op): Likewise.
4070 * fold-const.c (optimize_bit_field_compare): Likewise.
4071 (fold_truth_andor_1): Likewise.
4072 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
4073 Update for new type of m_mode.
4074 (get_best_mode): As above.
4076 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4077 Alan Hayward <alan.hayward@arm.com>
4078 David Sherwood <david.sherwood@arm.com>
4080 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
4081 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
4082 (store_bit_field): Check is_a <scalar_int_mode> before calling
4083 strict_volatile_bitfield_p.
4084 (extract_bit_field): Likewise.
4086 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4087 Alan Hayward <alan.hayward@arm.com>
4088 David Sherwood <david.sherwood@arm.com>
4090 * target.def (cstore_mode): Return a scalar_int_mode.
4091 * doc/tm.texi: Regenerate.
4092 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
4093 * targhooks.h (default_cstore_mode): Likewise.
4094 * targhooks.c (default_cstore_mode): Likewise, using a forced
4096 * expmed.c (emit_cstore): Expect the target of the cstore to be
4099 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4100 Alan Hayward <alan.hayward@arm.com>
4101 David Sherwood <david.sherwood@arm.com>
4103 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
4105 (niter_desc): Likewise mode.
4106 (iv_analyze): Add a mode parameter.
4108 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
4109 and change its type to scalar_int_mode.
4110 * loop-iv.c: Update commentary at head of file.
4111 (iv_constant): Pass the mode paraeter before the rtx it describes
4112 and change its type to scalar_int_mode. Remove VOIDmode handling.
4113 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
4114 (iv_extend): Likewise.
4115 (shorten_into_mode): Likewise.
4116 (iv_add): Use scalar_int_mode.
4117 (iv_mult): Likewise.
4118 (iv_shift): Likewise.
4119 (canonicalize_iv_subregs): Likewise.
4120 (get_biv_step_1): Pass the outer_mode parameter before the rtx
4121 it describes and change its mode to scalar_int_mode. Also change
4122 the type of the returned inner_mode to scalar_int_mode.
4123 (get_biv_step): Likewise, turning outer_mode from a pointer
4124 into a direct parameter. Update call to get_biv_step_1.
4125 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
4126 iv_constant and get_biv_step.
4127 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
4128 and change its type to scalar_int_mode. Don't initialise iv->mode
4129 to VOIDmode and remove later checks for its still being VOIDmode.
4130 Update calls to iv_analyze_op and iv_analyze_expr. Check
4131 is_a <scalar_int_mode> when changing the mode under consideration.
4132 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
4133 Update call to iv_analyze_expr.
4134 (iv_analyze_op): Add a mode parameter. Reject subregs whose
4135 inner register is not also a scalar_int_mode. Update call to
4137 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
4138 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
4139 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
4140 separate mode class checks. Update calls to iv_analyze. Remove
4141 fix-up of VOIDmodes after iv_analyze_biv.
4142 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
4143 don't have a scalar_int_mode. Update call to biv_p.
4145 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4146 Alan Hayward <alan.hayward@arm.com>
4147 David Sherwood <david.sherwood@arm.com>
4149 * cfgexpand.c (convert_debug_memory_address): Use
4150 as_a <scalar_int_mode>.
4151 * combine.c (expand_compound_operation): Likewise.
4152 (make_extraction): Likewise.
4153 (change_zero_ext): Likewise.
4154 (simplify_comparison): Likewise.
4155 * cse.c (cse_insn): Likewise.
4156 * dwarf2out.c (minmax_loc_descriptor): Likewise.
4157 (mem_loc_descriptor): Likewise.
4158 (loc_descriptor): Likewise.
4159 * expmed.c (init_expmed_one_mode): Likewise.
4160 (synth_mult): Likewise.
4161 (emit_store_flag_1): Likewise.
4162 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
4163 of a comparison with size.
4164 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
4165 (reduce_to_bit_field_precision): Likewise.
4166 * function.c (expand_function_end): Likewise.
4167 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
4168 * loop-doloop.c (doloop_modify): Likewise.
4169 * optabs.c (expand_binop): Likewise.
4170 (expand_unop): Likewise.
4171 (expand_copysign_absneg): Likewise.
4172 (prepare_cmp_insn): Likewise.
4173 (maybe_legitimize_operand): Likewise.
4174 * recog.c (const_scalar_int_operand): Likewise.
4175 * rtlanal.c (get_address_mode): Likewise.
4176 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4177 (simplify_cond_clz_ctz): Likewise.
4178 * tree-nested.c (get_nl_goto_field): Likewise.
4179 * tree.c (build_vector_type_for_mode): Likewise.
4180 * var-tracking.c (use_narrower_mode): Likewise.
4182 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4183 Alan Hayward <alan.hayward@arm.com>
4184 David Sherwood <david.sherwood@arm.com>
4186 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
4187 * builtins.c (expand_builtin_signbit): Use it.
4188 * cfgexpand.c (expand_debug_expr): Likewise.
4189 * dojump.c (do_jump): Likewise.
4190 (do_compare_and_jump): Likewise.
4191 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
4192 * expmed.c (make_tree): Likewise.
4193 * expr.c (expand_expr_real_2): Likewise.
4194 (expand_expr_real_1): Likewise.
4195 (try_casesi): Likewise.
4196 * fold-const-call.c (fold_const_call_ss): Likewise.
4197 * fold-const.c (unextend): Likewise.
4198 (extract_muldiv_1): Likewise.
4199 (fold_single_bit_test): Likewise.
4200 (native_encode_int): Likewise.
4201 (native_encode_string): Likewise.
4202 (native_interpret_int): Likewise.
4203 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
4204 * internal-fn.c (expand_addsub_overflow): Likewise.
4205 (expand_neg_overflow): Likewise.
4206 (expand_mul_overflow): Likewise.
4207 (expand_arith_overflow): Likewise.
4208 * match.pd: Likewise.
4209 * stor-layout.c (layout_type): Likewise.
4210 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
4211 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
4212 * tree-ssanames.c (get_range_info): Likewise.
4213 * tree-switch-conversion.c (array_value_type) Likewise.
4214 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
4215 (vect_recog_divmod_pattern): Likewise.
4216 (vect_recog_mixed_size_cond_pattern): Likewise.
4217 * tree-vrp.c (extract_range_basic): Likewise.
4218 (simplify_float_conversion_using_ranges): Likewise.
4219 * tree.c (int_fits_type_p): Likewise.
4220 * ubsan.c (instrument_bool_enum_load): Likewise.
4221 * varasm.c (mergeable_string_section): Likewise.
4222 (narrowing_initializer_constant_valid_p): Likewise.
4223 (output_constant): Likewise.
4225 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4226 Alan Hayward <alan.hayward@arm.com>
4227 David Sherwood <david.sherwood@arm.com>
4229 * machmode.h (NARROWEST_INT_MODE): New macro.
4230 * expr.c (alignment_for_piecewise_move): Use it instead of
4231 GET_CLASS_NARROWEST_MODE (MODE_INT).
4232 (push_block): Likewise.
4233 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
4235 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
4237 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4238 Alan Hayward <alan.hayward@arm.com>
4239 David Sherwood <david.sherwood@arm.com>
4241 * postreload.c (move2add_valid_value_p): Change the type of the
4242 mode parameter to scalar_int_mode.
4243 (move2add_use_add2_insn): Add a mode parameter and use it instead
4245 (move2add_use_add3_insn): Likewise.
4246 (reload_cse_move2add): Update accordingly.
4248 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4249 Alan Hayward <alan.hayward@arm.com>
4250 David Sherwood <david.sherwood@arm.com>
4252 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
4254 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
4255 * optabs.c (expand_unop): Likewise.
4257 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4258 Alan Hayward <alan.hayward@arm.com>
4259 David Sherwood <david.sherwood@arm.com>
4261 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
4262 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
4263 (popcount_loc_descriptor): Likewise.
4264 (bswap_loc_descriptor): Likewise.
4265 (rotate_loc_descriptor): Likewise.
4266 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
4267 calling the functions above.
4269 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4270 Alan Hayward <alan.hayward@arm.com>
4271 David Sherwood <david.sherwood@arm.com>
4273 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
4275 (try_combine): Likewise.
4276 (simplify_if_then_else): Likewise.
4277 * cse.c (cse_insn): Likewise.
4278 * dwarf2out.c (mem_loc_descriptor): Likewise.
4279 * emit-rtl.c (gen_lowpart_common): Likewise.
4280 * simplify-rtx.c (simplify_truncation): Likewise.
4281 (simplify_binary_operation_1): Likewise.
4282 (simplify_const_relational_operation): Likewise.
4283 (simplify_ternary_operation): Likewise.
4284 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
4286 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4287 Alan Hayward <alan.hayward@arm.com>
4288 David Sherwood <david.sherwood@arm.com>
4290 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
4291 * reload.c (push_reload): Likewise.
4292 (find_reloads): Likewise.
4294 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4295 Alan Hayward <alan.hayward@arm.com>
4296 David Sherwood <david.sherwood@arm.com>
4298 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
4299 (make_compound_operation_int): Likewise.
4300 (change_zero_ext): Likewise.
4301 * expr.c (convert_move): Likewise.
4302 (convert_modes): Likewise.
4303 * fwprop.c (forward_propagate_subreg): Likewise.
4304 * loop-iv.c (get_biv_step_1): Likewise.
4305 * optabs.c (widen_operand): Likewise.
4306 * postreload.c (move2add_valid_value_p): Likewise.
4307 * recog.c (simplify_while_replacing): Likewise.
4308 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4309 (simplify_binary_operation_1): Likewise. Remove redundant
4310 mode equality check.
4312 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4313 Alan Hayward <alan.hayward@arm.com>
4314 David Sherwood <david.sherwood@arm.com>
4316 * combine.c (combine_simplify_rtx): Add checks for
4317 is_a <scalar_int_mode>.
4318 (simplify_if_then_else): Likewise.
4319 (make_field_assignment): Likewise.
4320 (simplify_comparison): Likewise.
4321 * ifcvt.c (noce_try_bitop): Likewise.
4322 * loop-invariant.c (canonicalize_address_mult): Likewise.
4323 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4325 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4326 Alan Hayward <alan.hayward@arm.com>
4327 David Sherwood <david.sherwood@arm.com>
4329 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
4330 is_a <scalar_int_mode> instead of != BLKmode.
4332 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4333 Alan Hayward <alan.hayward@arm.com>
4334 David Sherwood <david.sherwood@arm.com>
4336 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
4337 instead of != VOIDmode.
4338 * combine.c (if_then_else_cond): Likewise.
4339 (change_zero_ext): Likewise.
4340 * dwarf2out.c (mem_loc_descriptor): Likewise.
4341 (loc_descriptor): Likewise.
4342 * rtlanal.c (canonicalize_condition): Likewise.
4343 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
4345 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4346 Alan Hayward <alan.hayward@arm.com>
4347 David Sherwood <david.sherwood@arm.com>
4349 * simplify-rtx.c (simplify_binary_operation_1): Use
4350 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
4352 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4353 Alan Hayward <alan.hayward@arm.com>
4354 David Sherwood <david.sherwood@arm.com>
4356 * wide-int.h (int_traits<unsigned char>) New class.
4357 (int_traits<unsigned short>) Likewise.
4358 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
4359 Use GET_MODE_UNIT_PRECISION and remove redundant test for
4361 * combine.c (set_nonzero_bits_and_sign_copies): Use
4362 is_a <scalar_int_mode>.
4363 (find_split_point): Likewise.
4364 (combine_simplify_rtx): Likewise.
4365 (simplify_logical): Likewise.
4366 (expand_compound_operation): Likewise.
4367 (expand_field_assignment): Likewise.
4368 (make_compound_operation): Likewise.
4369 (extended_count): Likewise.
4370 (change_zero_ext): Likewise.
4371 (simplify_comparison): Likewise.
4372 * dwarf2out.c (scompare_loc_descriptor): Likewise.
4373 (ucompare_loc_descriptor): Likewise.
4374 (minmax_loc_descriptor): Likewise.
4375 (mem_loc_descriptor): Likewise.
4376 (loc_descriptor): Likewise.
4377 * expmed.c (init_expmed_one_mode): Likewise.
4378 * lra-constraints.c (lra_constraint_offset): Likewise.
4379 * optabs.c (prepare_libcall_arg): Likewise.
4380 * postreload.c (move2add_note_store): Likewise.
4381 * reload.c (operands_match_p): Likewise.
4382 * rtl.h (load_extend_op): Likewise.
4383 * rtlhooks.c (gen_lowpart_general): Likewise.
4384 * simplify-rtx.c (simplify_truncation): Likewise.
4385 (simplify_unary_operation_1): Likewise.
4386 (simplify_binary_operation_1): Likewise.
4387 (simplify_const_binary_operation): Likewise.
4388 (simplify_const_relational_operation): Likewise.
4389 (simplify_subreg): Likewise.
4390 * stor-layout.c (bitwise_mode_for_mode): Likewise.
4391 * var-tracking.c (adjust_mems): Likewise.
4392 (prepare_call_arguments): Likewise.
4394 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4395 Alan Hayward <alan.hayward@arm.com>
4396 David Sherwood <david.sherwood@arm.com>
4398 * machmode.h (is_int_mode): New fuction.
4399 * combine.c (find_split_point): Use it.
4400 (combine_simplify_rtx): Likewise.
4401 (simplify_if_then_else): Likewise.
4402 (simplify_set): Likewise.
4403 (simplify_shift_const_1): Likewise.
4404 (simplify_comparison): Likewise.
4405 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
4406 * cse.c (notreg_cost): Likewise.
4407 (cse_insn): Likewise.
4408 * cselib.c (cselib_lookup_1): Likewise.
4409 * dojump.c (do_jump_1): Likewise.
4410 (do_compare_rtx_and_jump): Likewise.
4411 * dse.c (get_call_args): Likewise.
4412 * dwarf2out.c (rtl_for_decl_init): Likewise.
4413 (native_encode_initializer): Likewise.
4414 * expmed.c (emit_store_flag_1): Likewise.
4415 (emit_store_flag): Likewise.
4416 * expr.c (convert_modes): Likewise.
4417 (store_field): Likewise.
4418 (expand_expr_real_1): Likewise.
4419 * fold-const.c (fold_read_from_constant_string): Likewise.
4420 * gimple-ssa-sprintf.c (get_format_string): Likewise.
4421 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
4422 * optabs.c (expand_binop): Likewise.
4423 (expand_unop): Likewise.
4424 (expand_abs_nojump): Likewise.
4425 (expand_one_cmpl_abs_nojump): Likewise.
4426 * simplify-rtx.c (mode_signbit_p): Likewise.
4427 (val_signbit_p): Likewise.
4428 (val_signbit_known_set_p): Likewise.
4429 (val_signbit_known_clear_p): Likewise.
4430 (simplify_relational_operation_1): Likewise.
4431 * tree.c (vector_type_mode): Likewise.
4433 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4434 Alan Hayward <alan.hayward@arm.com>
4435 David Sherwood <david.sherwood@arm.com>
4437 * machmode.h (smallest_mode_for_size): Fix formatting.
4438 (smallest_int_mode_for_size): New function.
4439 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
4440 instead of smallest_mode_for_size.
4441 * combine.c (make_extraction): Likewise.
4442 * config/arc/arc.c (arc_expand_movmem): Likewise.
4443 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
4444 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
4445 * config/s390/s390.c (s390_expand_insv): Likewise.
4446 * config/sparc/sparc.c (assign_int_registers): Likewise.
4447 * config/spu/spu.c (spu_function_value): Likewise.
4448 (spu_function_arg): Likewise.
4449 * coverage.c (get_gcov_type): Likewise.
4450 (get_gcov_unsigned_t): Likewise.
4451 * dse.c (find_shift_sequence): Likewise.
4452 * expmed.c (store_bit_field_1): Likewise.
4453 * expr.c (convert_move): Likewise.
4454 (store_field): Likewise.
4455 * internal-fn.c (expand_arith_overflow): Likewise.
4456 * optabs-query.c (get_best_extraction_insn): Likewise.
4457 * optabs.c (expand_twoval_binop_libfunc): Likewise.
4458 * stor-layout.c (layout_type): Likewise.
4459 (initialize_sizetypes): Likewise.
4460 * targhooks.c (default_get_mask_mode): Likewise.
4461 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
4463 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4464 Alan Hayward <alan.hayward@arm.com>
4465 David Sherwood <david.sherwood@arm.com>
4467 * machmode.h (opt_mode::else_blk): New function.
4468 (int_mode_for_mode): Declare.
4469 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
4470 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
4472 * cfgexpand.c (expand_debug_expr): Likewise.
4473 * combine.c (gen_lowpart_or_truncate): Likewise.
4474 (gen_lowpart_for_combine): Likewise.
4475 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
4476 * config/avr/avr.c (avr_to_int_mode): Likewise.
4477 (avr_out_plus_1): Likewise.
4478 (avr_out_plus): Likewise.
4479 (avr_out_round): Likewise.
4480 * config/i386/i386.c (ix86_split_to_parts): Likewise.
4481 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
4482 (s390_expand_vcond): Likewise.
4483 * config/spu/spu.c (spu_split_immediate): Likewise.
4484 (spu_expand_mov): Likewise.
4485 * dse.c (get_stored_val): Likewise.
4486 * expmed.c (store_bit_field_1): Likewise.
4487 (convert_extracted_bit_field): Use int_mode_for_mode instead of
4489 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
4490 (extract_low_bits): Likewise.
4491 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
4492 handling rather than repeating the check.
4493 (emit_group_store): Likewise.
4494 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
4495 * optabs.c (expand_absneg_bit): Likewise.
4496 (expand_copysign_absneg): Likewise.
4497 (expand_copysign_bit): Likewise.
4498 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
4499 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
4500 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
4501 * var-tracking.c (prepare_call_arguments): Likewise.
4502 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
4503 int_mode_for_mode instead of mode_for_size.
4504 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
4506 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4507 Alan Hayward <alan.hayward@arm.com>
4508 David Sherwood <david.sherwood@arm.com>
4510 * machmode.h (int_mode_for_size): New function.
4511 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
4512 instead of mode_for_size.
4513 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
4515 * combine.c (expand_field_assignment): Use int_mode_for_size
4516 instead of mode_for_size.
4517 (make_extraction): Likewise.
4518 (simplify_shift_const_1): Likewise.
4519 (simplify_comparison): Likewise.
4520 * dojump.c (do_jump): Likewise.
4521 * dwarf2out.c (mem_loc_descriptor): Likewise.
4522 * emit-rtl.c (init_derived_machine_modes): Likewise.
4523 * expmed.c (flip_storage_order): Likewise.
4524 (convert_extracted_bit_field): Likewise.
4525 * expr.c (copy_blkmode_from_reg): Likewise.
4526 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
4527 * internal-fn.c (expand_mul_overflow): Likewise.
4528 * lower-subreg.c (simple_move): Likewise.
4529 * optabs-libfuncs.c (init_optabs): Likewise.
4530 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4531 * tree.c (vector_type_mode): Likewise.
4532 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
4533 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
4534 * tree-vect-generic.c (expand_vector_parallel): Likewise.
4535 * tree-vect-stmts.c (vectorizable_load): Likewise.
4536 (vectorizable_store): Likewise.
4538 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4539 Alan Hayward <alan.hayward@arm.com>
4540 David Sherwood <david.sherwood@arm.com>
4542 * coretypes.h (pod_mode): New type.
4543 (scalar_int_mode_pod): New typedef.
4544 * machmode.h (pod_mode): New class.
4545 (int_n_data_t::m): Change type to scalar_int_mode_pod.
4546 * genmodes.c (emit_mode_int_n): Update accordingly.
4547 * lower-subreg.h (target_lower_subreg): Change type to
4548 scalar_int_mode_pod.
4549 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
4550 scalar_int_mode_pod.
4552 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4553 Alan Hayward <alan.hayward@arm.com>
4554 David Sherwood <david.sherwood@arm.com>
4556 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
4557 machine_mode to scalar_int_mode.
4558 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
4559 (rs6000_option_override_internal): Remove cast to int.
4560 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
4561 machine_mode to scalar_int_mode.
4562 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
4563 (rs6000_option_override_internal): Remove cast to int.
4564 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
4565 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
4567 * config/s390/s390.c (s390_expand_builtin): Likewise.
4568 * coretypes.h (scalar_int_mode): New type.
4569 (opt_scalar_int_mode): New typedef.
4570 * machmode.h (scalar_int_mode): New class.
4571 (scalar_int_mode::includes_p): New function.
4572 (byte_mode): Change type to scalar_int_mode.
4573 (word_mode): Likewise.
4574 (ptr_mode): Likewise.
4575 * emit-rtl.c (byte_mode): Likewise.
4576 (word_mode): Likewise.
4577 (ptr_mode): Likewise.
4578 (init_derived_machine_modes): Update accordingly.
4579 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
4580 and MODE_PARTIAL_INT.
4581 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
4582 opt_scalar_int_mode.
4584 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4585 Alan Hayward <alan.hayward@arm.com>
4586 David Sherwood <david.sherwood@arm.com>
4588 * target.def (libgcc_floating_mode_supported_p): Take a
4590 * doc/tm.texi: Regenerate.
4591 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
4593 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
4594 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
4597 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4598 Alan Hayward <alan.hayward@arm.com>
4599 David Sherwood <david.sherwood@arm.com>
4601 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
4602 * doc/tm.texi: Regenerate.
4603 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
4604 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
4605 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
4606 * targhooks.h (default_floatn_mode): Likewise.
4607 * targhooks.c (default_floatn_mode): Likewise.
4608 * tree.c (build_common_tree_nodes): Update accordingly.
4610 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4611 Alan Hayward <alan.hayward@arm.com>
4612 David Sherwood <david.sherwood@arm.com>
4614 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
4615 (mode_iterator::iterate_p): Likewise.
4616 (mode_iterator::get_wider): Likewise.
4617 * expr.c (init_expr_target): Use opt_scalar_float_mode.
4619 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4620 Alan Hayward <alan.hayward@arm.com>
4621 David Sherwood <david.sherwood@arm.com>
4623 * coretypes.h (opt_scalar_float_mode): New typedef.
4624 * machmode.h (float_mode_for_size): New function.
4625 * emit-rtl.c (double_mode): Delete.
4626 (init_emit_once): Use float_mode_for_size.
4627 * stor-layout.c (layout_type): Likewise.
4628 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
4630 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4631 Alan Hayward <alan.hayward@arm.com>
4632 David Sherwood <david.sherwood@arm.com>
4634 * output.h (assemble_real): Take a scalar_float_mode.
4635 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
4636 * config/arm/arm.md (consttable_4): Likewise.
4637 (consttable_8): Likewise.
4638 (consttable_16): Likewise.
4639 * config/mips/mips.md (consttable_float): Likewise.
4640 * config/s390/s390.c (s390_output_pool_entry): Likewise.
4641 * varasm.c (assemble_real): Take a scalar_float_mode.
4642 (output_constant_pool_2): Update accordingly.
4643 (output_constant): Likewise.
4645 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4646 Alan Hayward <alan.hayward@arm.com>
4647 David Sherwood <david.sherwood@arm.com>
4649 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
4650 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
4651 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
4652 (native_encode_real): Likewise.
4653 (native_interpret_real): Likewise.
4654 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
4655 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
4657 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4658 Alan Hayward <alan.hayward@arm.com>
4659 David Sherwood <david.sherwood@arm.com>
4661 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
4662 <scalar_float_mode>. Simplify.
4663 (gen_extend_conv_libfunc): Likewise.
4665 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4666 Alan Hayward <alan.hayward@arm.com>
4667 David Sherwood <david.sherwood@arm.com>
4669 * coretypes.h (scalar_float_mode): New type.
4670 * machmode.h (mode_traits::from_int): Use machine_mode if
4671 USE_ENUM_MODES is defined.
4672 (is_a): New function.
4674 (dyn_cast): Likewise.
4675 (scalar_float_mode): New class.
4676 (scalar_float_mode::includes_p): New function.
4677 (is_float_mode): Likewise.
4678 * gdbhooks.py (MachineModePrinter): New class.
4679 (build_pretty_printer): Use it for scalar_float_mode.
4680 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
4681 (format_helper::format_helper): Turn into a template.
4682 * genmodes.c (get_mode_class): New function.
4683 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
4684 or machine_mode if none.
4685 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
4686 as_a <scalar_float_mode>.
4687 * dwarf2out.c (mem_loc_descriptor): Likewise.
4688 (insert_float): Likewise.
4689 (add_const_value_attribute): Likewise.
4690 * simplify-rtx.c (simplify_immed_subreg): Likewise.
4691 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
4692 (expand_unop): Update accordingly.
4693 (expand_abs_nojump): Likewise.
4694 (expand_copysign_absneg): Take a scalar_float_mode.
4695 (expand_copysign_bit): Likewise.
4696 (expand_copysign): Update accordingly.
4698 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4699 Alan Hayward <alan.hayward@arm.com>
4700 David Sherwood <david.sherwood@arm.com>
4702 * coretypes.h (opt_mode): New class.
4703 * machmode.h (opt_mode): Likewise.
4704 (opt_mode::else_void): New function.
4705 (opt_mode::require): Likewise.
4706 (opt_mode::exists): Likewise.
4707 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
4708 (GET_MODE_2XWIDER_MODE): Likewise.
4709 (mode_iterator::get_wider): Update accordingly.
4710 (mode_iterator::get_2xwider): Likewise.
4711 (mode_iterator::get_known_wider): Likewise, turning into a template.
4712 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
4713 forcing a wider mode to exist.
4714 * config/cr16/cr16.h (LONG_REG_P): Likewise.
4715 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
4716 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
4717 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
4718 * lower-subreg.c (init_lower_subreg): Likewise.
4719 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
4720 on the final iteration.
4721 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
4722 a wider mode exists before asking for a move pattern.
4723 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
4724 forcing a wider mode to exist.
4725 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
4726 returning false if no such mode exists.
4727 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
4728 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
4729 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
4730 Avoid checking for a MODE_INT if we already know the mode is not a
4732 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
4733 forcing a wider mode to exist.
4734 (expmed_mult_highpart_optab): Likewise.
4735 (expmed_mult_highpart): Likewise.
4736 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
4738 * lto-streamer-in.c (lto_input_mode_table): Likewise.
4739 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
4740 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
4741 * internal-fn.c (expand_mul_overflow): Update use of
4742 GET_MODE_2XWIDER_MODE.
4743 * omp-low.c (omp_clause_aligned_alignment): Likewise.
4744 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
4745 GET_MODE_WIDER_MODE.
4746 (convert_plusminus_to_widen): Likewise.
4747 * tree-switch-conversion.c (array_value_type): Likewise.
4748 * var-tracking.c (emit_note_insn_var_location): Likewise.
4749 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
4750 Return false inside rather than outside the loop if no wider mode
4752 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
4753 and GET_MODE_2XWIDER_MODE
4754 (can_compare_p): Use else_void.
4755 * gdbhooks.py (OptMachineModePrinter): New class.
4756 (build_pretty_printer): Use it for opt_mode.
4758 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4759 Alan Hayward <alan.hayward@arm.com>
4760 David Sherwood <david.sherwood@arm.com>
4762 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
4763 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
4765 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4766 Alan Hayward <alan.hayward@arm.com>
4767 David Sherwood <david.sherwood@arm.com>
4769 * machmode.h (mode_traits): New structure.
4770 (get_narrowest_mode): New function.
4771 (mode_iterator::start): Likewise.
4772 (mode_iterator::iterate_p): Likewise.
4773 (mode_iterator::get_wider): Likewise.
4774 (mode_iterator::get_known_wider): Likewise.
4775 (mode_iterator::get_2xwider): Likewise.
4776 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
4777 (FOR_EACH_MODE): Likewise.
4778 (FOR_EACH_MODE_FROM): Likewise.
4779 (FOR_EACH_MODE_UNTIL): Likewise.
4780 (FOR_EACH_WIDER_MODE): Likewise.
4781 (FOR_EACH_2XWIDER_MODE): Likewise.
4782 * builtins.c (expand_builtin_strlen): Use new mode iterators.
4783 * combine.c (simplify_comparison): Likewise
4784 * config/i386/i386.c (type_natural_mode): Likewise.
4785 * cse.c (cse_insn): Likewise.
4786 * dse.c (find_shift_sequence): Likewise.
4787 * emit-rtl.c (init_derived_machine_modes): Likewise.
4788 (init_emit_once): Likewise.
4789 * explow.c (hard_function_value): Likewise.
4790 * expmed.c (extract_fixed_bit_field_1): Likewise.
4791 (extract_bit_field_1): Likewise.
4792 (expand_divmod): Likewise.
4793 (emit_store_flag_1): Likewise.
4794 * expr.c (init_expr_target): Likewise.
4795 (convert_move): Likewise.
4796 (alignment_for_piecewise_move): Likewise.
4797 (widest_int_mode_for_size): Likewise.
4798 (emit_block_move_via_movmem): Likewise.
4799 (copy_blkmode_to_reg): Likewise.
4800 (set_storage_via_setmem): Likewise.
4801 (compress_float_constant): Likewise.
4802 * omp-low.c (omp_clause_aligned_alignment): Likewise.
4803 * optabs-query.c (get_best_extraction_insn): Likewise.
4804 * optabs.c (expand_binop): Likewise.
4805 (expand_twoval_unop): Likewise.
4806 (expand_twoval_binop): Likewise.
4807 (widen_leading): Likewise.
4808 (widen_bswap): Likewise.
4809 (expand_parity): Likewise.
4810 (expand_unop): Likewise.
4811 (prepare_cmp_insn): Likewise.
4812 (prepare_float_lib_cmp): Likewise.
4813 (expand_float): Likewise.
4814 (expand_fix): Likewise.
4815 (expand_sfix_optab): Likewise.
4816 * postreload.c (move2add_use_add2_insn): Likewise.
4817 * reg-stack.c (reg_to_stack): Likewise.
4818 * reginfo.c (choose_hard_reg_mode): Likewise.
4819 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
4820 * stor-layout.c (mode_for_size): Likewise.
4821 (smallest_mode_for_size): Likewise.
4822 (mode_for_vector): Likewise.
4823 (finish_bitfield_representative): Likewise.
4824 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
4825 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
4826 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
4827 * var-tracking.c (prepare_call_arguments): Likewise.
4829 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4830 Alan Hayward <alan.hayward@arm.com>
4831 David Sherwood <david.sherwood@arm.com>
4833 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
4834 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
4835 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
4836 * machmode.h (mode_size): Move earlier in file.
4837 (mode_precision): Likewise.
4838 (mode_inner): Likewise.
4839 (mode_nunits): Likewise.
4840 (mode_unit_size): Likewise.
4841 (unit_unit_precision): Likewise.
4842 (mode_wider): Likewise.
4843 (mode_2xwider): Likewise.
4844 (machine_mode): New class.
4845 (mode_to_bytes): New function.
4846 (mode_to_bits): Likewise.
4847 (mode_to_precision): Likewise.
4848 (mode_to_inner): Likewise.
4849 (mode_to_unit_size): Likewise.
4850 (mode_to_unit_precision): Likewise.
4851 (mode_to_nunits): Likewise.
4852 (GET_MODE_SIZE): Use mode_to_bytes.
4853 (GET_MODE_BITSIZE): Use mode_to_bits.
4854 (GET_MODE_PRECISION): Use mode_to_precision.
4855 (GET_MODE_INNER): Use mode_to_inner.
4856 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
4857 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
4858 (GET_MODE_NUNITS): Use mode_to_nunits.
4859 * system.h (ALWAYS_INLINE): New macro.
4860 * config/powerpcspe/powerpcspe-c.c
4861 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
4862 int for arg1_mode and arg2_mode.
4864 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4865 Alan Hayward <alan.hayward@arm.com>
4866 David Sherwood <david.sherwood@arm.com>
4868 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
4869 Prefix mode names with E_ in case statements.
4870 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4871 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
4872 (aarch64_split_simd_move): Likewise.
4873 (aarch64_gen_storewb_pair): Likewise.
4874 (aarch64_gen_loadwb_pair): Likewise.
4875 (aarch64_gen_store_pair): Likewise.
4876 (aarch64_gen_load_pair): Likewise.
4877 (aarch64_get_condition_code_1): Likewise.
4878 (aarch64_constant_pool_reload_icode): Likewise.
4879 (get_rsqrte_type): Likewise.
4880 (get_rsqrts_type): Likewise.
4881 (get_recpe_type): Likewise.
4882 (get_recps_type): Likewise.
4883 (aarch64_gimplify_va_arg_expr): Likewise.
4884 (aarch64_simd_container_mode): Likewise.
4885 (aarch64_emit_load_exclusive): Likewise.
4886 (aarch64_emit_store_exclusive): Likewise.
4887 (aarch64_expand_compare_and_swap): Likewise.
4888 (aarch64_gen_atomic_cas): Likewise.
4889 (aarch64_emit_bic): Likewise.
4890 (aarch64_emit_atomic_swap): Likewise.
4891 (aarch64_emit_atomic_load_op): Likewise.
4892 (aarch64_evpc_trn): Likewise.
4893 (aarch64_evpc_uzp): Likewise.
4894 (aarch64_evpc_zip): Likewise.
4895 (aarch64_evpc_ext): Likewise.
4896 (aarch64_evpc_rev): Likewise.
4897 (aarch64_evpc_dup): Likewise.
4898 (aarch64_gen_ccmp_first): Likewise.
4899 (aarch64_gen_ccmp_next): Likewise.
4900 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
4901 (alpha_emit_xfloating_libcall): Likewise.
4902 (emit_insxl): Likewise.
4903 (alpha_arg_type): Likewise.
4904 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
4905 (arc_preferred_simd_mode): Likewise.
4906 (arc_secondary_reload): Likewise.
4907 (get_arc_condition_code): Likewise.
4908 (arc_print_operand): Likewise.
4909 (arc_legitimate_constant_p): Likewise.
4910 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4911 * config/arc/arc.md (casesi_load): Likewise.
4912 (casesi_compact_jump): Likewise.
4913 * config/arc/predicates.md (proper_comparison_operator): Likewise.
4914 (cc_use_register): Likewise.
4915 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4916 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
4917 (arm_init_iwmmxt_builtins): Likewise.
4918 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
4919 (neon_expand_vector_init): Likewise.
4920 (arm_attr_length_move_neon): Likewise.
4921 (maybe_get_arm_condition_code): Likewise.
4922 (arm_emit_vector_const): Likewise.
4923 (arm_preferred_simd_mode): Likewise.
4924 (arm_output_iwmmxt_tinsr): Likewise.
4925 (thumb1_output_casesi): Likewise.
4926 (thumb2_output_casesi): Likewise.
4927 (arm_emit_load_exclusive): Likewise.
4928 (arm_emit_store_exclusive): Likewise.
4929 (arm_expand_compare_and_swap): Likewise.
4930 (arm_evpc_neon_vuzp): Likewise.
4931 (arm_evpc_neon_vzip): Likewise.
4932 (arm_evpc_neon_vrev): Likewise.
4933 (arm_evpc_neon_vtrn): Likewise.
4934 (arm_evpc_neon_vext): Likewise.
4935 (arm_validize_comparison): Likewise.
4936 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
4937 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
4938 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
4939 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
4940 (c6x_preferred_simd_mode): Likewise.
4941 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
4942 (epiphany_rtx_costs): Likewise.
4943 * config/epiphany/predicates.md (proper_comparison_operator):
4945 * config/frv/frv.c (condexec_memory_operand): Likewise.
4946 (frv_emit_move): Likewise.
4947 (output_move_single): Likewise.
4948 (output_condmove_single): Likewise.
4949 (frv_hard_regno_mode_ok): Likewise.
4950 (frv_matching_accg_mode): Likewise.
4951 * config/h8300/h8300.c (split_adds_subs): Likewise.
4952 (h8300_rtx_costs): Likewise.
4953 (h8300_print_operand): Likewise.
4954 (compute_mov_length): Likewise.
4955 (output_logical_op): Likewise.
4956 (compute_logical_op_length): Likewise.
4957 (compute_logical_op_cc): Likewise.
4958 (h8300_shift_needs_scratch_p): Likewise.
4959 (output_a_shift): Likewise.
4960 (compute_a_shift_length): Likewise.
4961 (compute_a_shift_cc): Likewise.
4962 (expand_a_rotate): Likewise.
4963 (output_a_rotate): Likewise.
4964 * config/i386/i386.c (classify_argument): Likewise.
4965 (function_arg_advance_32): Likewise.
4966 (function_arg_32): Likewise.
4967 (function_arg_64): Likewise.
4968 (function_value_64): Likewise.
4969 (ix86_gimplify_va_arg): Likewise.
4970 (ix86_legitimate_constant_p): Likewise.
4971 (put_condition_code): Likewise.
4972 (split_double_mode): Likewise.
4973 (ix86_avx256_split_vector_move_misalign): Likewise.
4974 (ix86_expand_vector_logical_operator): Likewise.
4975 (ix86_split_idivmod): Likewise.
4976 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
4977 (ix86_build_const_vector): Likewise.
4978 (ix86_build_signbit_mask): Likewise.
4979 (ix86_match_ccmode): Likewise.
4980 (ix86_cc_modes_compatible): Likewise.
4981 (ix86_expand_branch): Likewise.
4982 (ix86_expand_sse_cmp): Likewise.
4983 (ix86_expand_sse_movcc): Likewise.
4984 (ix86_expand_int_sse_cmp): Likewise.
4985 (ix86_expand_vec_perm_vpermi2): Likewise.
4986 (ix86_expand_vec_perm): Likewise.
4987 (ix86_expand_sse_unpack): Likewise.
4988 (ix86_expand_int_addcc): Likewise.
4989 (ix86_split_to_parts): Likewise.
4990 (ix86_vectorize_builtin_gather): Likewise.
4991 (ix86_vectorize_builtin_scatter): Likewise.
4992 (avx_vpermilp_parallel): Likewise.
4993 (inline_memory_move_cost): Likewise.
4994 (ix86_tieable_integer_mode_p): Likewise.
4995 (x86_maybe_negate_const_int): Likewise.
4996 (ix86_expand_vector_init_duplicate): Likewise.
4997 (ix86_expand_vector_init_one_nonzero): Likewise.
4998 (ix86_expand_vector_init_one_var): Likewise.
4999 (ix86_expand_vector_init_concat): Likewise.
5000 (ix86_expand_vector_init_interleave): Likewise.
5001 (ix86_expand_vector_init_general): Likewise.
5002 (ix86_expand_vector_set): Likewise.
5003 (ix86_expand_vector_extract): Likewise.
5004 (emit_reduc_half): Likewise.
5005 (ix86_emit_i387_round): Likewise.
5006 (ix86_mangle_type): Likewise.
5007 (ix86_expand_round_sse4): Likewise.
5008 (expand_vec_perm_blend): Likewise.
5009 (canonicalize_vector_int_perm): Likewise.
5010 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
5011 (expand_vec_perm_1): Likewise.
5012 (expand_vec_perm_interleave3): Likewise.
5013 (expand_vec_perm_even_odd_pack): Likewise.
5014 (expand_vec_perm_even_odd_1): Likewise.
5015 (expand_vec_perm_broadcast_1): Likewise.
5016 (ix86_vectorize_vec_perm_const_ok): Likewise.
5017 (ix86_expand_vecop_qihi): Likewise.
5018 (ix86_expand_mul_widen_hilo): Likewise.
5019 (ix86_expand_sse2_abs): Likewise.
5020 (ix86_expand_pextr): Likewise.
5021 (ix86_expand_pinsr): Likewise.
5022 (ix86_preferred_simd_mode): Likewise.
5023 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
5024 * config/i386/sse.md (*andnot<mode>3): Likewise.
5025 (<mask_codefor><code><mode>3<mask_name>): Likewise.
5026 (*<code><mode>3): Likewise.
5027 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
5028 (ia64_expand_atomic_op): Likewise.
5029 (ia64_arg_type): Likewise.
5030 (ia64_mode_to_int): Likewise.
5031 (ia64_scalar_mode_supported_p): Likewise.
5032 (ia64_vector_mode_supported_p): Likewise.
5033 (expand_vec_perm_broadcast): Likewise.
5034 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
5035 (iq2000_function_arg_advance): Likewise.
5036 (iq2000_function_arg): Likewise.
5037 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
5038 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
5039 (m68k_libcall_value): Likewise.
5040 (m68k_function_value): Likewise.
5041 (sched_attr_op_type): Likewise.
5042 * config/mcore/mcore.c (mcore_output_move): Likewise.
5043 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
5045 (microblaze_function_arg): Likewise.
5046 * config/mips/mips.c (mips16_build_call_stub): Likewise.
5047 (mips_print_operand): Likewise.
5048 (mips_mode_ok_for_mov_fmt_p): Likewise.
5049 (mips_vector_mode_supported_p): Likewise.
5050 (mips_preferred_simd_mode): Likewise.
5051 (mips_expand_vpc_loongson_even_odd): Likewise.
5052 (mips_expand_vec_unpack): Likewise.
5053 (mips_expand_vi_broadcast): Likewise.
5054 (mips_expand_vector_init): Likewise.
5055 (mips_expand_vec_reduc): Likewise.
5056 (mips_expand_msa_cmp): Likewise.
5057 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
5058 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
5059 (cc_flags_for_mode): Likewise.
5060 * config/msp430/msp430.c (msp430_print_operand): Likewise.
5061 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
5062 (nds32_output_casesi_pc_relative): Likewise.
5063 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
5064 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
5065 (nvptx_gen_unpack): Likewise.
5066 (nvptx_gen_pack): Likewise.
5067 (nvptx_gen_shuffle): Likewise.
5068 (nvptx_gen_wcast): Likewise.
5069 (nvptx_preferred_simd_mode): Likewise.
5070 * config/pa/pa.c (pa_secondary_reload): Likewise.
5071 * config/pa/predicates.md (base14_operand): Likewise.
5072 * config/powerpcspe/powerpcspe-c.c
5073 (altivec_resolve_overloaded_builtin): Likewise.
5074 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
5076 (rs6000_preferred_simd_mode): Likewise.
5077 (output_vec_const_move): Likewise.
5078 (rs6000_expand_vector_extract): Likewise.
5079 (rs6000_split_vec_extract_var): Likewise.
5080 (reg_offset_addressing_ok_p): Likewise.
5081 (rs6000_legitimate_offset_address_p): Likewise.
5082 (rs6000_legitimize_address): Likewise.
5083 (rs6000_emit_set_const): Likewise.
5084 (rs6000_const_vec): Likewise.
5085 (rs6000_emit_move): Likewise.
5086 (spe_build_register_parallel): Likewise.
5087 (rs6000_darwin64_record_arg_recurse): Likewise.
5088 (swap_selector_for_mode): Likewise.
5089 (spe_init_builtins): Likewise.
5090 (paired_init_builtins): Likewise.
5091 (altivec_init_builtins): Likewise.
5092 (do_load_for_compare): Likewise.
5093 (rs6000_generate_compare): Likewise.
5094 (rs6000_expand_float128_convert): Likewise.
5095 (emit_load_locked): Likewise.
5096 (emit_store_conditional): Likewise.
5097 (rs6000_output_function_epilogue): Likewise.
5098 (rs6000_handle_altivec_attribute): Likewise.
5099 (rs6000_function_value): Likewise.
5100 (emit_fusion_gpr_load): Likewise.
5101 (emit_fusion_p9_load): Likewise.
5102 (emit_fusion_p9_store): Likewise.
5103 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
5104 (fusion_gpr_mem_load): Likewise.
5105 (fusion_addis_mem_combo_load): Likewise.
5106 (fusion_addis_mem_combo_store): Likewise.
5107 * config/rs6000/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/rs6000-c.c (altivec_resolve_overloaded_builtin):
5113 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
5114 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
5115 (rs6000_preferred_simd_mode): Likewise.
5116 (output_vec_const_move): Likewise.
5117 (rs6000_expand_vector_extract): Likewise.
5118 (rs6000_split_vec_extract_var): Likewise.
5119 (reg_offset_addressing_ok_p): Likewise.
5120 (rs6000_legitimate_offset_address_p): Likewise.
5121 (rs6000_legitimize_address): Likewise.
5122 (rs6000_emit_set_const): Likewise.
5123 (rs6000_const_vec): Likewise.
5124 (rs6000_emit_move): Likewise.
5125 (rs6000_darwin64_record_arg_recurse): Likewise.
5126 (swap_selector_for_mode): Likewise.
5127 (paired_init_builtins): Likewise.
5128 (altivec_init_builtins): Likewise.
5129 (rs6000_expand_float128_convert): Likewise.
5130 (emit_load_locked): Likewise.
5131 (emit_store_conditional): Likewise.
5132 (rs6000_output_function_epilogue): Likewise.
5133 (rs6000_handle_altivec_attribute): Likewise.
5134 (rs6000_function_value): Likewise.
5135 (emit_fusion_gpr_load): Likewise.
5136 (emit_fusion_p9_load): Likewise.
5137 (emit_fusion_p9_store): Likewise.
5138 * config/rx/rx.c (rx_gen_move_template): Likewise.
5139 (flags_from_mode): Likewise.
5140 * config/s390/predicates.md (s390_alc_comparison): Likewise.
5141 (s390_slb_comparison): Likewise.
5142 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
5143 (s390_vector_mode_supported_p): Likewise.
5144 (s390_cc_modes_compatible): Likewise.
5145 (s390_match_ccmode_set): Likewise.
5146 (s390_canonicalize_comparison): Likewise.
5147 (s390_emit_compare_and_swap): Likewise.
5148 (s390_branch_condition_mask): Likewise.
5149 (s390_rtx_costs): Likewise.
5150 (s390_secondary_reload): Likewise.
5151 (__SECONDARY_RELOAD_CASE): Likewise.
5152 (s390_expand_cs): Likewise.
5153 (s390_preferred_simd_mode): Likewise.
5154 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
5155 * config/sh/sh.c (sh_print_operand): Likewise.
5156 (dump_table): Likewise.
5157 (sh_secondary_reload): Likewise.
5158 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
5159 * config/sh/sh.md (casesi_worker_1): Likewise.
5160 (casesi_worker_2): Likewise.
5161 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
5162 (fcc_comparison_operator): Likewise.
5163 * config/sparc/sparc.c (sparc_expand_move): Likewise.
5164 (emit_soft_tfmode_cvt): Likewise.
5165 (sparc_preferred_simd_mode): Likewise.
5166 (output_cbranch): Likewise.
5167 (sparc_print_operand): Likewise.
5168 (sparc_expand_vec_perm_bmask): Likewise.
5169 (vector_init_bshuffle): Likewise.
5170 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
5171 (spu_vector_mode_supported_p): Likewise.
5172 (spu_expand_insv): Likewise.
5173 (spu_emit_branch_or_set): Likewise.
5174 (spu_handle_vector_attribute): Likewise.
5175 (spu_builtin_splats): Likewise.
5176 (spu_builtin_extract): Likewise.
5177 (spu_builtin_promote): Likewise.
5178 (spu_expand_sign_extend): Likewise.
5179 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
5180 (tilegx_simd_int): Likewise.
5181 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
5182 (tilepro_simd_int): Likewise.
5183 * config/v850/v850.c (const_double_split): Likewise.
5184 (v850_print_operand): Likewise.
5185 (ep_memory_offset): Likewise.
5186 * config/vax/vax.c (vax_rtx_costs): Likewise.
5187 (vax_output_int_move): Likewise.
5188 (vax_output_int_add): Likewise.
5189 (vax_output_int_subtract): Likewise.
5190 * config/visium/predicates.md (visium_branch_operator): Likewise.
5191 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
5192 (visium_print_operand_address): Likewise.
5193 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
5194 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
5195 (xtensa_expand_conditional_branch): Likewise.
5196 (xtensa_copy_incoming_a7): Likewise.
5197 (xtensa_output_literal): Likewise.
5198 * dfp.c (decimal_real_maxval): Likewise.
5199 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
5201 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5202 Alan Hayward <alan.hayward@arm.com>
5203 David Sherwood <david.sherwood@arm.com>
5205 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
5206 (mode_nunits_inline): Likewise.
5207 (mode_inner_inline): Likewise.
5208 (mode_unit_size_inline): Likewise.
5209 (mode_unit_precision_inline): Likewise.
5210 (emit_insn_modes_h): Likewise. Also emit a #define of the
5212 (emit_mode_wider): Add an E_ prefix to mode names.
5213 (emit_mode_complex): Likewise.
5214 (emit_mode_inner): Likewise.
5215 (emit_mode_adjustments): Likewise.
5216 (emit_mode_int_n): Likewise.
5217 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
5218 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
5219 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
5220 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
5221 (CRC32_BUILTIN, ENTRY): Likewise.
5222 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
5223 (aarch64_pop_regs): Likewise.
5224 (aarch64_process_components): Likewise.
5225 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
5226 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
5227 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
5228 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
5229 * config/arm/arm.c (arm_init_libfuncs): Likewise.
5230 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
5232 * config/i386/i386-builtin.def (pcmpestr): Likewise.
5233 (pcmpistr): Likewise.
5234 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
5235 * config/mmix/mmix.c (mmix_output_condition): Likewise.
5236 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
5238 * config/rl78/rl78.c (mduc_regs): Likewise.
5239 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
5240 (htm_expand_builtin): Likewise.
5241 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
5242 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
5243 * config/xtensa/xtensa.c (print_operand): Likewise.
5244 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
5245 (NUM_MODE_VECTOR_INT): Likewise.
5246 * genoutput.c (null_operand): Likewise.
5247 (output_operand_data): Likewise.
5248 * genrecog.c (print_parameter_value): Likewise.
5249 * lra.c (debug_operand_data): Likewise.
5251 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5253 * dwarf2out.c (scompare_loc_descriptor_wide)
5254 (scompare_loc_descriptor_narrow): New functions, split out from...
5255 (scompare_loc_descriptor): ...here.
5256 * expmed.c (emit_store_flag_int): New function, split out from...
5257 (emit_store_flag): ...here.
5259 2017-08-30 Richard Biener <rguenther@suse.de>
5261 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
5262 (dwarf2out_early_finish): Move setting of AT_pubnames from
5263 early debug output to early finish.
5265 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
5267 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
5268 and -mdata-region to the assembler.
5270 2017-08-30 Richard Biener <rguenther@suse.de>
5272 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
5274 (gen_subprogram_die): Add DW_AT_object_pointer only early.
5275 (dwarf2out_early_global_decl): Only generate a DIE for the
5276 abstract origin if it doesn't already exist or is a declaration DIE.
5277 (resolve_addr): Do not add the linkage name twice when
5278 generating a stub DIE for the DW_TAG_GNU_call_site target.
5280 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5282 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5283 Use machine_mode rather than int for arg1_mode.
5285 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
5288 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
5289 that the second argument of the built-in functions to unpack
5290 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
5291 switch statement instead a lot of if statements.
5292 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
5293 Allow 64-bit values to be in Altivec registers as well as
5294 traditional floating point registers.
5295 (pack<mode>, FMOVE128_VSX iterator): Likewise.
5297 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
5299 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
5300 MAX_REGS_PER_ADDRESS == 1.
5302 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
5304 * config/i386/i386.opt (flag_fentry): Do not init to -1.
5305 * config/i386/i386.c (ix86_option_override_internal): Simplify
5306 setting of opts->x_flag_entry.
5308 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5309 Jakub Jelinek <jakub@redhat.com>
5310 Richard Biener <rguenther@suse.de>
5312 PR tree-optimization/81503
5313 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
5314 folded constant fits in the target type; reorder tests for clarity.
5316 2017-08-29 Martin Liska <mliska@suse.cz>
5318 * passes.def: Include pass_lower_switch.
5319 * stmt.c (dump_case_nodes): Remove and move to
5320 tree-switch-conversion.
5321 (case_values_threshold): Likewise.
5322 (expand_switch_as_decision_tree_p): Likewise.
5323 (emit_case_decision_tree): Likewise.
5324 (expand_case): Likewise.
5325 (balance_case_nodes): Likewise.
5326 (node_has_low_bound): Likewise.
5327 (node_has_high_bound): Likewise.
5328 (node_is_bounded): Likewise.
5329 (emit_case_nodes): Likewise.
5330 (struct simple_case_node): New struct.
5331 (add_case_node): Remove.
5332 (emit_case_dispatch_table): Use vector instead of case_list.
5333 (reset_out_edges_aux): Remove.
5334 (compute_cases_per_edge): Likewise.
5335 (expand_case): Build list of simple_case_node.
5336 (expand_sjlj_dispatch_table): Use it.
5337 * tree-switch-conversion.c (struct case_node): Moved from
5338 stmt.c and adjusted.
5339 (emit_case_nodes): Likewise.
5340 (node_has_low_bound): Likewise.
5341 (node_has_high_bound): Likewise.
5342 (node_is_bounded): Likewise.
5343 (case_values_threshold): Likewise.
5344 (reset_out_edges_aux): Likewise.
5345 (compute_cases_per_edge): Likewise.
5346 (add_case_node): Likewise.
5347 (dump_case_nodes): Likewise.
5348 (balance_case_nodes): Likewise.
5349 (expand_switch_as_decision_tree_p): Likewise.
5350 (emit_jump): Likewise.
5351 (emit_case_decision_tree): Likewise.
5352 (try_switch_expansion): Likewise.
5353 (do_jump_if_equal): Likewise.
5354 (emit_cmp_and_jump_insns): Likewise.
5355 (fix_phi_operands_for_edge): New function.
5356 (record_phi_operand_mapping): Likewise.
5357 (class pass_lower_switch): New pass.
5358 (pass_lower_switch::execute): New function.
5359 (make_pass_lower_switch): Likewise.
5360 (conditional_probability):
5361 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
5362 * tree-pass.h: Add make_pass_lower_switch.
5364 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
5367 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
5370 2017-08-29 Richard Biener <rguenther@suse.de>
5372 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
5373 we do not add a DW_AT_inline attribute twice.
5374 (gen_subprogram_die): Remove code setting DW_AT_inline on
5375 DECL_ABSTRACT_P nodes.
5377 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
5379 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
5380 calls to internal functions.
5381 (gimplify_modify_expr): Likewise.
5382 * tree-call-cdce.c (use_internal_fn): Likewise.
5383 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
5384 (convert_to_divmod): Set the nothrow flag.
5385 * tree-if-conv.c (predicate_mem_writes): Likewise.
5386 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
5387 (vectorizable_call): Likewise.
5388 (vectorizable_store): Likewise.
5389 (vectorizable_load): Likewise.
5390 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
5391 (vect_recog_mask_conversion_pattern): Likewise.
5393 2017-08-29 Martin Liska <mliska@suse.cz>
5396 * gcc.c (driver_handle_option): Add new argument.
5397 * opts-common.c (handle_option): Pass
5398 target_option_override_hook.
5399 * opts-global.c (lang_handle_option): Add new option.
5400 (set_default_handlers): Add new argument.
5401 (decode_options): Likewise.
5402 * opts.c (target_handle_option): Likewise.
5403 (common_handle_option): Call target_option_override_hook.
5404 * opts.h (struct cl_option_handler_func): Add hook for
5405 target option override.
5406 (struct cl_option_handlers): Likewise.
5407 (set_default_handlers): Add new argument.
5408 (decode_options): Likewise.
5409 (common_handle_option): Likewise.
5410 (target_handle_option): Likewise.
5411 * toplev.c (toplev::main): Pass targetm.target_option.override
5414 2017-08-29 Richard Biener <rguenther@suse.de>
5415 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
5417 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
5418 life to the active subtree.
5420 2017-08-28 Jeff Law <law@redhat.com>
5422 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
5423 derive_equivalences.
5424 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
5426 (edge_info::derive_equivalences): New private member function
5428 * tree-ssa-dom.c (class edge_info): Changed from a struct
5429 to a class. Add ctor/dtor, methods and data members.
5430 (edge_info::edge_info): Renamed from allocate_edge_info.
5431 Initialize additional members.
5432 (edge_info::~edge_info): New.
5433 (free_dom_edge_info): Delete the edge info.
5434 (record_edge_info): Use new class & associated member functions.
5435 Tighten forms for testing for edge equivalences.
5436 (record_temporary_equivalences): Iterate over the simple
5437 equivalences rather than assuming there's only one per edge.
5438 (cprop_into_successor_phis): Iterate over the simple
5439 equivalences rather than assuming there's only one per edge.
5440 (optimize_stmt): Use operand_equal_p rather than pointer
5441 equality for mini-DSE code.
5443 2017-08-28 Nathan Sidwell <nathan@acm.org>
5445 * gcc.c (execute): Fold SIGPIPE handling into switch
5446 statement. Adjust internal error message.
5448 2017-08-28 Richard Biener <rguenther@suse.de>
5451 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
5452 Do nothing for removed DIEs.
5454 2017-08-28 Richard Biener <rguenther@suse.de>
5456 PR tree-optimization/81977
5457 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
5460 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
5463 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
5464 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
5465 using targetm.gen_mem_thread_fence.
5467 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
5470 * config/i386/i386.md (*<btsc><mode>): Change operand 2
5471 predicate to register_operand. Reorder operands.
5472 (*btr<mode>): Ditto.
5473 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
5474 (*btr<mode>_mask): Ditto.
5476 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
5478 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
5479 * config/rs6000/xmmintrin.h: New file.
5480 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
5482 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5485 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
5486 parameter and_insn and return it.
5487 (recombine_lvx_pattern): Insert a copy to ensure availability of
5488 the base register of the copied masking operation at the point of
5489 the instruction replacement.
5490 (recombine_stvx_pattern): Likewise.
5492 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
5494 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
5495 undocumented switches.
5496 (-mpower9-dform-vector): Likewise.
5497 (-mpower9-dform): Likewise.
5498 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
5499 comments to delete references to -mpower9-dform* switches.
5500 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
5501 Delete reference to -mpower9-dform* switches, test for
5502 -mpower9-vector instead.
5503 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
5504 (OTHER_P9_VECTOR_MASKS): Likewise.
5505 (POWERPC_MASKS): Likewise.
5506 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
5507 tests against -mpower9-dform* to -mpower9-vector. Delete code
5508 that checked for -mpower9-dform* consistancy with other options.
5509 Add test for -mpower9-misc to enable other power9 switches.
5510 (rs6000_init_hard_regno_mode_ok): Likewise.
5511 (rs6000_option_override_internal): Likewise.
5512 (rs6000_emit_prologue): Likewise.
5513 (rs6000_emit_epilogue): Likewise.
5514 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
5515 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
5516 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
5518 (emit_fusion_p9_store): Likewise.
5519 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
5520 resetting these macros if the assembler does not support ISA 3.0
5522 (TARGET_P9_DFORM_VECTOR): Likewise.
5523 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
5524 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
5526 2017-08-25 Alan Modra <amodra@gmail.com>
5528 PR rtl-optimization/81747
5529 * cse.c (cse_extended_basic_block): Don't attempt to record
5530 equivalences for degenerate conditional branches that branch
5531 to their fall-through.
5533 2017-08-24 Martin Sebor <msebor@redhat.com>
5536 * gimple-fold.c (size_must_be_zero_p): New function.
5537 (gimple_fold_builtin_memory_op): Call it.
5539 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
5541 * config/rs6000/mm_malloc.h: New file.
5543 2017-08-24 Bin Cheng <bin.cheng@arm.com>
5545 PR tree-optimization/81913
5546 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
5547 analysis when either IVs in condition can wrap.
5549 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
5551 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
5552 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
5554 2017-08-24 Richard Biener <rguenther@suse.de>
5557 * targhooks.c (default_target_can_inline_p): Properly
5558 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
5559 is present and always compare.
5560 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
5561 infer -mfpmath=sse from TARGET_SSE_P.
5562 (ix86_can_inline_p): Properly use target_option_default_node when
5563 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
5565 2017-08-24 Richard Biener <rguenther@suse.de>
5568 * dwarf2out.c (output_die): Handle flag_generate_offload like
5570 (output_comp_unit): Likewise.
5571 (gen_array_type_die): Likewise.
5572 (dwarf2out_early_finish): Likewise.
5573 (note_variable_value_in_expr): Likewise.
5574 (dwarf2out_finish): Likewise. Adjust assert.
5575 * cgraphunit.c (symbol_table::compile): Move setting of
5576 flag_generate_offload earlier ...
5577 (symbol_table::finalize_compilation_unit): ... here, before
5578 early debug finalization.
5580 2017-08-24 Richard Biener <rguenther@suse.de>
5582 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
5583 and ipa-fnsummary.h.
5584 (ix86_can_inline_p): When ix86_fpmath flags do not match
5585 check whether the callee uses FP math at all.
5587 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
5590 * tree-ssanames.c (get_nonzero_bits): Use element_precision
5591 instead of TYPE_PRECISION.
5593 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
5594 Alan Hayward <alan.hayward@arm.com>
5595 David Sherwood <david.sherwood@arm.com>
5597 * combine.c (make_extraction): Use subreg_offset_from_lsb.
5599 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
5601 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
5603 (ix86_frame::stack_realign_allocate): New field.
5604 (struct machine_frame_state): Modify comments.
5605 (machine_frame_state::sp_realigned_fp_end): New field.
5606 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
5608 (sp_valid_at): Add assertion to assure no attempt to access invalid
5609 offset of a realigned stack.
5610 (fp_valid_at): Likewise.
5611 (choose_baseaddr): Modify comments.
5612 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
5613 ix86_expand_prologue.
5614 (ix86_expand_prologue): Modify stack realignment and allocation.
5615 (ix86_expand_epilogue): Modify comments.
5616 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
5617 avx2_runtime, avx512f, and avx512f_runtime.
5619 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
5621 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
5622 (mstackrealign): Do not init to -1.
5623 * config/i386/i386.c (ix86_option_override_internal):
5624 Check opts_set, not opts when setting default value of
5625 opts->x_ix86_force_align_arg_pointer.
5627 2017-08-23 Richard Biener <rguenther@suse.de>
5629 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
5630 lang_hooks.decl_printable_name.
5631 * print-rtl-function.c (print_rtx_function): Likewise.
5632 * tree-pretty-print.c (dump_function_header): Likewise.
5634 2017-08-23 Richard Biener <rguenther@suse.de>
5637 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
5638 -g0 at compile-time.
5640 2017-08-23 Tamar Christina <tamar.christina@arm.com>
5643 * doc/sourcebuild.texi (Other hardware attributes):
5646 2017-08-23 Tamar Christina <tamar.christina@arm.com>
5649 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
5650 Fix single-use check.
5652 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
5654 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
5656 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
5658 * doc/install.texi: Modify to add more details on running selected
5661 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
5663 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
5664 is combined with -mabi=ms.
5665 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
5668 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5670 PR tree-optimization/81488
5671 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
5672 and cached_basis fields.
5673 (MAX_SPREAD): New constant.
5674 (alloc_cand_and_find_basis): Initialize new fields.
5675 (clear_visited): New function.
5676 (create_phi_basis_1): Rename from create_phi_basis, set visited
5677 and cached_basis fields.
5678 (create_phi_basis): New wrapper function.
5679 (phi_add_costs_1): Rename from phi_add_costs, add spread
5680 parameter, set visited field, short-circuit when limits reached.
5681 (phi_add_costs): New wrapper function.
5682 (record_phi_increments_1): Rename from record_phi_increments, set
5684 (record_phi_increments): New wrapper function.
5685 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
5686 (phi_incr_cost): New wrapper function.
5687 (all_phi_incrs_profitable_1): Rename from
5688 all_phi_incrs_profitable, set visited field.
5689 (all_phi_incrs_profitable): New wrapper function.
5691 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
5692 Alan Hayward <alan.hayward@arm.com>
5693 David Sherwood <david.sherwood@arm.com>
5695 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
5696 that takes the outer and inner modes.
5697 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
5698 comparison as the canonical test for a paradoxical subreg.
5699 * combine.c (simplify_set): Use paradoxical_subreg_p.
5700 (make_extraction): Likewise.
5701 (force_to_mode): Likewise.
5702 (rtx_equal_for_field_assignment_p): Likewise.
5703 (gen_lowpart_for_combine): Likewise.
5704 (simplify_comparison): Likewise.
5705 * cse.c (equiv_constant): Likewise.
5706 * expmed.c (store_bit_field_1): Likewise.
5707 * final.c (alter_subreg): Likewise.
5708 * fwprop.c (propagate_rtx): Likewise.
5709 (forward_propagate_subreg): Likewise.
5710 * ira-conflicts.c (ira_build_conflicts): Likewise.
5711 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
5712 * lra-constraints.c (curr_insn_transform): Likewise.
5713 (split_reg): Likewise.
5714 * lra-eliminations.c (move_plus_up): Likewise.
5715 (lra_eliminate_regs_1): Likewise.
5716 * recog.c (general_operand): Likewise.
5717 * ree.c (combine_reaching_defs): Likewise.
5718 * reload.c (push_reload): Likewise.
5719 (find_reloads): Likewise.
5720 * reload1.c (elimination_effects): Likewise.
5721 (compute_reload_subreg_offset): Likewise.
5722 (choose_reload_regs): Likewise.
5723 * rtlanal.c (subreg_lsb_1): Likewise.
5724 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5725 (simplify_subreg): Likewise.
5726 * var-tracking.c (track_loc_p): Likewise.
5727 * emit-rtl.c (byte_lowpart_offset): Likewise.
5728 (paradoxical_subreg_p): Delete out-of-line definition.
5730 2017-08-22 Jeff Law <law@redhat.com>
5732 PR tree-optimization/81741
5733 PR tree-optimization/71947
5734 * tree-ssa-dom.c: Include tree-inline.h.
5735 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
5736 equivalences if one is more expensive to compute than the other.
5737 * tree-ssa-scopedtables.h (class const_or_copies): Make
5738 record_const_or_copy_raw method private.
5739 (class avail_exprs_stack): New method simplify_binary_operation.
5740 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
5741 avail_exprs_stack::simplify_binary_operation as needed.
5742 (avail_exprs_stack::simplify_binary_operation): New function.
5744 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
5746 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
5747 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
5748 (DOT_SYMBOLS): Likewise.
5749 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
5750 (RELOCATABLE_NEEDS_FIXUP): Likewise.
5751 (RS6000_ABI_NAME): Likewise.
5752 (TARGET_CMODEL): Likewise.
5753 (TOC_SECTION_ASM_OP): Likewise.
5754 (SET_CMODEL): New macro.
5755 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
5757 2017-08-22 Richard Biener <rguenther@suse.de>
5759 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
5760 to allow for free-lang-data replacements similar to verify_type_variant.
5762 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
5763 Alan Hayward <alan.hayward@arm.com>
5764 David Sherwood <david.sherwood@arm.com>
5766 * config/aarch64/aarch64.md (casesi): Use DImode rather than
5767 VOIDmode for the LABEL_REF.
5769 2017-08-22 Richard Biener <rguenther@suse.de>
5771 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
5773 2017-08-22 Richard Biener <rguenther@suse.de>
5775 * common.opt (feliminate-dwarf2-dups): Ignore.
5776 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
5777 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
5778 same_die_p_wrap, compute_section_prefix,
5779 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
5780 (comdat_symbol_id, comdat_symbol_number): Likewise.
5781 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
5783 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
5784 (output_die): Mark unreachable path unreachable.
5785 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
5786 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
5787 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
5788 (dwarf2out_early_finish): Likewise.
5790 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
5792 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
5794 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
5797 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
5798 not VAR_P. Filter attribute warnings with OPT_Wattributes.
5799 (avr_attribute_table) <io, io_low, address>: Initialize
5800 .decl_required with true.
5802 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
5804 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
5805 undocumented debugging options.
5806 (-mvsx-scalar-double): Likewise.
5807 (-mallow-df-permute): Likewise.
5808 (-mvectorize-builtins): Likewise.
5809 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
5810 (rs6000_builtin_vectorized_function): Likewise.
5811 (rs6000_builtin_md_vectorized_function): Likewise.
5812 (rs6000_opt_vars): Likewise.
5814 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
5817 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
5818 (*btrq_imm): Rename from *btrq.
5819 (*btcq_imm): Rename from *btcq.
5820 (btsc): New code attribute.
5821 (*<btsc><mode>): New insn pattern.
5822 (*btr<mode>): Ditto.
5823 (*<btsc><mode>_mask): New insn_and_split pattern.
5824 (*btr<mode>_mask): Ditto.
5826 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5827 Alan Hayward <alan.hayward@arm.com>
5828 David Sherwood <david.sherwood@arm.com>
5830 * function.c (pad_below): Simplify padding calculation.
5832 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5833 Alan Hayward <alan.hayward@arm.com>
5834 David Sherwood <david.sherwood@arm.com>
5836 * target.def (function_prologue): Remove frame size argument.
5837 (function_epilogue): Likewise.
5838 * doc/tm.texi: Regenerate.
5839 * final.c (final_start_function): Update call to function_prologue.
5840 (final_end_function): Update call to function_epilogue.
5841 (default_function_pro_epilogue): Remove frame size argument.
5842 * output.h (default_function_pro_epilogue): Likewise.
5843 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
5844 (arm_output_function_prologue): Likewise.
5845 * config/frv/frv.c (frv_function_prologue): Likewise.
5846 (frv_function_epilogue): Likewise.
5847 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
5848 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
5849 (ia64_output_function_epilogue): Likewise.
5850 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
5851 (m32r_output_function_epilogue): Likewise.
5852 * config/microblaze/microblaze.c (microblaze_function_prologue)
5853 (microblaze_function_epilogue): Likewise.
5854 * config/mips/mips.c (mips_output_function_prologue): Likewise.
5855 (mips_output_function_epilogue): Likewise.
5856 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
5857 (mmix_target_asm_function_epilogue): Likewise.
5858 * config/msp430/msp430.c (msp430_start_function): Likewise.
5859 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
5860 (nds32_asm_function_epilogue): Likewise.
5861 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
5862 * config/pa/pa.c (pa_output_function_prologue): Likewise.
5863 (pa_output_function_epilogue): Likewise.
5864 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
5865 (rs6000_output_function_epilogue): Likewise.
5866 * config/rl78/rl78.c (rl78_start_function): Likewise.
5867 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
5868 (rs6000_output_function_epilogue): Likewise.
5869 * config/rx/rx.c (rx_output_function_prologue): Likewise.
5870 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
5871 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
5872 (sparc_asm_function_epilogue): Likewise.
5874 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5876 * tree.h (type_has_mode_precision_p): New function.
5877 * convert.c (convert_to_integer_1): Use it.
5878 * expr.c (expand_expr_real_2): Likewise.
5879 (expand_expr_real_1): Likewise.
5880 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
5881 * match.pd: Likewise.
5882 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
5883 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
5884 * tree-tailcall.c (process_assignment): Likewise.
5885 * tree-vect-loop.c (vectorizable_reduction): Likewise.
5886 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
5887 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
5888 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
5889 (vectorizable_assignment): Likewise.
5890 (vectorizable_shift): Likewise.
5891 (vectorizable_operation): Likewise.
5892 * tree-vrp.c (register_edge_assert_for_2): Likewise.
5894 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
5896 * match.pd: Add pow (C, x) simplification.
5898 2017-08-21 Richard Biener <rguenther@suse.de>
5900 PR tree-optimization/81900
5901 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
5902 for blocks with abnormal predecessors.
5903 (compute_antic): Do not set visited flag prematurely.
5905 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
5908 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
5910 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5912 * stor-layout.h (vector_type_mode): Move to...
5913 * tree.h (vector_type_mode): ...here.
5914 * stor-layout.c (vector_type_mode): Move to...
5915 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
5917 2017-08-21 Richard Biener <rguenther@suse.de>
5919 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
5920 register_external_die hooks.
5921 (debug_false_tree_charstarstar_uhwistar): Declare.
5922 (debug_nothing_tree_charstar_uhwi): Likewise.
5923 * debug.c (do_nothing_debug_hooks): Adjust.
5924 (debug_false_tree_charstarstar_uhwistar): New do nothing.
5925 (debug_nothing_tree_charstar_uhwi): Likewise.
5926 * dbxout.c (dbx_debug_hooks): Adjust.
5927 (xcoff_debug_hooks): Likewise.
5928 * sdbout.c (sdb_debug_hooks): Likewise.
5929 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
5930 * dwarf2out.c (macinfo_label_base): New global.
5931 (dwarf2out_register_external_die): New function for the
5932 register_external_die hook.
5933 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
5934 (dwarf2_debug_hooks): Use them.
5935 (dwarf2_lineno_debug_hooks): Adjust.
5936 (struct die_struct): Add with_offset flag.
5937 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
5938 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
5939 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
5940 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
5941 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
5942 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
5943 defining section names for the early LTO debug variants.
5944 (reset_indirect_string): New helper.
5945 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
5946 (print_dw_val): Add support for offsetted symbol references.
5947 (get_ultimate_context): Split out from is_cxx.
5948 (is_cxx): Use get_ultimate_context.
5949 (is_fortran): Add decl overload.
5950 (compute_comp_unit_symbol): Split out worker from
5951 compute_section_prefix.
5952 (compute_section_prefix): Call compute_comp_unit_symbol and
5953 set comdat_type_p here.
5954 (output_die): Skip DIE symbol output for the LTO added one.
5955 Handle DIE symbol references with offset.
5956 (output_comp_unit): Guard section name mangling properly.
5957 For LTO debug sections emit a symbol at the section beginning
5958 which we use to refer to its DIEs.
5959 (add_abstract_origin_attribute): For DIEs registered via
5960 dwarf2out_register_external_die directly refer to the early
5961 DIE rather than indirectly through the shadow one we created.
5962 Remove obsolete call to dwarf2out_abstract_function for
5963 non-function/block origins.
5964 (gen_array_type_die): When generating early LTO debug do
5965 not emit DW_AT_string_length.
5966 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
5967 late when in LTO. As suggested place a gcc_unreachable for
5968 the DECL_ABSTRACT_P case.
5969 (gen_subprogram_die): Avoid another specification DIE
5970 for early built declarations/definitions for the late LTO case.
5971 (gen_variable_die): Add type references for late duplicated VLA dies
5973 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
5974 we have the abstract instance already.
5975 (process_scope_var): Adjust decl DIE contexts in LTO which
5976 first puts them in limbo.
5977 (gen_decl_die): Do not generate type DIEs late apart from
5978 types for VLAs or for decls we do not yet have a DIE. Do not
5979 call dwarf2out_abstract_function late.
5980 (dwarf2out_early_global_decl): Make sure to create DIEs
5981 for abstract instances of a decl first.
5982 (dwarf2out_late_global_decl): Adjust comment.
5983 (output_macinfo_op): With multiple macro sections use
5984 macinfo_label_base to distinguish labels.
5985 (output_macinfo): Likewise. Update macinfo_label_base.
5986 Pass in the line info label.
5987 (note_variable_value_in_expr): When generating LTO resolve
5988 all variable values here by generating DIEs as needed.
5989 (init_sections_and_labels): Add early LTO debug flag parameter
5990 and generate different sections and names if set. Add generation
5991 counter for the labels so we can have multiple of them.
5992 (reset_dies): Helper to allow DIEs to be output multiple times.
5993 (dwarf2out_finish): When outputting DIEs to the fat part of an
5994 LTO object first reset DIEs.
5995 (dwarf2out_early_finish): Output early DIEs when generating LTO.
5996 (modified_type_die): Check for decl_ultimate_origin being self
5998 (gen_type_die_with_usage): Likewise.
5999 (gen_typedef_die): Allow decl_ultimate_origin being self.
6000 (set_decl_abstract_flags): Remove.
6001 (set_block_abstract_flags): Likewise.
6002 (dwarf2out_abstract_function): Treat the early generated DIEs
6003 as the abstract copy and only add DW_AT_inline and
6004 DW_AT_artificial here and call set_decl_origin_self.
6005 If the DIE has an abstract origin don't do anything.
6006 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
6007 if we have none yet (Go fails to build one, PR78628).
6008 (variably_modified_type_p): Prevent endless recursion for Ada
6009 cyclic pointer types.
6010 * lto-streamer-in.c: Include debug.h.
6011 (dref_queue): New global.
6012 (lto_read_tree_1): Stream in DIE references.
6013 (lto_input_tree): Register DIE references.
6014 (input_function): Stream DECL_DEBUG_ARGS.
6015 * lto-streamer-out.c: Include debug.h.
6016 (lto_write_tree_1): Output DIE references.
6017 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
6018 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
6019 (output_function): Stream DECL_DEBUG_ARGS.
6020 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
6021 Stream DECL_ABSTRACT_ORIGIN.
6022 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
6023 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
6024 DECL_CONTEXT for file-scope decls.
6025 * lto-streamer.h (struct dref_entry): Declare.
6026 (dref_queue): Likewise.
6027 * cfgexpand.c (pass_expand::execute): Do not call the
6028 outlining_inline_function hook here.
6029 * lto-wrapper.c (debug_obj): New global.
6030 (tool_cleanup): Unlink it if required.
6031 (debug_objcopy): New function.
6032 (run_gcc): Handle early debug sections in the IL files by
6033 extracting them to separate files, partially linkin them and
6034 feeding the result back as result to the linker.
6035 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
6036 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
6037 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
6038 sections into a separate segment.
6039 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
6041 (darwin_asm_dwarf_section): Likewise.
6042 (darwin_asm_output_dwarf_offset): Likewise.
6043 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
6045 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
6046 Alan Hayward <alan.hayward@arm.com>
6047 David Sherwood <david.sherwood@arm.com>
6049 * read-md.h (md_reader::record_potential_iterator_use): Replace
6050 pointer argument with an rtx and an index.
6051 * read-rtl.c (iterator_group::apply_iterator): Likewise.
6052 (apply_mode_iterator): Likewise.
6053 (apply_code_iterator): Likewise.
6054 (apply_int_iterator): Likewise.
6055 (apply_subst_iterator): Likewise.
6056 (record_iterator_use): Likewise.
6057 (record_attribute_use): Likewise.
6058 (md_reader::record_potential_iterator_use): Likewise. Update calls
6059 to record_iterator_use and apply_iterator.
6060 (iterator_use): Replace ptr with x and index.
6061 (attribute_use): Likewise.
6062 (apply_attribute_uses): Update calls to apply_iterator.
6063 (apply_iterators): Likewise. Update initialization of iterator_use.
6064 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
6065 and record_potential_iterator_use.
6066 (rtx_reader::read_rtx_operand): Likewise.
6068 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
6069 Alan Hayward <alan.hayward@arm.com>
6070 David Sherwood <david.sherwood@arm.com>
6072 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
6075 2017-08-21 Richard Biener <rguenther@suse.de>
6078 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
6079 at struct end conservatively when comparing common bases.
6081 2017-08-21 Richard Biener <rguenther@suse.de>
6083 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
6084 (mem_ref_in_stmt): Remove.
6085 (determine_max_movement): Use ref index to get at the reference.
6086 (invariantness_dom_walker::before_dom_children): Deal with
6087 lim data already initialized.
6088 (gather_mem_refs_stmt): Initialize lim data and record ref index.
6090 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
6092 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
6093 (TARGET_ISA_ROUND): Ditto.
6094 (TARGET_ROUND): Ditto.
6095 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
6096 * config/i386/i386.md: Ditto.
6097 * config/i386/sse.md: Ditto.
6098 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
6099 with OPTION_MASK_ISA_SSE4_1.
6101 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
6104 * doc/extend.texi (x86 Built-in Functions): Correct the name of
6105 __builtin_ia32_lzcnt_u16.
6107 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
6110 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
6111 (rs6000_set_current_function): Rewrite function to use it.
6113 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
6116 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
6117 and TYPE_WARN_IF_NOT_ALIGN.
6118 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
6119 (handle_warn_if_not_align): New.
6120 (place_union_field): Call handle_warn_if_not_align.
6121 (place_field): Call handle_warn_if_not_align.
6122 Copy TYPE_WARN_IF_NOT_ALIGN.
6123 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
6124 (layout_type): Likewise.
6125 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
6127 (tree_decl_common): Add warn_if_not_align.
6128 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
6129 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
6130 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
6131 (DECL_WARN_IF_NOT_ALIGN): Likewise.
6132 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
6133 * doc/extend.texi: Document warn_if_not_aligned attribute.
6134 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
6136 2017-08-17 Martin Liska <mliska@suse.cz>
6139 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
6140 (get_data_dependence): Use it as pointer type.
6141 (distribute_loop): Likewise.
6143 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6145 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
6146 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
6147 (p8_vmrgow_<mode>_direct): New define_insn.
6148 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
6149 handle endianness for vmrgew and vmrgow permute patterns.
6151 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
6153 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
6154 * config/rs6000/rs6000-cpus.def: Remove comment.
6155 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
6156 (POWERPC_MASKS): Likewise.
6157 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
6158 use of TARGET_VSX_TIMODE.
6159 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
6160 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
6161 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
6162 (rs6000_option_override_internal): Remove dead code.
6163 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
6164 (rs6000_legitimize_reload_address): Likewise.
6165 (rs6000_legitimate_address_p): Likewise.
6166 (rs6000_opt_masks): Delete "vsx-timode".
6167 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
6168 from function comment.
6169 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
6170 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
6171 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
6173 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
6174 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
6175 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
6178 (define_peephole2): Likewise.
6180 2017-08-17 Martin Sebor <msebor@redhat.com>
6183 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
6184 past the end of an array.
6185 (test_pp_format): Add test cases.
6187 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
6189 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
6190 missing ECF_NOTHROW flags.
6192 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
6195 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
6196 operands residing in integer registers.
6197 (*vsx_le_perm_load_<mode>): Likewise.
6198 (*vsx_le_perm_store_<mode>): Likewise.
6199 (define_peephole2): Add peepholes to optimize the above.
6201 2017-08-17 Marek Polacek <polacek@redhat.com>
6204 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
6205 to mimic what shorten_compare did. Change the return type to bool.
6206 (fold_cond_expr_with_comparison): Update call to
6207 operand_equal_for_comparison_p.
6208 (fold_ternary_loc): Likewise.
6210 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
6212 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
6214 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
6216 2017-08-17 Richard Biener <rguenther@suse.de>
6218 * tree-ssa-structalias.c (solve_graph): When propagating
6219 to successors update the graphs succ edges and avoid duplicate work.
6221 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
6224 * config/i386/i386.c (ix86_option_override_internal): Save target
6225 specific options after ix86_stack_protector_guard_reg was changed.
6227 2017-08-17 Richard Biener <rguenther@suse.de>
6229 PR tree-optimization/81827
6230 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
6231 (new_var_info): Initialize it conservatively.
6232 (get_call_vi): Mark register vars.
6233 (new_scalar_tmp_constraint_exp): Likewise.
6234 (handle_rhs_call): Likewise.
6235 (handle_const_call): Likewise.
6236 (create_function_info_for): Likewise.
6237 (solve_constraints): Sort varinfos to separate register from
6238 non-register vars to pack points-to solution bitmaps during
6241 2017-08-17 Marek Polacek <polacek@redhat.com>
6243 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
6245 2017-08-17 Richard Biener <rguenther@suse.de>
6247 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
6248 to true when overflow is undefined and we saturated the result.
6250 2017-08-17 Alan Modra <amodra@gmail.com>
6253 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
6254 Don't use store multiple if only one reg needs saving.
6255 (interesting_frame_related_regno): New function.
6256 (rs6000_frame_related): Don't emit frame info for regs that
6258 (rs6000_emit_epilogue): Likewise.
6260 2017-08-16 Nathan Sidwell <nathan@acm.org>
6262 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
6263 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
6264 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
6265 * tree.c (free_lang_data_in_type): Use else-if chain. Always
6266 clear TYPE_LANG_1. Remove obsolete member-function stripping.
6267 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
6268 (verify_type): Adjust for TYPE_BINFO move.
6269 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
6270 process TYPE_BINFO directly.
6271 (hash_tree): Likewise.
6272 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
6274 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
6277 2017-08-16 David Malcolm <dmalcolm@redhat.com>
6279 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
6281 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
6284 * config/i386/i386.md (*anddi_1_btr): Change predicates of
6285 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
6286 Add ix86_binary_operator_ok to insn constraint.
6287 (*iordi_1_bts): Ditto.
6288 (*xordi_1_btc): Ditto.
6289 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
6290 Update corresponding peephole2 pattern.
6294 2017-08-16 Bin Cheng <bin.cheng@arm.com>
6296 PR tree-optimization/81832
6297 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
6298 copy loop header which has IFN_LOOP_DIST_ALIAS call.
6300 2017-08-16 Marek Polacek <polacek@redhat.com>
6303 * fold-const.c (fold_indirect_ref_1): Restore original behavior
6304 regarding size_zero_node.
6306 2017-08-16 Martin Liska <mliska@suse.cz>
6309 * config.gcc: Respect previously set extra_objs in case
6312 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
6314 PR tree-optimization/81835
6315 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
6316 the phi SSA_NAME. Check that the condition in a COND_EXPR does
6317 not depend on the phi.
6319 2017-08-16 Alan Modra <amodra@gmail.com>
6321 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
6324 2017-08-16 Alan Modra <amodra@gmail.com>
6326 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
6327 (save_reg_p): ..into this. Update all callers.
6328 (first_reg_to_save): Simplify.
6330 2017-08-16 Alan Modra <amodra@gmail.com>
6332 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
6335 2017-08-15 Joseph Myers <joseph@codesourcery.com>
6339 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
6340 constant count if that count is less than 32.
6342 2017-08-15 Nathan Sidwell <nathan@acm.org>
6344 * gcc.c (execute): Emit friendlier message if inferior is killed
6345 by an external cause.
6347 2017-08-15 Richard Biener <rguenther@suse.de>
6349 PR tree-optimization/81790
6350 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
6351 CONSTRUCTORs from simplifying and VN.
6353 2017-08-14 Martin Sebor <msebor@redhat.com>
6355 * builtin-attrs.def: Add comments.
6357 2017-08-14 Martin Sebor <msebor@redhat.com>
6360 * doc/extend.texi (attribute nonstring): Document new attribute.
6362 2017-08-14 Martin Sebor <msebor@redhat.com>
6365 * tree-diagnostic.c (default_tree_printer): Handle %G.
6366 * gimple-pretty-print.h (percent_G_format): Declare new function.
6367 * gimple-pretty-print.c (percent_G_format): Define.
6368 * tree-pretty-print.c (percent_K_format): Add argument.
6370 2017-08-14 Martin Sebor <msebor@redhat.com>
6372 PR translation/79998
6373 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
6374 Remove a stray space.
6376 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
6379 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
6380 (*iordi_1_bts): Ditto.
6381 (*xordi_1_btc): Ditto.
6383 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6386 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
6387 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6389 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
6390 quoted strings, and make more translator-friendly.
6391 (darwin_rs6000_override_options): Likewise.
6392 (rs6000_option_override_internal): Likewise.
6393 (rs6000_return_in_memory): Fix overlong line.
6394 (init_cmulative_args): Use quoted strings, and make more
6395 translator-friendly.
6396 (rs6000_pass_by_reference): Fix overlong line.
6397 (def_builtin): Use quoted strings.
6398 (altivec_expand_predicate_builtin): Use quoted strings, and make
6399 more translator-friendly.
6400 (htm_expand_builtin): Use quoted strings.
6401 (cpu_expand_builtin): Use quoted strings, and make more
6402 translator-friendly.
6403 (altivec_expand_builtin): Likewise.
6404 (paired_expand_predicate_builtin): Likewise.
6405 (rs6000_invalid_builtin): Likewise.
6406 (builtin_function_type): Use quoted strings.
6407 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
6408 more translator-friendly.
6409 (rs6000_trampoline_init): Likewise.
6410 (rs6000_handle_altivec_attribute): Likewise.
6411 (rs6000_inner_target_options): Use quoted strings.
6412 (rs6000_disable_incompatible_switches): Likewise.
6413 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
6414 strings, and make more translator-friendly.
6415 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
6417 2017-08-14 Bin Cheng <bin.cheng@arm.com>
6419 PR tree-optimization/81799
6420 * tree-loop-distribution.c (version_loop_by_alias_check): Force
6421 cond_expr to simple gimple operand.
6423 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
6426 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
6428 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
6432 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
6433 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
6434 TARGET_GASISR_PROLOGUES.
6435 * config/avr/avr.c (avr_option_override): Same.
6436 (avr_pass_pre_proep::execute): Same.
6438 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
6441 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
6442 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
6444 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
6446 * config/i386/i386.md (*load_tp_<mode>): Redefine as
6447 define_insn_and_split. Split to a memory load from 0 in
6448 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
6449 using PTR mode iterator.
6450 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
6451 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
6452 (*add_tp_<mode>): Redefine as define_insn_and_split.
6453 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
6454 address space. Merge with *add_tp_x32 using PTR mode iterator.
6455 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
6456 Split to an add with a memory load from 0 in
6457 DEFAULT_TLS_SEG_REG address space.
6459 2017-08-12 Andrew Pinski <apinski@cavium.com>
6461 * config/aarch64/aarch64-option-extensions.def (rdma):
6462 Fix feature string to what Linux prints out in /proc/cpuinfo.
6464 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
6467 * dwarf2out.c (modified_type_die): For C typedef types that have
6468 an ultimate origin, process the ultimate origin instead of the
6470 (gen_typedef_die): Assert that input DECLs have no ultimate
6472 (gen_type_die_with_usage): For typedef variants that have an
6473 ultimate origin, just call gen_decl_die on the original DECL.
6474 (process_scope_var): Avoid creating DIEs for local typedefs and
6475 concrete static variables.
6477 2017-08-12 Alan Modra <amodra@gmail.com>
6481 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
6482 match gnu-user.h startfile.
6483 (ENDFILE_LINUX_SPEC): Similarly.
6485 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
6488 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
6490 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
6492 2017-08-11 Tamar Christina <tamar.christina@arm.com>
6493 * config/aarch64/aarch64.md (mov<mode>): Change.
6494 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
6495 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
6496 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
6498 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
6500 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
6501 for storage order barriers.
6503 2017-08-11 Martin Liska <mliska@suse.cz>
6506 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
6507 variables of void type.
6509 2017-08-11 Martin Liska <mliska@suse.cz>
6511 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
6512 TARGET_SUPPORTS_ALIASES.
6513 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
6514 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
6515 (optimize_weakref): Likewise.
6516 * symtab.c (symtab_node::noninterposable_alias): Likewise.
6517 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
6518 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
6520 2017-08-11 Martin Liska <mliska@suse.cz>
6523 * config/i386/i386.c (make_resolver_func): Do complete
6524 refactoring of the function.
6526 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
6529 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
6530 * config/i386/i386.c (ix86_stack_protect_guard): Use
6531 ix86_stack_protect_guard_symbol_str to generate varible declaration.
6532 * doc/invoke.texi (x86 Options): Document
6533 -mstack-protector-guard-symbol= option.
6535 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
6537 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
6538 * config/i386/i386.c (ix86_split_stack_guard): New function.
6539 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
6540 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
6541 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
6542 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
6543 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
6544 (split_stack_space_check): Call ix86_split_stack_guard.
6546 2017-08-10 Martin Sebor <msebor@redhat.com>
6548 * print-tree.c (print_node): Print location using the established
6550 Replace spaces with colons.
6551 (debug_raw, debug): Ditto.
6553 2017-08-10 Martin Sebor <msebor@redhat.com>
6556 * pretty-print.c (pp_format): Correct the handling of %s precision.
6558 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
6561 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
6563 (ix86_finalize_stack_frame_flags): This. Also clear
6564 frame_pointer_needed if -fno-omit-frame-pointer is used without
6566 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
6567 with ix86_finalize_stack_frame_flags.
6568 (ix86_expand_epilogue): Likewise.
6569 (ix86_expand_split_stack_prologue): Likewise.
6570 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
6572 2017-08-10 Martin Liska <mliska@suse.cz>
6575 * c-attribs.c (handle_target_attribute):
6576 Report warning for an empty string argument of target attribute.
6578 2017-08-09 Jakub Jelinek <jakub@redhat.com>
6581 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
6583 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
6584 or DECL_NONLOCAL labels.
6585 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
6586 or DECL_NONLOCAL labels here.
6588 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
6590 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
6591 to indicate when early gimple folding has been disabled.
6592 (rs6000_gimple_fold_builtin): Add debug content.
6593 (rs6000_invalid_builtin): Fix whitespace.
6594 (rs6000_expand_builtin): Fix whitespace.
6595 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
6597 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
6600 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
6601 SAVE_MULTIPLE if not all the registers that saves, should be saved.
6603 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
6605 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
6606 (qdf24xx): Likewise.
6607 * config/aarch64/aarch64.md: Include falkor.md.
6608 * config/aarch64/falkor.md: New.
6610 2017-08-09 Marek Polacek <polacek@redhat.com>
6613 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
6614 * diagnostic.c (emit_diagnostic): Add a comment.
6615 (emit_diagnostic_valist): New function.
6617 2017-08-09 Marek Polacek <polacek@redhat.com>
6620 * input.c (make_location): New overload.
6621 * input.h (make_location): Declare.
6623 2017-08-08 Alan Modra <amodra@gmail.com>
6624 H.J. Lu <hongjiu.lu@intel.com>
6627 * gcc.c (NO_PIE_SPEC): Delete.
6628 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
6630 (LINK_PIE_SPEC): ..to here.
6631 (LINK_COMMAND_SPEC): Support -no-pie.
6632 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
6633 chain of crtbegin*.o selection, update for PIE_SPEC changes and
6635 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
6636 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
6637 (ENDFILE_CRTEND_SPEC): Similarly.
6639 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
6642 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
6643 (mstack-protector-guard-offset=): Ditto.
6644 * config/i386/i386.c (ix86_option_override): Handle
6645 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
6647 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
6648 ix86_stack_protect_guard_offset variables.
6649 (TARGET_STACK_PROTECT_GUARD): Always define.
6650 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
6651 and -mstack-protector-guard-offset= options.
6653 2017-08-08 Bin Cheng <bin.cheng@arm.com>
6655 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
6656 boundary case for the last candidate.
6658 2017-08-08 Bin Cheng <bin.cheng@arm.com>
6660 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
6661 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
6663 2017-08-08 Tamar Christina <tamar.christina@arm.com>
6666 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
6667 * config/aarch64/aarch64-builtins.c
6668 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
6669 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
6670 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
6672 2017-08-08 Tamar Christina <tamar.christina@arm.com>
6673 Andrew Pinski <pinskia@gmail.com>
6676 * internal-fn.def (XORSIGN): New.
6677 * optabs.def (xorsign_optab): New.
6678 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
6679 (convert_expand_mult_copysign): New.
6680 (pass_optimize_widening_mul::execute): Call
6681 convert_expand_mult_copysign.
6683 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6685 PR tree-optimization/81354
6686 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
6687 Insert on edges rather than explicitly creating landing pads.
6688 (analyze_candidates_and_replace): Commit edge inserts.
6690 2017-08-08 Richard Biener <rguenther@suse.de>
6693 * tree-ssa-loop-niter.c: Include tree-dfa.h.
6694 (expand_simple_operations): Also look through ADDR_EXPRs with
6695 MEM_REF bases treating them as POINTER_PLUS_EXPR.
6697 2017-08-08 Richard Biener <rguenther@suse.de>
6699 PR tree-optimization/81723
6700 * tree-vect-slp.c (struct bst_traits): New hash traits.
6701 (bst_fail): New global.
6702 (vect_build_slp_tree_2): New worker, split out from ...
6703 (vect_build_slp_tree): ... this now wrapping it with using
6704 bst_fail set to cache SLP tree build fails. Properly handle
6706 (vect_analyze_slp_instance): Allocate and free bst_fail.
6708 2017-08-08 Martin Liska <mliska@suse.cz>
6711 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
6712 LABEL_DECLs that can be from a different function.
6714 2017-08-08 Bin Cheng <bin.cheng@arm.com>
6716 PR tree-optimization/81744
6717 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
6718 loop's number of iterations.
6720 2017-08-08 Martin Liska <mliska@suse.cz>
6722 * asan.c: Include header files.
6723 * attribs.c (build_decl_attribute_variant): New function moved
6725 (build_type_attribute_qual_variant): Likewise.
6726 (cmp_attrib_identifiers): Likewise.
6727 (simple_cst_list_equal): Likewise.
6728 (omp_declare_simd_clauses_equal): Likewise.
6729 (attribute_value_equal): Likewise.
6730 (comp_type_attributes): Likewise.
6731 (build_type_attribute_variant): Likewise.
6732 (lookup_ident_attribute): Likewise.
6733 (remove_attribute): Likewise.
6734 (merge_attributes): Likewise.
6735 (merge_type_attributes): Likewise.
6736 (merge_decl_attributes): Likewise.
6737 (merge_dllimport_decl_attributes): Likewise.
6738 (handle_dll_attribute): Likewise.
6739 (attribute_list_equal): Likewise.
6740 (attribute_list_contained): Likewise.
6741 * attribs.h (lookup_attribute): New function moved from tree.[ch].
6742 (lookup_attribute_by_prefix): Likewise.
6743 * bb-reorder.c: Include header files.
6744 * builtins.c: Likewise.
6745 * calls.c: Likewise.
6746 * cfgexpand.c: Likewise.
6747 * cgraph.c: Likewise.
6748 * cgraphunit.c: Likewise.
6749 * convert.c: Likewise.
6750 * dwarf2out.c: Likewise.
6751 * final.c: Likewise.
6752 * fold-const.c: Likewise.
6753 * function.c: Likewise.
6754 * gimple-expr.c: Likewise.
6755 * gimple-fold.c: Likewise.
6756 * gimple-pretty-print.c: Likewise.
6757 * gimple.c: Likewise.
6758 * gimplify.c: Likewise.
6759 * hsa-common.c: Likewise.
6760 * hsa-gen.c: Likewise.
6761 * internal-fn.c: Likewise.
6762 * ipa-chkp.c: Likewise.
6763 * ipa-cp.c: Likewise.
6764 * ipa-devirt.c: Likewise.
6765 * ipa-fnsummary.c: Likewise.
6766 * ipa-inline.c: Likewise.
6767 * ipa-visibility.c: Likewise.
6769 * lto-cgraph.c: Likewise.
6770 * omp-expand.c: Likewise.
6771 * omp-general.c: Likewise.
6772 * omp-low.c: Likewise.
6773 * omp-offload.c: Likewise.
6774 * omp-simd-clone.c: Likewise.
6775 * opts-global.c: Likewise.
6776 * passes.c: Likewise.
6777 * predict.c: Likewise.
6778 * sancov.c: Likewise.
6779 * sanopt.c: Likewise.
6780 * symtab.c: Likewise.
6781 * toplev.c: Likewise.
6782 * trans-mem.c: Likewise.
6783 * tree-chkp.c: Likewise.
6784 * tree-eh.c: Likewise.
6785 * tree-into-ssa.c: Likewise.
6786 * tree-object-size.c: Likewise.
6787 * tree-parloops.c: Likewise.
6788 * tree-profile.c: Likewise.
6789 * tree-ssa-ccp.c: Likewise.
6790 * tree-ssa-live.c: Likewise.
6791 * tree-ssa-loop.c: Likewise.
6792 * tree-ssa-sccvn.c: Likewise.
6793 * tree-ssa-structalias.c: Likewise.
6794 * tree-ssa.c: Likewise.
6795 * tree-streamer-in.c: Likewise.
6796 * tree-vectorizer.c: Likewise.
6797 * tree-vrp.c: Likewise.
6799 * ubsan.c: Likewise.
6800 * varasm.c: Likewise.
6801 * varpool.c: Likewise.
6802 * tree.c: Remove functions moved to attribs.[ch].
6804 * config/aarch64/aarch64.c: Add attrs.h header file.
6805 * config/alpha/alpha.c: Likewise.
6806 * config/arc/arc.c: Likewise.
6807 * config/arm/arm.c: Likewise.
6808 * config/avr/avr.c: Likewise.
6809 * config/bfin/bfin.c: Likewise.
6810 * config/c6x/c6x.c: Likewise.
6811 * config/cr16/cr16.c: Likewise.
6812 * config/cris/cris.c: Likewise.
6813 * config/darwin.c: Likewise.
6814 * config/epiphany/epiphany.c: Likewise.
6815 * config/fr30/fr30.c: Likewise.
6816 * config/frv/frv.c: Likewise.
6817 * config/ft32/ft32.c: Likewise.
6818 * config/h8300/h8300.c: Likewise.
6819 * config/i386/winnt.c: Likewise.
6820 * config/ia64/ia64.c: Likewise.
6821 * config/iq2000/iq2000.c: Likewise.
6822 * config/lm32/lm32.c: Likewise.
6823 * config/m32c/m32c.c: Likewise.
6824 * config/m32r/m32r.c: Likewise.
6825 * config/m68k/m68k.c: Likewise.
6826 * config/mcore/mcore.c: Likewise.
6827 * config/microblaze/microblaze.c: Likewise.
6828 * config/mips/mips.c: Likewise.
6829 * config/mmix/mmix.c: Likewise.
6830 * config/mn10300/mn10300.c: Likewise.
6831 * config/moxie/moxie.c: Likewise.
6832 * config/msp430/msp430.c: Likewise.
6833 * config/nds32/nds32-isr.c: Likewise.
6834 * config/nds32/nds32.c: Likewise.
6835 * config/nios2/nios2.c: Likewise.
6836 * config/nvptx/nvptx.c: Likewise.
6837 * config/pa/pa.c: Likewise.
6838 * config/pdp11/pdp11.c: Likewise.
6839 * config/powerpcspe/powerpcspe.c: Likewise.
6840 * config/riscv/riscv.c: Likewise.
6841 * config/rl78/rl78.c: Likewise.
6842 * config/rx/rx.c: Likewise.
6843 * config/s390/s390.c: Likewise.
6844 * config/sh/sh.c: Likewise.
6845 * config/sol2.c: Likewise.
6846 * config/sparc/sparc.c: Likewise.
6847 * config/spu/spu.c: Likewise.
6848 * config/stormy16/stormy16.c: Likewise.
6849 * config/tilegx/tilegx.c: Likewise.
6850 * config/tilepro/tilepro.c: Likewise.
6851 * config/v850/v850.c: Likewise.
6852 * config/vax/vax.c: Likewise.
6853 * config/visium/visium.c: Likewise.
6854 * config/xtensa/xtensa.c: Likewise.
6856 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
6859 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
6860 constraints since the -mupper-regs-* switches have been
6862 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
6863 into a vector from a double word element that was extracted from
6864 another vector, and eliminate extra XXPERMDI instructions.
6865 (vsx_concat_<mode>_2): Likewise.
6866 (vsx_concat_<mode>_3): Likewise.
6867 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
6868 concat to allow optimizing inserts from previous extracts.
6870 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
6872 * config/i386/i386.c (ix86_stack_protect_guard): Generate
6873 memory reference to a SSP offset in TLS address space.
6874 (ix86_print_operand) <case '@'>: Remove.
6875 (ix86_print_operand_punct_valid_p): Remove '@' code.
6876 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
6878 (stack_tls_protect_set_<mode>): Remove.
6879 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
6880 (stack_tls_protect_test_<mode>): Remove.
6881 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
6883 2017-08-07 Olivier Hainque <hainque@adacore.com>
6886 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
6888 2017-08-07 Douglas Rupp <rupp@adacore.com>
6890 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
6891 variable was referenced as multidir in command.
6893 2017-08-07 Jakub Jelinek <jakub@redhat.com>
6896 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
6899 2017-08-07 Martin Liska <mliska@suse.cz>
6901 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
6902 * config/rl78/rl78.c: Add include of attribs.h.
6903 * config/sh/sh.c: Likewise.
6904 * config/v850/v850.c: Likewise.
6906 2017-08-07 Tom de Vries <tom@codesourcery.com>
6909 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
6911 2017-08-07 Martin Liska <mliska@suse.cz>
6913 * config/mips/mips.c: Include attribs.h.
6915 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
6918 * doc/invoke.texi: Document change in behvaior for -Ofast for
6921 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
6923 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
6925 (aarch64_pop_regs): Likewise.
6926 (aarch64_gen_load_pair): Likewise.
6927 (aarch64_save_callee_saves): Likewise.
6928 (aarch64_restore_callee_saves): Likewise.
6930 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
6932 * config/i386/i386.c: Revert the last change.
6934 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
6937 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
6939 (ix86_finalize_stack_frame_flags): This. Also clear
6940 frame_pointer_needed if -fno-omit-frame-pointer is used without
6942 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
6943 with ix86_finalize_stack_frame_flags.
6944 (ix86_expand_epilogue): Likewise.
6945 (ix86_expand_split_stack_prologue): Likewise.
6947 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
6950 * config/i386/i386.c (get_builtin_code_for_version): Set priority
6951 to P_AES for Westmere.
6953 2017-08-07 Jonathan Yong <10walls@gmail.com>
6955 * config/i386/mingw.opt (fset-stack-executable): Removed.
6956 * config/i386/cygming.opt (fset-stack-executable): Moved
6958 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
6960 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
6962 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
6964 2017-08-07 Marek Polacek <polacek@redhat.com>
6967 * fold-const.c (fold_indirect_ref_1): Check type_domain.
6969 2017-08-07 Martin Liska <mliska@suse.cz>
6971 * attribs.h (canonicalize_attr_name): New function.
6972 (cmp_attribs): Move from c-format.c and adjusted.
6973 (is_attribute_p): Moved from tree.h.
6974 * tree-inline.c: Add new includes.
6975 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
6976 (private_is_attribute_p): Remove.
6977 (private_lookup_attribute): Likewise.
6978 (private_lookup_attribute_by_prefix): Simplify.
6979 (remove_attribute): Use is_attribute_p.
6980 * tree.h: Remove removed declarations.
6982 2017-08-07 Jakub Jelinek <jakub@redhat.com>
6985 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
6986 instead of computing it in the function. Formatting fix.
6987 (expand_case): Don't rely on default_edge being the first edge,
6988 clear it if removing it, pass default_edge to
6989 emit_case_dispatch_table.
6990 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
6993 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
6995 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
6996 insn in the function, emit NOP after the insn.
6998 2017-08-06 Tom de Vries <tom@codesourcery.com>
7000 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
7003 2017-08-06 Tom de Vries <tom@codesourcery.com>
7005 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
7008 2017-08-04 Yury Gribov <tetra2005@gmail.com>
7010 PR tree-optimization/57371
7011 * match.pd: New pattern.
7013 2017-08-04 Marek Polacek <polacek@redhat.com>
7016 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
7017 perform the computation in offset_int.
7019 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
7021 PR tree-optimization/81136
7022 * tree-vectorizer.h: Include tree-hash-traits.h.
7023 (vec_base_alignments): New typedef.
7024 (vec_info): Add a base_alignments field.
7025 (vect_record_base_alignments): Declare.
7026 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
7028 (DR_IS_CONDITIONAL_IN_STMT): New macro.
7029 (create_data_ref): Add an is_conditional_in_stmt argument.
7030 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
7031 the is_conditional_in_stmt field.
7032 (data_ref_loc): Add an is_conditional_in_stmt field.
7033 (get_references_in_stmt): Set the is_conditional_in_stmt field.
7034 (find_data_references_in_stmt): Update call to create_data_ref.
7035 (graphite_find_data_references_in_stmt): Likewise.
7036 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
7037 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
7038 (vect_record_base_alignment): New function.
7039 (vect_record_base_alignments): Likewise.
7040 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
7041 for nested statements even if we fail to compute a misalignment.
7042 Use pooled base alignments for unconditional references.
7043 (vect_find_same_alignment_drs): Compare base addresses instead
7045 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
7046 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
7048 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
7050 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
7051 Add an explicit name for the enum. Use auto_vec for slp_instances
7053 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
7055 (_bb_vec_info): Add a constructor and destructor.
7056 (vinfo_for_stmt): Return NULL for uids of -1 as well.
7057 (destroy_loop_vec_info): Delete.
7058 (vect_destroy_datarefs): Likewise.
7059 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
7060 (vec_info::vec_info): New function.
7061 (vec_info::~vec_info): Likewise.
7062 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
7063 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
7064 destroy_loop_vec_info.
7065 * tree-vect-loop.c (new_loop_vec_info): Replace with...
7066 (_loop_vec_info::_loop_vec_info): ...this.
7067 (destroy_loop_vec_info): Replace with...
7068 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
7069 the stmt_vec_infos. Leave handling of vec_info information to its
7070 destructor. Remove explicit vector releases.
7071 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
7072 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
7073 * tree-vect-slp.c (new_bb_vec_info): Replace with...
7074 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
7075 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
7076 (destroy_bb_vec_info): Replace with...
7077 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
7078 information to its destructor.
7079 (vect_slp_analyze_bb_1): Use new and delete instead of
7080 new_bb_vec_info and destroy_bb_vec_info.
7081 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
7084 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
7086 * tree-data-ref.h (subscript): Add access_fn field.
7087 (data_dependence_relation): Add could_be_independent_p.
7088 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
7089 (same_access_functions): Move to tree-data-ref.c.
7090 * tree-data-ref.c (ref_contains_union_access_p): New function.
7091 (access_fn_component_p): Likewise.
7092 (access_fn_components_comparable_p): Likewise.
7093 (dr_analyze_indices): Add a reference to access_fn_component_p.
7094 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
7096 (constant_access_functions): Likewise.
7097 (add_other_self_distances): Likewise.
7098 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
7099 (initialize_data_dependence_relation): Use XCNEW and remove
7100 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
7101 of access functions that have the same type. Allow the
7102 subsequence to end with different bases in some circumstances.
7103 Record the chosen access functions in SUB_ACCESS_FN.
7104 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
7105 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
7106 (subscript_dependence_tester_1): Likewise dra and drb.
7107 (build_classic_dist_vector): Update calls accordingly.
7108 (subscript_dependence_tester): Likewise.
7109 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
7110 DDR_COULD_BE_INDEPENDENT_P.
7111 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
7112 comp_alias_ddrs instead of may_alias_ddrs.
7113 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
7115 (vect_analyze_data_ref_dependence): Use it if
7116 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
7117 distance vectors if that fails.
7118 (dependence_distance_ge_vf): New function.
7119 (vect_prune_runtime_alias_test_list): Use it. Don't clear
7120 LOOP_VINFO_MAY_ALIAS_DDRS.
7122 2017-08-04 Richard Biener <rguenther@suse.de>
7125 * fold-const.c (fold_binary_loc): Properly restrict
7126 minus_var0 && minus_var1 case when associating undefined overflow
7129 2017-08-04 Tom de Vries <tom@codesourcery.com>
7131 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
7133 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7135 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7136 Don't start diagnostic messages with a capital letter.
7137 * config/rs6000/rs6000.c (rs6000_option_override_internal):
7139 (rs6000_invalid_builtin): Likewise.
7140 (rs6000_trampoline_init): Likewise.
7142 2017-08-03 Jakub Jelinek <jakub@redhat.com>
7145 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
7146 after setting changeable df flags.
7148 2017-08-03 Richard Biener <rguenther@suse.de>
7150 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
7151 up if the use is in USE - X.
7153 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
7155 * toplev.c (dumpfile.h): New include.
7156 (internal_error_reentered): New static function. Use it...
7157 (internal_error_function): ...here to handle reentered internal_error.
7159 2017-08-03 Richard Biener <rguenther@suse.de>
7162 * fold-const.c (split_tree): Add minus_var and minus_con
7163 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
7164 here but always use minus_*.
7165 (associate_trees): Assert we never associate with MINUS_EXPR
7166 and NULL first operand. Do not recurse for PLUS_EXPR operands
7167 when associating as MINUS_EXPR either.
7168 (fold_binary_loc): Track minus_var and minus_con.
7170 2017-08-03 Tom de Vries <tom@codesourcery.com>
7173 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
7174 ACCEL_COMPILER, apply finish_options on
7175 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
7177 2017-08-03 Tom de Vries <tom@codesourcery.com>
7180 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
7181 function_entry_patch_area_size > 0.
7183 2017-08-03 Jakub Jelinek <jakub@redhat.com>
7186 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
7187 instead of 10??LU, perform unit multiplication in wide_int,
7188 don't change alloc_object_size_limit if the limit is larger
7191 PR tree-optimization/81655
7192 PR tree-optimization/81588
7193 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
7194 the case when ranges[i].low and high are 1 for unsigned type with
7198 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
7199 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
7201 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
7203 * tree-vrp.h: Add include guard.
7205 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
7208 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
7209 (ud2): New insn pattern.
7210 * config/i386/i386.c (ix86_expand_epilogue):
7211 For naked functions, generate ud2 instead of trap insn.
7213 2017-08-02 Marek Polacek <polacek@redhat.com>
7216 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
7218 2017-08-02 Tom de Vries <tom@codesourcery.com>
7219 Cesar Philippidis <cesar@codesourcery.com>
7221 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
7222 Add missing edge probabilities.
7224 2017-08-02 Tamar Christina <tamar.christina@arm.com>
7226 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
7229 2017-08-02 Jakub Jelinek <jakub@redhat.com>
7232 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
7233 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
7234 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
7235 prologue_seq sequences - if any.
7237 2017-08-02 Richard Biener <rguenther@suse.de>
7239 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
7240 via vectors if supported, integer extracts via punning if supported
7241 or otherwise vector extracts.
7243 2017-08-02 Richard Biener <rguenther@suse.de>
7245 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
7247 (bitmap_insert_into_set): ... this.
7249 2017-08-02 Richard Biener <rguenther@suse.de>
7251 PR tree-optimization/81633
7253 2015-08-17 Alan Hayward <alan.hayward@arm.com>
7255 PR tree-optimization/71752
7256 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
7258 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
7260 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
7261 (machine_function::call_ms2sysv_pad_out): Remove field.
7262 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
7263 (ix86_compute_frame_layout): Likewise.
7265 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
7268 * config/i386/i386.c (ix86_set_func_type): Disallow naked
7269 attribute with interrupt attribute.
7271 2017-08-01 Andrew Pinski <apinski@cavium.com>
7273 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
7274 BIT_INSERT_EXPR's operand 1
7275 to see if the types precision matches.
7277 2017-08-01 Martin Liska <mliska@suse.cz>
7280 * builtins.c (expand_builtin_memcpy_args): Remove.
7281 (expand_builtin_memcpy): Call newly added function
7282 expand_builtin_memory_copy_args.
7283 (expand_builtin_memcpy_with_bounds): Likewise.
7284 (expand_builtin_mempcpy): Remove last argument.
7285 (expand_builtin_mempcpy_with_bounds): Likewise.
7286 (expand_builtin_memory_copy_args): New function created from
7287 expand_builtin_mempcpy_args with small modifications.
7288 (expand_builtin_mempcpy_args): Remove.
7289 (expand_builtin_stpcpy): Remove unused argument.
7290 (expand_builtin): Likewise.
7291 (expand_builtin_with_bounds): Likewise.
7293 2017-08-01 Martin Liska <mliska@suse.cz>
7296 Make mempcpy more optimal (PR middle-end/70140).
7298 2017-08-01 Jakub Jelinek <jakub@redhat.com>
7301 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
7302 __builtin_vec_cmpne verify both arguments are compatible vectors
7303 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
7304 verify arg1_type is a pointer or array type. For __builtin_vec_st,
7305 move computation of aligned to after checking the argument types.
7309 * config/rs6000/vsx.md (vextract_fp_from_shorth,
7310 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
7313 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
7314 Doug Rupp <rupp@adacore.com>
7315 Olivier Hainque <hainque@adacore.com>
7317 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
7318 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
7320 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
7321 for TARGET_OS_CPP_BUILTIN.
7322 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
7323 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
7325 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
7326 passing required abi options to the assembler for EABI configurations.
7327 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
7328 of .text.hot and .text.unlikely sections for kernel modules when
7329 using ARM style exceptions.
7330 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
7331 options. Add EXTRA_CC1_SPEC.
7332 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
7334 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
7336 (ARM_TARGET2_DWARF_FORMAT): Define.
7337 * config/arm/t-vxworks: Adjust multilib control to removal of the
7338 Diab command line options.
7340 2017-08-01 Martin Liska <mliska@suse.cz>
7342 PR gcov-profile/81561
7343 * gcov.c (unblock): Make unblocking safe as we need to preserve
7344 index correspondence of blocks and block_lists.
7346 2017-08-01 Richard Biener <rguenther@suse.de>
7348 PR tree-optimization/81181
7349 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
7350 (compute_antic): ... end of iteration here.
7352 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
7354 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
7355 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
7356 (ftree-slp-vectorize): Likewise.
7357 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
7358 can no longer be set independent of flag_tree_loop_vectorize.
7359 * omp-general.c (emp_max_vf): Likewise.
7360 * opts.c (enable_fdo_optimizations): Remove references to
7361 flag_tree_vectorize, these are now implicit.
7362 (common_handle_option): Remove handling for OPT_ftree_vectorize,
7363 and leave it for the options machinery.
7365 2017-08-01 Martin Liska <mliska@suse.cz>
7368 * builtins.c (expand_builtin_memcpy_args): Remove.
7369 (expand_builtin_memcpy): Call newly added function
7370 expand_builtin_memory_copy_args.
7371 (expand_builtin_memcpy_with_bounds): Likewise.
7372 (expand_builtin_mempcpy): Remove last argument.
7373 (expand_builtin_mempcpy_with_bounds): Likewise.
7374 (expand_builtin_memory_copy_args): New function created from
7375 expand_builtin_mempcpy_args with small modifications.
7376 (expand_builtin_mempcpy_args): Remove.
7377 (expand_builtin_stpcpy): Remove unused argument.
7378 (expand_builtin): Likewise.
7379 (expand_builtin_with_bounds): Likewise.
7381 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
7384 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
7385 print "ds:" only for immediates in generic address space.
7387 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
7390 * config/i386/i386.c (ix86_funciton_naked): New prototype.
7391 (ix86_function_ok_for_sibcall): Return false for naked functions.
7393 2017-08-01 Richard Biener <rguenther@suse.de>
7395 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
7396 (compute_antic): Seed worklist with exit block predecessors.
7397 * cfganal.c (dfs_find_deadend): For a cycle return the source
7398 of the edge closing it.
7400 2017-08-01 Tamar Christina <tamar.christina@arm.com>
7402 * config/aarch64/aarch64.c
7403 (aarch64_can_const_movi_rtx_p): Move 0 check.
7405 2017-08-01 Bin Cheng <bin.cheng@arm.com>
7407 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
7408 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
7410 * match.pd: Ditto in address comparison pattern.
7412 2017-08-01 Bin Cheng <bin.cheng@arm.com>
7414 PR tree-optimization/81627
7415 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
7416 closed ssa form for store-store chain.
7418 2017-08-01 Bin Cheng <bin.cheng@arm.com>
7420 PR tree-optimization/81620
7421 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
7422 for store-store chain.
7424 2017-08-01 Jakub Jelinek <jakub@redhat.com>
7426 PR tree-optimization/81588
7427 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
7428 ranges[i].in_p, invert comparison code ccode. For >/>=,
7429 swap rhs1 and rhs2 and comparison code unconditionally,
7430 for </<= don't do that. Don't swap rhs1/rhs2 again if
7431 ranges[i].in_p, instead invert comparison code ccode if
7432 opcode or oe->rank is BIT_IOR_EXPR.
7435 * optabs.def (vec_extract_optab, vec_init_optab): Change from
7436 a direct optab to conversion optab.
7437 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
7438 with GET_MODE_INNER as last argument instead of optab_handler.
7439 * expmed.c (extract_bit_field_1): Likewise. Use vector from
7440 vector extraction if possible and optab is available.
7441 * expr.c (store_constructor): Use convert_optab_handler instead
7442 of optab_handler. Use vector initialization from smaller
7443 vectors if possible and optab is available.
7444 * tree-vect-stmts.c (vectorizable_load): Likewise.
7445 * doc/md.texi (vec_extract, vec_init): Document that the optabs
7447 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
7448 of vec_init from half-sized vectors with the same element mode.
7449 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
7450 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
7451 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
7452 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
7453 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
7454 after mode in gen_vec_extract* calls.
7455 (vec_extract<mode>): Renamed to ...
7456 (vec_extract<mode><ssescalarmodelower>): ... this.
7457 (vec_extract<mode><ssehalfvecmodelower>): New expander.
7458 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
7459 element mode after mode in gen_vec_init* calls.
7460 (VEC_INIT_HALF_MODE): New mode iterator.
7461 (vec_init<mode>): Renamed to ...
7462 (vec_init<mode><ssescalarmodelower>): ... this.
7463 (vec_init<mode><ssehalfvecmodelower>): New expander.
7464 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
7465 (vec_extractv2sfsf): ... this.
7466 (vec_initv2sf): Renamed to ...
7467 (vec_initv2sfsf): ... this.
7468 (vec_extractv2si): Renamed to ...
7469 (vec_extractv2sisi): ... this.
7470 (vec_initv2si): Renamed to ...
7471 (vec_initv2sisi): ... this.
7472 (vec_extractv4hi): Renamed to ...
7473 (vec_extractv4hihi): ... this.
7474 (vec_initv4hi): Renamed to ...
7475 (vec_initv4hihi): ... this.
7476 (vec_extractv8qi): Renamed to ...
7477 (vec_extractv8qiqi): ... this.
7478 (vec_initv8qi): Renamed to ...
7479 (vec_initv8qiqi): ... this.
7480 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
7481 (vec_init<mode>): Renamed to ...
7482 (vec_init<mode><VEC_base_l>): ... this.
7483 (vec_extract<mode>): Renamed to ...
7484 (vec_extract<mode><VEC_base_l>): ... this.
7485 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
7486 (vec_initv2sfsf): ... this.
7487 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
7488 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
7489 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
7490 element mode after mode in gen_vec_init* calls.
7491 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
7492 (vec_init<mode><Vel>): ... this.
7493 (vec_extract<mode>): Renamed to ...
7494 (vec_extract<mode><Vel>): ... this.
7495 * config/aarch64/iterators.md (Vel): New mode attribute.
7496 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
7497 Add element mode after mode in gen_vec_extract* calls.
7498 * config/s390/vector.md (non_vec_l): New mode attribute.
7499 (vec_extract<mode>): Renamed to ...
7500 (vec_extract<mode><non_vec_l>): ... this.
7501 (vec_init<mode>): Renamed to ...
7502 (vec_init<mode><non_vec_l>): ... this.
7503 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
7504 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
7506 * config/arm/iterators.md (V_elem_l): New mode attribute.
7507 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
7508 (vec_extract<mode><V_elem_l>): ... this.
7509 (vec_extractv2di): Renamed to ...
7510 (vec_extractv2didi): ... this.
7511 (vec_init<mode>): Renamed to ...
7512 (vec_init<mode><V_elem_l>): ... this.
7513 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
7514 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
7515 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
7516 Add element mode after gen_vec_extract* calls.
7517 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
7518 (vec_init<mode><unitmode>): ... this.
7519 (vec_extract<mode>): Renamed to ...
7520 (vec_extract<mode><unitmode>): ... this.
7521 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
7522 (vec_init<mode><unitmode>): ... this.
7523 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
7524 (vec_initv2sfsf): ... this.
7525 (vec_extractv2sf): Renamed to ...
7526 (vec_extractv2sfsf): ... this.
7527 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
7528 Add element mode after gen_vec_extract* calls.
7529 * config/mips/mips.md (unitmode): New mode iterator.
7530 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
7531 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
7532 * config/spu/spu.md (inner_l): New mode attribute.
7533 (vec_init<mode>): Renamed to ...
7534 (vec_init<mode><inner_l>): ... this.
7535 (vec_extract<mode>): Renamed to ...
7536 (vec_extract<mode><inner_l>): ... this.
7537 * config/sparc/sparc.md (veltmode): New mode iterator.
7538 (vec_init<VMALL:mode>): Renamed to ...
7539 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
7540 * config/ia64/vect.md (vec_initv2si): Renamed to ...
7541 (vec_initv2sisi): ... this.
7542 (vec_initv2sf): Renamed to ...
7543 (vec_initv2sfsf): ... this.
7544 (vec_extractv2sf): Renamed to ...
7545 (vec_extractv2sfsf): ... this.
7546 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
7547 (vec_init<mode>): Renamed to ...
7548 (vec_init<mode><VEC_base_l>): ... this.
7549 (vec_extract<mode>): Renamed to ...
7550 (vec_extract<mode><VEC_base_l>): ... this.
7551 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
7552 (vec_initv2sfsf): ... this.
7553 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
7554 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
7555 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
7556 gen_vec_init* calls.
7558 2017-08-01 Richard Biener <rguenther@suse.de>
7560 PR tree-optimization/81297
7561 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
7562 TREE_OVERFLOW from INTEGER_CSTs.
7564 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
7566 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
7568 2017-07-31 Carl Love <cel@us.ibm.com>
7570 * config/rs6000/rs6000-c: Add support for built-in functions
7571 vector signed char vec_xl_be (signed long long, signed char *);
7572 vector unsigned char vec_xl_be (signed long long, unsigned char *);
7573 vector signed int vec_xl_be (signed long long, signed int *);
7574 vector unsigned int vec_xl_be (signed long long, unsigned int *);
7575 vector signed long long vec_xl_be (signed long long, signed long long *);
7576 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
7577 vector signed short vec_xl_be (signed long long, signed short *);
7578 vector unsigned short vec_xl_be (signed long long, unsigned short *);
7579 vector double vec_xl_be (signed long long, double *);
7580 vector float vec_xl_be (signed long long, float *);
7581 * config/rs6000/altivec.h (vec_xl_be): Add #define.
7582 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
7583 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
7585 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
7586 (altivec_expand_builtin): Add switch statement to call
7587 altivec_expand_xl_be for each builtin.
7588 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
7589 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
7590 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
7591 __builtin_vsx_le_be_v16qi.
7592 * doc/extend.texi: Update the built-in documentation file for the
7593 new built-in functions.
7595 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
7598 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
7600 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
7602 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7604 * config.gcc: Add z14.
7605 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
7606 CPU model numbers for z13s and z14.
7607 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
7609 * config/s390/s390-opts.h (enum processor_type): Rename
7610 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
7611 * config/s390/s390.c (processor_table): Add field for CPU name to
7612 be passed to Binutils.
7613 (s390_asm_output_machine_for_arch): Use the new field in
7614 processor_table for Binutils.
7615 (s390_expand_builtin): Replace arch12 with z14.
7616 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
7617 (s390_get_sched_attrmask): Likewise.
7618 (s390_get_unit_mask): Likewise.
7619 * config/s390/s390.opt: Add z14 to processor_type enum.
7621 2017-07-31 Martin Jambor <mjambor@suse.cz>
7624 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
7625 regardless of optimization level.
7627 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
7628 Martin Liska <mliska@suse.cz>
7630 * predict.def: Remove old comment and adjust probability.
7631 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
7634 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
7637 * config/i386/i386.c (ix86_function_naked): New function.
7638 (ix86_can_use_return_insn_p): Return false for naked functions.
7639 (ix86_expand_prologue): Skip prologue for naked functions.
7640 (ix86_expand_epilogue): Skip epilogue for naked functions
7641 and emit trap instruction.
7642 (ix86_warn_func_return): New function.
7643 (ix86_attribute_table): Add "naked" attribute specification.
7644 (TARGET_WARN_FUNC_RETURN): Define.
7645 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
7647 2017-07-31 Martin Liska <mliska@suse.cz>
7649 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
7650 (dump_gimple_bb_header): Always dump BB info.
7651 (pp_cfg_jump): Do not append info about BB when dumping a jump.
7653 2017-07-31 Martin Liska <mliska@suse.cz>
7656 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
7657 also with current_function_decl non-null equality.
7659 2017-07-31 Jakub Jelinek <jakub@redhat.com>
7662 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
7663 change type to the element type, instead add eltype variable and
7664 use it where we are interested in the element type.
7666 PR tree-optimization/81603
7667 * ipa-polymorphic-call.c
7668 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
7669 offset arithmetic in offset_int, bail out if the resulting bit offset
7670 doesn't fit into shwi.
7672 2017-07-31 Martin Liska <mliska@suse.cz>
7674 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
7675 (gimplify_save_expr): Fix comment.
7677 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
7680 * config/i386/i386.c (ix86_function_arg): Update arguments for
7682 (ix86_compute_frame_layout): Set the initial stack offset to
7683 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
7684 INCOMING_FRAME_SP_OFFSET.
7685 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
7686 stack before exception handler returns.
7687 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
7688 the 'ERROR_CODE' for exception handler.
7690 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
7692 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
7693 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
7694 (ASM_OUTPUT_REG_POP): Ditto.
7695 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
7696 instead of asm_fprintf to output pure string.
7698 2017-07-29 Jakub Jelinek <jakub@redhat.com>
7700 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
7701 to imported_module_or_decl hook.
7702 (debug_nothing_tree_tree_tree_bool): Remove.
7703 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
7704 * debug.c (do_nothing_debug_hooks): Use
7705 debug_nothing_tree_tree_tree_bool_bool instead of
7706 debug_nothing_tree_tree_tree_bool.
7707 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
7708 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
7709 * sdbout.c (sdb_debug_hooks): Likewise.
7710 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
7711 (gen_namespace_die): Add DW_AT_export_symbols attribute if
7713 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
7714 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
7715 attribute, don't add anything.
7717 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7719 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
7720 (fold_build2_stat_loc): Likewise.
7721 (fold_build3_stat_loc): Likewise.
7722 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
7723 (fold_build1_loc): Remove macro.
7724 (fold_build2_loc): Likewise.
7725 (fold_build3_loc): Likewise.
7727 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7729 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
7730 (gimple_build_debug_bind_source_stat): Likewise.
7731 * gimple.h (gimple_build_debug_bind): Remove macro.
7732 (gimple_build_debug_bind_source): Likewise.
7734 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7736 * bitmap.c (bitmap_alloc): Adjust.
7737 (bitmap_gc_alloc): Likewise.
7738 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
7740 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7742 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
7743 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
7744 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
7745 (bitmap_gc_alloc_stat): Likewise.
7746 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
7748 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7750 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
7751 * rtl.h (shallow_copy_rtx): Remove macro.
7753 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7755 * emit-rtl.c (gen_raw_REG): Adjust.
7756 * gengenrtl.c (gendef): Likewise.
7757 * rtl.c (rtx_alloc_stat): Remove _stat from name.
7758 * rtl.h (rtx_alloc): Remove macro.
7760 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7762 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
7763 (build_tree_list_stat): Likewise.
7764 * tree.h (build_tree_list): Remove macro.
7765 (build_tree_list_vec): Likewise.
7767 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7769 * tree.c (make_vector_stat): Remove _stat from name.
7770 (build_vector_stat): Likewise.
7771 * tree.h (make_vector_stat): Remove macro.
7772 (build_vector_stat): Likewise.
7774 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7776 * tree.h (build_var_debug_value): Remove prototype.
7778 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7780 * tree.c (tree_cons_stat): Remove _stat from name.
7781 * tree.h (tree_cons): Remove macro.
7783 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7785 * tree.c (build_vl_exp_stat): Remove _stat from name.
7786 * tree.h (build_vl_exp): Remove macro.
7788 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7790 * tree.c (build_decl_stat): Remove _stat from name.
7791 * tree.h (build_decl): Remove macro.
7793 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7795 * gimple.c (gimple_build_with_ops_stat): Adjust.
7796 (gimple_alloc_stat): Remove _stat from name.
7797 * gimple.h (gimple_alloc): Remove macro.
7799 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7801 * tree.c (make_tree_vec_stat): Remove _stat from name.
7802 (grow_tree_vec_stat): Likewise.
7803 * tree.h (make_tree_vec_stat): Adjust prototype.
7804 (grow_tree_vec_stat): Likewise.
7805 (make_tree_vec): Remove macro.
7806 (grow_tree_vec): Likewise.
7808 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7810 * fold-const.c (fold_build1_stat_loc): Adjust.
7811 (fold_build2_stat_loc): Likewise.
7812 (fold_build3_stat_loc): Likewise.
7813 * tree.c (build0_stat): Remove _stat from name.
7814 (build1_stat): Likewise.
7815 (build2_stat): Likewise.
7816 (build3_stat): Likewise.
7817 (build4_stat): Likewise.
7818 (build5_stat): Likewise.
7819 * tree.h (build1_loc): Remove macro, and rename _stat function
7821 (build2_loc): Likewise.
7822 (build3_loc): Likewise.
7823 (build4_loc): Likewise.
7824 (build5_loc): Likewise.
7826 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7828 * tree.c (make_int_cst_stat): Remove _stat from name.
7829 * tree.h (make_int_cst_stat): Adjust prototype.
7830 (make_int_cst): Remove macro.
7832 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7834 * tree.c (make_tre_binfo_stat): Remove _stat from name.
7835 * tree.h (make_tree_binfo_stat): Adjust prototype.
7836 (make_tree_binfo): Remove.
7838 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7840 * tree.c (copy_node_stat): Rename to copy_node.
7841 (build_distinct_type_copy): Adjust.
7842 * tree.h (copy_node_stat): Adjust prototype.
7843 (copy_node): Remove macro.
7845 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7847 * tree.c (make_node_stat): rename to make_node.
7848 (build_tree_list_stat): Adjust.
7849 (build0_stat): Likewise.
7850 (build2_stat): Likewise.
7851 (build3_stat): Likewise.
7852 (build4_stat): Likewise.
7853 (build5_stat): Likewise.
7854 (build_decl_stat): Likewise.
7855 * tree.h (make_node_stat): Adjust prototype.
7856 (make_node): remove macro.
7858 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
7860 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
7861 (PPC_FEATURE2_SCV): Likewise.
7862 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
7864 2017-07-28 Tamar Christina <tamar.christina@arm.com>
7866 * config/aarch64/aarch64.c
7867 (aarch64_internal_mov_immediate): Add new special pattern.
7868 * config/aarch64/aarch64.md (*movdi_aarch64):
7869 Add reg/32bit const mov case.
7871 2017-07-28 Tamar Christina <tamar.christina@arm.com>
7872 Richard Sandiford <richard.sandiford@linaro.org>
7874 * config/aarch64/aarch64.md (mov<mode>): Generalize.
7875 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
7876 Add integer and movi cases.
7877 (movi-split-hf-df-sf split, fp16): New.
7878 (enabled): Added TARGET_FP_F16INST.
7879 * config/aarch64/iterators.md (GPF_HF): New.
7880 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
7882 2017-07-28 Tamar Christina <tamar.christina@arm.com>
7884 * config/aarch64/aarch64.c
7885 (aarch64_simd_container_mode): Add prototype.
7886 (aarch64_expand_mov_immediate): Add HI support.
7887 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
7888 (aarch64_can_const_movi_rtx_p): New.
7889 (aarch64_preferred_reload_class):
7890 Remove restrictions of using FP registers for certain SIMD operations.
7891 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
7892 (aarch64_valid_floating_const): Add integer move validation.
7893 (aarch64_simd_imm_scalar_p): Remove.
7894 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
7895 (aarch64_legitimate_constant_p): Expand list of supported cases.
7896 * config/aarch64/aarch64-protos.h
7897 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
7898 (aarch64_reinterpret_float_as_int): New.
7899 (aarch64_simd_imm_scalar_p): Remove.
7900 * config/aarch64/constraints.md (Uvi): New.
7901 (Dd): Split into Ds and new Dd.
7902 * config/aarch64/aarch64.md (*movsi_aarch64):
7904 (*movdi_aarch64): Add SIMD mov case.
7906 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7908 * tree-predcom.c: (struct chain): Handle store-store chain in which
7909 stores for elimination only store loop invariant values.
7910 (execute_pred_commoning_chain): Ditto.
7911 (prepare_initializers_chain_store_elim): Ditto.
7912 (prepare_finalizers): Ditto.
7913 (is_inv_store_elimination_chain): New function.
7914 (initialize_root_vars_store_elim_1): New function.
7916 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7918 * tree-predcom.c: Revise general description of the pass.
7919 (enum chain_type): New enum type for store elimination.
7920 (struct chain): New field supporting store elimination.
7921 (struct component): Ditto.
7922 (dump_chain): Dump store-stores chain.
7923 (release_chain): Release resources.
7924 (split_data_refs_to_components): Compute and create component
7925 contains only stores for elimination.
7926 (get_chain_last_ref_at): New function.
7927 (make_invariant_chain): Initialization.
7928 (make_rooted_chain): Specify chain type in parameter and record it.
7929 (add_looparound_copies): Skip for store-stores chain.
7930 (determine_roots_comp): Compute type of chain and pass it to
7932 (initialize_root_vars_store_elim_2): New function.
7933 (finalize_eliminated_stores): New function.
7934 (remove_stmt): Handle store for elimination.
7935 (execute_pred_commoning_chain): Execute predictive commoning on
7937 (determine_unroll_factor): Skip unroll for store-stores chain.
7938 (prepare_initializers_chain_store_elim): New function.
7939 (prepare_initializers_chain): Hanlde store-store chain.
7940 (prepare_finalizers_chain, prepare_finalizers): New function.
7941 (tree_predictive_commoning_loop): Return integer value indicating
7942 if loop is unrolled or lcssa form is corrupted.
7943 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
7945 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7947 * tree-predcom.c (initialize_root): Delete.
7948 (execute_pred_commoning_chain): Initialize root vars and replace
7949 reference of non-combined chain directly, rather than call above
7952 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7954 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
7955 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
7957 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7959 * tree-predcom.c (struct chain): New field init_seq.
7960 (release_chain): Release init_seq.
7961 (prepare_initializers_chain): Record intialization stmts in above
7963 (insert_init_seqs): New function.
7964 (tree_predictive_commoning_loop): Call insert_init_seqs.
7966 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7968 * tree-predcom.c (determine_roots_comp): Skip trivial components.
7970 2017-07-28 Richard Biener <rguenther@suse.de>
7972 * match.pd: Remove superfluous :c.
7973 * genmatch.c (simplify::id): Add member.
7974 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
7976 (current_id): New global.
7977 (dt_node::parent): Move from ...
7978 (dt_operand::parent): ... here. Add for_id member.
7979 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
7980 (decision_tree::find_node): Relax order requirement when
7981 merging DT_TRUE nodes to ones inbetween the current simplify
7982 and the one we try to merge with. Add diagnostic whenever
7983 we need to enforce pattern order by not merging.
7984 (decision_tree::insert): Set current_id.
7985 (decision_tree::print_node): Dump parent node and for_id.
7986 (parser::last_id): Add member.
7987 (parser::push_simplify): Assign unique id.
7988 (parser::parser): Initialize last_id.
7990 2017-07-28 Martin Liska <mliska@suse.cz>
7993 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
7994 gimple_build_debug_bind.
7996 2017-07-28 Richard Biener <rguenther@suse.de>
7998 PR tree-optimization/81502
7999 * match.pd: Add pattern combining BIT_INSERT_EXPR with
8001 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
8003 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
8004 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
8005 for BIT_FIELD_REF args.
8006 * fold-const.c (make_bit_field_ref): Likewise.
8007 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
8009 2017-07-28 Jakub Jelinek <jakub@redhat.com>
8012 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
8013 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
8014 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
8015 Or it into SANITIZER_UNDEFINED.
8016 * ubsan.c: Include gimple-fold.h and varasm.h.
8017 (ubsan_expand_ptr_ifn): New function.
8018 (instrument_pointer_overflow): New function.
8019 (maybe_instrument_pointer_overflow): New function.
8020 (instrument_object_size): Formatting fix.
8021 (pass_ubsan::execute): Call instrument_pointer_overflow
8022 and maybe_instrument_pointer_overflow.
8023 * internal-fn.c (expand_UBSAN_PTR): New function.
8024 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
8025 * sanitizer.def (__ubsan_handle_pointer_overflow,
8026 __ubsan_handle_pointer_overflow_abort): New builtins.
8027 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
8028 * internal-fn.def (UBSAN_PTR): New internal function.
8029 * opts.c (sanitizer_opts): Add pointer-overflow.
8030 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
8031 * fold-const.c (build_range_check): Compute pointer range check in
8032 integral type if pointer arithmetics would be needed. Formatting
8035 2017-07-28 Martin Liska <mliska@suse.cz>
8038 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
8039 parameters that are of a variable-length.
8041 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
8043 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
8045 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
8046 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
8047 (CRT_CALL_STATIC_FUNCTION): Likewise.
8048 (ASM_DEFAULT_SPEC): New define.
8049 (ASM_SPEC32): Likewise.
8050 (ASM_SPEC64): Likewise.
8051 (ASM_SPEC_COMMON): Likewise.
8052 (ASM_SPEC): Likewise.
8053 (INVALID_64BIT): Likewise.
8054 (LINK_OS_DEFAULT_SPEC): Likewise.
8055 (LINK_OS_SPEC32): Likewise.
8056 (LINK_OS_SPEC64): Likewise.
8057 (POWERPC_LINUX): Likewise.
8058 (PTRDIFF_TYPE): Likewise.
8059 (RESTORE_FP_PREFIX): Likewise.
8060 (RESTORE_FP_SUFFIX): Likewise.
8061 (SAVE_FP_PREFIX): Likewise.
8062 (SAVE_FP_SUFFIX): Likewise.
8063 (SIZE_TYPE): Likewise.
8064 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
8065 (TARGET_64BIT): Likewise.
8066 (TARGET_64BIT): Likewise.
8067 (TARGET_AIX): Likewise.
8068 (WCHAR_TYPE_SIZE): Likewise.
8069 (WCHAR_TYPE): Undefine.
8070 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
8071 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
8072 (CPP_OS_RTEMS_SPEC): Delete.
8073 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
8074 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
8076 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
8078 2017-07-28 Jakub Jelinek <jakub@redhat.com>
8080 PR tree-optimization/81578
8081 * tree-parloops.c (build_new_reduction): Bail out if
8082 reduction_code isn't one of the standard OpenMP reductions.
8083 Move the details printing after that decision.
8085 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
8087 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
8088 related to reload_in_progress.
8089 (splat_input_operand): Likewise.
8090 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
8092 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
8094 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
8095 (TARGET_INSTANTIATE_DECLS): Likewise.
8096 (legitimate_indexed_address_p): Delete reload_in_progress code.
8097 (rs6000_debug_legitimate_address_p): Likewise.
8098 (rs6000_eliminate_indexed_memrefs): Likewise.
8099 (rs6000_emit_le_vsx_store): Likewise.
8100 (rs6000_emit_move_si_sf_subreg): Likewise.
8101 (rs6000_emit_move): Likewise.
8102 (register_to_reg_type): Likewise.
8103 (rs6000_pre_atomic_barrier): Likewise.
8104 (rs6000_machopic_legitimize_pic_address): Likewise.
8105 (rs6000_allocate_stack_temp): Likewise.
8106 (rs6000_address_for_fpconvert): Likewise.
8107 (rs6000_address_for_altivec): Likewise.
8108 (rs6000_secondary_memory_needed_rtx): Delete function.
8109 (rs6000_check_sdmode): Likewise.
8110 (rs6000_alloc_sdmode_stack_slot): Likewise.
8111 (rs6000_instantiate_decls): Likewise.
8112 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
8113 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
8114 Delete reload_in_progress.
8115 (*vec_reload_and_plus_<mptrsize>): Likewise.
8116 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
8117 (vsx_div_v2di): Likewise.
8118 (vsx_udiv_v2di): Likewise.
8120 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
8122 * config/rs6000/rs6000.opt (mlra): Replace with stub.
8123 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
8124 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
8125 (rs6000_debug_reg_global): Delete print of LRA status.
8126 (rs6000_option_override_internal): Delete dead LRA related code.
8127 (rs6000_lra_p): Delete function.
8128 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
8130 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
8132 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
8133 * config/riscv/rtems.h: New file.
8135 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8136 Sudakshina Das <sudi.das@arm.com>
8138 * config/aarch64/aarch64.md
8139 (define_split for and<mode>3nr_compare): Move
8140 non aarch64_logical_operand to a register.
8141 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
8142 register immediate operand to a register.
8143 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
8145 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
8148 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
8150 2017-07-27 Richard Biener <rguenther@suse.de>
8152 PR tree-optimization/81573
8153 PR tree-optimization/81494
8154 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
8155 multi defuse cycle case.
8157 2017-07-27 Richard Biener <rguenther@suse.de>
8159 PR tree-optimization/81571
8160 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
8163 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
8165 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
8166 earlier and only if MASK_FPU is set. Adjust formatting.
8168 2017-07-27 Martin Liska <mliska@suse.cz>
8170 * opt-functions.awk: Add validation of value of Init.
8171 * optc-gen.awk: Pass new argument.
8173 2017-07-27 Martin Liska <mliska@suse.cz>
8175 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
8176 Fix wrong condition.
8178 2017-07-27 Martin Liska <mliska@suse.cz>
8180 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
8181 BBs and edges seen by autoFDO.
8183 2017-07-27 Richard Biener <rguenther@suse.de>
8185 PR tree-optimization/81502
8186 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
8187 with incompatible but same sized type.
8188 (execute_update_addresses_taken): Likewise.
8190 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
8192 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
8193 flag_tree_loop_vectorize rather than flag_tree_vectorize.
8195 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8198 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
8199 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
8200 Change s_operand to memory_operand.
8202 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
8204 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
8205 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
8206 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
8207 Emit instructions rather than returning an expression. Handle TFmode
8208 and KFmode by casting to TImode.
8209 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
8210 (rs6000_emit_le_vsx_store): Likewise.
8211 * config/rs6000/vsx.md (VSX_TI): New iterator.
8212 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
8213 (*vsx_le_undo_permute_<mode>): Likewise.
8214 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
8215 emit the split sequence.
8216 (*vsx_le_perm_store_<mode>): Likewise.
8218 2017-07-27 Jakub Jelinek <jakub@redhat.com>
8220 PR tree-optimization/81555
8221 PR tree-optimization/81556
8222 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
8223 if true, force CHANGED for the recursive invocation.
8224 (reassociate_bb): Remember original length of ops array, pass
8225 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
8227 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
8228 attributes for noipa attribute. For naked attribute use
8229 lookup_attribute first before lookup_attribute_spec.
8230 * final.c (rest_of_handle_final): Disable IPA RA for functions with
8232 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
8233 for functions with noipa attribute.
8234 (cgraph_externally_visible_p): Return true for functions with noipa
8236 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
8237 for functions with noipa attribute.
8238 * doc/extend.texi: Document noipa function attribute.
8239 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
8240 also for functions with noipa attribute.
8241 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
8243 2017-07-26 Andrew Pinski <apinski@cavium.com>
8245 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
8246 vec_unalign_load_cost and vec_unalign_store_cost.
8248 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
8250 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
8251 -mvsx-small-integer option.
8252 (ISA_3_0_MASKS_IEEE): Likewise.
8253 (OTHER_VSX_VECTOR_MASKS): Likewise.
8254 (POWERPC_MASKS): Likewise.
8255 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
8256 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
8257 code, only testing for DImode being allowed in non-VSX floating
8259 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
8260 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
8262 (rs6000_option_override_internal): Delete -mvsx-small-integer.
8263 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
8264 TARGET_P8_VECTOR test.
8265 (rs6000_secondary_reload_simple_move): Likewise.
8266 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
8267 since TARGET_P9_VECTOR was already tested.
8268 (rs6000_opt_masks): Remove -mvsx-small-integer.
8269 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
8270 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
8272 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
8273 test for TARGET_VEXTRACTUB was used, and that uses
8275 (p9 extract splitter): Likewise.
8276 (vsx_extract_<mode>_di_p9): Likewise.
8277 (vsx_extract_<mode>_store_p9): Likewise.
8278 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
8279 for TARGET_P9_VECTOR was used. Delete code that is now dead with
8280 the elimination of TARGET_VSX_SMALL_INTEGER.
8281 (vsx_extract_<mode>_p8): Likewise.
8282 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
8283 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
8284 (vsx_set_<mode>_p9): Likewise.
8285 (vsx_set_v4sf_p9): Likewise.
8286 (vsx_set_v4sf_p9_zero): Likewise.
8287 (vsx_insert_extract_v4sf_p9): Likewise.
8288 (vsx_insert_extract_v4sf_p9_2): Likewise.
8289 * config/rs6000/rs6000.md (sign extend splitter): Change
8290 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
8291 (floatsi<mode>2_lfiwax_mem): Likewise.
8292 (floatunssi<mode>2_lfiwzx_mem): Likewise.
8293 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
8294 since a test for TARGET_P9_VECTOR was used.
8295 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
8296 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
8297 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
8298 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
8299 TARGET_P8_VECTOR test.
8300 (fix_trunc<mode>si2_stfiwx): Likewise.
8301 (fix_trunc<mode>si2_internal): Likewise.
8302 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
8303 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
8305 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
8306 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
8307 TARGET_P8_VECTOR test.
8308 (fixuns_trunc<mode>si2_stfiwx): Likewise.
8309 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
8310 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
8312 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
8313 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
8314 since a test for TARGET_P9_VECTOR was used.
8315 (splitter for loading small constants): Likewise.
8317 2017-07-26 Andrew Pinski <apinski@cavium.com>
8319 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
8322 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
8325 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
8326 (fp_valid_at): Likewise.
8328 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
8330 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
8331 (qdf24xx_addrcost_table): Likewise.
8332 (cortexa57_tunings): Update to use generic_branch_cost.
8333 (cortexa72_tunings): Likewise.
8334 (cortexa73_tunings): Likewise.
8335 (qdf24xx_tunings): Likewise.
8337 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
8339 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
8340 (thunderx2t99_branch_cost): Likewise.
8341 (cortexa35_tunings): Update to use generic_branch_cost.
8342 (cortexa53_tunings): Likewise.
8343 (cortexa57_tunings): Likewise.
8344 (cortexa72_tunings): Likewise.
8345 (cortexa73_tunings): Likewise.
8346 (thunderx2t99_tunings): Likewise.
8348 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
8350 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
8351 (sparc_option_override): Honour MASK_FSMULD.
8352 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
8353 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
8354 * config/sparc/sparc.opt (mfsmuld): New option.
8355 * doc/invoke.texi (mfsmuld): Document option.
8357 2017-07-26 Marek Polacek <polacek@redhat.com>
8360 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
8362 2017-07-26 Richard Biener <rguenther@suse.de>
8364 * gimple-match-head.c (do_valueize): Return OP if valueize
8366 (get_def): New helper to get at the def stmt of a SSA name
8368 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
8369 do_valueize to get at the def stmt.
8370 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
8372 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
8375 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
8377 2017-07-26 Martin Liska <mliska@suse.cz>
8380 * function.c (expand_function_start): Make expansion of
8381 nonlocal_goto_save_area after parm_birth_insn.
8383 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
8385 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
8386 from all CPU target flags enable members.
8388 2017-07-26 Richard Biener <rguenther@suse.de>
8390 * genmatch.c (dt_simplify::gen): Make iterator vars const.
8391 (decision_tree::gen): Make 'type' const.
8392 (write_predicate): Likewise.
8394 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
8396 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
8397 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
8398 (rs6000_option_override_internal): Likewise.
8399 (rs6000_expand_vector_set): Likewise.
8400 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
8401 (TARGET_UPPER_REGS_SF): Likewise.
8402 (TARGET_UPPER_REGS_DI): Likewise.
8403 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
8404 (TARGET_DIRECT_MOVE_64BIT): Likewise.
8405 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
8406 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
8407 (Splitters for DI constants in Altivec registers): Likewise.
8408 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
8409 (vsx_set_v4sf_p9): Likewise.
8410 (vsx_set_v4sf_p9_zero): Likewise.
8411 (vsx_insert_extract_v4sf_p9): Likewise.
8412 (vsx_insert_extract_v4sf_p9_2): Likewise.
8414 2017-07-25 Carl Love <cel@us.ibm.com>
8416 * doc/extend.texi: Update the built-in documentation file for the
8417 existing built-in functions
8418 vector signed char vec_cnttz (vector signed char);
8419 vector unsigned char vec_cnttz (vector unsigned char);
8420 vector signed short vec_cnttz (vector signed short);
8421 vector unsigned short vec_cnttz (vector unsigned short);
8422 vector signed int vec_cnttz (vector signed int);
8423 vector unsigned int vec_cnttz (vector unsigned int);
8424 vector signed long long vec_cnttz (vector signed long long);
8425 vector unsigned long long vec_cnttz (vector unsigned long long);
8427 2017-07-25 Andrew Pinski <apinski@cavium.com>
8429 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
8430 accesses where the use is for the first operand of a BIT_INSERT.
8432 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
8435 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
8436 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
8438 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
8440 * config/i386/gstabs.h: Delete.
8441 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
8443 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
8445 * config/i386/i386.c (ix86_decompose_address): Do not check for
8446 register RTX when looking at index_reg or base_reg.
8447 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
8449 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
8451 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
8452 to update EH info here.
8454 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
8456 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
8458 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
8460 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
8462 2017-07-25 Torsten Duwe <duwe@suse.de>
8464 * common.opt: Introduce -fpatchable-function-entry
8465 command line option, and its variables function_entry_patch_area_size
8466 and function_entry_patch_area_start.
8467 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
8468 including a two-value parser.
8469 * target.def (print_patchable_function_entry): New target hook.
8470 * targhooks.h (default_print_patchable_function_entry): New function.
8471 * targhooks.c (default_print_patchable_function_entry): Likewise.
8472 * toplev.c (process_options): Switch off IPA-RA if
8473 patchable function entries are being generated.
8474 * varasm.c (assemble_start_function): Look at the
8475 patchable-function-entry command line switch and current
8476 function attributes and maybe generate NOP instructions by
8477 calling the print_patchable_function_entry hook.
8478 * doc/extend.texi: Document patchable_function_entry attribute.
8479 * doc/invoke.texi: Document -fpatchable_function_entry
8480 command line option.
8481 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
8483 * doc/tm.texi: Re-generate.
8485 2017-07-25 Jakub Jelinek <jakub@redhat.com>
8488 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
8489 TARGET_AVX512DQ rather than TARGET_AVX512BW.
8491 2017-07-25 Tamar Christina <tamar.christina@arm.com>
8493 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
8495 2017-07-25 Richard Biener <rguenther@suse.de>
8497 PR tree-optimization/81455
8498 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
8499 not walk in cycles when looking for guards.
8501 2017-07-25 Richard Biener <rguenther@suse.de>
8503 PR tree-optimization/81529
8504 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
8505 when optimizing backedge uses.
8507 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
8509 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
8511 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
8512 to dl_section_ref. On AIX, append an expression to subtract
8513 the size of the section length to dl_section_ref.
8515 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
8517 * configure.ac: If any of the config.* scripts fail, exit 1.
8518 * configure: Regenerate.
8520 2017-07-25 Richard Biener <rguenther@suse.de>
8523 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
8524 of immediate uses, be more verbose on errors.
8526 2017-07-25 Richard Biener <rguenther@suse.de>
8528 PR tree-optimization/81510
8529 * tree-vect-loop.c (vect_is_simple_reduction): When the
8530 reduction stmt is not inside the loop bail out.
8532 2017-07-25 Richard Biener <rguenther@suse.de>
8534 PR tree-optimization/81303
8535 * tree-vect-loop-manip.c (vect_loop_versioning): Build
8536 profitability check against LOOP_VINFO_NITERSM1.
8538 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
8540 * domwalk.c (cmp_bb_postorder): Simplify.
8541 (sort_bbs_postorder): New function. Use it...
8542 (dom_walker::walk): ...here to optimize common cases.
8544 2017-07-25 Martin Liska <mliska@suse.cz>
8547 * ipa-visibility.c (function_and_variable_visibility): Make the
8548 redirection just on target that supports aliasing.
8549 Fix GNU coding style.
8551 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
8554 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
8556 * config/arm/rtems.h: Likewise.
8557 * config/bfin/rtems.h: Likewise.
8558 * config/i386/rtemself.h: Likewise.
8559 * config/lm32/rtems.h: Likewise.
8560 * config/m32c/rtems.h: Likewise.
8561 * config/m68k/rtemself.h: Likewise.
8562 * config/microblaze/rtems.h: Likewise.
8563 * config/mips/rtems.h: Likewise.
8564 * config/moxie/rtems.h: Likewise.
8565 * config/nios2/rtems.h: Likewise.
8566 * config/powerpcspe/rtems.h: Likewise.
8567 * config/rs6000/rtems.h: Likewise.
8568 * config/rtems.h: Likewise.
8569 * config/sh/rtems.h: Likewise.
8570 * config/sh/rtemself.h: Likewise.
8571 * config/sparc/rtemself.h: Likewise.
8573 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
8576 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
8577 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
8578 * tree-ssa-structalias.c (alias_get_name): Same.
8580 2017-07-25 Bin Cheng <bin.cheng@arm.com>
8583 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
8584 instructions if no du chain is found.
8586 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
8588 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
8590 2017-07-25 Richard Biener <rguenther@suse.de>
8593 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
8596 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
8598 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
8600 (ISA_2_7_MASKS_SERVER): Likewise.
8601 (ISA_3_0_MASKS_IEEE): Likewise.
8602 (OTHER_P8_VECTOR_MASKS): Likewise.
8603 (OTHER_VSX_VECTOR_MASKS): Likewise.
8604 (POWERPC_MASKS): Likewise.
8605 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
8606 duplicate list of options.
8607 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
8608 explicit -mupper-regs options.
8609 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
8610 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
8611 alias for -mupper-regs-df.
8612 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
8613 (rs6000_init_hard_regno_mode_ok): Likewise.
8614 (rs6000_option_override_internal): Likewise.
8615 (rs6000_opt_masks): Likewise.
8616 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
8617 options in terms of whether -mvsx or -mpower8-vector was used.
8618 (TARGET_UPPER_REGS_DI): Likewise.
8619 (TARGET_UPPER_REGS_SF): Likewise.
8620 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
8621 -mupper-regs-* options.
8623 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
8625 * passes.c (emergency_dump_function): Print some empty lines and a
8626 header before the RTL dump.
8628 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
8630 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
8632 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
8635 * config/aarch64/aarch64.c (aarch64_classify_symbol):
8636 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
8638 2017-07-24 Carl Love <cel@us.ibm.com>
8640 * config/rs6000/rs6000-c.c: Add support for built-in functions
8641 vector float vec_extract_fp32_from_shorth (vector unsigned short);
8642 vector float vec_extract_fp32_from_shortl (vector unsigned short);
8643 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
8644 vec_extract_fp_from_shortl): Add defines for the two builtins.
8645 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
8646 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
8648 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
8649 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
8650 * doc/extend.texi: Update the built-in documentation file for the
8651 new built-in function.
8653 2017-07-24 Jakub Jelinek <jakub@redhat.com>
8656 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
8658 * doc/generic.texi: Likewise.
8659 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
8660 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
8662 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
8664 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
8665 (aarch64_mls_elt_merge<mode>): Likewise.
8667 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
8669 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
8670 having __cxa_atexit.
8672 2017-07-23 Michael Collison <michael.collison@arm.com>
8674 * config/arm/arm.c (arm_option_override): Deprecate
8675 use of -mstructure-size-boundary.
8676 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
8677 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
8679 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8682 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
8683 Reduce cost estimate for direct moves.
8685 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
8688 * config/i386/i386.c (ix86_option_override_internal): Disable
8689 BMI, BMI2 and TBM instructions for -m16.
8691 2017-07-21 Carl Love <cel@us.ibm.com>
8693 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8694 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
8695 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
8696 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
8697 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
8698 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
8699 VMULOSW): New enum "unspec" values.
8700 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
8701 altivec_vmulosw): New patterns.
8702 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
8703 VMULOSW): Add definitions.
8705 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
8707 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
8708 (qdf24xx): Likewise.
8709 * config/aarch64/aarch64-options-extensions.def (rdma); New.
8710 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
8711 (AARCH64_FL_V8_1): Renumber.
8712 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
8713 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
8714 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
8715 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
8716 rdma to feature modifiers list.
8718 2017-07-21 Yury Gribov <tetra2005@gmail.com>
8721 * ipa-visibility (function_and_variable_visibility): Convert
8722 recursive PLT call to direct call if appropriate.
8724 2017-07-21 Andrew Pinski <apinski@cavium.com>
8726 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
8727 operand 1 to see if the types precision matches.
8728 * fold-const.c (operand_equal_p): Likewise.
8730 2017-07-21 Richard Biener <rguenther@suse.de>
8732 PR tree-optimization/81303
8733 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
8734 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
8735 (vect_peeling_hash_get_lowest_cost): Adjust.
8736 (vect_enhance_data_refs_alignment): Likewise. Use
8737 vect_get_peeling_costs_all_drs to compute the penalty for no
8738 peeling to match up costs.
8740 2017-07-21 Richard Biener <rguenther@suse.de>
8742 PR tree-optimization/81500
8743 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
8744 we didn't identify a reduction path.
8746 2017-07-21 Tom de Vries <tom@codesourcery.com>
8747 Cesar Philippidis <cesar@codesourcery.com>
8749 PR gcov-profile/81442
8750 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
8753 2017-07-21 Tom de Vries <tom@codesourcery.com>
8756 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
8758 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
8759 nvptx_override_options_after_change.
8761 2017-07-21 Ulrich Drepper <drepper@redhat.com>
8763 * dwarf2out.c (output_file_names): Avoid double testing for
8766 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
8768 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
8770 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
8772 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
8774 (try_crossjump_to_edge): Do not punt on partitioned functions.
8776 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
8778 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
8779 Put all BBs reachable only via paths crossing cold region to cold
8781 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
8783 2016-07-21 Richard Biener <rguenther@suse.de>
8785 PR tree-optimization/81303
8786 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
8787 into account prologue and epilogue iterations when raising
8788 min_profitable_iters to sth at least covering one vector iteration.
8790 2017-07-21 Tamar Christina <tamar.christina@arm.com>
8792 * config/arm/arm.c (arm_test_cpu_arch_dat):
8795 2017-07-20 Nathan Sidwell <nathan@acm.org>
8797 Remove TYPE_METHODS.
8798 * tree.h (TYPE_METHODS): Delete.
8799 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
8800 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
8801 (dbxout_type_methods): Scan TYPE_FIELDS.
8802 (dbxout_type): Don't check TYPE_METHODS here.
8803 * function.c (use_register_for_decl): Always ignore register for
8804 class types when not optimizing.
8805 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
8806 * tree.c (free_lang_data_in_type): Stitch out member functions and
8807 templates from TYPE_FIELDS.
8808 (build_distinct_type_copy, verify_type_variant,
8809 verify_type): Member fns are on TYPE_FIELDS.
8810 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
8811 * tree-pretty-print.c (dump_generic_node): Likewise.
8813 2017-07-20 Jakub Jelinek <jakub@redhat.com>
8816 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
8817 V2TImode and V4TImode.
8818 (ix86_expand_vector_extract): Likewise.
8819 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
8820 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
8821 (ssescalarmode): Handle V4TImode and V2TImode.
8822 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
8823 (*vec_extractv2ti, *vec_extractv4ti): New insns.
8824 (VEXTRACTI128_MODE): New mode iterator.
8825 (splitter for *vec_extractv?ti first element): New.
8826 (VEC_INIT_MODE): New mode iterator.
8827 (vec_init<mode>): Consolidate 3 expanders into one using
8828 VEC_INIT_MODE mode iterator.
8830 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
8832 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
8833 non_spilled_static_chain_regno_p.
8835 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
8837 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
8839 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
8841 * bb-reorder.c (connect_traces): Allow copying of blocks within
8844 2017-07-20 Richard Biener <rguenther@suse.de>
8846 * gimple.h (gimple_phi_result): Add gphi * overload.
8847 (gimple_phi_result_ptr): Likewise.
8848 (gimple_phi_arg): Likewise. Adjust index assert to only
8849 allow actual argument accesses rather than all slots available
8851 (gimple_phi_arg_def): Add gphi * overload.
8852 * tree-phinodes.c (make_phi_node): Initialize only actual
8854 (resize_phi_node): Clear memory not covered by old node,
8855 do not initialize excess argument slots.
8856 (reserve_phi_args_for_new_edge): Initialize new argument slot
8859 2017-07-20 Bin Cheng <bin.cheng@arm.com>
8861 PR tree-optimization/81388
8863 2016-07-21 Bin Cheng <bin.cheng@arm.com>
8865 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
8866 by removing computation of may_be_zero.
8868 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
8869 Tom de Vries <tom@codesourcery.com>
8872 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
8873 when gimple level profile disagrees with what RTL expander did.
8875 2017-07-20 Richard Biener <rguenther@suse.de>
8877 PR tree-optimization/61171
8878 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
8879 (vect_analyze_stmt): Add slp instance parameter.
8880 (vectorizable_reduction): Likewise.
8881 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
8882 (vect_is_simple_reduction): Deal with chains not detected
8883 as SLP reduction chain, specifically not properly associated
8884 chains containing a mix of plus/minus.
8885 (get_reduction_op): Remove.
8886 (get_initial_defs_for_reduction): Simplify, pass in whether
8887 this is a reduction chain, pass in the SLP node for the PHIs.
8888 (vect_create_epilog_for_reduction): Get the SLP instance as
8890 (vectorizable_reduction): Get the SLP instance as arg.
8891 During analysis remember the SLP node with the PHIs in the
8892 instance. Simplify getting at the vectorized reduction PHIs.
8893 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
8894 through SLP instance.
8895 (vect_slp_analyze_operations): Likewise.
8896 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
8897 (vect_transform_stmt): Likewise.
8899 2017-07-20 Tom de Vries <tom@codesourcery.com>
8901 PR tree-optimization/81489
8902 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
8903 read of phi arg location to before loop that modifies phi.
8905 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
8907 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
8910 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
8913 * except.c (execute): Fix ordering issue.
8915 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
8917 PR rtl-optimization/81423
8918 * combine.c (make_compound_operation_int): Don't try to optimize
8919 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
8921 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
8923 PR rtl-optimization/81423
8924 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
8925 with a constant that is -1 in the truncated to mode.
8927 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
8929 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
8930 (determine_unlikely_bbs): ... here.
8931 * predict.h (propagate_unlikely_bbs_forward): Declare.
8932 * cfgexpand.c (pass_expand::execute): Use it.
8933 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
8935 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
8936 propagate_unlikely_bbs_forward.
8938 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
8941 * except.c (maybe_add_nop_after_section_switch): New function.
8944 2017-07-19 Tom de Vries <tom@codesourcery.com>
8946 * gimple.h (gimple_phi_set_arg): Make assert more strict.
8948 2017-07-19 Tom de Vries <tom@codesourcery.com>
8950 * gimple.h (gimple_phi_arg): Make assert more strict.
8952 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
8954 * config.gcc (powerpc*-*-*): Add mmintrin.h.
8955 * config/rs6000/mmintrin.h: New file.
8956 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
8958 2017-07-19 Jakub Jelinek <jakub@redhat.com>
8960 PR tree-optimization/81346
8961 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
8963 2017-07-19 Tom de Vries <tom@codesourcery.com>
8965 * config/nvptx/nvptx.md (VECIM): Add V2DI.
8967 2017-07-19 Tom de Vries <tom@codesourcery.com>
8969 * config/nvptx/nvptx-modes.def: Add V2DImode.
8970 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
8971 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
8972 (nvptx_output_mov_insn): Handle lack of mov.b128.
8973 (nvptx_print_operand): Handle 'H' and 'L' codes.
8974 (nvptx_vector_mode_supported): Allow V2DImode.
8975 (nvptx_preferred_simd_mode): New function.
8976 (nvptx_data_alignment): New function.
8977 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
8978 nvptx_preferred_simd_mode.
8979 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
8981 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
8983 2017-07-19 Tom de Vries <tom@codesourcery.com>
8985 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
8986 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
8987 (nvptx_vector_mode_supported): New function. Allow V2SImode.
8988 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
8989 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
8990 (mov<VECIM>_insn): New define_insn.
8991 (define_expand "mov<VECIM>): New define_expand.
8993 2017-07-19 Tom de Vries <tom@codesourcery.com>
8995 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
8997 2017-07-19 Jakub Jelinek <jakub@redhat.com>
8999 PR tree-optimization/81346
9000 * fold-const.h (fold_div_compare, range_check_type): Declare.
9001 * fold-const.c (range_check_type): New function.
9002 (build_range_check): Use range_check_type.
9003 (fold_div_compare): No longer static, rewritten into
9004 a match.pd helper function.
9005 (fold_comparison): Don't call fold_div_compare here.
9006 * match.pd (X / C1 op C2): New optimization using fold_div_compare
9009 2017-07-19 Nathan Sidwell <nathan@acm.org>
9011 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
9012 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
9013 * tree.c (find_decls_types_r, verify_type): Use
9014 TYPE_{MIN,MAX}_VALUE_RAW.
9015 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
9016 (hash_tree): Likewise.
9017 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
9019 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
9022 2017-07-18 Tom de Vries <tom@codesourcery.com>
9025 * omp-expand.c (expand_omp_for_static_chunk): Handle
9026 equal-argument loop exit phi.
9028 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
9031 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
9032 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
9033 operand 2 predicate.
9034 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
9035 operand 2 predicate.
9036 (ror,rol -> rorx splitters): Use const_int_operand as
9037 operand 2 predicate.
9039 2017-06-18 Richard Biener <rguenther@suse.de>
9041 PR tree-optimization/81410
9042 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
9043 the gap in the ! slp_perm SLP case after each group.
9045 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
9048 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
9051 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
9054 * predict.c (set_even_probabilities): Cleanup; do not affect
9055 probabilities that are already known.
9056 (combine_predictions_for_bb): Call even when count is set.
9058 2017-07-18 Nathan Sidwell <nathan@acm.org>
9060 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
9063 2017-07-18 Bin Cheng <bin.cheng@arm.com>
9066 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
9067 optimization for loop niter analysis.
9069 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
9072 * config/avr/avr.c (avr_optimize_casesi): Don't use
9073 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
9075 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
9077 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
9078 body_cost_vec from _vect_peel_extended_info.
9079 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
9080 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
9083 2017-07-18 Bin Cheng <bin.cheng@arm.com>
9085 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
9087 2017-07-18 Richard Biener <rguenther@suse.de>
9089 PR tree-optimization/80620
9090 PR tree-optimization/81403
9091 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
9092 info when re-using a VN table entry.
9094 2017-07-18 Richard Biener <rguenther@suse.de>
9096 PR tree-optimization/81418
9097 * tree-vect-loop.c (vectorizable_reduction): Properly compute
9098 vectype_in. Verify that with lane-reducing reduction operations
9099 we have a single def-use cycle.
9101 2017-07-17 Carl Love <cel@us.ibm.com>
9103 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
9105 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9106 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
9107 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
9108 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
9109 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
9110 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
9111 VMULOSW): New enum "unspec" values.
9112 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
9113 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
9114 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
9115 altivec_vmulosw): New patterns.
9116 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
9117 VMULOSW): Add definitions.
9118 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
9120 * config/alpha/alpha.c: Include predict.h.
9122 2017-07-17 Yury Gribov <tetra2005@gmail.com>
9124 * tree-vrp.c (compare_assert_loc): Fix comparison function
9125 to return predictable results.
9127 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
9129 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
9132 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
9133 * doc/invoke.texi (mexpand-adddi): Update text.
9135 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
9137 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
9138 that also clobbers the CC register. The old expand code is moved
9140 (*arc_clzsi2): ... here.
9141 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
9142 the CC register. The old expand code is moved to ...
9143 (arc_ctzsi2): ... here.
9145 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
9147 * config/arc/arc.opt (mindexed-loads): Use initial value
9148 TARGET_INDEXED_LOADS_DEFAULT.
9149 (mauto-modify-reg): Use initial value
9150 TARGET_AUTO_MODIFY_REG_DEFAULT.
9151 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
9152 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
9153 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
9154 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
9156 2017-07-17 Martin Liska <mliska@suse.cz>
9159 * opts.c (finish_options): Do not allow -fgnu-tm
9160 w/ -fsanitize={kernel-,}address. Say sorry.
9162 2017-07-17 Bin Cheng <bin.cheng@arm.com>
9165 * tree-loop-distribution.c (classify_partition): Only assert on
9166 numer of iterations.
9167 (merge_dep_scc_partitions): Delete prameter. Update function call.
9168 (distribute_loop): Remove code handling loop with unknown niters.
9169 (pass_loop_distribution::execute): Skip loop with unknown niters.
9171 2017-07-17 Bin Cheng <bin.cheng@arm.com>
9174 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
9175 function sort_partitions_by_post_order.
9177 2017-07-17 Bin Cheng <bin.cheng@arm.com>
9179 PR tree-optimization/81374
9180 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
9181 the max index of basic blocks, rather than number of basic blocks.
9183 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
9185 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
9187 (arc_legitimate_pic_operand_p): Likewise.
9188 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
9190 (arc_needs_pcl_p): Likewise.
9191 (arc_legitimate_pc_offset_p): Likewise.
9192 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
9193 function is also used in constrains.md.
9194 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
9195 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
9196 PLUS. Only return true/false in known cases, otherwise assert.
9197 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
9198 is already called in arc_legitimate_constant_p.
9199 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
9201 (LEGITIMATE_PIC_OPERAND_P): Use
9202 arc_raw_symbolic_reference_mentioned_p function.
9203 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
9208 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
9209 Andrew Burgess <andrew.burgess@embecosm.com>
9211 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
9212 (arc_return_address_register): New function.
9213 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
9214 (arc_handle_fndecl_attribute): Add naked attribute.
9215 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
9216 (TARGET_WARN_FUNC_RETURN): Likewise.
9217 (arc_allocate_stack_slots_for_args): New function.
9218 (arc_warn_func_return): Likewise.
9219 (machine_function): Change type fn_type.
9220 (arc_compute_function_type): Consider new naked function type,
9221 change function return type.
9222 (arc_must_save_register): Adapt to handle new
9223 arc_compute_function_type's return type.
9224 (arc_expand_prologue): Likewise.
9225 (arc_expand_epilogue): Likewise.
9226 (arc_return_address_regs): Delete.
9227 (arc_return_address_register): New function.
9228 (arc_epilogue_uses): Use above function.
9229 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
9230 (arc_function_type): Change encoding, add naked type.
9231 (ARC_INTERRUPT_P): Change to handle the new encoding.
9232 (ARC_FAST_INTERRUPT_P): Likewise.
9233 (ARC_NORMAL_P): Define.
9234 (ARC_NAKED_P): Likewise.
9235 (arc_compute_function_type): Delete prototype.
9236 * config/arc/arc.md (in_ret_delay_slot): Use
9237 arc_return_address_register function.
9238 (simple_return): Likewise.
9239 (p_return_i): Likewise.
9241 2017-07-17 Jakub Jelinek <jakub@redhat.com>
9243 PR tree-optimization/81428
9244 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
9245 can't be built for those types.
9247 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
9249 Remove stuff dead since r239246.
9251 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
9252 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
9253 (avr_inform_devices): Remove dead stuff.
9255 2017-07-17 Tamar Christina <tamar.christina@arm.com>
9257 * config/arm/arm_neon.h: Fix softp typo.
9259 2017-07-17 Jakub Jelinek <jakub@redhat.com>
9261 PR tree-optimization/81365
9262 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
9263 aggregate moves onto bb predecessor edges, make sure there are no
9264 loads that could alias the lhs in between the start of bb and the
9267 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
9270 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
9271 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
9272 [LSHIFTRT, outer_code = TRUNCATE]: Same.
9274 2017-07-17 Jakub Jelinek <jakub@redhat.com>
9276 PR tree-optimization/81396
9277 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
9278 (init_symbolic_number): Initialize it to 1.
9279 (perform_symbolic_merge): Add n_ops from both operands into the new
9281 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
9282 without base_addr as useless if they need more than one operation.
9283 (bswap_replace): Handle !bswap case for NULL base_addr.
9285 2017-07-17 Tom de Vries <tom@codesourcery.com>
9288 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
9291 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
9293 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
9294 conditional builtin define __FIX_LEON3FT_B2BST.
9296 2017-07-17 Daniel Cederman <cederman@gaisler.com>
9298 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
9299 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
9302 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
9304 PR rtl-optimization/81424
9305 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
9306 to remove potential trapping from operands if -fnon-call-exceptions.
9308 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9310 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
9311 profile_proability for scalling.
9312 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
9314 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9316 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
9318 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9320 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
9321 fixpoint arithmetics.
9323 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9325 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
9326 fixpoint arithmetics.
9328 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9330 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
9331 fixpoint arithmetics.
9333 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9335 * profile-count.h (profile_probability::from_reg_br_prob_note,
9336 profile_probability::to_reg_br_prob_note): New functions.
9337 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
9338 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
9339 * predict.c (probability_reliable_p): Update.
9340 (edge_probability_reliable_p): Update.
9341 (br_prob_note_reliable_p): Update.
9342 (invert_br_probabilities): Update.
9343 (add_reg_br_prob_note): New function.
9344 (combine_predictions_for_insn): Update.
9345 * asan.c (asan_clear_shadow): Update.
9346 * cfgbuild.c (compute_outgoing_frequencies): Update.
9347 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
9348 (update_br_prob_note): Update.
9349 (rtl_verify_edges): Update.
9350 (purge_dead_edges): Update.
9351 (fixup_reorder_chain): Update.
9352 * emit-rtl.c (try_split): Update.
9353 * ifcvt.c (cond_exec_process_insns): Update.
9354 (cond_exec_process_if_block): Update.
9355 (dead_or_predicable): Update.
9356 * internal-fn.c (expand_addsub_overflow): Update.
9357 (expand_neg_overflow): Update.
9358 (expand_mul_overflow): Update.
9359 * loop-doloop.c (doloop_modify): Update.
9360 * loop-unroll.c (compare_and_jump_seq): Update.
9361 * optabs.c (emit_cmp_and_jump_insn_1): Update.
9362 * predict.h: Update.
9363 * reorg.c (mostly_true_jump): Update.
9365 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
9366 * config/alpha/alpha.c (emit_unlikely_jump): Update.
9367 * config/arc/arc.c: (emit_unlikely_jump): Update.
9368 * config/arm/arm.c: (emit_unlikely_jump): Update.
9369 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
9370 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
9371 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
9372 (ix86_print_operand): Update.
9373 (ix86_split_fp_branch): Update.
9374 (predict_jump): Update.
9375 * config/ia64/ia64.c (ia64_print_operand): Update.
9376 * config/mmix/mmix.c (mmix_print_operand): Update.
9377 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
9378 (rs6000_expand_split_stack_prologue): Update.
9379 * config/rs6000/rs6000.c: Update.
9380 * config/s390/s390.c (s390_expand_vec_strlen): Update.
9381 (s390_expand_vec_movstr): Update.
9382 (s390_expand_cs_tdsi): Update.
9383 (s390_expand_split_stack_prologue): Update.
9384 * config/sh/sh.c (sh_print_operand): Update.
9385 (expand_cbranchsi4): Update.
9386 (expand_cbranchdi4): Update.
9387 * config/sparc/sparc.c (output_v9branch): Update.
9388 * config/spu/spu.c (get_branch_target): Update.
9389 (ea_load_store_inline): Update.
9390 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
9391 * config/tilepro/tilepro.c: Update.
9393 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
9395 * gimplify.c (mostly_copy_tree_r): Revert latest change.
9396 (gimplify_save_expr): Likewise.
9398 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
9400 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
9402 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
9404 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
9406 * timevar.def (TV_IPA_FNSUMMARY): Define.
9408 2017-07-16 Daniel Cederman <cederman@gaisler.com>
9410 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
9411 to back store errata sensitive sequence from being generated.
9412 (sqrtdf2_fix): Likewise.
9414 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
9416 * tree-ssa-threadupdate.c (compute_path_counts,
9417 update_joiner_offpath_counts): Use profile_probability.
9419 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
9422 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
9424 * config/arm/arm-c.c (arm_cpu_builtins): Define
9425 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
9427 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
9429 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9430 array entries to represent __ieee128 versions of the
9431 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
9432 scalar_extract_sig, and scalar_insert_exp built-in functions.
9433 (altivec_resolve_overloaded_builtin): Add special case handling
9434 for the __builtin_scalar_insert_exp function, as represented by
9435 the P9V_BUILTIN_VEC_VSIEDP constant.
9436 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
9437 exponent support for __ieee128 argument.
9438 (VSESQP): Add scalar extract signature support for __ieee128
9440 (VSTDCNQP): Add scalar test negative support for __ieee128
9442 (VSIEQP): Add scalar insert exponent support for __int128 argument
9443 with __ieee128 result.
9444 (VSIEQPF): Add scalar insert exponent support for __ieee128
9445 argument with __ieee128 result.
9446 (VSTDCQP): Add scalar test data class support for __ieee128
9448 (VSTDCNQP): Add overload support for scalar test negative with
9450 (VSTDCQP): Add overload support for scalar test data class
9452 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
9454 (UNSPEC_VSX_SIEXPQP): New constant.
9455 (xsxexpqp): New insn for VSX scalar extract exponent quad
9457 (xsxsigqp): New insn for VSX scalar extract significand quad
9459 (xsiexpqpf): New insn for VSX scalar insert exponent quad
9460 precision with floating point argument.
9461 (xststdcqp): New expand for VSX scalar test data class quad
9463 (xststdcnegqp): New expand for VSX scalar test negative quad
9465 (xststdcqp): New insn to match expansions for VSX scalar test data
9466 class quad precision and VSX scalar test negative quad precision.
9467 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
9468 special case operand checking to enforce that second operand of
9469 VSX scalar test data class with quad precision argument is a 7-bit
9471 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
9472 prototypes and descriptions of __ieee128 versions of
9473 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
9474 scalar_test_data_class, and scalar_test_neg built-in functions.
9476 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9478 PR tree-optimization/81162
9479 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
9480 replace a negate with an add.
9482 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
9484 * doc/invoke.texi (arm/-mcpu): Document +crypto.
9486 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
9488 * config/arm/arm-c.c (arm_cpu_builtins): Define
9489 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
9491 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
9493 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
9494 (armv8-r): Set ARM Cortex-R52 as default CPU.
9495 * config/arm/arm-tables.opt: Regenerate.
9496 * config/arm/arm-tune.md: Regenerate.
9497 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
9499 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
9500 extension for -mcpu=cortex-r52.
9502 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
9504 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
9505 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
9506 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
9507 (fp-armv8): Define it as FP_ARMv8 only.
9508 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
9509 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
9511 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
9512 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
9513 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
9514 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
9515 than TARGET_FPU_ARMV8.
9516 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
9517 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
9518 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
9520 * config/arm/neon.md (neon_vrint): Likewise.
9521 (neon_vcvt): Likewise.
9522 (neon_<fmaxmin_op><mode>): Likewise.
9523 (<fmaxmin><mode>3): Likewise.
9524 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
9525 * config/arm/predicates.md (arm_cond_move_operator): Check against
9526 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
9528 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
9530 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
9533 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9535 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
9536 loop in comment with memset.
9538 2017-07-14 Martin Liska <mliska@suse.cz>
9540 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
9541 * dwarf2out.c (is_java): Remove the function.
9542 (output_pubname): Remove usage of the function.
9543 (lower_bound_default): Remove usage of DW_LANG_Java.
9544 (gen_compile_unit_die): Likewise.
9545 * gcc.c: Remove compiler defaults for .java and .zip files.
9546 * gimple-expr.c (remove_suffix): Change as there's no longer
9547 extension than 4-letter one.
9548 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
9549 (gimplify_save_expr): Likewise.
9550 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
9551 as it's possible even for other languages than Java.
9552 * langhooks.h (struct lang_hooks): Remove Java from a comment.
9553 * lto-opts.c (lto_write_options): Remove reference to Java.
9554 * opts.c (strip_off_ending): Update file extension handling.
9555 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
9556 * tree-eh.c (lower_resx): Likewise.
9557 * tree.c (free_lang_data_in_type): Remove dead code.
9558 (find_decls_types_r): Likewise.
9559 (build_common_builtin_nodes): Remove Java from a comment.
9560 (verify_type): Remove dead code.
9561 * varasm.c (assemble_external): Remove Java from a comment.
9563 2017-07-14 Martin Liska <mliska@suse.cz>
9565 * opts.c (finish_options): Add quotes.
9566 (common_handle_option): Likewise.
9568 2017-07-14 Martin Liska <mliska@suse.cz>
9570 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
9571 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
9573 * dwarf2out.c (lower_bound_default): Do not handle
9575 (gen_compile_unit_die): Likewise.
9576 * gcc.c: Remove default extension binding for GNU Pascal.
9577 * stmt.c: Remove Pascal language from a comment.
9578 * xcoffout.c: Likewise.
9580 2017-07-13 David Malcolm <dmalcolm@redhat.com>
9583 * diagnostic-show-locus.c (fixit_cmp): New function.
9584 (layout::layout): Sort m_fixit_hints.
9585 (column_range::column_range): Assert that the values are valid.
9586 (struct char_span): New struct.
9587 (correction::overwrite): New method.
9588 (struct source_line): New struct.
9589 (line_corrections::add_hint): Add assertions. Reimplement memcpy
9590 calls in terms of classes source_line and char_span, and
9591 correction::overwrite.
9592 (selftest::test_overlapped_fixit_printing_2): New function.
9593 (selftest::diagnostic_show_locus_c_tests): Call it.
9595 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
9597 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
9598 early if there is no lhs.
9600 2017-07-13 Martin Liska <mliska@suse.cz>
9602 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
9603 (gen_reference_type_die): Likewise.
9604 * stor-layout.c: Remove Pascal-related comment.
9606 2017-07-13 Martin Liska <mliska@suse.cz>
9608 * opts.c (finish_options): Add quotes to error messages.
9609 (parse_sanitizer_options): Likewise.
9611 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
9613 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
9615 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
9617 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
9619 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
9621 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
9623 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
9625 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
9628 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
9629 provides the hardware capability bits, define the macro
9630 __BUILTIN_CPU_SUPPORTS__.
9631 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
9632 if GLIBC does not provide the hardware capability bits. Add a
9633 gcc_unreachable call if the built-in cpu function is neither
9634 __builtin_cpu_is nor __builtin_cpu_supports.
9635 (rs6000_get_function_versions_dispatcher): Change the warning
9636 that an old GLIBC is used which does not export the capability
9637 bits to be an error.
9638 * doc/extend.texi (target_clones attribute): Document the
9639 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
9640 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
9641 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
9642 the macros defined by GCC if the newer GLIBC is available.
9644 2017-07-12 Jeff Law <law@redhat.com>
9646 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
9647 remaining includes slightly.
9648 * config/riscv/riscv-builtins.c: Include profile-count.h.
9650 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
9653 * config/avr/avr.c (avr_set_current_function): In diagnostic
9654 messages: Quote keywords and (parts of) identifiers.
9655 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
9658 2017-07-12 Carl Love <cel@us.ibm.com>
9660 * config/rs6000/rs6000-c.c: Add support for built-in functions
9661 vector bool char vec_revb (vector bool char);
9662 vector bool short vec_revb (vector short char);
9663 vector bool int vec_revb (vector bool int);
9664 vector bool long long vec_revb (vector bool long long);
9665 * doc/extend.texi: Update the built-in documentation file for the
9666 new built-in functions.
9668 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9670 * config/s390/s390.md: Remove movcc splitter.
9672 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9674 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
9675 load/store on condition.
9677 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
9680 * config/avr/avr.c (avr_encode_section_info)
9681 [progmem && !TREE_READONLY]: Error if progmem object needs
9684 2017-07-11 Michael Collison <michael.collison@arm.com>
9686 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
9689 2017-07-11 Carl Love <cel@us.ibm.com>
9691 * config/rs6000/rs6000-c.c: Add support for builtins
9692 vector unsigned int vec_parity_lsbb (vector signed int);
9693 vector unsigned int vec_parity_lsbb (vector unsigned int);
9694 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
9695 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
9696 vector unsigned long long vec_parity_lsbb (vector signed long long);
9697 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
9698 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
9699 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
9700 * doc/extend.texi: Update the built-in documentation file for the
9701 new built-in functions.
9703 2017-07-11 David Malcolm <dmalcolm@redhat.com>
9705 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
9706 (layout::m_primary_loc): New field.
9707 (layout::layout): Initialize new field. Move location filtering
9708 logic from here to...
9709 (layout::maybe_add_location_range): ...this new method. Add
9710 support for filtering to just the lines already specified by other
9712 (layout::will_show_line_p): New method.
9713 (gcc_rich_location::add_location_if_nearby): New method.
9714 (selftest::test_add_location_if_nearby): New test function.
9715 (selftest::diagnostic_show_locus_c_tests): Call it.
9716 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
9719 2017-07-11 Tom de Vries <tom@codesourcery.com>
9721 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
9722 (bb_first_real_insn): New function.
9723 (nvptx_single): Add extra initialization of broadcasted condition
9726 2017-07-11 Nathan Sidwell <nathan@acm.org>
9728 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
9730 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
9732 * doc/extend.texi (AVR Function Attributes): Remove weblink to
9733 Binutils doc as TEXI will mess them up.
9734 * doc/invoke.texi (AVR Options): Same here.
9736 2017-07-11 Daniel Cederman <cederman@gaisler.com>
9738 * config/sparc/sparc.opt (mfix-ut700): New option.
9739 (mfix-gr712rc): Likewise.
9740 (sparc_fix_b2bst): New variable.
9741 * doc/invoke.texi (SPARC options): Document them.
9742 (ARM options): Fix warnings.
9743 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
9744 instructions to prevent sequences that can trigger the store-store
9745 errata for certain LEON3FT processors.
9746 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
9747 (sparc_option_override): Set sparc_fix_b2bst appropriately.
9748 * config/sparc/sparc.md (fix_b2bst): New attribute.
9749 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
9751 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
9754 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
9756 (*rsqrtsf2_sse): Ditto.
9758 (div<mode>3): Macroize insn from divdf3 and divsf3
9759 using MODEF mode iterator.
9761 2017-07-10 Martin Sebor <msebor@redhat.com>
9763 PR tree-optimization/80397
9764 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
9765 instead of testing for equality to INTEGER_TYPE.
9767 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
9769 * config.gcc: Remove uclibc from arc target spec.
9771 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
9773 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
9775 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
9778 * lto-wrapper.c (remove_option): New function.
9779 (merge_and_complain): Merge PIC/PIE options more realistically.
9781 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
9783 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
9787 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
9788 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
9789 * config.in: Regenerate.
9790 * configure: Regenerate.
9791 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
9792 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
9793 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
9794 (TARGET_GASISR_PROLOGUES): ...target mask.
9795 * common/config/avr/avr-common.c
9796 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
9797 Set -mgas-isr-prologues.
9798 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
9799 INSERT_PASS_BEFORE for it.
9800 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
9801 * config/avr/avr.c (avr_option_override)
9802 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
9803 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
9804 (avr_attribute_table) <no_gccisr>: Add new function attribute.
9805 (avr_set_current_function) <is_no_gccisr>: Init machine field.
9806 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
9808 (make_avr_pass_pre_proep): New function.
9809 (emit_push_sfr) <treg>: Add argument to function and use it
9811 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
9812 and set machine->gasisr.yes.
9813 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
9814 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
9815 __gcc_isr.n_pushed to .L__stack_usage.
9816 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
9817 (avr_asm_final_postscan_insn): ...this new static function.
9818 * config/avr/avr.h (machine_function)
9819 <is_no_gccisr, use_L__stack_usage>: New fields.
9820 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
9821 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
9822 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
9823 (gasisr, *gasisr): New expander and insn.
9824 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
9825 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
9826 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
9828 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
9830 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
9833 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
9835 Move jump-tables out of .text again.
9838 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
9839 (ASM_OUTPUT_ADDR_VEC): New function.
9840 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
9841 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
9842 INSN_ADDRESSes as asm comment.
9843 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
9844 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
9845 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
9846 * config/avr/avr.md (*tablejump): Adjust comment.
9847 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
9848 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
9850 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
9851 (avr_output_addr_vec): New proto.
9852 (avr_log_t) <insn_addresses>: New field.
9854 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
9857 * config/i386/i386.c (ix86_function_arg_advance): Set
9858 outgoing_args_on_stack to true if there are outgoing arguments
9860 (ix86_function_arg): Likewise.
9861 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
9862 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
9863 * config/i386/i386.h (machine_function): Add
9864 outgoing_args_on_stack.
9866 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
9868 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
9870 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
9872 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
9874 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
9875 (REAL_H): Remove $(MACHMODE_H).
9876 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
9878 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
9879 $(MACHMODE_H) and double-int.h.
9880 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
9882 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
9885 2017-07-07 Andrew Pinski <apinski@cavium.com>
9887 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
9888 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
9890 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
9892 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
9893 Add warning if GCC was not configured to link against a GLIBC that
9894 exports the hardware capability bits.
9895 (make_resolver_func): Make resolver function private and not a
9896 COMDAT function. Create the name with clone_function_name instead
9897 of make_unique_name.
9900 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
9901 correct operand in doing the split.
9903 2017-07-07 Carl Love <cel@us.ibm.com>
9905 * config/rs6000/rs6000-c: Add support for built-in function
9906 vector unsigned short vec_pack_to_short_fp32 (vector float,
9908 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
9909 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
9910 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
9911 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
9912 (convert_4f32_8i16): Add define_expand.
9913 * doc/extend.texi: Update the built-in documentation file for the
9914 new built-in function.
9916 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9918 * config/sparc/m8.md: New file.
9919 * config/sparc/sparc.md: Include m8.md.
9921 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9923 * config/sparc/sparc.opt: New option -mvis4b.
9924 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
9925 (sparc_option_override): Handle VIS4B.
9926 (enum sparc_builtins): Define
9927 SPARC_BUILTIN_DICTUNPACK{8,16,32},
9928 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
9929 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
9930 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
9931 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
9932 (check_constant_argument): New function.
9933 (sparc_vis_init_builtins): Define builtins
9934 __builtin_vis_dictunpack{8,16,32},
9935 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
9936 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
9937 __builtin_vis_fpcmpde{8,16,32}shl and
9938 __builtin_vis_fpcmpur{8,16,32}shl.
9939 (sparc_expand_builtin): Check that the constant operands to
9940 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
9941 constant and in range.
9942 * config/sparc/sparc-c.c (sparc_target_macros): Handle
9944 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
9945 (SPARC_IMM5_P): Likewise.
9946 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
9947 (enabled): Handle vis4b.
9948 (UNSPEC_DICTUNPACK): New unspec.
9949 (UNSPEC_FPCMPSHL): Likewise.
9950 (UNSPEC_FPUCMPSHL): Likewise.
9951 (UNSPEC_FPCMPDESHL): Likewise.
9952 (UNSPEC_FPCMPURSHL): Likewise.
9953 (cpu_feature): New CPU feature `vis4b'.
9954 (dictunpack{8,16,32}): New insns.
9955 (FPCSMODE): New mode iterator.
9956 (fpcscond): New code iterator.
9957 (fpcsucond): Likewise.
9958 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
9959 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
9960 (fpcmpde{8,16,32}{si,di}shl): Likewise.
9961 (fpcmpur{8,16,32}{si,di}shl): Likewise.
9962 * config/sparc/constraints.md: Define constraints `q' for unsigned
9963 2-bit integer constants and `t' for unsigned 5-bit integer
9965 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
9967 (imm5_operand_dictunpack16): Likewise.
9968 (imm5_operand_dictunpack32): Likewise.
9969 (imm2_operand): Likewise.
9970 * doc/invoke.texi (SPARC Options): Document -mvis4b.
9971 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
9972 ditunpack* and fpcmp*shl builtins.
9974 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9976 * config.gcc: Handle m8 in --with-{cpu,tune} options.
9977 * config.in: Add HAVE_AS_SPARC6 define.
9978 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
9980 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
9982 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
9983 (CPP_CPU_SPEC): Handle m8.
9984 (ASM_CPU_SPEC): Likewise.
9985 * config/sparc/sparc-opts.h (enum processor_type): Add
9987 * config/sparc/sparc.c (m8_costs): New struct.
9988 (sparc_option_override): Handle TARGET_CPU_m8.
9989 (sparc32_initialize_trampoline): Likewise.
9990 (sparc64_initialize_trampoline): Likewise.
9991 (sparc_issue_rate): Likewise.
9992 (sparc_register_move_cost): Likewise.
9993 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
9994 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
9995 (ASM_CPU64_DEFAULT_SPEC): Likewise.
9996 (CPP_CPU_SPEC): Handle M8.
9997 (ASM_CPU_SPEC): Likewise.
9998 (AS_M8_FLAG): Define.
9999 * config/sparc/sparc.md: Add m8 to the cpu attribute.
10000 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
10001 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
10003 * configure: Regenerate.
10004 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
10007 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10009 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
10011 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
10012 ("*movdi_insn_sp32"): Do not set v3pipe.
10013 ("*movsi_insn"): Likewise.
10014 ("*movdi_insn_sp64"): Likewise.
10015 ("*movsf_insn"): Likewise.
10016 ("*movdf_insn_sp32"): Likewise.
10017 ("*movdf_insn_sp64"): Likewise.
10018 ("*zero_extendsidi2_insn_sp64"): Likewise.
10019 ("*sign_extendsidi2_insn"): Likewise.
10020 ("*mov<VM32:mode>_insn"): Likewise.
10021 ("*mov<VM64:mode>_insn_sp64"): Likewise.
10022 ("*mov<VM64:mode>_insn_sp32"): Likewise.
10023 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
10024 ("<vlop:code><VL:mode>3"): Likewise.
10025 ("*not_<vlop:code><VL:mode>3"): Likewise.
10026 ("*nand<VL:mode>_vis"): Likewise.
10027 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
10028 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
10029 ("one_cmpl<VL:mode>2"): Likewise.
10030 ("faligndata<VM64:mode>_vis"): Likewise.
10031 ("alignaddrsi_vis"): Likewise.
10032 ("alignaddrdi_vis"): Likweise.
10033 ("alignaddrlsi_vis"): Likewise.
10034 ("alignaddrldi_vis"): Likewise.
10035 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
10036 ("bmaskdi_vis"): Likewise.
10037 ("bmasksi_vis"): Likewise.
10038 ("bshuffle<VM64:mode>_vis"): Likewise.
10039 ("cmask8<P:mode>_vis"): Likewise.
10040 ("cmask16<P:mode>_vis"): Likewise.
10041 ("cmask32<P:mode>_vis"): Likewise.
10042 ("pdistn<P:mode>_vis"): Likewise.
10043 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
10045 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10047 * config/sparc/sparc.md ("subtype"): New insn attribute.
10048 ("*wrgsr_sp64"): Set insn subtype.
10049 ("*rdgsr_sp64"): Likewise.
10050 ("alignaddrsi_vis"): Likewise.
10051 ("alignaddrdi_vis"): Likewise.
10052 ("alignaddrlsi_vis"): Likewise.
10053 ("alignaddrldi_vis"): Likewise.
10054 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
10055 ("fexpand_vis"): Likewise.
10056 ("fpmerge_vis"): Likewise.
10057 ("faligndata<VM64:mode>_vis"): Likewise.
10058 ("bshuffle<VM64:mode>_vis"): Likewise.
10059 ("cmask8<P:mode>_vis"): Likewise.
10060 ("cmask16<P:mode>_vis"): Likewise.
10061 ("cmask32<P:mode>_vis"): Likewise.
10062 ("fchksm16_vis"): Likewise.
10063 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
10064 ("fmean16_vis"): Likewise.
10065 ("fp<plusminus_insn>64_vis"): Likewise.
10066 ("<plusminus_insn>v8qi3"): Likewise.
10067 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
10068 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
10069 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
10070 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
10071 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
10072 ("*movqi_insn"): Likewise.
10073 ("*movhi_insn"): Likewise.
10074 ("*movsi_insn"): Likewise.
10075 ("movsi_pic_gotdata_op"): Likewise.
10076 ("*movdi_insn_sp32"): Likewise.
10077 ("*movdi_insn_sp64"): Likewise.
10078 ("movdi_pic_gotdata_op"): Likewise.
10079 ("*movsf_insn"): Likewise.
10080 ("*movdf_insn_sp32"): Likewise.
10081 ("*movdf_insn_sp64"): Likewise.
10082 ("*zero_extendhisi2_insn"): Likewise.
10083 ("*zero_extendqihi2_insn"): Likewise.
10084 ("*zero_extendqisi2_insn"): Likewise.
10085 ("*zero_extendqidi2_insn"): Likewise.
10086 ("*zero_extendhidi2_insn"): Likewise.
10087 ("*zero_extendsidi2_insn_sp64"): Likewise.
10088 ("ldfsr"): Likewise.
10089 ("prefetch_64"): Likewise.
10090 ("prefetch_32"): Likewise.
10091 ("tie_ld32"): Likewise.
10092 ("tie_ld64"): Likewise.
10093 ("*tldo_ldub_sp32"): Likewise.
10094 ("*tldo_ldub1_sp32"): Likewise.
10095 ("*tldo_ldub2_sp32"): Likewise.
10096 ("*tldo_ldub_sp64"): Likewise.
10097 ("*tldo_ldub1_sp64"): Likewise.
10098 ("*tldo_ldub2_sp64"): Likewise.
10099 ("*tldo_ldub3_sp64"): Likewise.
10100 ("*tldo_lduh_sp32"): Likewise.
10101 ("*tldo_lduh1_sp32"): Likewise.
10102 ("*tldo_lduh_sp64"): Likewise.
10103 ("*tldo_lduh1_sp64"): Likewise.
10104 ("*tldo_lduh2_sp64"): Likewise.
10105 ("*tldo_lduw_sp32"): Likewise.
10106 ("*tldo_lduw_sp64"): Likewise.
10107 ("*tldo_lduw1_sp64"): Likewise.
10108 ("*tldo_ldx_sp64"): Likewise.
10109 ("*mov<VM32:mode>_insn"): Likewise.
10110 ("*mov<VM64:mode>_insn_sp64"): Likewise.
10111 ("*mov<VM64:mode>_insn_sp32"): Likewise.
10113 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10115 * config/sparc/sparc.md ("type"): New insn type viscmp.
10116 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
10118 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
10119 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
10120 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
10121 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
10123 ("n7_vis_logical_11cycle"): Likewise.
10124 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
10125 * config/sparc/niagara2.md ("niag3_vis": Likewise.
10126 * config/sparc/niagara.md ("niag_vis"): Likewise.
10127 * config/sparc/ultra3.md ("us3_fga"): Likewise.
10128 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
10130 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
10132 * config/sparc/sparc.md: New instruction type `bmask'.
10133 (bmaskdi_vis): Use the `bmask' type.
10134 (bmasksi_vis): Likewise.
10135 * config/sparc/ultra3.md (us3_array): Likewise.
10136 * config/sparc/niagara7.md (n7_array): Likewise.
10137 * config/sparc/niagara4.md (n4_array): Likewise.
10138 * config/sparc/niagara2.md (niag2_vis): Likewise.
10139 (niag3_vis): Likewise.
10140 * config/sparc/niagara.md (niag_vis): Likewise.
10142 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
10144 * ipa-comdats.c: Remove optimize check from gate.
10145 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
10146 for functions not optimized.
10147 (ipa_fn_summary_read): Skip optimize check.
10148 (ipa_fn_summary_write): Likewise.
10149 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
10151 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
10153 (can_inline_edge_p): Check flag_pcc_struct_return for match.
10154 (check_callers): Give up on caller which is not optimized.
10155 (inline_small_functions): Likewise.
10156 (ipa_inline): Do not give up when not optimizing.
10157 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
10158 away unoptimizes cdtors.
10159 (whole_program_function_and_variable_visibility): Do
10160 ipa_discover_readonly_nonaddressable_vars in LTO mode.
10161 * ipa.c (process_references): Do not check optimize.
10162 (symbol_table::remove_unreachable_nodes): Update optimize check.
10163 (set_writeonly_bit): Update optimize check.
10164 (pass_ipa_cdtor_merge::gate): Do not check optimize.
10165 (pass_ipa_single_use::gate): Remove.
10167 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
10169 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
10170 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
10171 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
10172 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
10173 permute_load, permute_store, adjust_extract, adjust_splat,
10174 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
10175 replace_swap_with_copy, dump_swap_insn_table,
10176 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
10177 recombine_lvx_pattern, recombine_stvx_pattern,
10178 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
10179 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
10180 to file rs6000-p8swap.c.
10181 * config/rs6000/rs6000-p8swap.c: New file.
10182 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
10183 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
10184 and rs6000*-*-* targets.
10186 2017-07-06 David Malcolm <dmalcolm@redhat.com>
10188 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
10190 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
10192 * lto-wrapper.c (merge_and_complain): Do not merge
10193 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
10194 fsigned_zeros, ftrapping_math, fwrapv.
10195 (append_compiler_options): Do not track these options.
10196 (append_linker_options): Likewie
10198 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
10200 * cgraphunit.c (cgraph_node::finalize_function): When
10201 !flag_toplevel_reorde set no_reorder flag.
10202 (varpool_node::finalize_decl): Likewise.
10203 (symbol_table::compile): Drop no toplevel reorder path.
10205 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
10207 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
10208 edges; zero probability is not better than uninitialized.
10210 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
10212 * asan.h (asan_sanitize_allocas_p): Declare.
10213 * asan.c (asan_sanitize_allocas_p): New function.
10214 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
10215 (handle_builtin_alloca): Likewise.
10216 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
10217 if !asan_sanitize_allocas_p.
10218 * params.def (asan-instrument-allocas): Add new option.
10219 * params.h (ASAN_PROTECT_ALLOCAS): Define.
10220 * opts.c (common_handle_option): Disable allocas sanitization for
10223 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
10225 * asan.c: Include gimple-fold.h.
10226 (get_last_alloca_addr): New function.
10227 (handle_builtin_stackrestore): Likewise.
10228 (handle_builtin_alloca): Likewise.
10229 (asan_emit_allocas_unpoison): Likewise.
10230 (get_mem_refs_of_builtin_call): Add new parameter, remove const
10231 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
10232 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
10233 (instrument_builtin_call): Pass gimple iterator to
10234 get_mem_refs_of_builtin_call.
10235 (last_alloca_addr): New global.
10236 * asan.h (asan_emit_allocas_unpoison): Declare.
10237 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
10238 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
10239 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
10240 if function calls alloca.
10241 * gimple-fold.c (replace_call_with_value): Remove static keyword.
10242 * gimple-fold.h (replace_call_with_value): Declare.
10243 * internal-fn.c: Include asan.h.
10244 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
10245 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
10247 2017-07-06 David Malcolm <dmalcolm@redhat.com>
10249 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
10250 (C_SELFTEST_FLAGS): New.
10251 (CPP_SELFTEST_FLAGS): New.
10252 (SELFTEST_DEPS): New, from deps of s-selftest.
10253 (C_SELFTEST_DEPS): New, from deps of s-selftest.
10254 (CPP_SELFTEST_DEPS): New.
10255 (selftest): Add dependency on s-selftest-c++.
10256 (s-selftest): Rename to...
10257 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
10258 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
10259 than SELFTEST_FLAGS.
10260 (selftest-gdb): Rename to...
10261 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
10263 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
10264 (selftest-valgrind): Rename to...
10265 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
10267 (selftest-valgrind): Reintroduce as an alias for
10268 selftest-c-valgrind.
10269 (s-selftest-c++): New.
10270 (selftest-c++-gdb): New.
10271 (selftest-c++-valgrind): New.
10273 2017-07-06 Olivier Hainque <hainque@adacore.com>
10275 * gcc.c (process_command): When deciding if undefined variables
10276 should be ignored when processing specs, accept "gcc -v" as well.
10278 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
10280 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
10281 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
10283 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
10285 * config/arm/arm-cpus.in (armv8-r): Add new entry.
10286 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
10287 * config/arm/arm-tables.opt: Regenerate.
10288 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
10290 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
10292 2017-07-06 Carl Love <cel@us.ibm.com>
10294 * ChangeLog: Clean up from mid air collision
10296 2017-07-06 Carl Love <cel@us.ibm.com>
10298 * config/rs6000/rs6000-c.c: Add support for built-in functions
10299 vector signed int vec_subc (vector signed int, vector signed int);
10300 vector signed __int128 vec_subc (vector signed __int128,
10301 vector signed __int128);
10302 vector unsigned __int128 vec_subc (vector unsigned __int128,
10303 vector unsigned __int128);
10304 vector signed int vec_sube (vector signed int, vector signed int,
10305 vector signed int);
10306 vector unsigned int vec_sube (vector unsigned int,
10307 vector unsigned int,
10308 vector unsigned int);
10309 vector signed __int128 vec_sube (vector signed __int128,
10310 vector signed __int128,
10311 vector signed__int128);
10312 vector unsigned __int128 vec_sube (vector unsigned __int128,
10313 vector unsigned __int128,
10314 vector unsigned __int128);
10315 vector signed int vec_subec (vector signed int, vector signed int,
10316 vector signed int);
10317 vector unsigned int vec_subec (vector unsigned int,
10318 vector unsigned int,
10319 vector unsigned int);
10320 vector signed __int128 vec_subec (vector signed __int128,
10321 vector signed __int128,
10322 vector signed__int128);
10323 vector unsigned __int128 vec_subec (vector unsigned __int128,
10324 vector unsigned __int128,
10325 vector unsigned __int128);
10326 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
10327 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
10328 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
10329 BU_ALTIVEC_OVERLOAD_X definitions.
10330 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
10331 * doc/extend.texi: Update the built-in documentation file for the new
10332 built-in functions.
10334 2017-07-06 David Malcolm <dmalcolm@redhat.com>
10337 * diagnostic-show-locus.c (layout::layout): Use start and finish
10338 spelling location for the start and finish of each range.
10339 * genmatch.c (linemap_client_expand_location_to_spelling_point):
10340 Add unused aspect param.
10341 * input.c (expand_location_1): Add "aspect" param, and use it
10342 to access the correct part of the location.
10343 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
10345 (expand_location_to_spelling_point): Likewise.
10346 (linemap_client_expand_location_to_spelling_point): Add "aspect"
10347 param, and pass it to expand_location_1.
10349 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
10351 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
10352 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
10353 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
10354 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
10355 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
10356 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
10357 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
10358 _mm_maskz_getmant_ss): New intrinsics.
10359 (__builtin_ia32_getexpss128_mask): Changed to ...
10360 __builtin_ia32_getexpss128_round ... this.
10361 (__builtin_ia32_getexpsd128_mask): Changed to ...
10362 __builtin_ia32_getexpsd128_round ... this.
10363 * config/i386/i386-builtin-types.def
10364 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
10365 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
10366 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
10367 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
10368 __builtin_ia32_getmantss_mask_round): New builtins.
10369 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
10370 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
10371 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
10372 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
10373 * config/i386/sse.md
10374 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
10375 avx512f_sgetexp<mode><mask_scalar_name>
10376 <round_saeonly_scalar_name> ... this.
10377 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
10378 %0, %1, %2<round_saeonly_op3>}): Changed to ...
10379 vgetexp<ssescalarmodesuffix>
10380 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
10381 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
10382 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
10383 avx512f_vgetmant<mode><mask_scalar_name>
10384 <round_saeonly_scalar_name> ... this.
10385 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
10386 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
10387 vgetmant<ssescalarmodesuffix>
10388 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
10389 %0<mask_scalar_operand4>, %1, %2
10390 <round_saeonly_scalar_mask_op4>, %3} ... this.
10391 * config/i386/subst.md (mask_scalar_operand4,
10392 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
10393 round_saeonly_scalar_nimm_predicate): New subst attributes.
10395 2017-07-06 Julia Koval <julia.koval@intel.com>
10397 * config/i386/i386.c (ix86_erase_embedded_rounding):
10398 Remove code for old rounding pattern.
10400 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
10402 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
10404 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
10406 * doc/sourcebuild.texi (Test Directives, Variants of
10407 dg-require-support): Add documentation for dg-require-stack-check.
10409 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
10411 * config/i386/subst.md (mask_scalar, round_scalar,
10412 round_saeonly_scalar): New meta-templates.
10413 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
10414 round_scalar_mask_operand3, round_scalar_mask_op3,
10415 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
10416 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
10417 round_saeonly_scalar_constraint,
10418 round_saeonly_scalar_prefix): New subst attribute.
10419 * config/i386/sse.md
10420 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
10421 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
10422 <round_scalar_name> ... this.
10423 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
10424 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
10425 <round_scalar_name> ... this.
10426 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
10427 <sse>_vm<code><mode>3<mask_scalar_name>
10428 <round_saeonly_scalar_name> ... this.
10429 (v<plusminus_mnemonic><ssescalarmodesuffix>
10430 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
10431 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
10432 v<plusminus_mnemonic><ssescalarmodesuffix>
10433 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
10434 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
10435 (v<multdiv_mnemonic><ssescalarmodesuffix>
10436 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
10437 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
10438 v<multdiv_mnemonic><ssescalarmodesuffix>
10439 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
10440 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
10441 (v<maxmin_float><ssescalarmodesuffix>
10442 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
10443 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
10444 v<maxmin_float><ssescalarmodesuffix>
10445 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
10446 %0<mask_scalar_operand3>, %1, %<iptr>2
10447 <round_saeonly_scalar_mask_op3>} ... this.
10449 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
10451 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
10452 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
10454 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
10455 Alan Hayward <alan.hayward@arm.com>
10456 David Sherwood <david.sherwood@arm.com>
10458 * combine.c (simplify_if_then_else): Remove "enum" before
10460 * compare-elim.c (can_eliminate_compare): Likewise.
10461 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
10463 (aarch64_lookup_simd_builtin_type): Likewise.
10464 (aarch64_simd_builtin_type): Likewise.
10465 (aarch64_init_simd_builtin_types): Likewise.
10466 (aarch64_simd_expand_args): Likewise.
10467 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
10469 (aarch64_reverse_mask): Likewise.
10470 (aarch64_simd_emit_reg_reg_move): Likewise.
10471 (aarch64_gen_adjusted_ldpstp): Likewise.
10472 (aarch64_ccmp_mode_to_code): Likewise.
10473 (aarch64_operands_ok_for_ldpstp): Likewise.
10474 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10475 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
10477 (aarch64_min_divisions_for_recip_mul): Likewise.
10478 (aarch64_reassociation_width): Likewise.
10479 (aarch64_get_condition_code_1): Likewise.
10480 (aarch64_simd_emit_reg_reg_move): Likewise.
10481 (aarch64_simd_attr_length_rglist): Likewise.
10482 (aarch64_reverse_mask): Likewise.
10483 (aarch64_operands_ok_for_ldpstp): Likewise.
10484 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10485 (aarch64_gen_adjusted_ldpstp): Likewise.
10486 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
10488 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
10489 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
10490 (arm_lookup_simd_builtin_type): Likewise.
10491 (arm_simd_builtin_type): Likewise.
10492 (arm_init_simd_builtin_types): Likewise.
10493 (arm_expand_builtin_args): Likewise.
10494 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
10495 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
10496 (ft32_setup_incoming_varargs): Likewise.
10497 (ft32_function_arg): Likewise.
10498 (ft32_function_arg_advance): Likewise.
10499 (ft32_pass_by_reference): Likewise.
10500 (ft32_arg_partial_bytes): Likewise.
10501 (ft32_valid_pointer_mode): Likewise.
10502 (ft32_addr_space_pointer_mode): Likewise.
10503 (ft32_addr_space_legitimate_address_p): Likewise.
10504 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
10506 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
10507 (ix86_emit_outlined_ms2sysv_restore): Likewise.
10508 (iamcu_alignment): Likewise.
10509 (canonicalize_vector_int_perm): Likewise.
10510 (ix86_noce_conversion_profitable_p): Likewise.
10511 (ix86_mpx_bound_mode): Likewise.
10512 (ix86_operands_ok_for_move_multiple): Likewise.
10513 * config/microblaze/microblaze-protos.h
10514 (microblaze_expand_conditional_branch_reg): Likewise.
10515 * config/microblaze/microblaze.c
10516 (microblaze_expand_conditional_branch_reg): Likewise.
10517 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
10519 (rs6000_reassociation_width): Likewise.
10520 (rs6000_invalid_binary_op): Likewise.
10521 (fusion_p9_p): Likewise.
10522 (emit_fusion_p9_load): Likewise.
10523 (emit_fusion_p9_store): Likewise.
10524 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
10526 (riscv_hard_regno_mode_ok_p): Likewise.
10527 (riscv_address_insns): Likewise.
10528 (riscv_split_symbol): Likewise.
10529 (riscv_legitimize_move): Likewise.
10530 (riscv_function_value): Likewise.
10531 (riscv_hard_regno_nregs): Likewise.
10532 (riscv_expand_builtin): Likewise.
10533 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
10534 (riscv_build_integer): Likewise.
10535 (riscv_split_integer): Likewise.
10536 (riscv_legitimate_constant_p): Likewise.
10537 (riscv_cannot_force_const_mem): Likewise.
10538 (riscv_regno_mode_ok_for_base_p): Likewise.
10539 (riscv_valid_base_register_p): Likewise.
10540 (riscv_valid_offset_p): Likewise.
10541 (riscv_valid_lo_sum_p): Likewise.
10542 (riscv_classify_address): Likewise.
10543 (riscv_legitimate_address_p): Likewise.
10544 (riscv_address_insns): Likewise.
10545 (riscv_load_store_insns): Likewise.
10546 (riscv_force_binary): Likewise.
10547 (riscv_split_symbol): Likewise.
10548 (riscv_force_address): Likewise.
10549 (riscv_legitimize_address): Likewise.
10550 (riscv_move_integer): Likewise.
10551 (riscv_legitimize_const_move): Likewise.
10552 (riscv_legitimize_move): Likewise.
10553 (riscv_address_cost): Likewise.
10554 (riscv_subword): Likewise.
10555 (riscv_output_move): Likewise.
10556 (riscv_canonicalize_int_order_test): Likewise.
10557 (riscv_emit_int_order_test): Likewise.
10558 (riscv_function_arg_boundary): Likewise.
10559 (riscv_pass_mode_in_fpr_p): Likewise.
10560 (riscv_pass_fpr_single): Likewise.
10561 (riscv_pass_fpr_pair): Likewise.
10562 (riscv_get_arg_info): Likewise.
10563 (riscv_function_arg): Likewise.
10564 (riscv_function_arg_advance): Likewise.
10565 (riscv_arg_partial_bytes): Likewise.
10566 (riscv_function_value): Likewise.
10567 (riscv_pass_by_reference): Likewise.
10568 (riscv_setup_incoming_varargs): Likewise.
10569 (riscv_print_operand): Likewise.
10570 (riscv_elf_select_rtx_section): Likewise.
10571 (riscv_save_restore_reg): Likewise.
10572 (riscv_for_each_saved_reg): Likewise.
10573 (riscv_register_move_cost): Likewise.
10574 (riscv_hard_regno_mode_ok_p): Likewise.
10575 (riscv_hard_regno_nregs): Likewise.
10576 (riscv_class_max_nregs): Likewise.
10577 (riscv_memory_move_cost): Likewise.
10578 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
10579 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
10580 (rl78_addr_space_address_mode): Likewise.
10581 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10583 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
10584 (rs6000_reassociation_width): Likewise.
10585 (rs6000_invalid_binary_op): Likewise.
10586 (fusion_p9_p): Likewise.
10587 (emit_fusion_p9_load): Likewise.
10588 (emit_fusion_p9_store): Likewise.
10589 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
10590 (ok_for_simple_move_operands): Likewise.
10591 (ok_for_simple_move_strict_operands): Likewise.
10592 (ok_for_simple_arith_logic_operands): Likewise.
10593 (visium_legitimize_reload_address): Likewise.
10594 (visium_select_cc_mode): Likewise.
10595 (output_cbranch): Likewise.
10596 (visium_split_double_move): Likewise.
10597 (visium_expand_copysign): Likewise.
10598 (visium_expand_int_cstore): Likewise.
10599 (visium_expand_fp_cstore): Likewise.
10600 * config/visium/visium.c (visium_pass_by_reference): Likewise.
10601 (visium_function_arg): Likewise.
10602 (visium_function_arg_advance): Likewise.
10603 (visium_libcall_value): Likewise.
10604 (visium_setup_incoming_varargs): Likewise.
10605 (visium_legitimate_constant_p): Likewise.
10606 (visium_legitimate_address_p): Likewise.
10607 (visium_legitimize_address): Likewise.
10608 (visium_secondary_reload): Likewise.
10609 (visium_register_move_cost): Likewise.
10610 (visium_memory_move_cost): Likewise.
10611 (prepare_move_operands): Likewise.
10612 (ok_for_simple_move_operands): Likewise.
10613 (ok_for_simple_move_strict_operands): Likewise.
10614 (ok_for_simple_arith_logic_operands): Likewise.
10615 (visium_function_value_1): Likewise.
10616 (rtx_ok_for_offset_p): Likewise.
10617 (visium_legitimize_reload_address): Likewise.
10618 (visium_split_double_move): Likewise.
10619 (visium_expand_copysign): Likewise.
10620 (visium_expand_int_cstore): Likewise.
10621 (visium_expand_fp_cstore): Likewise.
10622 (visium_split_cstore): Likewise.
10623 (visium_select_cc_mode): Likewise.
10624 (visium_split_cbranch): Likewise.
10625 (output_cbranch): Likewise.
10626 (visium_print_operand_address): Likewise.
10627 * expmed.c (flip_storage_order): Likewise.
10628 * expmed.h (emit_cstore): Likewise.
10629 (flip_storage_order): Likewise.
10630 * genrecog.c (validate_pattern): Likewise.
10631 * hsa-gen.c (gen_hsa_addr): Likewise.
10632 * internal-fn.c (expand_arith_overflow): Likewise.
10633 * ira-color.c (allocno_copy_cost_saving): Likewise.
10634 * lra-assigns.c (find_hard_regno_for_1): Likewise.
10635 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
10636 (process_invariant_for_inheritance): Likewise.
10637 * lra-eliminations.c (move_plus_up): Likewise.
10638 * omp-low.c (lower_oacc_reductions): Likewise.
10639 * simplify-rtx.c (simplify_subreg): Likewise.
10640 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
10641 (TARGET_CHKP_BOUND_MODE): Likewise..
10642 * targhooks.c (default_chkp_bound_mode): Likewise.
10643 (default_setup_incoming_vararg_bounds): Likewise.
10644 * targhooks.h (default_chkp_bound_mode): Likewise.
10645 (default_setup_incoming_vararg_bounds): Likewise.
10646 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
10647 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
10648 (have_whole_vector_shift): Likewise.
10649 * tree-vect-stmts.c (vectorizable_load): Likewise.
10650 * doc/tm.texi: Regenerate.
10652 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
10654 Graceful degrade if Binutils PR21472 is not available.
10657 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
10658 .rodata in flash test fails.
10659 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
10660 * confgure: Regenerate.
10661 * config.in: Regenerate.
10662 * config/avr/avr.c (avr_asm_named_section)
10663 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
10664 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
10665 (avr_asm_init_sections): Same.
10667 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
10669 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
10670 (fma<VH:mode>4_intrinsic): Likewise.
10671 (*fmsub<VCVTF:mode>4): Likewise.
10672 (*fmsub<VH:mode>4_intrinsic): Likewise.
10674 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
10677 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
10678 Don't depend on "optimize > 0".
10679 (out_movhi_r_mr, out_movqi_mr_r): Same.
10680 (out_movhi_mr_r, out_movqi_r_mr): Same.
10681 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
10682 io_address_operand on "optimize > 0".
10684 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10686 * tree-loop-distribution.c: Add general explanantion on the pass.
10687 (generate_loops_for_partition): Mark distributed loop.
10688 (pg_add_dependence_edges): New parameter. Handle alias data
10689 dependence specially and record it in the parameter if asked.
10690 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
10691 (init_partition_graph_vertices, add_partition_graph_edge): New.
10692 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
10693 (free_partition_graph_vdata, build_partition_graph): New.
10694 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
10695 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
10696 (data_ref_segment_size, latch_dominated_by_data_ref): New.
10697 (compute_alias_check_pairs, version_loop_by_alias_check): New.
10698 (version_for_distribution_p, finalize_partitions): New.
10699 (distribute_loop): Handle alias data dependence specially. Factor
10700 out loop fusion code as functions and call these functions.
10702 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10704 * tree-loop-distribution.c (classify_partition): New parameter and
10705 better handle reduction statement.
10706 (rdg_build_partitions): Revise comment.
10707 (distribute_loop): Compute statements in all partitions and pass it
10708 to classify_partition.
10710 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10712 * tree-loop-distribution.c (enum partition_type): New.
10713 (struct partition): New field type.
10714 (partition_merge_into): Add parameter. Update partition type.
10715 (data_dep_in_cycle_p, update_type_for_merge): New functions.
10716 (build_rdg_partition_for_vertex): Compute partition type.
10717 (rdg_build_partitions): Dump partition type.
10718 (distribute_loop): Update calls to partition_merge_into.
10720 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10722 * tree-loop-distribution.c (struct ddr_hasher): New.
10723 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
10725 (classify_partition): Call get_data_dependence.
10726 (pg_add_dependence_edges): Ditto.
10727 (distribute_loop): Release data dependence hash table.
10729 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10731 * tree-loop-distribution.c (ref_base_address): Delete.
10732 (similar_memory_accesses): Rename ...
10733 (share_memory_accesses): ... to this. Check if partitions access
10734 the same memory reference.
10735 (distribute_loop): Call share_memory_accesses.
10737 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10739 * tree-loop-distribution.c (struct partition): New field recording
10740 its data reference.
10741 (partition_alloc, partition_free): Init and release data refs.
10742 (partition_merge_into): Merge data refs.
10743 (build_rdg_partition_for_vertex): Collect data refs for partition.
10744 (pg_add_dependence_edges): Change parameters from vector to bitmap.
10746 (distribute_loop): Remve data refs from vertice data of partition
10749 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10751 * tree-loop-distribution.c (params.h): Include header file.
10752 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
10753 (datarefs_vec): New global var.
10754 (create_rdg_vertices): Use datarefs_vec directly.
10755 (free_rdg): Don't free data references.
10756 (build_rdg): Update use. Don't free data references.
10757 (distribute_loop): Compute global variable for data references.
10758 Bail out if there are too many data references.
10760 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10762 * tree-loop-distribution.c (loop_nest): New global var.
10763 (build_rdg): Use loop directly, rather than loop nest.
10764 (pg_add_dependence_edges): Remove loop nest parameter. Use global
10766 (distribute_loop): Compute global variable loop nest. Update use.
10768 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10770 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
10771 (partition_merge_into): New parameter. Dump reason for fusion.
10772 (distribute_loop): Update use of partition_merge_into.
10774 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10776 * tree-loop-distribution.c (bb_top_order_index): New.
10777 (bb_top_order_index_size, bb_top_order_cmp): New.
10778 (stmts_from_loop): Use topological order.
10779 (pass_loop_distribution::execute): Compute and release topological
10780 order for basic blocks.
10782 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10784 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
10787 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10789 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
10790 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
10791 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
10792 * internal-fn.def (LOOP_DIST_ALIAS): New.
10793 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
10794 (fold_loop_internal_call): ... this.
10795 (vect_loop_dist_alias_call): New function.
10796 (set_uid_loop_bbs): Call fold_loop_internal_call.
10797 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
10800 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
10803 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
10804 Require dead FLAGS_REG at the beginning of a peephole.
10806 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
10809 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
10810 arguments in the call to __builtin_ia32_sbb_u32.
10811 (_subborrow_u64): Swap _X and _Y arguments in the call to
10812 __builtin_ia32_sbb_u64.
10814 2017-07-04 Jakub Jelinek <jakub@redhat.com>
10817 * tree-vrp.c (compare_assert_loc): Turn into a function template
10818 with stable template parameter. Only test if a->e is NULL,
10819 !a->e == !b->e has been verified already. Use e == NULL or
10820 e != NULL instead of e or ! e tests. If stable is true, don't use
10821 iterative_hash_expr, on the other side allow a or b or both NULL
10822 and sort the NULLs last.
10823 (process_assert_insertions): Sort using compare_assert_loc<false>
10824 instead of compare_assert_loc, later sort using
10825 compare_assert_loc<true> before calling process_assert_insertions_for
10826 in a loop. Use break instead of continue once seen NULL pointer.
10828 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
10830 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
10831 Cortex-R7 and Cortex-R8 processors.
10833 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
10835 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
10836 uninitialized while src is not.
10838 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
10840 * common/config/arm/arm-common.c: Adjust include path for
10842 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
10843 (arm-cpu.h): Create in build directory. Adjust dependency rules.
10844 (arm-cpu-data.h): Likewise.
10845 (arm-cpu-cdata.h): Likewise.
10846 * config/arm/arm-cpu.h: Delete.
10847 * config/arm/arm-cpu-cdata.h: Delete.
10848 * config/arm/arm-cpu-data.h: Delete.
10850 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
10852 * config/arm/arm-cpus.in (cortex-a55): New.
10853 (cortex-a75): Likewise.
10854 (cortex-a75.cortex-a55): Likewise.
10855 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
10857 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
10858 * config/arm/arm-cpu-cdata.h: Regenerate.
10859 * config/arm/arm-cpu-data.h: Regenerate.
10860 * config/arm/arm-cpu.h: Regenerate.
10861 * config/arm/arm-tables.opt: Regenerate.
10862 * config/arm/arm-tune.md: Regenerate.
10864 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
10866 * haifa-sched.c (sched_create_recovery_edges): Update profile.
10868 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
10870 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
10873 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
10875 PR tree-optimization/81292
10876 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
10877 full_string_p, also call adjust_related_strinfos if the adjustment
10878 is simple, otherwise invalidate related strinfos.
10880 2017-07-04 Martin Liska <mliska@suse.cz>
10883 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
10884 newly created variable as DECL_IGNORED_P.
10886 2017-07-04 Martin Liska <mliska@suse.cz>
10889 * ipa-inline.c (inline_small_functions):
10890 Use xstrdup_for_dump.
10892 2017-07-04 Tom de Vries <tom@codesourcery.com>
10894 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
10896 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
10899 * config/darwin.c (darwin_function_switched_text_sections):
10902 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
10904 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
10906 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
10908 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
10910 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10912 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
10913 min_profitable_iters, and th as inclusive lower bounds.
10914 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
10915 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
10916 for min_profitable_iters and min_profitable_estimate.
10917 (vect_transform_loop): Treat th as an inclusive lower bound.
10918 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
10920 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
10923 * config/darwin.c (darwin_function_switched_text_sections):
10924 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
10925 in two pieces, and suppress the use of buf.
10927 2017-07-03 Nathan Sidwell <nathan@acm.org>
10929 * hash-table.h (hash_table_mod1): Fix indentation.
10931 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10933 PR middle-end/81290
10934 * predict.c (force_edge_cold): Be more careful about propagation
10936 * profile-count.h (profile_probability::guessed,
10937 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
10939 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
10941 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
10943 * doc/invoke.texi (rcpc architecture extension): Document it.
10945 2017-07-03 Richard Biener <rguenther@suse.de>
10947 PR tree-optimization/60510
10948 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
10949 the scalar reduction PHI and use it.
10950 (vectorizable_reduction): Properly guard the single_defuse_cycle
10951 path for non-SLP reduction chains where we cannot use it.
10952 Rework reduc_def/index and vector type deduction. Rework
10953 vector operand gathering during reduction op code-gen.
10954 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
10955 chains dissolve the chain and leave it to non-SLP reduction
10958 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10960 * tree-data-ref.h (dr_alignment): Declare.
10961 * tree-data-ref.c (dr_alignment): New function.
10962 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
10963 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
10965 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
10967 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10969 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
10970 and base_misalignment fields.
10971 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
10972 * tree-data-ref.c: Include builtins.h.
10973 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
10974 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
10975 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
10976 * tree-vect-data-refs.c: Include tree-cfg.h.
10977 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
10978 fields instead of calculating an alignment here.
10979 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
10980 innermost_loop_behavior fields.
10982 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10984 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
10986 (DR_STEP_ALIGNMENT): New macro.
10987 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
10988 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
10989 (create_data_ref): Print it.
10990 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
10991 to tell whether the step preserves vector (mis)alignment.
10992 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
10993 Move the check for an integer step and generalise to all INTEGER_CST.
10994 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
10995 Print the outer step alignment.
10997 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10999 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
11000 with offset_alignment.
11001 (DR_ALIGNED_TO): Delete.
11002 (DR_OFFSET_ALIGNMENT): New macro.
11003 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
11004 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
11005 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
11006 (create_data_ref): Likewise.
11007 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
11008 (vect_analyze_data_refs): Likewise.
11009 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
11010 creating dummy innermost behavior.
11012 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11014 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
11015 with a "innermost_loop_behavior *" and refeence tree.
11016 * tree-data-ref.c (dr_analyze_innermost): Likewise.
11017 (create_data_ref): Update call accordingly.
11018 * tree-predcom.c (find_looparound_phi): Likewise.
11020 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11022 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
11023 fields with dr_wrt_vec_loop.
11024 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
11025 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
11026 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
11027 (vect_dr_behavior): New function.
11028 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
11029 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
11030 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
11031 track whether the step preserves the misalignment.
11032 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
11033 Use vect_dr_behavior.
11034 (vect_setup_realignment): Update call accordingly.
11035 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
11036 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
11037 call to vect_create_addr_base_for_vector_ref.
11038 (vect_create_cond_for_align_checks): Likewise.
11039 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
11040 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
11041 (vect_recog_mask_conversion_pattern): Likewise.
11042 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
11043 (new_stmt_vec_info): Remove redundant zeroing.
11045 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
11047 * common/config/arm/arm-common.c (arm_be8_option): New function.
11048 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
11049 (ISA_ARMv6): Add isa_bit_be8.
11050 * config/arm/arm.h (arm_be8_option): Add prototype.
11051 (BE8_SPEC_FUNCTION): New define.
11052 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
11053 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
11054 (mlittle-endian): Similarly.
11055 (mbe8, mbe32): New options.
11056 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
11057 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
11059 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11061 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
11063 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11065 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
11066 (cleanup_tree_cfg_bb): Use it.
11067 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
11069 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
11071 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11074 * loop-doloop.c (add_test): Update profile.
11076 2017-07-03 Martin Liska <mliska@suse.cz>
11079 * sanopt.c (rewrite_usage_of_param): New function.
11080 (sanitize_rewrite_addressable_params): Likewise.
11081 (pass_sanopt::execute): Call rewrite_usage_of_param.
11083 2017-07-03 Richard Biener <rguenther@suse.de>
11085 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
11086 back to using VIEW_CONVERT_EXPR.
11088 2017-07-03 Martin Liska <mliska@suse.cz>
11091 * doc/extend.texi: Document when a resolver function is
11092 generated for target_clones.
11094 2017-07-03 Martin Liska <mliska@suse.cz>
11096 * asan.c (asan_emit_stack_protection): Unpoison just red zones
11097 and shadow memory of auto variables which are subject of
11098 use-after-scope sanitization.
11099 (asan_expand_mark_ifn): Add do set only when is_poison.
11101 2016-07-03 Richard Biener <rguenther@suse.de>
11103 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
11105 (vect_force_simple_reduction): Record reduction def -> phi mapping.
11106 (vectorizable_reduction): Perform reduction PHI creation when
11107 visiting a reduction PHI and adjust and simplify code generation
11108 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
11109 (vect_transform_loop): Visit reduction PHIs.
11110 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
11111 defs into the SLP tree.
11112 (vect_build_slp_tree): Reduction defs terminate the recursion.
11113 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
11115 (vect_get_vec_defs_for_stmt_copy): Export.
11116 (vect_get_vec_defs): Likewise.
11117 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
11119 (vect_get_vec_defs_for_stmt_copy): Declare.
11120 (vect_get_vec_defs): Likewise.
11122 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
11124 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
11125 parameter with a "loop" parameter and use it instead of the
11126 loop containing DR_STMT. Don't check simple_iv when doing
11127 BB analysis. Describe the two analysis modes in the comment.
11129 2017-07-03 Tom de Vries <tom@codesourcery.com>
11131 PR tree-optimization/69468
11132 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
11133 (find_same_succ_bb): Handle ignore_edge_flags.
11135 2017-07-03 Tom de Vries <tom@codesourcery.com>
11137 PR tree-optimization/81192
11138 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
11140 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
11142 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
11144 2017-07-03 Tom de Vries <tom@codesourcery.com>
11146 PR tree-optimization/81192
11147 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
11148 BB_SAME_SUCC (bb) == NULL.
11150 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11152 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
11155 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11157 * dumpfile.c: Include profile-count.h
11158 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
11160 (insert_cond_bb): Update profile.
11161 * tree-cfg.h (insert_cond_bb): Update prototype.
11162 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
11163 * tree-dump.c: Do not include tree-cfg.
11165 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11167 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
11169 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11171 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
11174 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
11176 * tree-complex.c (expand_complex_div_wide): update profile.
11178 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
11179 Alan Hayward <alan.hayward@arm.com>
11180 David Sherwood <david.sherwood@arm.com>
11182 * Makefile.in (MACHMODE_H): Remove insn-modes.h
11183 (CORETYPES_H): New define.
11184 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
11185 (insn-modes-inline.h, s-modes-inline-h): New rules.
11186 (generated_files): Add insn-modes-inline.h.
11187 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
11188 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
11189 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
11190 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
11191 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
11192 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
11193 (build/gencodes.o, build/genconditions.o): Likewise.
11194 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
11195 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
11196 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
11197 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
11198 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
11199 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
11200 * coretypes.h: Include everything up to real.h for generators.
11201 Include insn-modes.h first. Include wide-int-print.h after
11202 wide-int.h. Include insn-modes-inline.h and then machmode.h.
11203 * machmode.h: Don't include insn-modes.h here.
11204 * function-tests.c: Remove includes of signop.h, machmode.h,
11205 double-int.h and wide-int.h.
11207 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
11209 * optc-save-gen.awk: Likewise.
11210 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
11211 * godump.c: Remove include of wide-int-print.h.
11212 * pretty-print.h: Likewise.
11213 * wide-int-print.cc: Likewise.
11214 * wide-int.cc: Likewise.
11215 * hash-map-tests.c: Remove include of signop.h.
11216 * hash-set-tests.c: Likewise.
11217 * rtl-tests.c: Likewise.
11218 * mkconfig.sh: Remove include of machmode.h.
11219 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
11221 (emit_insn_modes_inline_h): ...this new function. Emit the code
11222 into an insn-modes-inline.h header file, adding appropriate
11223 include guards and end comments.
11224 (emit_insn_modes_c_header): Remove include of machmode.h.
11225 (emit_min_insn_modes_c_header): Include coretypes.h rather than
11227 (main): Handle -i flag and call emit_insn_modes_inline_h when
11230 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
11232 * tree-ssa-strlen.c (strinfo): Rename the length field to
11233 nonzero_chars. Add a full_string_p field.
11234 (compare_nonzero_chars, zero_length_string_p): New functions.
11235 (get_addr_stridx): Add an offset_out parameter.
11236 Use compare_nonzero_chars.
11237 (get_stridx): Update accordingly. Use compare_nonzero_chars.
11238 (new_strinfo): Update after above changes to strinfo.
11239 (set_endptr_and_length): Set full_string_p.
11240 (get_string_length): Update after above changes to strinfo.
11241 (unshare_strinfo): Update call to new_strinfo.
11242 (maybe_invalidate): Likewise.
11243 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
11244 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
11245 as a uhwi instead of an shwi. Update after above changes to
11246 strinfo and new_strinfo.
11247 (zero_length_string): Assert that chainsi contains full strings.
11248 Use zero_length_string_p. Update call to new_strinfo.
11249 (adjust_related_strinfos): Update after above changes to strinfo.
11250 Copy full_string_p from origsi.
11251 (adjust_last_stmt): Use zero_length_string_p.
11252 (handle_builtin_strlen): Update after above changes to strinfo and
11253 new_strinfo. Install the lhs as the string length if the previous
11254 entry didn't describe a full string.
11255 (handle_builtin_strchr): Update after above changes to strinfo
11257 (handle_builtin_strcpy): Likewise.
11258 (handle_builtin_strcat): Likewise.
11259 (handle_builtin_malloc): Likewise.
11260 (handle_pointer_plus): Likewise.
11261 (handle_builtin_memcpy): Likewise. Track nonzero characters
11262 that aren't necessarily followed by a nul terminator.
11263 (handle_char_store): Likewise.
11265 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
11267 PR tree-optimization/80769
11268 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
11269 for malloc and calloc. Document the new invariant that all related
11270 strinfos have delayed lengths or none do.
11271 (verify_related_strinfos): Move earlier in file.
11272 (set_endptr_and_length): New function, split out from...
11273 (get_string_length): ...here. Also set the lengths of related
11275 (zero_length_string): Assert that chainsi has known (rather than
11277 (adjust_related_strinfos): Likewise.
11279 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
11281 PR tree-optimization/81136
11282 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
11283 assert that two references with the same misalignment have the same
11284 compile-time misalignment if those compile-time misalignments
11287 2017-07-01 Andi Kleen <ak@linux.intel.com>
11289 * print-tree.c (print_node): Print all attributes.
11291 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11293 * cfg.c (scale_bbs_frequencies): New function.
11294 * cfg.h (scale_bbs_frequencies): Declare it.
11295 * cfgloopanal.c (single_likely_exit): Cleanup.
11296 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
11298 (scale_loop_profile): Likewise.
11299 (loop_version): Likewise.
11300 (create_empty_loop_on_edge): Update.
11301 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
11302 scale_loop_frequencies, scale_loop_profile, loopify,
11303 loop_version): Update prototypes.
11304 * modulo-sched.c (sms_schedule): Update.
11305 * predict.c (unlikely_executed_edge_p): Also check probability.
11306 (probably_never_executed_edge_p): Fix typo.
11307 * tree-if-conv.c (version_loop_for_if_conversion): Update.
11308 * tree-parloops.c (gen_parallel_loop): Update.
11309 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
11310 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
11311 * tree-ssa-loop-split.c (split_loop): Update.
11312 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
11313 * tree-vect-loop-manip.c (vect_do_peeling): Update.
11314 (vect_loop_versioning): Update.
11315 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
11317 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11319 * trans-mem.c (split_bb_make_tm_edge): Update profile.
11321 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11323 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
11324 to keep profile consistent.
11326 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11328 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
11329 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
11330 * profile-count.h (max_safe_multiplier): Make unsigned.
11331 (profile_count::guessed_zero): New.
11333 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11335 * bb-reorder.c (fix_up_crossing_landing_pad,
11336 fix_crossing_conditional_branches): Use make_single_succ_edge
11337 to keep profile consistent.
11339 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11341 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
11344 2017-07-01 Jakub Jelinek <jakub@redhat.com>
11347 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
11348 the right scopes, make sure cond_jump isn't preserved between multiple
11349 iterations. Search for fallthru edge whenever there are 3+ edges and
11350 use find_fallthru_edge for it.
11352 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
11354 Patch by Alexander Monakov <amonakov@ispras.ru>
11355 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
11356 probabilities consistently.
11358 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
11360 * pa.c (pa_expand_compare_and_swap_loop): Update call of
11361 emit_cmp_and_jump_insns.
11363 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
11366 * tree-inline.c (expand_call_inline): Combine profile statuses.
11368 2017-06-30 Andrew Pinski <apinski@cavium.com>
11370 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
11371 fold_stmt returned true.
11373 2017-06-30 Nathan Sidwell <nathan@acm.org>
11375 * ggc.h (empty_string): Delete.
11376 * cfgexpand.c (expand_asm_stmt): Use plain "".
11377 * optabs.c (expand_asm_memory_barrier): Likewise.
11378 * stringpool.c (empty_string): Delete.
11379 (digit_vector, digit_string): Delete.
11380 (ggc_alloc_string): Use plain "", don't optimize single digit
11381 strings. Use ggc_alloc_atomic.
11383 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
11385 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
11386 comparison set and one other set, use the cost of the non-comparison
11389 2017-06-30 Nathan Sidwell <nathan@acm.org>
11391 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
11394 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
11396 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
11397 loops. Remove now unneeded calls to gimple_switch_set_label() that
11398 just set removed labels to NULL_TREE.
11400 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
11402 * tree-ssanames.c (set_range_info_raw): Abstract from ...
11403 (set_range_info): ...here. Only call set_range_info_raw if domain
11405 (set_nonzero_bits): Call set_range_info_raw.
11406 * tree-ssanames.h (set_range_info_raw): New.
11408 2017-06-30 Jakub Jelinek <jakub@redhat.com>
11411 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
11412 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
11413 of nonimmediate_operand and <store_mask_constraint> instead of m
11414 for the input operand. For V8FI iterator, always split if input
11415 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
11416 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
11417 <store_mask_predicate> instead of register_operand and
11418 <store_mask_constraint> instead of v for the input operand. Make
11419 sure both operands aren't MEMs for if not <mask_applied>.
11421 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
11423 * lto-wrapper.c (copy_file) Close both file descriptors before
11426 2017-06-30 Martin Liska <mliska@suse.cz>
11429 * multiple_target.c (create_dispatcher_calls): Make ifunc
11430 also for function that don't have calls or are not referenced.
11432 2017-06-30 Richard Biener <rguenther@suse.de>
11434 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
11435 analyze the first scalar stmt. Move vector type computation
11436 for the BB case here from ...
11437 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
11438 live operation processing in the SLP case properly.
11440 2017-06-30 Richard Biener <rguenther@suse.de>
11442 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
11444 2017-06-30 Martin Liska <mliska@suse.cz>
11447 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
11448 before BUILT_IN_UNWIND_RESUME when ASAN is used.
11450 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
11452 * doc/invoke.texi (AArch64): Add missing options and remove redundant
11455 2017-06-30 Richard Biener <rguenther@suse.de>
11457 PR tree-optimization/81249
11458 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
11459 condition reduction result to original scalar type.
11461 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11463 * profile-count.h (enum profile_quality): Fix typos and whitespace
11466 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11468 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
11469 type for branch probabilities.
11471 2017-06-29 Julian Brown <julian@codesourcery.com>
11472 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11474 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
11475 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
11476 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
11477 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
11479 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11481 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
11482 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
11483 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
11484 CC usage from generic code to here.
11485 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
11486 CC usage into the target macros.
11488 2017-06-29 Maya Rashish <coypu@sdf.org>
11490 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
11493 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
11495 * arm/arm-builtins.c: Include profile-count.h
11496 * except.c (sjlj_emit_function_enter): Use
11497 profile_probability::unlikely.
11499 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
11501 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
11502 and tocrel_offset be pointer args rather than implicitly using
11504 (legitimate_constant_pool_address_p, rs6000_emit_move,
11505 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
11506 tocrel_offset and use in toc_relative_expr_p call.
11507 (print_operand, print_operand_address): Use static tocrel_base_oac
11508 and tocrel_offset_oac.
11509 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
11512 2017-06-29 Maya Rashish <coypu@sdf.org>
11514 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
11516 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
11518 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
11519 objects, take into account only the alignment of 'op0' and 'mode1' if
11522 2017-06-29 Steve Ellcey <sellcey@cavium.com>
11524 * ccmp.c (ccmp_tree_comparison_p): New function.
11525 (ccmp_candidate_p): Update to use above function.
11526 (get_compare_parts): New function.
11527 (expand_ccmp_next): Update to use new functions.
11528 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
11530 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
11531 take mode as argument.
11532 * ccmp.h (expand_ccmp_expr): Add mode as argument.
11533 * expr.c (expand_expr_real_1): Pass mode as argument.
11535 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
11537 * combine.c (combine_instructions): Print insns to dump_file, together
11540 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
11542 * asan.c (asan_emit_stack_protection): Update.
11543 (create_cond_insert_point): Update.
11544 * auto-profile.c (afdo_propagate_circuit): Update.
11545 * basic-block.h (struct edge_def): Turn probability to
11546 profile_probability.
11547 (EDGE_FREQUENCY): Update.
11548 * bb-reorder.c (find_traces_1_round): Update.
11549 (better_edge_p): Update.
11550 (sanitize_hot_paths): Update.
11551 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
11552 (make_single_succ_edge): Update.
11553 (check_bb_profile): Update.
11554 (dump_edge_info): Update.
11555 (update_bb_profile_for_threading): Update.
11556 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
11557 probabilitycount to 0.
11558 * cfgbuild.c (compute_outgoing_frequencies): Update.
11559 * cfgcleanup.c (try_forward_edges): Update.
11560 (outgoing_edges_match): Update.
11561 (try_crossjump_to_edge): Update.
11562 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
11563 (expand_gimple_tailcall): Update.
11564 (construct_init_block): Use make_single_succ_edge.
11565 (construct_exit_block): Use make_single_succ_edge.
11566 * cfghooks.c (verify_flow_info): Update.
11567 (redirect_edge_succ_nodup): Update.
11568 (split_edge): Update.
11569 (account_profile_record): Update.
11570 * cfgloopanal.c (single_likely_exit): Update.
11571 * cfgloopmanip.c (scale_loop_profile): Update.
11572 (set_zero_probability): Remove.
11573 (duplicate_loop_to_header_edge): Update.
11574 * cfgloopmanip.h (loop_version): Update prototype.
11575 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
11576 (force_nonfallthru_and_redirect): Update.
11577 (update_br_prob_note): Update.
11578 (rtl_verify_edges): Update.
11579 (purge_dead_edges): Update.
11580 (rtl_lv_add_condition_to_bb): Update.
11581 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
11582 * cgraphunit.c (init_lowered_empty_function): Update.
11583 (cgraph_node::expand_thunk): Update.
11584 * cilk-common.c: Include profile-count.h
11585 * dojump.c (inv): Remove.
11586 (jumpifnot): Update.
11587 (jumpifnot_1): Update.
11588 (do_jump_1): Update.
11590 (do_jump_by_parts_greater_rtx): Update.
11591 (do_compare_rtx_and_jump): Update.
11592 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
11593 do_jump_1. do_compare_rtx_and_jump): Update prototype.
11594 * dwarf2cfi.c: Include profile-count.h
11595 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
11596 (sjlj_emit_dispatch_table): Likewise.
11597 * explow.c: Include profile-count.h
11598 * expmed.c (emit_store_flag_force): Update.
11599 (do_cmp_and_jump): Update.
11600 * expr.c (compare_by_pieces_d::generate): Update.
11601 (compare_by_pieces_d::finish_mode): Update.
11602 (emit_block_move_via_loop): Update.
11603 (store_expr_with_bounds): Update.
11604 (store_constructor): Update.
11605 (expand_expr_real_2): Update.
11606 (expand_expr_real_1): Update.
11607 * expr.h (try_casesi, try_tablejump): Update prototypes.
11608 * gimple-pretty-print.c (dump_probability): Update.
11609 (dump_profile): New.
11610 (dump_gimple_label): Update.
11611 (dump_gimple_bb_header): Update.
11612 * graph.c (draw_cfg_node_succ_edges): Update.
11613 * hsa-gen.c (convert_switch_statements): Update.
11614 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
11615 (find_if_case_1): Update.
11616 (find_if_case_2): Update.
11617 * internal-fn.c (expand_arith_overflow_result_store): Update.
11618 (expand_addsub_overflow): Update.
11619 (expand_neg_overflow): Update.
11620 (expand_mul_overflow): Update.
11621 (expand_vector_ubsan_overflow): Update.
11622 * ipa-cp.c (good_cloning_opportunity_p): Update.
11623 * ipa-split.c (split_function): Use make_single_succ_edge.
11624 * ipa-utils.c (ipa_merge_profiles): Update.
11625 * loop-doloop.c (add_test): Update.
11626 (doloop_modify): Update.
11627 * loop-unroll.c (compare_and_jump_seq): Update.
11628 (unroll_loop_runtime_iterations): Update.
11629 * lra-constraints.c (lra_inheritance): Update.
11630 * lto-streamer-in.c (input_cfg): Update.
11631 * lto-streamer-out.c (output_cfg): Update.
11632 * mcf.c (adjust_cfg_counts): Update.
11633 * modulo-sched.c (sms_schedule): Update.
11634 * omp-expand.c (expand_omp_for_init_counts): Update.
11635 (extract_omp_for_update_vars): Update.
11636 (expand_omp_ordered_sink): Update.
11637 (expand_omp_for_ordered_loops): Update.
11638 (expand_omp_for_generic): Update.
11639 (expand_omp_for_static_nochunk): Update.
11640 (expand_omp_for_static_chunk): Update.
11641 (expand_cilk_for): Update.
11642 (expand_omp_simd): Update.
11643 (expand_omp_taskloop_for_outer): Update.
11644 (expand_omp_taskloop_for_inner): Update.
11645 * omp-simd-clone.c (simd_clone_adjust): Update.
11646 * optabs.c (expand_doubleword_shift): Update.
11647 (expand_abs): Update.
11648 (emit_cmp_and_jump_insn_1): Update.
11649 (expand_compare_and_swap_loop): Update.
11650 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
11651 * predict.c (predictable_edge_p): Update.
11652 (edge_probability_reliable_p): Update.
11653 (set_even_probabilities): Update.
11654 (combine_predictions_for_insn): Update.
11655 (combine_predictions_for_bb): Update.
11656 (propagate_freq): Update.
11657 (estimate_bb_frequencies): Update.
11658 (force_edge_cold): Update.
11659 * profile-count.c (profile_count::dump): Add missing space into dump.
11660 (profile_count::debug): Add newline.
11661 (profile_count::differs_from_p): Explicitly convert to unsigned.
11662 (profile_count::stream_in): Update.
11663 (profile_probability::dump): New member function.
11664 (profile_probability::debug): New member function.
11665 (profile_probability::differs_from_p): New member function.
11666 (profile_probability::differs_lot_from_p): New member function.
11667 (profile_probability::stream_in): New member function.
11668 (profile_probability::stream_out): New member function.
11669 * profile-count.h (profile_count_quality): Rename to ...
11670 (profile_quality): ... this one.
11671 (profile_probability): New.
11672 (profile_count): Update.
11673 * profile.c (compute_branch_probabilities): Update.
11674 * recog.c (peep2_attempt): Update.
11675 * sched-ebb.c (schedule_ebbs): Update.
11676 * sched-rgn.c (find_single_block_region): Update.
11677 (compute_dom_prob_ps): Update.
11678 (schedule_region): Update.
11679 * sel-sched-ir.c (compute_succs_info): Update.
11680 * stmt.c (struct case_node): Update.
11681 (do_jump_if_equal): Update.
11682 (get_outgoing_edge_probs): Update.
11683 (conditional_probability): Update.
11684 (emit_case_dispatch_table): Update.
11685 (expand_case): Update.
11686 (expand_sjlj_dispatch_table): Update.
11687 (emit_case_nodes): Update.
11688 * targhooks.c: Update.
11689 * tracer.c (better_p): Update.
11690 (find_best_successor): Update.
11691 * trans-mem.c (expand_transaction): Update.
11692 * tree-call-cdce.c: Update.
11693 * tree-cfg.c (gimple_split_edge): Upate.
11694 (move_sese_region_to_fn): Upate.
11695 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
11696 * tree-eh.c (lower_resx): Upate.
11697 (cleanup_empty_eh_move_lp): Upate.
11698 * tree-if-conv.c (version_loop_for_if_conversion): Update.
11699 * tree-inline.c (copy_edges_for_bb): Update.
11700 (copy_cfg_body): Update.
11701 * tree-parloops.c (gen_parallel_loop): Update.
11702 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
11703 (gimple_gen_time_profiler): Update.
11704 * tree-ssa-dce.c (remove_dead_stmt): Update.
11705 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
11706 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
11707 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
11708 (unloop_loops): Update.
11709 (try_peel_loop): Update.
11710 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
11711 * tree-ssa-loop-split.c (connect_loops): Update.
11712 (split_loop): Update.
11713 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
11714 (hoist_guard): Update.
11715 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
11716 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
11717 (value_replacement): Update.
11718 * tree-ssa-reassoc.c (branch_fixup): Update.
11719 * tree-ssa-tail-merge.c (replace_block_by): Update.
11720 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
11721 (create_edge_and_update_destination_phis): Update.
11722 (compute_path_counts): Update.
11723 (recompute_probabilities): Update.
11724 (update_joiner_offpath_counts): Update.
11725 (freqs_to_counts_path): Update.
11726 (duplicate_thread_path): Update.
11727 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
11728 (struct switch_conv_info): Update.
11729 (gen_inbound_check): Update.
11730 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
11731 (vect_do_peeling): Update.
11732 (vect_loop_versioning): Update.
11733 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
11734 (optimize_mask_stores): Update.
11735 * ubsan.c (ubsan_expand_null_ifn): Update.
11736 * value-prof.c (gimple_divmod_fixed_value): Update.
11737 (gimple_divmod_fixed_value_transform): Update.
11738 (gimple_mod_pow2): Update.
11739 (gimple_mod_pow2_value_transform): Update.
11740 (gimple_mod_subtract): Update.
11741 (gimple_mod_subtract_transform): Update.
11742 (gimple_ic): Update.
11743 (gimple_stringop_fixed_value): Update.
11744 (gimple_stringops_transform): Update.
11745 * value-prof.h: Update.
11747 2017-06-29 Carl Love <cel@us.ibm.com>
11749 * config/rs6000/rs6000-c.c: Add support for built-in functions
11750 vector signed int vec_signed (vector float);
11751 vector signed long long vec_signed (vector double);
11752 vector signed int vec_signed2 (vector double, vector double);
11753 vector signed int vec_signede (vector double);
11754 vector signed int vec_signedo (vector double);
11755 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
11756 instruction generator.
11757 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
11758 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
11759 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
11761 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
11762 vunsignede_v2df): Add define_expands.
11763 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
11764 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
11765 VEC_UNSIGNEDO): Add definitions.
11766 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
11767 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
11768 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
11769 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
11770 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
11771 * config/rs6000/altivec.h (vec_signed, vec_signed2,
11772 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
11773 vec_unsignede, vec_unsignedo): Add builtin defines.
11774 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
11776 * doc/extend.texi: Update the built-in documentation file for the
11777 new built-in functions.
11779 2017-06-29 Richard Biener <rguenther@suse.de>
11781 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
11782 reduction chains to LOOP_VINFO_REDUCTIONS.
11783 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
11784 SLP reductions after processing reduction chains.
11786 2017-06-29 Nathan Sidwell <nathan@acm.org>
11788 * builtins.c (fold_builtin_FUNCTION): Use
11789 lang_hooks.decl_printable_name.
11791 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
11793 PR middle-end/81194
11794 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
11795 with only one label.
11796 * stmt.c (expand_case): Assert NCASES is greater than one.
11798 2017-06-29 Richard Biener <rguenther@suse.de>
11800 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
11802 (group_case_labels): Likewise.
11803 (find_taken_edge): Push sanity checking on val to workers...
11804 (find_taken_edge_cond_expr): ... here
11805 (find_taken_edge_switch_expr): ... and here, handle cases
11806 with just a default label.
11807 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
11808 (group_case_labels): Likewise.
11809 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
11810 group_case_labels does anything cleanup the CFG again.
11812 2017-06-29 Bin Cheng <bin.cheng@arm.com>
11814 PR tree-optimization/81196
11815 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
11816 exit condition comparing two IVs.
11818 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
11820 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
11821 profile to the dummy entry at the end of the list of architectures.
11822 * config/arm/arm-cpu-cdata.h: Regenerated.
11824 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11825 Michael Collison <michael.collison@arm.com>
11828 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
11830 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
11831 (*aarch64_reg_<mode>3_minus_mask): New pattern.
11832 (*aarch64_<optab>_reg_di3_mask2): New pattern.
11833 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
11834 of shift when the shift amount is masked with constant equal to
11835 the size of the mode.
11836 * config/aarch64/predicates.md (subreg_lowpart_operator): New
11839 2017-06-29 Martin Liska <mliska@suse.cz>
11841 * config/i386/i386.opt: Change range from [1,5] to [0,5].
11843 2017-06-29 Yury Gribov <tetra2005@gmail.com>
11846 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
11849 (edge_growth_cache_entry::edge_growth_cache_entry): New
11851 (reset_edge_growth_cache): Update to use constructor.
11853 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
11855 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
11856 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
11857 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
11859 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
11861 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
11862 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
11864 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
11866 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
11867 (*-linux-uclibc*): Add t-uclibc tmake_file.
11868 * config/t-musl: New.
11869 * config/t-uclibc: New.
11871 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
11873 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
11875 (gen_comm_data): Emit architectural setting of arch_prof.
11876 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
11878 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
11879 (armv8-m.base, armv8-m.main): Likewise.
11880 * arm-protos.h (arm_build_target): Add profile field.
11881 (arch_option): Likewise.
11882 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
11884 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
11885 arm_active_target.profile.
11887 2017-06-28 Richard Biener <rguenther@suse.de>
11889 PR middle-end/81227
11890 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
11891 TYPE_OVERFLOW_WRAPS.
11892 * match.pd (negate_expr_p): Likewise.
11893 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
11894 fold_build2, not fold_binary.
11896 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
11898 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
11899 Convert memory address to Pmode.
11900 (aarch64_print_operand): Assert MEM operands are always Pmode.
11902 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
11905 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
11906 Remove redundant if.
11907 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
11908 * config/arm/aarch-common-protos.h
11909 (aarch_forward_to_shift_is_not_shifted_re): Remove.
11910 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
11912 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
11915 * multiple_target.c (create_dispatcher_calls): Set the default
11916 clone to be static, not public.
11918 2017-06-28 Richard Biener <rguenther@suse.de>
11920 * tree-vect-loop.c (vectorizable_reduction): Move special
11921 cond reduction IV var creation ...
11922 (vect_create_epilog_for_reduction): ... here. Remove induction_index
11923 parameter. Use STMT_VINFO_VECTYPE.
11924 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
11927 2017-06-28 Martin Liska <mliska@suse.cz>
11930 * ipa-visibility.c (non_local_p): Handle visibility.
11932 2017-06-28 Martin Liska <mliska@suse.cz>
11935 * common.opt: Add IntegerRange to various options.
11936 * opt-functions.awk (integer_range_info): New function.
11937 * optc-gen.awk: Add integer_range_info to cl_options struct.
11938 * opts-common.c (decode_cmdline_option): Handle
11939 CL_ERR_INT_RANGE_ARG.
11940 (cmdline_handle_error): Likewise.
11941 * opts.c (print_filtered_help): Show valid interval in
11942 when --help is provided.
11943 * opts.h (struct cl_option): Add range_min and range_max fields.
11944 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
11946 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
11948 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
11949 (x * C EQ/NE y * C): New transformation.
11951 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
11953 * genmultilib (combination_space): Accept '+' in option names.
11955 2017-06-28 Martin Liska <mliska@suse.cz>
11958 * asan.c (instrument_derefs): Bail out inner references
11959 that are hard register variables.
11961 2017-06-28 Jakub Jelinek <jakub@redhat.com>
11964 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
11965 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
11967 2017-06-28 Richard Biener <rguenther@suse.de>
11969 * tree-vectorizer.h (vect_get_vec_defs): Remove.
11970 (vect_get_slp_defs): Adjust.
11971 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
11973 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
11975 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
11976 get_initial_defs_for_reduction instead of vect_get_vec_defs.
11977 (vectorizable_reduction): Adjust.
11978 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
11980 (vect_get_slp_defs): Likewise.
11981 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
11982 (vectorizable_bswap): Adjust.
11983 (vectorizable_call): Likewise.
11984 (vectorizable_conversion): Likewise.
11985 (vectorizable_assignment): Likewise.
11986 (vectorizable_shift): Likewise.
11987 (vectorizable_operation): Likewise.
11988 (vectorizable_store): Likewise.
11989 (vectorizable_condition): Likewise.
11990 (vectorizable_comparison): Likewise.
11992 2017-06-28 Michael Collison <michael.collison@arm.com>
11995 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
11997 (arm_gen_movmemqi): Removed unused variable 'i'.
11998 Convert 'for' loop into 'while' loop.
11999 (arm_expand_prologue): Remove last unnecessary set of insn.
12000 (thumb_pop): Remove unused variable 'pushed_words'.
12001 (thumb_exit): Remove last unnecessary set of regs_to_pop.
12003 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12005 * config/s390/predicates.md: Use s390_rel_address_ok_p.
12006 * config/s390/s390-protos.h: Add prototype of
12007 s390_rel_address_ok_p.
12008 * config/s390/s390.c (s390_got_symbol): New function.
12009 (s390_rel_address_ok_p): New function.
12010 (legitimize_pic_address): Use s390_rel_address_ok_p.
12011 (s390_load_got): Use s390_got_symbol.
12012 (s390_option_override): Issue error if
12013 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
12014 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
12016 * config/s390/s390.opt: New option mpic-data-is-text-relative.
12018 2017-06-27 Andrew Pinski <apinski@cavium.com>
12020 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
12021 (X * copysign (1.0, X)): New pattern.
12022 (X * copysign (1.0, -X)): New pattern.
12023 (copysign (-1.0, CST)): New pattern.
12025 2017-06-27 Joseph Myers <joseph@codesourcery.com>
12027 * genmultilib (combination_space): Remove variable.
12028 Validate reuse rules against regular expression for any sequence
12029 of multilib options in any order.
12031 2017-06-27 Michael Collison <michael.collison@arm.com>
12033 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
12034 call aarch64_split_simd_combine.
12035 * (aarch64_combine_internal<mode>): Delete pattern.
12036 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
12037 Allow register and subreg operands.
12039 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
12041 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
12042 specific need, just fallback on defaults.
12043 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
12045 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
12046 Olivier Hainque <hainque@adacore.com>
12048 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
12050 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
12051 targets. Pick a default if no particular attempt applied.
12052 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
12055 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
12057 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
12058 (x86_64-wrs-vxworks7): Likewise.
12060 2017-06-27 Marek Polacek <polacek@redhat.com>
12063 * ubsan.c (instrument_null): Check get_base_address's result for null.
12065 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
12067 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
12069 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
12071 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
12072 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
12073 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
12074 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
12075 New function types.
12076 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
12077 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
12078 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
12079 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
12080 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
12081 BUILT_IN_FEUPDATEENV): New builtins.
12082 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
12083 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
12084 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
12085 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
12087 (builtin_structptr_types): Adjust size.
12088 * tree.c (builtin_structptr_types): Add four entries.
12090 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
12091 Olivier Hainque <hainque@adacore.com>
12093 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
12094 (TLS_SYM): New local macro, forcing reference to __tls__ on
12095 link command lines for VxWorks 7 RTPs, triggering initialization
12097 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
12098 OS features TLS support, true for RTPs on VxWorks 7.
12099 * config/vxworks.c (vxworks_override_options): Setup emutls
12102 2017-06-27 Jakub Jelinek <jakub@redhat.com>
12104 * predict.c (test_prediction_value_range): Use -1U instead of -1
12105 to avoid narrowing conversion warning.
12106 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
12107 to avoid narrowing conversion warning.
12108 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
12110 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
12112 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
12114 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
12115 64bit configurations.
12116 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
12117 (SIZE_TYPE): Likewise.
12118 * config/vxworks.c (vxworks_emutls_var_fields): Use
12119 long_unsigned_type_node instead of unsigned_type_node as the offset
12120 field type, which is "pointer" mode in emutls.c.
12122 2017-06-27 Jakub Jelinek <jakub@redhat.com>
12125 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
12127 PR middle-end/81207
12128 * gimple-fold.c (replace_call_with_call_and_fold): Handle
12129 gimple_vuse copying separately from gimple_vdef copying.
12131 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
12133 * value-prof.c (free_hist): Remove call to memset and the enclosing if
12136 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
12137 Olivier Hainque <hainque@adacore.com>
12139 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
12140 for all vxworks7 targets.
12141 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
12142 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
12143 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
12144 variations for VX6/VX7 and 32/64bits later on in ...
12145 (VXWORKS_LIB_SPEC): Leverage new macros.
12146 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
12147 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
12149 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
12151 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
12152 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
12154 2017-06-26 Carl Love <cel@us.ibm.com>
12156 * config/rs6000/rs6000-c.c: Add support for built-in functions
12157 vector bool char vec_reve (vector bool char);
12158 vector signed char vec_reve (vector signed char);
12159 vector unsigned char vec_reve (vector unsigned char);
12160 vector bool int vec_reve (vector bool int);
12161 vector signed int vec_reve (vector signed int);
12162 vector unsigned int vec_reve (vector unsigned int);
12163 vector bool long long vec_reve (vector bool long long);
12164 vector signed long long vec_reve (vector signed long long);
12165 vector unsigned long long vec_reve (vector unsigned long long);
12166 vector bool short vec_reve (vector bool short);
12167 vector signed short vec_reve (vector signed short);
12168 vector double vec_reve (vector double);
12169 vector float vec_reve (vector float);
12170 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
12171 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
12172 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
12173 (altivec_vreve): New pattern.
12174 * config/rs6000/altivec.h (vec_reve): New define.
12175 * doc/extend.texi (vec_rev): Update the built-in documentation file
12176 for the new built-in functions.
12178 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12180 PR tree-optimization/71815
12181 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
12183 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
12185 (slsr_process_phi): Likewise.
12186 (replace_uncond_cands_and_profitable_phis): Don't replace a
12187 multiply candidate with a stride of 1 (copy or cast).
12188 (phi_incr_cost): Call uses_consumed_by_stmt rather than
12190 (lowest_cost_path): Likewise.
12191 (total_savings): Likewise.
12193 2017-06-26 Richard Biener <rguenther@suse.de>
12196 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
12197 Use def_builtin_pure for all gather builtins.
12199 2017-06-26 Richard Biener <rguenther@suse.de>
12201 PR tree-optimization/81203
12202 * tree-tailcall.c (find_tail_calls): Do not move stmts into
12203 non-dominating BBs.
12205 2017-06-26 Marek Polacek <polacek@redhat.com>
12208 * doc/invoke.texi: Document -Wmultistatement-macros.
12210 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
12212 * doc/sourcebuild.texi (ARM-specific attributes): Document new
12213 arm_neon_ok_no_float_abi effective target.
12215 2017-06-26 Richard Biener <rguenther@suse.de>
12217 PR tree-optimization/80928
12218 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
12219 (copy_bbs): Set BB_DUPLICATED flag early.
12220 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
12222 (execute_on_shrinking_pred): Likewise.
12223 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
12224 BB_DUPLICATED blocks.
12225 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
12226 iterate over all PHIs considering removal of *gsi.
12228 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
12230 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
12233 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
12235 * config/rs6000/rs6000-string.c: (expand_block_clear,
12236 do_load_for_compare, select_block_compare_mode,
12237 compute_current_alignment, expand_block_compare,
12238 expand_strncmp_align_check, expand_strn_compare,
12239 expand_block_move, rs6000_output_load_multiple)
12240 Move functions related to string/block move/compare
12241 to a separate file.
12242 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
12243 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
12244 for this function which is now used in two files.
12245 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
12246 * config.gcc: Add rs6000-string.o to extra_objs for
12247 targets powerpc*-*-* and rs6000*-*-*.
12249 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
12252 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
12253 32-bit, since indexed is not valid for DImode.
12254 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
12255 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
12256 (define_peephole2 for Altivec d-form load): Add 32-bit support.
12257 (define_peephole2 for Altivec d-form store): Likewise.
12260 * multiple_target.c (create_dispatcher_calls): Only create the
12261 dispatcher call if the function is the default clone of a
12262 versioned function.
12264 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
12266 PR middle-end/80902
12267 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
12268 a call, force the call to not be a tail call.
12270 2017-06-23 Jeff Law <law@redhat.com>
12272 * doc/contrib.texi: Add entry for Steven Pemberton's work on
12275 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
12277 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
12278 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
12279 handling for early expansion of vector shifts (sl,sr,sra,rl).
12280 (builtin_function_type): Add vector shift right instructions
12281 to the unsigned argument list.
12283 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
12285 rtl-optimizatoin/79286
12286 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
12287 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
12288 trap. PIC register plus a const unspec without offset can never trap.
12290 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
12292 * tree.h (builtin_structptr_type): New type.
12293 (builtin_structptr_types): Declare new array.
12294 * tree.c (builtin_structptr_types): New array.
12295 (free_lang_data, build_common_tree_nodes): Use it.
12297 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
12300 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
12303 2017-06-22 Matt Turner <mattst88@gmail.com>
12305 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
12306 Lake models to skylake case. Assume skylake for unknown
12307 models with clflushopt.
12309 2017-06-22 Jeff Law <law@redhat.com>
12311 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
12312 frame sizes that do not satisfy aarch64_uimm12_shift.
12314 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
12316 * profile-count.h (apply_probability,
12317 apply_scale, probability_in): Fix checks for zero.
12319 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12321 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
12322 * doc/cppdiropts.texi (-I @var{dir}): Document it.
12324 2016-06-22 Richard Biener <rguenther@suse.de>
12326 * tree-vect-loop.c (vect_model_reduction_cost): Handle
12327 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
12328 REDUC_MAX_EXPR support.
12329 (vectorizable_reduction): Likewise.
12330 (vect_create_epilog_for_reduction): Likewise.
12332 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
12334 * match.pd (A / (1 << B) -> A >> B): New.
12335 * generic-match-head.c: Include optabs-tree.h.
12336 * gimple-match-head.c: Likewise.
12337 * optabs-tree.h (target_supports_op_p): New.
12338 * optabs-tree.c (target_supports_op_p): New.
12340 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12342 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
12343 $gcc_cv_ld --help output.
12344 (gcc_cv_ld_demangle): Likewise.
12345 (gcc_cv_ld_eh_frame_hdr): Likewise.
12346 (gcc_cv_ld_pie): Likewise.
12347 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
12348 (gcc_cv_ld_buildid): Likewise.
12349 (gcc_cv_ld_sysroot): Likewise.
12350 (ld_bndplt_support): Likewise.
12351 (ld_pushpopstate_support): Likewise.
12352 * configure: Regenerate.
12353 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
12355 2017-06-21 Jakub Jelinek <jakub@redhat.com>
12358 * config/i386/sse.md (round<mode>2): Renumber match_dup and
12359 operands indexes to avoid gap between operands and match_dups.
12361 2017-06-21 Andrew Pinski <apinski@cavium.com>
12363 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
12364 Increment Arith_shift and Arith_shift_reg by 1.
12365 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
12367 * config/aarch64/aarch64.c (thunderx_tunings): Enable
12368 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
12369 (aarch64_strip_extend): Add new argument and test for it.
12370 (aarch64_cheap_mult_shift_p): New function.
12371 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
12372 add a cost if it is true.
12373 Update calls to aarch64_strip_extend.
12374 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
12376 2017-06-21 Andrew Pinski <apinski@cavium.com>
12378 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
12380 (thunderxt88): Likewise.
12381 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
12382 (thunderx_prefetch_tune): New variable.
12383 (thunderx2t99_prefetch_tune): Update for the correct values.
12384 (thunderxt88_tunings): New variable.
12385 (thunderx_tunings): Use thunderx_prefetch_tune instead of
12386 generic_prefetch_tune.
12387 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
12389 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12391 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
12392 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
12393 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
12394 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
12395 (aarch64_atomic_cas<mode>, GPI): Likewise.
12397 2017-06-21 Martin Liska <mliska@suse.cz>
12399 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
12400 statements on cold and hot labels.
12401 * predict.c (tree_estimate_probability_bb): Remove the
12402 prediction from this place.
12404 2017-06-21 Martin Liska <mliska@suse.cz>
12406 PR tree-optimization/79489
12407 * gimplify.c (maybe_add_early_return_predict_stmt): New
12409 (gimplify_return_expr): Call the function.
12410 * predict.c (tree_estimate_probability_bb): Remove handling
12412 * predict.def: Update comment about early return predictor.
12413 * gimple-predict.h (is_gimple_predict): New function.
12414 * predict.def: Change default value of early return to 66.
12415 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
12417 * passes.def: Put pass_strip_predict_hints to the beginning of
12420 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
12422 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
12423 FUNCTION_DECL declarations.
12424 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
12426 (dwaf2out_decl): Likewise.
12427 * godump.c (go_early_global_decl): Skip call to the real debug hook
12428 for FUNCTION_DECL declarations.
12429 * passes.c (rest_of_decl_compilation): Skip call to the
12430 early_global_decl debug hook for FUNCTION_DECL declarations, unless
12431 -fdump-go-spec is passed.
12433 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
12435 * config/i386/i386.c (struct builtin_isa): New field pure_p.
12436 Reorder for compactness.
12437 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
12438 (def_builtin_pure, def_builtin_pure2): New functions.
12439 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
12441 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
12443 * match.pd (nop_convert): New predicate.
12444 ((A +- CST1) +- CST2): Allow some NOP conversions.
12446 2017-06-21 Jakub Jelinek <jakub@redhat.com>
12449 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
12450 with ctors/dtors if GOVD_SHARED is set.
12452 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
12454 * config/aarch64/aarch64.md (movti_aarch64):
12455 Emit mov rather than orr.
12456 (movtf_aarch64): Likewise.
12457 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
12458 Emit mov rather than orr.
12460 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
12462 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
12463 Swap alternatives, make integer dup more expensive.
12465 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
12467 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
12468 Return true for non-tls symbols.
12470 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
12472 * config/aarch64/aarch64-cores.def (cortex-a55): New.
12473 (cortex-a75): Likewise.
12474 (cortex-a75.cortex-a55): Likewise.
12475 * config/aarch64/aarch64-tune.md: Regenerate.
12476 * doc/invoke.texi (-mtune): Document new values for -mtune.
12478 2017-06-21 Tom de Vries <tom@codesourcery.com>
12480 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
12481 stack_size feature.
12482 (Effective-Target Keywords, Other attributes): Suggest using
12483 dg-add-options stack_size feature to get stack limit in stack_size
12484 effective target documentation.
12486 2017-06-21 Julian Brown <julian@codesourcery.com>
12487 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
12489 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
12490 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
12491 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
12493 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
12494 attribute type list for neon_multiply.
12495 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
12496 attribute type list for neon_multiply.
12497 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
12498 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
12499 attribute type list for neon_multiply.
12500 * config/arm/types.md (crypto_pmull): Add.
12501 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
12502 attribute type list.
12504 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
12506 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
12509 2017-06-20 Jakub Jelinek <jakub@redhat.com>
12511 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
12512 to make sure not to dereference a NULL cost_classes_ptr pointer.
12514 2017-06-20 Carl Love <cel@us.ibm.com>
12516 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12517 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
12518 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
12519 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
12520 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
12521 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
12522 VMULOSW): New enum "unspec" values.
12523 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
12524 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
12525 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
12526 altivec_vmulosw): New patterns.
12527 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
12528 VMULOSW): Add definitions.
12530 2017-06-20 Julia Koval <julia.koval@intel.com>
12532 * config/i386/i386.c: Fix rounding expand for new pattern.
12533 * config/i386/subst.md: Fix pattern (parallel -> unspec).
12535 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
12537 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
12538 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
12540 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
12542 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
12545 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
12547 * config/aarch64/aarch64-cores.def: Rearrange to sort by
12548 architecture, then by implementer ID.
12549 * config/aarch64/aarch64-tune.md: Regenerate.
12551 2017-06-20 Richard Biener <rguenther@suse.de>
12553 PR middle-end/81097
12554 * fold-const.c (split_tree): Fold to type before negating.
12556 2017-06-20 David Malcolm <dmalcolm@redhat.com>
12558 * diagnostic-show-locus.c
12559 (selftest::test_fixit_deletion_affecting_newline): New function.
12560 (selftest::diagnostic_show_locus_c_tests): Call it.
12562 2017-06-20 Andreas Schwab <schwab@suse.de>
12565 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
12568 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
12570 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
12571 __ARM_FEATURE_COPROC according to support.
12573 2017-06-20 Jakub Jelinek <jakub@redhat.com>
12575 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
12576 Rewritten to avoid overflow for > 32-bit pointers.
12579 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
12580 by removing enum keyword.
12581 (ubsan_type_descriptor): Likewise. Formatting fix.
12584 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
12585 splitter): Require TARGET_SSE2 in the condition.
12587 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
12590 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
12591 for doing vector set of SFmode on ISA 3.0.
12592 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
12593 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
12595 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
12596 SFmode value into a V4SF variable that was extracted from another
12597 V4SF variable without converting the element to double precision
12598 and back to single precision vector format.
12599 (vsx_insert_extract_v4sf_p9_2): Likewise.
12601 2017-06-19 Jakub Jelinek <jakub@redhat.com>
12603 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
12604 in UWHI to avoid undefined overflow.
12607 * ubsan.h (enum ubsan_encode_value_phase): New.
12608 (ubsan_encode_value): Change second argument to
12609 enum ubsan_encode_value_phase with default value of
12610 UBSAN_ENCODE_VALUE_GENERIC.
12611 * ubsan.c (ubsan_encode_value): Change second argument to
12612 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
12613 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
12614 create_tmp_var_raw instead of create_tmp_var and use a
12616 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
12617 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
12618 ubsan_encode_value callers.
12621 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
12622 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
12623 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
12625 2017-06-19 Richard Biener <rguenther@suse.de>
12627 PR middle-end/81118
12628 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
12629 estimates if we changed anything.
12631 2017-06-19 Richard Biener <rguenther@suse.de>
12633 PR tree-optimization/80887
12634 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
12635 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
12636 simplified lookups, then reset mprts_hook.
12637 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
12639 (try_to_simplify): Likewise.
12641 2017-06-19 Martin Liska <mliska@suse.cz>
12644 * gimplify.c (gimplify_switch_expr):
12645 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
12647 2017-06-19 Martin Liska <mliska@suse.cz>
12649 * doc/install.texi: Document that PGO runs in 4 stages.
12651 2017-06-19 Martin Liska <mliska@suse.cz>
12654 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
12655 to dispatcher function name.
12656 * multiple_target.c (replace_function_decl): New function.
12657 (create_dispatcher_calls): Redirect both edges and references.
12659 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
12661 * profile-count.c (profile_count::dump): Dump quality.
12662 (profile_count::differs_from_p): Update for unsigned val.
12663 * profile-count.h (profile_count_quality): New enum.
12664 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
12666 2017-06-19 Richard Biener <rguenther@suse.de>
12668 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
12669 struct function as arg.
12670 (estimate_numbers_of_iterations): Export overload with loop arg.
12671 (free_numbers_of_iterations_estimates_loop): Use an overload of
12672 free_numbers_of_iterations_estimates instead.
12673 * tree-cfg.c (remove_bb): Adjust.
12674 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
12675 * tree-parloops.c (gen_parallel_loop): Likewise.
12676 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
12678 (tree_unroll_loops_completely): Likewise.
12679 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
12680 Use an overload instead and export.
12681 (estimated_loop_iterations): Adjust.
12682 (max_loop_iterations): Likewise.
12683 (likely_max_loop_iterations): Likewise.
12684 (estimate_numbers_of_iterations): Take struct function as arg
12686 (loop_exits_before_overflow): Adjust.
12687 (free_numbers_of_iterations_estimates_loop): Use an overload.
12688 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
12689 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
12691 2017-06-19 Richard Biener <rguenther@suse.de>
12694 * ipa-prop.c (find_constructor_constant_at_offset): Handle
12695 RANGE_EXPR conservatively.
12697 2017-06-16 Carl Love <cel@us.ibm.com>
12699 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12700 definitions for vec_float, vec_float2, vec_floato,
12701 vec_floate built-ins.
12702 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
12703 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
12705 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
12706 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
12707 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
12708 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
12709 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
12710 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
12711 vec_floato): Add builtin defines.
12712 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
12713 Update the built-in documentation file for the new built-in
12716 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12718 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
12719 (mthumb): Mark as the negative of -marm.
12721 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12723 * doc/invoke.texi (ARM Options, -mcpu): Document supported
12725 (ARM Options, -mtune): Document that this accepts the same
12726 extension options as -mcpu.
12727 (ARM Options, -mfpu): Document addition of -mfpu=auto.
12729 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12731 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
12732 permitted extensions.
12734 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12736 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
12737 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
12738 (armv8-m.main): Add option +nodsp.
12739 * config/arm/arm-cpu-cdata.h: Regenerated.
12741 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12743 * config/arm/t-fuchsia: New file.
12744 * config.gcc (arm*-*-fuchsia*): Use it.
12746 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12748 * config/arm/t-symbian: Rewrite for new option infrastructure.
12750 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12752 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
12753 (MULTILIB_REQUIRED): Likewise.
12755 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12757 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
12758 (MULTILIB_RESUE): Likewise.
12759 (MULTILIB_MATCHES): Likewise.
12760 (MULTLIB_REQUIRED): Likewise.
12762 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12764 * config/arm/t-rtems: Rewrite for new option framework.
12766 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12768 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
12769 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
12770 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
12771 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
12772 * config/arm/t-multilib: ... here.
12773 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
12774 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
12775 armv7-a and armv8*-a when A-profile libraries have not been built.
12776 * config/arm/t-rmprofile: Rewrite.
12778 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12780 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
12781 with a backslash. Remove the backslash after substituting unescaped
12783 * doc/fragments.texi (MULTILIB_REUSE): Document it.
12785 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12787 * config.gcc: (arm*-*-*): When building a-profile libraries, force
12788 the driver to pass through the default setting of -mfloat-abi.
12789 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
12791 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
12792 (all_feat_combs): New rule.
12793 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
12795 * config/arm/t-aprofile: Rewrite.
12797 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12799 * config/arm/arm.h (FPUTYPE_AUTO): Define.
12800 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
12801 fpu is not specified by the user/command-line.
12802 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
12803 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
12804 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
12805 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
12806 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
12807 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
12809 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12811 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
12812 * config/arm/t-arm-elf: Rewritten.
12814 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12816 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
12817 have some floating-point instructions.
12818 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
12819 (TARGET_MAYBE_HARD_FLOAT): New macro.
12820 * config/arm/arm-builtins.c (arm_init_builtins): Use
12821 TARGET_MAYBE_HARD_FLOAT.
12822 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
12824 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12826 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
12827 (configargs.h): Include it.
12828 (arm_print_hint_for_fpu_option): New function.
12829 (arm_parse_fpu_option): New function.
12830 (candidate_extension): New class.
12831 (arm_canon_for_multilib): New function.
12832 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
12833 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
12834 (ARCH_CANONICAL_SPECS): New macro.
12835 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
12837 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12839 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
12840 are set after handling multilib fragments. Set target_cpu_default2
12843 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12845 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
12847 (arm*-*-*): Set target_cpu_default2 to a quoted string.
12848 * config/arm/parsecpu.awk (check_cpu): Validate any extension
12850 (check_arch): Likewise.
12851 * config/arm/arm.c (arm_configure_build_target): Handle
12852 TARGET_CPU_DEFAULT being a string constant. Scan any feature
12853 options in the default.
12855 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12857 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
12858 when an option is an alias of another.
12859 * config/arm/parsecpu.awk (optalias): New parser token.
12860 (gen_comm_data): Mark non-alias options as such. Emit entries
12861 for extension aliases.
12862 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
12863 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
12864 (armv6kz, armv6zk, armv6t2): Likewise.
12865 (armv7): Make vfpv3-d16 an alias.
12866 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
12868 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
12869 Sort in canonical order.
12870 (armv8-a): Sort in canonical order.
12871 (armv8.1-a, armv8.2-a): Likewise.
12872 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
12874 (cortex-a9): Sort in canonical order.
12875 * config/arm/arm.c (selftests.h): Include it.
12876 (arm_test_cpu_arch_data): New function.
12877 (arm_run_self_tests): New function.
12878 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
12879 (targetm): Move declaration to the end of the file.
12880 * arm-cpu-cdata.h: Regenerated.
12882 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12884 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
12885 call to target_mode_check describing the type of option passed.
12886 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
12887 (arm_target_thumb_only): Use arm_parse_arch_option_name or
12888 arm_parse_cpu_option_name to match parameters against list of
12890 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
12891 arm_arch_core_flags data structure.
12892 * config/arm/arm-cpu_cdata.h: Regenerated.
12894 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12896 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
12898 (arm_print_hint_for_cpu_option): Likewise.
12899 (arm_print_hint_for_arch_option): Likewise.
12900 (arm_parse_cpu_option_name): Likewise.
12901 (arm_parse_arch_option_name): Likewise.
12902 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
12903 of entries in the all_fpus list.
12904 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
12905 (arm_parse_cpu_option_name): Declare.
12906 (arm_parse_arch_option_name): Declare.
12907 (arm_parse_option_features): Declare.
12908 (arm_intialize_isa): Declare.
12909 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
12911 (gen_comm_data): ... here. Make definitions non-static.
12912 * config/arm/arm-cpu-data.h: Regenerated.
12913 * config/arm/arm-cpu-cdata.h: Regenerated.
12915 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12917 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
12918 (cpu_arch_extension): New structure.
12919 (cpu_arch_option, arch_option, cpu_option): New structures.
12920 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
12921 architecture types.
12922 (gen_data): Generate new format data tables.
12923 * config/arm/arm.c (cpu_tune): New structure.
12924 (cpu_option, processors): Delete.
12925 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
12926 (arm_print_hint_for_cpu_option): ... this and ...
12927 (arm_print_hint_for_arch_option): ... this.
12928 (arm_parse_arch_cpu_name): Delete. Replace with ...
12929 (arm_parse_cpu_option_name): ... this and ...
12930 (arm_parse_arch_option_name): ... this.
12931 (arm_unrecognized_feature): Change type of target parameter to
12933 (arm_parse_arch_cpu_features): Delete. Replace with ...
12934 (arm_parse_option_features): ... this.
12935 (arm_configure_build_target): Rework to use new configuration data
12937 (arm_print_tune_info): Rework for new configuration data tables.
12938 * config/arm/arm-cpu-data.h: Regenerated.
12939 * config/arm/arm-cpu.h: Regenerated.
12941 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12943 * Makefile.in (OBJS): Move sbitmap.o from here ...
12944 (OBJS-libcommon): ... to here.
12946 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12948 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
12949 (ISA_ALL_CRYPTO): New macro.
12950 (ISA_ALL_SIMD): New macro
12951 (ISA_ALL_FP): New macro.
12952 * config/arm/arm.c (fpu_bitlist): Update initializer.
12953 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
12955 (arm9e): Add fpu. Add option for nofp
12956 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
12957 (arm926ej-s, arm1026ej-s): Likewise.
12958 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
12959 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
12960 neon-fp16, neon-vfpv4, nofp and nosimd.
12961 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
12962 (cortex-a8): Add fpu. Add option for nofp.
12963 (cortex-a9): Add fpu. Add options for nosimd and nofp.
12964 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
12965 (cortex-r4f): Add fpu.
12966 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
12967 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
12969 (cortex-r8): Likewise.
12970 (cortex-m4): Add fpu. Add option for nofp.
12971 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
12972 (cortex-a17.cortex-a7): Likewise.
12973 (cortex-a32): Add fpu. Add options for crypto and nofp.
12974 (cortex-a35, cortex-a53): Likewise.
12975 (cortex-a57): Add fpu. Add option for crypto.
12976 (cortex-a72, cortex-a73): Likewise.
12977 (exynos-m1): Likewise.
12978 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
12979 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
12980 (cortex-m33): Add fpu. Add option for nofp.
12981 * config/arm/arm-cpu-cdata.h: Regenerated
12982 * config/arm/arm-cpu-data.h: Regenerated.
12984 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12986 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
12987 (armv5te, armv5tej): Likewise.
12988 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
12989 (armv7): Add options fp and vfpv3-d16.
12990 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
12991 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
12993 (armv7ve): Likewise.
12994 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
12995 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
12996 (armv8-a): Add nocrypto option.
12997 (armv8.1-a, armv8.2-a): Likewise.
12998 (armv8-m.main): add options fp, fp.dp and nofp.
13000 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13002 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
13004 (armv8-a+crc): Delete.
13005 (armv8.1-a): Add options simd, crypto and nofp.
13006 (armv8.2-a): Add options fp16, simd, crypto and nofp.
13007 (armv8.2-a+fp16): Delete.
13008 (armv8-m.main): Add option dsp.
13009 (armv8-m.main+dsp): Delete.
13010 (cortex-a8): Add fpu. Add nofp option.
13011 (cortex-a9): Add fpu. Add nofp and nosimd options.
13012 * config/arm/parsecpu.awk (gen_data): Generate option tables and
13013 link to main cpu and architecture data structures.
13014 (gen_comm_data): Only put isa attributes from the main architecture
13016 (option): New statement for architecture and CPU entries.
13017 * arm.c (struct cpu_option): New structure.
13018 (struct processors): Add entry for options.
13019 (arm_unrecognized_feature): New function.
13020 (arm_parse_arch_cpu_name): Ignore any characters after the first
13022 (arm_parse_arch_cpu_feature): New function.
13023 (arm_configure_build_target): Separate out any CPU and architecture
13024 features and parse separately. Don't error out if -mfpu=auto is
13025 used with only an architecture string.
13026 (arm_print_asm_arch_directives): New function.
13027 (arm_file_start): Call it.
13028 * config/arm/arm-cpu-cdata.h: Regenerated.
13029 * config/arm/arm-cpu-data.h: Likewise.
13030 * config/arm/arm-tables.opt: Likewise.
13032 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13034 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
13035 assembler when it is not -mfpu=auto.
13037 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13039 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
13040 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
13041 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
13042 (ASM_CPU_SPEC): Rewrite.
13043 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
13044 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
13045 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
13046 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
13047 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
13048 copied string is NUL-terminated. Also strip any characters prefixed
13050 (arm_rewrite_selected_arch): New function.
13051 (arm_rewrite_march): New function.
13053 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
13055 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
13056 (x_arm_cpu_string, x_arm_tune_string): Likewise.
13057 (march, mcpu, mtune): Convert to string-based options.
13058 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
13059 (arm_parse_arch_cpu_name): New function.
13060 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
13061 identify selected architecture or CPU.
13062 (arm_option_save): New function.
13063 (TARGET_OPTION_SAVE): Redefine.
13064 (arm_option_restore): Restore string options.
13065 (arm_option_print): Print string options.
13067 2017-06-16 Martin Sebor <msebor@redhat.com>
13069 PR tree-optimization/80933
13070 PR tree-optimization/80934
13071 * builtins.c (fold_builtin_3): Do not handle bcmp here.
13072 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
13073 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
13074 (gimple_fold_builtin): Call them.
13076 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13078 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
13079 as unlikely; update profile.
13081 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13083 * predict.c (force_edge_cold): Handle declaring edges impossible
13086 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13088 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
13090 (try_unroll_loop_completely): Fix reporting.
13092 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13094 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
13096 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
13099 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
13100 if given a non-constant argument for an intrinsic which requires a
13103 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13105 * profile.c (compare_freqs): New function.
13106 (branch_prob): Sort edge list.
13107 (find_spanning_tree): Assume that the list is priority sorted.
13109 2017-06-16 Richard Biener <rguenther@suse.de>
13111 PR tree-optimization/81090
13112 * passes.def (pass_record_bounds): Remove.
13113 * tree-pass.h (make_pass_record_bounds): Likewise.
13114 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
13115 make_pass_record_bounds): Likewise.
13116 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
13117 not free niter estimates at the beginning but at the end.
13118 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
13120 2017-06-16 Richard Biener <rguenther@suse.de>
13122 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
13123 initializer to workaround ICE in host GCC 4.8.
13125 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
13127 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
13129 (clone_inlined_nodes): Update.
13131 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
13133 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
13134 prefetch settings, and enable prefetching by default at -O3.
13136 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
13138 * config/aarch64/aarch64.c (aarch64_override_options_internal):
13139 Set flag_prefetch_loop_arrays according to tuning data.
13141 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
13143 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
13144 New tune structure.
13145 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
13146 [Unrelated to main purpose of the patch] Place the pointer field last
13147 to enable type checking errors when tune structure are wrongly merged.
13148 * config/aarch64/aarch64.c (generic_prefetch_tune,)
13149 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
13150 (thunderx2t99_prefetch_tune): New tune constants.
13151 (tune_params *_tunings): Update all tunings (no functional change).
13152 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
13153 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
13154 from tunings structures.
13156 2017-06-16 Jakub Jelinek <jakub@redhat.com>
13159 * ubsan.c (instrument_null): Add T argument, use it instead
13160 of computing it based on IS_LHS.
13161 (instrument_object_size): Likewise.
13162 (pass_ubsan::execute): Adjust instrument_null and
13163 instrument_object_size callers to pass gimple_get_lhs or
13164 gimple_assign_rhs1 result to it. Use instrument_null instead of
13165 calling get_base_address and instrument_mem_ref. Handle
13166 aggregate call arguments for object-size sanitization.
13168 2017-06-16 Yury Gribov <tetra2005@gmail.com>
13170 PR tree-optimization/81089
13171 * tree-vrp.c (is_masked_range_test): Validate operands of
13174 2017-06-15 Martin Sebor <msebor@redhat.com>
13177 * dumpfile.c (dump_register): Avoid calling memset to initialize
13178 a class with a default ctor.
13179 * gcc.c (struct compiler): Remove const qualification.
13180 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
13181 * hash-table.h: Ditto.
13182 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
13184 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
13185 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
13187 * params.h (struct param_info): Make struct members non-const.
13188 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
13189 with default initialization.
13190 * vec.h (vec_copy_construct, vec_default_construct): New helper
13192 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
13193 with vec_copy_construct.
13194 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
13195 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
13196 * doc/invoke.texi (-Wclass-memaccess): Document.
13198 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
13200 * emit-rtl.h (is_leaf): Update comment about local
13201 register allocator.
13203 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13206 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
13207 for a variable to have a section before checking if the section has a
13209 Set section to.persistent if persistent attribute is set.
13210 Warn if .persistent attribute is used on an automatic variable.
13212 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
13214 PR rtl-optimization/80474
13215 * reorg.c (update_block): Do not ignore instructions in a delay slot.
13217 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
13219 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
13222 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
13224 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
13225 (casesi): Emit bounds checking as RTL.
13226 (casesi_internal_mips16_<mode>): Remove bounds checking.
13228 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
13230 * config/xtensa/xtensa.c (xtensa_option_override): Append
13231 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
13232 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
13233 xtensa_doloop_hooks): Define unconditionally.
13234 (xtensa_reorg_loops): Only call reorg_loops in the presence of
13236 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
13237 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
13238 for it in xtensa_option_override.
13239 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
13240 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
13242 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
13244 * doc/cppopts.texi: Document '-' special value to -MF.
13246 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
13248 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
13249 (cortex_a53_fconst): Likewise.
13250 (cortex_a53_fpmul): Likewise.
13251 (cortex_a53_f_load_64): Likewise.
13252 (cortex_a53_f_load_many): Likewise.
13253 (cortex_a53_advsimd_alu): Likewise.
13254 (cortex_a53_advsimd_alu_q): Likewise.
13255 (cortex_a53_advsimd_mul): Likewise.
13256 (cortex_a53_advsimd_mul_q): Likewise.
13257 (fpmac bypass): Add new bypass for fpmac-fpmac case.
13258 Add missing fmul, r2f_cvt and fconst cases.
13260 2017-06-14 Richard Biener <rguenther@suse.de>
13262 PR middle-end/81088
13263 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
13265 (fold_binary_loc): When associating do not treat pre-existing
13266 TREE_OVERFLOW on literal constants as a reason to allow
13267 TREE_OVERFLOW on associated literal constants.
13269 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
13271 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
13272 (MASK_FEATURES): New macro.
13273 * config/sparc/sparc.c (sparc_option_override): Remove the special
13274 handling of -mfpu and generalize it to all MASK_FEATURES switches.
13276 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
13278 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
13279 a division of 0 if non-call exceptions are enabled.
13281 2017-06-14 Andrew Pinski <apinski@cavium.com>
13282 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
13285 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
13286 Improve vector initialization code gen for only variable case.
13288 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
13290 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
13292 2017-06-14 Richard Biener <rguenther@suse.de>
13294 PR tree-optimization/81083
13295 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
13298 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13300 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
13301 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
13302 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
13303 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
13304 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
13305 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
13307 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13309 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
13310 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
13312 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13314 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
13316 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13318 * config/rs6000/t-rtems: Don't handle SPE.
13320 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13322 * config/rs6000/t-linux: Don't handle SPE.
13324 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13326 * config/rs6000/eabispe.h: Delete file.
13328 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13330 * config/rs6000/t-spe: Delete file.
13332 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13334 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
13335 (rs6000_legitimate_offset_address_p): Return false for anything in
13336 V2SImode or V2SFmode.
13338 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13340 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
13341 except V2SF and V2SI. Rearrange the vector modes, and add comments.
13342 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
13344 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
13345 (rs6000_legitimate_offset_address_p): Ditto.
13346 (rs6000_emit_move): Ditto.
13347 (rs6000_init_builtins): Remove V4HI_type_node.
13349 2017-06-13 Martin Liska <mliska@suse.cz>
13352 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
13353 (gate_asan): Likewise.
13354 * asan.h (asan_no_sanitize_address_p): Remove the function.
13355 (sanitize_flags_p): New function.
13356 * builtins.def: Fix coding style.
13357 * common.opt: Use renamed enum value.
13358 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
13359 * doc/extend.texi: Document no_sanitize attribute.
13360 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
13361 to SANITIZE_UNDEFINED_NONDEFAULT.
13362 * gcc.c (sanitize_spec_function): Use the renamed enum value.
13363 * gimple-fold.c (optimize_atomic_compare_exchange_p):
13364 Use sanitize_flags_p.
13365 * gimplify.c (gimplify_function_tree): Likewise.
13366 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
13367 * opts.c (parse_no_sanitize_attribute): New function.
13368 (common_handle_option): Use renamed enum value.
13369 * opts.h (parse_no_sanitize_attribute): Declare.
13370 * tree.c (sanitize_flags_p): New function.
13371 * tree.h: Declared here.
13372 * tsan.c: Use sanitize_flags_p.
13373 * ubsan.c (ubsan_expand_null_ifn): Likewise.
13374 (instrument_mem_ref): Likewise.
13375 (instrument_bool_enum_load): Likewise.
13376 (do_ubsan_in_current_function): Remove the function.
13377 (pass_ubsan::execute): Use sanitize_flags_p.
13378 * ubsan.h: Remove do_ubsan_in_current_function
13379 * tree-cfg.c (print_no_sanitize_attr_value): New function.
13380 (dump_function_to_file): Use it here.
13382 2017-06-13 Martin Jambor <mjambor@suse.cz>
13384 PR tree-optimization/80803
13385 PR tree-optimization/81063
13386 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
13387 (propagate_subaccesses_across_link): Enqueue subtree whenever
13388 necessary instead of relying on the caller.
13390 2017-06-13 Martin Jambor <mjambor@suse.cz>
13392 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
13393 that have a first_link.
13394 (sort_and_splice_var_accesses): Do not check first_link before
13396 (subtree_mark_written_and_enqueue): Likewise.
13397 (propagate_all_subaccesses): Likewise and do not stop at first
13398 parent with a first_link.
13400 2017-06-13 Martin Jambor <mjambor@suse.cz>
13402 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
13405 2017-06-13 Yury Gribov <tetra2005@gmail.com>
13407 * match.pd: New pattern.
13409 2017-06-13 Yury Gribov <tetra2005@gmail.com>
13411 * tree-vrp.c (is_masked_range_test): New function.
13412 (register_edge_assert_for): Determine ranges for
13415 2017-06-13 Yury Gribov <tetra2005@gmail.com>
13417 PR tree-optimization/67328
13418 * fold-const.c (maskable_range_p): New function.
13419 (build_range_check): Generate bittests if possible.
13421 2017-06-13 Martin Liska <mliska@suse.cz>
13423 * gimple-pretty-print.c (dump_probability): Add new argument.
13424 (dump_edge_probability): Dump both probability and count.
13425 (dump_gimple_label): Likewise.
13426 (dump_gimple_bb_header): Likewise.
13428 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
13431 * config/avr/avr-devices.c: Fix indentation.
13432 * config/avr/gen-avr-mmcu-specs.c: Dito.
13434 2017-06-13 Richard Biener <rguenther@suse.de>
13436 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
13437 instead get vector type from stmt_info.
13438 (vectorizable_reduction): Adjust. Remove dead code.
13440 2017-06-13 Richard Biener <rguenther@suse.de>
13442 PR middle-end/81065
13443 * fold-const.c (extract_muldiv_1): Remove bogus distribution
13444 case of C * (x * C2 + C3).
13445 (fold_addr_of_array_ref_difference): Properly fold index difference.
13447 2017-06-12 David S. Miller <davem@davemloft.net>
13450 * config/sparc/sparc.md (return expander): Emit frame blockage if
13451 function uses alloca.
13453 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
13455 * combine.c (make_field_assignment): Check len rather than the mode
13456 precision when calling force_to_mode.
13458 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
13460 Support multilibs and devices that see flash in RAM address range.
13463 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
13464 (avr_mcu_t) <flash_pm_offset>: New field.
13465 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
13466 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
13467 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
13468 (AVR_TINY_PM_OFFSET): Remove macro.
13469 * config/avr/avr.opt (-mshort-calls): New option.
13470 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
13471 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
13472 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
13473 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
13474 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
13475 instead of avr_arch->have_jmp_call.
13476 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
13477 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
13478 avr_arch->flash_pm_offset to define.
13479 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
13480 new field flash_pm_offset. Add entry for avrxmega3.
13481 (avr_texinfo): Add entry for avrxmega3.
13482 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
13483 attiny212, attiny214,
13484 attiny412, attiny414, attiny416, attiny417,
13485 attiny814, attiny816, attiny817,
13486 attiny1614, attiny1616, attiny1617,
13487 attiny3214, attiny3216, attiny3217.
13488 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
13489 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
13490 (avr_print_operand_address) [AVR_TINY]: Same.
13491 (avr_asm_init_sections) <readonly_data_section>: Only patch
13492 callback if avr_arch->flash_pm_offset = 0.
13493 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
13494 for rodata if avr_arch->flash_pm_offset != 0.
13495 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
13496 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
13497 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
13498 (m_options): Append opt_rcall.
13499 (m_dirnames): Append dir_rcall.
13500 * config/avr/t-multilib: Regenerate.
13502 * configure.ac [target=avr]: Check whether avrxmega3 default
13503 linker description file works as needed.
13504 * configure: Regenerate.
13505 * doc/avr-mmcu.texi: Regenerate.
13506 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
13507 <__AVR_ARCH__>: Document avrxmega3 and 103.
13508 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
13509 <__AVR_SHORT_CALLS__>: Document it.
13510 <__AVR_PM_BASE_ADDRESS__>: Document it.
13511 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
13512 (AVR Variable Attributes) <progmem>: Document this is
13513 not needed for avrxmega3.
13514 (AVR Named Address Spaces) <__flash>: Dito.
13516 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
13518 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
13520 2017-06-12 Doug Rupp <rupp@adacore.com>
13522 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
13523 Append vxworks-stdint.h to the tm_file list.
13524 * config/vxworks-stdint.h: New file.
13526 2017-06-12 Martin Liska <mliska@suse.cz>
13528 PR tree-optimization/81041
13529 * tree-profile.c (gimple_gen_ic_func_profiler):
13530 Create an extra BB in profile-generate
13531 (gimple_gen_time_profiler): Likewise.
13533 2017-06-12 Jakub Jelinek <jakub@redhat.com>
13535 PR tree-optimization/81003
13536 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
13537 (update_range_test): Use it instead of force_gimple_operand_gsi.
13539 2017-06-12 Richard Biener <rguenther@suse.de>
13541 PR tree-optimization/81053
13542 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
13543 with backedge value not defined in loop. Simplify def stmt
13546 2017-06-11 Tom de Vries <tom@codesourcery.com>
13549 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
13551 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
13552 nvptx_cannot_force_const_mem.
13554 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
13556 * opts.c (finish_options): Move test for flag_split_stack after
13557 it has been initialized.
13559 2017-06-11 Jason Merrill <jason@redhat.com>
13561 * tree.h (id_equal): New.
13562 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
13563 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
13564 instead of strcmp of IDENTIFIER_POINTER.
13566 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
13568 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
13569 (mark_all_inlined_calls_cdtor): Fix formating.
13570 (inline_transform): Rescale profile before inlining.
13572 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
13574 * cgraph.h (cgraph_edge::clone): Update prototype.
13575 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
13576 (cgraph_node::create_clone): Update.
13577 (cgraph_node::create_version_clone): Update.
13578 * tree-inline.c (copy_bb): Update.
13579 (expand_call_inline): Update.
13581 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
13583 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
13584 factored out from ...
13585 (rs6000_emit_prologue): ... here.
13587 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
13589 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
13590 factored out from ...
13591 (rs6000_emit_prologue): ... here.
13593 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
13595 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
13597 (handle_missing_profiles): Fix computation of tp_first_run.
13598 (counts_to_freqs): Do not touch freqs when count is 0.
13600 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
13602 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
13605 2017-06-10 Tom de Vries <tom@codesourcery.com>
13607 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
13608 attributes): Document signal effective target.
13610 2017-06-10 Tom de Vries <tom@codesourcery.com>
13612 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
13613 Document effective target stack_size.
13615 2017-06-09 David Malcolm <dmalcolm@redhat.com>
13617 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
13618 to the edit_context if they can be auto-applied.
13620 2017-06-9 Ian Lance Taylor <iant@golang.org>
13622 * opts.c (finish_options): If -fsplit-stack, disable implicit
13623 -forder-blocks-and-partition.
13624 * doc/invoke.texi (Optimize Options): Document that when using
13625 -fsplit-stack -forder-blocks-and-partition is not implicitly
13628 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
13630 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
13631 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
13632 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
13633 * builtins.def (abort, trap, unreachable): Declare cold.
13634 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
13635 * tree-core.h (ECF_COLD): New.
13636 * tree.c (set_call_expr_flags): Handle ECF_COLD.
13637 (build_common_builtin_nodes): Mark unreachable and abort as cold.
13639 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
13641 * predict.c (unlikely_executed_stmt_p): Cleanup.
13643 2017-06-09 Richard Biener <rguenther@suse.de>
13645 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
13646 model if the ref is always written to.
13648 2017-06-09 Tamar Christina <tamar.christina@arm.com>
13650 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
13652 2017-06-09 Tamar Christina <tamar.christina@arm.com>
13654 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
13657 2017-06-09 Tom de Vries <tom@codesourcery.com>
13660 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
13661 "target cannot support label values" when encountering LABEL_REF.
13663 2017-06-09 Martin Liska <mliska@suse.cz>
13665 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
13666 (gimple_gen_ic_func_profiler): Emit direct comparison
13667 of __gcov_indirect_call_callee with NULL.
13668 (gimple_gen_time_profiler): Change probability from
13669 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
13671 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
13673 * profile.c (edge_gcov_counts): Turn to pointer.
13674 (compute_branch_probabilities, compute_branch_probabilities): Update.
13675 (branch_prob): Do not clear edge_gcov_count.
13676 * profile.h (edge_gcov_counts): Turn to pointer.
13677 (edge_gcov_count): Update.
13679 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
13681 * gimple.h (gimple_check_failed): Mark cold.
13683 2017-06-09 Richard Biener <rguenther@suse.de>
13685 PR tree-optimization/66623
13686 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
13687 refactor check_reduction into two parts, properly computing
13688 whether we have to check reduction validity for outer loop
13691 2017-06-09 Richard Biener <rguenther@suse.de>
13693 PR tree-optimization/79483
13694 * graphite-scop-detection.c (order): New global.
13695 (get_order): Compute bb to order mapping that satisfies code
13696 generation constraints.
13697 (cmp_pbbs): New helper.
13698 (build_scops): Start domwalk at entry block, sort generated
13701 2017-06-09 Richard Biener <rguenther@suse.de>
13703 PR middle-end/81007
13704 * ipa-polymorphic-call.c
13705 (ipa_polymorphic_call_context::restrict_to_inner_class):
13706 Skip FIELD_DECLs with error_mark_node type.
13707 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
13710 2017-06-09 Martin Liska <mliska@suse.cz>
13712 * predict.c (struct branch_predictor): New struct.
13713 (test_prediction_value_range): New test.
13714 (predict_c_tests): New function.
13715 * selftest-run-tests.c (selftest::run_tests): Run the function.
13716 * selftest.h: Declare new tests.
13718 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
13721 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
13722 gen_add3_insn did not fail.
13723 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
13724 r0, construct that number in a temporary reg and add that reg to r0.
13725 If asked to put the result in r0 as well, fail.
13727 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
13729 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13730 for early expansion of vec_eqv.
13732 2017-06-08 Jakub Jelinek <jakub@redhat.com>
13734 PR middle-end/81005
13735 * ubsan.c (instrument_null): Avoid pointless code temporary.
13736 (pass_ubsan::execute): Instrument aggregate arguments of calls.
13738 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
13742 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
13745 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
13748 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13750 * predict.c (unlikely_executed_edge_p): Move ahead.
13751 (probably_never_executed_edge_p): Use it.
13753 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
13755 PR middle-end/79988
13756 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
13757 gimple_call_builtin_p call.
13759 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13761 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
13762 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
13763 rtl_check_failed_type2, rtl_check_failed_code1,
13764 rtl_check_failed_code2, rtl_check_failed_code_mode,
13765 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
13766 rtvec_check_failed_bounds, rtl_check_failed_flag,
13767 _fatal_insn_not_found, _fatal_insn): Likewise.
13768 * tree.h (tree_contains_struct_check_failed,
13769 tree_check_failed, tree_not_check_failed,
13770 tree_class_check_failed, tree_range_check_failed,
13771 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
13772 tree_vec_elt_check_failed, phi_node_elt_check_failed,
13773 tree_operand_check_failed, omp_clause_check_failed,
13774 omp_clause_operand_check_failed, omp_clause_range_check_failed):
13777 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13779 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
13780 flag_branch_probabilities.
13781 * ipa-inline.c (edge_badness): Likewise.
13782 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
13783 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
13784 * predict.c (maybe_hot_frequency_p): Likewise.
13785 (probably_never_executed): Likewise.
13786 * sched-ebb.c (schedule_ebbs): Likewise.
13787 * sched-rgn.c (find_single_block_region): Likewise.
13788 * tracer.c (tail_duplicate): Likewise.
13790 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13792 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
13793 longer requires x_flag_profile_use.
13795 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13797 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
13798 instead of flag_reorder_blocks_and_partition.
13799 * dbxout.c (dbxout_function_end): Likewise.
13800 * dwarf2out.c (gen_subprogram_die): Likewise.
13801 * haifa-sched.c (sched_create_recovery_edges): Likewise.
13802 * hw-doloop.c (reorg_loops): Likewise.
13803 * varasm.c (assemble_start_function,
13804 assemble_end_function): Likewise.
13805 (decide_function_section): Do not check for
13806 flag_reorder_blocks_and_partition.
13808 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
13810 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
13812 (chkp_get_hard_register_fake_addr_expr): Ditto.
13813 (chkp_build_addr_expr): Add check for hard reg case.
13814 (chkp_parse_array_and_component_ref): Ditto.
13815 (chkp_find_bounds_1): Ditto.
13816 (chkp_process_stmt): Don't generate bounds store for
13819 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13821 * predict.c (maybe_hot_bb_p): Do not check profile status.
13822 (maybe_hot_edge_p): Likewise.
13823 (probably_never_executed): Check for zero counts even if profile
13825 (unlikely_executed_edge_p): New function.
13826 (unlikely_executed_stmt_p): New function.
13827 (unlikely_executed_bb_p): New function.
13828 (set_even_probabilities): Use unlikely predicates.
13829 (combine_predictions_for_bb): Likewise.
13830 (predict_paths_for_bb): Likewise.
13831 (predict_paths_leading_to_edge): Likewise.
13832 (determine_unlikely_bbs): New function.
13833 (estimate_bb_frequencies): Use it.
13834 (compute_function_frequency): Use zero counts even if profile is
13836 * profile-count.h: Fix typo.
13838 2017-08-08 Julia Koval <julia.koval@intel.com>
13840 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
13841 _mm512_mask_cvtsepi16_storeu_epi8,
13842 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
13843 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
13844 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
13845 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
13846 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
13847 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
13848 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
13849 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
13850 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
13851 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
13852 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
13853 __builtin_ia32_pmovuswb256mem_mask,
13854 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
13855 __builtin_ia32_pmovwb512mem_mask): New builtins.
13857 2017-08-08 Julia Koval <julia.koval@intel.com>
13859 PR target/73350,80862
13860 * config/i386/subst.md (round): Fix round pattern.
13861 * config/i386/i386.c (ix86_erase_embedded_rounding):
13862 Fix erasing rounding for the fixed pattern.
13864 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13866 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
13868 2017-06-08 Martin Liska <mliska@suse.cz>
13870 PR gcov-profile/80911
13871 * gcov.c (block_info::block_info): New constructor.
13873 2017-06-07 Carl Love <cel@us.ibm.com>
13875 * config/rs6000/rs6000-c: The return type of the following
13876 built-in functions was implemented as int not long long. Fix sign
13877 of return value for the unsigned version of vec_mulo and vec_mule.
13878 vector unsigned long long vec_bperm (vector unsigned long long,
13879 vector unsigned char)
13880 vector signed long long vec_mule (vector signed int,
13882 vector unsigned long long vec_mule (vector unsigned int,
13883 vector unsigned int)
13884 vector signed long long vec_mulo (vector signed int,
13886 vector unsigned long long vec_mulo (vector unsigned int,
13887 vector unsigned int)
13888 * doc/extend.texi: Fix the documentation for the built-in
13891 2017-06-07 Carl Love <cel@us.ibm.com>
13894 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
13897 2017-06-07 Carl Love <cel@us.ibm.com>
13899 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
13900 support, Generate doublehv for signed int/float for BE case only.
13902 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
13904 * doc/invoke.texi (mcx16): Rewrite.
13906 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13908 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
13909 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
13910 *mov<mode>_softfloat, and an anonymous splitter): Use
13911 nonimmediate_operand instead of rs6000_nonimmediate_operand.
13913 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13915 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
13917 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
13918 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
13919 (rs6000_debug_reg_global): Adjust.
13920 (rs6000_init_hard_regno_mode_ok): Adjust.
13921 (rs6000_dbx_register_number): Adjust.
13922 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
13923 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
13924 Remove SPE_ACC and SPEFSCR.
13925 (REG_ALLOC_ORDER): Ditto.
13926 (FRAME_POINTER_REGNUM): Change to 111.
13927 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
13928 (REG_CLASS_NAMES): Ditto.
13929 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
13930 (REGISTER_NAMES): Ditto.
13931 (ADDITIONAL_REG_NAMES): Ditto.
13932 (rs6000_reg_names): Ditto.
13933 * config/rs6000/rs6000.md: Renumber some register number
13936 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13938 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
13940 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
13941 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
13943 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
13944 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
13945 Delete the SPE high registers.
13946 (REG_ALLOC_ORDER): Ditto.
13947 (enum reg_class): Remove SPE_HIGH_REGS.
13948 (REG_CLASS_NAMES): Ditto.
13949 (REG_CLASS_CONTENTS): Delete the SPE high registers.
13950 (REGISTER_NAMES): Ditto.
13951 (rs6000_reg_names): Ditto.
13952 * doc/tm.texi.in: Remove SPE as example.
13953 * doc/tm.texi: Regenerate.
13955 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13957 * config/rs6000/8540.md (ppc8540_brinc): Delete.
13958 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
13959 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
13960 * config/rs6000/rs6000.md (type): Remove "brinc".
13962 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13964 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
13965 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
13966 * config/rs6000/linuxspe.h: Delete file.
13967 * config/rs6000/rs6000.md: Don't include spe.md.
13968 * config/rs6000/spe.h: Delete file.
13969 * config/rs6000/spe.md: Delete file.
13970 * config/rs6000/t-rs6000: Remove spe.md.
13972 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13974 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
13975 (reg_or_none500mem_operand): Delete.
13976 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
13977 instead of reg_or_none500mem_operand.
13979 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13981 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
13982 handling of SPE flags.
13983 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
13985 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13987 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
13989 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
13990 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
13991 paired_divv2sf3): Similar.
13992 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
13993 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
13994 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
13996 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
13997 Rename the paired_* instruction patterns.
13998 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
14000 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
14001 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
14002 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
14003 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
14004 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
14005 PAIRED_VECTOR_MODE.
14006 (struct machine_function): Delete field spe_insn_chain_scanned_p.
14007 (spe_func_has_64bit_regs_p): Delete.
14008 (spe_expand_predicate_builtin): Delete.
14009 (spe_expand_evsel_builtin): Delete.
14010 (TARGET_DWARF_REGISTER_SPAN): Do not define.
14011 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
14012 (invalid_e500_subreg): Delete.
14013 (rs6000_legitimize_address): Always force_reg op2 as well, for
14014 paired single memory accesses.
14015 (rs6000_member_type_forces_blk): Delete.
14016 (rs6000_spe_function_arg): Delete.
14017 (rs6000_expand_unop_builtin): Delete SPE handling.
14018 (rs6000_expand_binop_builtin): Ditto.
14019 (spe_expand_stv_builtin): Delete.
14020 (bdesc_2arg_spe): Delete.
14021 (spe_expand_builtin): Delete.
14022 (spe_expand_predicate_builtin): Delete.
14023 (spe_expand_evsel_builtin): Delete.
14024 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
14025 (spe_init_builtins): Delete.
14026 (spe_func_has_64bit_regs_p): Delete.
14027 (savres_routine_name): Delete "info" parameter. Adjust callers.
14028 (rs6000_emit_stack_reset): Ditto.
14029 (rs6000_dwarf_register_span): Delete.
14030 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
14031 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
14032 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
14034 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
14036 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
14037 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
14038 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
14039 mulv2sf3, divv2sf3): Delete expanders.
14041 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14043 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
14045 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14047 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
14048 * config/rs6000/rs6000.c: Ditto.
14050 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14052 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
14053 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
14054 comparison_operator.
14056 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14058 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
14059 * config/rs6000/rs6000.opt: Ditto.
14060 * config/rs6000/t-rtems: Ditto.
14062 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14064 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
14065 TARGET_E500_SINGLE by 0, simplify.
14066 * config/rs6000/rs6000.c: Ditto.
14067 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
14068 (spe_build_register_parallel): Delete.
14069 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
14070 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
14071 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
14072 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
14073 (E500_CONVERT): Delete.
14074 * config/rs6000/spe.md: Remove many patterns and all define_constants.
14076 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
14078 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
14079 * config/rs6000/dfp.md: Ditto.
14080 (negdd2, *negdd2_fpr): Merge.
14081 (absdd2, *absdd2_fpr): Merge.
14082 (negtd2, *negtd2_fpr): Merge.
14083 (abstd2, *abstd2_fpr): Merge.
14084 * config/rs6000/e500.h: Delete file.
14085 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
14086 TARGET_FPRS by 1 and simplify.
14087 * config/rs6000/rs6000-c.c: Ditto.
14088 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
14089 TARGET_DF_SPE by 0.
14090 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
14092 * config/rs6000/rs6000.md: Ditto.
14093 (floatdidf2, *floatdidf2_fpr): Merge.
14094 (move_from_CR_gt_bit): Delete.
14095 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
14096 (E500_CR_IOR_COMPARE): Delete.
14097 (All patterns that require !TARGET_FPRS): Delete.
14098 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
14100 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14102 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
14104 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14106 * graphds.c (add_edge): Intitialize edge's attached data.
14107 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
14108 pointer parameter. Call pointed function on each edge during
14109 graph traversing. Skip traversing the edge when the function
14111 (graphds_dfs, graphds_scc): Ditto.
14112 (for_each_edge): New parameter. Pass the new parameter to callback
14114 * graphds.h (skip_edge_callback): New function pointer type.
14115 (graphds_dfs, graphds_scc): New function pointer parameter.
14116 (graphds_edge_callback, for_each_edge): New parameter.
14118 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14120 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
14121 out code checking if runtime alias check is possible to below ...
14122 Call the new function.
14123 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
14124 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
14126 2017-06-07 Marek Polacek <polacek@redhat.com>
14129 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
14130 TYPE_OVERFLOW_WRAPS check.
14132 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14134 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
14135 if versioning is required.
14136 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
14137 peeling with the check for versioning.
14139 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14141 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
14142 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
14143 Set true to new parameter if new ssa variable is defined.
14144 (vect_gen_vector_loop_niters): Refactor. Set range information
14145 for the new vector loop bound variable.
14146 (vect_do_peeling): Ditto.
14148 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14150 * tree-affine.c (ssa.h): Include header file.
14151 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
14152 has wrapping overflow behavior.
14154 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14156 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
14158 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14160 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
14161 (tree_to_aff_combination): ... here.
14163 2017-06-07 Bin Cheng <bin.cheng@arm.com>
14165 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
14166 reg_pressure model function.
14167 (ivopts_global_cost_for_size): Delete.
14168 (determine_set_costs, iv_ca_recount_cost): Call new model function
14169 ivopts_estimate_reg_pressure.
14171 2017-06-07 Tamar Christina <tamar.christina@arm.com>
14173 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
14174 expensive than udiv. Remove floating point cases from mod.
14176 2017-06-07 Tamar Christina <tamar.christina@arm.com>
14178 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
14179 Increase idiv cost.
14181 2017-06-07 Tamar Christina <tamar.christina@arm.com>
14183 * config/aarch64/aarch64.md
14184 (copysignsf3): Fix mask generation.
14186 2017-06-07 Jakub Jelinek <jakub@redhat.com>
14188 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
14190 (class dump_manager): Add register_dumps method.
14191 * dumpfile.c: Include langhooks.h.
14192 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
14193 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
14194 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
14195 (dump_manager::dump_register): Start with 512 entries instead of 32.
14196 (dump_manager::register_dumps): New method.
14197 * toplev.c (general_init): Instead of invoking register_dumps
14198 langhook, invoke register_dumps method on the dump manager.
14199 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
14202 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
14204 * doc/md.texi: Clarify the restrictions on a define_insn condition.
14205 Say that # requires an associated define_split to exist, and that
14206 the define_split must be suitable for use after register allocation.
14208 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14210 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
14211 (compute_outgoing_frequencies): Also initialize zero counts.
14212 (find_many_sub_basic_blocks): Do not produce uninitialized profile
14213 around loops; preserve more of profile when nothing changes.
14215 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
14217 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
14219 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
14220 * config/arm/arm-cpu-cdata.h: Regenerate.
14221 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
14222 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
14223 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
14224 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
14225 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
14227 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
14229 * config/arm/t-rmprofile: Likewise.
14230 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
14232 2017-06-06 David S. Miller <davem@davemloft.net>
14235 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
14236 blockage if function uses alloca.
14238 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
14240 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
14241 New "uid" fields to hold pretty-print IDs of group and ref.
14242 Memory references are now identified as <group_id>:<ref_id>
14243 instead of using [random] addresses.
14244 (dump_mem_details): Simplify, no functional change.
14245 (dump_mem_ref): Simplify and make output more concise.
14246 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
14247 (find_or_create_group): Initialize group uid.
14248 (record_ref): Initialize ref uid. Improve debug output.
14249 (prune_group_by_reuse, should_issue_prefetch_p,)
14250 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
14251 (mark_nontemporal_store, determine_loop_nest_reuse):
14252 Improve debug output.
14254 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
14256 * dbgcnt.def (prefetch): New debug counter.
14257 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
14258 (schedule_prefetches): Stop issueing prefetches if debug counter
14261 2017-06-06 Tom de Vries <tom@codesourcery.com>
14263 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
14264 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
14266 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14268 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
14269 Use aarch64_reg_or_zero predicate for operand 4.
14270 (aarch64_compare_and_swap<mode> define_insn_and_split):
14271 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
14272 (aarch64_store_exclusive<mode>): Likewise for operand 2.
14274 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
14276 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
14277 (arm_compute_save_core_reg_mask): This.
14278 (thumb1_compute_save_reg_mask): Rename into ...
14279 (thumb1_compute_save_core_reg_mask): This.
14280 (arm_compute_save_reg0_reg12_mask): Adapt comment.
14281 (arm_compute_frame_layout): Likewise.
14283 2017-06-06 Richard Biener <rguenther@suse.de>
14285 PR tree-optimization/80974
14286 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
14287 keep or clear leaders SSA info.
14289 2017-06-06 Tom de Vries <tom@codesourcery.com>
14291 * config/nvptx/nvptx.c (split_mode_p): New function.
14292 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
14294 2017-06-06 Tom de Vries <tom@codesourcery.com>
14296 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
14298 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14301 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
14304 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14306 * shrink-wrap.c (handle_simple_exit): Update profile.
14307 (try_shrink_wrapping): Upate profile.
14309 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14311 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
14312 (tree_guess_outgoing_edge_probabilities): New.
14313 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
14314 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
14316 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14318 * ipa-split.c (split_function): Initialize return bb profile.
14320 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14322 * profile.c (compute_branch_probabilities): Also initialize
14323 EXIT_BLOCK profile.
14325 2017-06-06 Richard Biener <rguenther@suse.de>
14327 PR tree-optimization/80928
14328 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
14329 (vect_analyze_loop_operations): Properly guard analysis for
14331 (vect_transform_loop): Likewise.
14332 (vect_analyze_loop_2): Also reset SLP type on PHIs.
14333 (vect_model_induction_cost): Do not cost for pure SLP.
14334 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
14335 of induction in inner loop vectorization.
14336 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
14337 (vect_get_and_check_slp_defs): Handle vect_induction_def.
14338 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
14340 (vect_analyze_slp_cost_1): Cost induction.
14341 (vect_detect_hybrid_slp_stmts): Handle PHIs.
14342 (vect_get_slp_vect_defs): Likewise.
14343 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
14344 (vect_transform_stmt): Handle SLP reductions.
14345 * tree-vectorizer.h (vectorizable_induction): Adjust.
14347 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
14349 * config/rs6000/rs6000.c (make_resolver_func): Update
14350 init_lowered_empty_function call.
14352 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
14354 * doc/invoke.texi: Document the -fprofile-abs-path option.
14355 * common.opt (fprofile-abs-path): New option.
14356 * gcov-io.h (gcov_write_filename): Declare.
14357 * gcov-io.c (gcov_write_filename): New function.
14358 * coverage.c (coverage_begin_function): Use gcov_write_filename.
14359 * profile.c (output_location): Likewise.
14361 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
14363 * shring-wrap.c: Revert accidental commit.
14365 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
14367 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
14369 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
14371 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
14373 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
14374 profile in callgraph edge.
14375 * profile-count.h (apply_probability): If THIS is 0, then result is 0
14376 (apply_scale): Likewise.
14377 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
14378 Also scale profile when inlining function with zero profile.
14379 (initialize_cfun): Update exit block profile even when it is zero.
14380 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
14381 when profile is read.
14383 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
14385 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
14386 (CLONE_*): New constants to define the processors we can generate
14387 code for with the target_clone attribute.
14388 (rs6000_clone_map): New array to identify which clone processors
14389 the current program is running on.
14390 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
14391 target_clone attribute.
14392 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
14393 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
14394 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
14395 (cpu_expand_builtin): Add support for target_clone attribute.
14396 (rs6000_valid_attribute_p): Allow "default" attribute.
14397 (get_decl_name): New debug function to simplify printing the
14398 current function name in debugging statements.
14399 (rs6000_clone_priority): New functions to support the target_clone
14400 attribute, and be able to generate code to switch between ISA 2.05
14401 through ISA 3.0 (power6 through power9).
14402 (rs6000_compare_version_priority): Likewise.
14403 (rs6000_get_function_versions_dispatcher): Likewise.
14404 (make_resolver_func): Likewise.
14405 (add_condition_to_bb): Likewise.
14406 (dispatch_function_versions): Likewise.
14407 (rs6000_generate_version_dispatcher_body): Likewise.
14408 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
14409 (fusion_gpr_load_p): Fix a spacing issue.
14410 * doc/extend.texi (Common Function Attributes): Document that the
14411 PowerPC supports the target_clone attribute.
14413 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
14415 * config/arm/arm.h: explain F symbol found in description of ARM
14416 register allocation in its legend.
14418 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
14420 * config/mips/frame-header-opt.c: Include profile-count.h.
14421 * config/riscv/riscv.c: Include profile-count.h
14423 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
14425 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
14427 (sm_set_flag_if_changed): Add bbs field.
14428 (execute_sm_if_changed_flag_set): Pass BBS.
14429 (execute_sm): Update.
14431 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14433 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
14436 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14438 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
14439 (peephole2): New peephole2 to emit the above.
14440 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
14442 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14444 * config/aarch64/aarch64.c (define_peephole2 above
14445 *sub_<shift>_<mode>): New peephole.
14447 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
14449 * config/i386/i386.c (make_resolver_func): Update.
14450 * Makefile.in: Add profile-count.h and profile-count.o
14451 * auto-profile.c (afdo_indirect_call): Update to new API.
14452 (afdo_set_bb_count): Update.
14453 (afdo_propagate_edge): Update.
14454 (afdo_propagate_circuit): Update.
14455 (afdo_calculate_branch_prob): Update.
14456 (afdo_annotate_cfg): Update.
14457 * basic-block.h: Include profile-count.h
14458 (struct edge_def): Turn count to profile_count.
14459 (struct basic_block_def): Likewie.
14460 (REG_BR_PROB_BASE): Move to profile-count.h
14461 (RDIV): Move to profile-count.h
14462 * bb-reorder.c (max_entry_count): Turn to profile_count.
14463 (find_traces): Update.
14464 (rotate_loop):Update.
14465 (connect_traces):Update.
14466 (sanitize_hot_paths):Update.
14467 * bt-load.c (migrate_btr_defs): Update.
14468 * cfg.c (RDIV): Remove.
14469 (init_flow): Use alloc_block.
14470 (alloc_block): Uninitialize count.
14471 (unchecked_make_edge): Uninitialize count.
14472 (check_bb_profile): Update.
14473 (dump_edge_info): Update.
14474 (dump_bb_info): Update.
14475 (update_bb_profile_for_threading): Update.
14476 (scale_bbs_frequencies_int): Update.
14477 (scale_bbs_frequencies_gcov_type): Update.
14478 (scale_bbs_frequencies_profile_count): New.
14479 * cfg.h (update_bb_profile_for_threading): Update.
14480 (scale_bbs_frequencies_profile_count): Declare.
14481 * cfgbuild.c (compute_outgoing_frequencies): Update.
14482 (find_many_sub_basic_blocks): Update.
14483 * cfgcleanup.c (try_forward_edges): Update.
14484 (try_crossjump_to_edge): Update.
14485 * cfgexpand.c (expand_gimple_tailcall): Update.
14486 (construct_exit_block): Update.
14487 * cfghooks.c (verify_flow_info): Update.
14488 (dump_bb_for_graph): Update.
14489 (split_edge): Update.
14490 (make_forwarder_block): Update.
14491 (duplicate_block): Update.
14492 (account_profile_record): Update.
14493 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
14494 (get_estimated_loop_iterations): Update.
14495 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
14496 (single_likely_exit): Update.
14497 * cfgloopmanip.c (scale_loop_profile): Update.
14499 (set_zero_probability): Update.
14500 (lv_adjust_loop_entry_edge): Update.
14501 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
14502 (purge_dead_edges): Update.
14503 (rtl_account_profile_record): Update.
14504 * cgraph.c (cgraph_node::create): Uninitialize count.
14505 (symbol_table::create_edge): Uninitialize count.
14506 (cgraph_update_edges_for_call_stmt_node): Update.
14507 (cgraph_edge::dump_edge_flags): Update.
14508 (cgraph_node::dump): Update.
14509 (cgraph_edge::maybe_hot_p): Update.
14510 * cgraph.h: Include profile-count.h
14511 (create_clone), create_edge, create_indirect_edge): Update.
14512 (cgraph_node): Turn count to profile_count.
14513 (cgraph_edge0: Likewise.
14514 (make_speculative, clone): Update.
14515 (create_edge): Update.
14516 (init_lowered_empty_function): Update.
14517 * cgraphclones.c (cgraph_edge::clone): Update.
14518 (duplicate_thunk_for_node): Update.
14519 (cgraph_node::create_clone): Update.
14520 * cgraphunit.c (cgraph_node::analyze): Update.
14521 (cgraph_node::expand_thunk): Update.
14522 * final.c (dump_basic_block_info): Update.
14523 * gimple-streamer-in.c (input_bb): Update.
14524 * gimple-streamer-out.c (output_bb): Update.
14525 * graphite.c (print_global_statistics): Update.
14526 (print_graphite_scop_statistics): Update.
14527 * hsa-brig.c: Include basic-block.h.
14528 * hsa-dump.c: Include basic-block.h.
14529 * hsa-gen.c (T sum_slice): Update.
14530 (convert_switch_statements):Update.
14531 * hsa-regalloc.c: Include basic-block.h.
14532 * ipa-chkp.c (chkp_produce_thunks): Update.
14533 * ipa-cp.c (struct caller_statistics): Update.
14534 (init_caller_stats): Update.
14535 (gather_caller_stats): Update.
14536 (ipcp_cloning_candidate_p): Update.
14537 (good_cloning_opportunity_p): Update.
14538 (get_info_about_necessary_edges): Update.
14539 (dump_profile_updates): Update.
14540 (update_profiling_info): Update.
14541 (update_specialized_profile): Update.
14542 (perhaps_add_new_callers): Update.
14543 (decide_about_value): Update.
14544 (ipa_cp_c_finalize): Update.
14545 * ipa-devirt.c (struct odr_type_warn_count): Update.
14546 (struct decl_warn_count): Update.
14547 (struct final_warning_record): Update.
14548 (possible_polymorphic_call_targets): Update.
14549 (ipa_devirt): Update.
14550 * ipa-fnsummary.c (redirect_to_unreachable): Update.
14551 * ipa-icf.c (sem_function::merge): Update.
14552 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
14553 * ipa-inline.c (compute_uninlined_call_time): Update.
14554 (compute_inlined_call_time): Update.
14555 (want_inline_small_function_p): Update.
14556 (want_inline_self_recursive_call_p): Update.
14557 (edge_badness): Update.
14558 (lookup_recursive_calls): Update.
14559 (recursive_inlining): Update.
14560 (inline_small_functions): Update.
14561 (dump_overall_stats): Update.
14562 (dump_inline_stats): Update.
14563 * ipa-profile.c (ipa_profile_generate_summary): Update.
14564 (ipa_propagate_frequency): Update.
14565 (ipa_profile): Update.
14566 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
14567 * ipa-utils.c (ipa_merge_profiles): Update.
14568 * loop-doloop.c (doloop_modify): Update.
14569 * loop-unroll.c (report_unroll): Update.
14570 (unroll_loop_runtime_iterations): Update.
14571 * lto-cgraph.c (lto_output_edge): Update.
14572 (lto_output_node): Update.
14573 (input_node): Update.
14574 (input_edge): Update.
14575 (merge_profile_summaries): Update.
14576 * lto-streamer-in.c (input_cfg): Update.
14577 * lto-streamer-out.c (output_cfg): Update.
14578 * mcf.c (create_fixup_graph): Update.
14579 (adjust_cfg_counts): Update.
14580 (sum_edge_counts): Update.
14581 * modulo-sched.c (sms_schedule): Update.
14582 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
14583 * predict.c (maybe_hot_count_p): Update.
14584 (probably_never_executed): Update.
14585 (dump_prediction): Update.
14586 (combine_predictions_for_bb): Update.
14587 (propagate_freq): Update.
14588 (handle_missing_profiles): Update.
14589 (counts_to_freqs): Update.
14590 (rebuild_frequencies): Update.
14591 (force_edge_cold): Update.
14592 * predict.h: Include profile-count.h
14593 (maybe_hot_count_p, counts_to_freqs): UPdate.
14594 * print-rtl-function.c: Do not include cfg.h
14595 * print-rtl.c: Include basic-block.h
14596 * profile-count.c: New file.
14597 * profile-count.h: New file.
14598 * profile.c (is_edge_inconsistent): Update.
14599 (correct_negative_edge_counts): Update.
14600 (is_inconsistent): Update.
14601 (set_bb_counts): Update.
14602 (read_profile_edge_counts): Update.
14603 (compute_frequency_overlap): Update.
14604 (compute_branch_probabilities): Update; Initialize and deinitialize
14606 (branch_prob): Update.
14607 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
14608 (edge_gcov_count): New.
14609 (bb_gcov_count): New.
14610 * shrink-wrap.c (try_shrink_wrapping): Update.
14611 * tracer.c (better_p): Update.
14612 * trans-mem.c (expand_transaction): Update.
14613 (ipa_tm_insert_irr_call): Update.
14614 (ipa_tm_insert_gettmclone_call): Update.
14615 * tree-call-cdce.c: Update.
14616 * tree-cfg.c (gimple_duplicate_sese_region): Update.
14617 (gimple_duplicate_sese_tail): Update.
14618 (gimple_account_profile_record): Update.
14619 (execute_fixup_cfg): Update.
14620 * tree-inline.c (copy_bb): Update.
14621 (copy_edges_for_bb): Update.
14622 (initialize_cfun): Update.
14623 (freqs_to_counts): Update.
14624 (copy_cfg_body): Update.
14625 (expand_call_inline): Update.
14626 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
14627 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
14628 (try_unroll_loop_completely): Update.
14629 (try_peel_loop): Update.
14630 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
14631 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
14632 * tree-ssa-loop-split.c (connect_loops): Update.
14633 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
14634 * tree-ssa-reassoc.c (branch_fixup): Update.
14635 * tree-ssa-tail-merge.c (replace_block_by): Update.
14636 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
14637 (compute_path_counts): Update.
14638 (update_profile): Update.
14639 (recompute_probabilities): Update.
14640 (update_joiner_offpath_counts): Update.
14641 (estimated_freqs_path): Update.
14642 (freqs_to_counts_path): Update.
14643 (clear_counts_path): Update.
14644 (ssa_fix_duplicate_block_edges): Update.
14645 (duplicate_thread_path): Update.
14646 * tree-switch-conversion.c (case_bit_test_cmp): Update.
14647 (struct switch_conv_info): Update.
14648 * tree-tailcall.c (decrease_profile): Update.
14649 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
14650 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
14651 * value-prof.c (check_counter): Update.
14652 (gimple_divmod_fixed_value): Update.
14653 (gimple_mod_pow2): Update.
14654 (gimple_mod_subtract): Update.
14655 (gimple_ic_transform): Update.
14656 (gimple_stringop_fixed_value): Update.
14657 * value-prof.h (gimple_ic): Update.
14659 2017-06-02 Carl Love <cel@us.ibm.com>
14661 * config/rs6000/rs6000-c: Add support for built-in functions
14662 vector double vec_doublee (vector signed int);
14663 vector double vec_doublee (vector unsigned int);
14664 vector double vec_doublee (vector float);
14665 vector double vec_doubleh (vector signed int);
14666 vector double vec_doubleh (vector unsigned int);
14667 vector double vec_doubleh (vector float);
14668 vector double vec_doublel (vector signed int);
14669 vector double vec_doublel (vector unsigned int);
14670 vector double vec_doublel (vector float);
14671 vector double vec_doubleo (vector signed int);
14672 vector double vec_doubleo (vector unsigned int);
14673 vector double vec_doubleo (vector float);.
14674 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
14675 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
14677 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
14678 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
14679 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
14681 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
14682 vec_doublel, vec_doubleh.
14683 * doc/extend.texi: Update the built-in documentation file for the
14684 new built-in functions.
14686 2017-06-02 David Malcolm <dmalcolm@redhat.com>
14689 * ipa-inline-analysis.c (free_growth_caches): Set
14690 edge_removal_hook_holder to NULL after removing it.
14692 2017-06-02 Sudakshina Das <sudi.das@arm.com>
14694 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
14695 comparision with zero.
14697 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
14698 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14699 for early expansion of vec_min and vec_max builtins.
14700 (builtin_function_type): Add min/max unsigned variants to those
14701 identified as having unsigned arguments.
14703 2017-06-02 Olivier Hainque <hainque@adacore.com>
14705 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
14707 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14709 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
14710 Use VALL_F16 iterator rather than VALL.
14712 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14714 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
14715 Emit CBNZ inside loop when doing a strong exchange and comparing
14716 against zero. Generate the CC flags after the loop.
14718 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
14720 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
14721 (dl_section_ref): New.
14722 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
14723 On AIX, append an expression to subtract the size of the
14724 section length to dl_section_ref.
14726 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
14728 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14729 for early expansion of vector absolute builtins.
14731 2017-06-02 Richard Biener <rguenther@suse.de>
14733 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
14734 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
14736 2017-06-02 Richard Biener <rguenther@suse.de>
14738 PR tree-optimization/80948
14739 * tree-tailcall.c (find_tail_calls): Track stmts to move in
14740 stmt order as well.
14742 2017-06-02 Richard Biener <rguenther@suse.de>
14744 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
14746 * tree-vect-stmts.c (process_use): Do not mark backedge defs
14747 for inductions as relevant.
14749 2017-06-02 Richard Biener <rguenther@suse.de>
14751 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
14752 (vectorizable_induction): ... this. Remove dead code.
14754 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
14756 * builtins. (expand_builtin_alloca): Remove second parameter and
14757 infer its value from the first parameter instead.
14758 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
14760 2017-06-02 Jakub Jelinek <jakub@redhat.com>
14762 PR rtl-optimization/80903
14763 * loop-doloop.c (add_test): Unshare sequence.
14765 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
14767 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
14769 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
14771 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
14773 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
14774 xlogue_layout::get_instance, logue_layout::xlogue_layout,
14775 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
14776 (xlogue_layout::get_stub_rtx): Make static.
14777 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
14778 (xlogue_layout::compute_stub_managed_regs): Rename to...
14779 (xlogue_layout::count_stub_managed_regs): ...this.
14780 (xlogue_layout::is_stub_managed_reg): New function.
14781 (xlogue_layout::m_stub_names): Rename to...
14782 (xlogue_layout::s_stub_names): ...this, make static.
14783 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
14784 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
14785 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
14786 xlogue_layout::s_stub_names): Instantiate statics.
14787 (stub_managed_regs): Remove.
14788 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
14789 (disable_call_ms2sysv_xlogues): Rename to...
14790 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
14791 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
14793 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
14794 change after reload_completed.
14795 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
14797 (ix86_expand_prologue): Likewise.
14798 (ix86_expand_epilogue): Likewise.
14799 (ix86_expand_split_stack_prologue): Likewise.
14800 (ix86_compute_frame_layout): Remove frame parameter ...
14801 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
14802 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
14804 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
14805 (ix86_frame): Move from here ...
14806 * config/i386/i386.h (ix86_frame): ... to here.
14807 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
14808 complete ix86_frame data structure instead. Remove some_ld_name.
14810 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
14812 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
14813 symbols that hold a DECL_VALUE_EXPR.
14815 2017-06-01 Martin Jambor <mjambor@suse.cz>
14817 PR tree-optimization/80898
14818 * tree-sra.c (process_subtree_disqualification): Removed.
14819 (disqualify_candidate): Do not acll
14820 process_subtree_disqualification.
14821 (subtree_mark_written_and_enqueue): New function.
14822 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
14823 RHS has been disqualified and re-queue LHS if necessary. Apart
14824 from that, ignore disqualified RHS.
14826 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14828 * config/s390/s390.c (s390_emit_epilogue): Disable early return
14829 address fetch for z10 or later.
14831 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14833 * config/arc/arc.md (tst_movb): Add guard when splitting.
14835 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14837 * config/arc/arc.c (arc_can_eliminate): Test against
14838 arc_frame_pointer_needed.
14840 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14842 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
14843 to prevent store reordering.
14844 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
14845 (type): Add block type.
14846 (stack_tie): Define special instruction to be used in
14849 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14851 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
14852 constraint. It is not valid for the pattern.
14853 (noncommutative_binary_comparison): Likewise.
14855 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14857 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
14860 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14862 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
14863 be used by the reg-alloc.
14865 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14867 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
14868 reg-alloc when having mul64 or mul32x16 instructions.
14869 (mulsidi3): Likewise.
14870 (umulsidi3): Likewise.
14871 (mulsi32x16): New pattern.
14872 (mulsi64): Likewise.
14873 (mulsidi64): Likewise.
14874 (umulsidi64): Likewise.
14875 (MUL32x16_REG): Define.
14876 (mul64_600): Use MUL32x16_REG.
14877 (mac64_600): Likewise.
14878 (umul64_600): Likewise.
14879 (umac64_600): Likewise.
14881 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14883 * config/arc/arc.md (mulsi3_700): Make it commutative.
14885 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
14887 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
14889 (*sign_extendsidi2_insn): Likewise for movstosw.
14891 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
14893 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
14894 the type of the input discriminant value. Convert the
14895 discriminant value of signedness vary.
14897 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
14899 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
14900 Add to -Wall section.
14902 2017-06-01 Richard Biener <rguenther@suse.de>
14904 PR middle-end/66313
14905 * fold-const.c (fold_plusminus_mult_expr): If the factored
14906 factor may be zero use a wrapping type for the inner operation.
14907 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
14908 and handle moved defs.
14909 (process_assignment): Properly guard the unary op case. Return a
14910 tri-state indicating that moving the stmt before the call may allow
14911 to continue. Pass through to_move.
14912 (find_tail_calls): Handle moving unrelated defs before
14915 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
14918 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
14919 splitter result in the canonical way.
14921 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
14923 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
14924 also for 32bit target. Update insn attributes.
14925 (zero-extendsidi2 splitter): Allow all registers for operand 1.
14927 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
14929 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
14930 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
14931 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
14932 (_mm_maskz_min_ss): New intrinsics.
14934 2017-05-31 Martin Liska <mliska@suse.cz>
14936 * tree-vect-loop.c (vect_create_epilog_for_reduction):
14937 Change comment style to one we normally use.
14938 (vectorizable_reduction): Likewise.
14939 (vectorizable_induction): Likewise.
14940 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
14941 (vectorizable_call): Likewise.
14942 (vectorizable_simd_clone_call): Likewise.
14943 (vectorizable_conversion): Likewise.
14944 (vectorizable_assignment): Likewise.
14945 (vectorizable_shift): Likewise.
14946 (vectorizable_operation): Likewise.
14947 (vectorizable_store): Likewise.
14948 (vectorizable_load): Likewise.
14949 * tree-vectorizer.h: Likewise.
14951 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
14953 * passes.c (emergency_dump_function): New.
14954 * tree-pass.h (emergency_dump_function): Declare.
14955 * plugin.c (plugins_internal_error_function): Remove.
14956 * plugin.h (plugins_internal_error_function): Remove declaration.
14957 * toplev.c (internal_error_function): New static function. Use it...
14958 (general_init): ...here.
14960 2017-05-31 Graham Markall <graham.markall@embecosm.com>
14962 * config/arc/arc.c (arc_print_operand): Handle constant operands.
14963 (arc_rtx_costs): Add costs for new patterns.
14964 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
14965 * config/arc/predicates.md: Add _1_2_3_operand predicate.
14967 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
14969 * tree-ssa-strlen.c (get_next_strinfo): New function.
14970 (get_stridx_plus_constant): Use it.
14971 (zero_length_string): Likewise.
14972 (adjust_related_strinfos): Likewise.
14973 (adjust_last_stmt): Likewise.
14975 2017-05-31 Richard Biener <rguenther@suse.de>
14978 * config/i386/i386.c (ix86_expand_builtin): Remove assert
14979 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
14981 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
14983 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
14984 loop_vinfo argument and use of dependence distance vectors.
14985 Check instead whether the two references differ only in their
14986 initial value and assume that they have the same alignment if the
14987 difference is a multiple of the vector alignment.
14988 (vect_analyze_data_refs_alignment): Update call accordingly.
14990 2017-05-31 Martin Liska <mliska@suse.cz>
14993 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
14995 2017-05-31 Bin Cheng <bin.cheng@arm.com>
14997 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
14998 (create_intersect_range_checks): Move from ...
14999 * tree-data-ref.c (create_intersect_range_checks_index)
15000 (create_intersect_range_checks): ... to here.
15001 (create_runtime_alias_checks): New function factored from ...
15002 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
15003 here. Call above function.
15004 * tree-data-ref.h (create_runtime_alias_checks): New function.
15006 2017-05-31 Bin Cheng <bin.cheng@arm.com>
15008 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
15009 segment length for dr_b and compute it in wide_int.
15011 2017-05-31 Richard Biener <rguenther@suse.de>
15013 PR tree-optimization/80906
15014 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
15015 and pass through iv_map.
15016 (copy_bb_and_scalar_dependences): Adjust.
15017 (translate_pending_phi_nodes): Likewise.
15018 (copy_loop_close_phi_args): Handle code-generating IVs instead
15021 2017-05-30 David Malcolm <dmalcolm@redhat.com>
15023 * diagnostic-color.c (color_dict): Add "type-diff".
15024 (parse_gcc_colors): Update comment.
15025 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
15026 -fdiagnostics-show-template-tree and -fno-elide-type.
15027 (GCC_COLORS): Add type-diff to example.
15029 (-fdiagnostics-show-template-tree): New.
15030 (-fno-elide-type): New.
15031 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
15032 the pp_format_decoder callback. Call any m_format_postprocessor's
15034 (pretty_printer::pretty_printer): Initialize
15035 m_format_postprocessor.
15036 (pretty_printer::~pretty_printer): Delete any
15037 m_format_postprocessor.
15038 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
15039 (class format_postprocessor): New class.
15040 (struct pretty_printer::format_decoder): Document the new parameters.
15041 (struct pretty_printer::m_format_postprocessor): New field.
15042 * tree-diagnostic.c (default_tree_printer): Update for new
15043 bool and const char ** params.
15044 * tree-diagnostic.h (default_tree_printer): Likewise.
15046 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
15048 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
15049 (lwa_operand): Delete rs6000_gen_cell_microcode test.
15050 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
15051 rs6000_gen_cell_microcode code.
15052 (rs6000_final_prescan_insn): Delete.
15053 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
15054 "warn-cell-microcode" entries.
15055 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
15056 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
15057 throughout. Change cc_reg_not_micro_cr0_operand to
15058 cc_reg_not_cr0_operand throughout.
15059 (*extendhi<mode>2_noload): Delete.
15060 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
15061 (mwarn-cell-microcode): Delete.
15062 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
15063 -mgen-cell-microcode and -mwarn-cell-microcode.
15065 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
15068 * config/i386/constraints.md (Yd): New constraint.
15070 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
15071 and (?Yd, r) alternatives. Update insn attributes.
15072 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
15073 and (?*Yd, r) alternatives. Update insn attributes.
15074 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
15076 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
15078 * gimplify.c (gimplify_modify_expr): Don't create a
15079 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
15082 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
15084 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
15086 2017-05-30 Richard Biener <rguenther@suse.de>
15088 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
15089 and reduc_def fields.
15090 (STMT_VINFO_REDUC_TYPE): New define.
15091 (STMT_VINFO_REDUC_DEF): Likewise.
15092 (vect_force_simple_reduction): Adjust prototype.
15093 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
15094 (vect_is_simple_reduction): Remove check_reduction argument.
15095 (vect_force_simple_reduction): Adjust and set
15096 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
15097 (vectorizable_reduction): Do not re-do reduction analysis
15098 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
15099 * tree-parloops.c (gather_scalar_reductions): Adjust.
15101 2017-05-30 Richard Biener <rguenther@suse.de>
15103 PR middle-end/80901
15104 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
15107 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
15109 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
15110 Introduce unknown_misalignment parameter and remove vf.
15111 (vect_peeling_hash_get_lowest_cost):
15112 Pass unknown_misalignment parameter.
15113 (vect_enhance_data_refs_alignment):
15114 Fix unsupportable data ref treatment.
15116 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
15118 * tree-vect-data-refs.c (vect_get_data_access_cost):
15119 Workaround for SLP handling.
15120 (vect_enhance_data_refs_alignment):
15121 Compute costs for doing no peeling at all, compare to the best
15122 peeling costs so far and avoid peeling if cheaper.
15124 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
15126 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
15127 Return peeling info and set costs to zero for unlimited cost
15129 (vect_enhance_data_refs_alignment): Also inspect all datarefs
15130 with unknown misalignment. Compute and costs for unknown
15131 misalignment, compare them to the costs for known misalignment
15132 and choose the cheapest for peeling.
15134 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
15136 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
15137 (vect_get_peeling_costs_all_drs): Create function.
15138 (vect_peeling_hash_get_lowest_cost):
15139 Use vect_get_peeling_costs_all_drs.
15140 (vect_peeling_supportable): Create function.
15141 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
15143 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
15145 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
15146 DR_HAS_NEGATIVE_STEP.
15147 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
15148 (vect_enhance_data_refs_alignment): Use.
15149 (vect_duplicate_ssa_name_ptr_info): Use.
15150 * tree-vectorizer.h (dr_misalignment): Use.
15151 (known_alignment_for_access_p): Use.
15153 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
15156 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
15158 (has_section_name): New function.
15160 2017-05-30 Martin Liska <mliska@suse.cz>
15163 * auto-profile.c (get_function_decl_from_block): Fix
15166 2017-05-30 Richard Biener <rguenther@suse.de>
15168 PR middle-end/80876
15169 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
15171 2017-05-30 Martin Liska <mliska@suse.cz>
15173 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
15174 * dumpfile.h (struct dump_file_info): Remove ctors.
15176 2017-05-30 Martin Liska <mliska@suse.cz>
15178 * predict.def: Fix GNU coding style.
15180 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
15182 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
15183 Mark 'to' argument with ATTRIBUTE_UNUSED.
15185 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
15187 * config/xtensa/xtensa.c (xtensa_emit_call): Use
15188 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
15189 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
15192 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
15194 * doc/install.texi (Options specification): Restore entry of
15195 --enable-sjlj-exceptions.
15197 2017-05-27 Michael Eager <eager@eagercon.com>
15200 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
15202 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
15204 * config/microblaze/microblaze.h
15205 (FIXED_REGISTERS): Update in macro.
15206 (CALL_USED_REGISTERS): Update in macro.
15208 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
15210 * doc/install.texi: Add links to macOS binary distributions.
15212 2017-05-27 Jakub Jelinek <jakub@redhat.com>
15216 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
15218 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
15220 2017-05-26 Martin Liska <mliska@suse.cz>
15222 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
15224 2017-05-26 Martin Liska <mliska@suse.cz>
15226 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
15227 always leading ';; '.
15228 (dump_bb_info): Likewise.
15229 (brief_dump_cfg): Likewise.
15230 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
15231 * dumpfile.c: Remove usage of TDF_VERBOSE.
15232 * dumpfile.h (enum dump_kind): Likewise.
15233 (dump_gimple_bb_header): Do not use TDF_COMMENT.
15234 * print-tree.c (debug_verbose): Remove.
15235 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
15236 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
15237 * tree-diagnostic.c (default_tree_printer): Replace
15238 TDF_DIAGNOSTIC with TDF_SLIM.
15240 2017-05-26 Bin Cheng <bin.cheng@arm.com>
15242 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
15243 in parameter loop, rather than loop_vinfo.
15244 (create_intersect_range_checks): Ditto.
15245 (vect_create_cond_for_alias_checks): Update call to above functions.
15247 2017-05-26 Bin Cheng <bin.cheng@arm.com>
15249 PR tree-optimization/80815
15250 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
15251 for merging runtime alias checks. Handle negative DR_STEPs.
15253 2017-05-26 Bin Cheng <bin.cheng@arm.com>
15255 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
15257 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
15258 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
15259 out code pruning runtime alias checks.
15260 * tree-data-ref.c (prune_runtime_alias_test_list): New function
15261 factored out from above.
15262 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
15264 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
15266 (prune_runtime_alias_test_list): New decalaration.
15268 2017-05-26 Bin Cheng <bin.cheng@arm.com>
15270 * tree-vect-data-refs.c (compare_tree): Rename and move ...
15271 * tree-data-ref.c (data_ref_compare_tree): ... to here.
15272 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
15273 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
15274 (operator==, comp_dr_with_seg_len_pair): Ditto.
15275 (vect_prune_runtime_alias_test_list): Ditto.
15277 2017-05-26 Martin Liska <mliska@suse.cz>
15280 * params.def: Bound partial-inlining-entry-probability param.
15282 2017-05-26 Marek Polacek <polacek@redhat.com>
15285 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
15288 2017-05-26 Richard Biener <rguenther@suse.de>
15290 PR tree-optimization/80842
15291 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
15294 2017-05-26 Richard Biener <rguenther@suse.de>
15296 PR tree-optimization/80844
15297 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
15299 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
15301 * doc/md.texi (Machine Constraints): Update x86 family
15302 machine constraints section to match 'config/i386/constraints.md'.
15304 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
15306 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
15308 2017-05-25 Nathan Sidwell <nathan@acm.org>
15310 * doc/invoke.texi (--enable-languages): Update documentation.
15312 2017-05-25 Martin Liska <mliska@suse.cz>
15314 * dumpfile.c: Add TDF_FOLDING.
15315 * dumpfile.h (enum dump_kind): Likewise.
15316 * genmatch.c (dt_simplify::gen_1): Use it.
15318 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
15320 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
15322 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
15324 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
15325 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
15327 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
15329 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
15330 * match.pd (X == C): Rewrite it here.
15331 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
15332 with_certain_nonzero_bits2): New predicates.
15333 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
15335 2017-05-24 Nathan Sidwell <nathan@acm.org>
15337 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
15340 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
15343 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
15345 * config/powerpcspe: New port. Files are copied from the rs6000
15346 port, with "rs6000" in filenames replaced by "powerpcspe".
15348 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
15350 PR rtl-optimization/80754
15351 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
15353 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
15355 * config/sparc/sparc.md (length): Return the correct value for -mflat
15356 sibcalls to match output_sibcall.
15358 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
15362 * config/rs6000/rs6000.c (struct machine_function): Add new field
15364 (rs6000_get_separate_components): Init that field, use it.
15365 (rs6000_components_for_bb): Use the field.
15367 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
15369 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
15371 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
15373 PR middle-end/80823
15374 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
15376 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15379 * config/s390/s390.c (s390_check_qrst_address): Check incoming
15380 address against address_operand predicate.
15381 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
15383 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
15385 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
15386 parameters passed indirectly.
15388 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
15390 * config/i386/i386.md (*movdi_internal): Remove SSE4
15391 alternative 18 (?r, *v). Update insn attributes.
15392 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
15393 Update insn attributes.
15394 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
15395 Update insn attributes.
15396 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
15397 alternative 1 (r, v). Remove isa attribute.
15398 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
15399 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
15400 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
15402 2017-05-23 Tom de Vries <tom@codesourcery.com>
15404 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
15407 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
15409 * cgraphunit.c (symbol_table::process_new_functions): Update.
15410 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
15411 (inline_generate_summary): Rename to ...
15412 (ipa_fn_summary_generate): ... this one.
15413 (inline_read_summary): Rename to ...
15414 (ipa_fn_summary_read): ... this one.
15415 (inline_write_summary): Rename to ...
15416 (ipa_fn_summary_write): ... this one.
15417 (inline_free_summary): Rename to ...
15418 (ipa_free_fn_summary): ... this one.
15419 (pass_data_local_fn_summary, pass_local_fn_summary,
15420 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
15421 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
15422 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
15423 make_pass_ipa_fn_summary): New.
15424 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
15425 inline_write_summary, inline_free_summary): Remove.
15426 (ipa_free_fn_summary) : New.
15427 * ipa-inline.c (ipa_inline): Update.
15428 (pass_ipa_inline): Do not generate summaries.
15429 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
15431 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
15432 and add pass_ipa_fn_summary.
15433 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
15435 (make_pass_inline_parameters): Remove.
15437 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
15439 * omp-low.c (struct omp_context): Remove "default_kind" member.
15442 * omp-offload.c (execute_oacc_device_lower): Remove the
15443 parallelism dimensions function attributes for unparallelized
15444 OpenACC kernels constructs.
15446 2017-05-23 Martin Liska <mliska@suse.cz>
15448 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
15450 (cgraph_edge::make_speculative): Likewise.
15451 (cgraph_edge::resolve_speculation): Likewise.
15452 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
15453 (cgraph_node::dump): Likewise.
15454 * cgraph.h: Likewise.
15455 * cgraphunit.c (analyze_functions): Likewise.
15456 (symbol_table::compile): Likewise.
15457 * ipa-cp.c (print_all_lattices): Likewise.
15458 (determine_versionability): Likewise.
15459 (initialize_node_lattices): Likewise.
15460 (ipcp_verify_propagated_values): Likewise.
15461 (estimate_local_effects): Likewise.
15462 (update_profiling_info): Likewise.
15463 (create_specialized_node): Likewise.
15464 (perhaps_add_new_callers): Likewise.
15465 (decide_about_value): Likewise.
15466 (decide_whether_version_node): Likewise.
15467 (identify_dead_nodes): Likewise.
15468 (ipcp_store_bits_results): Likewise.
15469 * ipa-devirt.c (dump_targets): Likewise.
15470 (ipa_devirt): Likewise.
15471 * ipa-icf.c (sem_item::dump): Likewise.
15472 (sem_function::equals): Likewise.
15473 (sem_variable::equals): Likewise.
15474 (sem_item_optimizer::read_section): Likewise.
15475 (sem_item_optimizer::execute): Likewise.
15476 (congruence_class::dump): Likewise.
15477 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
15478 (dump_inline_summary): Likewise.
15479 (estimate_node_size_and_time): Likewise.
15480 (inline_analyze_function): Likewise.
15481 * ipa-inline-transform.c (inline_call): Likewise.
15482 * ipa-inline.c (report_inline_failed_reason): Likewise.
15483 (want_early_inline_function_p): Likewise.
15484 (edge_badness): Likewise.
15485 (update_edge_key): Likewise.
15486 (inline_small_functions): Likewise.
15487 * ipa-profile.c (ipa_profile): Likewise.
15488 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
15489 (ipa_make_edge_direct_to_target): Likewise.
15490 (remove_described_reference): Likewise.
15491 (ipa_impossible_devirt_target): Likewise.
15492 (propagate_controlled_uses): Likewise.
15493 (ipa_print_node_params): Likewise.
15494 (ipcp_transform_function): Likewise.
15495 * ipa-pure-const.c (pure_const_read_summary): Likewise.
15496 (propagate_pure_const): Likewise.
15497 * ipa-reference.c (generate_summary): Likewise.
15498 (read_write_all_from_decl): Likewise.
15499 (propagate): Likewise.
15500 (ipa_reference_read_optimization_summary): Likewise.
15501 * ipa-utils.c (ipa_merge_profiles): Likewise.
15502 * ipa.c (walk_polymorphic_call_targets): Likewise.
15503 (symbol_table::remove_unreachable_nodes): Likewise.
15504 (ipa_single_use): Likewise.
15505 * passes.c (execute_todo): Likewise.
15506 * predict.c (drop_profile): Likewise.
15507 * symtab.c (symtab_node::get_dump_name): New function.
15508 (symtab_node::dump_name): Likewise.
15509 (symtab_node::dump_asm_name): Likewise.
15510 (symtab_node::dump_references): Likewise.
15511 (symtab_node::dump_referring): Likewise.
15512 (symtab_node::dump_base): Likewise.
15513 (symtab_node::debug_symtab): Likewise.
15514 * tree-sra.c (convert_callers_for_node): Likewise.
15515 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
15516 * value-prof.c (init_node_map): Likewise.
15518 2017-05-23 Martin Liska <mliska@suse.cz>
15520 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
15521 and symtab_node::debug_symtab to symbol_table::debug.
15522 * cgraphunit.c (analyze_functions): Use the renamed function.
15523 (symbol_table::compile): Likewise.
15524 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
15525 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
15526 * passes.c (execute_todo): Likewise.
15527 * symtab.c (symbol_table::dump): New function.
15528 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
15530 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
15532 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
15533 that nonconst implies exec.
15535 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
15537 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
15538 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
15539 (inline_edge_summary_vec): Turn into ...
15540 (ipa_call_summaries): ... this one.
15541 (redirect_to_unreachable, edge_set_predicate,
15542 evaluate_properties_for_edge, inline_summary_alloc,
15543 reset_ipa_call_summary, reset_inline_summary,
15544 inline_summary_t::duplicate): Update.
15545 (inline_edge_duplication_hook): Turn to ...
15546 (ipa_call_summary_t::duplicate): ... this one.
15547 (inline_edge_removal_hook): Turn to ...
15548 (ipa_call_summary_t::remove): ... this one.
15549 (dump_inline_edge_summary): Turn to ...
15550 (dump_ipa_call_summary): ... this one.
15551 (estimate_function_body_sizes): Update.
15552 (inline_update_callee_summaries): Update.
15553 (remap_edge_change_prob): Update.
15554 (remap_edge_summaries): Update.
15555 (inline_merge_summary): Update.
15556 (do_estimate_edge_time): Update.
15557 (inline_generate_summary): Update.
15558 (inline_read_section): Update.
15559 (inline_read_summary): Update.
15560 (inline_free_summary): Update.
15561 * ipa-inline.c (can_inline_edge_p): Update.
15562 (compute_inlined_call_time): Update.
15563 (want_inline_small_function_p): Update.
15564 (edge_badness): Update.
15565 (early_inliner): Update.
15566 * ipa-inline.h (inline_edge_summary): Turn to ...
15567 (ipa_call_summary): ... this one.
15568 (ipa_call_summary_t): New class.
15569 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
15570 (ipa_call_summaries): New.
15571 (inline_edge_summary): Remove.
15572 (estimate_edge_growth): Update.
15573 * ipa-profile.c (ipa_propagate_frequency_1): Update.
15574 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
15575 * ipa-split.c (execute_split_functions): Update.
15576 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
15578 2017-05-23 Tom de Vries <tom@codesourcery.com>
15580 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
15581 attributes): Document rdrand effective target.
15583 2017-05-23 Tom de Vries <tom@codesourcery.com>
15585 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
15586 attributes): Sort alphabetically.
15588 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
15590 * config/avr/genmultilib.awk: Use gsub instead of gensub.
15592 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
15595 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
15596 V2DF/V2DI splat into two separate patterns, one that handles
15597 registers, and the other that only handles memory. Drop support
15598 for splatting from a GPR on ISA 2.07 and then splitting the
15599 splat into direct move and splat.
15600 (vsx_splat_<mode>_reg): Likewise.
15601 (vsx_splat_<mode>_mem): Likewise.
15603 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
15605 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
15607 2017-05-22 Jakub Jelinek <jakub@redhat.com>
15609 PR middle-end/80809
15610 * omp-low.c (finish_taskreg_remap): New function.
15611 (finish_taskreg_scan): If unit size of ctx->record_type
15612 is non-constant, unshare the size expression and replace
15613 decls in it with possible outer var refs.
15615 PR middle-end/80809
15616 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
15617 GOVD_SHARED rather than GOVD_PRIVATE with it.
15618 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
15619 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
15621 PR middle-end/80853
15622 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
15623 as last argument to build_outer_var_ref for pointer bases of array
15624 section reductions.
15626 2017-05-19 Martin Sebor <msebor@redhat.com>
15628 * print-tree.c (print_node): Print DECL_READ_P flag.
15630 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15632 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
15633 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
15634 * cgraph.c: Likewise.
15635 * cgraphunit.c: Likewise.
15636 * gengtype.c: Likewise.
15637 * ipa-cp.c: Likewise.
15638 * ipa-devirt.c: Likewise.
15639 * ipa-icf.c: Likewise.
15640 * ipa-predicate.c: Likewise.
15641 * ipa-profile.c: Likewise.
15642 * ipa-prop.c: Likewise.
15643 * ipa-split.c: Likewise.
15645 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
15646 edge_predicate_pool, dump_inline_hints,
15647 inline_summary::account_size_time, redirect_to_unreachable,
15648 edge_set_predicate, set_hint_predicate,
15649 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
15650 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
15651 inline_summary_t::remove, remap_hint_predicate_after_duplication,
15652 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
15653 ipa_call_summary_t::remove, initialize_growth_caches,
15654 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
15655 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
15656 mark_modified, unmodified_parm_1, unmodified_parm,
15657 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
15658 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
15659 compute_bb_predicates, will_be_nonconstant_expr_predicate,
15660 will_be_nonconstant_predicate, record_modified_bb_info,
15661 get_minimal_bb, record_modified, param_change_prob,
15662 phi_result_unknown_predicate, predicate_for_phi_result,
15663 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
15664 estimate_function_body_sizes, compute_inline_parameters,
15665 compute_inline_parameters_for_curren, pass_data_inline_parameters,
15666 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
15667 inline_update_callee_summaries, remap_edge_change_prob,
15668 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
15669 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
15670 inline_analyze_function, inline_summary_t::insert,
15671 inline_generate_summary, read_ipa_call_summary, inline_read_section,
15672 inline_read_summary, write_ipa_call_summary, inline_write_summary,
15673 inline_free_summary): Move to ipa-fnsummary.h
15674 (predicate_t): Remove.
15675 * ipa-fnsummary.c: New file.
15676 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
15677 (enum inline_hints_vals, inline_hints, agg_position_info,
15678 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
15679 inline_summaries, ipa_call_summary, ipa_call_summary_t,
15680 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
15681 dump_inline_summary, dump_inline_hints, inline_generate_summary,
15682 inline_read_summary, inline_write_summary, inline_free_summary,
15683 inline_analyze_function, initialize_inline_failed,
15684 inline_merge_summary, inline_update_overall_summary,
15685 compute_inline_parameters): Move to ipa-fnsummary.h
15686 * ipa-fnsummary.h: New file.
15687 * ipa-inline-transform.h: Include ipa-inline.h.
15688 * ipa-inline.c: LIkewise.
15690 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15692 * ipa-inline.c (edge_badness): Use inlined_time instead of
15693 inline_summaries->get.
15695 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15697 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
15699 2017-05-22 Nathan Sidwell <nathan@acm.org>
15701 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
15702 (fdump-lang): Document 'raw' option.
15703 * dumpfile.h (TDI_tu): Delete.
15704 * dumpfile.c (dump_files): Remove translation-unit.
15705 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
15707 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
15709 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
15710 command option from $(AWK) call.
15711 * config/avr/genmultilib.awk: Simplify and rewrite so that it
15712 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
15713 [FORMAT]: Remove handling of variable.
15714 * config/avr/t-multilib: Regenerate.
15716 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15718 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
15720 (dump_inline_summary): Do not print self_time.
15721 (estimate_function_body_sizes): Do not set self_time.
15722 (compute_inline_parameters): Likewise.
15723 (inline_read_section, inline_write_summary): Do not stream self_time.
15724 * ipa-inline.h (inline_summary): Drop self_time.
15726 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15728 * ipa-inline-analysis.c (account_size_time): Rename to ...
15729 (inline_summary::account_size_time): ... this one.
15730 (reset_ipa_call_summary): Turn to ...
15731 (ipa_call_summary::reset): ... this one.
15732 (reset_inline_summary): Turn to ...
15733 (inline_summary::reset): ... this one.
15734 (inline_summary_t::remove): Update.
15735 (inline_summary_t::duplicate): Update.
15736 (ipa_call_summary_t::remove): Update.
15737 (dump_inline_summary): Update.
15738 (estimate_function_body_sizes): Update.
15739 (compute_inline_parameters): Update.
15740 (estimate_node_size_and_time): Update.
15741 (inline_merge_summary): Update.
15742 (inline_update_overall_summary): Update.
15743 (inline_read_section): Update.
15744 (inline_write_summary): Update.
15745 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
15746 add account_size_time and reset member functions.
15747 (ipa_call_summary): Add reset function.
15748 * ipa-predicate.h (predicate::operator &): Constify.
15750 2017-05-22 Richard Biener <rguenther@suse.de>
15752 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
15754 2017-05-19 Jason Merrill <jason@redhat.com>
15756 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
15758 2017-05-19 Marek Polacek <polacek@redhat.com>
15761 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
15762 TYPE_OVERFLOW_WRAPS checks.
15764 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
15766 * tree-core.h (enum omp_clause_default_kind): Add
15767 "OMP_CLAUSE_DEFAULT_PRESENT".
15768 * tree-pretty-print.c (dump_omp_clause): Handle it.
15769 * gimplify.c (enum gimplify_omp_var_data): Add
15770 "GOVD_MAP_FORCE_PRESENT".
15771 (gimplify_adjust_omp_clauses_1): Map it to
15772 "GOMP_MAP_FORCE_PRESENT".
15773 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
15775 * gimplify.c (oacc_default_clause): Clarify.
15777 2017-05-19 Nathan Sidwell <nathan@acm.org>
15779 LANG_HOOK_REGISTER_DUMPS
15780 * toplev.c (general_init): Call register dump lang hook.
15781 * doc/invoke.texi: Document -fdump-lang option family.
15782 * dumpfile.c (dump_files): Remove class dump here.
15783 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
15784 * dumpfile.h (tree_dump_index): Remove TDI_class.
15785 * langhooks-def.h (lhd_register_dumps): Declare.
15786 (LANG_HOOKS_REGISTER_DUMPS): Define.
15787 (LANG_HOOKS_INITIALIZER): Add it.
15788 * langhooks.c (lhd_register_dumps): Define.
15789 * langhooks.h (struct lang_hooks): Add register_dumps.
15791 2017-05-19 Nathan Sidwell <nathan@acm.org>
15793 * context.h (context::set_passes): New.
15794 * context.c (context::context): Do not create pass manager.
15795 * toplev.c (general_init): Create pass manager here.
15797 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
15799 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
15800 use this splitter if two add or or instructions would also work for
15801 the constant we want to generate.
15803 2017-05-19 Richard Biener <rguenther@suse.de>
15806 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
15807 predicate evaluation.
15809 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
15811 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
15813 * ipa-inline.c (want_inline_small_function_p): Do not cast to
15816 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
15818 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
15819 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
15820 (inline_edge_summary_vec): Turn into ...
15821 (ipa_call_summaries): ... this one.
15822 (redirect_to_unreachable, edge_set_predicate,
15823 evaluate_properties_for_edge, inline_summary_alloc,
15824 reset_ipa_call_summary, reset_inline_summary,
15825 inline_summary_t::duplicate): Update.
15826 (inline_edge_duplication_hook): Turn to ...
15827 (ipa_call_summary_t::duplicate): ... this one.
15828 (inline_edge_removal_hook): Turn to ...
15829 (ipa_call_summary_t::remove): ... this one.
15830 (dump_inline_edge_summary): Turn to ...
15831 (dump_ipa_call_summary): ... this one.
15832 (estimate_function_body_sizes): Update.
15833 (inline_update_callee_summaries): Update.
15834 (remap_edge_change_prob): Update.
15835 (remap_edge_summaries): Update.
15836 (inline_merge_summary): Update.
15837 (do_estimate_edge_time): Update.
15838 (inline_generate_summary): Update.
15839 (inline_read_section): Update.
15840 (inline_read_summary): Update.
15841 (inline_free_summary): Update.
15842 * ipa-inline.c (can_inline_edge_p): Update.
15843 (compute_inlined_call_time): Update.
15844 (want_inline_small_function_p): Update.
15845 (edge_badness): Update.
15846 (early_inliner): Update.
15847 * ipa-inline.h (inline_edge_summary): Turn to ...
15848 (ipa_call_summary): ... this one.
15849 (ipa_call_summary_t): New class.
15850 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
15851 (ipa_call_summaries): New.
15852 (inline_edge_summary): Remove.
15853 (estimate_edge_growth): Update.
15854 * ipa-profile.c (ipa_propagate_frequency_1): Update.
15855 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
15856 * ipa-split.c (execute_split_functions): Update.
15857 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
15859 2017-05-19 Richard Biener <rguenther@suse.de>
15861 PR middle-end/80764
15862 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
15864 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
15866 * config/rs6000/rs6000.c (struct machine_function): Add field
15867 fpr_is_wrapped_separately.
15868 (rs6000_get_separate_components): Use 64 components. Handle the
15869 new FPR components.
15870 (rs6000_components_for_bb): Handle the FPR components.
15871 (rs6000_emit_prologue_components): Handle the FPR components.
15872 (rs6000_emit_epilogue_components): Handle the FPR components.
15873 (rs6000_set_handled_components): Handle the FPR components.
15874 (rs6000_emit_prologue): Don't output prologue code for those FPRs
15875 that are already separately shrink-wrapped.
15876 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
15877 that are already separately shrink-wrapped.
15879 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
15882 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
15885 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
15886 (define_peephole2 for Altivec d-form load): Add peepholes to catch
15887 cases where the register allocator uses a move and an offsettable
15888 memory operation to/from a FPR register on ISA 2.06/2.07.
15889 (define_peephole2 for Altivec d-form store): Likewise.
15891 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
15894 * config/i386/mmx.md (*mov<mode>_internal): Enable
15895 alternatives 11, 12, 13 and 14 also for 32bit targets.
15896 Remove alternatives 15, 16, 17 and 18.
15897 * config/i386/sse.md (vec_concatv2di): Change
15898 alternative (!x, *y) to (x, ?!*Yn).
15900 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
15902 * dumpfile.h (enum dump_kind): Remove stray comma.
15904 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
15906 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
15907 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
15908 predicate::num_conditions
15909 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
15910 (CHANGED): turn into predicate::changed.
15911 (agg_position_info): Move to ipa-predicate.h
15912 (add_condition, predicate::add_clause, predicate::operator &=,
15913 predicate::or_with, predicate::evaluate, predicate::probability,
15914 dump_condition, dump_clause, predicate::dump,
15915 predicate::remap_after_duplication, predicate::remap_after_inlining,
15916 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
15917 (evaluate_conditions_for_known_args): Update.
15918 (set_cond_stmt_execution_predicate): Update.
15919 * ipa-inline.h: Include ipa-predicate.h
15920 (condition, inline_param_summary, conditions, agg_position_info,
15921 predicate): Move to ipa-predicate.h
15922 * ipa-predicate.c: New file.
15923 * ipa-predicate.h: New file.
15925 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
15927 * final.c (leaf_function_p): Check we are not in a sequence.
15929 2017-05-18 Martin Liska <mliska@suse.cz>
15931 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
15932 * dumpfile.c (dump_register): Use new enum dump_kind.
15933 (get_dump_file_name): Likewise.
15934 (dump_enable_all): Likewise.
15935 (dump_switch_p_1): Likewise.
15936 (enable_rtl_dump_file): Remove usage of TDF_RTL.
15937 * dumpfile.h (enum dump_kind): New enum type.
15938 (struct dump_file_info): Create constructor and
15939 format fields and comments.
15940 * passes.c (pass_manager::register_one_dump_file):
15942 * statistics.c (statistics_early_init): Likewise.
15943 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
15944 TDF_TREE with TDF_SLIM.
15945 (gather_memory_references_ref): Likewise.
15947 2017-05-18 Martin Liska <mliska@suse.cz>
15949 * vec.h (struct vnull): Use it.
15951 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
15953 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
15954 (true_predicate, false_predicate, true_predicate_p,
15955 false_predicate_p): Remove.
15956 (single_cond_predicate, not_inlined_predicate): Turn to member function
15958 (add_condition): Update.
15959 (add_clause): Turn to...
15960 (predicate::add_clause): ... this one; update; allow passing NULL
15962 (and_predicates): Turn to ...
15963 (predicate::operator &=): ... this one.
15964 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
15965 (or_predicates): Turn to ...
15966 (predicate::or_with): ... this one.
15967 (evaluate_predicate): Turn to ...
15968 (predicate::evaluate): ... this one.
15969 (predicate_probability): Turn to ...
15970 (predicate::probability): ... this one.
15971 (dump_condition): Update.
15972 (dump_predicate): Turn to ...
15973 (predicate::dump): ... this one.
15974 (account_size_time): Update.
15975 (edge_set_predicate): Update.
15976 (set_hint_predicate): UPdate.
15977 (evaluate_conditions_for_known_args): Update.
15978 (evaluate_properties_for_edge): Update.
15979 (remap_predicate_after_duplication): Turn to...
15980 (predicate::remap_after_duplication): ... this one.
15981 (remap_hint_predicate_after_duplication): Update.
15982 (inline_summary_t::duplicate): UPdate.
15983 (dump_inline_edge_summary): Update.
15984 (dump_inline_summary): Update.
15985 (set_cond_stmt_execution_predicate): Update.
15986 (set_switch_stmt_execution_predicate): Update.
15987 (compute_bb_predicates): Update.
15988 (will_be_nonconstant_expr_predicate): Update.
15989 (will_be_nonconstant_predicate): Update.
15990 (phi_result_unknown_predicate): Update.
15991 (predicate_for_phi_result): Update.
15992 (array_index_predicate): Update.
15993 (estimate_function_body_sizes): Update.
15994 (estimate_node_size_and_time): Update.
15995 (estimate_ipcp_clone_size_and_time): Update.
15996 (remap_predicate): Rename to ...
15997 (predicate::remap_after_inlining): ... this one.
15998 (remap_hint_predicate): Update.
15999 (inline_merge_summary): Update.
16000 (inline_update_overall_summary): Update.
16001 (estimate_size_after_inlining): Update.
16002 (read_predicate): Rename to ...
16003 (predicate::stream_in): ... this one.
16004 (read_inline_edge_summary): Update.
16005 (write_predicate): Rename to ...
16006 (predicate::stream_out): ... this one.
16007 (write_inline_edge_summary): Update.
16008 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
16009 (clause_t): Turn to uint32_t
16010 (predicate): Turn to class; implement constructor and operators
16012 (size_time_entry): Update.
16013 (inline_summary): Update.
16014 (inline_edge_summary): Update.
16016 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
16018 * fold-const.c (fold_binary_loc): Move transformation...
16019 * match.pd (C - X CMP X): ... here.
16021 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
16023 * config/sparc/sparc.c (sparc_option_override): Set function
16024 alignment for -mcpu=niagara7 to 64 to match the I$ line.
16025 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
16027 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
16029 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
16031 2017-05-18 Marek Polacek <polacek@redhat.com>
16034 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
16035 (pass_ubsan::execute): Call gimple_assign_single_p instead of
16036 gimple_assign_load_p.
16038 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
16040 PR middle-end/80692
16041 * real.c (do_compare): Give decimal_do_compare preference over
16042 comparing just the signs.
16044 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
16046 * doc/md.texi (Canonicalization of Instructions): Describe the
16047 canonical form of instructions that inherently set a condition
16050 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
16052 PR middle-end/80775
16053 * tree-cfg.c: Move deletion of unreachable case statements to after
16054 the merging of consecutive case labels.
16056 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
16058 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
16059 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
16060 restoring of callee-saved registers.
16062 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
16064 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
16065 * config/visium/visium.c (single_set_and_flags): Likewise.
16066 * config/visium/visium.md (Substitutions): Likewise.
16068 2017-05-17 Martin Liska <mliska@suse.cz>
16070 * cfg.c: Introduce dump_flags_t type and
16071 use it instead of int type.
16073 * cfghooks.c: Likewise.
16074 * cfghooks.h (struct cfg_hooks): Likewise.
16075 * cfgrtl.c: Likewise.
16076 * cfgrtl.h: Likewise.
16077 * cgraph.c (cgraph_node::get_body): Likewise.
16078 * coretypes.h: Likewise.
16079 * domwalk.c: Likewise.
16080 * domwalk.h: Likewise.
16081 * dumpfile.c (struct dump_option_value_info): Likewise.
16082 (dump_enable_all): Likewise.
16083 (dump_switch_p_1): Likewise.
16084 (opt_info_switch_p): Likewise.
16085 * dumpfile.h (enum tree_dump_index): Likewise.
16086 (struct dump_file_info): Likewise.
16087 * genemit.c: Likewise.
16088 * generic-match-head.c: Likewise.
16089 * gengtype.c (open_base_files): Likewise.
16090 * gimple-pretty-print.c: Likewise.
16091 * gimple-pretty-print.h: Likewise.
16092 * graph.c (print_graph_cfg): Likewise.
16093 * graphite-scop-detection.c (dot_all_sese): Likewise.
16094 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
16095 * loop-unroll.c (report_unroll): Likewise.
16096 * passes.c (pass_manager::register_one_dump_file): Likewise.
16097 * print-tree.c: Likewise.
16098 * statistics.c: Likewise.
16099 * tree-cfg.c: Likewise.
16100 * tree-cfg.h: Likewise.
16101 * tree-dfa.c: Likewise.
16102 * tree-dfa.h: Likewise.
16103 * tree-dump.c (dump_function): Likewise.
16104 * tree-dump.h (struct dump_info): Likewise.
16105 * tree-pretty-print.c: Likewise.
16106 * tree-pretty-print.h: Likewise.
16107 * tree-ssa-live.c: Likewise.
16108 * tree-ssa-live.h: Likewise.
16109 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
16110 * tree-vect-loop.c: Likewise.
16111 * tree-vect-slp.c: Likewise.
16113 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
16114 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16116 PR tree-optimization/80457
16117 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
16118 of all arguments to a statement as scalar_to_vec operations.
16119 (vectorizable_call): Adjust call to vect_model_simple_cost for
16121 (vectorizable_conversion): Likewise.
16122 (vectorizable_assignment): Likewise.
16123 (vectorizable_shift): Likewise.
16124 (vectorizable_operation): Likewise.
16125 (vectorizable_comparison): Likewise.
16126 (vect_is_simple_cond): Record the def types for operands.
16127 (vectorizable_condition): Likewise, call vect_model_simple_cost.
16128 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
16129 for statement argument count.
16131 2017-05-16 Carl Love <cel@us.ibm.com>
16133 * config/rs6000/rs6000-c: Add support for built-in functions
16134 vector unsigned long long vec_bperm (vector unsigned long long,
16135 vector unsigned char)
16136 vector signed long long vec_mule (vector signed int,
16138 vector unsigned long long vec_mule (vector unsigned int,
16139 vector unsigned int)
16140 vector signed long long vec_mulo (vector signed int,
16142 vector unsigned long long vec_mulo (vector unsigned int,
16143 vector unsigned int)
16144 vector signed char vec_sldw (vector signed char,
16145 vector signed char,
16147 vector unsigned char vec_sldw (vector unsigned char,
16148 vector unsigned char,
16150 vector signed short vec_sldw (vector signed short,
16151 vector signed short,
16153 vector unsigned short vec_sldw (vector unsigned short,
16154 vector unsigned short,
16156 vector signed int vec_sldw (vector signed int,
16159 vector unsigned int vec_sldw (vector unsigned int,
16160 vector unsigned int,
16162 vector signed long long vec_sldw (vector signed long long,
16163 vector signed long long,
16165 vector unsigned long long vec_sldw (vector unsigned long long,
16166 vector unsigned long long,
16168 * config/rs6000/rs6000-c: Add support for built-in functions
16169 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
16170 * config/rs6000/altivec.h: Add defintion for vec_sldw.
16171 * doc/extend.texi: Update the built-in documentation for the
16172 new built-in functions.
16174 2017-05-16 Marek Polacek <polacek@redhat.com>
16178 * tree.c (save_expr): Don't fold the expression.
16180 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
16182 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
16183 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
16184 and (?*y,m). Update insn attributes.
16186 2017-05-16 Martin Liska <mliska@suse.cz>
16188 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
16189 flags argument of print_gimple_stmt, print_gimple_expr,
16190 print_generic_stmt and print_generic_expr.
16191 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
16192 * coretypes.h: Likewise.
16193 * except.c (dump_eh_tree): Likewise.
16194 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
16195 * gimple-pretty-print.h: Likewise.
16196 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
16197 (backprop::push_to_worklist): Likewise.
16198 (backprop::pop_from_worklist): Likewise.
16199 (backprop::process_use): Likewise.
16200 (backprop::intersect_uses): Likewise.
16201 (note_replacement): Likewise.
16202 * gimple-ssa-store-merging.c
16203 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
16204 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
16205 (pass_store_merging::execute): Likewise.
16206 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
16207 (ssa_base_cand_dump_callback): Likewise.
16208 (dump_incr_vec): Likewise.
16209 (replace_refs): Likewise.
16210 (replace_mult_candidate): Likewise.
16211 (create_add_on_incoming_edge): Likewise.
16212 (create_phi_basis): Likewise.
16213 (insert_initializers): Likewise.
16214 (all_phi_incrs_profitable): Likewise.
16215 (introduce_cast_before_cand): Likewise.
16216 (replace_one_candidate): Likewise.
16217 * gimplify.c (gimplify_expr): Likewise.
16218 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
16219 (set_rename): Likewise.
16220 (rename_uses): Likewise.
16221 (copy_loop_phi_nodes): Likewise.
16222 (add_close_phis_to_merge_points): Likewise.
16223 (copy_loop_close_phi_args): Likewise.
16224 (copy_cond_phi_args): Likewise.
16225 (graphite_copy_stmts_from_block): Likewise.
16226 (translate_pending_phi_nodes): Likewise.
16227 * graphite-poly.c (print_pdr): Likewise.
16228 (dump_gbb_cases): Likewise.
16229 (dump_gbb_conditions): Likewise.
16230 (print_scop_params): Likewise.
16231 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
16232 (build_cross_bb_scalars_use): Likewise.
16233 (gather_bbs::before_dom_children): Likewise.
16234 * hsa-dump.c (dump_hsa_immed): Likewise.
16235 * ipa-cp.c (print_ipcp_constant_value): Likewise.
16236 (get_replacement_map): Likewise.
16237 * ipa-inline-analysis.c (dump_condition): Likewise.
16238 (estimate_function_body_sizes): Likewise.
16239 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
16240 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
16241 * ipa-prop.c (ipa_dump_param): Likewise.
16242 (ipa_print_node_jump_functions_for_edge): Likewise.
16243 (ipa_modify_call_arguments): Likewise.
16244 (ipa_modify_expr): Likewise.
16245 (ipa_dump_param_adjustments): Likewise.
16246 (ipa_dump_agg_replacement_values): Likewise.
16247 (ipcp_modif_dom_walker::before_dom_children): Likewise.
16248 * ipa-pure-const.c (check_stmt): Likewise.
16249 (pass_nothrow::execute): Likewise.
16250 * ipa-split.c (execute_split_functions): Likewise.
16251 * omp-offload.c (dump_oacc_loop_part): Likewise.
16252 (dump_oacc_loop): Likewise.
16253 * trans-mem.c (tm_log_emit): Likewise.
16254 (tm_memopt_accumulate_memops): Likewise.
16255 (dump_tm_memopt_set): Likewise.
16256 (dump_tm_memopt_transform): Likewise.
16257 * tree-cfg.c (gimple_verify_flow_info): Likewise.
16258 (print_loop): Likewise.
16259 * tree-chkp-opt.c (chkp_print_addr): Likewise.
16260 (chkp_gather_checks_info): Likewise.
16261 (chkp_get_check_result): Likewise.
16262 (chkp_remove_check_if_pass): Likewise.
16263 (chkp_use_outer_bounds_if_possible): Likewise.
16264 (chkp_reduce_bounds_lifetime): Likewise.
16265 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
16266 (chkp_mark_completed_bounds): Likewise.
16267 (chkp_register_incomplete_bounds): Likewise.
16268 (chkp_mark_invalid_bounds): Likewise.
16269 (chkp_maybe_copy_and_register_bounds): Likewise.
16270 (chkp_build_returned_bound): Likewise.
16271 (chkp_get_bound_for_parm): Likewise.
16272 (chkp_build_bndldx): Likewise.
16273 (chkp_get_bounds_by_definition): Likewise.
16274 (chkp_generate_extern_var_bounds): Likewise.
16275 (chkp_get_bounds_for_decl_addr): Likewise.
16276 * tree-chrec.c (chrec_apply): Likewise.
16277 * tree-data-ref.c (dump_data_reference): Likewise.
16278 (dump_subscript): Likewise.
16279 (dump_data_dependence_relation): Likewise.
16280 (analyze_overlapping_iterations): Likewise.
16281 * tree-inline.c (expand_call_inline): Likewise.
16282 (tree_function_versioning): Likewise.
16283 * tree-into-ssa.c (dump_defs_stack): Likewise.
16284 (dump_currdefs): Likewise.
16285 (dump_names_replaced_by): Likewise.
16286 (dump_update_ssa): Likewise.
16287 (update_ssa): Likewise.
16288 * tree-object-size.c (pass_object_sizes::execute): Likewise.
16289 * tree-parloops.c (build_new_reduction): Likewise.
16290 (try_create_reduction_list): Likewise.
16291 (ref_conflicts_with_region): Likewise.
16292 (oacc_entry_exit_ok_1): Likewise.
16293 (oacc_entry_exit_single_gang): Likewise.
16294 * tree-pretty-print.h: Likewise.
16295 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
16296 (get_scalar_evolution): Likewise.
16297 (add_to_evolution): Likewise.
16298 (get_loop_exit_condition): Likewise.
16299 (analyze_evolution_in_loop): Likewise.
16300 (analyze_initial_condition): Likewise.
16301 (analyze_scalar_evolution): Likewise.
16302 (instantiate_scev): Likewise.
16303 (number_of_latch_executions): Likewise.
16304 (gather_chrec_stats): Likewise.
16305 (final_value_replacement_loop): Likewise.
16306 (scev_const_prop): Likewise.
16307 * tree-sra.c (dump_access): Likewise.
16308 (disqualify_candidate): Likewise.
16309 (create_access): Likewise.
16310 (reject): Likewise.
16311 (maybe_add_sra_candidate): Likewise.
16312 (create_access_replacement): Likewise.
16313 (analyze_access_subtree): Likewise.
16314 (analyze_all_variable_accesses): Likewise.
16315 (sra_modify_assign): Likewise.
16316 (initialize_constant_pool_replacements): Likewise.
16317 (find_param_candidates): Likewise.
16318 (decide_one_param_reduction): Likewise.
16319 (replace_removed_params_ssa_names): Likewise.
16320 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
16321 * tree-ssa-copy.c (dump_copy_of): Likewise.
16322 (copy_prop_visit_cond_stmt): Likewise.
16323 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
16324 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
16325 (record_equivalences_from_stmt): Likewise.
16326 * tree-ssa-dse.c (compute_trims): Likewise.
16327 (delete_dead_call): Likewise.
16328 (delete_dead_assignment): Likewise.
16329 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
16330 (forward_propagate_into_cond): Likewise.
16331 (pass_forwprop::execute): Likewise.
16332 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
16333 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
16335 (move_computations_worker): Likewise.
16336 (execute_sm): Likewise.
16337 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
16338 (remove_exits_and_undefined_stmts): Likewise.
16339 (remove_redundant_iv_tests): Likewise.
16340 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
16341 (adjust_iv_update_pos): Likewise.
16342 * tree-ssa-math-opts.c (bswap_replace): Likewise.
16343 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
16344 (value_replacement): Likewise.
16345 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
16346 * tree-ssa-pre.c (print_pre_expr): Likewise.
16347 (get_representative_for): Likewise.
16348 (create_expression_by_pieces): Likewise.
16349 (insert_into_preds_of_block): Likewise.
16350 (eliminate_insert): Likewise.
16351 (eliminate_dom_walker::before_dom_children): Likewise.
16352 (eliminate): Likewise.
16353 (remove_dead_inserted_code): Likewise.
16354 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
16355 * tree-ssa-reassoc.c (get_rank): Likewise.
16356 (eliminate_duplicate_pair): Likewise.
16357 (eliminate_plus_minus_pair): Likewise.
16358 (eliminate_not_pairs): Likewise.
16359 (undistribute_ops_list): Likewise.
16360 (eliminate_redundant_comparison): Likewise.
16361 (update_range_test): Likewise.
16362 (optimize_range_tests_var_bound): Likewise.
16363 (optimize_vec_cond_expr): Likewise.
16364 (rewrite_expr_tree): Likewise.
16365 (rewrite_expr_tree_parallel): Likewise.
16366 (linearize_expr): Likewise.
16367 (break_up_subtract): Likewise.
16368 (linearize_expr_tree): Likewise.
16369 (attempt_builtin_powi): Likewise.
16370 (attempt_builtin_copysign): Likewise.
16371 (transform_stmt_to_copy): Likewise.
16372 (transform_stmt_to_multiply): Likewise.
16373 (dump_ops_vector): Likewise.
16374 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
16375 (print_scc): Likewise.
16376 (set_ssa_val_to): Likewise.
16377 (visit_reference_op_store): Likewise.
16378 (visit_use): Likewise.
16379 (sccvn_dom_walker::before_dom_children): Likewise.
16380 (run_scc_vn): Likewise.
16381 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
16383 (expr_hash_elt::print): Likewise.
16384 (const_and_copies::pop_to_marker): Likewise.
16385 (const_and_copies::record_const_or_copy_raw): Likewise.
16386 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
16387 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
16388 (dump_predicates): Likewise.
16389 (find_uninit_use): Likewise.
16390 (warn_uninitialized_phi): Likewise.
16391 (pass_late_warn_uninitialized::execute): Likewise.
16392 * tree-ssa.c (verify_vssa): Likewise.
16393 (verify_ssa): Likewise.
16394 (maybe_optimize_var): Likewise.
16395 * tree-vrp.c (dump_value_range): Likewise.
16396 (dump_all_value_ranges): Likewise.
16397 (dump_asserts_for): Likewise.
16398 (register_edge_assert_for_2): Likewise.
16399 (vrp_visit_cond_stmt): Likewise.
16400 (vrp_visit_switch_stmt): Likewise.
16401 (vrp_visit_stmt): Likewise.
16402 (vrp_visit_phi_node): Likewise.
16403 (simplify_cond_using_ranges_1): Likewise.
16404 (fold_predicate_in): Likewise.
16405 (evrp_dom_walker::before_dom_children): Likewise.
16406 (evrp_dom_walker::push_value_range): Likewise.
16407 (evrp_dom_walker::pop_value_range): Likewise.
16408 (execute_early_vrp): Likewise.
16410 2017-05-16 Richard Biener <rguenther@suse.de>
16412 * dwarf2out.c (loc_list_from_tree_1): Do not create
16413 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
16415 2017-05-16 Richard Biener <rguenther@suse.de>
16417 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
16419 (note_variable_value_in_expr): If we resolved the decl ref
16420 do not push to the stack.
16422 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
16424 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
16425 operations in fast-math mode.
16426 (vaddq_f16): Likewise.
16427 (vmul_f16): Likewise.
16428 (vmulq_f16): Likewise.
16429 (vsub_f16): Likewise.
16430 (vsubq_f16): Likewise.
16431 * config/arm/neon.md (add<mode>3): New.
16433 (fma:<VH:mode>3): New. Also remove outdated comment.
16436 2017-05-16 Martin Liska <mliska@suse.cz>
16440 * ipa-devirt.c (warn_types_mismatch): Fix typo.
16441 (odr_types_equivalent_p): Likewise.
16443 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
16445 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
16447 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
16450 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
16451 non-interunit SSE move alternatives with '?'.
16452 (zero-extendsidi peephole2): New peephole to skip intermediate
16453 general register in SSE zero-extend sequence.
16455 2017-05-15 Jeff Law <law@redhat.com>
16457 * reorg.c (relax_delay_slots): Create a new variable to hold
16458 the temporary target rather than clobbering TARGET_LABEL.
16460 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
16461 missing argument to extract_bit_field call.
16462 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
16464 2017-05-15 Martin Liska <mliska@suse.cz>
16467 * gcc.c (process_command): Do not allow empty argument of -o option.
16469 2017-05-15 Renlin Li <renlin.li@arm.com>
16471 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
16472 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
16473 * config/aarch64/constraints.md (Usf): Add long call check.
16474 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
16475 (call_value): Likewise.
16476 (sibcall): Likewise.
16477 (sibcall_value): Likewise.
16479 (call_value_insn): New.
16480 (sibcall_insn): Update rtx pattern.
16481 (sibcall_value_insn): Likewise.
16482 (call_internal): Remove.
16483 (call_value_internal): Likewise.
16484 (sibcall_internal): Likewise.
16485 (sibcall_value_internal): Likewise.
16486 (call_reg): Likewise.
16487 (call_symbol): Likewise.
16488 (call_value_reg): Likewise.
16489 (call_value_symbol): Likewise.
16491 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
16494 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
16496 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
16498 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
16499 compatible with CCGOCmode and with CCZmode.
16501 2017-05-14 Martin Sebor <msebor@redhat.com>
16503 PR middle-end/77671
16504 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
16505 (gimple_fold_builtin_snprintf): Same.
16506 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
16507 (gimple_fold_builtin_snprintf): Same.
16508 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
16509 of character types.
16510 (is_call_safe): New function.
16511 (try_substitute_return_value): Call it.
16512 (try_simplify_call): New function.
16513 (pass_sprintf_length::handle_gimple_call): Call it.
16515 2017-05-14 Martin Sebor <msebor@redhat.com>
16517 PR middle-end/80669
16518 * builtins.c (expand_builtin_stpncpy): Simplify.
16520 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
16522 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
16523 * config/i386/i386.h
16524 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
16525 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
16526 (struct machine_function): Add new members call_ms2sysv,
16527 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
16528 (struct machine_frame_state): New fields sp_realigned and
16529 sp_realigned_offset.
16530 * config/i386/i386.c
16531 (enum xlogue_stub): New enum.
16532 (enum xlogue_stub_sets): New enum.
16533 (class xlogue_layout): New class.
16534 (struct ix86_frame): New fields stack_realign_allocate_offset,
16535 stack_realign_offset and outlined_save_offset. Modify comments to
16536 detail stack layout when using out-of-line stubs.
16537 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
16538 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
16539 -mcall-ms2sysv-xlogues.
16540 (stub_managed_regs): New static variable.
16541 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
16542 registers managed by out-of-line stub.
16543 (disable_call_ms2sysv_xlogues): New function.
16544 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
16545 m->call_ms2sysv when appropriate and compute frame layout for
16547 (sp_valid_at, fp_valid_at): New inline functions.
16548 (choose_basereg): New function.
16549 (choose_baseaddr): Add align parameter, use choose_basereg and modify
16551 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
16552 Use align parameter of choose_baseaddr to generated aligned SSE movs
16554 (pro_epilogue_adjust_stack): Modify to track
16555 machine_frame_state::sp_realigned.
16556 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
16557 (ix86_nsaved_sseregs): Likewise.
16558 (ix86_emit_save_regs): Likewise.
16559 (ix86_emit_save_regs_using_mov): Likewise.
16560 (ix86_emit_save_sse_regs_using_mov): Likewise.
16561 (get_scratch_register_on_entry): Likewise.
16562 (gen_frame_set): New function.
16563 (gen_frame_load): Likewise.
16564 (gen_frame_store): Likewise.
16565 (emit_outlined_ms2sysv_save): Likewise.
16566 (emit_outlined_ms2sysv_restore): Likewise.
16567 (ix86_expand_prologue): Modify stack re-alignment code and call
16568 emit_outlined_ms2sysv_save when appropriate.
16569 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
16570 parameter rtx_insn *insn, which allows the function to be used to only
16571 generate the notes.
16572 (ix86_expand_epilogue): Modify validity checks of frame and stack
16573 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
16574 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
16575 * config/i386/predicates.md
16576 (save_multiple): New predicate.
16577 (restore_multiple): Likewise.
16578 * config/i386/sse.md
16579 (save_multiple<mode>): New pattern.
16580 (save_multiple_realign<mode>): Likewise.
16581 (restore_multiple<mode>): Likewise.
16582 (restore_multiple_and_return<mode>): Likewise.
16583 (restore_multiple_leave_return<mode>): Likewise.
16584 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
16586 2017-05-14 Julia Koval <julia.koval@intel.com>
16588 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
16589 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
16590 (__builtin_ia32_xsetbv): New builtins.
16591 * config/i386/i386.c (ix86_expand_special_args_builtin):
16593 (ix86_expand_builtin): Special expand for new intrinsics.
16594 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
16595 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
16596 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
16598 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16600 * cfganal.c (inverted_post_order_compute): Change argument type
16602 * cfganal.h (inverted_post_order_compute): Adjust prototype.
16603 * df-core.c (rest_of_handle_df_initialize): Adjust.
16604 (rest_of_handle_df_finish): Likewise.
16605 (df_analyze_1): Likewise.
16606 (df_analyze): Likewise.
16607 (loop_inverted_post_order_compute): Change argument to be a vec *.
16608 (df_analyze_loop): Adjust.
16609 (df_get_n_blocks): Likewise.
16610 (df_get_postorder): Likewise.
16611 * df.h (struct df_d): Change field to be a vec.
16612 * lcm.c (compute_laterin): Adjust.
16613 (compute_available): Likewise.
16614 * lra-lives.c (lra_create_live_ranges_1): Likewise.
16615 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
16616 * tree-ssa-pre.c (compute_antic): Likewise.
16618 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16620 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
16621 (depth_first_search::depth_first_search): Change structure init
16622 function to this constructor.
16623 (depth_first_search::add_bb): Rename function to this member.
16624 (depth_first_search::execute): Likewise.
16625 (flow_dfs_compute_reverse_finish): Adjust.
16627 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16629 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
16630 (longest_simple_path): Likewise.
16631 * shrink-wrap.c (spread_components): Likewise.
16632 (disqualify_problematic_components): Likewise.
16633 (emit_common_heads_for_components): Likewise.
16634 (emit_common_tails_for_components): Likewise.
16635 (insert_prologue_epilogue_for_components): Likewise.
16637 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16639 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
16642 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16644 * df-core.c (df_set_blocks): Start using auto_bitmap.
16645 (df_compact_blocks): Likewise.
16646 * df-problems.c (df_rd_confluence_n): Likewise.
16647 * df-scan.c (df_insn_rescan_all): Likewise.
16648 (df_process_deferred_rescans): Likewise.
16649 (df_update_entry_block_defs): Likewise.
16650 (df_update_exit_block_uses): Likewise.
16651 (df_entry_block_bitmap_verify): Likewise.
16652 (df_exit_block_bitmap_verify): Likewise.
16653 (df_scan_verify): Likewise.
16654 * lra-constraints.c (lra_constraints): Likewise.
16655 (undo_optional_reloads): Likewise.
16656 (lra_undo_inheritance): Likewise.
16657 * lra-remat.c (calculate_gen_cands): Likewise.
16658 (do_remat): Likewise.
16659 * lra-spills.c (assign_spill_hard_regs): Likewise.
16660 (spill_pseudos): Likewise.
16661 * tree-ssa-pre.c (bitmap_set_and): Likewise.
16662 (bitmap_set_subtract_values): Likewise.
16664 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16666 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
16667 management with auto_bitmap.
16668 (fix_inter_tick): Likewise.
16669 (fix_recovery_deps): Likewise.
16670 * ira.c (add_store_equivs): Likewise.
16671 (find_moveable_pseudos): Likewise.
16672 (split_live_ranges_for_shrink_wrap): Likewise.
16673 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
16674 (rtx_reuse_manager::seen_def_p): Likewise.
16675 (rtx_reuse_manager::set_seen_def): Likewise.
16676 * print-rtl.h (class rtx_reuse_manager): Likewise.
16678 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16680 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
16682 (migrate_btr_def): Likewise.
16683 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
16684 * df-core.c (loop_post_order_compute): Likewise.
16685 (loop_inverted_post_order_compute): Likewise.
16686 * hsa-common.h: Likewise.
16687 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
16688 * init-regs.c (initialize_uninitialized_regs): Likewise.
16689 * ipa-inline.c (resolve_noninline_speculation): Likewise.
16690 (inline_small_functions): Likewise.
16691 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
16692 * ira.c (combine_and_move_insns): Likewise.
16693 (build_insn_chain): Likewise.
16694 * loop-invariant.c (find_invariants): Likewise.
16695 * lower-subreg.c (propagate_pseudo_copies): Likewise.
16696 * predict.c (tree_predict_by_opcode): Likewise.
16697 (predict_paths_leading_to): Likewise.
16698 (predict_paths_leading_to_edge): Likewise.
16699 (estimate_loops_at_level): Likewise.
16700 (estimate_loops): Likewise.
16701 * shrink-wrap.c (try_shrink_wrapping): Likewise.
16702 (spread_components): Likewise.
16703 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
16704 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
16705 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
16706 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
16707 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
16708 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
16709 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
16710 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
16711 (mark_threaded_blocks): Likewise.
16712 (thread_through_all_blocks): Likewise.
16713 * tree-ssa.c (verify_ssa): Likewise.
16714 (execute_update_addresses_taken): Likewise.
16715 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
16717 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16719 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
16721 (post_order_compute): Likewise.
16722 (inverted_post_order_compute): Likewise.
16723 (pre_and_rev_post_order_compute_fn): Likewise.
16725 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16727 * genrecog.c (int_set::int_set): Explicitly construct our
16728 auto_vec base class.
16729 * vec.h (auto_vec::auto_vec): New constructor.
16731 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16733 * bitmap.h (class auto_bitmap): New constructor taking
16734 bitmap_obstack * argument.
16736 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16738 * bitmap.h (class auto_bitmap): Change type of m_bits to
16739 bitmap_head, and adjust ctor / dtor and member operators.
16741 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
16743 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
16744 when returned register mode doesn't match original mode.
16746 2017-05-12 Jeff Law <law@redhat.com>
16747 Jakub Jelinek <jakub@redhat.com>
16749 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
16750 we look for cc setter after the compare-elim changes.
16751 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
16752 within the vector to match what compare-elim now expects.
16753 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
16754 (xorsi3_flags, one_cmplsi2_flags): Likewise.
16756 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
16757 after the compare-elim changes.
16758 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
16759 the vector to match what compare-elim now expects.
16760 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
16761 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
16762 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
16763 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
16764 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
16766 * config/visium/visium.c (single_set_and_flags): Fix where
16767 we look for cc setter after the compare-elim changes.
16768 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
16769 with the vector to match what compare-elim now expects.
16770 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
16771 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
16772 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
16773 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
16774 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
16775 (neg<mode>2_insn_set_overflow): Likewise.
16777 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
16779 PR middle-end/79794
16780 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
16781 maybe_expand_insn call, set ops[0].target. If still set after call,
16782 set alt_rtl. Add extra arg to recursive calls.
16783 (extract_bit_field): Add alt_rtl argument. Pass to
16785 * expmed.h (extract_bit_field): Fix prototype.
16786 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
16787 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
16788 to extract_bit_field_calls.
16789 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
16790 Pass alt_rtl to extract_bit_field calls.
16791 * calls.c (store_unaligned_arguments_into_psuedos)
16792 load_register_parameters): Pass extra NULL to extract_bit_field calls.
16793 * optabs.c (maybe_legitimize_operand): Clear op->target when call
16795 * optabs.h (struct expand_operand): Add target bitfield.
16797 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
16799 * compare-elim.c (try_eliminate_compare): Canonicalize
16800 operation with embedded compare to
16801 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
16802 (set (reg) (operation)].
16804 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
16806 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
16809 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
16810 cost of adding a carry flag for ADC instruction.
16811 [case MINUS]: Ignore the cost of subtracting a carry flag
16812 for SBB instruction.
16814 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
16816 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
16818 * config/rs6000/bmiintrin.h: New file.
16819 * config/rs6000/bmi2intrin.h: New file.
16820 * config/rs6000/x86intrin.h: New file.
16822 2017-05-12 Jeff Law <law@redhat.com>
16824 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
16827 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
16829 PR middle-end/80707
16830 * tree-cfg.c: Remove cfg edges of unreachable case statements.
16832 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
16834 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
16835 early expansion of vector divide builtins.
16836 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
16837 builtins identified as having unsigned arguments.
16839 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
16841 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
16842 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
16843 expansion of vector logical operations (and, andc, or, xor,
16846 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
16848 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
16849 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
16851 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
16853 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
16854 early GIMPLE expansion of vector multiplies.
16856 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
16858 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
16859 TARGET_HAVE_MOVT conditional.
16860 (movt splitter): Likewise.
16862 2017-05-12 Richard Biener <rguenther@suse.de>
16864 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
16865 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
16866 Fold all stmts not inplace.
16868 2017-05-12 Richard Biener <rguenther@suse.de>
16870 PR tree-optimization/80713
16871 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
16872 inserted_exprs bit for not removed stmts.
16874 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
16876 PR middle-end/69921
16877 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
16878 parallelized" attribute for parallelized OpenACC kernels.
16879 * omp-offload.c (execute_oacc_device_lower): Use it.
16881 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
16882 Set "oacc kernels" attribute.
16883 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
16884 parameter. Adjust all users.
16885 (oacc_fn_attrib_kernels_p): Remove function.
16886 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
16887 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
16888 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
16889 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
16890 assert "oacc kernels" attribute is set.
16892 2017-05-11 Carl Love <cel@us.ibm.com>
16894 * config/rs6000/rs6000-c: Add support for built-in functions
16895 vector unsigned char vec_popcnt (vector signed char)
16896 vector unsigned char vec_popcnt (vector unsigned char)
16897 vector unsigned short vec_popcnt (vector signed short)
16898 vector unsigned short vec_popcnt (vector unsigned short)
16899 vector unsigned int vec_popcnt (vector signed int)
16900 vector unsigned int vec_popcnt (vector unsigned int)
16901 vector unsigned long long vec_popcnt (vector signed long long)
16902 vector unsigned long long vec_popcnt (vector unsigned long long)
16903 vector signed long long vec_slo (vector signed long long,
16904 vector signed char)
16905 vector signed long long vec_slo (vector signed long long,
16906 vector unsigned char)
16907 vector unsigned long long vec_slo (vector unsigned long long,
16908 vector signed char)
16909 vector unsigned long long vec_slo (vector unsigned long long,
16910 vector unsigned char)
16911 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
16912 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
16913 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
16914 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
16915 * doc/extend.texi: Update the built-in documentation file for the
16916 new built-in functions.
16918 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
16920 * attribs.h (sorted_attr_string): Move machine independent
16921 functions for target clone support from the i386 port to common
16922 code. Rename ix86_function_versions to common_function_versions.
16923 Rename make_name to make_unique_name.
16924 (common_function_versions): Likewise.
16925 (make_unique_name): Likewise.
16926 (make_dispatcher_decl): Likewise.
16927 (is_function_default_version): Likewise.
16928 * attribs.c (attr_strcmp): Likewise.
16929 (sorted_attr_string): Likewise.
16930 (common_function_versions): Likewise.
16931 (make_unique_name): Likewise.
16932 (make_dispatcher_decl): Likewise.
16933 (is_function_default_version): Likewise.
16934 * config/i386/i386.c (attr_strcmp): Likewise.
16935 (sorted_attr_string): Likewise.
16936 (ix86_function_versions): Likewise.
16937 (make_name): Likewise.
16938 (make_dispatcher_decl): Likewise.
16939 (is_function_default_version): Likewise.
16940 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
16942 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16945 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
16946 Account for direct move costs for vec_construct of integer
16949 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
16952 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
16953 (UNSPEC_STX_ATOMIC): Ditto.
16954 (loaddi_via_sse): New insn.
16955 (storedi_via_sse): Ditto.
16956 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
16957 Update corresponding peephole2 patterns.
16958 (atomic_storedi_fpu): Ditto.
16960 2017-05-11 Julia Koval <julia.koval@intel.com>
16962 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
16963 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
16965 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
16966 (__builtin_ia32_rsqrt14ss_mask): New builtins.
16967 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
16969 2017-05-11 Nathan Sidwell <nathan@acm.org>
16971 * graphite-poly.c: Include dumpfile.h.
16973 * dumpfle.h (dump_function): Declare here ...
16974 * tree-dump.h (dump_function): ... not here.
16975 * dumpfile.c: #include tree-cfg.h.
16976 (dump_function): Move here from ...
16977 * tree-dump.c (dump_function): ... here.
16978 * gimplify.c: #include splay-tree.h, not tree-dump.h.
16979 * graphite-poly.c: Don't include tree-dump.h.
16980 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
16981 * print-tree.c: Likewise.
16982 * stor-layout.c: Likewise.
16983 * tree-nested.c: Likewise.
16985 * dumpfile.c (dump_start): Use TDF_FLAGS.
16986 (dump_enable_all): Fix TDF_KIND check thinko.
16988 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
16990 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16991 array entries to represent two legal parameterizations of the
16992 overloaded __builtin_cmpb function, as represented by the
16993 P6_OV_BUILTIN_CMPB constant.
16994 (altivec_resolve_overloaded_builtin): Add special case handling
16995 for the __builtin_cmpb function, as represented by the
16996 P6_OV_BUILTIN_CMPB constant.
16997 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
16998 (BU_P6_64BIT_2): New macro.
16999 (BU_P6_OVERLOAD_2): New macro
17000 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
17001 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
17002 (CMPB): Add overload support to represent both 32-bit and 64-bit
17003 compare-bytes function.
17004 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
17005 support for TARGET_CMPB.
17006 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
17007 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
17008 documentation of the __builtin_cmpb overloaded built-in function.
17010 2017-05-11 Richard Biener <rguenther@suse.de>
17012 PR tree-optimization/80705
17013 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
17014 bases are not vectorizable.
17016 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17018 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
17019 when counting register pressure.
17021 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17023 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
17024 (iv_ca_more_deps): Renamed to ...
17025 (iv_ca_compare_deps): ... this.
17026 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
17028 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17030 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
17032 (determine_group_iv_costs): ... here.
17033 (find_inv_vars_cb): Record inv var if it's not recorded before.
17035 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17037 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
17038 (get_shiftadd_cost): Ditto.
17040 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17042 * tree-ssa-address.c: Include header file.
17043 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
17045 (add_to_parts): Refactor.
17046 (addr_to_parts): New parameter. Update use of move_hint_to_base.
17047 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
17050 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17052 PR tree-optimization/53090
17053 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
17055 (extract_cond_operands): Detect condition with IV on both sides
17056 and return COMP_IV_EXPR_2.
17057 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
17058 (rewrite_use_compare): Simplify by removing call to function
17059 extract_cond_operands.
17061 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17063 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
17064 (extract_cond_operands): Detect condition comparing against non-
17065 invariant bound and return appropriate enum value.
17066 (find_interesting_uses_cond): Update use of extract_cond_operands.
17067 Handle its return value accordingly.
17068 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
17070 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17072 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
17073 nonlinear iv_use computation in loop invariant sensitive way.
17075 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17077 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
17078 (find_iv_candidates): Call relate_compare_use_with_all_cands.
17080 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17082 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
17083 (dump_cand): Support iv_cand.inv_exprs.
17084 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
17086 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
17089 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17091 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
17093 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
17094 as local function. Include necessary header files.
17095 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
17097 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17099 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
17101 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17103 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
17104 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
17105 RSHIFT_EXPR and BIT_NOT_EXPR.
17107 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17109 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
17110 (adjust_setup_cost): New parameter supporting round up adjustment.
17111 (struct address_cost_data): Delete.
17112 (force_expr_to_var_cost): Don't bound cost with spill_cost.
17113 (split_address_cost, ptr_difference_cost): Delete.
17114 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
17115 (struct ainc_cost_data): New struct.
17116 (get_address_cost_ainc): New function.
17117 (get_address_cost, get_computation_cost): Reimplement.
17118 (determine_group_iv_cost_address): Record inv_expr for all uses of
17120 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
17121 (iv_ca_has_deps): Reimplemented to ...
17122 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
17124 (iv_ca_extend): Call iv_ca_more_deps.
17126 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17128 * tree-ssa-address.c (struct mem_address): Move to header file.
17129 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
17130 * tree-ssa-address.h (struct mem_address): Move from C file.
17131 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
17133 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17135 * tree-affine.h (aff_combination_type): New interface.
17136 (aff_combination_zero_p): Remove static.
17137 (aff_combination_const_p): New interface.
17138 (aff_combination_singleton_var_p): New interfaces.
17140 2017-05-11 Richard Biener <rguenther@suse.de>
17142 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
17143 Skip unreachable blocks and destinations.
17144 (eliminate): Move stmt removal and fixup ...
17145 (fini_eliminate): ... here. Skip inserted exprs.
17146 (pass_pre::execute): Move fini_pre after fini_eliminate.
17147 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
17148 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
17149 PRE to get rid of dead code that has invalid SSA form and
17150 split critical edges again.
17152 2017-05-11 Bin Cheng <bin.cheng@arm.com>
17154 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
17156 2017-05-11 Richard Biener <rguenther@suse.de>
17158 * passes.c (execute_function_todo): Verify loops if they are
17159 said to be up-to-date.
17160 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
17161 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
17163 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
17166 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
17167 handle calling assemble_external ourself.
17170 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
17171 modes with zero size. Enhance comment.
17173 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17175 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
17176 built-ins for vec_xl and vec_xst with short and char pointer
17179 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
17181 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
17182 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
17183 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
17184 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
17185 (_mm_maskz_min_round_ss): New intrinsics.
17186 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
17187 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
17188 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
17189 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
17190 (__builtin_ia32_minss_mask_round): New builtins.
17191 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
17192 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
17193 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
17195 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
17196 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
17198 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
17201 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
17203 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
17204 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
17205 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
17206 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
17207 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
17208 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
17209 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
17210 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
17211 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
17212 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
17213 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
17214 (__builtin_ia32_mulss_mask_round): New builtins.
17215 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
17216 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
17217 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
17219 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
17220 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
17222 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
17225 2017-05-10 Julia Koval <julia.koval@intel.com>
17227 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
17228 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
17229 (_mm256_setr_m128i): New intrinsics.
17231 2017-05-10 Julia Koval <julia.koval@intel.com>
17233 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
17234 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
17235 (_mm_maskz_rcp14_ss): New intrinsics.
17236 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
17237 (__builtin_ia32_rcp14ss_mask): New builtins.
17238 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
17240 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
17242 PR tree-optimization/51513
17243 * tree-cfg.c (gimple_seq_unreachable_p): New function.
17244 (assert_unreachable_fallthru_edge_p): Use it.
17245 (group_case_labels_stmt): Likewise.
17246 * tree-cfg.h: Prototype it.
17247 * stmt.c: Include cfghooks.h and tree-cfg.h.
17248 (emit_case_dispatch_table) <gap_label>: New local variable.
17249 Use it to fill dispatch table gaps.
17250 Test for default_label before updating probabilities.
17251 (expand_case) <default_label>: Remove unneeded initialization.
17252 Test for unreachable default case statement and remove its edge.
17253 Set default_label accordingly.
17254 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
17256 2017-05-10 Carl Love <cel@us.ibm.com>
17258 * config/rs6000/rs6000-c: Add support for built-in functions
17259 vector signed char vec_neg (vector signed char)
17260 vector signed short int vec_neg (vector short int)
17261 vector signed int vec_neg (vector signed int)
17262 vector signed long long vec_neg (vector signed long long)
17263 vector float vec_neg (vector float)
17264 vector double vec_neg (vector double)
17265 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
17267 * config/rs6000/altivec.h: Add define for vec_neg
17268 * doc/extend.texi: Update the built-in documentation for the
17269 new built-in functions.
17271 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17273 PR tree-optimization/77644
17274 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
17276 2017-05-10 Nathan Sidwell <nathan@acm.org>
17278 * dumpfile.h (TDI_lang_all): New.
17279 (TDF_KIND): New. Renumber others
17280 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
17282 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
17284 (get_dump_file_name): Adjust suffix generation.
17285 (dump_enable_all): Use TDF_KIND.
17286 * doc/invoke.texi (-fdump-lang-all): Document.
17288 * dumpfile.h: Tabify.
17290 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
17293 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
17294 Move member access before delete.
17296 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
17298 * tree-inline.c (expand_call_inline): Split block at stmt
17301 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
17304 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
17305 are now unused after splitting mov{sf,sd}_hardfloat.
17306 (f32_lr2): Likewise.
17307 (f32_lm): Likewise.
17308 (f32_lm2): Likewise.
17309 (f32_li): Likewise.
17310 (f32_li2): Likewise.
17311 (f32_lv): Likewise.
17312 (f32_sr): Likewise.
17313 (f32_sr2): Likewise.
17314 (f32_sm): Likewise.
17315 (f32_sm2): Likewise.
17316 (f32_si): Likewise.
17317 (f32_si2): Likewise.
17318 (f32_sv): Likewise.
17319 (f32_dm): Likewise.
17320 (f32_vsx): Likewise.
17321 (f32_av): Likewise.
17322 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
17323 For movsf, order stores so the VSX stores occur before the GPR
17324 store which encourages the register allocator to use a traditional
17325 FPR instead of a GPR. For movsd, order the stores so that the GPR
17326 store comes before the VSX stores to allow the power6 to work.
17327 This is due to the power6 not having a 32-bit integer store
17328 instruction from a FPR.
17329 (movsf_hardfloat): Likewise.
17330 (movsd_hardfloat): Likewise.
17332 2017-05-09 Martin Sebor <msebor@redhat.com>
17334 PR translation/80280
17335 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
17338 PR translation/80280
17339 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
17340 data member added in r247778.
17341 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
17343 2017-05-09 Nathan Sidwell <nathan@acm.org>
17345 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
17347 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
17350 2017-05-09 Marek Polacek <polacek@redhat.com>
17352 * doc/invoke.texi: Fix typo.
17354 2017-05-09 Richard Biener <rguenther@suse.de>
17356 * tree-vrp.c (vrp_val_is_max): Adjust comment.
17357 (vrp_val_is_min): Likewise.
17358 (set_value_range_to_value): Likewise.
17359 (set_value_range_to_nonnegative): Likewise.
17360 (gimple_assign_nonzero_p): Likewise.
17361 (gimple_stmt_nonzero_p): Likewise.
17362 (vrp_int_const_binop): Likewise. Remove unreachable case.
17363 (adjust_range_with_scev): Adjust comments.
17364 (compare_range_with_value): Likewise.
17365 (extract_range_from_phi_node): Likewise.
17366 (test_for_singularity): Likewise.
17368 2017-05-09 Richard Biener <rguenther@suse.de>
17370 * tree-vrp.c (get_single_symbol): Add assert that we don't
17371 get overflowed constants as invariant part.
17372 (compare_values_warnv): Add comment before the TREE_NO_WARNING
17373 checks. Use wi::cmp instead of recursing for integer constants.
17374 (compare_values): Just ignore whether we assumed undefined
17375 overflow instead of failing the compare.
17376 (extract_range_for_var_from_comparison_expr): Add comment before the
17377 TREE_NO_WARNING sets.
17378 (test_for_singularity): Likewise.
17379 (extract_range_from_comparison): Do not disable optimization
17380 when we assumed undefined overflow.
17381 (extract_range_basic): Remove init of unused var.
17383 2017-05-09 Richard Biener <rguenther@suse.de>
17385 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
17386 (extract_range_from_multiplicative_op_1): Adjust.
17387 (extract_range_from_binary_expr_1): Use int_const_binop.
17389 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
17392 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
17393 rs6000_store_data_bypass_p in seven define_bypass directives and
17394 in several comments.
17395 * config/rs6000/rs6000-protos.h: Add prototype for
17396 rs6000_store_data_bypass_p function.
17397 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
17398 function implements slightly different (rs6000-specific) semantics
17399 than store_data_bypass_p, returning false rather than aborting
17400 with assertion error when arguments do not satisfy the
17401 requirements of store data bypass.
17402 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
17403 rs6000_store_data_bypass_p.
17405 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
17407 * config/xtensa/xtensa-protos.h
17408 (xtensa_initial_elimination_offset): New declaration.
17409 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
17410 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
17411 macro definition, add case for FRAME_POINTER_REGNUM when
17412 FRAME_GROWS_DOWNWARD.
17413 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
17414 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
17415 xtensa_initial_elimination_offset.
17417 2017-05-08 Nathan Sidwell <nathan@acm.org>
17419 * doc/invoke.texi: Alphabetize -fdump options.
17421 2017-05-08 Martin Sebor <msebor@redhat.com>
17423 PR translation/80280
17424 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
17426 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
17428 * target.def (compute_frame_layout): New optional target hook.
17429 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
17430 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
17431 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
17433 * reload1.c (verify_initial_elim_offsets): Likewise.
17434 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
17435 (use_simple_return_p): Call arm_compute_frame_layout if needed.
17436 (arm_get_frame_offsets): Split up into this ...
17437 (arm_compute_frame_layout): ... and this function.
17439 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
17441 * config/aarch64/constraints.md (Usa): New constraint.
17442 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
17444 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
17446 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
17447 with_multilib_list after it has been checked.
17449 2017-05-08 Richard Biener <rguenther@suse.de>
17451 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
17452 (bitmap_set_subtract_values): Likewise.
17454 2017-05-08 Richard Biener <rguenther@suse.de>
17456 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
17457 (gimple_assign_nonzero): ... this and remove strict_overflow_p
17459 (gimple_stmt_nonzero_warnv_p): Rename to ...
17460 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
17462 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
17463 (extract_range_basic): Adjust, do not disable propagation on
17464 strict overflow sensitive simplification.
17465 (vrp_visit_cond_stmt): Likewise.
17467 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
17469 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
17470 body size unconditionally.
17472 2017-05-07 Jeff Law <law@redhat.com>
17475 2017-05-06 Jeff Law <law@redhat.com>
17476 PR tree-optimization/78496
17477 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
17480 PR tree-optimization/78496
17481 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
17482 (simplify_stmt_using_ranges): Call it.
17483 (vrp_dom_walker::before_dom_children): Extract equivalences
17484 from an ASSERT_EXPR with an equality comparison against a
17487 2017-05-06 Jeff Law <law@redhat.com>
17489 PR tree-optimization/78496
17490 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
17493 PR tree-optimization/78496
17494 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
17495 (simplify_stmt_using_ranges): Call it.
17496 (vrp_dom_walker::before_dom_children): Extract equivalences
17497 from an ASSERT_EXPR with an equality comparison against a
17500 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
17502 * lra-constraints.c (lra_copy_reg_equiv): New function.
17503 (split_reg): Use it to copy equivalence information from the
17504 original register to the spill register.
17506 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
17508 PR rtl-optimization/75964
17509 * simplify-rtx.c (simplify_const_relational_operation): Remove
17510 invalid handling of comparisons of integer ABS.
17512 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
17514 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
17515 initialize to zero.
17516 (init_regs): Remove declaration.
17517 (function_arg_advance_32): Initialize error_p as boolean variable.
17519 2017-05-05 Nathan Sidwell <nathan@acm.org>
17521 * store-motion.c (remove_reachable_equiv_notes): Reformat long
17522 lines. Use for (;;).
17524 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17526 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
17527 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
17528 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
17529 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
17530 VF=2 that require versioning.
17532 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17534 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
17537 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17539 * diagnostic.h (diagnostic_override_option_index): Convert from
17540 macro to inline function.
17542 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17544 * diagnostic.c (last_module_changed_p): New function.
17545 (set_last_module): New function.
17546 (diagnostic_report_current_module): Convert macro usage to
17547 the above functions.
17548 * diagnostic.h (diagnostic_context::last_module): Strengthen
17549 from const line_map * to const line_map_ordinary *.
17550 (diagnostic_last_module_changed): Delete macro.
17551 (diagnostic_set_last_module): Delete macro.
17553 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17555 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
17556 with diagnostic_report_diagnostic.
17557 (diagnostic_n_impl_richloc): Likewise.
17558 * diagnostic.h (report_diagnostic): Delete macro.
17559 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
17560 with diagnostic_report_diagnostic.
17561 * substring-locations.c (format_warning_va): Likewise.
17563 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17565 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
17566 save/restor of format_spec. Move option-printing code to...
17567 (print_option_information): ...this new function, and
17568 reimplement by simply printing to the pretty_printer,
17569 rather than appending to the format string.
17571 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17573 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
17574 handling logic into...
17575 (update_effective_level_from_pragmas): ...this new function.
17577 2017-05-04 Andrew Waterman <andrew@sifive.com>
17579 * config/riscv/riscv.opt (mstrict-align): New option.
17580 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
17581 (SLOW_UNALIGNED_ACCESS): Define.
17582 (riscv_slow_unaligned_access): Declare.
17583 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
17585 (riscv_slow_unaligned_access): New variable.
17586 (rocket_tune_info): Set slow_unaligned_access to true.
17587 (optimize_size_tune_info): Set slow_unaligned_access to false.
17588 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
17589 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
17590 (riscv_option_override): Set riscv_slow_unaligned_access.
17591 * doc/invoke.texi: Add -mstrict-align to RISC-V.
17593 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
17595 * config/riscv/riscv.md: Unify indentation.
17597 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
17602 * config/rs6000/rs6000.md (u code attribute): Add FIX and
17604 (extendsi<mode>2): Add support for doing sign extension via
17605 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
17606 don't have ISA 3.0 instructions.
17607 (extendsi<mode>2 splitter): Likewise.
17608 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
17609 generate the normal insns since SImode can now go in vector
17610 registers. Disallow the special UNSPECs needed for previous
17611 machines to hide SImode being used. Add new insns
17612 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
17613 (fix_trunc<mode>si2_stfiwx): Likewise.
17614 (fix_trunc<mode>si2_internal): Likewise.
17615 (fixuns_trunc<mode>si2): Likewise.
17616 (fixuns_trunc<mode>si2_stfiwx): Likewise.
17617 (fctiw<u>z_<mode>_smallint): Likewise.
17618 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
17619 of floating point to 32-bit integer from doing a direct move to
17620 the GPR registers to do a store.
17621 (fctiwz_<mode>): Break long line.
17623 2017-05-05 Bin Cheng <bin.cheng@arm.com>
17625 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
17626 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
17627 (addr_list, addr_offset_valid_p): New.
17628 (split_address_groups): Check offset validity with above function.
17629 (gt-tree-ssa-loop-ivopts.h): Include header file.
17631 2017-05-05 Nathan Sidwell <nathan@acm.org>
17633 * config.gcc (arm*-*-*): Add missing 'fi'.
17635 2017-05-05 Steve Ellcey <sellcey@cavium.com>
17637 * doc/invoke.texi (-fopt-info): Explicitly say order of options
17638 included in -fopt-info does not matter.
17639 * doc/optinfo.texi (-fopt-info): Fix description of default
17640 behavour. Explicitly say order of options included in -fopt-info
17643 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
17645 * config.gcc: Allow combinations of aprofile and rmprofile values for
17646 --with-multilib-list.
17647 * config/arm/t-multilib: New file.
17648 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
17649 variables. Remove setting of ISA and floating-point ABI in
17650 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
17651 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
17652 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
17653 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
17655 * config/arm/t-rmprofile: Likewise except for the matches changes.
17656 * doc/install.texi (--with-multilib-list): Document the combination of
17657 aprofile and rmprofile values and warn about pitfalls in doing that.
17659 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
17661 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
17662 (movdi_aarch64): Likewise.
17664 2017-05-05 Jakub Jelinek <jakub@redhat.com>
17666 PR tree-optimization/80632
17667 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
17669 (build_arrays): Initialize it for virtual phis.
17670 (fix_phi_nodes): Use it for virtual phis.
17672 PR tree-optimization/80558
17673 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
17674 [x, y] op z into [x op, y op z] for op & or | if conditions
17677 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
17678 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
17681 * config/arm/arm.md (use_literal_pool): Remove.
17682 (64-bit immediate split): No longer takes cost into consideration
17683 if arm_disable_literal_pool is enabled.
17684 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
17685 used when arm_disable_literal_pool is enabled.
17686 (arm_max_const_double_inline_cost): Remove use of
17687 arm_disable_literal_pool.
17688 (push_minipool_fix): Add assert.
17689 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
17690 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
17691 (no_literal_pool_sf_immediate): New.
17693 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17695 PR tree-optimization/80613
17696 * tree-ssa-dce.c (propagate_necessity): Remove cases for
17697 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
17699 2017-05-05 Richard Biener <rguenther@suse.de>
17701 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
17703 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
17705 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
17706 of this flag from insn conditions due to removal from r247495.
17708 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
17710 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
17712 (arm_early_store_addr_dep_ptr): Likewise.
17713 * config/arm/aarch-common-protos.h
17714 (arm_early_load_addr_dep_ptr): Add prototype.
17715 (arm_early_store_addr_dep_ptr): Likewise.
17716 * config/arm/cortex-a53.md: Add new bypasses.
17718 2017-05-05 Jakub Jelinek <jakub@redhat.com>
17720 * tree.c (next_type_uid): Change type to unsigned.
17721 (type_hash_canon): Decrement back next_type_uid if
17722 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
17723 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
17726 2017-05-04 Martin Sebor <msebor@redhat.com>
17728 * builtins.c: Fix a trivial typo in a comment.
17730 PR middle-end/79234
17731 * builtins.c (check_sizes): Adjust to handle reading past the end.
17732 Avoid printing excessive upper bound of ranges. Use %E to print
17733 tree nodes instead of converting them to %wu.
17734 (expand_builtin_memchr): New function.
17735 (compute_dest_size): Rename...
17736 (compute_objsize): ...to this.
17737 (expand_builtin_memcpy): Adjust.
17738 (expand_builtin_mempcpy): Adjust.
17739 (expand_builtin_strcat): Adjust.
17740 (expand_builtin_strcpy): Adjust.
17741 (check_strncat_sizes): Adjust.
17742 (expand_builtin_strncat): Adjust.
17743 (expand_builtin_strncpy): Adjust and simplify.
17744 (expand_builtin_memset): Adjust.
17745 (expand_builtin_bzero): Adjust.
17746 (expand_builtin_memcmp): Adjust.
17747 (expand_builtin): Handle memcmp.
17748 (maybe_emit_chk_warning): Check strncat just once.
17750 2017-05-04 Martin Sebor <msebor@redhat.com>
17752 PR preprocessor/79214
17753 PR middle-end/79222
17754 PR middle-end/79223
17755 * builtins.c (check_sizes): Add inlining context and issue
17756 warnings even when -Wno-system-headers is set.
17757 (check_strncat_sizes): Same.
17758 (expand_builtin_strncat): Same.
17759 (expand_builtin_memmove): New function.
17760 (expand_builtin_stpncpy): Same.
17761 (expand_builtin): Handle memmove and stpncpy.
17763 2017-05-04 Bin Cheng <bin.cheng@arm.com>
17765 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
17766 which is not used any more.
17768 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
17770 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
17772 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
17774 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
17775 (cortexa53_tunings): Likewise.
17776 (cortexa57_tunings): Likewise.
17777 (cortexa72_tunings): Likewise.
17778 (cortexa73_tunings): Likewise.
17780 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
17782 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
17783 Set loop alignment to 8.
17785 2017-05-04 Martin Sebor <msebor@redhat.com>
17787 PR translation/80280
17788 * builtins.c (expand_builtin_object_size): Add missing quoting to
17789 %D and like directives.
17790 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
17791 (hsa_type_for_tree_type): Same.
17792 (verify_function_arguments): Same.
17793 * symtab.c (symbol_table::change_decl_assembler_name): Same.
17794 * varasm.c (get_section): Same.
17797 2017-05-04 Martin Sebor <msebor@redhat.com>
17799 PR translation/80280
17800 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
17802 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
17804 * config/aarch64/aarch64.c (generic_addrcost_table):
17805 Change HI/TI mode setting.
17807 2017-05-04 Martin Jambor <mjambor@suse.cz>
17809 PR tree-optimization/80622
17810 * tree-sra.c (comes_initialized_p): New function.
17811 (build_accesses_from_assign): Only set write lazily when
17812 comes_initialized_p is false.
17813 (analyze_access_subtree): Use comes_initialized_p.
17814 (propagate_subaccesses_across_link): Assert !comes_initialized_p
17815 instead of testing for PARM_DECL.
17817 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17819 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
17820 constraint on operand 0 to allow more general addressing modes.
17821 Adjust output template.
17822 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
17824 * config/aarch64/aarch64-protos.h
17825 (aarch64_address_valid_for_prefetch_p): Declare prototype.
17826 * config/aarch64/constraints.md (Dp): New address constraint.
17827 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
17830 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
17832 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
17833 update use of estimate_ipcp_clone_size_and_time.
17834 (estimate_local_effects): Update use of
17835 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
17836 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
17837 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
17838 Return nonspecialized time.
17840 2017-05-04 Richard Biener <rguenther@suse.de>
17842 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
17843 for the last VUSE which def dominates the PHI. Directly call
17845 (get_continuation_for_phi_1): Remove.
17847 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
17849 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
17850 to explain the use of truncating division. Cap the number of
17851 iterations to the maximum given by nb_iterations_upper_bound,
17854 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
17856 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
17857 * configure: Regenerate.
17858 * config.in: Regenerate.
17859 * config/i386/driver-mingw32.c: new file.
17860 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
17861 * config.host: Link driver-mingw32.o on MinGW host.
17862 * doc/install.texi: Document new --enable-mingw-wildcard configure
17865 2017-05-04 Marek Polacek <polacek@redhat.com>
17867 PR tree-optimization/80612
17868 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
17870 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
17871 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
17873 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
17874 (movt splitter): Likewise.
17875 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
17876 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
17877 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
17878 block for Thumb-1 with MOVT.
17879 (thumb2_legitimate_address_p): Move code block ...
17880 (can_avoid_literal_pool_for_label_p): ... into this new function.
17881 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
17883 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
17884 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
17885 "M-profile targets with the MOVT instruction".
17887 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
17889 * config/arm/arm-builtins.c (arm_init_builtins): Rename
17890 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
17891 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
17893 2017-05-04 Martin Liska <mliska@suse.cz>
17895 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
17896 variable cond_code.
17898 2017-05-04 Richard Biener <rguenther@suse.de>
17900 * tree.c (array_at_struct_end_p): Handle arrays at struct
17901 end with flexarrays more conservatively. Refactor and treat
17902 arrays of arrays or aggregates more strict. Fix
17903 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
17904 * tree.c (array_at_struct_end_p): Adjust prototype.
17905 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
17906 * gimple-fold.c (get_range_strlen): Likewise.
17907 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
17909 2017-05-04 Richard Biener <rguenther@suse.de>
17911 PR tree-optimization/31130
17912 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
17914 (supports_overflow_infinity): Likewise.
17915 (is_negative_overflow_infinity): Likewise.
17916 (is_positive_overflow_infinity): Likewise.
17917 (is_overflow_infinity): Likewise.
17918 (stmt_overflow_infinity): Likewise.
17919 (overflow_infinity_range_p): Likewise.
17920 (usable_range_p): Remove as always returning true.
17921 (make_overflow_infinity): Remove.
17922 (negative_overflow_infinity): Likewise.
17923 (positive_overflow_infinity): Likewise.
17924 (avoid_overflow_infinity): Likewise.
17925 (set_value_range): Adjust accordingly.
17926 (set_value_range_to_nonnegative): Likewise, remove now unused
17927 overflow_infinity arg.
17928 (vrp_operand_equal_p): Adjust.
17929 (update_value_range): Likewise.
17930 (range_int_cst_singleton_p): Likewise.
17931 (operand_less_p): Likewise.
17932 (compare_values_warnv): Likewise.
17933 (extract_range_for_var_from_comparison_expr): Likewise.
17934 (vrp_int_const_binop): Likewise.
17935 (zero_nonzero_bits_from_vr): Likewise.
17936 (extract_range_from_multiplicative_op_1): Likewise.
17937 (extract_range_from_binary_expr_1): Likewise.
17938 (extract_range_from_unary_expr): Likewise.
17939 (extract_range_from_comparison): Likewise.
17940 (extract_range_basic): Likewise.
17941 (adjust_range_with_scev): Likewise.
17942 (compare_ranges): Likewise.
17943 (compare_range_with_value): Likewise.
17944 (dump_value_range): Likewise.
17945 (test_for_singularity): Likewise, remove strict_overflow_p parameter
17947 (simplify_cond_using_ranges): Adjust.
17949 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
17951 * brig-builtins.def: Added a builtin for class_f64.
17952 * builtin-types.def: Added a builtin type needed by class_f64.
17954 2017-05-03 Jason Merrill <jason@redhat.com>
17956 * timevar.def: Add TV_CONSTEXPR.
17958 2017-05-03 David Malcolm <dmalcolm@redhat.com>
17960 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
17962 2017-05-03 Martin Jambor <mjambor@suse.cz>
17964 * ipa-prop.c (ipa_update_after_lto_read): Removed.
17965 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
17966 * ipa-cp.c (ipcp_propagate_stage): Do not call
17967 ipa_update_after_lto_read.
17968 * ipa-inline.c (ipa_inline): Likewise.
17970 2017-05-03 Martin Jambor <mjambor@suse.cz>
17972 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
17973 tag. Added a default constructor and a destructor.
17974 (ipa_edge_args_sum_t): New class;
17975 (ipa_edge_args_sum): Declare.
17976 (ipa_edge_args_vector): Remove declaration.
17977 (IPA_EDGE_REF): Use ipa_edge_args_sum.
17978 (ipa_free_edge_args_substructures): Remove declaration.
17979 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
17980 (ipa_edge_args_info_available_for_edge_p): Likewise.
17981 * ipa-prop.c (ipa_edge_args_vector): Removed.
17982 (edge_removal_hook_holder): Likewise.
17983 (edge_duplication_hook_holder): Likewise.
17984 (ipa_edge_args_sum): New variable.
17985 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
17986 ipa_edge_args_vector.
17987 (ipa_free_edge_args_substructures): Likewise.
17988 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
17989 ipa_edge_args_vector.
17990 (ipa_edge_removal_hook): Turned into method
17991 ipa_edge_args_sum_t::remove.
17992 (ipa_edge_duplication_hook): Turned into method
17993 ipa_edge_args_sum_t::duplicate.
17994 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
17995 registering edge hooks.
17996 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
17997 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
17998 ipa_edge_args_sum instead of ipa_edge_args_vector.
17999 * ipa-profile.c (ipa_profile): Likewise.
18001 2017-05-03 Martin Jambor <mjambor@suse.cz>
18003 * symbol-summary.h (function_summary): New method exists.
18004 (function_summary::symtab_removal): Deallocate through release.
18005 (call_summary): New class.
18006 (gt_ggc_mx): New overload.
18007 (gt_pch_nx): Likewise.
18008 (gt_pch_nx): Likewise.
18010 2017-05-03 Jeff Law <law@redhat.com>
18012 PR tree-optimization/78496
18013 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
18014 from simplify_cond_using_ranges. Split off code to walk
18015 backwards through casts into ...
18016 (simplify_cond_using_ranges_2): New function.
18017 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
18018 (execute_vrp): After identifying jump threads, call
18019 simplify_cond_using_ranges_2.
18021 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
18024 * ipa-inline.h (inline_summary): Add ctor.
18025 (create_ggc): Do not use ggc_cleared_alloc.
18027 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
18028 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18030 * gcc.c (handle_braces): Support escaping in switch matching
18032 * doc/invoke.texi (Spec Files): Document it.
18033 Remove superfluous @code markup in items.
18035 2017-05-03 David Malcolm <dmalcolm@redhat.com>
18037 * diagnostic-show-locus.c (struct column_range): New struct.
18038 (get_affected_columns): New function.
18039 (get_printed_columns): New function.
18040 (struct correction): New struct.
18041 (correction::ensure_capacity): New function.
18042 (correction::ensure_terminated): New function.
18043 (struct line_corrections): New struct.
18044 (line_corrections::~line_corrections): New dtor.
18045 (line_corrections::add_hint): New function.
18046 (layout::print_trailing_fixits): Reimplement in terms of the new
18048 (selftest::test_overlapped_fixit_printing): New function.
18049 (selftest::diagnostic_show_locus_c_tests): Call it.
18051 2017-05-03 Nathan Sidwell <nathan@acm.org>
18053 Canonicalize canonical type hashing
18054 * tree.h (type_hash_canon_hash): Declare.
18055 * tree.c (type_hash_list, attribute_hash_list): Move into
18056 type_hash_canon_hash.
18057 (build_type_attribute_qual_variant): Break out hash code calc into
18058 type_hash_canon_hash.
18059 (type_hash_canon_hash): New. Generic type hash computation.
18060 (build_range_type_1, build_array_type_1, build_function_type,
18061 build_method_type_directly, build_offset_type, build_complex_type,
18062 make_vector_type): Call it.
18064 2017-05-03 Richard Biener <rguenther@suse.de>
18066 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
18067 When all DRs have unknown misaligned do not always peel
18068 when there is a store but apply the same costing model as if
18069 there were only loads.
18071 2017-05-03 Richard Biener <rguenther@suse.de>
18074 PR tree-optimization/80492
18075 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
18076 compare_base_decls returning dont-know properly.
18078 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
18080 * config/arm/iterators.md (CCSI): New mode iterator.
18081 (arch): New mode attribute.
18082 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
18083 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
18084 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
18085 code iterator for success result mode.
18086 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
18087 the corresponding new insn generators.
18089 2017-05-03 Bin Cheng <bin.cheng@arm.com>
18092 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18093 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
18095 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
18097 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
18098 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
18099 (DDR_A): Wrap DDR argument in brackets.
18100 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
18101 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
18102 (DDR_REVERSED_P): Likewise.
18104 2017-05-03 Jakub Jelinek <jakub@redhat.com>
18106 PR tree-optimization/79472
18107 * tree-switch-conversion.c (struct switch_conv_info): Add
18108 contiguous_range and default_case_nonstandard fields.
18109 (collect_switch_conv_info): Compute contiguous_range and
18110 default_case_nonstandard fields, don't clear final_bb if
18111 contiguous_range and only the default case doesn't have the required
18113 (check_all_empty_except_final): Set default_case_nonstandard instead
18114 of failing if contiguous_range and the default case doesn't have empty
18116 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
18117 and only the default case doesn't have the required constants. Skip
18119 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
18120 if default_case_nonstandard.
18121 (build_constructors): Build constant 1 just once. Assert that default
18122 values aren't inserted in between cases if contiguous_range. Skip
18124 (build_arrays): Skip virtual phis.
18125 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
18126 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
18127 Handle virtual phis.
18128 (gen_inbound_check): Handle default_case_nonstandard case.
18129 (process_switch): Adjust check_final_bb caller. Call
18130 gather_default_values with the first non-default case instead of
18131 default case if default_case_nonstandard.
18133 2017-05-02 Nathan Sidwell <nathan@acm.org>
18135 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
18136 check. Fix formatting.
18138 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
18140 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
18141 errors when comparing specialized and unspecialized times.
18143 2017-05-02 David Malcolm <dmalcolm@redhat.com>
18145 * diagnostic-show-locus.c
18146 (layout::should_print_annotation_line_p): Make private.
18147 (layout::print_annotation_line): Make private.
18148 (layout::annotation_line_showed_range_p): Make private.
18149 (layout::show_ruler): Make private.
18150 (layout::print_source_line): Make private. Pass in line and
18151 line_width, rather than calling location_get_source_line. Drop
18153 (layout::print_leading_fixits): New method.
18154 (layout::print_any_fixits): Rename to...
18155 (layout::print_trailing_fixits): ...this, and make private.
18156 Don't print newline fixits.
18157 (diagnostic_show_locus): Move logic for printing one row into...
18158 (layout::print_line): ...this new function. Move the
18159 location_get_source_line call and error-handling from
18160 print_source_line to here. Call print_leading_fixits, and rename
18161 print_any_fixits to print_trailing_fixits.
18162 (selftest::test_fixit_insert_containing_newline): Update now that
18163 newlines are partially supported.
18164 (selftest::test_fixit_insert_containing_newline_2): New test.
18165 (selftest::test_fixit_replace_containing_newline): Update comments.
18166 (selftest::diagnostic_show_locus_c_tests): Call the new test.
18167 * edit-context.c (class added_line): New class.
18168 (class edited_line): Describe newline handling in comment.
18169 (edited_line::actually_edited_p): New method.
18170 (edited_line::print_content): Delete redundant decl.
18171 (edited_line::m_predecessors): New field.
18172 (edited_file::print_content): Call edited_line::print_content.
18173 (edited_file::print_diff): Update to support newlines.
18174 (edited_file::print_diff_hunk): Likewise.
18175 (edited_file::print_run_of_changed_lines): New function.
18176 (edited_file::print_diff_line): Convert to...
18177 (print_diff_line): ...this.
18178 (edited_file::get_effective_line_count): New function.
18179 (edited_line::edited_line): Initialize new field m_predecessors.
18180 (edited_line::~edited_line): Clean up m_predecessors.
18181 (edited_line::apply_fixit): Handle newlines.
18182 (edited_line::get_effective_line_count): New function.
18183 (edited_line::print_content): New function.
18184 (edited_line::print_diff_lines): New function.
18185 (selftest::test_applying_fixits_insert_containing_newline): New
18187 (selftest::test_applying_fixits_replace_containing_newline): New
18189 (selftest::insert_line): New function.
18190 (selftest::test_applying_fixits_multiple_lines): Add example of
18192 (selftest::edit_context_c_tests): Call the new tests.
18194 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18196 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
18197 parameter cand. Update dump information.
18198 (get_computation_cost): Update uses.
18200 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18202 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
18203 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
18204 (get_computation_at, rewrite_use_address): Update use of
18205 get_computation_aff.
18207 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18209 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
18210 (get_computation): Delete.
18211 (get_computation_cost): Implement like get_computation_cost_at.
18212 Use get_computation_at.
18213 (get_computation_cost_at): Delete.
18214 (rewrite_use_nonlinear_expr): Use get_computation_at.
18215 (rewrite_use_compare, remove_unused_ivs): Ditto.
18217 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18219 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
18221 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18223 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
18224 (ivopts_global_cost_for_size): Rename parameter and update uses.
18225 (iv_ca_recount_cost): Update uses.
18226 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
18227 candidates seperately in n_invs and n_cands.
18228 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
18230 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18232 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
18233 (find_inv_vars_cb): New.
18234 (find_depends): Renamed to ...
18235 (find_inv_vars): ... this.
18236 (add_candidate_1, force_var_cost): Call find_inv_vars.
18237 (split_address_cost, determine_group_iv_cost_cond): Ditto.
18239 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18241 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
18242 inv_vars. Add inv_exprs.
18243 (struct iv_cand): Rename depends_on to inv_vars.
18244 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
18245 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
18246 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
18247 (dump_cand): Dump inv_vars.
18248 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
18249 (record_invariant, find_depends, add_candidate_1): Ditto.
18250 (set_group_iv_cost, force_var_cost): Ditto.
18251 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
18252 (get_computation_cost_at, get_computation_cost): Ditto.
18253 (determine_group_iv_cost_generic): Ditto.
18254 (determine_group_iv_cost_address): Ditto.
18255 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
18256 (determine_group_iv_costs): Ditto.
18257 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
18258 (iv_ca_set_remove_invariants): Renamed to ...
18259 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
18260 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
18261 (iv_ca_set_add_invariants): Renamed to ...
18262 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
18263 (iv_ca_set_cp): Use iv_ca_set_add_invs.
18264 (iv_ca_has_deps): Support inv_vars and inv_exprs.
18265 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
18266 (create_new_ivs): Remove useless dump.
18268 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18270 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
18272 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
18273 (iv_ca_set_no_cp, create_new_iv): Ditto.
18275 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18277 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
18279 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
18281 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
18282 function tree_check2.
18284 2017-05-02 Martin Liska <mliska@suse.cz>
18286 * doc/gcov.texi: Add missing preposition.
18287 * gcov.c (function_info::function_info): Properly fill up
18288 all member variables.
18290 2017-05-02 Tamar Christina <tamar.christina@arm.com>
18292 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
18294 2017-05-02 Tamar Christina <tamar.christina@arm.com>
18296 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
18298 2017-05-02 Martin Liska <mliska@suse.cz>
18301 * lto-streamer-in.c (lto_read_tree_1): Remove
18302 LTO_STREAMER_DEBUG.
18303 * lto-streamer.c (struct tree_hash_entry): Likewise.
18304 (struct tree_entry_hasher): Likewise.
18305 (tree_entry_hasher::hash): Likewise.
18306 (tree_entry_hasher::equal): Likewise.
18307 (lto_streamer_init): Likewise.
18308 (lto_orig_address_map): Likewise.
18309 (lto_orig_address_get): Likewise.
18310 (lto_orig_address_remove): Likewise.
18311 * lto-streamer.h: Likewise.
18312 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
18313 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
18315 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
18317 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
18318 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
18319 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
18320 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
18321 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
18322 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
18323 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
18324 (mm_maskz_sub_ss): New intrinsics.
18325 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
18326 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
18327 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
18328 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
18329 (__builtin_ia32_subss_mask_round): New builtins.
18330 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
18331 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
18332 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
18334 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
18335 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
18337 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
18340 2017-05-02 Martin Jambor <mjambor@suse.cz>
18342 PR tree-optimization/78687
18343 * tree-sra.c (access): New field parent.
18344 (process_subtree_disqualification): New function.
18345 (disqualify_candidate): Call it.
18346 (build_accesses_from_assign): Reset write flag if creating an
18348 (build_access_subtree): Fill in parent field and also prpagate
18349 down grp_write flag.
18350 (create_artificial_child_access): New parameter set_grp_write, set
18351 grp_write to its value.
18352 (propagate_subaccesses_across_link): Also propagate grp_write flag
18354 (propagate_all_subaccesses): Push the closest parent back to work
18355 queue if add_access_to_work_queue returned true.
18357 2017-05-02 Richard Biener <rguenther@suse.de>
18359 * common.opt (fstrict-overflow): Alias negative to fwrapv.
18360 * doc/invoke.texi (fstrict-overflow): Remove all traces of
18361 -fstrict-overflow documentation.
18362 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
18363 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
18364 flag_strict_overflow.
18365 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
18366 * lto-opts.c (lto_write_options): Do not stream it.
18367 * lto-wrapper.c (merge_and_complain): Do not handle it.
18368 * opts.c (default_options_table): Do not set -fstrict-overflow.
18369 (finish_options): Likewise do not clear it when sanitizing.
18370 * simplify-rtx.c (simplify_const_relational_operation): Do not
18371 test flag_strict_overflow.
18373 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
18375 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
18376 using enabled attribute.
18377 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
18378 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
18379 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
18380 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
18381 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
18382 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
18383 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
18384 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
18385 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
18386 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
18388 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
18390 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
18392 2017-05-02 Richard Biener <rguenther@suse.de>
18394 PR tree-optimization/80591
18396 2017-04-10 Richard Biener <rguenther@suse.de>
18398 * tree-ssa-structalias.c (find_func_aliases): Properly handle
18401 2017-05-02 Richard Biener <rguenther@suse.de>
18403 PR tree-optimization/80549
18404 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
18405 (cleanup_tree_cfg_noloop): Create forwarders to known loop
18406 headers if they do not have a preheader.
18408 2017-05-02 Martin Liska <mliska@suse.cz>
18411 * common.opt: Fix typo.
18412 * doc/invoke.texi: Likewise.
18414 2017-05-01 Jan Beulich <jbeulich@suse.com>
18416 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
18417 swapping, add (x,x,m,x,n) alternative.
18419 2017-05-01 Nathan Sidwell <nathan@acm.org>
18421 * calls.c (combine_pending_stack_adjustment_and_call): Remove
18422 unnecessary unadjusted_alignment check.
18424 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
18427 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
18429 * gimplify.c (gimplify_cilk_detach): New function.
18430 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
18431 * tree-core.h: Document EXPR_CILK_SPAWN.
18432 * tree.h (EXPR_CILK_SPAWN): Define.
18434 2017-05-01 David Malcolm <dmalcolm@redhat.com>
18436 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
18437 to use new fixit_hint representation, using the "replace" logic.
18438 (get_line_span_for_fixit_hint): Likewise.
18439 (layout::print_any_fixits): Likewise.
18440 (selftest::test_one_liner_many_fixits): Rename to...
18441 (selftest::test_one_liner_many_fixits_1): ...this, and update
18442 comment and expected output to reflect that the multiple fix-it
18443 hints are now consolidated into one insertion.
18444 (selftest::test_one_liner_many_fixits_2): New test.
18445 (selftest::test_diagnostic_show_locus_one_liner): Update for
18447 (selftest::test_fixit_consolidation): Update for fix-it API
18449 * diagnostic.c (print_parseable_fixits): Likewise.
18450 * edit-context.c (edited_line::m_line_events): Convert from
18451 auto_vec <line_event *> to auto_vec <line_event>.
18452 (class line_event): Convert from abstract base class to a concrete
18453 class, taking over the role of replace_event.
18454 (class insert_event): Delete.
18455 (class replace_event): Rename to class line_event. Convert to
18457 (edit_context::add_fixits): Reimplement.
18458 (edit_context::apply_insert): Delete.
18459 (edit_context::apply_replace): Rename to...
18460 (edit_context::apply_fixit): ...this. Convert to half-open range.
18461 (edited_file::apply_insert): Delete.
18462 (edited_file::apply_replace): Rename to...
18463 (edited_file::apply_fixit): ...this.
18464 (edited_line::~edited_line): Drop deletion of events.
18465 (edited_line::apply_insert): Delete.
18466 (edited_line::apply_replace): Rename to...
18467 (edited_line::apply_fixit): ...this. Convert to half-open range.
18468 Update for change to type of m_line_events.
18469 * edit-context.h (edit_context::apply_insert): Delete.
18470 (edit_context::apply_replace): Rename to...
18471 (edit_context::apply_fixit): ...this.
18473 2017-05-01 Martin Sebor <msebor@redhat.com>
18475 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
18478 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
18481 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
18482 __get_cpuid_max returns 0.
18483 (__get_cpuid_count): Ditto.
18485 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
18487 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
18488 replacement expression is another instance of one of its arguments.
18490 2017-05-01 Jakub Jelinek <jakub@redhat.com>
18493 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
18494 check for stack push/pop autoinc.
18495 * config/i386/i386.c (ix86_agi_dependent): Return false
18496 if the only reason why modified_in_p returned true is that
18497 addr is SP based and set_insn is a push or pop.
18499 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
18501 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
18504 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
18507 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
18508 (account_size_time): Use two predicates - exec_pred and
18510 (evaluate_conditions_for_known_args): Compute both clause and
18512 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
18513 (inline_summary_t::duplicate): Update.
18514 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
18516 (compute_inline_parameters): Likewise.
18517 (estimate_edge_size_and_time): Update caluclation of time.
18518 (estimate_node_size_and_time): Compute both time and nonspecialized
18520 (estimate_ipcp_clone_size_and_time): Update.
18521 (inline_merge_summary): Update.
18522 (do_estimate_edge_time): Update.
18523 (do_estimate_edge_size): Update.
18524 (do_estimate_edge_hints): Update.
18525 (inline_read_section, inline_write_summary): Stream both new predicates.
18526 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
18528 (compute_inlined_call_time): Cleanup.
18529 (big_speedup_p): Update.
18530 (edge_badness): Update.
18531 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
18532 (size_time_entry): Replace predicate by exec_predicate and
18533 nonconst_predicate.
18534 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
18535 (estimate_edge_time): Return also nonspec_time.
18536 (reset_edge_growth_cache): Update.
18538 2017-04-29 Jakub Jelinek <jakub@redhat.com>
18540 PR rtl-optimization/80491
18541 * ifcvt.c (noce_process_if_block): When looking for x setter
18542 with missing else_bb, don't check only the insn right before
18543 cond_earliest, but look for the last insn that x is modified in
18544 within the same bb.
18546 PR rtl-optimization/80491
18547 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
18549 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
18551 PR tree-optimization/80487
18552 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
18554 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18556 PR tree-optimization/79697
18557 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
18558 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
18559 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
18561 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
18562 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
18564 2017-04-28 Martin Sebor <msebor@redhat.com>
18566 PR tree-optimization/80523
18567 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
18568 (init_target_to_host_charmap, target_to_host, target_strtol10): New
18570 (maybe_warn, format_directive, parse_directive): Use new functions.
18571 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
18573 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
18575 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
18577 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
18579 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
18580 target_header_dir): Set correctly.
18581 * configure: Regenerated.
18582 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
18583 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
18584 instead of SYSTEM_HEADER_DIR.
18586 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
18588 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
18589 (estimate_local_effects): Likewise.
18590 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
18591 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
18592 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
18593 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
18594 do_estimate_edge_time, estimate_edge_time): Likewise.
18595 * ipa-inline-analysis.c (estimate_node_size_and_time,
18596 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
18597 (estimate_time_after_inlining): Remove.
18599 2017-04-28 Martin Liska <mliska@suse.cz>
18601 * doc/gcov.texi: Enhance documentation of gcov.
18603 2017-04-28 Martin Liska <mliska@suse.cz>
18605 * doc/gcov.texi: Sort options in alphabetic order.
18606 * doc/gcov-dump.texi: Likewise.
18607 * doc/gcov-tool.texi: Likewise.
18608 * gcov.c (print_usage): Likewise.
18609 * gcov-dump.c (print_usage): Likewise.
18610 * gcov-tool.c (print_merge_usage_message): Likewise.
18611 (print_rewrite_usage_message): Likewise.
18612 (print_overlap_usage_message): Likewise.
18614 2017-04-28 Martin Liska <mliska@suse.cz>
18616 PR gcov-profile/53915
18617 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
18619 2017-04-28 Martin Liska <mliska@suse.cz>
18621 PR gcov-profile/79891
18622 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
18623 is marked by compiler as living on a line.
18624 (get_cycles_count): Remove usage of the union.
18625 (output_intermediate_file): Likewise.
18626 (find_source): Fix GNU coding style.
18627 (accumulate_line_counts): Remove old non-all block mode.
18628 (output_lines): Remove usage of the union.
18629 * profile.c (output_location): Include all BBs, even if
18630 belonging to a same line (and file) as a previous BB.
18632 2017-04-28 Martin Liska <mliska@suse.cz>
18634 * gcov.c (process_args): Handle new argument 'w'.
18635 (read_graph_file): Assign ID to BBs.
18636 (output_branch_count): Display BB # if verbose flag is set.
18637 (output_lines): Likewise for arcs.
18638 (print_usage): Add '--verbose' option help.
18639 * doc/gcov.texi: Document --verbose (-w) option.
18641 2017-04-28 Martin Liska <mliska@suse.cz>
18643 * gcov.c (struct block_location_info): New struct.
18644 (process_file): Fill up the new structure.
18645 (read_graph_file): Replace usage of encoding by the newly added
18647 (add_line_counts): Likewise.
18648 (accumulate_line_counts): Remove usage of the union.
18649 (function_info::function_info): New function.
18650 (function_info::~function_info): Likewise.
18651 (process_file): Call delete instead of release_function.
18652 (release_function): Release the function.
18653 (release_structures): Call delete instead of release_function.
18654 (solve_flow_graph): Replace usage of num_blocks.
18655 (find_exception_blocks): Likewise.
18656 (output_lines): Fix GNU coding style.
18658 2017-04-28 Martin Liska <mliska@suse.cz>
18661 * coverage.c (coverage_remove_note_file): New function.
18662 * coverage.h: Declare the function.
18663 * toplev.c (finalize): Clean if an error has been seen.
18665 2017-04-28 Martin Liska <mliska@suse.cz>
18667 PR gcov-profile/80031
18668 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
18669 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
18670 * gcov.c (read_graph_file): Read just number of blocks.
18671 * profile.c (branch_prob): Do not stream 0 flags per a basic
18674 2017-04-28 Martin Liska <mliska@suse.cz>
18676 * gcov-dump.c (tag_*): Add new argument to declarations.
18677 (dump_gcov_file): Likewise.
18678 (tag_blocks): Add and use new argument depth.
18679 (tag_arcs): Likewise.
18680 (tag_lines): Likewise.
18681 (tag_counters): Likewise.
18682 (tag_summary): Likewise.
18683 (dump_working_sets): Use depth to do a proper indentation.
18685 2017-04-28 Jakub Jelinek <jakub@redhat.com>
18688 * cgraph.h (symtab_node::debug_symtab): No longer inline.
18689 * symtab.c (symtab_node::debug_symtab): Move definition here.
18691 2017-04-28 Richard Biener <rguenther@suse.de>
18693 * lto-streamer.h (LTO_major_version): Bump to 7.
18695 2017-04-28 Richard Biener <rguenther@suse.de>
18697 * tree-vrp.c (assert_info): New struct.
18698 (add_assert_info): New helper.
18699 (register_edge_assert_for_2): Refactor to add asserts to a vector
18701 (register_edge_assert_for_1): Likewise.
18702 (register_edge_assert_for): Likewise.
18703 (finish_register_edge_assert_for): New helper actually registering
18704 asserts where live on edge.
18705 (find_conditional_asserts): Adjust.
18706 (find_switch_asserts): Likewise.
18707 (evrp_dom_walker::try_find_new_range): Generalize.
18708 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
18710 2017-04-27 Marek Polacek <polacek@redhat.com>
18713 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
18714 arg10 and arg11 to itype.
18716 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
18718 * doc/extend.texi (Object Size Checking): Improve grammar.
18720 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
18723 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
18724 that the logic for permitting reciprocal estimates matches that
18727 2017-04-27 Jakub Jelinek <jakub@redhat.com>
18730 * tree.c (type_cache_hasher::equal): Only compare
18731 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
18732 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
18733 non-aggregate element types.
18734 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
18735 about the flag on ARRAY_TYPEs in the comment, formatting fix.
18737 2017-04-27 Richard Biener <rguenther@suse.de>
18739 PR middle-end/80533
18740 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
18741 stripping ARRAY_REFs from MEM_EXPR make sure we're not
18742 keeping a reference to a trailing array.
18744 2017-04-27 Richard Biener <rguenther@suse.de>
18746 PR middle-end/80539
18747 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
18748 being in loop-closed SSA form conservatively.
18749 (chrec_fold_multiply_poly_poly): Likewise.
18751 2017-04-27 Tamar Christina <tamar.christina@arm.com>
18753 PR middle-end/79665
18754 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
18755 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
18757 2017-04-27 Jakub Jelinek <jakub@redhat.com>
18760 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
18761 (aarch64_function_arg_alignment): Return unsigned int again, but still
18762 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
18763 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
18764 Don't emit -Wpsabi note.
18765 (aarch64_function_arg_boundary): Likewise.
18766 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
18769 2017-04-26 Nathan Sidwell <nathan@acm.org>
18771 * tree.h (crc32_unsigned_n): Declare.
18772 (crc32_unsigned, crc32_unsigned): Make inline.
18773 * tree.c (crc32_unsigned_bits): Replace with ...
18774 (crc32_unsigned_n): ... this.
18775 (crc32_unsigned, crc32_byte): Remove.
18776 (crc32_string): Remove unnecessary braces.
18778 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
18780 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
18781 * ipa-inline-analysis.c (MAX_TIME): Remove.
18782 (account_size_time): Use sreal for time.
18783 (dump_inline_summary): Update.
18784 (estimate_function_body_sizes): Update.
18785 (estimate_edge_size_and_time): Update.
18786 (estimate_calls_size_and_time): Update.
18787 (estimate_node_size_and_time): Update.
18788 (inline_merge_summary): Update.
18789 (inline_update_overall_summary): Update.
18790 (estimate_time_after_inlining): Update.
18791 (inline_read_section): Update.
18792 (inline_write_summary): Update.
18793 * ipa-inline.c (compute_uninlined_call_time): Update.
18794 (compute_inlined_call_time): Update.
18795 (recursive_inlining): Update.
18796 (inline_small_functions): Update.
18797 (dump_overall_stats): Update.
18798 * ipa-inline.h: Include sreal.h.
18799 (size_time_entry): Turn time to sreal.
18800 (inline_summary): Turn self_time nad time to sreal.
18802 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
18804 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
18806 (sreal::stream_out, sreal::stream_in): New.
18807 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
18809 2017-04-25 Jakub Jelinek <jakub@redhat.com>
18811 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
18814 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
18817 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
18819 (dimode_scalar_chain::compute_convert_gain): Ditto.
18820 (dimode_scalar_chain::make_vector_copies): Ditto.
18821 (dimode_scalar_chain::convert_reg): Ditto.
18822 (dimode_scalar_chain::convert_insn): Ditto.
18823 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
18824 (VI248_AVX512BW_1): New mode iterator.
18825 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
18826 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
18829 2017-04-25 Martin Sebor <msebor@redhat.com>
18831 PR tree-optimization/80497
18832 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
18833 constants are representable in HOST_WIDE_INT.
18834 (parse_directive): Ditto.
18836 2017-04-25 Martin Sebor <msebor@redhat.com>
18839 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
18840 (new_zero_array): Adjust signature.
18841 (dom_info::dom_init): Used unsigned rather that size_t.
18842 (dom_info::dom_info): Same.
18844 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18845 Jakub Jelinek <jakub@redhat.com>
18848 * config/arm/arm.c: Include gimple.h.
18849 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
18850 returns negative, increment ncrn only if it returned positive.
18851 (arm_needs_doubleword_align): Return int instead of bool,
18852 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
18853 members, but if there is any such non-FIELD_DECL
18854 > PARM_BOUNDARY aligned decl, return -1 instead of false.
18855 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
18856 returns negative, increment nregs only if it returned positive.
18857 (arm_setup_incoming_varargs): Likewise.
18858 (arm_function_arg_boundary): Emit -Wpsabi note if
18859 arm_needs_doubleword_align returns negative, return
18860 DOUBLEWORD_ALIGNMENT only if it returned positive.
18862 2017-04-25 Marek Polacek <polacek@redhat.com>
18865 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
18866 first argument to type.
18868 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
18871 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
18872 type checks to test for compatibility instead of equality.
18874 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18875 Jakub Jelinek <jakub@redhat.com>
18878 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
18880 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
18881 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
18882 the alignment computation, but return their maximum in warn_alignment.
18883 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
18884 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
18886 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
18887 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
18890 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18892 * config/arc/simdext.md (dmpyh): Fix typo.
18894 2017-04-25 Richard Biener <rguenther@suse.de>
18896 PR tree-optimization/80492
18897 * alias.c (compare_base_decls): Handle registers with asm
18898 specification conservatively.
18899 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
18900 compare_base_decls returning dont-know properly.
18902 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18904 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
18905 (legitimate_offset_address_p): New function.
18906 (arc_legitimate_address_p): Use above function.
18908 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18910 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
18912 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18914 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
18915 ACCH registers whenever they are available.
18917 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18919 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
18920 double regs fix when not used.
18922 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18924 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
18926 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
18927 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
18929 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18931 * config/arc/arc.c (arc_output_addsi): Check for h-register class
18932 when emitting short ADD instructions.
18934 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18936 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
18938 (cmpsi_cc_c_insn): Likewise.
18939 (cbranchsi4_scratch): Compute proper instruction length using
18940 compact_hreg_operand.
18941 * config/arc/predicates.md (compact_hreg_operand): New predicate.
18943 2017-04-25 Richard Biener <rguenther@suse.de>
18945 PR middle-end/80509
18946 * passes.c (pass_manager::pass_manager): Initialize
18947 m_name_to_pass_map.
18949 2017-04-25 Richard Biener <rguenther@suse.de>
18951 PR tree-optimization/79201
18952 * tree-ssa-sink.c (statement_sink_location): Handle calls.
18954 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18957 * config/s390/vector.md: Split MEM->GPR vector moves for
18958 non-s_operand addresses.
18960 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18963 * config/s390/predicates.md (reload_const_wide_int_operand): New
18965 * config/s390/s390.md ("movti"): Remove d/P alternative.
18966 ("movti_bigconst"): New pattern definition.
18968 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
18971 * s390-protos.h (s390_expand_cs_hqi): Removed.
18972 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
18973 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
18974 modes as well as CCZ1mode and CCZmode.
18975 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
18976 signature of s390_emit_compare_and_swap.
18977 (s390_expand_cs_hqi): Likewise, make static.
18978 (s390_expand_cs_tdsi): Generate an explicit compare before trying
18979 compare-and-swap, in some cases.
18980 (s390_expand_cs): Wrapper function.
18981 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
18983 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
18984 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
18985 patterns for small and large integers. Forbid symref memory operands.
18986 Move expander to s390.c. Require cc register.
18987 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
18988 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
18989 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
18990 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
18991 symref memory operands. Remove CC mode and call s390_match_ccmode
18993 ("atomic_exchange<mode>"): Allow and implement all integer modes.
18995 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
18997 * config/s390/s390.md (define_peephole2): New peephole to help
18998 combining the load-and-test pattern with volatile memory.
19000 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
19002 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
19003 with CCZmode for TARGET_Z196.
19005 2017-04-25 Jakub Jelinek <jakub@redhat.com>
19007 PR rtl-optimization/80501
19008 * combine.c (make_compound_operation_int): Set subreg_code to SET
19009 even for AND with mask of the sign bit of mode.
19011 PR rtl-optimization/80500
19012 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
19013 sum's initial value.
19015 2017-04-25 Julian Brown <julian@codesourcery.com>
19016 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19018 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
19020 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
19022 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
19024 2017-04-25 Julian Brown <julian@codesourcery.com>
19025 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19027 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
19028 (thunderx2t99_sha): New Reservation.
19030 2017-04-25 Julian Brown <julian@codesourcery.com>
19031 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19033 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
19034 type for 1-element load.
19036 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
19038 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
19040 2017-04-24 Martin Jambor <mjambor@suse.cz>
19042 PR tree-optimization/80293
19043 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
19044 char arrays not totally scalarizable if it is false.
19045 (analyze_all_variable_accesses): Pass correct value in the new
19046 parameter. Add a statistics counter.
19048 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
19050 PR middle-end/79931
19051 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
19053 2017-04-24 Richard Biener <rguenther@suse.de>
19055 PR tree-optimization/80494
19056 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
19057 out for complex types.
19059 2017-04-24 Richard Biener <rguenther@suse.de>
19061 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
19062 * tree-ssa-sccvn.c (print_scc): Print SCC size.
19063 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
19064 (DFS): Adjust and never fail.
19065 (sccvn_dom_walker::fail): Remove.
19066 (sccvn_dom_walker::before_dom_children): Adjust.
19067 (run_scc_vn): Likewise and never fail.
19068 * tree-ssa-pre.c (pass_pre::execute): Adjust.
19069 (pass_fre::execute): Likewise.
19071 2017-04-24 Richard Biener <rguenther@suse.de>
19073 PR tree-optimization/79725
19074 * tree-ssa-sink.c (statement_sink_location): Return whether
19075 failure reason was zero uses. Move that check later.
19076 (sink_code_in_bb): Deal with zero uses by removing the stmt
19079 2017-04-24 Richard Biener <rguenther@suse.de>
19082 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
19083 pointer-based references.
19085 2017-04-24 Richard Biener <rguenther@suse.de>
19088 * pass_manager.h (pass_manager::operator new): Remove.
19089 (pass_manager::operator delete): Likewise.
19090 * passes.c (pass_manager::operator new): Remove.
19091 (pass_manager::operator delete): Likewise.
19092 (pass_manager::pass_manager): Zero individual pass members.
19094 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
19097 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
19098 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
19099 Check "XEXP (src, 1)" operand here.
19100 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
19101 Check "XEXP (src, 1)" operand here.
19102 (dimode_scalar_chain::make_vector_copies): Detect count register
19103 of a shift instruction. Zero extend count register from QImode
19104 to DImode to satisfy vector shift pattern count operand predicate.
19105 Substitute vector shift count operand with a DImode copy.
19106 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
19109 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
19111 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
19112 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
19113 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
19114 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
19115 (UNSPEC_NOREX_MEM): Remove definition.
19117 2017-04-21 Richard Biener <rguenther@suse.de>
19119 PR tree-optimization/79547
19120 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
19121 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
19122 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
19123 without any constraints.
19125 2017-04-21 Richard Biener <rguenther@suse.de>
19127 PR tree-optimization/78847
19128 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
19130 2017-04-21 Richard Biener <rguenther@suse.de>
19132 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
19133 (build_distinct_type_copy): Likewise.
19134 (build_variant_type_copy): Likewise.
19135 * tree.c (build_qualified_type): Pass down mem-stat info.
19136 (build_distinct_type_copy): Likewise.
19137 (build_variant_type_copy): Likewise.
19139 2017-04-21 Richard Biener <rguenther@suse.de>
19141 PR tree-optimization/80237
19142 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
19144 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
19145 for a simplified result.
19147 2016-04-21 Richard Biener <rguenther@suse.de>
19149 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
19150 sth as strict as a simple_iv but a chrec without symbols and an
19151 operand defined in the loop we are peeling (and not some subloop).
19152 (propagate_constants_for_unrolling): Propagate all constants.
19154 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
19157 * config/i386/i386.c (print_reg): Remove assert for disalowed
19158 regno values, call output_operand_lossage instead.
19160 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
19163 * config/i386/constraints.md (Yc): New register constraint.
19164 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
19165 Use Yc constraint for alternative 2 of operand 0. Remove
19166 preferred_for_speed attribute.
19168 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
19170 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
19171 lastprivate clauses in SIMT case.
19173 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
19175 * doc/invoke.texi (-Wextra-semi): Document new warning option.
19177 2017-04-20 Richard Biener <rguenther@suse.de>
19179 PR tree-optimization/57796
19180 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
19181 as N scalar stores.
19182 (vect_model_load_cost): Cost gathers as N scalar loads.
19184 2017-04-20 Richard Biener <rguenther@suse.de>
19186 * ggc-page.c (ggc_allocated_p): Rename to ...
19187 (safe_lookup_page_table_entry): ... this and return the lookup
19189 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
19191 2017-04-20 Richard Biener <rguenther@suse.de>
19193 PR tree-optimization/80453
19194 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
19195 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
19196 from the conditions.
19197 (vn_phi_eq): Pass them down.
19198 (vn_phi_lookup): Record them.
19199 (vn_phi_insert): Likewise.
19201 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
19203 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
19204 uninitialized variable warning to avoid buffer overrun.
19206 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
19209 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
19210 is suppressed for '{ 0 }' in C.
19212 2017-04-20 Jakub Jelinek <jakub@redhat.com>
19214 * BASE-VER: Set to 8.0.0.
19216 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
19218 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
19219 priority .init_array and .fini_array section with SECTION_NOTYPE
19222 2017-04-20 Jakub Jelinek <jakub@redhat.com>
19224 PR middle-end/80423
19225 * tree.h (build_array_type): Add typeless_storage default argument.
19226 * tree.c (type_cache_hasher::equal): Also compare
19227 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
19228 (build_array_type): Add typeless_storage argument, set
19229 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
19231 (build_nonshared_array_type): Adjust build_array_type_1 caller.
19232 (build_array_type): Likewise. Add typeless_storage argument.
19234 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
19235 Jakub Jelinek <jakub@redhat.com>
19237 PR tree-optimization/80426
19238 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
19239 operation on symbolic operands, also compute the overflow for the
19240 invariant part when the operation degenerates into a negation.
19242 2017-04-19 Jakub Jelinek <jakub@redhat.com>
19245 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
19246 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
19249 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
19251 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
19254 * config/avr/avr.c (tree.h): Include it.
19255 (cgraph.h): Include it.
19256 (avr_encode_section_info): Don't warn for uninitialized progmem
19257 variable if it's just an alias.
19259 2017-04-19 Richard Biener <rguenther@suse.de>
19262 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
19263 when needed by AutoPGO.
19265 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
19268 * doc/lto.texi: Remove an extra 'that'.
19270 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
19272 PR rtl-optimization/80429
19273 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
19274 are only used in debug insns.
19276 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
19277 Vladimir Makarov <vmakarov@redhat.com>
19279 * config/sparc/predicates.md (input_operand): Add comment. Return
19280 true for any memory operand when LRA is in progress.
19281 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
19283 2017-04-18 Jeff Law <law@redhat.com>
19286 * mips.md ({return,simple_return}_internal): Do not overwrite
19289 2017-04-18 Jakub Jelinek <jakub@redhat.com>
19291 PR tree-optimization/80443
19292 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
19293 instead of adding 1, subtract -1 and similarly instead of subtracting
19296 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
19298 PR rtl-optimization/80357
19299 * haifa-sched.c (tmp_bitmap): New variable.
19300 (model_recompute): Handle duplicate use records.
19301 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
19302 (free_global_sched_pressure_data): Free it.
19304 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
19307 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
19308 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
19309 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
19310 instead of SYSTEM_HEADER_DIR.
19312 2017-04-18 Jeff Law <law@redhat.com>
19314 PR middle-end/80422
19315 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
19316 predecessors after walking up the insn chain.
19318 2017-04-18 Jakub Jelinek <jakub@redhat.com>
19321 * dwarf2out.c (modified_type_die): Try harder not to emit internal
19322 sizetype type into debug info.
19324 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
19327 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
19328 unneeded test for TARGET_UPPER_REGS_SF.
19329 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
19331 2017-04-18 Jakub Jelinek <jakub@redhat.com>
19334 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
19335 instead of gsi_after_labels.
19337 2017-04-18 Jeff Law <law@redhat.com>
19339 * regcprop.c (maybe_mode_change): Avoid creating copies of the
19343 2017-04-13 Jeff Law <law@redhat.com>
19344 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
19345 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
19347 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
19350 * config/avr/avr.c (intl.h): Include it.
19351 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
19353 2017-04-18 Martin Liska <mliska@suse.cz>
19355 PR gcov-profile/78783
19356 * gcov-tool.c (gcov_output_files): Validate that destination
19357 file is either removed by the tool or by a user.
19359 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
19360 Guy Benyei <guybe@mellanox.com>
19362 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
19363 block, and do not negate it, the stored id is already negative.
19365 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
19367 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
19369 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
19372 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
19373 masks of options that should be turned off if the VSX vector
19374 options are turned off.
19375 (OTHER_P8_VECTOR_MASKS): Likewise.
19376 (OTHER_VSX_VECTOR_MASKS): Likewise.
19377 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
19378 rs6000_disable_incompatible_switches to validate no type switches
19380 (rs6000_incompatible_switch): New function to disallow turning on
19381 other vector options if -mno-vsx, -mno-power8-vector, or
19382 -mno-power9-vector are specified.
19384 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
19386 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
19388 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
19390 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
19391 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
19392 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
19393 (ARG_POINTER_CFA_OFFSET): Likewise.
19395 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
19397 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
19398 conditions to take advantage of various optimizations.
19400 2017-04-13 Jeff Law <law@redhat.com>
19402 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
19403 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
19404 (zero_extendsidi2_dext): Likewise.
19406 2017-04-13 Jakub Jelinek <jakub@redhat.com>
19409 * fold-const.c (fold_ternary_loc): Revert
19410 use op0 instead of fold_convert_loc (loc, type, arg0) part of
19413 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
19415 PR rtl-optimization/80343
19416 * lra-remat.c (update_scratch_ops): Assign original hard reg to
19417 new scratch pseudo.
19419 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
19422 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
19423 to ubsan_encode_value.
19425 2017-04-13 Jeff Law <law@redhat.com>
19427 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
19428 appearing in DEBUG_INSNs.
19430 2017-04-13 Martin Liska <mliska@suse.cz>
19432 PR gcov-profile/80413
19433 * gcov-io.c (gcov_write_string): Copy to buffer just when
19434 allocated size is greater than zero.
19436 2017-04-13 Jakub Jelinek <jakub@redhat.com>
19439 * dwarf2out.c (decls_for_scope): Ignore declarations of
19440 current_function_decl in BLOCK_NONLOCALIZED_VARS.
19442 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
19445 * ipa-visibility.c (non_local_p): Fix typos.
19446 (localize_node): When localizing symbol in same comdat group,
19447 dissolve the group only when we know external symbols are going
19449 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
19451 2017-04-12 Jakub Jelinek <jakub@redhat.com>
19453 PR tree-optimization/79390
19454 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
19455 order does not result in usable sequence, retry with reversed operand
19461 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
19462 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
19463 op0 instead of fold_convert_loc (loc, type, arg0).
19465 2017-04-12 Jeff Law <law@redhat.com>
19467 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
19468 has a delay slot in the generated code.
19470 * config/cris/cris.md (cris_preferred_reload_class): Return
19471 GENNONACR_REGS rather than GENERAL_REGS.
19473 2017-04-12 Jakub Jelinek <jakub@redhat.com>
19476 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
19477 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
19478 signedness of the result type.
19480 2017-04-12 Richard Biener <rguenther@suse.de>
19481 Jeff Law <law@redhat.com>
19483 PR tree-optimization/80359
19484 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
19485 trim stores to TARGET_MEM_REFs.
19487 2017-04-12 Richard Biener <rguenther@suse.de>
19489 PR tree-optimization/79390
19490 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
19491 threading case even more.
19493 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
19496 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
19497 for quad_address_p for TImode, instead of just not indexed_address.
19499 2017-04-12 Richard Biener <rguenther@suse.de>
19500 Bernd Edlinger <bernd.edlinger@hotmail.de>
19502 PR middle-end/79671
19503 * alias.c (component_uses_parent_alias_set_from): Handle
19504 TYPE_TYPELESS_STORAGE.
19505 (get_alias_set): Likewise.
19506 * tree-core.h (tree_type_common): Add typeless_storage flag.
19507 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
19508 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
19509 for types containing members with TYPE_TYPELESS_STORAGE.
19510 (place_field): Likewise.
19511 (layout_type): Likewise for ARRAY_TYPE.
19512 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
19513 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
19514 TYPE_TYPELESS_STORAGE.
19515 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
19517 2017-04-12 Jakub Jelinek <jakub@redhat.com>
19520 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
19521 first argument to type.
19523 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19527 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
19528 CONST0_RTX (mode) rather than const0_rtx where appropriate.
19529 (rs6000_expand_binop_builtin): Likewise.
19530 (rs6000_expand_ternop_builtin): Likewise; also add missing
19531 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
19532 vshasigma built-ins.
19533 * doc/extend.texi: Document that vec_xxpermdi's third argument
19534 must be a constant.
19536 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
19538 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
19539 Use shift_const cost parameter when calculating gain of STV shifts.
19541 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
19543 PR rtl-optimization/70478
19544 * lra-constraints.c (process_alt_operands): Check memory for
19545 disfavoring memory insn operand.
19547 2017-04-11 Jakub Jelinek <jakub@redhat.com>
19549 PR middle-end/80100
19550 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
19551 left shift in unsigned HOST_WIDE_INT type.
19553 PR rtl-optimization/80385
19554 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
19555 (not (neg X)) into (plus X -1) for complex or non-integral modes.
19558 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
19559 if they have any depend clauses.
19561 2017-04-11 Martin Liska <mliska@suse.cz>
19564 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
19565 * ipa-split.c (split_function): Create a local comdat symbol
19566 if caller is in a comdat group.
19568 2017-04-11 Martin Liska <mliska@suse.cz>
19571 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
19574 2017-04-11 Martin Sebor <msebor@redhat.com>
19576 PR middle-end/80364
19577 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
19578 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
19580 (directive::set_width, directive::set_precision, format_character):
19582 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
19585 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
19588 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
19589 conflict, set target->arch_name instead of target->cpu_name.
19591 2017-04-11 Richard Biener <rguenther@suse.de>
19593 PR tree-optimization/80374
19594 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
19595 build_zero_cst, remove fold_convertible_p check again.
19597 2017-04-11 Martin Liska <mliska@suse.cz>
19600 * ubsan.c (instrument_object_size): Do not instrument register
19603 2017-04-11 Jakub Jelinek <jakub@redhat.com>
19606 * config/i386/i386-builtin-types.def
19607 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
19608 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
19609 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
19610 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
19611 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
19612 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
19613 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
19614 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
19615 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
19616 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
19617 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
19618 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
19619 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
19620 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
19621 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
19622 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
19623 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
19624 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
19625 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
19626 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
19627 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
19628 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
19629 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
19630 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
19631 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
19632 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
19633 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
19634 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
19635 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
19636 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
19637 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
19638 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
19639 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
19640 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
19641 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
19642 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
19643 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
19644 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
19645 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
19646 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
19647 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
19648 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
19649 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
19650 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
19651 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
19652 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
19654 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
19655 flag to second_arg_count, handle 4 argument function type _COUNT
19656 aliases, handle second_arg_count on second argument rather than last.
19658 2017-04-10 Jeff Law <law@redhat.com>
19660 PR tree-optimization/80374
19661 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
19662 record anything if we can not convert integer_zero_node to the
19665 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
19668 * config/rs6000/rs6000.c (rs6000_option_override_internal):
19669 Enhance special handling given to the TARGET_P9_MINMAX option in
19670 relation to certain other options.
19672 2017-04-10 Bin Cheng <bin.cheng@arm.com>
19674 PR tree-optimization/80153
19675 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
19676 remove POINTER_PLUS_EXPR's base part directly, rather than through
19679 2017-04-10 Richard Biener <rguenther@suse.de>
19680 Bin Cheng <bin.cheng@arm.com>
19682 PR tree-optimization/80153
19683 * tree-affine.c (aff_combination_to_tree): Get base pointer from
19684 the first element of pointer type aff_tree. Build result expr in
19686 (add_elt_to_tree): Convert to type unconditionally. Remove other
19687 fold_convert calls.
19688 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
19689 (rewrite_use_nonlinear_expr): Check invariant using iv information.
19691 2017-04-10 Richard Biener <rguenther@suse.de>
19693 * tree-ssa-structalias.c (find_func_aliases): Properly handle
19696 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
19698 PR rtl-optimization/70478
19699 * lra-constraints.c (curr_small_class_check): New.
19700 (update_and_check_small_class_inputs): New.
19701 (process_alt_operands): Update curr_small_class_check. Disfavor
19702 alternative insn memory operands. Check available regs for small
19705 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
19708 * config/mips/mips.opt (-mvirt): Update description.
19709 * doc/invoke.texi (-mvirt): Likewise.
19711 2017-04-10 Richard Biener <rguenther@suse.de>
19713 PR middle-end/80362
19714 * fold-const.c (fold_binary_loc): Look at unstripped ops when
19715 looking for NEGATE_EXPR in -A / -B to A / B folding.
19717 2017-04-10 Martin Liska <mliska@suse.cz>
19719 PR gcov-profile/80224
19720 * gcov.c (print_usage): Fix usage string.
19721 (get_gcov_intermediate_filename): Remove.
19722 (output_gcov_file): Use both for normal and intermediate format.
19723 (generate_results): Do not initialize special file for
19724 intermediate format.
19726 2017-04-10 Richard Biener <rguenther@suse.de>
19728 PR tree-optimization/80304
19729 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
19732 2017-04-10 Nathan Sidwell <nathan@acm.org>
19735 * config/rs6000/rs6000.c (rs6000_vector_type): New.
19736 (rs6000_init_builtins): Use it.
19738 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19740 * config/arm/arm.md (<mrc>): Add mode to SET source.
19741 (<mrrc>): Likewise.
19743 2017-04-10 Richard Biener <rguenther@suse.de>
19745 PR middle-end/80344
19746 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
19748 2017-04-10 Jakub Jelinek <jakub@redhat.com>
19751 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
19752 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
19753 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
19754 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
19755 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
19756 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
19757 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
19758 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
19759 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
19760 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
19761 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
19762 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
19763 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
19764 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
19765 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
19766 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
19767 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
19768 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
19769 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
19770 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
19771 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
19772 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
19773 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
19775 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
19777 PR rtl-optimization/70478
19778 * lra-constraints.c: Reverse the last patch.
19780 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
19782 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
19783 Add comment for WCHAR_T.
19785 2017-04-08 Martin Liska <mliska@suse.cz>
19788 2017-04-07 Martin Liska <mliska@suse.cz>
19791 * ipa-split.c (split_function): Add function part to a same comdat
19794 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19797 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
19799 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
19801 * rs6000/rs6000.c (vec_load_pendulum): Rename...
19802 (vec_pairing): ...to this.
19803 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
19804 (rs6000_sched_init): Adjust for name change.
19805 (struct rs6000_sched_context): Likewise.
19806 (rs6000_init_sched_context): Likewise.
19807 (rs6000_set_sched_context): Likewise.
19809 2017-04-07 Jakub Jelinek <jakub@redhat.com>
19815 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
19817 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
19818 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
19819 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
19821 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
19823 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
19825 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
19827 PR rtl-optimization/70703
19828 * ira-color.c (update_conflict_hard_regno_costs): Use
19829 int64_t instead of HOST_WIDE_INT.
19831 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
19833 PR rtl-optimization/70478
19834 * lra-constraints.c (process_alt_operands): Disfavor alternative
19835 insn memory operands.
19837 2017-04-07 Jeff Law <law@redhat.com>
19839 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
19840 CALL and NOTE_INSN_CALL_ARG_LOCATION.
19842 2017-04-07 Martin Liska <mliska@suse.cz>
19845 * config/aarch64/aarch64.c (aarch64_process_target_attr):
19846 Show error message instead of an ICE.
19848 2017-04-07 Martin Liska <mliska@suse.cz>
19851 * ipa-split.c (split_function): Add function part to a same comdat
19854 2017-04-07 Richard Biener <rguenther@suse.de>
19856 PR middle-end/80341
19857 * tree.c (get_unwidened): Also handle ! for_type case for
19859 * convert.c (do_narrow): Split out from ...
19860 (convert_to_integer_1): ... here. Do not pass final truncation
19861 type to get_unwidened for TRUNC_DIV_EXPR.
19863 2017-04-07 Richard Biener <rguenther@suse.de>
19865 * tree-affine.c (wide_int_ext_for_comb): Take type rather
19867 (aff_combination_const): Adjust.
19868 (aff_combination_scale): Likewise.
19869 (aff_combination_add_elt): Likewise.
19870 (aff_combination_add_cst): Likewise.
19871 (aff_combination_convert): Likewise.
19872 (add_elt_to_tree): Likewise. Remove unused argument.
19873 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
19875 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
19877 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
19879 * config/arm/arm.c (arm_default_short_enums): Use
19880 ARM_DEFAULT_SHORT_ENUMS.
19881 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
19883 2017-04-06 Jakub Jelinek <jakub@redhat.com>
19886 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
19887 members with redundant out-of-class redeclaration.
19889 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
19892 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
19893 * config/i386/i386.md (*zero_extendsidi2):
19894 Add (?*x,*x) and (?*v,*v) alternatives.
19896 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
19899 * config/i386/i386.c (ix86_expand_builtin)
19900 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
19901 mode from insn data. Convert operands to insn operand mode.
19902 Copy operands that don't satisfy insn predicate to a register.
19904 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
19906 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
19909 2017-04-06 Richard Biener <rguenther@suse.de>
19911 PR tree-optimization/80334
19912 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
19913 preserve alignment of accesses.
19915 2017-04-06 Richard Biener <rguenther@suse.de>
19917 PR tree-optimization/80262
19918 * tree-sra.c (build_ref_for_offset): Preserve address-space
19920 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
19921 Drop useless address-space information on MEM_REF offsets.
19923 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
19925 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
19927 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
19929 PR rtl-optimization/70703
19930 * ira-color.c (update_conflict_hard_regno_costs): Use
19931 HOST_WIDE_INT instead of long.
19933 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
19936 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
19937 not defined for x86_64 target. Add -mmmx target option when __SSE2__
19939 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
19940 for x86_64 target. Handle -m3dnowa option.
19942 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
19944 PR rtl-optimization/70703
19945 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
19946 (update_conflict_hard_regno_costs): Use long instead of unsigned
19947 arithmetic for cost calculation.
19949 2017-04-05 Jakub Jelinek <jakub@redhat.com>
19950 Bernd Edlinger <bernd.edlinger@hotmail.de>
19953 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
19956 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
19959 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
19960 ptr_mode with Pmode throughout.
19961 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
19962 into probe_stack_range and use DImode.
19964 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
19967 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
19968 call_eh_return is true.
19970 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19972 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
19973 Initialize last_match_fntype_index.
19975 2017-04-05 Jakub Jelinek <jakub@redhat.com>
19978 * tree-nvr.c: Include internal-fn.h.
19979 (pass_return_slot::execute): Ignore internal calls without
19982 2017-04-04 Jakub Jelinek <jakub@redhat.com>
19983 Richard Biener <rguenther@suse.de>
19986 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
19987 captures used multiple times, except for the last use.
19988 * generic-match-head.c: Include gimplify.h.
19990 2017-04-04 Jakub Jelinek <jakub@redhat.com>
19992 PR tree-optimization/79390
19993 * target.h (struct noce_if_info): Declare.
19994 * targhooks.h (default_noce_conversion_profitable_p): Declare.
19995 * target.def (noce_conversion_profitable_p): New target hook.
19996 * ifcvt.h (struct noce_if_info): New type, moved from ...
19997 * ifcvt.c (struct noce_if_info): ... here.
19998 (noce_conversion_profitable_p): Renamed to ...
19999 (default_noce_conversion_profitable_p): ... this. No longer
20001 (noce_try_store_flag_constants, noce_try_addcc,
20002 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
20003 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
20004 instead of noce_conversion_profitable_p.
20005 * config/i386/i386.c: Include ifcvt.h.
20006 (ix86_option_override_internal): Don't override
20007 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
20008 (ix86_noce_conversion_profitable_p): New function.
20009 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
20010 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
20011 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
20012 * doc/tm.texi: Regenerated.
20014 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20016 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
20019 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
20022 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
20023 instructions for small multiply cores.
20025 2017-04-04 Jeff Law <law@redhat.com>
20027 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
20029 (mips_expand_vec_perm_const): Initialize elements in orig_perm
20030 that are not set by the loop over the elements.
20032 2017-04-04 Jakub Jelinek <jakub@redhat.com>
20035 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
20036 int mode, convert_modes it to mode as unsigned, otherwise use
20037 lowpart_subreg to mode rather than SImode.
20038 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
20039 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
20040 Use DImode instead of SImode for the shift count operand.
20041 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
20044 2017-04-04 Richard Biener <rguenther@suse.de>
20046 PR middle-end/80281
20047 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
20048 arithmetic done for the negate or the plus. Simplify.
20049 (A - (-B) -> A + B): Likewise.
20050 * fold-const.c (split_tree): Make sure to not negate pointers.
20052 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
20054 PR rtl-optimization/60818
20055 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
20056 a compare of comparisons with the thing compared if this results
20057 in a different machine mode.
20059 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
20061 * alias.c (base_alias_check): Fix typo in comment.
20062 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
20063 * cgraphunit.c (symbol_table::compile): Likewise.
20064 * collect2.c (maybe_run_lto_and_relink): Likewise.
20065 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
20066 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
20067 * config/avr/avr.c (avr_map_op_t): Likewise.
20068 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
20069 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
20070 * config/epiphany/epiphany.md (movcc): Likewise.
20071 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
20072 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
20074 * config/mips/mips.c (mips_save_restore_reg): Likewise.
20075 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
20076 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
20077 * config/sh/sh.c (sh_rtx_costs): Likewise.
20078 * fold-const.c (fold_truth_andor): Likewise.
20079 * genautomata.c (collapse_flag): Likewise.
20080 * gengtype.h (struct type::u::s): Likewise.
20081 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
20082 * input.c (FORMAT_AMOUNT): Likewise.
20083 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
20084 (known_aggs_to_agg_replacement_list): Likewise.
20085 * ipa-inline-analysis.c: Likewise.
20086 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
20087 * ipa-polymorphic-call.c
20088 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
20089 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
20090 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
20092 * modulo-sched.c (apply_reg_moves): Likewise.
20093 * omp-expand.c (build_omp_regions_1): Likewise.
20094 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
20095 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
20096 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
20097 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
20098 * value-prof.c: Likewise.
20099 * var-tracking.c (val_reset): Likewise.
20101 2017-04-03 Richard Biener <rguenther@suse.de>
20103 PR tree-optimization/80275
20104 * fold-const.c (split_address_to_core_and_offset): Handle
20107 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
20109 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
20110 descriptors is at least equal to that of functions.
20112 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
20114 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
20116 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
20119 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
20120 (mov<IMOD4:mode>): New expander.
20121 (*mov<IMOD4:mode>_internal): New insn and split pattern.
20123 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
20125 PR rtl-optimization/79405
20126 * fwprop.c (propagations_left): New variable.
20127 (forward_propagate_into): Decrement it.
20128 (fwprop_init): Initialize it.
20129 (fw_prop): If the variable has reached zero, stop propagating.
20130 (fwprop_addr): Ditto.
20132 2017-03-31 Jakub Jelinek <jakub@redhat.com>
20135 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
20136 a FUNCTION_DECL, pass it as decl instead of origin to
20139 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
20141 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
20144 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
20147 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
20148 TARGET_VSX_SMALL_INTEGER.
20150 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20152 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
20153 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
20155 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
20157 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
20158 extraction from odd-numbered MSA register.
20160 2017-03-31 Jakub Jelinek <jakub@redhat.com>
20162 PR middle-end/80173
20163 * expmed.c (store_bit_field_1): Don't attempt to create
20164 a word subreg out of hard registers wider than word if they
20165 have HARD_REGNO_NREGS of 1 for their mode.
20167 PR middle-end/80163
20168 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
20169 conversions to integer types wider than word and pointer.
20172 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
20173 (rtx_equal_for_cselib_p): Pass 0 to it.
20174 * cselib.c (cselib_hasher::equal): Likewise.
20175 (rtx_equal_for_cselib_1): Add depth argument. If depth
20176 is 128, don't look up VALUE locs and punt. Increment
20177 depth in recursive calls when walking VALUE locs.
20179 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
20181 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
20182 (make_gcov_file_name): Use the canonical path name for generating
20184 (read_line): Fix handling of files with ascii null bytes.
20186 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
20188 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
20189 to initialise a vector register instead
20190 of using a const_int.
20192 2017-03-30 Jakub Jelinek <jakub@redhat.com>
20194 PR translation/80189
20195 * gimplify.c (omp_default_clause): Use %qs instead of %s in
20196 diagnostic messages.
20198 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
20201 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
20202 (dfp_diex_<mode>): Update mode of operand 1.
20203 * doc/extend.texi (dxex, dxexq): Document change to return type.
20204 (diex, diexq): Document change to argument type.
20206 2017-03-30 Martin Jambor <mjambor@suse.cz>
20209 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
20210 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
20211 it reflects the signature changes performed at the callee side.
20212 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
20213 to cgraph_build_function_type_skip_args.
20214 (build_function_decl_skip_args): Adjust call to the above function.
20216 2017-03-30 Jakub Jelinek <jakub@redhat.com>
20219 * config/i386/sse.md
20220 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
20221 register as dest whenever it is a MEM not rtx_equal_p to the
20222 corresponding dup operand, and when forcing into reg move the
20223 reg into the memory afterwards.
20224 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
20225 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
20226 for the force_reg mode.
20227 (avx512vl_vextractf128<mode>): Use register as dest either
20228 always when a MEM, or when it is a MEM not rtx_equal_p to the
20229 corresponding dup operand, or even not when it is a CONST_VECTOR
20230 depending on the mode and lo vs. hi.
20231 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
20233 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
20234 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
20235 Likewise. Require that operands[2] is even.
20236 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
20237 Remove extraneous parens. Require that operands[2] is a multiple
20239 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
20240 operands[0] is a MEM if <mask_applied>, the predicates/constraints
20241 disallow memory then.
20243 2017-03-30 Richard Biener <rguenther@suse.de>
20245 PR tree-optimization/77498
20246 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
20247 to non-constants over backedges.
20249 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
20251 PR rtl-optimization/80233
20252 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
20253 as last_combined_insn. Do not test for BARRIER_P separately.
20255 2017-03-29 Andreas Schwab <schwab@suse.de>
20258 * calls.c (prepare_call_address): Convert funexp to Pmode before
20259 copying to temp reg.
20261 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20263 PR tree-optimization/80158
20264 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
20265 Handle possible future case of more than one alternate
20267 (replace_rhs_if_not_dup): Likewise.
20268 (replace_one_candidate): Likewise.
20270 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
20272 PR rtl-optimization/80193
20273 * ira.c (ira): Do not check allocation for LRA.
20275 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
20277 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
20278 (nvptx_output_simt_exit): Declare.
20279 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
20280 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
20281 (init_softstack_frame): Move initialization of crtl->is_leaf to...
20282 (nvptx_declare_function_name): ...here. Emit declaration of local
20283 memory space buffer for omp_simt_enter insn.
20284 (nvptx_output_unisimt_switch): New.
20285 (nvptx_output_softstack_switch): New.
20286 (nvptx_output_simt_enter): New.
20287 (nvptx_output_simt_exit): New.
20288 * config/nvptx/nvptx.h (struct machine_function): New fields
20289 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
20290 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
20291 (UNSPECV_SIMT_EXIT): Ditto.
20292 (omp_simt_enter_insn): New insn.
20293 (omp_simt_enter): New expansion.
20294 (omp_simt_exit): New insn.
20295 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
20297 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
20298 (expand_GOMP_SIMT_ENTER_ALLOC): New.
20299 (expand_GOMP_SIMT_EXIT): New.
20300 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
20301 (GOMP_SIMT_ENTER_ALLOC): Ditto.
20302 (GOMP_SIMT_EXIT): Ditto.
20303 * target-insns.def (omp_simt_enter): New insn.
20304 (omp_simt_exit): Ditto.
20305 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
20307 (lower_rec_simd_input_clauses): Implement SIMT privatization.
20308 (lower_rec_input_clauses): Likewise.
20309 (lower_lastprivate_clauses): Handle SIMT privatization.
20311 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
20312 (ompdevlow_adjust_simt_enter): New.
20313 (find_simtpriv_var_op): New.
20314 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
20315 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
20317 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
20318 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
20319 (copy_decl_for_dup_finish): Ditto.
20321 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
20323 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
20326 * config/i386/i386.c (ix86_option_override_internal): Always
20327 allow -mpreferred-stack-boundary=3 for 64-bit targets.
20329 2017-03-28 Bin Cheng <bin.cheng@arm.com>
20331 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
20333 2017-03-28 Bin Cheng <bin.cheng@arm.com>
20335 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
20336 mark new edge's irreducible flag accordign to it.
20337 (vect_do_peeling): Check loop preheader edge's irreducible flag
20338 and pass it to function slpeel_add_loop_guard.
20340 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
20342 PR tree-optimization/80218
20343 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
20344 Update block frequencies and counts.
20346 2017-03-28 Richard Biener <rguenther@suse.de>
20348 PR tree-optimization/78644
20349 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
20350 of a simplification result we may not use it at all.
20352 2017-03-28 Richard Biener <rguenther@suse.de>
20355 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
20356 without arguments, generate default definition of a SSA name.
20358 2017-03-28 Richard Biener <rguenther@suse.de>
20360 PR middle-end/80222
20361 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
20362 TYPE_REF_CAN_ALIAS_ALL references.
20363 * fold-const.c (fold_indirect_ref_1): Likewise.
20365 2017-03-28 Martin Liska <mliska@suse.cz>
20368 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
20369 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
20371 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
20372 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
20374 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
20375 (EXTRA_SPECS): Define.
20376 (SUBTARGET_EXTRA_SPECS): Likewise.
20377 (SUBTARGET_CPP_SPEC): Likewise.
20378 * config/arc/elf.h (EXTRA_SPECS): Renamed to
20379 SUBTARGET_EXTRA_SPECS.
20380 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
20382 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
20384 * config/arc/simdext.md (vst64_insn): Update pattern.
20385 (vld32wh_insn): Likewise.
20386 (vld32wl_insn): Likewise.
20387 (vld64_insn): Likewise.
20388 (vld32_insn): Likewise.
20390 2017-03-28 Marek Polacek <polacek@redhat.com>
20393 * fold-const.c (fold_comparison): Use protected_set_expr_location
20394 instead of SET_EXPR_LOCATION.
20396 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
20398 * tree.c (add_expr): Avoid name lookup warning.
20400 2017-03-27 Jeff Law <law@redhat.com>
20402 PR tree-optimization/80216
20403 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
20404 function name. Limit recursion depth.
20405 (record_temporary_equivalences): Corresponding changes.
20407 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
20409 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
20412 2017-03-27 Jakub Jelinek <jakub@redhat.com>
20415 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
20417 * cfgcleanup.c (reg_note_cfa_p): New array.
20418 (insns_have_identical_cfa_notes): New function.
20419 (old_insns_match_p): Don't cross-jump in between /f
20420 and non-/f instructions. If both i1 and i2 are frame related,
20421 verify all CFA notes, their order and content.
20423 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
20426 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
20427 HImode and SImode with zero extend to DImode to one insn.
20428 (bswap<mode>2_extenddi): Likewise.
20429 (bswapsi2_extenddi): Likewise.
20430 (bswaphi2_extendsi): Likewise.
20431 (bswaphi2): Combine bswap HImode and SImode into one insn.
20432 Separate memory insns from swapping register.
20433 (bswapsi2): Likewise.
20434 (bswap<mode>2): Likewise.
20435 (bswaphi2_internal): Delete, no longer used.
20436 (bswapsi2_internal): Likewise.
20437 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
20438 store, and gpr<-gpr swap insns.
20439 (bswap<mode>2_store): Likewise.
20440 (bswaphi2_reg): Register only splitter, combine with the splitter.
20441 (bswaphi2 splitter): Likewise.
20442 (bswapsi2_reg): Likewise.
20443 (bswapsi2 splitter): Likewise.
20444 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
20445 the insns into load, store, and register/register insns.
20446 (bswapdi2_ldbrx): Likewise.
20447 (bswapdi2_load): Likewise.
20448 (bswapdi2_store): Likewise.
20449 (bswapdi2_reg): Likewise.
20451 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
20453 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
20454 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
20456 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
20459 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
20461 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
20462 special handling for target option conflicts between dform
20463 options (-mpower9-dform, -mpower9-dform-vector,
20464 -mpower9-dform-scalar) and -mno-direct-move.
20466 2017-03-27 Richard Biener <rguenther@suse.de>
20468 PR tree-optimization/80181
20469 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
20471 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
20473 * config/arc/predicates.md (move_double_src_operand): Replace the
20474 call to move_double_src_operand with a call to address_operand.
20476 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
20478 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
20479 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
20480 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
20482 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
20484 * config/arc/predicates.md (long_immediate_loadstore_operand):
20485 Consider scaled addresses cases.
20487 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
20489 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
20490 restored when in interrupt.
20491 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
20492 doesn't have delay slot.
20494 2017-03-27 Richard Biener <rguenther@suse.de>
20497 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
20498 inlined thunk clones.
20500 2017-03-27 Jakub Jelinek <jakub@redhat.com>
20503 * asan.c (instrument_derefs): Copy over last operand from
20504 original COMPONENT_REF to the new COMPONENT_REF with
20505 DECL_BIT_FIELD_REPRESENTATIVE.
20506 * ubsan.c (instrument_object_size): Likewise.
20508 2017-03-27 Richard Biener <rguenther@suse.de>
20510 PR tree-optimization/80170
20511 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
20512 sure DR/SCEV didnt fold in constants we do not see when looking
20513 at the reference base alignment.
20515 2017-03-27 Richard Biener <rguenther@suse.de>
20517 PR middle-end/80171
20518 * gimple-fold.c (fold_ctor_reference): Properly guard against
20519 NULL return value from canonicalize_constructor_val.
20521 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
20524 * config/i386/i386.c (ix86_expand_builtin)
20525 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
20526 flags reg setting and flags reg using instructions.
20527 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
20528 clobbering instructions to zero extend op2.
20530 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
20532 * doc/install.texi (Configuration) <--with-aix-soname>:
20533 Update link to AIX ld.
20535 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
20537 PR rtl-optimization/80160
20538 PR rtl-optimization/80159
20539 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
20540 reg_alternate_class into account.
20542 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
20545 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
20546 to consider in curr_insn_transform.
20548 2017-03-24 Jakub Jelinek <jakub@redhat.com>
20550 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
20551 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
20552 and emit_mode_inner.
20554 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20556 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
20557 argument to the overloaded builtin variants. Use the new flag to
20558 deprecate certain builtin variants.
20559 * config/s390/s390-builtin-types.def: Add new builtin types.
20560 * config/s390/s390-builtins.h: Support new flags field for
20561 overloaded builtins.
20562 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
20563 (s390_macro_to_expand): Enable vector float data type.
20564 (s390_cpu_cpp_builtins_internal): Indicate support of the new
20565 builtins by incrementing the __VEC__ version number.
20566 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
20568 (s390_resolve_overloaded_builtin): Emit error messages depending
20569 on the builtin flags.
20570 * config/s390/s390.c (s390_expand_builtin): Support additional
20571 flags argument. Change error message to match the messages
20572 emitted in s390-c.c.
20573 * config/s390/s390.md: New UNSPEC_* constants.
20574 (op_type): Add new instruction types.
20575 * config/s390/vecintrin.h: Add new builtins and test data class
20577 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
20578 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
20579 (VEC_INEXACT, VEC_NOINEXACT): New constants.
20580 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
20581 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
20582 ("vec_mergel<mode>"): V_HW -> VEC_HW.
20584 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
20585 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
20586 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
20587 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
20589 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
20590 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
20591 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
20592 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
20594 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
20595 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
20596 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
20597 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
20598 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
20599 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
20600 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
20602 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
20603 ("vec_scatter_element<V_HW_4:mode>_DI")
20604 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
20605 ("vec_fpint<mode>", "vflls")
20606 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
20607 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
20608 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
20609 ("*vec_cmphe<mode>_cc"): ... these.
20611 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
20612 mode constant instead of magic value.
20614 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20616 * config/s390/s390.c (s390_expand_vec_compare): Support other
20617 vector floating point modes than just V2DF.
20618 (s390_expand_vcond): Likewise.
20619 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
20620 (s390_cannot_change_mode_class): Prevent mode changes between TF
20621 and V1TF in vector registers.
20622 * config/s390/s390.md (DF, SF): New mode attributes.
20623 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
20624 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
20625 SFmode support for VRs.
20626 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
20628 (VFT, VF_HW): New mode iterators.
20629 (vw, sdx): New mode attributes.
20630 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
20631 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
20632 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
20633 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
20634 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
20635 also the new vector floating point modes. Renaming to ...
20637 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
20638 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
20639 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
20640 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
20641 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
20642 ("vec_unordered<mode>"): ... these.
20644 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
20645 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
20646 ("*vec_extendv2df"): New insn definitions.
20648 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20650 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
20651 ("mulditi3_2", "*muldi3_sign"): New patterns.
20652 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
20653 rename the pattern definition.
20655 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20657 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
20659 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
20661 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20663 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
20664 instruction if possible.
20665 * config/s390/vector.md (vec_halfnumelts): New mode
20667 ("*vec_vllezlf<mode>"): New pattern.
20669 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20671 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
20672 ("popcountv4si2", "popcountv2di2"): Rename to ...
20673 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
20674 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
20676 ("popcount<mode>2_vxe"): New pattern.
20678 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20680 * common/config/s390/s390-common.c (processor_flags_table): Add
20682 * config.gcc: Add arch12.
20683 * config/s390/driver-native.c (s390_host_detect_local_cpu):
20684 Default to arch12 for unknown CPU model numbers.
20685 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
20686 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
20687 PROCESSOR_max sanity check.
20688 * config/s390/s390-opts.h (enum processor_type): Add
20690 * config/s390/s390.c (processor_table): Add arch12.
20691 (s390_expand_builtin): Add check for B_VXE flag.
20692 (s390_issue_rate): Add PROCESSOR_ARCH12.
20693 (s390_get_sched_attrmask): Likewise.
20694 (s390_get_unit_mask): Likewise.
20695 (s390_sched_score): Enable z13 scheduling for arch12.
20696 (s390_sched_reorder): Likewise.
20697 (s390_sched_variable_issue): Likewise.
20698 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
20700 (s390_tune_attr): Use z13 scheduling also for arch12.
20701 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
20702 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
20703 (TARGET_VXE_P): New macros.
20704 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
20705 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
20706 * config/s390/s390.opt: Add arch12 as processor_type.
20708 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20710 * config/s390/s390.md
20711 ("fixuns_truncdddi2", "fixuns_trunctddi2")
20712 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
20713 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
20715 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
20716 Rename expanders to ...
20718 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
20719 ("fixuns_truncdddi2_emu"): ... these.
20721 ("fixuns_trunc<mode>si2_emu"): New expander.
20723 ("*fixuns_truncdfdi2_z13"): Rename to ...
20724 ("*fixuns_truncdfdi2_vx"): ... this.
20726 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20728 * config/s390/2964.md: Remove the single element vector compare
20729 instructions which are no longer used.
20730 * config/s390/s390.c (s390_select_ccmode): Remove handling of
20732 (s390_canonicalize_comparison): Remove handling of DFmode
20734 (s390_expand_vec_compare_scalar): Remove function.
20735 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
20736 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
20738 ("*cmp<mode>_ccs"): Add wfcdb instruction.
20740 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20742 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
20744 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
20745 will anyway by matched by mov<mode>_64dfp.
20747 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20749 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
20750 vlef/vstef. Add missing operand to vleif.
20752 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20754 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
20755 pair for all vector types with 64 bit elements.
20756 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
20757 * config/s390/vector.md (V_HW_64): ... here.
20758 (V_128_NOSINGLE): New mode iterator.
20759 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
20760 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
20761 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
20762 ("*vec_load_pairv2di"): Change to ...
20763 ("*vec_load_pair<mode>"): ... this one.
20765 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20767 * config/s390/constraints.md: Add comments.
20768 (jKK): Reject element sizes > 8 bytes.
20769 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
20771 * config/s390/s390.md: Add the s_operand checks formerly in
20772 s390_split_ok_p to various splitters where they are still
20774 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
20775 for 128 bit vectors. Plus two splitters.
20777 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20779 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
20782 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20785 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
20786 error if the boundary argument is not constant.
20788 2017-03-24 Jakub Jelinek <jakub@redhat.com>
20790 PR rtl-optimization/80112
20791 * loop-doloop.c (doloop_condition_get): Don't check condition
20792 if cmp isn't SET with IF_THEN_ELSE src.
20794 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20796 PR tree-optimization/80158
20797 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
20798 replacing a candidate statement, also replace it for the
20799 candidate's alternate interpretation.
20800 (replace_rhs_if_not_dup): Likewise.
20801 (replace_one_candidate): Likewise.
20803 2017-03-24 Richard Biener <rguenther@suse.de>
20805 PR tree-optimization/80167
20806 * graphite-isl-ast-to-gimple.c
20807 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
20809 (translate_isl_ast_to_gimple::get_rename): Likewise.
20811 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
20813 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
20814 handling of certain combinations of target options, including the
20815 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
20816 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
20818 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20821 * config/arm/arm.md (*load_multiple): Add reload_completed to
20822 matching condition.
20824 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20825 Richard Biener <rguenth@suse.de>
20827 PR tree-optimization/79908
20828 PR tree-optimization/80136
20829 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
20830 been cast away, gimplify_and_add suffices.
20832 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
20834 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
20836 2017-03-23 Richard Biener <rguenther@suse.de>
20838 PR tree-optimization/80032
20839 * gimplify.c (gimple_push_cleanup): Forced unconditional
20840 cleanups still have to go to the conditional_cleanups
20843 2017-03-22 Jakub Jelinek <jakub@redhat.com>
20845 PR tree-optimization/80072
20846 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
20848 (next_operand_entry_id): Change type to unsigned int.
20849 (sort_by_operand_rank): Make sure to return the right return value
20850 even if unsigned fields are bigger than INT_MAX.
20851 (struct oecount): Change cnt and id type to unsigned int.
20852 (oecount_hasher::equal): Formatting fix.
20853 (oecount_cmp): Make sure to return the right return value
20854 even if unsigned fields are bigger than INT_MAX.
20855 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
20858 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
20859 TREE_READONLY on result if writing it more than once.
20862 * doc/invoke.texi (-fsanitize=thread): Document that with
20863 -fnon-call-exceptions atomics are not able to throw
20867 * tsan.c: Include tree-eh.h.
20868 (instrument_builtin_call): Call maybe_clean_eh_stmt or
20869 maybe_clean_or_replace_eh_stmt where needed.
20870 (instrument_memory_accesses): Add cfg_changed argument.
20871 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
20872 if it returned true.
20873 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
20875 PR rtl-optimization/63191
20876 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
20877 wrapper function, moved the whole old content into ...
20878 (ix86_delegitimize_address_1): ... this. New inline function.
20879 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
20880 true as last argument instead of ix86_delegitimize_address.
20882 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
20884 * config/aarch64/aarch64.c (generic_branch_cost): Copy
20885 cortexa57_branch_cost.
20887 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
20889 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
20891 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20894 * doc/md.texi (Constraints): Document wA constraint.
20895 * config/rs6000/constraints.md (wA): New.
20896 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
20897 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
20898 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
20899 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
20901 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
20904 * gimplify.c (is_oacc_declared): New function.
20905 (oacc_default_clause): Use it to set default flags for acc declared
20906 variables inside parallel regions.
20907 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
20908 declared variables.
20909 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
20910 declare attribute to any decl as necessary.
20912 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
20915 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
20916 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
20917 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
20918 (arm_arch_lpae): This.
20919 * config/arm/arm.c (arm_arch7ve): Rename into ...
20920 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
20921 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
20924 2017-03-22 Martin Liska <mliska@suse.cz>
20927 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
20928 error message instead of an ICE.
20930 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20932 * doc/extend.texi (6.11 Additional Floating Types): Revise.
20934 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
20936 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
20938 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
20941 2017-03-21 Martin Sebor <msebor@redhat.com>
20943 * doc/extend.texi: Use "cannot" instead of "can't."
20944 * doc/hostconfig.texi: Same.
20945 * doc/install.texi: Same.
20946 * doc/invoke.texi: Same.
20947 * doc/loop.texi: Same.
20948 * doc/md.texi: Same.
20949 * doc/objc.texi: Same.
20950 * doc/rtl.texi: Same.
20951 * doc/tm.texi: Same.
20952 * doc/tm.texi.in: Same.
20953 * doc/trouble.texi: Same.
20955 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
20958 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
20959 (collect_checksum_attributes): Set it.
20960 (die_checksum_ordered): Use it.
20962 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20964 PR tree-optimization/79908
20965 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
20966 change: For a VA_ARG whose LHS has been cast away, use
20967 force_gimple_operand to construct the side effects.
20969 2017-03-21 David Malcolm <dmalcolm@redhat.com>
20971 PR translation/80001
20972 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
20973 more amenable to translation.
20974 (oacc_loop_auto_partitions): Likewise.
20976 2017-03-21 Marek Polacek <polacek@redhat.com>
20977 Martin Sebor <msebor@redhat.com>
20979 PR tree-optimization/80109
20980 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
20981 on INTEGRAL_TYPE_P.
20983 2017-03-21 Jakub Jelinek <jakub@redhat.com>
20984 Segher Boessenkool <segher@kernel.crashing.org>
20987 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
20988 check reg_used_between_p between insn and one of succ or succ2
20989 depending on if succ is artificial insn not inserted into insn
20992 2017-03-21 Martin Liska <mliska@suse.cz>
20994 PR gcov-profile/80081
20995 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
20996 * doc/gcc.texi: Include gcov-dump stuff.
20997 * doc/gcov-dump.texi: New file.
20999 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
21001 PR rtl-optimization/79150
21002 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
21003 conditional jump, if the jump is the last insn of the loop.
21005 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21006 Richard Biener <rguenth@suse.de>
21008 PR tree-optimization/79908
21009 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
21010 been cast away, use force_gimple_operand to construct the side
21013 2017-03-21 Martin Liska <mliska@suse.cz>
21015 PR libfortran/79956
21016 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
21019 2017-03-21 Brad Spengler <spender@grsecurity.net>
21022 * plugin.c (htab_hash_plugin): New function.
21023 (add_new_plugin): Use it and adjust.
21024 (parse_plugin_arg_opt): Adjust.
21025 (init_one_plugin): Likewise.
21027 2017-03-21 Richard Biener <rguenther@suse.de>
21029 PR tree-optimization/80032
21030 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
21031 if set force the cleanup to happen unconditionally.
21032 (gimplify_target_expr): Push inserted clobbers with force_uncond
21033 to avoid them being removed by control-dependent DCE.
21035 2017-03-21 Richard Biener <rguenther@suse.de>
21037 PR tree-optimization/80122
21038 * tree-inline.c (copy_bb): Do not expans va-arg packs or
21039 va_arg_pack_len when the inlined call stmt requires pack
21041 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
21043 2017-03-21 Jakub Jelinek <jakub@redhat.com>
21046 * tsan.c (instrument_builtin_call): If the memory model argument
21047 is not a constant, assume it is valid.
21050 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
21053 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
21055 PR rtl-optimization/79910
21056 * combine.c (can_combine_p): Do not allow combining an I0 or I1
21057 if its dest is used by an insn before I2 (other than the combined
21058 insns themselves, which are properly handled already).
21060 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
21063 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
21065 * combine.c (record_used_regs): New static function.
21066 (try_combine): Handle situations where there is an additional
21067 instruction between I2 and I3 which needs to have a LOG_LINK
21071 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
21073 * combine.c (try_combine): Delete redundant i1 test. Call
21074 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
21076 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21079 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
21080 alternatives 13/14.
21082 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21084 PR tree-optimization/80054
21085 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
21086 the optimization if a PHI or any of its arguments is not dominated
21087 by the candidate's basis. Use gphi* rather than gimple* as
21089 (replace_profitable_candidates): Clean up a gimple* variable that
21090 should be a gphi* variable.
21092 2017-03-20 Martin Sebor <msebor@redhat.com>
21095 * doc/extend.texi (attribute constructor): Document present limitation.
21097 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
21100 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
21101 __POWER9_VECTOR__ #ifdef control, change template definition to
21102 use Power9-specific built-in function.
21103 (vec_any_eq): Likewise.
21104 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
21105 to control outcomes from this test.
21106 (vector_ae_<mode>p): For VEC_F modes, likewise.
21108 2017-03-20 Ian Lance Taylor <iant@google.com>
21110 * config/i386/i386.c (ix86_function_regparm): Save an extra
21111 register for -fsplit-stack with DECL_STATIC_CHAIN.
21113 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
21116 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
21117 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
21119 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
21121 * config/riscv/riscv.c (riscv_print_operand): Use "fence
21123 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
21126 2017-03-20 Marek Polacek <polacek@redhat.com>
21129 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
21131 2017-03-20 Richard Biener <rguenther@suse.de>
21133 PR tree-optimization/80113
21134 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
21135 allocate extra SSA name for PHI def.
21136 (add_close_phis_to_outer_loops): Likewise.
21137 (add_close_phis_to_merge_points): Likewise.
21138 (copy_loop_close_phi_args): Likewise.
21139 (copy_cond_phi_nodes): Likewise.
21141 2017-03-20 Martin Liska <mliska@suse.cz>
21143 PR middle-end/79753
21144 * tree-chkp.c (chkp_build_returned_bound): Do not build
21145 returned bounds for a LHS that's not a BOUNDED_P type.
21147 2017-03-20 Martin Liska <mliska@suse.cz>
21151 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
21152 COMPLEX_CST and VECTOR_CST.
21154 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21157 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
21158 target operand. A new splitter adds the clobber statement in case
21159 the target operand is dead anyway.
21161 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
21163 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
21164 to age-old versions of binutils and glibc.
21166 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
21168 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
21170 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
21172 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
21174 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
21176 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
21177 requirement for binutils 2.13.
21179 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
21181 * combine.c (try_combine): Delete redundant i1 test. Call
21182 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
21184 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
21186 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
21187 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
21189 <riscv64-*-elf>: Re-arrange section
21190 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
21191 <riscv32-*-linux>: Likewise.
21192 <riscv64-*-elf>: Likewise
21193 <riscv64-*-linux>: Likewise.
21195 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
21198 * aarch64.opt(verbose-cost-dump): Fix typo.
21200 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
21203 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
21204 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
21206 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
21208 * reload.c (find_reloads): When reloading a nonoffsettable address,
21209 use RELOAD_OTHER for it and its address reloads.
21211 PR rtl-optimization/79910
21212 * combine.c (record_used_regs): New static function.
21213 (try_combine): Handle situations where there is an additional
21214 instruction between I2 and I3 which needs to have a LOG_LINK
21217 2017-03-17 Jeff Law <law@redhat.com>
21219 PR tree-optimization/71437
21220 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
21221 conditional in the hash table first.
21222 (vrp_dom_walker::before_dom_children): Extract condition from
21223 ASSERT_EXPR. Record condition, its inverion and any implied
21224 conditions as well.
21226 2017-03-17 Marek Polacek <polacek@redhat.com>
21227 Markus Trippelsdorf <markus@trippelsdorf.de>
21229 PR tree-optimization/80079
21230 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
21233 2017-03-17 Richard Biener <rguenther@suse.de>
21235 PR middle-end/80075
21236 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
21237 Properly verify the LHS before the RHS possibly claims to be
21239 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
21242 2017-03-17 Martin Jambor <mjambor@suse.cz>
21244 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
21245 (List of -O2 options): Likewise.
21246 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
21249 2017-03-17 Tom de Vries <tom@codesourcery.com>
21251 * gcov-dump.c (print_usage): Print bug_report_url.
21253 2017-03-17 Richard Biener <rguenther@suse.de>
21255 PR middle-end/80050
21256 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
21257 (parser::peek): Likewise.
21259 2017-03-17 Richard Biener <rguenther@suse.de>
21261 PR tree-optimization/80048
21262 * sese.c (free_sese_info): Properly release rename_map and
21263 copied_bb_map elements.
21265 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
21267 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
21268 Add linked-list forward and backlinks. Insert on
21269 construction, remove on destruction.
21270 (class pass_store_merging): Add m_stores_head field.
21271 (pass_store_merging::terminate_and_process_all_chains):
21272 Iterate over m_stores_head list.
21273 (pass_store_merging::terminate_all_aliasing_chains):
21275 (pass_store_merging::execute): Check for debug stmts first.
21276 Push new chains onto the m_stores_head stack.
21278 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
21281 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
21282 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
21283 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
21285 2017-03-16 Jeff Law <law@redhat.com>
21287 PR tree-optimization/71437
21288 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
21289 member function. Implementation moved into after_dom_children
21290 member function and into the threader's thread_outgoing_edges
21292 (dom_opt_dom_walker::after_dom_children): Simplify by moving
21293 some code into new thread_outgoing_edges.
21294 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
21295 definition. Simplify marker handling (do it here). Assume we always
21296 have the available expression and the const/copies tables.
21297 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
21299 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
21300 * tree-vrp.c (equiv_stack): No longer file scoped.
21301 (vrp_dom_walker): New class.
21302 (vrp_dom_walker::before_dom_children): New member function.
21303 (vrp_dom_walker::after_dom_children): Likewise.
21304 (identify_jump_threads): Setup domwalker. Use it rather than
21305 walking edges in a random order by hand. Simplify setup/finalization.
21306 (finalize_jump_threads): Remove.
21307 (vrp_finalize): Do not call identify_jump_threads here.
21308 (execute_vrp): Do it here instead and call thread_through_all_blocks
21311 PR tree-optimization/71437
21312 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
21314 (simplify_stmt_for_jump_threading): Add basic_block argument. All
21316 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
21317 (dom_opt_dom_walker::thread_across_edge): Remove
21318 handle_dominating_asserts argument. All callers changed.
21319 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
21320 changes. Remove calls to lhs_of_dominating_assert. Other
21321 uses of handle_dominating_asserts turn into unconditional code
21322 (simplify_control_stmt_condition_1): Likewise.
21323 (simplify_control_stmt_condition): Likewise.
21324 (thread_through_normal_block, thread_across_edge): Likewise.
21325 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
21326 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
21327 object if it is not an SSA_NAME.
21328 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
21329 before calling into the VRP specific simplifiers.
21330 (identify_jump_threads): Remove handle_dominating_asserts
21333 2017-03-16 Jakub Jelinek <jakub@redhat.com>
21336 * tree-diagnostic.c (default_tree_printer): No longer static.
21337 * tree-diagnostic.h (default_tree_printer): New prototype.
21339 2017-03-16 Tamar Christina <tamar.christina@arm.com>
21341 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
21342 Change ins into fmov.
21344 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21346 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
21347 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
21348 Use h_con constraint for operand 1.
21349 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
21350 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
21352 2017-03-15 Jeff Law <law@redhat.com>
21354 PR tree-optimization/71437
21355 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
21356 (record_temporary_equivalences): Use it.
21358 PR tree-optimization/71437
21359 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
21360 tree-ssa-scopedtables.
21361 (lookup_avail_expr, build_and_record_new_cond): Likewise.
21362 (record_conditions, record_cond, vuse_eq): Likewise.
21363 (record_edge_info): Adjust to API tweak of record_conditions.
21364 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
21365 (record_temporary_equivalences, optimize_stmt): Likewise.
21366 (eliminate_redundant_computations): Likewise.
21367 (record_equivalences_from_stmt): Likewise.
21368 * tree-ssa-scopedtables.c: Include options.h and params.h.
21369 (vuse_eq): New function, moved from tree-ssa-dom.c
21370 (build_and_record_new_cond): Likewise.
21371 (record_conditions): Likewise. Accept vector of conditions rather
21372 than edge_equivalence structure for first argument.
21373 for the first argument.
21374 (avail_exprs_stack::lookup_avail_expr): New member function, moved
21375 from tree-ssa-dom.c.
21376 (avail_exprs_stack::record_cond): Likewise.
21377 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
21378 from tree-ssa-dom.c.
21379 (avail_exprs_stack): Add new member functions lookup_avail_expr
21381 (record_conditions): Declare.
21383 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
21386 * lra-constraints.c (process_alt_operands): Increase reject for
21387 reloading an input/output operand.
21389 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
21392 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
21393 insns to convert from signed/unsigned char/short to IEEE 128-bit
21395 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
21397 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
21400 * config/i386/i386.c (ix86_vector_duplicate_value): Create
21401 subreg of inner mode for values already in registers.
21403 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
21405 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
21406 iteration reg is used after the loop.
21408 2017-03-14 Martin Sebor <msebor@redhat.com>
21410 PR tree-optimization/79800
21411 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
21412 precision in negative-positive range.
21413 (format_floating): Call non-const overload with adjusted precision.
21415 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
21418 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
21421 2017-03-14 Martin Sebor <msebor@redhat.com>
21423 PR middle-end/80020
21424 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
21425 * builtins.def (aligned_alloc): Use it.
21428 * Makefile.in (GTFILES): Add calls.c.
21429 * calls.c: Include "gt-calls.h".
21431 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
21433 PR rtl-optimization/79728
21434 * regs.h (struct target_regs): New field
21435 x_contains_allocatable_regs_of_mode.
21436 (contains_allocatable_regs_of_mode): New macro.
21437 * reginfo.c (init_reg_sets_1): Initialize it, and change
21438 contains_reg_of_mode so it includes global regs as well.
21439 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
21440 rather than contains_regs_of_mode.
21442 2017-03-14 Martin Liska <mliska@suse.cz>
21444 * doc/invoke.texi: Document options that can't be combined with
21445 -fcheck-pointer-bounds.
21447 2017-03-14 Martin Liska <mliska@suse.cz>
21449 PR middle-end/79831
21450 * doc/invoke.texi (-Wchkp): Document the option.
21452 2017-03-14 Martin Liska <mliska@suse.cz>
21454 * Makefile.in: Install gcov-dump.
21456 2017-03-14 Martin Liska <mliska@suse.cz>
21458 * multiple_target.c (expand_target_clones): Bail out for
21459 an invalid attribute.
21461 2017-03-14 Richard Biener <rguenther@suse.de>
21463 * alias.c (struct alias_set_entry): Pack properly.
21464 * cfgloop.h (struct loop): Likewise.
21465 * cse.c (struct set): Likewise.
21466 * ipa-utils.c (struct searchc_env): Likewise.
21467 * loop-invariant.c (struct invariant): Likewise.
21468 * lra-remat.c (struct cand): Likewise.
21469 * recog.c (struct change_t): Likewise.
21470 * rtl.h (struct address_info): Likewise.
21471 * symbol-summary.h (function_summary): Likewise.
21472 * tree-loop-distribution.c (struct partition): Likewise.
21473 * tree-object-size.c (struct object_size_info): Likewise.
21474 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
21475 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
21476 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
21477 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
21478 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
21479 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
21480 (struct _stmt_vec_info): Likewise.
21482 2017-03-14 Martin Liska <mliska@suse.cz>
21485 * multiple_target.c (create_dispatcher_calls): Check that
21486 a target can create a function dispatcher.
21488 2017-03-14 Martin Liska <mliska@suse.cz>
21491 * multiple_target.c (expand_target_clones): Drop local.local
21492 flag for default implementation.
21494 2017-03-14 Richard Biener <rguenther@suse.de>
21496 PR tree-optimization/80030
21497 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
21499 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
21501 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
21502 gcc_fallthrough() instead of __attribute__((fallthrough));
21504 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
21506 * doc/gcc.texi: Remove "up" link to (DIR).
21507 * doc/gccint.texi: Ditto.
21509 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
21511 * doc/install.texi (Specific) <avr>: Remove reference to
21514 2017-03-13 Jeff Law <law@redhat.com>
21516 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
21517 attribute rather than comments.
21519 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
21520 match_scratch operand is highest.
21522 2017-03-13 Martin Liska <mliska@suse.cz>
21524 PR middle-end/78339
21525 * ipa-pure-const.c (warn_function_noreturn): If the declarations
21526 is a CHKP clone, use original declaration.
21528 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
21530 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
21531 (arc_conditional_register_usage): Use a different allocation order
21532 when optimizing for size.
21533 * common/config/arc/arc-common.c (arc_option_optimization_table):
21534 Section anchors default on when optimizing for size.
21536 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
21538 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
21540 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
21542 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
21543 * config/arc/arc.md (cpu_facility): Add cd variant.
21544 (*movqi_insn): Add code density variant.
21545 (*movhi_insn): Likewise.
21546 (*movqi_insn): Likewise.
21547 (*addsi3_mixed): Likewise.
21548 (subsi3_insn): Likewise.
21550 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
21552 * config/arc/arc.md (movsi_cond_exec): Update constraint.
21554 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
21556 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
21557 expressions with MINUS and UNARY ops.
21559 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21562 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
21564 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
21565 between vec_select and vector argument.
21566 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
21567 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
21568 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
21569 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
21570 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
21571 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
21573 2017-03-13 Richard Biener <rguenther@suse.de>
21576 * params.def (vect-max-peeling-for-alignment): Fix typo.
21578 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
21580 * doc/install.texi (Specific) <mips-*-*>: Remove description of
21581 issue that only occurred with binutils below 2.18.
21583 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
21585 * doc/install.texi (Specific) <cris-axis-elf>: No longer
21586 refer to binutils 2.11/2.12 minimum.
21588 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
21590 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
21591 ftp.kernel.org and simplify binutils requirement.
21593 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
21595 * doc/invoke.texi (Warning Options): Fix spelling of link-time
21597 (Optimize Options): Ditto. Also remove redundancy.
21599 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21601 PR translation/79848
21602 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
21604 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
21605 to G_ to avoid double translation.
21607 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21609 PR translation/79923
21610 * auto-profile.c (get_combined_location): Convert leading
21611 character of diagnostics to lower case and remove trailing period.
21612 (read_profile): Likewise for various diagnostics.
21613 * config/arm/arm.c (arm_option_override): Remove trailing period
21614 from various diagnostics.
21615 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
21616 (msp430_expand_delay_cycles): Likewise.
21618 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21621 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
21622 full command-line argument, rather than just "str".
21623 (aarch64_validate_march): Likewise.
21624 (aarch64_validate_mtune): Likewise.
21626 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
21628 PR rtl-optimization/78911
21629 * lra-assigns.c (must_not_spill_p): New function.
21630 (spill_for): Use it.
21632 2017-03-10 Jakub Jelinek <jakub@redhat.com>
21634 PR tree-optimization/79981
21635 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
21636 ATOMIC_COMPARE_EXCHANGE ifn result.
21637 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
21638 IFN_ATOMIC_COMPARE_EXCHANGE.
21640 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21643 * opts.c (parse_sanitizer_options): Add missing question mark to
21644 "did you mean" message.
21646 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21648 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
21650 (VMULEUH_UNS): Likewise.
21651 (VMULOUB_UNS): Likewise.
21652 (VMULOUH_UNS): Likewise.
21653 * config/rs6000/rs6000.c (builtin_function_type): Remove
21654 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
21656 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21659 * read-rtl-function.c (function_reader::read_rtx_operand): Update
21660 x with result of extra_parsing_for_operand_code_0.
21661 (function_reader::extra_parsing_for_operand_code_0): Convert
21662 return type from void to rtx, returning x. When reading
21663 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
21664 larger size containing struct block_symbol.
21666 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
21668 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
21669 -mfloat128-hardware without -m64.
21671 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
21674 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
21675 entries to the case statement that marks unsigned arguments to
21676 overloaded functions.
21678 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
21680 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
21681 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
21683 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
21686 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
21687 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
21689 2017-03-10 Martin Liska <mliska@suse.cz>
21693 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
21694 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
21697 2017-03-10 Olivier Hainque <hainque@adacore.com>
21699 * tree-switch-conversion (array_value_type): Start by resetting
21700 candidate type to it's main variant.
21702 2017-03-10 Jakub Jelinek <jakub@redhat.com>
21704 PR rtl-optimization/79909
21705 * combine.c (try_combine): Use simplify_replace_rtx on individual
21706 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
21707 of the whole CALL_INSN_FUNCTION_USAGE.
21709 PR tree-optimization/79972
21710 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
21711 get_range_info on SSA_NAMEs. Formatting fixes.
21713 2017-03-10 Richard Biener <rguenther@suse.de>
21714 Jakub Jelinek <jakub@redhat.com>
21716 PR tree-optimization/77975
21717 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
21718 edge to be constant.
21719 (get_val_for): For constant x return it. Formatting fix.
21720 (loop_niter_by_eval): Avoid pointless looping if the next iteration
21721 would use the same bases as the current one.
21723 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21725 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
21726 instead of vec_select for V1TImode.
21727 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
21729 (VSX_LE_128): Add V1TI to this mode iterator.
21730 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
21731 (*vsx_le_perm_store_<mode>): Likewise.
21732 (pre-reload splitter for VSX stores): Likewise.
21733 (post-reload splitter for VSX stores): Likewise.
21734 (*vsx_xxpermdi2_le_<mode>): Likewise.
21735 (*vsx_lxvd2x2_le_<mode>): Likewise.
21736 (*vsx_stxvd2x2_le_<mode>): Likewise.
21738 2017-03-09 Michael Eager <eager@eagercon.com>
21740 Correct failures with --enable-checking=yes,rtl.
21742 * config/microblaze/microblaze.c (microblaze_expand_shift):
21743 Replace GET_CODE test with CONST_INT_P and INTVAL test with
21744 test for const0_rtx.
21745 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
21746 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
21748 2017-03-09 Richard Biener <rguenther@suse.de>
21750 PR tree-optimization/79977
21751 * graphite-scop-detection.c (scop_detection::merge_sese):
21752 Handle the case of extra exits to blocks dominating the entry.
21754 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
21756 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
21759 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
21761 PR rtl-optimization/79949
21762 * lra-constraints.c (process_alt_operands): Check memory when
21763 trying to predict a cycle. Print about the overall increase.
21765 2017-03-09 Richard Biener <rguenther@suse.de>
21767 PR middle-end/79971
21768 * gimple-expr.c (useless_type_conversion_p): Preserve
21769 TYPE_SATURATING for fixed-point types.
21771 2017-03-09 Richard Biener <rguenther@suse.de>
21774 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
21775 alignment of BLKmode params.
21777 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21780 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
21781 (VALL_NO_V2Q): Likewise.
21783 * config/aarch64/aarch64-simd.md
21784 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
21786 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
21787 VALL_NO_V2Q mode iterator.
21788 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
21790 2017-03-09 Martin Liska <mliska@suse.cz>
21792 PR tree-optimization/79631
21793 * tree-chkp-opt.c (chkp_is_constant_addr): Call
21794 tree_int_cst_sign_bit just for INTEGER constants.
21796 2017-03-09 Martin Liska <mliska@suse.cz>
21800 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
21803 2017-03-09 Marek Polacek <polacek@redhat.com>
21806 * tree.c (inchash::add_expr): Handle TREE_VEC.
21808 2017-03-09 Martin Liska <mliska@suse.cz>
21811 (chkp_narrow_size_and_offset): New function.
21812 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
21813 (void chkp_parse_bit_field_ref): New function.
21814 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
21815 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
21817 2017-03-09 Martin Liska <mliska@suse.cz>
21820 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
21821 (chkp_find_bounds_1): Remove gcc_unreachable.
21823 2017-03-09 Jakub Jelinek <jakub@redhat.com>
21826 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
21827 BUILT_IN_SYNC*, determine the access type from the size suffix and
21828 always build a MEM_REF with that type. Handle forgotten
21829 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
21832 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
21833 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
21834 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
21835 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
21836 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
21837 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
21838 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
21839 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
21840 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
21841 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
21842 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
21843 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
21844 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
21845 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
21846 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
21847 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
21848 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
21849 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
21850 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
21851 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
21852 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
21853 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
21854 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
21855 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
21856 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
21857 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
21858 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
21859 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
21860 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
21861 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
21862 definitions outside of __OPTIMIZE__ guarded section.
21865 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
21866 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
21867 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
21868 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
21871 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21873 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
21874 ("vfenez<mode>"): Add missing constraints.
21876 2017-03-08 Martin Sebor <msebor@redhat.com>
21879 * config/nds32/nds32.c (nds32_option_override):
21880 Fix misspelled diagnostic.
21882 2017-03-08 Jakub Jelinek <jakub@redhat.com>
21885 * gimplify.c (gimplify_omp_for): Replace index var in outer
21886 taskloop statement with an artificial variable and add
21887 OMP_CLAUSE_PRIVATE clause for it.
21889 2017-03-08 Richard Biener <rguenther@suse.de>
21891 PR tree-optimization/79955
21892 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
21893 for accesses that are completely outside of the variable.
21895 2017-03-08 Andrew Haley <aph@redhat.com>
21897 PR tree-optimization/79943
21898 * tree-ssa-loop-split.c (compute_new_first_bound): When
21899 calculating the new upper bound, (END-BEG) should be added, not
21902 2017-03-08 Jakub Jelinek <jakub@redhat.com>
21904 * config/avr/avr.md (setmemhi): Make sure match_dup
21905 operand number comes before match_scratch.
21907 2017-03-08 Richard Biener <rguenther@suse.de>
21909 PR tree-optimization/79920
21910 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
21911 with ncopies == 1 to ...
21912 (vect_transform_slp_perm_load): ... here. Properly compute
21913 all element loads by iterating VF times over the group. Do
21914 not handle ncopies (computed in a broken way) in
21915 vect_create_mask_and_perm.
21917 2017-03-08 Jakub Jelinek <jakub@redhat.com>
21920 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
21921 is a uniform vector, use uniform_vector_p return value instead of
21922 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
21924 2017-03-07 Marek Polacek <polacek@redhat.com>
21926 PR middle-end/79809
21927 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
21928 (alloca_call_type): Likewise.
21930 2017-03-07 Martin Liska <mliska@suse.cz>
21932 * gcov.c (process_args): Put comment to correct location.
21934 2017-03-07 Martin Liska <mliska@suse.cz>
21936 PR middle-end/68270
21937 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
21938 Use array_at_struct_end_p instead of DECL_CHAIN (field).
21939 (chkp_narrow_bounds_for_field): Likewise.
21940 (chkp_parse_array_and_component_ref): Pass one more argument to
21943 2017-03-07 Richard Biener <rguenther@suse.de>
21945 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
21948 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
21950 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
21951 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
21953 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21956 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
21957 to end of description.
21958 (PARAM_MAX_STORES_TO_MERGE): Likewise.
21960 2017-03-07 Jakub Jelinek <jakub@redhat.com>
21962 PR rtl-optimization/79901
21963 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
21965 (*avx512f_<code><mode>3<mask_name>): ... this.
21966 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
21967 iterator instead of VI8_AVX2_AVX512BW.
21969 PR rtl-optimization/79901
21970 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
21971 min/max expander, expand it using expand_vec_cond_expr.
21974 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
21977 2017-03-06 Jakub Jelinek <jakub@redhat.com>
21980 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
21981 to void * for PCH reasons.
21982 * dwarf2out.c (output_loc_operands, output_die): Cast
21983 v.val_vec.array to unsigned char *.
21985 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
21988 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
21991 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
21993 PR rtl-optimization/79571
21994 * lra-constraints.c (process_alt_operands): Calculate static
21995 reject and subtract it from overall when only addresses will be
21998 2017-03-06 Julia Koval <julia.koval@intel.com>
22001 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
22002 incoming stack boundary to 128 for 64-bit targets.
22004 2017-03-06 Richard Biener <rguenther@suse.de>
22006 PR tree-optimization/79894
22007 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
22008 to NULL after folding it.
22010 2017-03-06 Richard Biener <rguenther@suse.de>
22012 PR tree-optimization/79824
22013 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
22014 check disabling peeling for gaps.
22016 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
22018 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
22019 attributes): Document gettimeofday.
22021 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
22023 * config/s390/s390.c (s390_option_override_internal): Set
22024 PARAM_MIN_VECT_LOOP_BOUND
22026 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
22028 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
22029 * config/s390/s390.md: Likewise.
22031 2017-03-06 Jakub Jelinek <jakub@redhat.com>
22034 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
22035 (<avx2_avx512>_perm<mode>): Rename to ...
22036 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
22038 (<avx512>_perm<mode>_mask): Rename to ...
22039 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
22041 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
22042 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
22043 instead of VI8F_256_512.
22044 (avx512f_perm<mode>): New define_expand.
22045 (avx512f_perm<mode>_mask): Likewise.
22046 (avx512f_perm<mode>_1<mask_name>): New define_insn.
22047 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
22049 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
22051 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
22052 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
22054 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
22056 2017-03-06 Martin Liska <mliska@suse.cz>
22059 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
22060 when having a SSA NAME w/o VAR_DECL assigned to it.
22062 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
22064 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
22065 msa_dpsub_<su>_d): Fix MODE for vec_select.
22067 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
22069 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
22071 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
22073 2017-03-06 Richard Biener <rguenther@suse.de>
22075 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
22076 * plugin.c (register_plugin_info): Likewise.
22077 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
22079 2017-03-05 Jakub Jelinek <jakub@redhat.com>
22081 * config/i386/sse.md (sse_storehps, sse_storelps,
22082 avx_<castmode><avxsizesuffix>_<castmode>,
22083 avx512f_<castmode><avxsizesuffix>_<castmode>,
22084 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
22085 in condition that at least one operand is not a MEM.
22087 2017-03-03 Jakub Jelinek <jakub@redhat.com>
22089 PR middle-end/79805
22090 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
22091 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
22093 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
22094 gimple_call_nothrow_p flag based on whether original builtin can throw.
22095 If it can, emit following stmts on the fallthrough edge.
22096 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
22097 don't create new bb if inserting just debug stmts on the edge, try to
22098 insert them on the fallthru bb or just reset debug stmts.
22100 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
22103 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
22104 restore recog_data (including the operand rtxes inside it) around
22105 the call to get_insn_template.
22107 2017-03-03 Martin Sebor <msebor@redhat.com>
22109 PR tree-optimization/79699
22110 * context.c (context::~context): Free MPFR caches to avoid
22111 a memory leak on program exit.
22113 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22115 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
22116 Use wide_int::ulow () instead of .elt (0).
22118 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
22120 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
22121 (*pushxf): Limit oF constraint to 32bit targets and add oC
22122 constraint for 64bit targets.
22123 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
22124 (*pushdf): Change rmF constraint to rmC.
22126 2017-03-03 Martin Liska <mliska@suse.cz>
22128 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
22129 Remove unused variable.
22131 2017-03-03 Jakub Jelinek <jakub@redhat.com>
22134 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
22135 is a memory operand, increase num_memory.
22136 (ix86_expand_args_builtin): Likewise.
22138 2017-03-03 Jan Hubicka <jh@suse.cz>
22141 * ipa-devirt.c (maybe_record_node): Properly handle
22142 __cxa_pure_virtual visibility.
22144 2017-03-03 Martin Liska <mliska@suse.cz>
22146 PR tree-optimization/79803
22147 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
22149 (pass_loop_prefetch::execute): Disabled optimization if an
22150 assumption about L1 cache size is not met.
22152 2017-03-03 Martin Liska <mliska@suse.cz>
22154 PR rtl-optimization/79574
22155 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
22156 (hash_scan_set): Likewise.
22157 (dump_hash_table): Likewise.
22158 (hoist_code): Likewise.
22160 2017-03-03 Richard Biener <rguenther@suse.de>
22162 * fixed-value.c (fixed_from_string): Restore use of elt (1)
22163 in place of uhigh ().
22164 (fixed_convert_from_real): Likewise.
22166 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
22169 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
22171 2017-03-03 Richard Biener <rguenther@suse.de>
22173 PR middle-end/79818
22174 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
22175 TYPE_OVERFLOW_UNDEFINED check.
22177 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22179 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
22181 (vector_ae_<mode>_p): Likewise.
22182 (vector_nez_<mode>_p): Likewise.
22183 (vector_ne_v2di_p): Likewise.
22184 (vector_ae_v2di_p): Likewise.
22185 (vector_ne_<mode>_p): Likewise.
22186 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
22188 (vsx_tsqrt<mode>2_fe): Likewise.
22190 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
22193 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
22195 2017-03-02 Jakub Jelinek <jakub@redhat.com>
22197 PR rtl-optimization/79780
22198 * cprop.c (one_cprop_pass): When second and further conditional trap
22199 in a single basic block is turned into an unconditional trap, turn it
22200 into a deleted note to avoid RTL verification failures.
22202 2017-03-02 Richard Biener <rguenther@suse.de>
22204 * fold-const.c (const_binop): Use ulow () instead of elt (0).
22206 2017-03-02 Richard Biener <rguenther@suse.de>
22208 PR tree-optimization/79345
22210 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
22211 param and abort the walk, returning -1 if it is hit.
22212 (walk_aliased_vdefs): Take a limit param and pass it on.
22213 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
22214 defaulting to 0 and return a signed int.
22215 * tree-ssa-uninit.c (struct check_defs_data): New struct.
22216 (check_defs): New helper.
22217 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
22218 about uninitialized memory.
22219 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
22220 bogus uninitialized warning.
22221 (fixed_convert_from_real): Likewise.
22223 2017-03-02 Bin Cheng <bin.cheng@arm.com>
22225 PR tree-optimization/66768
22226 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
22227 iv_use if base object can't be determined.
22229 2017-03-02 Jakub Jelinek <jakub@redhat.com>
22231 PR tree-optimization/79345
22232 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
22233 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
22234 (get_pattern_stats): Initialize it.
22235 * genemit.c (gen_expand): Verify match_scratch numbers come after
22236 match_operand/match_dup numbers.
22237 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
22238 match_scratch numbers.
22239 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
22241 * config/s390/s390.md (trunctdsd2): Likewise.
22243 2017-03-02 Richard Biener <rguenther@suse.de>
22245 * wide-int.h (wide_int_storage::operator=): Implement in terms
22248 2017-03-02 Richard Biener <rguenther@suse.de>
22250 PR tree-optimization/79777
22251 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
22252 the to insert expression to sth existing.
22254 2017-03-01 Martin Sebor <msebor@redhat.com>
22256 PR middle-end/79692
22257 * gimple-ssa-sprintf.c
22258 (directive::known_width_and_precision): New function.
22259 (format_integer): Use it.
22260 (get_mpfr_format_length): Consider the full range of precision
22261 when computing %g output with the # flag. Set the likely byte
22262 count to 3 rather than 1 when precision is indeterminate.
22263 (format_floating): Correct the lower bound of precision.
22265 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22267 * doc/invoke.texi: Document default code model for 64-bit Linux.
22269 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22272 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
22273 udiv rather than div since input pattern is unsigned.
22275 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
22277 * config/i386/i386.c (print_reg): Warn for values of
22278 unsupported size in integer register.
22280 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
22283 * config/rs6000/predicates.md (current_file_function_operand): Do
22284 not allow self calls to be local if the function is replaceable.
22286 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
22289 * config/rs6000/altivec.h (vec_ctz and others): Change the
22290 preprocessor macro that controls conditional compilation from
22291 _ARCH_PWR9 to __POWER9_VECTOR__.
22292 (vec_all_ne): Change parameterization of __altivec_scalar_pred
22293 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
22294 control (instead of _ARCH_PWR9 control) so that template
22295 definition uses power9-specific function.
22296 (vec_any_eq): Likewise.
22297 (vec_all_ne): Change macro definition to use a power9-specific
22298 expansion under #ifdef __POWER9_VECTOR__ control (instead of
22299 _ARCH_PWR9 control).
22300 (vec_any_eq) Likewise.
22301 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
22302 expansion for CMPNEF to remove support for xvcmpnesp instruction.
22303 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
22304 support for xvcmpnedp instruction.
22305 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
22306 macro expansion so that Power9 implementation of vec_all_ne does
22307 not use the AltiVec predicate framework.
22308 (VCMPNEH_P): Likewise.
22309 (VCMPNEW_P): Likewise.
22310 (VCMPNED_P): Likewise.
22311 (VCMPNEFP_P): Likewise.
22312 (VCMPNEDP_P): Likewise.
22313 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
22314 implementation of vec_any_eq to not use AltiVec predicate
22316 (VCMPAEH_P): Likewise.
22317 (VCMPAEW_P): Likewise.
22318 (VCMPAED_P): Likewise.
22319 (VCMPAEFP_P): Likewise.
22320 (VCMPAEDP_P): Likewise.
22321 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
22322 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
22323 not use the AltiVec predicate framework.
22324 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
22325 of vec_any_eq to not use AltiVec predicate framework.
22326 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
22327 support for predefined __POWER9_VECTOR__ macro to indicate that
22328 Power9 instruction selection is enabled.
22329 (altivec_overloaded_builtins): Remove extraneous
22330 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
22331 function argument types RS6000_BTI_bool_V16QI and
22332 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
22333 entry for overloaded function argument types RS6000_BTI_bool_V4SI
22334 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
22335 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
22336 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
22337 Power9 for implementations of vec_cmpne. Change the signature for
22338 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
22339 (representing vec_all_ne) to remove the previously described first
22340 argument of type RS6000_BTI_INTSI, as this was an artifact of
22341 reliance on the AltiVec predicate framework, which is no longer
22342 used in the implementation of these functions. Add
22343 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
22344 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
22345 since, unlike the AltiVec predicate framework implementation, we
22346 do not share function descriptors between vec_alle and vec_anyeq.
22347 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
22348 set of modes that receive special treatment even when
22349 TARGET_P9_VECTOR is true. The special treatment emits code that
22350 does not depend on Power9 instructions.
22351 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
22352 define_expand to not rely on AltiVec predicate framework.
22353 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
22355 (vector_ne_v2di_p): Change this define_expand to not rely on
22356 AltiVec predicate framework.
22357 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
22359 (vector_ne_<mode>_p): Change this define_expand to not rely on
22360 AltiVec predicate framework.
22361 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
22363 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
22364 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
22365 define_insn pattern.
22366 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
22367 define_insn pattern because the xvcmpne<VSs>. instruction is not
22369 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
22370 instruction is not supported.
22372 2017-03-01 Jakub Jelinek <jakub@redhat.com>
22374 * config/nvptx/nvptx.c: Include intl.h.
22376 2017-03-01 Martin Jambor <mjambor@suse.cz>
22379 * ipa-prop.h (ipa_bits): Removed field known.
22380 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
22381 to pointers. Adjusted their comments to warn about their sharing.
22382 (ipcp_transformation_summary): Change bits to a vector of pointers.
22383 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
22384 (ipa_get_ipa_bits_for_value): Declare.
22385 * tree-vrp.h (value_range): Mark as GTY((for_user)).
22386 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
22387 (ipa_bits_hash_table): Likewise.
22388 (ipa_vr_ggc_hash_traits): Likewise.
22389 (ipa_vr_hash_table): Likewise.
22390 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
22391 being pointers and vr_known being removed.
22392 (ipa_set_jf_unknown): Likewise.
22393 (ipa_get_ipa_bits_for_value): New function.
22394 (ipa_set_jfunc_bits): Likewise.
22395 (ipa_get_value_range): New overloaded functions.
22396 (ipa_set_jfunc_vr): Likewise.
22397 (ipa_compute_jump_functions_for_edge): Use the above functions to
22398 construct bits and vr parts of jump functions.
22399 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
22400 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
22402 (ipcp_grow_transformations_if_necessary): Also allocate
22403 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
22405 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
22406 them. Fix too long lines.
22407 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
22408 vr_known being removed.
22409 (ipa_read_jump_function): Use new setter functions to construct bits
22410 and vr parts of jump functions or set them to NULL.
22411 (write_ipcp_transformation_info): Adjust for bits being pointers.
22412 (read_ipcp_transformation_info): Likewise.
22413 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
22415 Include gt-ipa-prop.h.
22416 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
22418 (ipcp_store_bits_results): Likewise.
22419 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
22420 Do not write to existing jump functions but use a temporary instead.
22422 2017-03-01 Jakub Jelinek <jakub@redhat.com>
22425 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
22426 attempt to use its first operand as BIT_FIELD_REF base.
22428 2017-03-01 Richard Biener <rguenther@suse.de>
22430 PR middle-end/79721
22431 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
22432 interpolating formula in wrapping arithmetic.
22433 (chrec_apply): Convert chrec_evaluate return value to wanted type.
22435 2017-03-01 Jakub Jelinek <jakub@redhat.com>
22437 PR tree-optimization/79734
22438 * tree-vect-generic.c (expand_vector_condition): Optimize
22439 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
22440 Handle VEC_COND_EXPR where comparison has different inner width from
22441 type's inner width.
22443 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
22445 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
22446 markup, and similar issues. Remove @opindex entries for things
22447 that aren't options. Add missing -mmpy-option entries.
22449 2017-02-28 Jakub Jelinek <jakub@redhat.com>
22451 PR tree-optimization/79737
22452 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
22453 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
22454 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
22455 instead of byte_size. Formatting fix.
22456 (shift_bytes_in_array_right): Formatting fix.
22458 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
22461 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
22462 condition on optimize for the leaf function test.
22464 2017-02-28 Martin Liska <mliska@suse.cz>
22467 * read-rtl-function.c (function_reader::handle_unknown_directive):
22468 Bail out when one uses -flto.
22470 2017-02-28 Martin Liska <mliska@suse.cz>
22472 * common.opt: Replace space with tabular for options of <number>
22474 * config/i386/i386.opt: Show <number> value for
22475 -mlarge-data-threshold.
22476 * opts.c (print_filtered_help): Do not display number in hexadecimal
22479 2017-02-28 Martin Liska <mliska@suse.cz>
22481 * common.opt: Fix --help=option -Q for options which are of
22484 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
22486 * config/i386/i386.c (print_reg): Error out for values
22487 of 8-bit size in invalid integer register.
22489 2017-02-28 Martin Sebor <msebor@redhat.com>
22491 PR tree-optimization/79691
22492 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
22494 2017-02-28 Jakub Jelinek <jakub@redhat.com>
22497 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
22498 gcc_unreachable with output_operand_lossage.
22500 2017-02-28 Richard Biener <rguenther@suse.de>
22502 PR tree-optimization/79740
22503 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
22505 (visit_nary_op): Insert the nary into the hashtable if we
22506 pattern-matched sth.
22507 * tree-ssa-pre.c (eliminate_insert): Robustify.
22509 2017-02-28 Richard Biener <rguenther@suse.de>
22511 PR middle-end/79731
22512 * fold-const.c (decode_field_reference): Reject out-of-bound
22515 2017-02-28 Jakub Jelinek <jakub@redhat.com>
22517 * config/i386/i386.c: Include intl.h.
22518 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
22519 instead of just cond ? "..." : "...".
22520 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
22521 * coverage.c (read_counts_file): Likewise.
22522 * omp-offload.c: Include intl.h.
22523 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
22524 of just cond ? "..." : "...".
22525 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
22526 of just cond ? "..." : "...".
22528 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
22531 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
22533 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
22534 'tune for' CPU name.
22535 * config/arm/arm-cpu-data.h: Regenerated.
22537 2017-02-28 Richard Biener <rguenther@suse.de>
22539 PR tree-optimization/79732
22540 * tree-inline.c (expand_call_inline): Do not shadow var.
22542 2017-02-28 Richard Biener <rguenther@suse.de>
22544 PR tree-optimization/79723
22545 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
22546 address-space properly.
22548 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
22550 * doc/optinfo.texi (Optimization groups): Fix option used for
22552 * doc/invoke.texi (-fopt-info): Document "omp".
22553 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
22554 (OPTGROUP_ALL): Add OPTGROUP_OMP.
22555 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
22556 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
22557 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
22559 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
22561 * dumpfile.c (optgroup_options): Instead of "openmp", associate
22562 OPTGROUP_OMP with "omp".
22564 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
22567 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
22568 for arithmetic shift of unsigned V2DI.
22570 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
22572 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
22573 arc/linux.h headers.
22574 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
22575 (LINK_SPEC): Likewise.
22576 (ARC_TLS_EXTRA_START_SPEC): Likewise.
22577 (EXTRA_SPECS): Likewise.
22578 (STARTFILE_SPEC): Likewise.
22579 (ENDFILE_SPEC): Likewise.
22580 (LIB_SPEC): Likewise.
22581 (TARGET_SDATA_DEFAULT): Likewise.
22582 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
22583 (MULTILIB_DEFAULTS): Likewise.
22584 (DWARF2_UNWIND_INFO): Likewise.
22585 * config/arc/big.h: New file.
22586 * config/arc/elf.h: Likewise.
22587 * config/arc/linux.h: Likewise.
22588 * config/arc/t-uClibc: Remove.
22590 2017-02-27 Bin Cheng <bin.cheng@arm.com>
22592 PR tree-optimization/77536
22593 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
22594 (tree_transform_and_unroll_loop): Use above function to compute the
22595 estimated niter of unrolled loop and use it when scaling profile.
22596 Also use count info rather than frequency if it's non-zero.
22597 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
22598 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
22599 (vect_transform_loop): Call above function.
22601 2017-02-27 Richard Biener <rguenther@suse.de>
22603 PR tree-optimization/45397
22604 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
22605 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
22606 (visit_nary_op): Add pattern matching for CSEing sign-changed
22607 or truncated operations with wider ones.
22609 2017-02-27 Richard Biener <rguenther@suse.de>
22611 PR tree-optimization/79690
22612 * tree-vect-stmts.c (vectorizable_store): Use vector type
22613 built from the DR with address-space.
22615 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
22617 * doc/invoke.texi (Optimize Options): Refine the description
22618 of asan-use-after-return.
22620 2017-02-25 Alan Modra <amodra@gmail.com>
22622 PR rtl-optimization/79584
22623 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
22624 base, not ad->base_term, the reg within base. Remove assertion
22625 that ad->base == ad->base_term. Replace gen_int_mode using
22626 bogus mode with const0_rtx.
22628 2017-02-25 Jakub Jelinek <jakub@redhat.com>
22630 PR middle-end/79396
22631 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
22632 FMA_EXPR like tcc_binary or tcc_unary.
22634 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
22637 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
22639 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
22640 (output_loc_operands): Handle DW_OP_call_ref and
22641 DW_OP_GNU_variable_value.
22642 (struct variable_value_struct): New type.
22643 (struct variable_value_hasher): Likewise.
22644 (variable_value_hash): New variable.
22645 (string_types): Remove.
22646 (copy_loc_descr): New function.
22647 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
22648 (prepend_loc_descr_to_each): New function.
22649 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
22650 instead of add_loc_descr_to_each if the first argument is single
22651 location list and the second has multiple.
22652 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
22653 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
22654 when looking for variable value which doesn't have other location info.
22655 (loc_list_from_tree): Formatting fix.
22656 (gen_array_type_die): Simplify DW_AT_string_length handling.
22657 (adjust_string_types): Remove.
22658 (gen_subprogram_die): Don't call adjust_string_types nor test/set
22659 string_types. Call resolve_variable_values.
22660 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
22661 (resolve_addr_in_expr): Likewise. Add A argument.
22662 (copy_deref_exprloc): Remove deref argument. Adjust for the
22663 original expression being DW_OP_GNU_variable_value with optionally
22664 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
22665 optionally after it.
22666 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
22667 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
22668 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
22669 (variable_value_hasher::hash, variable_value_hasher::equal): New
22671 (resolve_variable_value_in_expr, resolve_variable_value,
22672 resolve_variable_values, note_variable_value_in_expr,
22673 note_variable_value): New functions.
22674 (dwarf2out_early_finish): Call note_variable_value on all toplevel
22677 2017-02-24 Jakub Jelinek <jakub@redhat.com>
22680 * opts.h (handle_generated_option): Add GENERATED_P argument.
22681 * opts-common.c (handle_option): Adjust function comment.
22682 (handle_generated_option): Add GENERATED_P argument, pass it to
22684 (control_warning_option): Pass false to handle_generated_option
22686 * opts.c (maybe_default_option): Pass true to handle_generated_option
22688 * optc-gen.awk: Likewise.
22690 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22692 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
22693 a REG, look at the REG it is a SUBREG of.
22694 (splitter for cmpeqsi_t): Ditto.
22696 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22698 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
22699 the special USEs with the pattern of the insn, not the insn itself.
22701 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
22704 * doc/invoke.texi: Document -mload-store-pairs.
22706 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22707 Sandra Loosemore <sandra@codesourcery.com>
22709 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
22710 argument isn't a CONST_INT.
22711 (nios2_alternate_compare_const): Assert op is a CONST_INT.
22712 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
22713 (nios2_validate_compare): Bypass alternate compare logic if *op2
22714 is not a CONST_INT.
22715 (ldstwm_operation_p): Return false if first_base is not a REG or
22716 if first_offset is not a CONST_INT.
22718 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22720 * config/cris/cris.md: Use correct operand in a define_peephole2.
22722 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22724 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
22726 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22728 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
22729 this_insn if it is an INSN or JUMP_INSN.
22730 (force_offsettable): Look at base, not at addr.
22731 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
22732 on things that aren't necessarily CONST_INTs.
22734 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
22736 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
22737 -mfpmath=sse is the default also for x86-32 targets with SSE2
22738 instruction set when @option{-ffast-math} is enabled
22740 2017-02-24 Jeff Law <law@redhat.com>
22742 PR rtl-optimizatoin/79286
22743 * ira.c (update_equiv_regs): Drop may_trap_p exception to
22746 2017-02-24 Richard Biener <rguenther@suse.de>
22748 PR tree-optimization/79389
22749 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
22752 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
22754 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
22755 function comment to reflect reality.
22756 (loop_exits_before_overflow): Fix typo in function description.
22758 2017-02-24 Richard Biener <rguenther@suse.de>
22760 PR tree-optimization/79389
22761 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
22762 properly that a threading opportunity exists. Detect conditional
22763 copy/constant propagation opportunities.
22765 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
22767 * config/visium/visium.md (type): Add trap.
22768 (b): New mode attribute.
22769 (*btst): Rename into...
22770 (*btst<mode>): ...this and adjust.
22771 (*cbranchsi4_btst_insn): Rename into...
22772 (*cbranch<mode>4_btst_insn): ...this and adjust.
22773 (trap): New define_insn.
22775 2017-02-23 Jakub Jelinek <jakub@redhat.com>
22777 PR tree-optimization/79389
22778 * ifcvt.c (struct noce_if_info): Add rev_cond field.
22779 (noce_reversed_cond_code): New function.
22780 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
22781 reversed_comparison_code. Formatting fix.
22782 (noce_try_store_flag): Test rev_cond != NULL in addition to
22783 reversed_comparison_code.
22784 (noce_try_store_flag_constants): Likewise.
22785 (noce_try_store_flag_mask): Likewise.
22786 (noce_try_addcc): Use rev_cond if non-NULL instead of
22787 reversed_comparison_code.
22788 (noce_try_cmove_arith): Likewise. Formatting fixes.
22789 (noce_try_minmax, noce_try_abs): Clear rev_cond.
22790 (noce_find_if_block): Initialize rev_cond.
22791 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
22792 instead of false as last argument never attempt to reverse it
22795 2017-02-23 Bin Cheng <bin.cheng@arm.com>
22797 PR tree-optimization/79663
22798 * tree-predcom.c (combine_chains): Process refs in reverse order
22799 only for ZERO length chains, and add explaining comment.
22801 2017-02-23 Jeff Law <law@redhat.com>
22803 PR tree-optimization/79578
22804 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
22805 in call to operand_equal_p.
22807 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
22810 * config/i386/cpuid.h: Fix another undefined behavior.
22812 2017-02-23 Richard Biener <rguenther@suse.de>
22814 PR tree-optimization/79683
22815 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
22816 vector types for data-refs.
22818 2017-02-23 Martin Liska <mliska@suse.cz>
22820 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
22822 2017-02-23 Jakub Jelinek <jakub@redhat.com>
22824 PR middle-end/79665
22825 * internal-fn.c (get_range_pos_neg): Moved to ...
22826 * tree.c (get_range_pos_neg): ... here. No longer static.
22827 * tree.h (get_range_pos_neg): New prototype.
22828 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
22829 are known to be in between 0 and signed maximum inclusive, try to
22830 expand both unsigned and signed divmod and use the cheaper one from
22833 2017-02-22 Jeff Law <law@redhat.com>
22835 PR tree-optimization/79578
22836 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
22837 to compare base operands.
22839 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
22842 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
22843 gpc_reg_operand instead of fpr_reg_operand.
22845 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
22847 * config/mips/mips.c (mips_return_in_memory): Force FP
22848 vector types to be returned in memory for o32 ABI.
22850 2017-02-22 Jakub Jelinek <jakub@redhat.com>
22852 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
22853 instead of DW_TAG_member for static data member declarations and don't
22854 set no_linkage_name for static inline data members.
22855 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
22858 2017-02-22 Martin Liska <mliska@suse.cz>
22860 * doc/invoke.texi: Replace inequality signs with square brackets
22863 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22865 PR tree-optimization/68644
22866 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
22868 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
22871 * lra-constraints.c (simplify_operand_subreg): Handle
22872 WORD_REGISTER_OPERATIONS targets.
22874 2017-02-22 Jakub Jelinek <jakub@redhat.com>
22877 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
22878 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
22879 elimination by swapping fld*.
22881 2017-02-22 Richard Biener <rguenther@suse.de>
22883 PR tree-optimization/79673
22884 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
22885 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
22886 irrelevant address-space qualifiers and avoiding a
22887 ADDR_SPACE_CONVERT_EXPR from fold_convert.
22889 2017-02-22 Richard Biener <rguenther@suse.de>
22891 PR tree-optimization/79666
22892 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
22893 to not symbolically negate if that may introduce undefined
22896 2017-02-22 Martin Liska <mliska@suse.cz>
22899 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
22900 * data-streamer-out.c (streamer_write_gcov_count_stream):
22902 * value-prof.c (stream_out_histogram_value): Make assert more
22903 precise based on type of counter.
22905 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
22908 * config/i386/i386.md (standard_x87sse_constant_load splitter):
22909 Use nonimmediate_operand instead of memory_operand for operand 1.
22910 (float-extend standard_x87sse_constant_load splitter): Ditto.
22912 2017-02-21 Jeff Law <law@redhat.com>
22914 PR tree-optimization/79621
22915 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
22916 blocks with edges to themselves.
22918 2017-02-21 Jakub Jelinek <jakub@redhat.com>
22921 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
22922 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
22923 Use gimple_call_builtin_p.
22926 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
22927 on temporarily removed DEBUG_INSNs.
22929 PR tree-optimization/79649
22930 * tree-loop-distribution.c (classify_partition): Give up on
22931 non-generic address space loads/stores.
22933 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
22935 * doc/loop.texi (Loop manipulation): Remove nonexistent
22936 tree_ssa_loop_version from the documentation.
22937 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
22939 2017-02-21 Jakub Jelinek <jakub@redhat.com>
22942 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
22943 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
22944 * config/rs6000/rs6000.c: Include except.h.
22945 (rs6000_expand_split_stack_prologue): Call
22946 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
22948 2017-02-21 Martin Jambor <mjambor@suse.cz>
22951 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
22952 have been analyzed.
22954 2017-02-21 Martin Jambor <mjambor@suse.cz>
22956 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
22957 for backward compatibility only.
22958 * doc/invoke.texi (Option Summary): Remove all references to
22959 -fipa-cp-alignment.
22961 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
22965 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
22967 * lra-constraints.c (curr_insn_transform): Handle
22968 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
22970 2017-02-21 Martin Liska <mliska@suse.cz>
22972 * config/i386/i386.opt: Replace -masm-dialect with -masm.
22974 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
22976 PR translation/79638
22977 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
22979 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
22982 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
22983 (arm_function_ok_for_sibcall): Return false for an indirect call by
22984 descriptor if all the argument registers are used.
22985 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
22986 alignment of the function.
22988 2017-02-21 Jakub Jelinek <jakub@redhat.com>
22990 PR tree-optimization/61441
22991 * simplify-rtx.c (simplify_const_unary_operation): For
22992 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
22993 the sNaN unmodified.
22995 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
22997 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
22998 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
22999 instead of SYSTEM_HEADER_DIR.
23001 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
23002 Martin Liška <mliska@suse.cz>
23004 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
23005 Fix typos and grammar, use active voice, and clarify.
23007 2017-02-20 Marek Polacek <polacek@redhat.com>
23009 PR middle-end/79537
23010 * gimplify.c (gimplify_expr): Handle unused *&&L;.
23013 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
23015 2017-02-20 Jakub Jelinek <jakub@redhat.com>
23018 * config/i386/i386.c (ix86_expand_builtin): Handle
23019 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
23020 ix86_builtins_isa[fcode].isa as a requirement of those
23021 flags and any other flag in the bitmask.
23022 (ix86_init_mmx_sse_builtins): Use 0 instead of
23023 ~OPTION_MASK_ISA_64BIT as mask.
23024 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
23025 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
23026 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
23027 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
23029 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
23032 * lra-constraints.c (split_reg): Check requested split mode
23033 is supported by the register.
23035 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
23037 * lra-constraints.c (simplify_operand_subreg): Remove early
23040 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
23043 * lra-constraints.c (curr_insn_transform): Tighten condition
23044 for converting SUBREG reloads from OP_OUT to OP_INOUT.
23046 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
23049 * lra-constraints.c (curr_insn_transform): Handle
23050 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
23052 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
23055 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
23057 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
23059 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
23062 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
23065 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23067 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
23070 2017-02-18 Jakub Jelinek <jakub@redhat.com>
23072 * final.c (last_columnnum, override_columnnum): New variables.
23073 (final_start_function): Set last_columnnum, pass it to begin_prologue
23074 hook and pass 0 to dwarf2out_begin_prologue.
23075 (final_scan_insn): Update override_columnnum. Pass last_columnnum
23076 to source_line debug hook.
23077 (notice_source_line): Compute last_columnnum and for debug_column_info
23078 return true on column changes.
23079 * debug.h (struct gcc_debug_hooks): Add column argument to
23080 source_line and begin_prologue hooks.
23081 (debug_nothing_int_charstar_int_bool): Remove prototype.
23082 (debug_nothing_int_int_charstar,
23083 debug_nothing_int_int_charstar_int_bool): New prototypes.
23084 (dwarf2out_begin_prologue): Add column argument.
23085 * debug.c (do_nothing_debug_hooks): Adjust source_line and
23086 begin_prologue hooks.
23087 (debug_nothing_int_charstar_int_bool): Remove.
23088 (debug_nothing_int_int_charstar,
23089 debug_nothing_int_int_charstar_int_bool): New functions.
23090 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
23091 through to dwarf2out_source_line.
23092 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
23093 (dwarf2out_source_line): Add column argument, emit it if requested.
23094 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
23096 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
23097 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
23098 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
23099 through to dwarf2out_begin_prologue.
23100 (vmsdbgout_source_line): Add column argument, pass it through to
23101 dwarf2out_source_line.
23102 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
23103 dbxout_source_line caller.
23104 (dbxout_source_line): Add column argument.
23106 * common.opt (gno-column-info, gcolumn-info): New options.
23107 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
23108 (check_die): Also test for multiple DW_AT_decl_column attributes.
23109 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
23110 DW_AT_decl_column if requested.
23111 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
23113 (gen_variable_die): Likewise.
23114 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
23115 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
23118 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
23119 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
23120 (ix86_handle_option): Handle OPT_m3dnowa.
23121 * doc/invoke.texi (-m3dnowa): Document.
23122 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
23123 -m3dnowa instead of -m3dnow -march=athlon.
23126 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
23127 instead of gcc_assert for K, r and R code checks. Formatting fixes.
23129 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23132 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
23133 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
23134 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
23135 generator for vsx_xxpermdi_<mode>_be.
23136 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
23137 force big-endian semantics.
23138 (vsx_xxpermdi_<mode>_be): New define_expand with same
23139 implementation as previous version of vsx_xxpermdi_<mode>.
23141 2017-02-17 Jakub Jelinek <jakub@redhat.com>
23143 PR tree-optimization/79327
23144 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
23145 variable, its initialization and use.
23147 2017-02-17 Julia Koval <julia.koval@intel.com>
23149 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
23150 (OPTION_MASK_ISA_PKU_UNSET): New.
23151 (ix86_handle_option): Handle -mrdpid.
23152 * config/i386/cpuid.h (bit_RDPID): New.
23153 * config/i386/driver-i386.c (host_detect_local_cpu):
23154 Detect RDPID feature.
23155 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
23156 * config/i386/i386-c.c (ix86_target_macros_internal):
23158 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
23159 (ix86_valid_target_attribute_inner_p): Add "rdpid".
23160 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
23161 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
23162 * config/i386/i386.md (define_insn "rdpid"): New.
23163 * config/i386/i386.opt Add -mrdpid.
23164 * config/i386/immintrin.h (_rdpid_u32): New.
23166 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
23168 PR rtl-optimization/79541
23169 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
23170 instead of transforming it into USE.
23172 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
23174 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
23175 If HONOR_SNANS (SFmode) force the input to a register.
23176 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
23177 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
23178 an frsp or similar insn.
23180 2017-02-17 Martin Liska <mliska@suse.cz>
23182 PR rtl-optimization/79577
23183 * params.def (selsched-max-sched-times): Increase minimum to 1.
23185 2017-02-17 Martin Liska <mliska@suse.cz>
23187 PR rtl-optimization/79574
23188 * gcse.c (want_to_gcse_p): Prevent integer overflow.
23190 2017-02-17 Martin Liska <mliska@suse.cz>
23192 PR tree-optimization/79529
23193 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
23194 ssa_defined_default_def_p to handle cases which are implicitly
23196 * tree-ssa.c (ssa_defined_default_def_p): New function.
23197 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
23198 which are implicitly defined.
23199 * tree-ssa.h (ssa_defined_default_def_p): Declare.
23201 2017-02-17 Richard Biener <rguenther@suse.de>
23203 PR middle-end/79576
23204 * params.def (max-ssa-name-query-depth): Limit to 10.
23206 2017-02-17 Richard Biener <rguenther@suse.de>
23208 PR tree-optimization/79552
23209 * tree-ssa-structalias.c (visit_loadstore): Properly verify
23212 2017-02-17 Richard Biener <rguenther@suse.de>
23215 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
23217 2017-02-17 Marek Polacek <polacek@redhat.com>
23219 PR middle-end/79536
23220 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
23221 (fold_negate_expr): New wrapper.
23223 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
23225 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
23226 Correct terminology and de-emphasize pre-standard behavior.
23228 2017-02-16 Alan Modra <amodra@gmail.com>
23230 PR rtl-optimization/79286
23231 * ira.c (def_dominates_uses): New function.
23232 (update_equiv_regs): Don't create an equivalence for insns that
23233 may trap where the register def does not dominate the use.
23235 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
23237 PR rtl-optimization/78127
23238 * lra.c (lra): Call lra_eliminate before finish the loop after
23241 2017-02-16 Richard Biener <rguenther@suse.de>
23243 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
23245 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
23246 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
23247 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
23248 (isl_val_int_from_wi): New function.
23249 (extract_affine_gmp): Rename to ...
23250 (extract_affine_wi): ... this, take a widest_int.
23251 (extract_affine_int): Just wrap extract_affine_wi.
23252 (add_param_constraints): Use isl_val_int_from_wi.
23253 (add_loop_constraints): Likewise, and extract_affine_wi.
23255 2017-02-15 Jeff Law <law@redhat.com>
23257 PR middle-end/79521
23258 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
23259 ira_init_register_move_cost_if_necessary.
23261 2017-02-15 Martin Sebor <msebor@redhat.com>
23263 PR middle-end/32003
23264 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
23265 removed in a prior commit.
23267 2017-02-15 Bin Cheng <bin.cheng@arm.com>
23269 PR tree-optimization/79347
23270 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
23271 counters during peeling.
23273 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
23275 * Makefile.in (site.exp): Remove "set ISLVER".
23277 2017-02-15 Jakub Jelinek <jakub@redhat.com>
23280 * real.c (real_from_integer): Call real_convert even for decimal.
23282 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
23285 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
23287 2017-02-14 Andrew Pinski <apinski@cavium.com>
23289 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
23290 cores and change the partno/implementer to be correct.
23291 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
23292 the 'B" as the implementer.
23293 * config/aarch64/aarch64-tune.md: Regenerate.
23295 2017-02-14 Carl Love <cel@us.ibm.com>
23297 * config/rs6000/rs6000.c: Add case statement entry to make the
23298 xvcvuxdsp built-in argument unsigned.
23299 * config/rs6000/vsx.md: Fix the source and return operand types so they
23300 match the instruction definitions from the ISA document. Fix typo
23301 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
23304 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
23307 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
23308 member early_clobber_alts.
23309 * lra-lives.c (reg_early_clobber_p): New.
23310 (process_bb_lives): Use it.
23311 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
23312 (debug_operand_data): Initialize early_clobber_alts.
23313 (setup_operand_alternative): Set up early_clobber_alts.
23314 (collect_non_operand_hard_regs): Ditto. Pass early clobber
23315 alternatives to new_insn_reg.
23316 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
23318 (lra_update_insn_regno_info): Pass the new arg.
23320 2017-02-14 Jakub Jelinek <jakub@redhat.com>
23322 PR middle-end/79505
23323 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
23324 (new_oacc_loop_raw): Don't clear already cleared fields.
23327 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
23328 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
23329 _mm512_prefetch_i64gather_ps): New inline functions and macros.
23331 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
23334 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
23336 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
23339 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
23340 the extra instruction to the right place to store 128-bit constant
23343 2017-02-14 Martin Sebor <msebor@redhat.com>
23345 PR middle-end/79448
23346 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
23347 warning for strings of unknown length.
23349 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
23351 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
23353 2017-02-14 Jeff Law <law@redhat.com>
23356 * ira-costs.c (scan_one_insn): Initialize register move costs
23357 for pseudos seen in USE/CLOBBER insns.
23359 PR tree-optimization/79095
23360 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
23361 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
23362 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
23363 if the operands are known to be not equal, then the resulting range
23365 (intersect_ranges): If the new range is ~[0,0] and the old range is
23366 wide, then prefer ~[0,0].
23367 * tree-vrp.c (overflow_comparison_p_1): New function.
23368 (overflow_comparison_p): New function.
23369 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
23370 if NAME is used in an overflow test.
23371 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
23372 overflow check that can be expressed as an equality test, then adjust
23373 ops to be that equality test.
23375 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23377 * config/s390/s390-builtin-types.def: Remove flags argument.
23378 * config/s390/s390.c (s390_init_builtins): Likewise.
23380 2017-02-14 Martin Liska <mliska@suse.cz>
23382 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
23383 vector. Fix trailing white spaces.
23385 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
23387 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
23390 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23392 PR rtl-optimization/68664
23393 * config/arm/arm.c (arm_sched_can_speculate_insn):
23394 New function. Declare prototype.
23395 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
23397 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23399 PR rtl-optimization/68664
23400 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
23402 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
23404 2017-02-14 Amit Pawar <amit.pawar@amd.com>
23406 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
23407 max skip bytes for function, loop and jump.
23409 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23411 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
23412 ABS_EXPR for gimple dump.
23414 2017-02-14 Jakub Jelinek <jakub@redhat.com>
23417 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
23419 PR tree-optimization/79408
23420 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
23421 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
23422 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
23423 also if rhs1 is INTEGER_CST.
23425 2017-02-14 Richard Biener <rguenther@suse.de>
23427 PR middle-end/79432
23428 * tree-into-ssa.c (insert_phi_nodes): When the function can
23429 have abnormal edges rewrite SSA names with broken use-def
23430 dominance out of SSA and register them for PHI insertion.
23432 2017-02-13 Martin Sebor <msebor@redhat.com>
23434 PR middle-end/79496
23435 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
23436 clearing info.nowrite flag when snprintf size argument is a range.
23438 2017-02-13 Jakub Jelinek <jakub@redhat.com>
23440 * cprop.c (cprop_jump): Add missing space in string literal.
23441 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
23442 (get_constraint_for_component_ref): Likewise.
23443 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
23444 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
23445 * lra-constraints.c (process_alt_operands): Likewise.
23446 * ipa-inline.c (inline_small_functions): Likewise.
23447 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
23448 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
23449 * trans-mem.c (diagnose_tm_1_op): Likewise.
23450 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
23451 (grid_parallel_clauses_gridifiable): Likewise.
23453 * config/nvptx/mkoffload.c (process): Add space in between
23456 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
23457 "MOD4_SSE_REGS" and "ALL_REGS".
23459 * spellcheck.c (test_data): Add , in between "foo" and "food".
23461 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23464 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
23465 boundary crossing check and subsequent code generation agree.
23467 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23469 * config/aarch64/aarch64.c (has_memory_op): Delete.
23470 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
23473 2017-02-13 Jakub Jelinek <jakub@redhat.com>
23475 PR rtl-optimization/79388
23476 PR rtl-optimization/79450
23477 * combine.c (distribute_notes): When removing TEM_INSN for which
23478 corresponding dest has last value recorded, invalidate that last
23481 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23483 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
23484 of explicit '@'. Add missing assembly comment marker on branch costs
23487 2017-02-13 Nathan Sidwell <nathan@acm.org>
23489 * gengtype-lex.l (<in_struct>): Add '/'.
23491 2017-02-13 Martin Liska <mliska@suse.cz>
23494 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
23496 2017-02-13 Richard Biener <rguenther@suse.de>
23498 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
23500 * configure: Re-generate.
23501 * config.in: Likewise.
23502 * graphite-dependences.c: Simplify as if
23503 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
23504 * graphite-isl-ast-to-gimple.c: Likewise.
23505 * graphite-optimize-isl.c: Likewise.
23506 * graphite-poly.c: Likewise.
23507 * graphite-sese-to-poly.c: Likewise.
23508 * graphite.h: Likewise.
23509 * toplev.c: Include isl/version.h and use isl_version () for
23510 printing the ISL version.
23511 * doc/install.texi: Update ISL requirement.
23513 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
23515 * doc/standards.texi (Standards): Update reference to
23518 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
23520 * doc/extend.texi (Named Address Spaces): sourceware.org now
23522 * doc/install.texi (Binaries): Ditto.
23525 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
23527 * doc/cpp.texi: Replace "stringify"/"stringification" with C
23528 standard terminology "stringize"/"stringizing" throughout.
23529 * doc/cppinternals.texi: Likewise.
23531 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
23533 * doc/extend.texi: Fix some spelling mistakes and typos.
23534 * doc/invoke.texi: Likewise.
23536 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
23539 * params.def (inline-min-speedup) Change from 10 to 8.
23541 2017-02-11 Jakub Jelinek <jakub@redhat.com>
23543 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
23546 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
23549 * ipa-inline-analysis.c (get_minimal_bb): New function.
23550 (record_modified): Use it.
23551 (remap_edge_change_prob): Handle also ancestor functions.
23553 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
23555 * doc/contrib.texi (Contributors): Remove broken link into
23556 the Mauve CVS repository.
23558 2017-02-11 Jakub Jelinek <jakub@redhat.com>
23560 PR middle-end/79454
23561 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
23562 result computation whenever lhs doesn't have vector mode, not
23563 just when it has BLKmode.
23565 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
23567 * doc/makefile.texi (profiledbootstrap): Refer to the
23568 installation instructions only in textual form.
23570 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23573 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
23575 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
23577 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
23578 (Specific): Update mingw-w64 reference.
23580 (Specific): Remove broken link to Renesas RX processor.
23582 2017-02-10 Richard Biener <rguenther@suse.de>
23584 * toplev.c (process_options): Do not mention obsolete graphite
23585 options when printing sorry message about missing graphite support.
23586 Mention -floop-nest-optimize.
23588 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
23590 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
23591 (vtst_p16): Likewise.
23592 (vtstq_p8): Likewise.
23593 (vtstq_p16): Likewise.
23595 (vtstq_p64): Likewise.
23596 * config/arm/arm_neon.h (vgetq_lane_p64): New.
23597 (vset_lane_p64): New.
23598 (vsetq_lane_p64): New.
23600 2017-02-10 Jakub Jelinek <jakub@redhat.com>
23602 PR tree-optimization/79411
23603 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
23604 stmt operands are SSA_NAMEs used in abnormal phis.
23605 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
23608 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
23611 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
23614 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
23616 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
23618 2017-02-09 Jakub Jelinek <jakub@redhat.com>
23620 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
23624 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
23625 not arbitrary TREE_CONSTANT.
23628 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
23629 "omp declare target link" attribute unless is_global_var.
23630 * omp-offload.c (find_link_var_op): Likewise.
23632 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
23633 Chung-Lin Tang <cltang@codesourcery.com>
23635 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
23637 (gimplify_adjust_omp_clauses): Don't delete TILE.
23638 (gimplify_omp_for): Deal with TILE.
23639 * internal-fn.c (expand_GOACC_TILE): New function.
23640 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
23642 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
23643 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
23645 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
23646 avoid DIV for outermost collapse var.
23647 (expand_oacc_for): Insert tile element loop as needed. Adjust.
23648 Remove out of date comments, fix whitespace.
23649 * omp-general.c (omp_extract_for_data): Deal with tiling.
23650 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
23651 adjust OLF_DIM_BASE value.
23652 (struct omp_for_data): Add tiling field.
23653 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
23654 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
23655 for auto loops. Remove default auto determining, moved to
23656 oacc_loop_fixed_partitions.
23657 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
23658 stmts, add e_mask field.
23659 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
23660 (oacc_thread_numbers): Use oacc_dim_call.
23661 (oacc_xform_tile): New.
23662 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
23663 (finish_oacc_loop): Adjust for ifns vector.
23664 (oacc_loop_discover_walk): Append loop abstraction sites to list,
23665 add case for GOACC_TILE fns.
23666 (oacc_loop_xform_loop): Delete.
23667 (oacc_loop_process): Iterate over call list directly, and add
23668 handling for GOACC_TILE fns.
23669 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
23671 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
23672 vector partitioning to outer loops. Assign 2 partitions to loops
23673 when available. Add TILE handling.
23674 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
23675 (execite_oacc_device_lower): Process GOACC_TILE fns,
23676 ignore unknown specs.
23677 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
23678 * tree.c (omp_clause_num_ops): Adjust TILE ops.
23679 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
23681 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
23683 * configure.ac (ACX_BUGURL): Update.
23684 * configure: Regenerate.
23686 2017-02-09 Richard Biener <rguenther@suse.de>
23688 PR tree-optimization/69823
23689 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
23690 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
23692 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
23694 * config/arc/arc-c.def: Add __NPS400__ definition.
23695 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
23696 (TARGET_NPS400): Define.
23698 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
23700 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
23702 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
23703 pointer, arch_info.
23704 (arc_cpu_types): Fill the arch_info field with a pointer into the
23705 arc_arch_types table.
23706 (arc_selected_cpu): Declare.
23707 * config/arc/arc.c (arc_selected_cpu): Make global.
23708 (arc_selected_arch): Delete.
23709 (arc_base_cpu): Delete.
23710 (arc_override_options): Remove references to deleted variables,
23711 update access to arch information.
23712 (ARC_OPT): Update access to arch information.
23713 (ARC_OPTX): Likewise.
23714 * config/arc/arc.h (arc_base_cpu): Remove declaration.
23715 (TARGET_ARC600): Update access to arch information.
23716 (TARGET_ARC601): Likewise.
23717 (TARGET_ARC700): Likewise.
23718 (TARGET_EM): Likewise.
23719 (TARGET_HS): Likewise.
23720 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
23723 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
23726 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
23727 condition/operands for integer GE/LE/GEU/LEU operations.
23729 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
23731 PR translation/79397
23732 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
23735 2017-02-08 Martin Jambor <mjambor@suse.cz>
23738 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
23739 whether allocation happened.
23740 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
23741 nothing was allocated.
23743 2017-02-08 Jakub Jelinek <jakub@redhat.com>
23745 PR tree-optimization/79408
23746 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
23747 constant, but SSA_NAME with a known integer range, use the minimum
23748 of that range instead of op1 to determine if modulo can be replaced
23749 with its first operand.
23751 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23753 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
23755 2017-02-08 Richard Biener <rguenther@suse.de>
23757 PR tree-optimization/71824
23758 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
23759 Check all loops contained in the merged region.
23761 2017-02-07 Andrew Pinski <apinski@cavium.com>
23763 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
23765 2017-02-07 Andrew Pinski <apinski@cavium.com>
23767 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
23768 (thunderxt88): Likewise.
23769 (thunderxt81): Disable LSE and change v8.1 to v8.
23770 (thunderxt83): Likewise.
23772 2017-02-07 Jakub Jelinek <jakub@redhat.com>
23773 Richard Biener <rguenther@suse.de>
23775 PR middle-end/79399
23776 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
23777 type from int to size_t.
23778 * ira-costs.c (struct_costs_size): Change type from int to size_t.
23780 2017-02-07 Jakub Jelinek <jakub@redhat.com>
23782 PR rtl-optimization/79386
23783 * cprop.c (bypass_conditional_jumps): Initialize
23784 bypass_last_basic_block already before splitting bbs after
23785 unconditional traps...
23786 (bypass_conditional_jumps): ... rather than here.
23789 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
23790 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
23791 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
23792 fix -masm=intel patterns.
23794 2017-02-07 Richard Biener <rguenther@suse.de>
23796 PR tree-optimization/79256
23797 PR middle-end/79278
23798 * builtins.c (get_object_alignment_2): Use min_align_of_type
23799 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
23800 and ADJUST_FIELD_ALIGN.
23802 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
23804 * doc/tm.texi: Regenerate.
23805 * stor-layout.c (layout_decl): Adjust.
23806 (update_alignment_for_field): Likewise.
23807 (place_field): Likewise.
23808 (min_align_of_type): Likewise.
23809 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
23810 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
23811 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
23812 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
23813 * config/frv/frv.c (frv_adjust_field_align): Likewise.
23814 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
23815 * config/i386/i386.c (x86_field_alignment): Likewise.
23816 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
23817 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
23818 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
23819 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
23820 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
23821 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
23825 2017-01-30 Richard Biener <rguenther@suse.de>
23827 PR tree-optimization/79256
23828 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
23829 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
23832 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
23834 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
23835 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
23836 builtins to SImode and emit a zero-extend, if necessary.
23838 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
23840 * docs/invoke.texi (RISC-V Options): Alphabetize.
23842 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
23844 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
23847 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
23849 * config/riscv/riscv.c: New file.
23850 * common/config/riscv/riscv-common.c: Likewise.
23851 * config.gcc: Likewise.
23852 * config/riscv/constraints.md: Likewise.
23853 * config/riscv/elf.h: Likewise.
23854 * config/riscv/generic.md: Likewise.
23855 * config/riscv/linux.h: Likewise.
23856 * config/riscv/multilib-generator: Likewise.
23857 * config/riscv/peephole.md: Likewise.
23858 * config/riscv/pic.md: Likewise.
23859 * config/riscv/predicates.md: Likewise.
23860 * config/riscv/riscv-builtins.c: Likewise.
23861 * config/riscv/riscv-c.c: Likewise.
23862 * config/riscv/riscv-ftypes.def: Likewise.
23863 * config/riscv/riscv-modes.def: Likewise.
23864 * config/riscv/riscv-opts.h: Likewise.
23865 * config/riscv/riscv-protos.h: Likewise.
23866 * config/riscv/riscv.h: Likewise.
23867 * config/riscv/riscv.md: Likewise.
23868 * config/riscv/riscv.opt: Likewise.
23869 * config/riscv/sync.md: Likewise.
23870 * config/riscv/t-elf-multilib: Likewise.
23871 * config/riscv/t-linux: Likewise.
23872 * config/riscv/t-linux-multilib: Likewise.
23873 * config/riscv/t-riscv: Likewise.
23874 * configure.ac: Likewise.
23875 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
23876 Waterman as RISC-V maintainers.
23877 * doc/install.texi: Add RISC-V entries.
23878 * doc/invoke.texi: Add RISC-V options section.
23879 * doc/md.texi: Add RISC-V constraints section.
23880 * configure: Regenerated.
23882 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
23885 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
23886 false values to be constant vectors with all 0 or all 1 bits set.
23887 (vcondu<mode><mode>): Likewise.
23888 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
23890 (fpmask_comparison_operator): Update comment.
23891 (vecint_comparison_operator): New predicate.
23892 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
23893 vector conditionals when the true and false values are constant
23894 vectors with all 0 bits or all 1 bits set.
23896 2017-02-06 Martin Sebor <msebor@redhat.com>
23898 PR tree-optimization/79376
23899 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
23901 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
23903 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
23904 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
23905 to simplify split condition.
23907 2017-02-06 Jakub Jelinek <jakub@redhat.com>
23909 * omp-expand.c (oxpand_omp_atomic_fetch_op,
23910 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
23913 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
23915 PR rtl-optimization/68664
23916 * target.def (can_speculate_insn): New hook.
23917 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
23918 * doc/tm.texi: Regenerate.
23919 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
23920 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
23921 (rs6000_sched_can_speculate_insn): New function.
23923 2017-02-06 Jakub Jelinek <jakub@redhat.com>
23925 PR tree-optimization/79284
23926 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
23927 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
23928 vectorizable_mask_load_store, vectorizable_operation,
23929 vect_is_simple_cond, get_same_sized_vectype): Use it instead
23930 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
23931 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
23932 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
23933 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
23934 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
23935 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
23936 is_gimple_assign (stmt). Replace another such test with
23937 is_gimple_assign (stmt).
23939 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
23942 * config/avr/avr.c (rtl-iter.h): Include it.
23943 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
23944 (avr_legitimate_combined_insn): ...and implementation.
23946 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
23948 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
23949 * config/s390/s390.c (s390_const_operand_ok)
23950 (s390_canonicalize_comparison, s390_extract_part)
23951 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
23952 (s390_contiguous_bitmask_p, s390_rtx_costs)
23953 (legitimize_pic_address): Likewise.
23954 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
23955 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
23956 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
23957 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
23958 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
23960 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
23962 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
23963 REGNO($0) == REGNO($1).
23965 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23967 * config/s390/linux.h(SIZE_TYPE): Add comment.
23969 2017-02-06 Julian Brown <julian@codesourcery.com>
23970 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23971 Virendra Pathak <virendra.pathak@broadcom.com>
23973 * config/aarch64/aarch64-cores.def: Change the scheduler
23975 * config/aarch64/aarch64.md: Include thunderx2t99.md.
23976 * config/aarch64/thunderx2t99.md: New file.
23978 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
23980 * doc/standards.texi (Go Language): Update link to language
23983 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
23985 * tree-eh.c (lower_resx): Sanitize profile.
23986 (cleanup_empty_eh_move_lp): Likewise.
23988 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
23991 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
23993 * cfgloopmanip.h (loop_version): Update prototype.
23994 * modulo-sched.c (sms_schedule): Update call of loop_version.
23995 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
23996 * tree-parloops.c (gen_parallel_loop): Likewise.
23997 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
23998 * tree-ssa-loop-split.c (split_loop): Likewise.
23999 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
24000 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
24002 2017-02-05 Martin Liska <mliska@suse.cz>
24005 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
24007 (print_operand_address): Initialize a struct to zero.
24009 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
24011 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
24012 garbage collector only in textual form.
24014 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
24016 * doc/extend.texi (x86 specific memory model extensions for
24017 transactional memory): Simplify a phrase.
24019 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
24022 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
24023 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
24024 (atomic_storedi_1): Likewise.
24026 2017-02-04 Jakub Jelinek <jakub@redhat.com>
24028 PR tree-optimization/79338
24029 * tree-parloops.c (gather_scalar_reductions): Don't call
24030 vect_analyze_loop_form for loop->inner before destroying loop's
24033 2017-02-03 Martin Sebor <msebor@redhat.com>
24035 PR tree-optimization/79327
24036 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
24037 when precision has resulted in leading zeros.
24038 (format_integer): Adjust the likely counter to assume an unknown
24039 argument that may be zero is non-zero.
24041 2017-02-03 Jason Merrill <jason@redhat.com>
24044 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
24045 avoid copying non-taken branch.
24047 2017-02-03 Jakub Jelinek <jakub@redhat.com>
24049 PR tree-optimization/79340
24050 * tree-vect-loop.c (vectorizable_reduction): Release
24051 vec_defs elements after safe_splicing them into other vectors.
24054 PR tree-optimization/79327
24055 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
24056 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
24058 (format_integer): Use wide_int_to_tree instead of build_int_cst
24059 + to_?hwi. If argmin is NULL, just set argmin and argmax to
24060 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
24061 of shortest and longest sequence.
24063 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
24065 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
24066 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
24068 2017-02-03 Walter Lee <walt@tilera.com>
24071 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
24072 after initial stackframe link reg save.
24073 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
24075 2017-02-03 Jakub Jelinek <jakub@redhat.com>
24078 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
24079 wu for stxssp alternative.
24081 2017-02-03 Martin Sebor <msebor@redhat.com>
24083 PR tree-optimization/79352
24084 * gimple-fold.c (get_range_strlen): Add argument.
24085 (get_range_strlen): Change return type to bool.
24086 (get_maxval_strlen): Pass in a dummy argument.
24087 * gimple-fold.h (get_range_strlen): Change return type to bool.
24088 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
24089 * tree.h (array_at_struct_end_p): Add argument.
24090 * tree.c (array_at_struct_end_p): Handle it.
24092 2017-02-03 Martin Liska <mliska@suse.cz>
24095 * multiple_target.c (create_dispatcher_calls): Redirect edge
24096 from a caller of a dispatcher.
24097 (expand_target_clones): Make the clones local.
24098 (ipa_target_clone): Do both target clones and resolvers.
24099 (ipa_dispatcher_calls): Remove the pass.
24100 (pass_dispatcher_calls::gate): Likewise.
24101 (make_pass_dispatcher_calls): Likewise.
24102 * passes.def (pass_target_clone): Put as very first IPA early
24105 2017-02-03 Martin Liska <mliska@suse.cz>
24107 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
24108 in case of a function with ifunc attribute.
24110 2017-02-03 Martin Liska <mliska@suse.cz>
24112 * cgraph.c (cgraph_node::dump): Dump function version info.
24113 * symtab.c (symtab_node::dump_base): Add missing new line.
24115 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
24117 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
24118 (ifcombine_ifandif): Use it.
24120 2017-02-03 Martin Liska <mliska@suse.cz>
24122 * doc/invoke.texi: Document default value for
24123 use-after-scope-direct-emission-threshold.
24125 2017-02-03 Martin Liska <mliska@suse.cz>
24127 PR tree-optimization/79339
24128 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
24129 (format_floating): Likewise.
24131 2017-02-03 Martin Liska <mliska@suse.cz>
24134 * ipa-prop.c (ipa_node_params_t::insert): Remove current
24136 (ipa_node_params_t::remove): Likewise.
24137 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
24138 initialization from removed ipa_node_params_t::insert.
24139 (ipa_node_params::~ipa_node_params): Move from removed
24140 ipa_node_params_t::release.
24141 * symbol-summary.h (symbol_summary::m_released): New member.
24142 Do not release a summary twice. Do not allow to call finalizer
24143 for types of a summary that live in GGC memory.
24145 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24147 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
24150 2017-02-02 Martin Sebor <msebor@redhat.com>
24152 PR middle-end/79275
24153 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
24154 (format_string): Tighten up the range of output for non-constant
24155 strings and correct the expected range for wide non-constant strings.
24157 2017-02-02 Martin Sebor <msebor@redhat.com>
24159 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
24161 PR middle-end/32003
24162 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
24164 (-fdump-tree-@var): Add to index and document how to come up
24165 with pass-specific option and dump file names.
24166 (-fdump-passes): Clarify where to look for output.
24168 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
24170 PR middle-end/77445
24171 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
24172 statistics of the analyzed path; allow threading for speed when
24173 any of BBs along the path are optimized for speed.
24175 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
24177 PR middle-end/78468
24178 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
24179 settings of the virtual registers.
24182 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
24184 * explow.c (get_dynamic_stack_size): Take known alignment of stack
24185 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
24188 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24190 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
24191 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
24193 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24195 * config/s390/s390.md: Add missing comments with the expanded
24197 * config/s390/vector.md: Likewise.
24198 * config/s390/vx-builtins.md: Likewise.
24200 2017-02-02 Jakub Jelinek <jakub@redhat.com>
24203 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
24204 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
24205 conditions on a single line.
24207 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24209 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
24210 __S390_VX__ to __VX__.
24212 2017-02-01 Andrew Pinski <apinski@cavium.com>
24214 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
24215 stmt_info to record_stmt_cost.
24216 (vect_get_known_peeling_cost): Pass stmt_info if known to
24218 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
24219 cpu_vector_cost field into
24220 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
24221 field into vec_int_stmt_cost and vec_fp_stmt_cost.
24222 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
24223 splitting of scalar_stmt_cost and vec_stmt_cost.
24224 (thunderx_vector_cost): Likewise.
24225 (cortexa57_vector_cost): LIkewise.
24226 (exynosm1_vector_cost): Likewise.
24227 (xgene1_vector_cost): Likewise.
24228 (thunderx2t99_vector_cost): Improve after the splitting of the two
24230 (aarch64_builtin_vectorization_cost): Update for the splitting of
24231 scalar_stmt_cost and vec_stmt_cost.
24233 2017-02-01 Torvald Riegel <triegel@redhat.com>
24234 Richard Henderson <rth@redhat.com>
24236 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
24237 conditional on existance of a fast atomic load.
24238 * optabs-query.c (can_atomic_load_p): New function.
24239 * optabs-query.h (can_atomic_load_p): Declare it.
24240 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
24241 no fast atomic load is available for the particular size of access.
24242 (expand_atomic_compare_and_swap): Likewise.
24243 (expand_atomic_load): Likewise.
24244 (expand_atomic_store): Likewise.
24245 (expand_atomic_fetch_op): Likewise.
24246 * testsuite/lib/target-supports.exp
24247 (check_effective_target_sync_int_128): Remove x86 because it provides
24248 no fast atomic load.
24249 (check_effective_target_sync_int_128_runtime): Likewise.
24251 2017-02-01 Richard Biener <rguenther@suse.de>
24253 * graphite.c: Include tree-vectorizer.h for find_loop_location.
24254 (graphite_transform_loops): Provide opt-info for optimized nests.
24255 * tree-parloop.c (parallelize_loops): Provide opt-info for
24256 parallelized loops.
24258 2017-02-01 Richard Biener <rguenther@suse.de>
24260 PR middle-end/79315
24261 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
24262 was not set before.
24264 2017-02-01 Richard Biener <rguenther@suse.de>
24266 PR tree-optimization/71824
24267 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
24268 Verify the loops are valid in the merged SESE region.
24269 (scop_detection::can_represent_loop_1): Check analyzing the
24270 evolution of the number of iterations in the region succeeds.
24272 2017-01-31 Ian Lance Taylor <iant@golang.org>
24274 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
24275 REG_ARGS_SIZE note to 32-bit push insns and call insn.
24277 2017-01-31 David Malcolm <dmalcolm@redhat.com>
24279 PR preprocessor/79210
24280 * input.c (get_substring_ranges_for_loc): Replace line_width
24281 assertion with error-handling.
24283 2017-01-31 Richard Biener <rguenther@suse.de>
24285 PR tree-optimization/77318
24286 * graphite-sese-to-poly.c (extract_affine): Fix assert.
24287 (create_pw_aff_from_tree): Take loop parameter.
24288 (add_condition_to_pbb): Pass loop of the condition to
24289 create_pw_aff_from_tree.
24291 2017-01-31 Jakub Jelinek <jakub@redhat.com>
24293 * config/s390/s390.c (s390_asan_shadow_offset): New function.
24294 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
24296 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
24300 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
24302 (convert_int_to_float128): Likewise.
24303 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
24304 (convert_int_to_float128): Likewise.
24305 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
24306 (UNSPEC_IEEE128_CONVERT): Likewise.
24307 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
24308 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
24309 Use local variables for IBM extended format.
24310 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
24311 (fix_trunc<mode>si2_fprs): Likewise.
24312 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
24313 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
24314 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
24315 to know that we can now have integers of all sizes in vector
24317 (fix<uns>_<mode>di2_hw): Likewise.
24318 (float<uns>_<mode>si2_hw): Likewise.
24319 (fix_<mode>si2_hw): Likewise.
24320 (fixuns_<mode>si2_hw): Likewise.
24321 (float<uns>_<mode>di2_hw): Likewise.
24322 (float_<mode>di2_hw): Likewise.
24323 (float_<mode>si2_hw): Likewise.
24324 (floatuns_<mode>di2_hw): Likewise.
24325 (floatuns_<mode>si2_hw): Likewise.
24326 (xscvqp<su>wz_<mode>): Delete, no longer used.
24327 (xscvqp<su>dz_<mode>): Likewise.
24328 (xscv<su>dqp_<mode>): Likewise.
24329 (ieee128_mfvsrd_64bit): Likewise.
24330 (ieee128_mfvsrd_32bit): Likewise.
24331 (ieee128_mfvsrwz): Likewise.
24332 (ieee128_mtvsrw): Likewise.
24333 (ieee128_mtvsrd_64bit): Likewise.
24334 (ieee128_mtvsrd_32bit): Likewise.
24336 2017-01-31 Martin Liska <mliska@suse.cz>
24339 * ipa-prop.c (ipa_free_all_node_params): Call release method
24340 instead of ~sumbol_summary to not to trigger double times
24343 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
24345 PR tree-optimization/71691
24346 * bitmap.h (class auto_bitmap): New.
24347 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
24348 is_maybe_undefined instead of ssa_undefined_value_p.
24350 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24352 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
24353 __S390_ARCH_LEVEL__ to __ARCH__.
24355 2017-01-31 Jakub Jelinek <jakub@redhat.com>
24357 PR tree-optimization/79267
24358 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
24359 if should_remove_lhs_p is true.
24361 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
24364 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
24365 (add_alignment_attribute): New.
24366 (base_type_die): Add alignment attribute.
24367 (subrange_type_die): Likewise.
24368 (modified_type_die): Likewise.
24369 (gen_array_type_die): Likewise.
24370 (gen_descr_array_type_die: Likewise.
24371 (gen_enumeration_type_die): Likewise.
24372 (gen_subprogram_die): Likewise.
24373 (gen_variable_die): Likewise.
24374 (gen_field_die): Likewise.
24375 (gen_ptr_to_mbr_type_die): Likewise.
24376 (gen_struct_or_union_type_die): Likewise.
24377 (gen_subroutine_type_die): Likewise.
24378 (gen_typedef_die): Likewise.
24379 (base_type_cmp): Compare alignment attribute.
24381 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24384 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
24385 (setb_unsigned) New pattern for setb with CCUNS.
24386 * config/rs6000/rs6000.c (expand_block_compare): Use a different
24387 subfc./subfe sequence to avoid overflow problems. Generate a
24388 shorter sequence with cmpld/setb for power9.
24389 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
24390 for generating subfc. instruction.
24391 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
24392 now uses this instruction.
24394 2017-01-30 Ian Lance Taylor <iant@google.com>
24397 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
24398 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
24400 2017-01-30 Martin Sebor <msebor@redhat.com>
24402 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
24403 Move constant to the right of a relational operator.
24404 (get_mpfr_format_length, format_character, format_string): Ditto.
24405 (should_warn_p, maybe_warn): Same.
24407 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
24409 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
24412 * asan.c (get_translation_unit_decl): Remove function.
24413 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
24415 2017-01-30 Martin Liska <mliska@suse.cz>
24417 PR gcov-profile/79259
24418 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
24419 -fprofile-generate.
24421 2017-01-30 Martin Liska <mliska@suse.cz>
24424 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
24425 Initialize variables with NULL value.
24427 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
24430 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
24432 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
24434 2017-01-30 Richard Biener <rguenther@suse.de>
24436 PR tree-optimization/79276
24437 * tree-vrp.c (process_assert_insertions): Properly adjust common
24438 when removing a duplicate.
24440 * gcc.dg/torture/pr79276.c: New testcase.
24442 2017-01-30 Richard Biener <rguenther@suse.de>
24444 PR tree-optimization/79256
24445 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
24446 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
24448 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
24450 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
24453 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
24454 ("*r<noxa>sbg_<mode>_sll_bitmask")
24455 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
24456 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
24457 Use contiguous_bitmask_nowrap_operand.
24459 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24462 * config/rs6000/altivec.h (vec_xl): Revise #define.
24463 (vec_xst): Likewise.
24465 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
24467 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
24469 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
24471 PR rtl-optimization/79194
24472 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
24473 traps before call to bypass_conditional_jumps.
24475 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
24477 PR tree-optimization/71374
24478 * lra-constraints.c (check_conflict_input_operands): New.
24479 (match_reload): Use it.
24481 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
24484 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
24485 account to calculate conflict_set.
24487 2017-01-27 Bin Cheng <bin.cheng@arm.com>
24489 PR rtl-optimization/78559
24490 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
24491 other_insn in combine.
24493 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
24495 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
24496 uint16_type_node for BT_UINT16.
24498 2017-01-27 David Malcolm <dmalcolm@redhat.com>
24500 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
24501 "RTL Tests" to menu.
24502 (GIMPLE Tests): New node.
24503 (RTL Tests): New node.
24505 2017-01-27 Richard Biener <rguenther@suse.de>
24507 PR tree-optimization/79245
24508 * tree-loop-distribution.c (distribute_loop): Apply cost
24509 modeling also to detected patterns.
24511 2017-01-27 Richard Biener <rguenther@suse.de>
24513 PR tree-optimization/71433
24514 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
24515 (compare_assert_loc): New function.
24516 (process_assert_insertions): Sort and optimize assert locations
24517 to remove duplicates and push down identical assertions on
24518 edges to their destination block.
24520 2017-01-27 Richard Biener <rguenther@suse.de>
24522 PR tree-optimization/79244
24523 * tree-vrp.c (remove_range_assertions): Forcefully propagate
24524 out SSA names even if abnormal.
24526 2017-01-27 Jakub Jelinek <jakub@redhat.com>
24528 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
24529 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
24530 instead of MPFR_RNDN.
24532 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
24535 * arm.c (arm_option_override): Don't call build_target_option_node
24536 until after doing all option overrides.
24537 (arm_valid_target_attribute_tree): Likewise.
24539 2017-01-27 Martin Liska <mliska@suse.cz>
24541 * doc/invoke.texi (-fprofile-arcs): Document profiling support
24542 for {cd}tors and C++ {cd}tors.
24544 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
24546 * config/s390/s390.md ("*setmem_long_and")
24547 ("*setmem_long_and_31z"): Use zero_extend instead of and.
24549 2017-01-26 Martin Sebor <msebor@redhat.com>
24551 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
24554 2017-01-26 Martin Sebor <msebor@redhat.com>
24556 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
24557 HAVE_DFmode before using XFmode or DFmode.
24558 (parse_directive): Avoid using the z length modifier to avoid
24559 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
24561 PR middle-end/78703
24562 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
24563 to accept adjustment as an array.
24564 (get_int_range): New function.
24565 (struct directive): Make width and prec arrays.
24566 (directive::set_width, directive::set_precision): Call get_int_range.
24567 (format_integer, format_floating): Handle width and precision ranges.
24568 (format_string, parse_directive): Same.
24570 2017-01-26 Jakub Jelinek <jakub@redhat.com>
24573 * dwarf2out.c (generate_skeleton_bottom_up): For children with
24574 comdat_type_p set, just clone them, but keep the children in the
24578 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
24579 which have direct callers with -fvar-tracking-assignments enabled
24581 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
24582 inside of type units.
24584 2017-01-26 Martin Sebor <msebor@redhat.com>
24586 PR middle-end/78703
24587 * gimple-ssa-sprintf.c (struct result_range): Add likely and
24589 (struct format_result): Replace number_chars, number_chars_min,
24590 and number_chars_max with a single member of struct result_range.
24592 (format_result::operator+=): Adjust.
24593 (struct fmtresult): Remove bounded. Handle likely and unlikely
24595 (fmtresult::adjust_for_width_or_precision): New function.
24596 (fmtresult:type_max_digits): New function.
24597 (bytes_remaining): Handle likely and unlikely counters.
24598 (min_bytes_remaining): Remove.
24599 (format_percent): Simplify.
24600 (format_integer, format_floating): Set likely and unlikely counters.
24601 (get_string_length, format_character, format_string): Same.
24602 (format_plain, should_warn_p): New function.
24603 (maybe_warn): Call should_warn_p. Update diagnostic messages
24604 and handle those for all directives, including plain strings.
24605 (format_directive): Handle likely and unlikely counters.
24606 Remove unnecessary quoting from diagnostics. Add an informational
24608 (add_bytes): Remove.
24609 (pass_sprintf_length::compute_format_length): Simplify.
24610 (try_substitute_return_value): Handle likely and unlikely counters.
24612 2017-01-26 Carl Love <cel@us.ibm.com>
24614 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
24615 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
24617 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
24620 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
24621 endianess for subregs into account.
24622 * lra-constraints.c (lra_constraints): Do risky transformations
24623 always on the first iteration.
24624 * lra-lives.c (check_pseudos_live_through_calls): Add arg
24625 last_call_used_reg_set.
24626 (process_bb_lives): Define and use last_call_used_reg_set.
24627 * lra.c (lra): Always continue after lra_constraints on the first
24630 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
24632 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
24634 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
24636 2017-01-26 Jakub Jelinek <jakub@redhat.com>
24638 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
24639 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
24640 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
24641 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
24642 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
24643 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
24644 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
24645 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
24646 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
24648 2017-01-26 Marek Polacek <polacek@redhat.com>
24651 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
24652 for the third operand.
24654 2017-01-26 Jakub Jelinek <jakub@redhat.com>
24656 PR middle-end/79236
24657 * omp-low.c (struct omp_context): Add simt_stmt field.
24658 (scan_omp_for): Return omp_context *.
24659 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
24660 context to the _simt_ SIMD stmt.
24661 (lower_omp_for): For combined SIMD with sibling _simt_
24662 SIMD, make sure to use the same decls in _looptemp_
24663 clauses as in the sibling.
24665 2017-01-26 David Sherwood <david.sherwood@arm.com>
24667 PR middle-end/79212
24668 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
24671 2017-01-26 Jakub Jelinek <jakub@redhat.com>
24674 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
24675 emit fld b; fld a; if possible.
24677 * brig-builtins.def: Update copyright years.
24678 * config/arm/arm_acle_builtins.def: Update copyright years.
24680 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
24683 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
24684 constraint instead of o for the stxsd instruction.
24686 2017-01-25 Carl Love <cel@us.ibm.com>
24688 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
24689 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
24691 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
24693 * doc/invoke.texi (C++ Dialect Options): Fix typo.
24695 2017-01-25 Richard Biener <rguenther@suse.de>
24697 PR tree-optimization/69264
24698 * target.def (vector_alignment_reachable): Improve documentation.
24699 * doc/tm.texi: Regenerate.
24700 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
24702 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
24703 earlier changes with respect to TYPE_USER_ALIGN.
24704 (vector_alignment_reachable_p): Likewise. Improve dumping.
24706 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24709 * config/arm/arm.md (xordi3): Force constant operand into a register
24712 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24714 * doc/invoke.texi (-fstore-merging): Correct default optimization
24715 levels at which it is enabled.
24716 (-O): Move -fstore-merging from list to...
24719 2017-01-25 Richard Biener <rguenther@suse.de>
24722 * omp-expand.c: Include debug.h.
24723 (expand_omp_taskreg): Make sure to generate early debug before
24724 outlining anything from a function.
24725 (expand_omp_target): Likewise.
24726 (grid_expand_target_grid_body): Likewise.
24728 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
24731 * asan.c (get_translation_unit_decl): New function.
24732 (asan_add_global): Extract modules file name from globals
24733 TRANSLATION_UNIT_DECL name.
24735 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
24738 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
24739 for long calls with APCS frame and VFP.
24741 2017-01-24 David Malcolm <dmalcolm@redhat.com>
24743 * cfg.c (original_copy_tables_initialized_p): New function.
24744 * cfg.h (original_copy_tables_initialized_p): New decl.
24745 * cfgrtl.c (relink_block_chain): Guard the call to
24746 free_original_copy_tables with a call to
24747 original_copy_tables_initialized_p.
24748 * cgraph.h (symtab_node::native_rtl_p): New decl.
24749 * cgraphunit.c (symtab_node::native_rtl_p): New function.
24750 (symtab_node::needed_p): Don't assert for early assembly output
24751 for __RTL functions.
24752 (cgraph_node::finalize_function): Set "force_output" for __RTL
24754 (cgraph_node::analyze): Bail out early for __RTL functions.
24755 (analyze_functions): Update assertion to support __RTL functions.
24756 (cgraph_node::expand): Bail out early for __RTL functions.
24757 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
24759 * function.h (struct function): Update comment for field
24761 * gimple-expr.c: Include "tree-pass.h".
24762 (gimple_has_body_p): Return false for __RTL functions.
24763 * Makefile.in (OBJS): Add run-rtl-passes.o.
24764 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
24766 (gcc::pass_manager::get_clean_slate): New accessor.
24767 * passes.c: Include "insn-addr.h".
24768 (should_skip_pass_p): Add logging. Update logic for running
24769 "expand" to be compatible with both __GIMPLE and __RTL. Guard
24770 property-provider override so it is only done for gimple passes.
24772 (skip_pass): New function.
24773 (execute_one_pass): Call skip_pass when skipping passes.
24774 * read-md.c (md_reader::read_char): Support filtering
24775 the input to a subset of line numbers.
24776 (md_reader::md_reader): Initialize fields
24777 m_first_line and m_last_line.
24778 (md_reader::read_file_fragment): New function.
24779 * read-md.h (md_reader::read_file_fragment): New decl.
24780 (md_reader::m_first_line): New field.
24781 (md_reader::m_last_line): New field.
24782 * read-rtl-function.c (function_reader::create_function): Only
24783 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
24784 curr_properties. Set DECL_INITIAL to a dummy block.
24785 (read_rtl_function_body_from_file_range): New function.
24786 * read-rtl-function.h (read_rtl_function_body_from_file_range):
24788 * run-rtl-passes.c: New file.
24789 * run-rtl-passes.h: New file.
24791 2017-01-24 Jeff Law <law@redhat.com>
24793 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
24796 2017-01-24 Bin Cheng <bin.cheng@arm.com>
24798 PR tree-optimization/79159
24799 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
24800 (record_nonwrapping_iv): Improve boundary using above function if no
24801 value range information.
24803 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
24804 Martin Jambor <mjambor@suse.cz>
24806 * brig-builtins.def: New file.
24807 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
24808 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
24809 (DEF_HSAIL_SAT_BUILTIN): Likewise.
24810 (DEF_HSAIL_INTR_BUILTIN): Likewise.
24811 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
24812 * builtin-types.def (BT_INT8): New.
24813 (BT_INT16): Likewise.
24814 (BT_UINT8): Likewise.
24815 (BT_UINT16): Likewise.
24816 (BT_FN_ULONG): Likewise.
24817 (BT_FN_UINT_INT): Likewise.
24818 (BT_FN_UINT_ULONG): Likewise.
24819 (BT_FN_UINT_LONG): Likewise.
24820 (BT_FN_UINT_PTR): Likewise.
24821 (BT_FN_ULONG_PTR): Likewise.
24822 (BT_FN_INT8_FLOAT): Likewise.
24823 (BT_FN_INT16_FLOAT): Likewise.
24824 (BT_FN_UINT32_FLOAT): Likewise.
24825 (BT_FN_UINT16_FLOAT): Likewise.
24826 (BT_FN_UINT8_FLOAT): Likewise.
24827 (BT_FN_UINT64_FLOAT): Likewise.
24828 (BT_FN_UINT16_UINT32): Likewise.
24829 (BT_FN_UINT32_UINT16): Likewise.
24830 (BT_FN_UINT16_UINT16_UINT16): Likewise.
24831 (BT_FN_INT_PTR_INT): Likewise.
24832 (BT_FN_UINT_PTR_UINT): Likewise.
24833 (BT_FN_LONG_PTR_LONG): Likewise.
24834 (BT_FN_ULONG_PTR_ULONG): Likewise.
24835 (BT_FN_VOID_UINT64_UINT64): Likewise.
24836 (BT_FN_UINT8_UINT8_UINT8): Likewise.
24837 (BT_FN_INT8_INT8_INT8): Likewise.
24838 (BT_FN_INT16_INT16_INT16): Likewise.
24839 (BT_FN_INT_INT_INT): Likewise.
24840 (BT_FN_UINT_FLOAT_UINT): Likewise.
24841 (BT_FN_FLOAT_UINT_UINT): Likewise.
24842 (BT_FN_ULONG_UINT_UINT): Likewise.
24843 (BT_FN_ULONG_UINT_PTR): Likewise.
24844 (BT_FN_ULONG_ULONG_ULONG): Likewise.
24845 (BT_FN_UINT_UINT_UINT): Likewise.
24846 (BT_FN_VOID_UINT_PTR): Likewise.
24847 (BT_FN_UINT_UINT_PTR: Likewise.
24848 (BT_FN_UINT32_UINT64_PTR): Likewise.
24849 (BT_FN_INT_INT_UINT_UINT): Likewise.
24850 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
24851 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
24852 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
24853 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
24854 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
24855 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
24856 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
24857 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
24858 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
24859 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
24860 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
24861 * doc/frontends.texi: List BRIG FE.
24862 * doc/install.texi (Testing): Add BRIG tesring requirements.
24863 * doc/invoke.texi (Overall Options): Mention BRIG.
24864 * doc/standards.texi (Standards): Doucment BRIG HSA version.
24866 2017-01-24 Richard Biener <rguenther@suse.de>
24868 PR translation/79208
24869 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
24871 2017-01-24 Martin Jambor <mjambor@suse.cz>
24874 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
24875 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
24876 and known_contexts.
24878 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
24880 PR middle-end/79123
24881 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
24882 casts from signed to unsigned really don't have a range.
24884 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
24886 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
24887 GMP_RNDx for compatiblity.
24889 2017-01-24 Martin Liska <mliska@suse.cz>
24892 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
24893 that would prevent us to call alloca with -1 as argument.
24895 2017-01-24 Jakub Jelinek <jakub@redhat.com>
24897 * dwarf2out.c (output_compilation_unit_header, output_file_names):
24898 Avoid -Wformat-security warning.
24900 2017-01-23 Andrew Pinski <apinski@cavium.com>
24902 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
24905 2017-01-23 Martin Sebor <msebor@redhat.com>
24907 PR middle-end/78703
24908 * gimple-ssa-sprintf.c (warn_level): New global.
24909 (format_integer): Use it here and throughout the rest of the file.
24910 Use the same switch to compute sign as base.
24911 (maybe_warn): New function.
24912 (format_directive): Factor out warnings into maybe_warn.
24913 Add debugging output. Use warn_level.
24914 (add_bytes): Use warn_level.
24915 (pass_sprintf_length::compute_format_length): Add debugging output.
24916 (try_substitute_return_value): Same.
24917 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
24919 PR middle-end/78703
24920 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
24921 (struct fmtresult, format_integer, format_floating): Adjust.
24922 (fmtresult::fmtresult): Set max correctly in two argument ctor.
24923 (get_string_length, format_string,format_directive): Same.
24924 (pass_sprintf_length::compute_format_length): Same.
24925 (try_substitute_return_value): Simplify slightly.
24927 PR middle-end/78703
24928 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
24929 (fmtresult::operator+=): Outlined.
24930 (struct fmtresult): Add ctors.
24931 (struct conversion_spec): Rename...
24932 (struct directive): ...to this. Add and remove data members.
24933 (directive::set_width, directive::set_precision): New functions.
24934 (format_percent): Use fmtresult ctor.
24935 (get_width_and_precision): Remove.
24936 (format_integer): Make naming changes. Avoid computing width and
24938 (format_floating): Same. Adjust indentation.
24939 (format_character, format_none): New functions.
24940 (format_string): Moved character handling to format_character.
24941 (format_directive): Remove arguments, change return type.
24942 (parse_directive): New function.
24943 (pass_sprintf_length::compute_format_length): Move directive
24944 parsing to parse_directive.
24946 2017-01-23 Jakub Jelinek <jakub@redhat.com>
24948 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
24949 (assign_assembler_name_if_needed): ... this.
24950 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
24951 (assign_assembler_name_if_needed): ... this.
24952 (free_lang_data_in_cgraph): Adjust callers.
24953 * cgraphunit.c (cgraph_node::analyze): Likewise.
24954 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
24957 2017-01-23 Richard Biener <rguenther@suse.de>
24959 PR tree-optimization/79088
24960 PR tree-optimization/79188
24961 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
24962 resetting loop bounds after last path deletion. Reset loop
24963 bounds of the target loop, make code match the comments.
24964 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
24965 Make sure loops need no fixups.
24967 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
24969 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
24970 exponent support with double type for first argument.
24971 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
24972 type returned by __builtin_vec_extract_sig,
24973 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
24974 functions from "vector int" to "vector unsigned int" or from
24975 "vector long long int" to "vector unsigned long long int".
24976 Changed type returned by __builtin_vec_extract_exp,
24977 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_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 return type of __builtin_vec_test_data_class,
24981 __builtin_vec_test_data_class_sp, and
24982 __builtin_vec_test_data_class_dp from "vector int" to
24983 "vector bool int" or from "vector long long int" to "vector bool
24984 long long int" and changed second argument type from "unsigned
24985 int" to "int". Added new overloaded function forms "vector float
24986 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
24987 "vector float __builtin_vec_insert_exp_sp (vector float, vector
24988 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
24989 double, vector unsigned long long int)" and "vector double
24990 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
24991 long int)". Changed return type of
24992 __builtin_scalar_test_data_class and
24993 __builtin_scalar_test_data_class_sp and
24994 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
24995 int" and changed second argument from "unsigned int" to "int".
24996 Changed type returned by __builtin_scalar_test_neg,
24997 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
24998 from "int" to "bool int". Added new overloaded function form
24999 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
25000 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
25001 exponent double-precision with floating point first argument.
25002 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
25003 documentation of scalar_test_data_class, scalar_test_neg,
25004 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
25005 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
25006 vec_test_data_class built-in functions to reflect refinements in
25007 their type signatures.
25009 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
25011 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
25013 (aarch64_elf_asm_destructor): Likewise.
25015 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
25017 PR rtl-optimization/78634
25018 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
25019 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
25020 * ifcvt.c (noce_try_cmove): Add missing cost check.
25022 PR rtl-optimization/71724
25023 * combine.c (if_then_else_cond): Look for situations where it is
25024 beneficial to undo the work of one of the recursive calls.
25026 2017-01-23 Bin Cheng <bin.cheng@arm.com>
25028 PR tree-optimization/70754
25029 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
25030 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
25031 combined stmt before it if not NULL.
25032 (combine_chains): Process refs reversely and compute dominance point
25035 2017-01-23 Martin Liska <mliska@suse.cz>
25037 PR tree-optimization/79196
25038 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
25039 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
25041 (strlen_optimize_stmt): Call the renamed function.
25043 2017-01-23 Michael Matz <matz@suse.de>
25045 PR tree-optimization/78384
25046 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
25048 2017-01-23 Richard Biener <rguenther@suse.de>
25050 PR tree-optimization/79186
25051 * tree-vrp.c (register_new_assert_for): Make sure we've seen
25052 both incoming edges before moving an assert.
25054 2017-01-23 Martin Jambor <mjambor@suse.cz>
25056 * ipa-prop.c (load_from_param_1): Removed.
25057 (load_from_unmodified_param): Bits from load_from_param_1 put back
25059 (load_from_param): Removed.
25060 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
25061 with stmt. Reverted back to use of load_from_unmodified_param.
25063 2017-01-23 Martin Jambor <mjambor@suse.cz>
25066 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
25067 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
25068 field a pointer to garbage collected vector, mark lattices and
25069 ipcp_orig_node with GTY((skip)).
25070 (ipa_get_param_count): Adjust to descriptors being a pointer.
25071 (ipa_get_param): Likewise.
25072 (ipa_get_type): Likewise.
25073 (ipa_get_param_move_cost): Likewise.
25074 (ipa_set_param_used): Likewise.
25075 (ipa_get_controlled_uses): Likewise.
25076 (ipa_set_controlled_uses): Likewise.
25077 (ipa_is_param_used): Likewise.
25078 (ipa_node_params_t): Move into garbage collector. New methods insert
25080 (ipa_node_params_sum): Annotate wth GTY(()).
25081 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
25083 (ipa_load_from_parm_agg): Adjust declaration.
25084 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
25085 * ipa-profile.c (ipa_profile): Likewise.
25086 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
25087 (ipa_populate_param_decls): Make descriptors parameter garbage
25089 (ipa_dump_param): Adjust to descriptors being a pointer.
25090 (ipa_alloc_node_params): Likewise.
25091 (ipa_initialize_node_params): Likewise.
25092 (load_from_param_1): Make descriptors parameter garbage collected.
25093 (load_from_unmodified_param): Likewise.
25094 (load_from_param): Likewise.
25095 (ipa_load_from_parm_agg): Likewise.
25096 (ipa_node_params::~ipa_node_params): Removed.
25097 (ipa_free_all_node_params): Remove call to delete operator.
25098 (ipa_node_params_t::insert): New.
25099 (ipa_node_params_t::remove): Likewise.
25100 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
25101 copy known_csts and known_contexts vectors.
25102 (ipa_read_node_info): Adjust to descriptors being a pointer.
25103 (ipcp_modif_dom_walker): Make m_descriptors field garbage
25105 (ipcp_transform_function): Make descriptors variable garbage
25108 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
25110 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
25111 * config/i386/avx512dqintrin.h: Ditto.
25112 * config/i386/avx512fintrin.h: Ditto.
25113 * config/i386/i386.c: Handle new builtins.
25114 * config/i386/i386-builtin.def: Add new builtins.
25115 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
25116 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
25118 2017-01-23 Jakub Jelinek <jakub@redhat.com>
25119 Martin Liska <mliska@suse.cz>
25121 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
25122 * asan.c (asan_expand_poison_ifn): Support stores and use
25123 appropriate ASAN report function.
25124 * internal-fn.c (expand_ASAN_POISON_USE): New function.
25125 * internal-fn.def (ASAN_POISON_USE): Declare.
25126 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
25127 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
25128 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
25129 ASAN_POISON calls w/o LHS.
25130 * tree-ssa.c (execute_update_addresses_taken): Create clobber
25131 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
25132 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
25133 * gimplify.c (asan_poison_variables): Add attribute
25134 use_after_scope_memory to variables that really needs to live
25136 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
25137 having the attribute.
25139 2017-01-23 Martin Liska <mliska@suse.cz>
25141 * asan.c (create_asan_shadow_var): New function.
25142 (asan_expand_poison_ifn): Likewise.
25143 * asan.h (asan_expand_poison_ifn): New declaration.
25144 * internal-fn.c (expand_ASAN_POISON): Likewise.
25145 * internal-fn.def (ASAN_POISON): New builtin.
25146 * sanopt.c (pass_sanopt::execute): Expand
25147 asan_expand_poison_ifn.
25148 * tree-inline.c (copy_decl_for_dup_finish): Make function
25150 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
25151 * tree-ssa.c (is_asan_mark_p): New function.
25152 (execute_update_addresses_taken): Rewrite local variables
25153 (identified just by use-after-scope as addressable) into SSA.
25155 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
25157 * doc/install.texi (Specific): opensource.apple.com uses https
25158 now. Remove trailing slash.
25160 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
25162 * README.Portability: Remove note on an Irix compatibility issue.
25164 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
25166 * gcov.c (INCLUDE_ALGORITHM): Define.
25167 (INCLUDE_VECTOR): Define.
25168 No longer include <vector> and <algorithm> directly.
25170 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
25172 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
25174 * doc/invoke.texi (Code Gen Options): Ditto.
25176 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
25179 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
25181 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
25183 rtl-optimization/79125
25184 * cprop.c (local_cprop_pass): Handle cases where we make an
25185 unconditional trap.
25187 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
25191 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
25192 read from, for big endian.
25194 2017-01-20 Jiong Wang <jiong.wang@arm.com>
25196 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
25197 register pauth builtins for LP64 only.
25199 2017-01-20 Marek Polacek <polacek@redhat.com>
25202 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
25205 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
25207 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
25209 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
25210 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
25211 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
25213 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25216 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
25217 in big-endian mode when they are not a single duplicated value.
25219 2017-01-20 Richard Biener <rguenther@suse.de>
25221 * BASE-VER: Bump to 7.0.1.
25223 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
25225 * omp-low.c (omplow_simd_context): New struct. Use it...
25226 (lower_rec_simd_input_clauses): ...here and...
25227 (lower_rec_input_clauses): ...here to hold common data. Adjust all
25228 references to idx, lane, max_vf, is_simt.
25230 2017-01-20 Graham Markall <graham.markall@embecosm.com>
25232 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
25235 2017-01-20 Martin Jambor <mjambor@suse.cz>
25237 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
25238 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
25240 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
25241 (GTFILES): Rename hsa.c to hsa-common.c.
25242 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
25243 * hsa-dump.c: Likewise.
25244 * hsa-gen.c: Likewise.
25245 * hsa-regalloc.c: Likewise.
25246 * ipa-hsa.c: Likewise.
25247 * omp-expand.c: Likewise.
25248 * omp-low.c: Likewise.
25249 * toplev.c: Likewise.
25251 2017-01-20 Marek Polacek <polacek@redhat.com>
25254 * doc/invoke.texi: Document -Wduplicated-branches.
25255 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
25256 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
25257 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
25258 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
25259 return 0 only when not OEP_LEXICOGRAPHIC.
25260 (fold_build_cleanup_point_expr): Use the expression
25261 location when building CLEANUP_POINT_EXPR.
25262 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
25263 * tree.c (add_expr): Handle error_mark_node.
25265 2017-01-20 Martin Liska <mliska@suse.cz>
25268 * tree-profile.c (init_ic_make_global_vars): Do not call
25270 (gimple_init_gcov_profiler): Likewise.
25272 2017-01-20 Martin Liska <mliska@suse.cz>
25275 * cgraph.h (maybe_create_reference): Remove argument and
25277 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
25279 * ipa-cp.c (create_specialized_node): Likewise.
25280 * symtab.c (symtab_node::maybe_create_reference): Handle
25281 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
25283 2017-01-20 Martin Liska <mliska@suse.cz>
25285 * read-rtl-function.c (function_reader::create_function): Use
25286 build_decl instread of build_decl_stat.
25288 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
25290 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
25291 * config/i386/avx512dqintrin.h: Ditto.
25292 * config/i386/avx512fintrin.h: Ditto.
25293 * config/i386/i386-builtin-types.def: Add new types.
25294 * config/i386/i386.c: Handle new types.
25295 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
25296 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
25297 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
25298 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
25299 (__builtin_ia32_kshiftridi): New.
25300 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
25302 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
25306 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
25307 define to rs6000_init_stack_protect_guard.
25308 (rs6000_init_stack_protect_guard): New function.
25310 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
25311 Yunqiang Su <yunqiang.su@imgtec.com>
25313 * config.gcc (supported_defaults): Add madd4.
25314 (with_madd4): Add validation.
25315 (all_defaults): Add madd4.
25316 * config/mips/mips.opt (mmadd4): New option.
25317 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
25319 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
25321 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
25322 (ISA_HAS_FUSED_MADD4): Likewise.
25323 * doc/invoke.texi (-mmadd4): Document the new option.
25324 * doc/install.texi (--with-madd4): Document the new option.
25326 2017-01-19 Jiong Wang <jiong.wang@arm.com>
25328 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
25329 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
25330 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
25331 (aarch64_init_pauth_hint_builtins): New.
25332 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
25333 (aarch64_expand_builtin): Expand new builtins.
25335 2017-01-19 Jiong Wang <jiong.wang@arm.com>
25337 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
25338 * combine-stack-adj.c (no_unhandled_cfa): Handle
25339 REG_CFA_TOGGLE_RA_MANGLE.
25340 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
25341 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
25342 info for return address signing.
25343 (aarch64_expand_epilogue): Likewise.
25345 2017-01-19 Jiong Wang <jiong.wang@arm.com>
25347 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
25348 * config/aarch64/aarch64-protos.h
25349 (aarch64_return_address_signing_enabled): New declaration.
25350 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
25352 (aarch64_expand_prologue): Sign return address before it's pushed onto
25354 (aarch64_expand_epilogue): Authenticate return address fetched from
25356 (aarch64_override_options): Sanity check for ILP32 and ISA level.
25357 (aarch64_attributes): New function attributes for "sign-return-address".
25358 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
25359 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
25360 ("*do_return"): Generate combined instructions according to key index.
25361 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
25362 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
25364 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
25365 * config/aarch64/aarch64.opt (msign-return-address=): New.
25366 * doc/extend.texi (AArch64 Function Attributes): Documents
25367 "sign-return-address=".
25368 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
25370 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
25372 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
25373 overall option summary.
25375 2017-01-19 Jiong Wang <jiong.wang@arm.com>
25377 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
25378 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
25379 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
25380 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
25382 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
25384 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
25385 -mpower9-minmax by default for -mcpu=power9.
25386 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
25387 128-bit floating point.
25389 2017-01-20 Alan Modra <amodra@gmail.com>
25391 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
25392 optimizing for size.
25394 2017-01-20 Alan Modra <amodra@gmail.com>
25397 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
25398 for strcmp and strncmp from corresponding builtin decl.
25400 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
25402 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
25403 instead of i386/rtems-64.h.
25404 * config/i386/rtems-64.h: Remove.
25406 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
25410 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
25412 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
25414 2017-01-19 Tamar Christina <tamar.christina@arm.com>
25416 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
25417 Change int to HOST_WIDE_INT.
25418 * config/aarch64/aarch64-protos.h
25419 (aarch64_simd_gen_const_vector_dup): Likewise.
25420 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
25422 2017-01-19 David Malcolm <dmalcolm@redhat.com>
25424 * langhooks-def.h (lhd_type_for_size): New decl.
25425 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
25426 * langhooks.c (lhd_type_for_size): New function, taken from
25429 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
25431 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
25432 define_bypass for CR latency.
25433 (power9-cracked-alu): Update bypass latency and remove power9-branch.
25434 (power9-alu2): Add define_bypass for CR latency.
25436 (power9-mul): Update insn latency.
25437 (power9-mul-compare): Update insn latency, bypass latency and remove
25440 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25442 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
25444 * config/aarch64/aarch64.md
25445 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
25446 aarch64_nopcrelative_literal_loads.
25447 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
25449 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
25451 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
25452 TARGET_LOONGSON_3A.
25453 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
25455 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
25458 * config.gcc (supported_defaults): Add lxc1-sxc1.
25459 (with_lxc1_sxc1): Add validation.
25460 (all_defaults): Add lxc1-sxc1.
25461 * config/mips/mips.opt (mlxc1-sxc1): New option.
25462 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
25464 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
25465 __mips_no_lxc1_sxc1.
25466 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
25467 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
25468 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
25470 2017-01-19 Richard Biener <rguenther@suse.de>
25472 PR tree-optimization/72488
25473 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
25474 sure to restore SSA info.
25475 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
25477 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
25479 PR rtl-optimization/79121
25480 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
25481 of the inner type when shifting an extended value.
25483 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
25486 * symtab.c (symtab_node::equal_address_to): Fix comparing of
25487 interposable aliases.
25489 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
25492 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
25493 Use the evmergelohi instruction.
25494 (mov_si<mode>_e500_subreg4_2_le): Likewise.
25495 (mov_sitf_e500_subreg8_2_be): Likewise.
25496 (mov_sitf_e500_subreg12_2_le): Likewise.
25497 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
25498 (mov_si<mode>_e500_subreg4_2_be): Likewise.
25499 (mov_sitf_e500_subreg8_2_le): Likewise.
25500 (mov_sitf_e500_subreg12_2_be): Likewise.
25502 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25504 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
25505 attribute from vecsimple to vecperm.
25506 (altivec_vbpermq2): Likewise.
25508 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25511 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
25513 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25514 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
25515 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
25516 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
25517 case where N arg is SIZE_MAX.
25518 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
25519 (cmpstrsi): Add pattern.
25521 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
25523 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
25524 __builtin_vec_revb builtins.
25525 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
25526 built-in functions to support generation of the ISA 3.0 XXBR<x>
25527 vector byte reverse instructions.
25528 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
25529 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
25530 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
25531 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
25532 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
25533 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
25534 (P9V_BUILTIN_VEC_REVB): Likewise.
25535 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
25536 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
25537 (p9_xxbrq_v16qi): Likewise.
25538 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
25539 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
25540 (p9_xxbrh_v8hi): Likewise.
25541 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
25542 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
25543 vec_revb built-in functions.
25545 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
25547 PR rtl-optimization/78952
25548 * config/i386/i386.md (any_extract): New code iterator.
25549 (*insvqi_2): Use any_extract for source operand.
25550 (*insvqi_3): Use any_shiftrt for source operand.
25552 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
25554 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
25556 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
25558 2017-01-18 Matthias Klose <doko@ubuntu.com>
25560 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
25562 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25564 * config/rs6000/altivec.h (vec_bperm): Change #define.
25565 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
25566 (altivec_vbpermq2): New define_insn.
25567 (altivec_vbpermd): Likewise.
25568 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
25569 function interface.
25570 (VBPERMD): Likewise.
25571 (VBPERM): New polymorphic function interface.
25572 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
25573 Add entries for P9V_BUILTIN_VEC_VBPERM.
25574 * doc/extend.texi: Add interfaces for vec_bperm.
25576 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25578 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
25579 first letter of error messages.
25580 (s390_resolve_overloaded_builtin): Likewise.
25581 * config/s390/s390.c (s390_expand_builtin): Likewise.
25582 (s390_invalid_arg_for_unprototyped_fn): Likewise.
25583 (s390_valid_target_attribute_inner_p): Likewise.
25584 * config/s390/s390.md ("tabort"): Likewise.
25586 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
25588 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
25589 (ISA_AVOID_DIV_HILO): New macro.
25590 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
25591 (ISA_HAS_DDIV): Likewise.
25593 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
25595 * doc/invoke.texi (fabi-version): Correct number of occurrences.
25597 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
25599 * doc/invoke.texi (fabi-version): Spelling fix.
25601 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
25604 * doc/invoke.texi (fabi-version): Mention mangling fix for
25607 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
25610 * doc/invoke.texi (fabi-version): Document discriminator mangling.
25612 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
25615 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
25616 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
25618 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
25620 (stack_protect_test): Ditto.
25621 * config/rs6000/rs6000.opt (mstack-protector-guard=,
25622 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
25624 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
25625 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
25626 -mstack-protector-guard-offset=.
25627 (RS/6000 and PowerPC Options): Ditto.
25629 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
25631 * config/i386/i386.h (MASK_CLASS_P): New define.
25632 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
25633 there are no registers from different register sets also when
25634 mask registers are used. Update function comment.
25635 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
25636 to (*k/*r) and (*k/*km) alternatives.
25638 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
25640 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
25641 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
25642 (EH_RETURN_HANDLER_RTX): New define.
25643 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
25644 Force frame pointer in EH return functions.
25645 (aarch64_expand_epilogue): Add barrier for eh_return.
25646 (aarch64_final_eh_return_addr): Remove.
25647 (aarch64_eh_return_handler_rtx): New function.
25648 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
25650 (aarch64_eh_return_handler_rtx): New prototype.
25652 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25654 * config/rs6000/altivec.h (vec_rlmi): New #define.
25655 (vec_vrlnm): Likewise.
25656 (vec_rlnm): Likewise.
25657 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
25658 (UNSPEC_VRLNM): Likewise.
25659 (VIlong): New mode iterator.
25660 (altivec_vrl<VI_char>mi): New define_insn.
25661 (altivec_vrl<VI_char>nm): Likewise.
25662 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
25664 (VRLDNM): Likewise.
25665 (RLNM): New polymorphic function entry.
25666 (VRLWMI): New monomorphic function entry.
25667 (VRLDMI): Likewise.
25668 (RLMI): New polymorphic function entry.
25669 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
25670 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
25671 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
25674 2017-01-17 Jakub Jelinek <jakub@redhat.com>
25677 * dwarf2out.c (field_byte_offset): Restore the
25678 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
25679 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
25680 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
25683 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
25686 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
25688 2017-01-17 Jakub Jelinek <jakub@redhat.com>
25691 * dwarf2out.c (add_data_member_location_attribute): For constant
25692 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
25693 instead of DW_AT_data_member_location, DW_AT_bit_offset and
25694 DW_AT_byte_size attributes.
25696 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
25698 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
25699 after forcing to constant memory when the code model is medium.
25701 2017-01-17 Julia Koval <julia.koval@intel.com>
25704 * config/i386/avx512fintrin.h
25705 (_mm512_i32gather_ps): Change __addr type to void const*.
25706 (_mm512_mask_i32gather_ps): Ditto.
25707 (_mm512_i32gather_pd): Ditto.
25708 (_mm512_mask_i32gather_pd): Ditto.
25709 (_mm512_i64gather_ps): Ditto.
25710 (_mm512_mask_i64gather_ps): Ditto.
25711 (_mm512_i64gather_pd): Ditto.
25712 (_mm512_mask_i64gather_pd): Ditto.
25713 (_mm512_i32gather_epi32): Ditto.
25714 (_mm512_mask_i32gather_epi32): Ditto.
25715 (_mm512_i32gather_epi64): Ditto.
25716 (_mm512_mask_i32gather_epi64): Ditto.
25717 (_mm512_i64gather_epi32): Ditto.
25718 (_mm512_mask_i64gather_epi32): Ditto.
25719 (_mm512_i64gather_epi64): Ditto.
25720 (_mm512_mask_i64gather_epi64): Ditto.
25721 (_mm512_i32scatter_ps): Change __addr type to void*.
25722 (_mm512_mask_i32scatter_ps): Ditto.
25723 (_mm512_i32scatter_pd): Ditto.
25724 (_mm512_mask_i32scatter_pd): Ditto.
25725 (_mm512_i64scatter_ps): Ditto.
25726 (_mm512_mask_i64scatter_ps): Ditto.
25727 (_mm512_i64scatter_pd): Ditto.
25728 (_mm512_mask_i64scatter_pd): Ditto.
25729 (_mm512_i32scatter_epi32): Ditto.
25730 (_mm512_mask_i32scatter_epi32): Ditto.
25731 (_mm512_i32scatter_epi64): Ditto.
25732 (_mm512_mask_i32scatter_epi64): Ditto.
25733 (_mm512_i64scatter_epi32): Ditto.
25734 (_mm512_mask_i64scatter_epi32): Ditto.
25735 (_mm512_i64scatter_epi64): Ditto.
25736 (_mm512_mask_i64scatter_epi64): Ditto.
25737 * config/i386/avx512pfintrin.h
25738 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
25739 (_mm512_mask_prefetch_i32gather_ps): Ditto.
25740 (_mm512_mask_prefetch_i64gather_pd): Ditto.
25741 (_mm512_mask_prefetch_i64gather_ps): Ditto.
25742 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
25743 (_mm512_prefetch_i32scatter_ps): Ditto.
25744 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
25745 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
25746 (_mm512_prefetch_i64scatter_pd): Ditto.
25747 (_mm512_prefetch_i64scatter_ps): Ditto.
25748 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
25749 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
25750 * config/i386/avx512vlintrin.h
25751 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
25752 (_mm_mmask_i32gather_ps): Ditto.
25753 (_mm256_mmask_i32gather_pd): Ditto.
25754 (_mm_mmask_i32gather_pd): Ditto.
25755 (_mm256_mmask_i64gather_ps): Ditto.
25756 (_mm_mmask_i64gather_ps): Ditto.
25757 (_mm256_mmask_i64gather_pd): Ditto.
25758 (_mm_mmask_i64gather_pd): Ditto.
25759 (_mm256_mmask_i32gather_epi32): Ditto.
25760 (_mm_mmask_i32gather_epi32): Ditto.
25761 (_mm256_mmask_i32gather_epi64): Ditto.
25762 (_mm_mmask_i32gather_epi64): Ditto.
25763 (_mm256_mmask_i64gather_epi32): Ditto.
25764 (_mm_mmask_i64gather_epi32): Ditto.
25765 (_mm256_mmask_i64gather_epi64): Ditto.
25766 (_mm_mmask_i64gather_epi64): Ditto.
25767 (_mm256_i32scatter_ps): Change __addr type to void*.
25768 (_mm256_mask_i32scatter_ps): Ditto.
25769 (_mm_i32scatter_ps): Ditto.
25770 (_mm_mask_i32scatter_ps): Ditto.
25771 (_mm256_i32scatter_pd): Ditto.
25772 (_mm256_mask_i32scatter_pd): Ditto.
25773 (_mm_i32scatter_pd): Ditto.
25774 (_mm_mask_i32scatter_pd): Ditto.
25775 (_mm256_i64scatter_ps): Ditto.
25776 (_mm256_mask_i64scatter_ps): Ditto.
25777 (_mm_i64scatter_ps): Ditto.
25778 (_mm_mask_i64scatter_ps): Ditto.
25779 (_mm256_i64scatter_pd): Ditto.
25780 (_mm256_mask_i64scatter_pd): Ditto.
25781 (_mm_i64scatter_pd): Ditto.
25782 (_mm_mask_i64scatter_pd): Ditto.
25783 (_mm256_i32scatter_epi32): Ditto.
25784 (_mm256_mask_i32scatter_epi32): Ditto.
25785 (_mm_i32scatter_epi32): Ditto.
25786 (_mm_mask_i32scatter_epi32): Ditto.
25787 (_mm256_i32scatter_epi64): Ditto.
25788 (_mm256_mask_i32scatter_epi64): Ditto.
25789 (_mm_i32scatter_epi64): Ditto.
25790 (_mm_mask_i32scatter_epi64): Ditto.
25791 (_mm256_i64scatter_epi32): Ditto.
25792 (_mm256_mask_i64scatter_epi32): Ditto.
25793 (_mm_i64scatter_epi32): Ditto.
25794 (_mm_mask_i64scatter_epi32): Ditto.
25795 (_mm256_i64scatter_epi64): Ditto.
25796 (_mm256_mask_i64scatter_epi64): Ditto.
25797 (_mm_i64scatter_epi64): Ditto.
25798 (_mm_mask_i64scatter_epi64): Ditto.
25799 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
25800 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
25801 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
25802 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
25803 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
25804 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
25805 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
25806 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
25807 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
25808 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
25809 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
25810 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
25811 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
25812 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
25813 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
25814 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
25815 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
25816 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
25817 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
25818 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
25819 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
25820 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
25821 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
25822 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
25823 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
25824 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
25825 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
25826 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
25827 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
25828 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
25829 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
25830 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
25831 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
25832 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
25833 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
25834 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
25835 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
25836 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
25837 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
25838 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
25839 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
25840 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
25841 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
25842 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
25843 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
25844 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
25845 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
25846 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
25847 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
25848 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
25849 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
25850 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
25851 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
25852 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
25853 definitions accordingly.
25855 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
25856 Kuan-Lin Chen <kuanlinchentw@gmail.com>
25859 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
25862 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
25865 * ira-conflicts.c (ira_build_conflicts): Update total conflict
25866 hard regs for inner regno.
25868 2017-01-17 Martin Liska <mliska@suse.cz>
25871 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
25872 assumption and add comment.
25874 2017-01-17 Nathan Sidwell <nathan@acm.org>
25876 * ipa-visibility.c (localize_node): New function, broken out of ...
25877 (function_and_variable_visibility): ... here. Call it.
25879 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
25881 PR middle-end/77445
25882 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
25883 correctly set frequency of oudgoing edge.
25884 (duplicate_thread_path): Fix profile updating.
25886 2017-01-17 Jakub Jelinek <jakub@redhat.com>
25889 * configure.ac: Add GCC_BASE_VER.
25890 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
25891 version from BASE-VER file.
25892 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
25893 (gcc.o): Depend on $(BASEVER).
25894 * common.opt (dumpfullversion): New option.
25895 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
25896 * doc/invoke.texi: Document -dumpfullversion.
25897 * doc/install.texi: Document --with-gcc-major-version-only.
25898 * configure: Regenerated.
25900 2017-01-17 Richard Biener <rguenther@suse.de>
25902 PR tree-optimization/71433
25903 * tree-vrp.c (register_new_assert_for): Merge same asserts
25904 on all incoming edges.
25905 (process_assert_insertions_for): Handle insertions at the
25908 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
25910 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
25911 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
25913 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
25916 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
25919 2017-01-17 Alan Modra <amodra@gmail.com>
25922 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
25923 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
25924 symbolic stack limit when pic.
25926 2017-01-16 Martin Sebor <msebor@redhat.com>
25928 PR tree-optimization/78608
25929 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
25931 2017-01-16 Jeff Law <law@redhat.com>
25934 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
25935 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
25936 for several include directories that may be relative to sysroot.
25937 * config/i386/x-mingw32 (gplus_includedir): Define.
25938 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
25939 (native_system_includedir): Likewise.
25940 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
25941 override if TARGET_SYSTEM_ROOT is defined.
25942 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
25944 PR tree-optimization/79090
25945 PR tree-optimization/33562
25946 PR tree-optimization/61912
25947 PR tree-optimization/77485
25948 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
25949 and computed trims into the dump file.
25951 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
25953 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
25955 2017-01-16 Jakub Jelinek <jakub@redhat.com>
25958 * gimplify.c (gimplify_init_constructor): If want_value and
25959 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
25963 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
25964 sequence. Formatting fixes.
25965 (doloop_optimize): Formatting fixes.
25968 * gcc.c (debug_level_greater_than_spec_func): New function.
25969 (static_spec_functions): Add debug-level-gt spec function.
25970 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
25972 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
25973 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
25974 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
25975 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
25976 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
25977 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
25979 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
25981 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
25982 QImode fixups to general and mask registers only.
25984 2017-01-16 Carl Love <cel@us.ibm.com>
25986 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
25987 for built-in functions
25988 vector signed char vec_nabs (vector signed char)
25989 vector signed short vec_nabs (vector signed short)
25990 vector signed int vec_nabs (vector signed int)
25991 vector signed long long vec_nabs (vector signed long long)
25992 vector float vec_nabs (vector float)
25993 vector double vec_nabs (vector double)
25994 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
25996 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
25997 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
25998 * doc/extend.texi: Update the documentation file for the new built-in
26001 2017-01-16 Martin Sebor <msebor@redhat.com>
26003 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
26006 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26008 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
26009 UNSPEC_VSX__XXSPLTD to require special splat handling.
26011 2017-01-16 David Malcolm <dmalcolm@redhat.com>
26014 * system.h: Poison strndup.
26016 2017-01-16 Alan Modra <amodra@gmail.com>
26019 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
26022 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
26024 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
26026 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
26028 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
26029 call recog here. Assert that INSN_CODE (insn) is non-negative.
26031 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
26034 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
26036 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
26037 in the currently scheduled RTL fragment.
26039 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
26041 PR rtl-optimization/78751
26042 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
26045 2017-01-14 Jeff Law <law@redhat.com>
26047 PR tree-optimization/79090
26048 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
26049 variable length stores.
26050 (compute_trims): Delete dead assignment to *trim_tail.
26051 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
26054 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
26056 PR rtl-optimization/78626
26057 PR rtl-optimization/78727
26058 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
26059 of a block, and split such blocks after everything else is finished.
26061 2017-01-14 Alan Modra <amodra@gmail.com>
26064 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
26065 target legitimate_combined_insn.
26066 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
26067 (rs6000_legitimate_combined_insn): New function.
26068 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
26070 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
26071 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
26072 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
26074 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
26076 * doc/frontends.texi (G++ and GCC): Remove references to Java.
26078 2017-01-13 Jeff Law <law@redhat.com>
26080 PR tree-optimization/33562
26081 PR tree-optimization/61912
26082 PR tree-optimization/77485
26083 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
26085 (delete_dead_assignment): Likewise.
26086 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
26087 statement to delete_dead_call and delete_dead_assignment.
26089 2017-01-13 David Malcolm <dmalcolm@redhat.com>
26092 * substring-locations.c (format_warning_va): Strengthen case 1 so
26093 that both endpoints of the substring must be within the format
26094 range for just the substring to be printed.
26096 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
26098 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
26099 * config/i386/i386.c (ix86_target_string): Add missing options
26100 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
26101 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
26102 flags_other and ix86_target_other to flags2_other. Display unknown
26104 (ix86_valid_target_attribute_inner_p): Add missing options and
26105 reorder options by implied ISAs, as in ix86_target_string.
26107 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
26109 * hash-table.h (hash_table::too_empty_p): New function.
26110 (hash_table::expand): Use it.
26111 (hash_table::traverse): Likewise.
26112 (hash_table::empty_slot): Use sizeof (value_type) instead of
26113 sizeof (PTR) to convert bytes to elements. Shrink the table
26114 if the current size is excessive for the current number of
26117 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
26119 * ira-costs.c (record_reg_classes): Break from the inner loop
26120 early once alt_fail is known to be true. Update outer loop
26121 handling accordingly.
26123 2017-01-13 Jeff Law <law@redhat.com>
26125 * tree-ssa-dse.c (decrement_count): New function.
26126 (increment_start_addr, maybe_trim_memstar_call): Likewise.
26127 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
26128 when we know the partially dead statement is a mem* function.
26130 PR tree-optimization/61912
26131 PR tree-optimization/77485
26132 * tree-ssa-dse.c: Include expr.h.
26133 (maybe_trim_constructor_store): New function.
26134 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
26136 PR tree-optimization/33562
26137 PR tree-optimization/61912
26138 PR tree-optimization/77485
26139 * doc/invoke.texi: Document new dse-max-object-size param.
26140 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
26141 * tree-ssa-dse.c: Include params.h.
26142 (dse_store_status): New enum.
26143 (initialize_ao_ref_for_dse): New, partially extracted from
26145 (valid_ao_ref_for_dse, normalize_ref): New.
26146 (setup_live_bytes_from_ref, compute_trims): Likewise.
26147 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
26148 (maybe_trim_partially_dead_store): Likewise.
26149 (maybe_trim_complex_store): Likewise.
26150 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
26151 Track what bytes live from the original store. Return tri-state
26152 for dead, partially dead or live.
26153 (dse_dom_walker): Add constructor, destructor and new private members.
26154 (delete_dead_call, delete_dead_assignment): New extracted from
26156 (dse_optimize_stmt): Make a member of dse_dom_walker.
26157 Use initialize_ao_ref_for_dse.
26159 PR tree-optimization/33562
26160 PR tree-optimization/61912
26161 PR tree-optimization/77485
26162 * sbitmap.h (bitmap_count_bits): Prototype.
26163 (bitmap_clear_range, bitmap_set_range): Likewise.
26164 * sbitmap.c (bitmap_clear_range): New function.
26165 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
26167 2017-01-13 Martin Liska <mliska@suse.cz>
26170 * function.c (set_cfun): Add new argument force.
26171 * function.h (set_cfun): Likewise.
26172 * ipa-inline-transform.c (inline_call): Use the function when
26173 strict alising from is dropped for function we inline to.
26175 2017-01-13 Richard Biener <rguenther@suse.de>
26177 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
26178 for dumping GIMPLE INTEGER_CSTs.
26180 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26182 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
26183 to 201112L since C++17.
26185 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
26188 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
26189 if -fsanitize=kernel-address is present.
26191 2017-01-13 Richard Biener <rguenther@suse.de>
26193 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
26194 as _Literal ( type ) number in case usual suffixes do not
26195 preserve all information.
26197 2017-01-13 Richard Biener <rguenther@suse.de>
26199 PR tree-optimization/77283
26200 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
26201 and ssa-iterators.h.
26202 (is_feasible_trace): Implement a cost model based on joiner
26205 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
26208 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
26209 char or short to __float128/_Float128 directly.
26211 2017-01-12 Martin Sebor <msebor@redhat.com>
26213 to -Wformat-overflow.
26214 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
26215 (min_bytes_remaining): Same.
26216 (get_string_length): Same.
26217 (format_string): Same.
26218 (format_directive): Same.
26220 (pass_sprintf_length::handle_gimple_call): Same.
26222 2017-01-12 Jakub Jelinek <jakub@redhat.com>
26224 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
26225 info.nowrite calls with no lhs that can't throw. Return bool
26226 whether gsi_remove has been called or not.
26227 (pass_sprintf_length::handle_gimple_call): Return bool whether
26228 try_substitute_return_value called gsi_remove. Formatting fix.
26229 (pass_sprintf_length::execute): Don't use gsi_remove if
26230 handle_gimple_call returned true.
26233 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
26234 be removed due to side-effects, don't remove following barrier nor
26235 turn the successor edge into fallthru edge.
26237 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26240 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
26241 element-reversing loads and stores as not swappable.
26243 2017-01-12 Nathan Sidwell <nathan@acm.org>
26244 Nicolai Stange <nicstange@gmail.com>
26246 * combine.c (try_combine): Don't ignore result of overlap checking
26247 loop. Combine overlap & asm check into single loop.
26249 2017-01-12 Richard Biener <rguenther@suse.de>
26251 * tree-pretty-print.c (dump_generic_node): Provide -gimple
26252 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
26254 2017-01-12 Richard Biener <rguenther@suse.de>
26256 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
26257 and TS_TARGET_OPTION directly derive from TS_BASE.
26258 * tree-core.h (tree_optimization_option): Derive from tree_base.
26259 (tree_target_option): Likewise.
26261 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
26263 * config/i386/i386.c (memory_address_length): Increase len
26264 only when rip_relative_addr_p returns false.
26266 2017-01-11 Julia Koval <julia.koval@intel.com>
26268 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
26269 (OPTION_MASK_ISA_SGX_SET): New.
26270 (ix86_handle_option): Handle OPT_msgx.
26271 * config.gcc: Added sgxintrin.h.
26272 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
26273 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
26274 * config/i386/i386.c (ix86_target_string): Add -msgx.
26276 (ix86_option_override_internal): Handle new options.
26277 (ix86_valid_target_attribute_inner_p): Add sgx.
26278 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
26279 * config/i386/i386.opt: Add msgx.
26280 * config/i386/sgxintrin.h: New file.
26281 * config/i386/x86intrin.h: Add sgxintrin.h.
26283 2017-01-11 Jakub Jelinek <jakub@redhat.com>
26286 * fold-const.c (maybe_nonzero_address): Return 1 for function
26288 (tree_single_nonzero_warnv_p): Don't handle function local objects
26292 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
26295 2017-01-11 David Malcolm <dmalcolm@redhat.com>
26298 * opts.c: Include "spellcheck.h"
26299 (struct string_fragment): New struct.
26300 (struct edit_distance_traits<const string_fragment &>): New
26302 (get_closest_sanitizer_option): New function.
26303 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
26305 2017-01-11 Jakub Jelinek <jakub@redhat.com>
26307 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
26309 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
26310 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
26311 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
26312 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
26313 for initial die_offset if dwarf_split_debug_info.
26314 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
26315 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
26317 (output_skeleton_debug_sections): Formatting fix. Use
26318 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
26319 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
26321 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
26323 * config/arm/cortex-a53.md: Add bypasses for
26324 cortex_a53_r2f_cvt.
26325 (cortex_a53_r2f): Only use for transfers.
26326 (cortex_a53_f2r): Likewise.
26327 (cortex_a53_r2f_cvt): Add reservation for conversions.
26328 (cortex_a53_f2r_cvt): Likewise.
26330 2017-01-11 Tamar Christina <tamar.christina@arm.com>
26332 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
26333 to all inlined functions, change static to extern.
26335 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
26338 * config/arm/arm.c (legitimize_pic_address): Handle reference to
26340 (arm_assemble_integer): Likewise.
26342 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
26344 * config.gcc: Use new awk script to check CPU, FPU and architecture
26345 parameters for --with-... options.
26346 * config/arm/parsecpu.awk: New file
26347 * config/arm/arm-cpus.in: New file.
26348 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
26350 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
26352 * config/arm/t-arm: Update dependency rules.
26353 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
26354 of processing .def files.
26355 * config/arm/genopt.sh: Deleted.
26356 * config/arm/gentune.sh: Deleted.
26357 * config/arm/arm-cores.def: Deleted.
26358 * config/arm/arm-arches.def: Deleted.
26359 * config/arm/arm-fpus.def: Deleted.
26360 * config/arm/arm-tune.md: Regenerated.
26361 * config/arm/arm-tables.opt: Regenerated.
26362 * config/arm/arm-cpu.h: New generated file.
26363 * config/arm/arm-cpu-data.h: New generated file.
26364 * config/arm/arm-cpu-cdata.h: New generated file.
26366 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
26369 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
26371 (input_varpool_node): Unpack dynamically_initialized bit.
26373 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
26375 PR rtl-optimization/79032
26376 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
26377 the alignment of the adjusted memory reference against that of MODE,
26378 instead of the alignment of the original memory reference.
26380 2017-01-11 Martin Jambor <mjambor@suse.cz>
26382 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
26384 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
26385 decorated functions.
26387 2017-01-11 Richard Biener <rguenther@suse.de>
26389 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
26390 set range/nonnull info for PHI results. Do not set it on
26391 stmts marked for removal.
26393 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
26395 * expr.c (store_field): In the bitfield case, fetch the return value
26396 from the registers before applying a single big-endian adjustment.
26397 Always do a final load for a BLKmode value not larger than a word.
26399 2017-01-10 David Malcolm <dmalcolm@redhat.com>
26402 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
26403 that we correctly handle column numbers greater than
26404 LINE_MAP_MAX_COLUMN_NUMBER.
26406 2017-01-10 Martin Sebor <msebor@redhat.com>
26408 PR middle-end/78245
26409 * gimple-ssa-sprintf.c (get_destination_size): Call
26410 {init,fini}object_sizes.
26411 * tree-object-size.c (addr_object_size): Adjust.
26412 (pass_through_call): Adjust.
26413 (pass_object_sizes::execute): Adjust.
26414 * tree-object-size.h (fini_object_sizes): Declare.
26416 2017-01-10 Martin Sebor <msebor@redhat.com>
26418 PR tree-optimization/78775
26419 * builtins.c (get_size_range): Move...
26420 * calls.c: ...to here.
26421 (alloc_max_size): Accept zero argument.
26422 (operand_signed_p): Remove.
26423 (maybe_warn_alloc_args_overflow): Call get_size_range.
26424 * calls.h (get_size_range): Declare.
26426 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
26428 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
26429 from TI's devices.csv file as of September 2016.
26430 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
26432 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
26434 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
26435 * doc/invoke.texi: Likewise.
26436 * doc/md.texi: Likewise.
26437 * doc/objc.texi: Likewise.
26439 2017-01-10 Joshua Conner <joshconner@google.com>
26441 * config/arm/fuchsia-elf.h: New file.
26442 * config/fuchsia.h: New file.
26443 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
26444 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
26446 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
26448 2016-01-10 Richard Biener <rguenther@suse.de>
26450 PR tree-optimization/79034
26451 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
26452 Propagate out degenerate PHIs in the joiner.
26454 2017-01-10 Martin Liska <mliska@suse.cz>
26456 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
26457 (sort_congruence_classes_by_decl_uid): Likewise.
26458 (sort_congruence_class_groups_by_decl_uid): Likewise.
26459 (sem_item_optimizer::merge_classes): Sort class, groups in these
26460 classes and members in the groups by DECL_UID of declarations.
26461 This would make merge operations stable.
26463 2017-01-10 Martin Liska <mliska@suse.cz>
26465 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
26466 usage of m_classes_vec.
26467 (sem_item_optimizer::~sem_item_optimizer): Likewise.
26468 (sem_item_optimizer::get_group_by_hash): Likewise.
26469 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
26470 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
26471 (sem_item_optimizer::verify_classes): Likewise.
26472 (sem_item_optimizer::process_cong_reduction): Likewise.
26473 (sem_item_optimizer::dump_cong_classes): Likewise.
26474 (sem_item_optimizer::merge_classes): Likewise.
26475 * ipa-icf.h (congruence_class_hash): Rename from
26476 congruence_class_group_hash. Remove declaration of m_classes_vec.
26478 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
26480 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
26481 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
26482 * config.gcc: Add avx512vpopcntdqintrin.h.
26483 * config/i386/avx512vpopcntdqintrin.h: New.
26484 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
26485 * config/i386/i386-builtin-types.def: Add new types.
26486 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
26487 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
26488 __builtin_ia32_vpopcountq_v8di_mask): New.
26489 * config/i386/i386-c.c (ix86_target_macros_internal): Define
26490 __AVX512VPOPCNTDQ__.
26491 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
26492 (PTA_AVX512VPOPCNTDQ): Define.
26493 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
26494 TARGET_AVX512VPOPCNTDQ_P): Define.
26495 * config/i386/i386.opt: Add mavx512vpopcntdq.
26496 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
26497 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
26499 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
26501 PR middle-end/77484
26502 * predict.def (PRED_CALL): Set to 67.
26504 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
26506 * expr.c (store_field): In the bitfield case, if the value comes from
26507 a function call and is of an aggregate type returned in registers, do
26508 not modify the field mode; extract the value in all cases if the mode
26509 is BLKmode and the size is not larger than a word.
26511 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
26514 * config/i386/cpuid.h: Fix undefined behavior.
26516 2017-01-04 Jeff Law <law@redhat.com>
26518 PR tree-optimization/79007
26519 PR tree-optimization/67955
26520 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
26521 conservative for pt.null when flag_non_call_exceptions is on.
26523 2017-01-09 Jakub Jelinek <jakub@redhat.com>
26525 PR translation/79019
26526 PR translation/79020
26527 * params.def (PARAM_INLINE_MIN_SPEEDUP,
26528 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
26529 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
26531 * config/avr/avr.opt (maccumulate-args): Likewise.
26532 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
26533 * common.opt (freport-bug): Likewise.
26534 * cif-code.def (CIF_FINAL_ERROR): Likewise.
26535 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
26536 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
26537 translatable string.
26538 * config/i386/i386.c (function_value_32): Likewise.
26539 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
26540 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
26542 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
26543 * common/config/msp430/msp430-common.c (msp430_handle_option):
26545 * symtab.c (symtab_node::verify_base): Likewise.
26546 * opts.c (set_debug_level): Likewise.
26547 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
26548 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
26549 missing whitespace to translatable strings.
26550 * config/avr/avr.md (bswapsi2): Fix typo in comment.
26551 * config/sh/superh.h: Likewise.
26552 * config/i386/xopintrin.h: Likewise.
26553 * config/i386/znver1.md: Likewise.
26554 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
26555 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
26556 * double-int.h (struct double_int): Likewise.
26557 * double-int.c (div_and_round_double): Likewise.
26558 * wide-int.cc: Likewise.
26559 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
26560 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
26561 * cfgcleanup.c (crossjumps_occured): Renamed to ...
26562 (crossjumps_occurred): ... this.
26563 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
26566 PR tree-optimization/78899
26567 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
26568 returning bool return struct loop *, NULL for failure and the new
26570 (versionable_outer_loop_p): Don't version outer loop if it has
26571 dont_vectorized bit set.
26572 (tree_if_conversion): When versioning outer loop, ensure
26573 tree_if_conversion is performed also on the inner loop of the
26574 non-vectorizable outer loop copy.
26575 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
26576 LOOP_VECTORIZED in inner loop of the scalar outer loop and
26577 prevent vectorization of it.
26578 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
26579 the outer loop vectorization of the non-scalar version is attempted
26580 before vectorization of the inner loop in scalar version. If
26581 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
26582 vectorization of its inner loop.
26583 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
26584 has 2 inner loops, rename also on edges from bb whose single pred
26585 is outer_loop->header. Fix typo in function comment.
26587 2017-01-09 Martin Sebor <msebor@redhat.com>
26590 * asan.c (asan_emit_stack_protection): Increase local buffer size
26591 to avoid snprintf truncation warning.
26593 2017-01-09 Andrew Pinski <apinski@cavium.com>
26595 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
26596 to reference thunderx2t99 for the tuning structure
26597 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
26599 (thunderx2t99_extra_costs): This.
26600 * config/aarch64/aarch64-tune.md: Regenerate.
26601 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
26602 (vulcan_addrcost_table): This.
26603 (vulcan_regmove_cost): Rename to ...
26604 (thunderx2t99_regmove_cost): This.
26605 (vulcan_vector_cost): Rename to ...
26606 (thunderx2t99_vector_cost): this.
26607 (vulcan_branch_cost): Rename to ...
26608 (thunderx2t99_branch_cost): This.
26609 (vulcan_tunings): Rename to ...
26610 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
26611 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
26613 2017-01-09 Martin Jambor <mjambor@suse.cz>
26617 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
26618 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
26619 (propagate_vr_accross_jump_function): Use the above function for all
26620 value range computations for pass-through jump functions and type
26621 converasion from explicit value range values.
26622 (ipcp_propagate_stage): Do not attempt to deduce types of formal
26623 parameters from TYPE_ARG_TYPES.
26624 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
26625 (ipa_write_node_info): Stream type of the actual argument.
26626 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
26628 2017-01-09 Martin Liska <mliska@suse.cz>
26631 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
26632 (lookup_compiler): Do not show error message with have_E.
26634 2017-01-09 Jakub Jelinek <jakub@redhat.com>
26636 PR tree-optimization/78938
26637 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
26638 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
26639 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
26640 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
26643 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26645 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
26648 2017-01-09 Richard Biener <rguenther@suse.de>
26650 PR tree-optimization/78997
26651 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
26652 name condition properly.
26654 2017-01-09 Richard Biener <rguenther@suse.de>
26657 * dwarf2out.c (is_cxx): New overload with context.
26658 (is_naming_typedef_decl): Use it.
26660 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
26662 * invoke.texi (Option Summary): Correct spacing in option lists
26663 and add line breaks to fix over-long lines.
26665 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
26667 PR middle-end/17660
26669 * extend.texi (Common Variable Attributes): Add xref to GCC
26670 Internals manual to explain mode attribute keywords.
26672 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
26675 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
26676 and Preprocessor Options.
26677 (Options for Linking): Document -pthread here....
26678 (RS/6000 and PowerPC Options): ...not here.
26679 (Solaris 2 Options): ...or here.
26680 * doc/cppopts.texi: Document -pthread.
26682 2017-01-08 Martin Sebor <msebor@redhat.com>
26684 PR middle-end/77708
26685 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
26686 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
26687 New member functions.
26688 (format_directive): Used them.
26690 (pass_sprintf_length::handle_gimple_call): Same.
26691 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
26692 to avoid truncation for any argument.
26693 (extract_affine_mul): Same.
26694 * tree.c (get_file_function_name): Same.
26696 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
26698 PR middle-end/77484
26699 * predict.def (PRED_INDIR_CALL): Set to 86.
26701 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
26703 PR preprocessor/54124
26704 * doc/cppopts.texi: Reformat -d subtable to list the full name
26705 of the options. Add cross-reference to the docs for the general
26706 compiler -d options.
26707 * doc/invoke.texi (Developer Options): Add cross-reference to the
26708 preprocessor-specific -d option documentation.
26710 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
26712 PR preprocessor/13498
26713 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
26714 redudant material, and reflect new command-line options.
26715 (System Headers): Likewise.
26717 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
26719 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
26720 -isystem, and -idirafter. Copy-edit.
26721 * doc/cppopts.texi: Copy-edit. Remove contradiction about
26722 default for -ftrack-macro-expansion. Delete obsolete and
26723 badly-formatted implementation details about -fdebug-cpp output.
26724 * doc/cppwarnopts.texi: Copy-edit.
26726 2017-01-07 David Malcolm <dmalcolm@redhat.com>
26729 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
26730 that the transition from a max line width >= 1<<10 to narrower
26731 lines works correctly.
26733 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
26735 * doc/options.texi (PerFunction): New.
26736 * opt-functions.awk (switch_flags): Map both Optimization and
26737 PerFunction to CL_OPTIMIZATION.
26738 * opth-gen.awk: Test for PerFunction flag along with
26740 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
26741 it only when the latter is present. Skip those that don't in
26742 the hash function generator.
26743 * common.opt (fvar-tracking): Mark as PerFunction instead of
26745 (fvar-tracking-assignments): Likewise.
26746 (fvar-tracking-assignments-toggle): Likewise.
26747 (fvar-tracking-uninit): Likewise.
26749 2017-01-07 Jakub Jelinek <jakub@redhat.com>
26751 PR translation/79018
26752 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
26755 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
26758 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
26759 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
26760 TARGET_LONG_JUMP_TABLE_OFFSETS.
26761 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
26762 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
26763 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
26764 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
26765 * config/m68k/m68k.md (tablejump expander): Likewise.
26766 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
26767 TARGET_LONG_JUMP_TABLE_OFFSETS.
26768 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
26769 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
26771 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
26772 David Holsgrove <david.holsgrove@xilinx.com>
26774 * common/config/microblaze/microblaze-common.c
26775 (TARGET_EXCEPT_UNWIND_INFO): Remove.
26776 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
26778 * config/microblaze/microblaze.c (microblaze_must_save_register)
26779 (microblaze_expand_epilogue, microblaze_return_addr): Handle
26781 (microblaze_eh_return): New function.
26782 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
26783 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
26784 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
26785 * config/microblaze/microblaze.md (eh_return): New pattern.
26787 2017-01-06 Jakub Jelinek <jakub@redhat.com>
26789 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
26790 GCC_DIAGNOSTIC_STRINGIFY): Define.
26792 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
26794 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26796 * config/arm/arm.md (<mcrr>): New.
26798 * config/arm/arm.c (arm_arch5te): New.
26799 (arm_option_override): Set arm_arch5te.
26800 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
26802 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
26803 (arm_mcrr_qualifiers): ... this. New.
26804 (MRRC_QUALIFIERS): Define to...
26805 (arm_mrrc_qualifiers): ... this. New.
26806 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
26808 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
26809 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
26810 (MRRCI, mrrc, MRRC): New.
26811 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
26812 VUNSPEC_MRRC2): New.
26814 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26816 * config/arm/arm.md (<mcr>): New.
26818 * config/arm/arm.c (arm_coproc_builtin_available): Add
26819 support for mcr, mrc, mcr2 and mrc2.
26820 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
26821 (arm_mcr_qualifiers): ... this. New.
26822 (MRC_QUALIFIERS): Define to ...
26823 (arm_mrc_qualifiers): ... this. New.
26824 (MCR_QUALIFIERS): Define to ...
26825 (arm_mcr_qualifiers): ... this. New.
26826 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
26828 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
26829 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
26830 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
26831 VUNSPEC_MRC2): New.
26833 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26835 * config/arm/arm.md (*ldc): New.
26839 * config/arm/arm.c (arm_coproc_builtin_available): Add
26840 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
26841 (arm_coproc_ldc_stc_legitimate_address): New.
26842 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
26843 'qualifier_const_pointer'.
26844 (LDC_QUALIFIERS): Define to...
26845 (arm_ldc_qualifiers): ... this. New.
26846 (STC_QUALIFIERS): Define to...
26847 (arm_stc_qualifiers): ... this. New.
26848 * config/arm/arm-protos.h
26849 (arm_coproc_ldc_stc_legitimate_address): New.
26850 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
26851 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
26852 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
26853 stc2, stcl, stc2l): New.
26854 * config/arm/constraints.md (Uz): New.
26855 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
26856 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
26857 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
26858 VUNSPEC_STC2L): New.
26860 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26862 * config/arm/arm.md (<cdp>): New.
26863 * config/arm/arm.c (neon_const_bounds): Rename this ...
26864 (arm_const_bounds): ... this.
26865 (arm_coproc_builtin_available): New.
26866 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
26867 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
26868 (CDP_QUALIFIERS): Define to...
26869 (arm_cdp_qualifiers): ... this. New.
26871 (arm_expand_builtin_args): Add case for 6 arguments.
26872 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
26873 (arm_const_bounds): ... this.
26874 (arm_coproc_builtin_available): New.
26875 * config/arm/arm_acle.h (__arm_cdp): New.
26877 * config/arm/arm_acle_builtins.def (cdp): New.
26879 * config/arm/iterators.md (CDPI,CDP,cdp): New.
26880 * config/arm/neon.md: Rename all 'neon_const_bounds' to
26881 'arm_const_bounds'.
26882 * config/arm/types.md (coproc): New.
26883 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
26884 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
26885 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
26886 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
26888 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26890 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
26891 (UBINOP_QUALIFIERS): New.
26893 (acle_builtin_data): New. Change comment.
26894 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
26895 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
26896 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
26897 arm_acle_builtins.def.
26898 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
26899 (arm_init_acle_builtins): New.
26900 (CRC32_BUILTIN): Remove.
26901 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
26902 crc32cb, crc32ch and crc32cw.
26903 (arm_init_crc32_builtins): Remove.
26904 (arm_init_builtins): Use arm_init_acle_builtins rather
26905 than arm_init_crc32_builtins.
26906 (arm_expand_acle_builtin): New.
26907 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
26908 * config/arm/arm_acle_builtins.def: New.
26910 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26912 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
26913 (arm_builtin_datum): ... this.
26914 (arm_init_neon_builtin): Rename to ...
26915 (arm_init_builtin): ... this. Add a new parameters PREFIX
26916 and USE_SIG_IN_NAME.
26917 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
26918 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
26919 'arm_builtin_datum'.
26920 (arm_init_vfp_builtins): Likewise.
26921 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
26922 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
26923 (arm_expand_neon_args): Rename to ...
26924 (arm_expand_builtin_args): ... this. Rename builtin_arg
26925 enum values and differentiate between ARG_BUILTIN_MEMORY
26926 and ARG_BUILTIN_NEON_MEMORY.
26927 (arm_expand_neon_builtin_1): Rename to ...
26928 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
26929 values, arm_expand_builtin_args and add bool parameter NEON.
26930 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
26931 (arm_expand_vfp_builtin): Likewise.
26932 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
26934 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
26936 PR middle-end/77484
26937 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
26938 * predict.c (tree_estimate_probability_bb): Reverse direction of
26939 polymorphic call predictor.
26941 2017-01-06 David Malcolm <dmalcolm@redhat.com>
26943 * passes.c (execute_one_pass): Split out pass-skipping logic into...
26944 (determine_pass_name_match): ...this new function and...
26945 (should_skip_pass_p): ...this new function.
26947 2017-01-06 Nathan Sidwell <nathan@acm.org>
26949 * ipa-visibility.c (function_and_variable_visibility): Reformat
26950 comments and long lines. Remove extrneous if.
26951 * symtab.c (symtab_node::make_decl_local): Fix code format.
26952 (symtab_node::set_section_for_node): Fix comment typo.
26954 2017-01-06 Martin Liska <mliska@suse.cz>
26957 * lra-constraints.c: Rename invariant to lra_invariant.
26958 * predict.c (set_even_probabilities): Initialize e to NULL.
26960 2017-01-05 Martin Sebor <msebor@redhat.com>
26962 PR tree-optimization/78910
26963 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
26964 (format_integer): Correct off-by-one error in the handling
26965 of precision with negative numbers in signed conversions..
26967 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
26969 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
26971 2017-01-05 Jakub Jelinek <jakub@redhat.com>
26973 PR tree-optimization/71016
26974 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
26975 factor_out_conditional_conversion. Formatting fix.
26976 (factor_out_conditional_conversion): Add cond_stmt argument.
26977 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
26978 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
26981 2017-01-05 David Malcolm <dmalcolm@redhat.com>
26983 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
26984 read-rtl-function.o, and selftest-rtl.o.
26985 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
26986 (selftest::aarch64_test_loading_full_dump): New function.
26987 (selftest::aarch64_run_selftests): New function.
26988 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
26989 selftest::aarch64_run_selftests.
26990 * config/i386/i386.c
26991 (selftest::ix86_test_loading_dump_fragment_1): New function.
26992 (selftest::ix86_test_loading_call_insn): New function.
26993 (selftest::ix86_test_loading_full_dump): New function.
26994 (selftest::ix86_test_loading_unspec): New function.
26995 (selftest::ix86_run_selftests): Call the new functions.
26996 * emit-rtl.c (maybe_set_max_label_num): New function.
26997 * emit-rtl.h (maybe_set_max_label_num): New decl.
26998 * function.c (instantiate_decls): Guard call to
26999 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
27000 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
27002 * gensupport.c (gen_reader::gen_reader): Pass "false"
27003 for new "compact" param of rtx_reader.
27004 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
27005 rather than an empty string for NULL strings.
27006 * read-md.c: Potentially include config.h rather than bconfig.h.
27007 Wrap include of errors.h with #ifdef GENERATOR_FILE.
27008 (have_error): New global, copied from errors.c.
27009 (md_reader::read_name): Rename to...
27010 (md_reader::read_name_1): ...this, adding "out_loc" param,
27011 and converting "missing name or number" to returning false, rather
27013 (md_reader::read_name): Reimplement in terms of read_name_1.
27014 (md_reader::read_name_or_nil): New function.
27015 (md_reader::read_string): Handle "(nil)" by returning NULL.
27016 (md_reader::md_reader): Add new param "compact".
27017 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
27018 (md_reader::read_file): New method.
27019 * read-md.h (md_reader::md_reader): Add new param "compact".
27020 (md_reader::read_file): New method.
27021 (md_reader::is_compact): New accessor.
27022 (md_reader::read_name): Convert return type from void to file_location.
27023 (md_reader::read_name_or_nil): New decl.
27024 (md_reader::read_name_1): New decl.
27025 (md_reader::m_compact): New field.
27026 (noop_reader::noop_reader): Pass "false" for new "compact" param
27028 (rtx_reader::rtx_reader): Add new "compact" param.
27029 (rtx_reader::read_rtx_operand): Make virtual and convert return
27030 type from void to rtx.
27031 (rtx_reader::read_until): New decl.
27032 (rtx_reader::handle_any_trailing_information): New virtual function.
27033 (rtx_reader::postprocess): New virtual function.
27034 (rtx_reader::finalize_string): New virtual function.
27035 (rtx_reader::m_in_call_function_usage): New field.
27036 (rtx_reader::m_reuse_rtx_by_id): New field.
27037 * read-rtl-function.c: New file.
27038 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
27039 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
27040 (selftest::verify_three_block_rtl_cfg): New decl.
27041 * read-rtl-function.h: New file.
27042 * read-rtl.c: Potentially include config.h rather than bconfig.h.
27043 For host, include function.h, memmodel.h, and emit-rtl.h.
27044 (one_time_initialization): New function.
27045 (struct compact_insn_name): New struct.
27046 (compact_insn_names): New array.
27047 (find_code): Handle insn codes in compact dumps.
27048 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
27049 (bind_subst_iter_and_attr): Likewise.
27050 (add_condition_to_string): Likewise.
27051 (add_condition_to_rtx): Likewise.
27052 (apply_attribute_uses): Likewise.
27053 (add_current_iterators): Likewise.
27054 (apply_iterators): Likewise.
27055 (initialize_iterators): Guard usage of apply_subst_iterator with
27056 #ifdef GENERATOR_FILE.
27057 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
27058 (md_reader::read_mapping): Likewise.
27059 (add_define_attr_for_define_subst): Likewise.
27060 (add_define_subst_attr): Likewise.
27061 (read_subst_mapping): Likewise.
27062 (check_code_iterator): Likewise.
27063 (rtx_reader::read_rtx): Likewise. Move one-time initialization
27065 (one_time_initialization): New function.
27066 (rtx_reader::read_until): New method.
27067 (read_flags): New function.
27068 (parse_reg_note_name): New function.
27069 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
27070 Handle reuse_rtx ids.
27071 Wrap iterator lookup within #ifdef GENERATOR_FILE.
27072 Add parsing support for RTL dumps, mirroring the special-cases in
27073 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
27074 values, and calling handle_any_trailing_information.
27075 (rtx_reader::read_rtx_operand): Convert return type from void
27076 to rtx, returning return_rtx. Handle case 'e'. Call
27077 finalize_string on XSTR and XTMPL fields.
27078 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
27079 "(nil)" values were omitted. Call the postprocess vfunc on the
27081 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
27082 class ctor. Initialize m_in_call_function_usage. Call
27083 one_time_initialization.
27084 * rtl-tests.c (selftest::test_uncond_jump): Call
27085 set_new_first_and_last_insn.
27086 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
27087 * selftest-rtl.c: New file.
27088 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
27089 (selftest::get_insn_by_uid): New decl.
27090 * selftest-run-tests.c (selftest::run_tests): Call
27091 read_rtl_function_c_tests.
27092 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
27093 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
27096 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
27098 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
27099 operands in a special way. Assert that pos+len <= mode precision.
27101 2017-01-05 Jakub Jelinek <jakub@redhat.com>
27103 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
27104 3 argument Alias with unlimited for the negative form.
27105 (fno-vect-cost-model): Removed.
27107 2017-01-05 Martin Liska <mliska@suse.cz>
27109 * hsa-gen.c (gen_hsa_divmod): New function.
27110 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
27112 2017-01-05 Martin Liska <mliska@suse.cz>
27115 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
27118 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27120 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
27121 small constant length operands.
27123 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27125 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
27126 between loop iterations.
27128 2017-01-05 Martin Liska <mliska@suse.cz>
27131 * gimplify.c (gimplify_decl_expr): Compare to
27132 asan_poisoned_variables instread of checking flags.
27133 (gimplify_target_expr): Likewise.
27134 (gimplify_expr): Likewise.
27135 (gimplify_function_tree): Conditionally initialize
27136 asan_poisoned_variables.
27138 2017-01-04 Jeff Law <law@redhat.com>
27140 PR tree-optimizatin/78812
27141 * rtl.h (contains_mem_rtx_p): Prototype.
27142 * ifcvt.c (containts_mem_rtx_p): Move from here to...
27143 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
27144 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
27145 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
27146 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
27148 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
27150 * input.c (assert_char_at_range): Default-initialize actual_range.
27152 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
27154 * df-scan.c (df_ref_create_structure): Make regno unsigned,
27155 to match the caller.
27157 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
27159 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
27160 insns after final jump in test to emit dummy move.
27162 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
27164 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
27165 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
27167 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
27169 * multiple_target.c (create_dispatcher_calls): Init e_next.
27170 * tree-ssa-loop-split.c (split_loop): Init border.
27171 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
27174 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
27179 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
27180 (altivec_register_operand): Do not return true if the operand
27181 contains a SUBREG mixing SImode and SFmode.
27182 (vsx_register_operand): Likewise.
27183 (vsx_reg_sfsubreg_ok): New predicate.
27184 (vfloat_operand): Do not return true if the operand contains a
27185 SUBREG mixing SImode and SFmode.
27186 (vint_operand): Likewise.
27187 (vlogical_operand): Likewise.
27188 (gpc_reg_operand): Likewise.
27189 (int_reg_operand): Likewise.
27190 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
27191 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
27192 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
27194 (rs6000_emit_move_si_sf_subreg): New helper function.
27195 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
27196 fixup SUBREGs involving SImode and SFmode.
27197 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
27198 numbers for the new peephole2 optimization.
27199 (peephole2 for SFmode unions): New peephole2 to optimize cases in
27200 the GLIBC math library that do AND/IOR/XOR operations on single
27201 precision floating point.
27202 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
27203 target macros to say whether we need to avoid SUBREGs mixing
27205 (TARGET_ALLOW_SF_SUBREG): Likewise.
27206 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
27207 (UNSPEC_SI_FROM_SF): Likewise.
27208 (iorxor): Change spacing.
27209 (and_ior_xor): New iterator for AND, IOR, and XOR.
27210 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
27211 (movdi_from_sf_zero_ext): Likewise.
27212 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
27213 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
27214 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
27215 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
27216 (fms<mode>4): Likewise.
27217 (fnma<mode>4): Likewise.
27218 (fnms<mode>4): Likewise.
27219 (nfma<mode>4): Likewise.
27220 (nfms<mode>4): Likewise.
27222 2017-01-04 Marek Polacek <polacek@redhat.com>
27225 * doc/invoke.texi: Document -Wpointer-compare.
27227 2017-01-04 Jakub Jelinek <jakub@redhat.com>
27229 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
27232 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
27233 descriptions for -gdwarf-5 and emit them as uleb128 instead of
27236 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
27239 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
27240 documentation of the powerpc_popcntb_ok attribute.
27241 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
27242 code to issue warning messages if a requested CPU configuration is
27243 not supported by the binary (assembler and loader) toolchain.
27244 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
27245 made to define a built-in function that has been disabled.
27246 (paired_init_builtins): Add assertion to prevent ICE if attempt is
27247 made to define a built-in function that has been disabled.
27248 (altivec_init_builtins): Add comment explaining why definition
27249 of the DST built-in functions is not preceded by an assertion
27250 check. Add assertions to prevent ICE if attempts are made to
27251 define an altivec predicate or an abs* built-in function that has
27253 (htm_init_builtins): Add comment explaining why definition of the
27254 htm built-in functions is not preceded by an assertion check.
27256 2017-01-04 Jeff Law <law@redhat.com>
27258 PR tree-optimizatin/67955
27259 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
27260 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
27261 the points-to solution does not include pt_null. Use DECL_PT_UID
27264 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
27266 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
27267 Use gen_int_mode instead of gen_lopwart for const_int operands.
27269 2017-01-04 Jakub Jelinek <jakub@redhat.com>
27271 PR tree-optimization/71563
27272 * match.pd: Simplify X << Y into X if Y is known to be 0 or
27273 out of range value - has low bits known to be zero.
27275 2017-01-04 Alan Modra <amodra@gmail.com>
27277 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
27278 * configure: Regenerate.
27279 * config.in: Regenerate.
27281 2017-01-04 Jakub Jelinek <jakub@redhat.com>
27284 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
27285 a substring of the message, but strcmp with the whole message. Ifdef
27286 ENABLE_NLS, translate the message first using dgettext.
27288 2017-01-03 Jeff Law <law@redhat.com>
27290 PR tree-optimizatin/78856
27291 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
27292 (mark_threaded_blocks): Remove code to truncate thread paths that
27293 cross multiple loop headers. Instead invalidate the cached loop
27294 iteration information and handle case of a thread path walking
27295 into an irreducible region.
27297 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
27300 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
27301 assertions. Add support for doing the signbit if the IEEE 128-bit
27302 floating point value is in a GPR.
27303 * config/rs6000/rs6000.md (Fsignbit): Delete.
27304 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
27305 Update the length attribute if the value is in a GPR.
27306 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
27307 the sign or zero extension instruction, since the value is always 0/1.
27308 (signbit<mode>2_dm2): Delete using <Fsignbit>.
27311 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
27312 extracting SImode to a GPR register so that we can generate a
27313 store, limit the vector to be in a traditional Altivec register
27314 for the vextuwrx instruction.
27316 2017-01-03 Ian Lance Taylor <iant@google.com>
27318 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
27320 2017-01-03 Martin Sebor <msebor@redhat.com>
27322 PR tree-optimization/78696
27323 * gimple-ssa-sprintf.c (format_floating): Correct handling of
27324 precision. Use MPFR for %f for greater fidelity. Correct handling
27326 (pass_sprintf_length::compute_format_length): Set width and precision
27327 specified by asrerisk to void_node for vararg functions.
27328 (try_substitute_return_value): Adjust dump output.
27330 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
27332 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
27334 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
27336 * doc/invoke.texi (SPARC options): Document -mlra as the default.
27337 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
27338 -mlra/-mno-lra was passed to the compiler.
27340 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
27342 PR rtl-optimization/65618
27343 * emit-rtl.c (try_split): Move initialization of "before" and
27344 "after" to just before the call to emit_insn_after_setloc.
27346 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
27348 * doc/md.texi (Standard Names): Remove reference to Java frontend.
27350 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
27352 * dwarf2out.c (gen_enumeration_type_die): When
27353 -gno-strict-dwarf, add a DW_AT_encoding attribute.
27355 2017-01-03 Jakub Jelinek <jakub@redhat.com>
27357 PR tree-optimization/78965
27358 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
27359 Change first argument from const call_info & to call_info &. For %n
27360 set info.nowrite to false.
27362 PR middle-end/78901
27363 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
27364 possibly throwing calls.
27366 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
27367 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
27368 and fns handling, rather than in a separate case SSA_NAME.
27370 2017-01-02 Jeff Law <law@redhat.com>
27372 * config/darwin-driver.c (darwin_driver_init): Const-correctness
27373 fixes for first_period and second_period variables.
27375 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
27378 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
27379 (*insvqi_1): New insn pattern.
27380 (*insvqi_1_mem_rex64): Ditto.
27381 (*insvqi_2): Ditto.
27382 (*insvqi_3): Rename from *insvqi.
27384 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
27386 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
27388 * doc/cfg.texi (Edges): Remove reference to Java.
27389 (Maintaining the CFG): Ditto.
27391 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
27393 PR middle-end/77674
27394 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
27395 transparent aliases.
27397 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
27399 PR middle-end/77484
27400 * predict.def (PRED_CALL): Update hitrate.
27401 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
27402 * predict.c (tree_estimate_probability_bb): Split CALL predictor
27403 into direct/indirect/polymorphic variants.
27405 2017-01-01 Jakub Jelinek <jakub@redhat.com>
27407 Update copyright years.
27409 * gcc.c (process_command): Update copyright notice dates.
27410 * gcov-dump.c (print_version): Ditto.
27411 * gcov.c (print_version): Ditto.
27412 * gcov-tool.c (print_version): Ditto.
27413 * gengtype.c (create_file): Ditto.
27414 * doc/cpp.texi: Bump @copying's copyright year.
27415 * doc/cppinternals.texi: Ditto.
27416 * doc/gcc.texi: Ditto.
27417 * doc/gccint.texi: Ditto.
27418 * doc/gcov.texi: Ditto.
27419 * doc/install.texi: Ditto.
27420 * doc/invoke.texi: Ditto.
27422 Copyright (C) 2017 Free Software Foundation, Inc.
27424 Copying and distribution of this file, with or without modification,
27425 are permitted in any medium without royalty provided the copyright
27426 notice and this notice are preserved.