1 2019-01-12 Svante Signell <svante.signell@gmail.com>
3 * config/i386/gnu.h (TARGET_THREAD_SSP_OFFSET): Define.
4 (TARGET_CAN_SPLIT_STACK): Define.
5 (TARGET_THREAD_SPLIT_STACK_OFFSET): Define.
7 2019-01-05 Jan Hubicka <hubicka@ucw.cz>
9 * params.def (inline-unit-growth): Set to 40.
11 2019-01-12 Jakub Jelinek <jakub@redhat.com>
13 * tree-ssa-loop-ivopts.c (find_inv_vars): Fix a comment typo.
15 2019-01-12 Tom de Vries <tdevries@suse.de>
17 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims_1): In offloading
18 region calling vector-partitionable routine, set default_vector_length
21 2019-01-12 Tom de Vries <tdevries@suse.de>
23 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims_1): Add an use new
24 variable default_vector_length.
26 2019-01-12 Tom de Vries <tdevries@suse.de>
29 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims_1): Apply defaults
30 from oacc_default_dims, as oacc_validate_dims would do it, and apply
33 2019-01-12 Tom de Vries <tdevries@suse.de>
35 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims_1)
36 (nvptx_goacc_validate_dims): Add used parameter.
37 * doc/tm.texi: Regenerate.
38 * omp-offload.c (oacc_parse_default_dims, oacc_validate_dims): Add
39 argument to call to targetm.goacc.validate_dims.
40 (default_goacc_validate_dims): Add used
42 * target.def (validate_dims): Add used parameter in DEFHOOK.
43 * targhooks.h (default_goacc_validate_dims): Add used parameter.
45 2019-01-11 Jakub Jelinek <jakub@redhat.com>
49 * tree-inline.h (struct copy_body_data): Add adjust_array_error_bounds
51 * tree-inline.c (remap_type_1): Formatting fix. If TYPE_MAX_VALUE of
52 ARRAY_TYPE's TYPE_DOMAIN is newly error_mark_node, replace it with
53 a dummy "omp dummy var" variable if id->adjust_array_error_bounds.
54 * omp-low.c (new_omp_context): Set cb.adjust_array_error_bounds.
56 2019-01-11 Vladimir Makarov <vmakarov@redhat.com>
58 PR rtl-optimization/87305
60 (setup_live_pseudos_and_spill_after_risky_transforms): Add code
61 for little endian pseudos used as paradoxical subreg.
63 2019-01-11 Jakub Jelinek <jakub@redhat.com>
65 PR tree-optimization/88693
66 * tree-ssa-strlen.c (get_min_string_length): Don't set *full_string_p
67 for STRING_CSTs that don't contain any NUL characters in the first
68 TREE_STRING_LENGTH bytes.
70 2019-01-11 Alan Modra <amodra@gmail.com>
74 * genattrtab.c (min_fn): Don't translate values.
75 (min_attr_value): Return INT_MAX when the value can't be calculated.
76 Return minimum among any values that can be calculated.
77 (max_attr_value): Adjust.
79 2019-01-11 Jakub Jelinek <jakub@redhat.com>
81 * Makefile.in (PLUGIN_HEADERS): Add $(INSN_ATTR_H).
83 2019-01-11 Steve Ellcey <sellcey@marvell.com>
85 * config/aarch64/aarch64.c (aarch64_simd_call_p): New function.
86 (aarch64_hard_regno_call_part_clobbered): Add insn argument.
87 (aarch64_return_call_with_max_clobbers): New function.
88 (TARGET_RETURN_CALL_WITH_MAX_CLOBBERS): New macro.
89 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Add insn
91 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): Ditto.
92 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Ditto.
93 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered): Ditto.
94 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): Ditto.
95 * cselib.c (cselib_process_insn): Add argument to
96 targetm.hard_regno_call_part_clobbered call.
97 * ira-conflicts.c (ira_build_conflicts): Ditto.
98 * ira-costs.c (ira_tune_allocno_costs): Ditto.
99 * lra-constraints.c (inherit_reload_reg): Ditto.
100 * lra-int.h (struct lra_reg): Add call_insn field, remove call_p field.
101 * lra-lives.c (check_pseudos_live_through_calls): Add call_insn
102 argument. Call targetm.return_call_with_max_clobbers.
103 Add argument to targetm.hard_regno_call_part_clobbered call.
104 (calls_have_same_clobbers_p): New function.
105 (process_bb_lives): Add call_insn and last_call_insn variables.
106 Pass call_insn to check_pseudos_live_through_calls.
107 Modify if stmt to check targetm.return_call_with_max_clobbers.
108 Update setting of flush variable.
109 (lra_create_live_ranges_1): Set call_insn to NULL instead of call_p
111 * lra.c (initialize_lra_reg_info_element): Set call_insn to NULL.
112 * regcprop.c (copyprop_hardreg_forward_1): Add argument to
113 targetm.hard_regno_call_part_clobbered call.
114 * reginfo.c (choose_hard_reg_mode): Ditto.
115 * regrename.c (check_new_reg_p): Ditto.
116 * reload.c (find_equiv_reg): Ditto.
117 * reload1.c (emit_reload_insns): Ditto.
118 * sched-deps.c (deps_analyze_insn): Ditto.
119 * sel-sched.c (init_regs_for_mode): Ditto.
120 (mark_unavailable_hard_regs): Ditto.
121 * targhooks.c (default_dwarf_frame_reg_mode): Ditto.
122 * target.def (hard_regno_call_part_clobbered): Add insn argument.
123 (return_call_with_max_clobbers): New target function.
124 * doc/tm.texi: Regenerate.
125 * doc/tm.texi.in (TARGET_RETURN_CALL_WITH_MAX_CLOBBERS): New hook.
126 * hooks.c (hook_bool_uint_mode_false): Change to
127 hook_bool_insn_uint_mode_false.
128 * hooks.h (hook_bool_uint_mode_false): Ditto.
130 2019-01-11 Steve Ellcey <sellcey@marvell.com>
132 * config/aarch64/aarch64.c (aarch64_simd_call_p): New function.
133 (aarch64_remove_extra_call_preserved_regs): New function.
134 (TARGET_REMOVE_EXTRA_CALL_PRESERVED_REGS): New macro.
135 * doc/tm.texi.in (TARGET_REMOVE_EXTRA_CALL_PRESERVED_REGS): New hook.
136 * doc/tm.texi: Regenerate.
137 * final.c (get_call_reg_set_usage): Call new hook.
138 * target.def (remove_extra_call_preserved_regs): New hook.
139 * targhooks.c (default_remove_extra_call_preserved_regs): New function.
140 * targhooks.h (default_remove_extra_call_preserved_regs): New function.
142 2019-01-11 Jakub Jelinek <jakub@redhat.com>
145 * passes.c (finish_optimization_passes): Call print_combine_total_stats
146 inside of pass_combine_1 dump rather than pass_profile_1.
148 2019-01-11 Tom de Vries <tdevries@suse.de>
150 * config/nvptx/nvptx.c (PTX_CTA_NUM_BARRIERS, PTX_PER_CTA_BARRIER)
151 (PTX_NUM_PER_CTA_BARRIER, PTX_FIRST_PER_WORKER_BARRIER)
152 (PTX_NUM_PER_WORKER_BARRIERS): Define.
153 (nvptx_apply_dim_limits): Prevent vector_length 64 and
156 2019-01-11 Tom de Vries <tdevries@suse.de>
158 * config/nvptx/nvptx.c (PTX_CTA_SIZE): Move up.
160 2019-01-11 Jan Beulich <jbeulich@suse.com>
162 * config/i386/i386.md (rex64suffix): Add L suffix for SI.
163 * config/i386/sse.md (cvtusi2<ssescalarmodesuffix>32<round_name>,
164 sse2_cvtsi2sd): Add {l}.
165 (sse2_cvtsi2sdq<round_name>): Make q conditional upon AT&T
168 2019-01-10 Jakub Jelinek <jakub@redhat.com>
171 * config/i386/sse.md (float<floatunssuffix>v2div2sf2): Turn into
173 (*float<floatunssuffix>v2div2sf2): New define_insn.
174 (float<floatunssuffix>v2div2sf2_mask): Turn into define_expand.
175 (*float<floatunssuffix>v2div2sf2_mask): New define_insn.
176 (*float<floatunssuffix>v2div2sf2_mask_1): Replace
177 subrtxes (const_vector:V2SF [(const_int 0) (const_int 0)]) with
178 match_operands with "const0_operand" "C".
180 2019-01-10 Tamar Christina <tamar.christina@arm.com>
182 * config/aarch64/aarch64-builtins.c
183 (aarch64_init_builtins): Move aarch64_init_fcmla_laneq_builtins...
184 (aarch64_init_simd_builtins): ...Here
186 2019-01-10 Vladimir Makarov <vmakarov@redhat.com>
188 PR rtl-optimization/87305
190 (setup_live_pseudos_and_spill_after_risky_transforms): Check
191 allocation for big endian pseudos used as paradoxical subregs and
192 spill them if it is wrong.
193 * lra-constraints.c (lra_constraints): Add a comment.
195 2019-01-10 Richard Biener <rguenther@suse.de>
197 PR tree-optimization/88792
198 * tree-ssa-pre.c (get_representative_for): Do not return a
201 2019-01-10 Jakub Jelinek <jakub@redhat.com>
205 * function.c (assign_stack_local_1): Revert the 2018-11-21 changes.
206 (assign_parm_setup_block): Do the argument slot realignment here
209 2019-01-10 Stefan Agner <stefan@agner.ch>
212 * config/arm/arm.c (arm_option_override_internal): Force
213 opts->x_inline_asm_unified to true only if TARGET_THUMB2_P.
215 2019-01-10 Jakub Jelinek <jakub@redhat.com>
218 * attribs.c (handle_dll_attribute): Clear TREE_STATIC after setting
221 2019-01-10 Tamar Christina <tamar.christina@arm.com>
223 * config/arm/arm-builtins.c
224 (enum arm_type_qualifiers): Add qualifier_lane_pair_index.
225 (MAC_LANE_PAIR_QUALIFIERS): New.
226 (arm_expand_builtin_args): Use it.
227 (arm_expand_builtin_1): Likewise.
228 * config/arm/arm-protos.h (neon_vcmla_lane_prepare_operands): New.
229 * config/arm/arm.c (neon_vcmla_lane_prepare_operands): New.
230 * config/arm/arm-c.c (arm_cpu_builtins): Add __ARM_FEATURE_COMPLEX.
231 * config/arm/arm_neon.h:
232 (vcadd_rot90_f16): New.
233 (vcaddq_rot90_f16): New.
234 (vcadd_rot270_f16): New.
235 (vcaddq_rot270_f16): New.
238 (vcmla_lane_f16): New.
239 (vcmla_laneq_f16): New.
240 (vcmlaq_lane_f16): New.
241 (vcmlaq_laneq_f16): New.
242 (vcmla_rot90_f16): New.
243 (vcmlaq_rot90_f16): New.
244 (vcmla_rot90_lane_f16): New.
245 (vcmla_rot90_laneq_f16): New.
246 (vcmlaq_rot90_lane_f16): New.
247 (vcmlaq_rot90_laneq_f16): New.
248 (vcmla_rot180_f16): New.
249 (vcmlaq_rot180_f16): New.
250 (vcmla_rot180_lane_f16): New.
251 (vcmla_rot180_laneq_f16): New.
252 (vcmlaq_rot180_lane_f16): New.
253 (vcmlaq_rot180_laneq_f16): New.
254 (vcmla_rot270_f16): New.
255 (vcmlaq_rot270_f16): New.
256 (vcmla_rot270_lane_f16): New.
257 (vcmla_rot270_laneq_f16): New.
258 (vcmlaq_rot270_lane_f16): New.
259 (vcmlaq_rot270_laneq_f16): New.
260 (vcadd_rot90_f32): New.
261 (vcaddq_rot90_f32): New.
262 (vcadd_rot270_f32): New.
263 (vcaddq_rot270_f32): New.
266 (vcmla_lane_f32): New.
267 (vcmla_laneq_f32): New.
268 (vcmlaq_lane_f32): New.
269 (vcmlaq_laneq_f32): New.
270 (vcmla_rot90_f32): New.
271 (vcmlaq_rot90_f32): New.
272 (vcmla_rot90_lane_f32): New.
273 (vcmla_rot90_laneq_f32): New.
274 (vcmlaq_rot90_lane_f32): New.
275 (vcmlaq_rot90_laneq_f32): New.
276 (vcmla_rot180_f32): New.
277 (vcmlaq_rot180_f32): New.
278 (vcmla_rot180_lane_f32): New.
279 (vcmla_rot180_laneq_f32): New.
280 (vcmlaq_rot180_lane_f32): New.
281 (vcmlaq_rot180_laneq_f32): New.
282 (vcmla_rot270_f32): New.
283 (vcmlaq_rot270_f32): New.
284 (vcmla_rot270_lane_f32): New.
285 (vcmla_rot270_laneq_f32): New.
286 (vcmlaq_rot270_lane_f32): New.
287 (vcmlaq_rot270_laneq_f32): New.
288 * config/arm/arm_neon_builtins.def (vcadd90, vcadd270, vcmla0, vcmla90,
289 vcmla180, vcmla270, vcmla_lane0, vcmla_lane90, vcmla_lane180, vcmla_lane270,
290 vcmla_laneq0, vcmla_laneq90, vcmla_laneq180, vcmla_laneq270,
291 vcmlaq_lane0, vcmlaq_lane90, vcmlaq_lane180, vcmlaq_lane270): New.
292 * config/arm/neon.md (neon_vcmla_lane<rot><mode>,
293 neon_vcmla_laneq<rot><mode>, neon_vcmlaq_lane<rot><mode>): New.
294 * config/arm/arm.c (arm_arch8_3, arm_arch8_4): New.
295 * config/arm/arm.h (TARGET_COMPLEX, arm_arch8_3, arm_arch8_4): New.
296 (arm_option_reconfigure_globals): Use them.
297 * config/arm/iterators.md (VDF, VQ_HSF): New.
299 (VF_constraint, rot, rotsplit1, rotsplit2): Add V4HF and V8HF.
300 * config/arm/neon.md (neon_vcadd<rot><mode>, neon_vcmla<rot><mode>): New.
301 * config/arm/unspecs.md (UNSPEC_VCADD90, UNSPEC_VCADD270,
302 UNSPEC_VCMLA, UNSPEC_VCMLA90, UNSPEC_VCMLA180, UNSPEC_VCMLA270): New.
304 2019-01-10 Tamar Christina <tamar.christina@arm.com>
306 * config/aarch64/aarch64-builtins.c (enum aarch64_type_qualifiers): Add qualifier_lane_pair_index.
307 (emit-rtl.h): Include.
308 (TYPES_QUADOP_LANE_PAIR): New.
309 (aarch64_simd_expand_args): Use it.
310 (aarch64_simd_expand_builtin): Likewise.
311 (AARCH64_SIMD_FCMLA_LANEQ_BUILTINS, aarch64_fcmla_laneq_builtin_datum): New.
312 (FCMLA_LANEQ_BUILTIN, AARCH64_SIMD_FCMLA_LANEQ_BUILTIN_BASE,
313 AARCH64_SIMD_FCMLA_LANEQ_BUILTINS, aarch64_fcmla_lane_builtin_data,
314 aarch64_init_fcmla_laneq_builtins, aarch64_expand_fcmla_builtin): New.
315 (aarch64_init_builtins): Add aarch64_init_fcmla_laneq_builtins.
316 (aarch64_expand_buildin): Add AARCH64_SIMD_BUILTIN_FCMLA_LANEQ0_V2SF,
317 AARCH64_SIMD_BUILTIN_FCMLA_LANEQ90_V2SF, AARCH64_SIMD_BUILTIN_FCMLA_LANEQ180_V2SF,
318 AARCH64_SIMD_BUILTIN_FCMLA_LANEQ2700_V2SF, AARCH64_SIMD_BUILTIN_FCMLA_LANEQ0_V4HF,
319 AARCH64_SIMD_BUILTIN_FCMLA_LANEQ90_V4HF, AARCH64_SIMD_BUILTIN_FCMLA_LANEQ180_V4HF,
320 AARCH64_SIMD_BUILTIN_FCMLA_LANEQ270_V4HF.
321 * config/aarch64/iterators.md (FCMLA_maybe_lane): New.
322 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins): Add __ARM_FEATURE_COMPLEX.
323 * config/aarch64/aarch64-simd-builtins.def (fcadd90, fcadd270, fcmla0, fcmla90,
324 fcmla180, fcmla270, fcmla_lane0, fcmla_lane90, fcmla_lane180, fcmla_lane270,
325 fcmla_laneq0, fcmla_laneq90, fcmla_laneq180, fcmla_laneq270,
326 fcmlaq_lane0, fcmlaq_lane90, fcmlaq_lane180, fcmlaq_lane270): New.
327 * config/aarch64/aarch64-simd.md (aarch64_fcmla_lane<rot><mode>,
328 aarch64_fcmla_laneq<rot>v4hf, aarch64_fcmlaq_lane<rot><mode>,aarch64_fcadd<rot><mode>,
329 aarch64_fcmla<rot><mode>): New.
330 * config/aarch64/arm_neon.h:
331 (vcadd_rot90_f16): New.
332 (vcaddq_rot90_f16): New.
333 (vcadd_rot270_f16): New.
334 (vcaddq_rot270_f16): New.
337 (vcmla_lane_f16): New.
338 (vcmla_laneq_f16): New.
339 (vcmlaq_lane_f16): New.
340 (vcmlaq_rot90_lane_f16): New.
341 (vcmla_rot90_laneq_f16): New.
342 (vcmla_rot90_lane_f16): New.
343 (vcmlaq_rot90_f16): New.
344 (vcmla_rot90_f16): New.
345 (vcmlaq_laneq_f16): New.
346 (vcmla_rot180_laneq_f16): New.
347 (vcmla_rot180_lane_f16): New.
348 (vcmlaq_rot180_f16): New.
349 (vcmla_rot180_f16): New.
350 (vcmlaq_rot90_laneq_f16): New.
351 (vcmlaq_rot270_laneq_f16): New.
352 (vcmlaq_rot270_lane_f16): New.
353 (vcmla_rot270_laneq_f16): New.
354 (vcmlaq_rot270_f16): New.
355 (vcmla_rot270_f16): New.
356 (vcmlaq_rot180_laneq_f16): New.
357 (vcmlaq_rot180_lane_f16): New.
358 (vcmla_rot270_lane_f16): New.
359 (vcadd_rot90_f32): New.
360 (vcaddq_rot90_f32): New.
361 (vcaddq_rot90_f64): New.
362 (vcadd_rot270_f32): New.
363 (vcaddq_rot270_f32): New.
364 (vcaddq_rot270_f64): New.
368 (vcmla_lane_f32): New.
369 (vcmla_laneq_f32): New.
370 (vcmlaq_lane_f32): New.
371 (vcmlaq_laneq_f32): New.
372 (vcmla_rot90_f32): New.
373 (vcmlaq_rot90_f32): New.
374 (vcmlaq_rot90_f64): New.
375 (vcmla_rot90_lane_f32): New.
376 (vcmla_rot90_laneq_f32): New.
377 (vcmlaq_rot90_lane_f32): New.
378 (vcmlaq_rot90_laneq_f32): New.
379 (vcmla_rot180_f32): New.
380 (vcmlaq_rot180_f32): New.
381 (vcmlaq_rot180_f64): New.
382 (vcmla_rot180_lane_f32): New.
383 (vcmla_rot180_laneq_f32): New.
384 (vcmlaq_rot180_lane_f32): New.
385 (vcmlaq_rot180_laneq_f32): New.
386 (vcmla_rot270_f32): New.
387 (vcmlaq_rot270_f32): New.
388 (vcmlaq_rot270_f64): New.
389 (vcmla_rot270_lane_f32): New.
390 (vcmla_rot270_laneq_f32): New.
391 (vcmlaq_rot270_lane_f32): New.
392 (vcmlaq_rot270_laneq_f32): New.
393 * config/aarch64/aarch64.h (TARGET_COMPLEX): New.
394 * config/aarch64/iterators.md (UNSPEC_FCADD90, UNSPEC_FCADD270,
395 UNSPEC_FCMLA, UNSPEC_FCMLA90, UNSPEC_FCMLA180, UNSPEC_FCMLA270): New.
398 * config/arm/types.md (neon_fcadd, neon_fcmla): New.
400 2019-01-09 Sandra Loosemore <sandra@codesourcery.com>
404 * config/pa/pa.c: Change "can not" to "cannot".
405 * gimple-ssa-evrp-analyze.c: Likewise.
406 * ipa-icf.c: Likewise.
407 * ipa-polymorphic-call.c: Likewise.
408 * ipa-pure-const.c: Likewise.
409 * lra-constraints.c: Likewise.
410 * lra-remat.c: Likewise.
411 * reload1.c: Likewise.
413 * tree-ssa-uninit.c: Likewise.
415 2019-01-09 Sandra Loosemore <sandra@codesourcery.com>
419 * Makefile.in: Mechanically replace "can not" with "cannot".
421 * builtins.c: Likewise.
423 * cgraph.c: Likewise.
424 * cgraph.h: Likewise.
425 * cgraphclones.c: Likewise.
426 * cgraphunit.c: Likewise.
427 * combine-stack-adj.c: Likewise.
428 * combine.c: Likewise.
429 * common/config/i386/i386-common.c: Likewise.
430 * config/aarch64/aarch64.c: Likewise.
431 * config/alpha/sync.md: Likewise.
432 * config/arc/arc.c: Likewise.
433 * config/arc/predicates.md: Likewise.
434 * config/arm/arm-c.c: Likewise.
435 * config/arm/arm.c: Likewise.
436 * config/arm/arm.h: Likewise.
437 * config/arm/arm.md: Likewise.
438 * config/arm/cortex-r4f.md: Likewise.
439 * config/csky/csky.c: Likewise.
440 * config/csky/csky.h: Likewise.
441 * config/darwin-f.c: Likewise.
442 * config/epiphany/epiphany.md: Likewise.
443 * config/i386/i386.c: Likewise.
444 * config/i386/sol2.h: Likewise.
445 * config/m68k/m68k.c: Likewise.
446 * config/mcore/mcore.h: Likewise.
447 * config/microblaze/microblaze.md: Likewise.
448 * config/mips/20kc.md: Likewise.
449 * config/mips/sb1.md: Likewise.
450 * config/nds32/nds32.c: Likewise.
451 * config/nds32/predicates.md: Likewise.
452 * config/pa/pa.c: Likewise.
453 * config/rs6000/e300c2c3.md: Likewise.
454 * config/rs6000/rs6000.c: Likewise.
455 * config/s390/s390.h: Likewise.
456 * config/sh/sh.c: Likewise.
457 * config/sh/sh.md: Likewise.
458 * config/spu/vmx2spu.h: Likewise.
460 * dbxout.c: Likewise.
461 * df-scan.c: Likewise.
462 * doc/cfg.texi: Likewise.
463 * doc/extend.texi: Likewise.
464 * doc/fragments.texi: Likewise.
465 * doc/gty.texi: Likewise.
466 * doc/invoke.texi: Likewise.
467 * doc/lto.texi: Likewise.
468 * doc/md.texi: Likewise.
469 * doc/objc.texi: Likewise.
470 * doc/rtl.texi: Likewise.
471 * doc/tm.texi: Likewise.
473 * emit-rtl.c: Likewise.
474 * emit-rtl.h: Likewise.
475 * except.c: Likewise.
476 * expmed.c: Likewise.
478 * fold-const.c: Likewise.
479 * genautomata.c: Likewise.
480 * gimple-fold.c: Likewise.
481 * hard-reg-set.h: Likewise.
483 * ipa-comdats.c: Likewise.
484 * ipa-cp.c: Likewise.
485 * ipa-devirt.c: Likewise.
486 * ipa-fnsummary.c: Likewise.
487 * ipa-icf.c: Likewise.
488 * ipa-inline-transform.c: Likewise.
489 * ipa-inline.c: Likewise.
490 * ipa-polymorphic-call.c: Likewise.
491 * ipa-profile.c: Likewise.
492 * ipa-prop.c: Likewise.
493 * ipa-pure-const.c: Likewise.
494 * ipa-reference.c: Likewise.
495 * ipa-split.c: Likewise.
496 * ipa-visibility.c: Likewise.
498 * ira-build.c: Likewise.
499 * ira-color.c: Likewise.
500 * ira-conflicts.c: Likewise.
501 * ira-costs.c: Likewise.
502 * ira-int.h: Likewise.
503 * ira-lives.c: Likewise.
506 * loop-invariant.c: Likewise.
507 * loop-unroll.c: Likewise.
508 * lower-subreg.c: Likewise.
509 * lra-assigns.c: Likewise.
510 * lra-constraints.c: Likewise.
511 * lra-eliminations.c: Likewise.
512 * lra-lives.c: Likewise.
513 * lra-remat.c: Likewise.
514 * lra-spills.c: Likewise.
516 * lto-cgraph.c: Likewise.
517 * lto-streamer-out.c: Likewise.
518 * postreload-gcse.c: Likewise.
519 * predict.c: Likewise.
520 * profile-count.h: Likewise.
521 * profile.c: Likewise.
524 * reload.c: Likewise.
525 * reload1.c: Likewise.
527 * resource.c: Likewise.
530 * rtlanal.c: Likewise.
531 * sched-deps.c: Likewise.
532 * sched-ebb.c: Likewise.
533 * sched-rgn.c: Likewise.
534 * sel-sched-ir.c: Likewise.
535 * sel-sched.c: Likewise.
536 * shrink-wrap.c: Likewise.
537 * simplify-rtx.c: Likewise.
538 * symtab.c: Likewise.
539 * target.def: Likewise.
540 * toplev.c: Likewise.
541 * tree-call-cdce.c: Likewise.
542 * tree-cfg.c: Likewise.
543 * tree-complex.c: Likewise.
544 * tree-core.h: Likewise.
545 * tree-eh.c: Likewise.
546 * tree-inline.c: Likewise.
547 * tree-loop-distribution.c: Likewise.
548 * tree-nrv.c: Likewise.
549 * tree-profile.c: Likewise.
550 * tree-sra.c: Likewise.
551 * tree-ssa-alias.c: Likewise.
552 * tree-ssa-dce.c: Likewise.
553 * tree-ssa-dom.c: Likewise.
554 * tree-ssa-forwprop.c: Likewise.
555 * tree-ssa-loop-im.c: Likewise.
556 * tree-ssa-loop-ivcanon.c: Likewise.
557 * tree-ssa-loop-ivopts.c: Likewise.
558 * tree-ssa-loop-niter.c: Likewise.
559 * tree-ssa-phionlycprop.c: Likewise.
560 * tree-ssa-phiopt.c: Likewise.
561 * tree-ssa-propagate.c: Likewise.
562 * tree-ssa-threadedge.c: Likewise.
563 * tree-ssa-threadupdate.c: Likewise.
564 * tree-ssa-uninit.c: Likewise.
565 * tree-ssanames.c: Likewise.
566 * tree-streamer-out.c: Likewise.
569 * vr-values.c: Likewise.
571 2019-01-09 Uroš Bizjak <ubizjak@gmail.com>
573 * config/i386/i386-protos.h (ix86_expand_xorsign): New prototype.
574 (ix86_split_xorsign): Ditto.
575 * config/i386/i386.c (ix86_expand_xorsign): New function.
576 (ix86_split_xorsign): Ditto.
577 * config/i386/i386.md (UNSPEC_XORSIGN): New unspec.
578 (xorsign<mode>3): New expander.
579 (xorsign<mode>3_1): New insn_and_split pattern.
580 * config/i386/sse.md (xorsign<mode>3): New expander.
582 2019-01-09 Eric Botcazou <ebotcazou@adacore.com>
584 * config/sparc/sparc.md (*tablejump_sp32): Merge into...
585 (*tablejump_sp64): Likewise.
586 (*tablejump<P:mode>): ...this.
587 (*call_address_sp32): Merge into...
588 (*call_address_sp64): Likewise.
589 (*call_address<P:mode>): ...this.
590 (*call_symbolic_sp32): Merge into...
591 (*call_symbolic_sp64): Likewise.
592 (*call_symbolic<P:mode>): ...this.
593 (call_value): Remove constraint and add predicate.
594 (*call_value_address_sp32): Merge into...
595 (*call_value_address_sp64): Likewise.
596 (*call_value_address<P:mode>): ...this.
597 (*call_value_symbolic_sp32): Merge into...
598 (*call_value_symbolic_sp64): Likewise.
599 (*call_value_symbolic<P:mode>): ...this.
600 (*sibcall_symbolic_sp32): Merge into...
601 (*sibcall_symbolic_sp64): Likewise.
602 (*sibcall_symbolic<P:mode>): ...this.
603 (sibcall_value): Remove constraint and add predicate.
604 (*sibcall_value_symbolic_sp32): Merge into...
605 (*sibcall_value_symbolic_sp64): Likewise.
606 (*sibcall_value_symbolic<P:mode>): ...this.
607 (window_save): Minor tweak.
608 (*branch_sp32): Merge into...
609 (*branch_sp64): Likewise.
610 (*branch<P:mode>): ...this.
612 2019-01-09 Eric Botcazou <ebotcazou@adacore.com>
613 James Clarke <jrtc27@jrtc27.com>
616 * config/sparc/sparc.c (sparc_legitimize_tls_address): Only use Pmode
617 consistently in TLS address generation and adjust code to the renaming
618 of patterns. Mark calls to __tls_get_addr as const.
619 * config/sparc/sparc.md (tgd_hi22): Turn into...
620 (tgd_hi22<P:mode>): ...this and use Pmode throughout.
621 (tgd_lo10): Turn into...
622 (tgd_lo10<P:mode>): ...this and use Pmode throughout.
623 (tgd_add32): Merge into...
624 (tgd_add64): Likewise.
625 (tgd_add<P:mode>): ...this and use Pmode throughout.
626 (tldm_hi22): Turn into...
627 (tldm_hi22<P:mode>): ...this and use Pmode throughout.
628 (tldm_lo10): Turn into...
629 (tldm_lo10<P:mode>): ...this and use Pmode throughout.
630 (tldm_add32): Merge into...
631 (tldm_add64): Likewise.
632 (tldm_add<P:mode>): ...this and use Pmode throughout.
633 (tldm_call32): Merge into...
634 (tldm_call64): Likewise.
635 (tldm_call<P:mode>): ...this and use Pmode throughout.
636 (tldo_hix22): Turn into...
637 (tldo_hix22<P:mode>): ...this and use Pmode throughout.
638 (tldo_lox10): Turn into...
639 (tldo_lox10<P:mode>): ...this and use Pmode throughout.
640 (tldo_add32): Merge into...
641 (tldo_add64): Likewise.
642 (tldo_add<P:mode>): ...this and use Pmode throughout.
643 (tie_hi22): Turn into...
644 (tie_hi22<P:mode>): ...this and use Pmode throughout.
645 (tie_lo10): Turn into...
646 (tie_lo10<P:mode>): ...this and use Pmode throughout.
647 (tie_ld64): Use DImode throughout.
648 (tie_add32): Merge into...
649 (tie_add64): Likewise.
650 (tie_add<P:mode>): ...this and use Pmode throughout.
651 (tle_hix22_sp32): Merge into...
652 (tle_hix22_sp64): Likewise.
653 (tle_hix22<P:mode>): ...this and use Pmode throughout.
654 (tle_lox22_sp32): Merge into...
655 (tle_lox22_sp64): Likewise.
656 (tle_lox22<P:mode>): ...this and use Pmode throughout.
657 (*tldo_ldub_sp32): Merge into...
658 (*tldo_ldub_sp64): Likewise.
659 (*tldo_ldub<P:mode>): ...this and use Pmode throughout.
660 (*tldo_ldub1_sp32): Merge into...
661 (*tldo_ldub1_sp64): Likewise.
662 (*tldo_ldub1<P:mode>): ...this and use Pmode throughout.
663 (*tldo_ldub2_sp32): Merge into...
664 (*tldo_ldub2_sp64): Likewise.
665 (*tldo_ldub2<P:mode>): ...this and use Pmode throughout.
666 (*tldo_ldsb1_sp32): Merge into...
667 (*tldo_ldsb1_sp64): Likewise.
668 (*tldo_ldsb1<P:mode>): ...this and use Pmode throughout.
669 (*tldo_ldsb2_sp32): Merge into...
670 (*tldo_ldsb2_sp64): Likewise.
671 (*tldo_ldsb2<P:mode>): ...this and use Pmode throughout.
672 (*tldo_ldub3_sp64): Use DImode throughout.
673 (*tldo_ldsb3_sp64): Likewise.
674 (*tldo_lduh_sp32): Merge into...
675 (*tldo_lduh_sp64): Likewise.
676 (*tldo_lduh<P:mode>): ...this and use Pmode throughout.
677 (*tldo_lduh1_sp32): Merge into...
678 (*tldo_lduh1_sp64): Likewise.
679 (*tldo_lduh1<P:mode>): ...this and use Pmode throughout.
680 (*tldo_ldsh1_sp32): Merge into...
681 (*tldo_ldsh1_sp64): Likewise.
682 (*tldo_ldsh1<P:mode>): ...this and use Pmode throughout.
683 (*tldo_lduh2_sp64): Use DImode throughout.
684 (*tldo_ldsh2_sp64): Likewise.
685 (*tldo_lduw_sp32): Merge into...
686 (*tldo_lduw_sp64): Likewise.
687 (*tldo_lduw<P:mode>): ...this and use Pmode throughout.
688 (*tldo_lduw1_sp64): Use DImode throughout.
689 (*tldo_ldsw1_sp64): Likewise.
690 (*tldo_ldx_sp64): Likewise.
691 (*tldo_stb_sp32): Merge into...
692 (*tldo_stb_sp64): Likewise.
693 (*tldo_stb<P:mode>): ...this and use Pmode throughout.
694 (*tldo_sth_sp32): Merge into...
695 (*tldo_sth_sp64): Likewise.
696 (*tldo_sth<P:mode>): ...this and use Pmode throughout.
697 (*tldo_stw_sp32): Merge into...
698 (*tldo_stw_sp64): Likewise.
699 (*tldo_stw<P:mode>): ...this and use Pmode throughout.
700 (*tldo_stx_sp64): Use DImode throughout.
702 2018-01-09 Sudakshina Das <sudi.das@arm.com>
704 * config/aarch64/aarch64.c (aarch64_override_options): Add case to
705 check configure option to set BTI and Return Address Signing.
706 * configure.ac: Add --enable-standard-branch-protection and
707 --disable-standard-branch-protection.
708 * configure: Regenerated.
709 * doc/install.texi: Document the same.
711 2018-01-09 Sudakshina Das <sudi.das@arm.com>
712 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
714 * config.gcc (aarch64*-*-*): Add aarch64-bti-insert.o.
715 * config/aarch64/aarch64.h: Update comment for TRAMPOLINE_SIZE.
716 * config/aarch64/aarch64.c (aarch64_asm_trampoline_template): Update
718 * config/aarch64/aarch64-bti-insert.c: New file.
719 * config/aarch64/aarch64-passes.def (INSERT_PASS_BEFORE): Insert bti
721 * config/aarch64/aarch64-protos.h (make_pass_insert_bti): Declare the
723 * config/aarch64/aarch64.md (unspecv): Add UNSPECV_BTI_NOARG,
724 UNSPECV_BTI_C, UNSPECV_BTI_J and UNSPECV_BTI_JC.
725 (bti_noarg, bti_j, bti_c, bti_jc): New define_insns.
726 * config/aarch64/t-aarch64: Add rule for aarch64-bti-insert.o.
728 2018-01-09 Sudakshina Das <sudi.das@arm.com>
730 * config/aarch64/aarch64-protos.h (aarch64_bti_enabled): Declare.
731 * config/aarch64/aarch64.c (aarch64_handle_no_branch_protection):
732 Disable bti for -mbranch-protection=none.
733 (aarch64_handle_standard_branch_protection): Enable bti for
734 -mbranch-protection=standard.
735 (aarch64_handle_bti_protection): Enable bti for "bti" in the string to
737 (aarch64_bti_enabled): Check if bti is enabled.
738 * config/aarch64/aarch64.opt: Declare target variable.
739 * doc/invoke.texi: Add bti to the -mbranch-protection documentation.
741 2018-01-09 Sudakshina Das <sudi.das@arm.com>
743 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use new
744 epilogue/prologue scratch registers EP0_REGNUM and EP1_REGNUM.
745 (aarch64_expand_epilogue): Likewise.
746 (aarch64_output_mi_thunk): Likewise
747 * config/aarch64/aarch64.h (REG_CLASS_CONTENTS): Change
748 TAILCALL_ADDR_REGS to x16 and x17.
749 * config/aarch64/aarch64.md: Define EP0_REGNUM and EP1_REGNUM.
751 2018-01-09 Sudakshina Das <sudi.das@arm.com>
753 * config/aarch64/aarch64-option-extensions.def: Define
754 AARCH64_OPT_EXTENSION for memtag, rng, sb, ssbs and predres.
755 * gcc/config/aarch64/aarch64.h (AARCH64_FL_RNG): New.
756 (AARCH64_FL_MEMTAG, ARCH64_FL_SB, AARCH64_FL_SSBS): New.
757 (AARCH64_FL_PREDRES): New.
758 (AARCH64_FL_FOR_ARCH8_5): Add AARCH64_FL_SB, AARCH64_FL_SSBS and
759 AARCH64_FL_PREDRES by default.
760 * gcc/doc/invoke.texi: Document rng, memtag, sb, ssbs and predres.
762 2018-01-09 Sudakshina Das <sudi.das@arm.com>
764 * config/aarch64/aarch64-arches.def: Define AARCH64_ARCH for
766 * gcc/config/aarch64/aarch64.h (AARCH64_FL_V8_5): New.
767 (AARCH64_FL_FOR_ARCH8_5, AARCH64_ISA_V8_5): New.
768 * gcc/doc/invoke.texi: Document ARMv8.5-A.
770 2019-01-09 Alejandro Martinez <alejandro.martinezvicente@arm.com>
772 * config/aarch64/aarch64-sve.md (copysign<mode>3): New define_expand.
773 (xorsign<mode>3): Likewise.
775 2019-01-09 Jelinek <jakub@redhat.com>
778 * tree.c (initializer_each_zero_or_onep) <case VECTOR_CST>: Use
779 vector_cst_elt instead of VECTOR_CST_ENCODED_ELT.
781 PR rtl-optimization/88331
782 * function.c (assign_stack_local_1): Don't set dynamic_align_addr if
783 not currently_expanding_to_rtl.
785 2019-01-09 Eric Botcazou <ebotcazou@adacore.com>
787 * doc/invoke.texi (-Os): Remove trailing spaces.
788 (-finline-functions): Remove reference to -O2.
790 2019-01-08 Jakub Jelinek <jakub@redhat.com>
792 PR rtl-optimization/79593
793 * config/i386/i386.md (reg = mem; mem = reg): New define_peephole2.
795 * config/rs6000/rs6000.c (rs6000_delegitimize_address): Delegitimize
796 UNSPEC_FUSION_GPR to its argument. Formatting fixes.
798 2019-01-08 Eric Botcazou <ebotcazou@adacore.com>
801 * config/sparc/sparc.c (function_arg_slotno): Set *PPREGNO & *PPADDING
805 * config/sparc/sparc.c (sparc_delegitimize_address): Deal with naked
806 UNSPECs and UNSPEC_MOVE_GOTDATA specifically.
808 2019-01-08 H.J. Lu <hongjiu.lu@intel.com>
811 * config/i386/i386.c (ix86_avx_u128_mode_exit): Call
812 ix86_avx_u128_mode_entry.
814 2019-01-08 Martin Liska <mliska@suse.cz>
816 PR tree-optimization/88753
817 * tree-switch-conversion.c (switch_conversion::build_one_array):
818 Come up with local variable constructor. Convert first to
819 type of constructor values.
821 2019-01-08 Richard Biener <rguenther@suse.de>
823 PR tree-optimization/86554
824 * tree-ssa-sccvn.c (eliminate_dom_walker, rpo_elim,
825 rpo_avail): Move earlier.
826 (visit_nary_op): When value-numbering to expressions
827 with different overflow behavior make sure there's an
828 available expression on the path.
830 2019-01-08 Sam Tebbs <sam.tebbs@arm.com>
832 * config/aarch64/aarch64.c (BRANCH_PROTECT_STR_MAX,
833 aarch64_parse_branch_protection,
834 struct aarch64_branch_protect_type,
835 aarch64_handle_no_branch_protection,
836 aarch64_handle_standard_branch_protection,
837 aarch64_validate_mbranch_protection,
838 aarch64_handle_pac_ret_protection,
839 aarch64_handle_attr_branch_protection,
840 accepted_branch_protection_string,
841 aarch64_pac_ret_subtypes,
842 aarch64_branch_protect_types,
843 aarch64_handle_pac_ret_leaf): Define.
844 (aarch64_override_options_after_change_1, aarch64_override_options):
845 Add check for accepted_branch_protection_string.
846 (aarch64_option_save): Save accepted_branch_protection_string.
847 (aarch64_option_restore): Save accepted_branch_protection_string.
848 * config/aarch64/aarch64.c (aarch64_attributes): Add branch-protection.
849 * config/aarch64/aarch64.opt: Add mbranch-protection. Deprecate
850 msign-return-address.
851 * doc/invoke.texi: Add mbranch-protection.
853 2019-01-08 Alan Modra <amodra@gmail.com>
856 * genattrtab.c (max_attr_value, min_attr_value, or_attr_value):
857 Delete "unknownp" parameter. Adjust callers. Handle
858 CONST_INT, PLUS, MINUS, and MULT.
859 (attr_value_aligned): Renamed from or_attr_value.
860 (min_attr_value): Return INT_MIN for unhandled rtl case..
861 (min_fn): ..and translate to INT_MAX here.
862 (write_length_unit_log): Modify to cope without "unknown".
863 (write_attr_value): Handle IF_THEN_ELSE.
865 2019-01-07 Richard Sandiford <richard.sandiford@arm.com>
867 * tree-vect-stmts.c (vectorizable_store): Don't use the dataref_offset
868 optimization for masked stores.
870 2019-01-07 Richard Sandiford <richard.sandiford@arm.com>
873 * tree-vect-loop.c (get_initial_defs_for_reduction): Pass the
874 output vector directly to duplicate_and_interleave instead of
875 going through a temporary. Postpone insertion of ctor_seq to
878 2019-01-07 Richard Earnshaw <rearnsha@arm.com>
881 * config/aarch64/aarch64.c (aarch64_expand_subvti): New parameter
882 unsigned_p. Handle signed and unsigned overflow correction as
884 * config/aarch64/aarch64-protos.h (aarch64_expand_subvti): Update
886 * config/aarch64/aarch64.md (addv<mode>4): Use aarch64_plus_operand
888 (add<mode>3_compareV_imm): Make this callable for expanding.
889 (subv<GPI:mode>4): Use register_operand for operand 1. Use
890 aarch64_plus_operand for operand 2.
891 (subv<GPI:mode>_insn): New insn pattern.
892 (subv<GPI:mode>_imm): Likewise.
893 (negv<GPI:mode>3): New expand pattern.
894 (negv<GPI:mode>_insn): New insn pattern.
895 (negv<GPI:mode>_cmp_only): Likewise.
896 (cmpv<GPI:mode>_insn): Likewise.
897 (subvti4): Use register_operand for operand 1. Update call to
898 aarch64_expand_subvti.
899 (usubvti4): Likewise.
900 (negvti3): New expand pattern.
901 (negdi_carryout): New insn pattern.
902 (negvdi_carryinV): New insn pattern.
903 (sub<mode3>_compare1_imm): Delete named insn pattern, make anonymous
904 version the named version.
905 (peepholes to convert to sub<mode3>_compare1_imm): Adjust order of
907 (usub<GPI:mode>3_carryinC, usub<GPI:mode>3_carryinC_z1): New insn
909 (usub<GPI:mode>3_carryinC_z2, usub<GPI:mode>3_carryinC): New insn
911 (sub<mode>3_carryinCV, sub<mode>3_carryinCV_z1_z2): Delete.
912 (sub<mode>3_carryinCV_z1, sub<mode>3_carryinCV_z2): Delete.
913 (sub<mode>3_carryinCV): Delete.
914 (sub<GPI:mode>3_carryinV): New expand pattern.
915 sub<mode>3_carryinV, sub<mode>3_carryinV_z2): New insn patterns.
917 2019-01-07 Richard Biener <rguenther@suse.de>
919 * tree-ssa-uncprop.c (ssa_equip_hash_traits): Remove in favor
920 of tree_operand_hash.
922 2019-01-07 Richard Sandiford <richard.sandiford@arm.com>
924 PR tree-optimization/88598
925 * tree.h (single_nonzero_element): Declare.
926 * tree.c (single_nonzero_element): New function.
927 * match.pd: Fold certain reductions of X & CST to X[I] & CST[I]
928 if I is the only nonzero element of CST.
930 2019-01-07 Richard Sandiford <richard.sandiford@arm.com>
932 PR tree-optimization/88598
933 * tree.h (initializer_each_zero_or_onep): Declare.
934 * tree.c (initializer_each_zero_or_onep): New function.
935 (signed_or_unsigned_type_for): Handle float types too.
936 (unsigned_type_for, signed_type_for): Update comments accordingly.
937 * match.pd: Fold x * { 0 or 1, 0 or 1, ...} to
938 x & { 0 or -1, 0 or -1, ... }.
940 2019-01-07 Jonathan Wakely <jwakely@redhat.com>
942 * doc/install.texi: Replace references to x86_64-unknown-linux-gnu
943 with x86_64-pc-linux-gnu.
945 2019-01-07 Tom de Vries <tdevries@suse.de>
948 * config/nvptx/nvptx.c (has_vector_partitionable_routine_calls_p): New
950 (nvptx_goacc_validate_dims): Force vl32 if calling vector-partitionable
953 2019-01-07 Jakub Jelinek <jakub@redhat.com>
955 * config/i386/sse.md (vec_extract<mode><ssehalfvecmodelower>): Use
956 V_256_512 iterator instead of V_512 and TARGET_AVX instead of
957 TARGET_AVX512F as condition.
960 * dwarf2out.c (const_ok_for_output_1): Remove redundant call to
961 const_not_ok_for_debug_p target hook.
962 (mem_loc_descriptor) <case UNSPEC>: Only call const_ok_for_output_1
963 on UNSPEC and subexpressions thereof if all subexpressions of the
964 UNSPEC are CONSTANT_P.
966 PR tree-optimization/88676
967 * tree-ssa-phiopt.c (two_value_replacement): New function.
968 (tree_ssa_phiopt_worker): Call it.
971 * cfgexpand.c (expand_stack_vars): Only align prev_offset to
972 ASAN_MIN_RED_ZONE_SIZE, not to maximum of that and alignb.
975 * tree-vect-generic.c: Include insn-config.h and recog.h.
976 (expand_vector_piecewise): Add defaulted ret_type argument,
977 if non-NULL, use that in preference to type for the result type.
978 (expand_vector_parallel): Formatting fix.
979 (do_vec_conversion, do_vec_narrowing_conversion,
980 expand_vector_conversion): New functions.
981 (expand_vector_operations_1): Call expand_vector_conversion
982 for VEC_CONVERT ifn calls.
983 * internal-fn.def (VEC_CONVERT): New internal function.
984 * internal-fn.c (expand_VEC_CONVERT): New function.
985 * fold-const-call.c (fold_const_vec_convert): New function.
986 (fold_const_call): Use it for CFN_VEC_CONVERT.
987 * doc/extend.texi (__builtin_convertvector): Document.
989 2019-01-07 Tom de Vries <tdevries@suse.de>
991 * config/nvptx/nvptx-protos.h (nvptx_output_red_partition): Declare.
992 * config/nvptx/nvptx.c (vector_red_size, vector_red_align,
993 vector_red_partition, vector_red_sym): New global variables.
994 (nvptx_option_override): Initialize vector_red_sym.
995 (nvptx_declare_function_name): Restore red_partition register.
996 (nvptx_file_end): Emit code to declare the vector reduction variables.
997 (nvptx_output_red_partition): New function.
998 (nvptx_expand_shared_addr): Add vector argument. Use it to handle
999 large vector reductions.
1000 (enum nvptx_builtins): Add NVPTX_BUILTIN_VECTOR_ADDR.
1001 (nvptx_init_builtins): Add VECTOR_ADDR.
1002 (nvptx_expand_builtin): Update call to nvptx_expand_shared_addr.
1003 Handle nvptx_expand_shared_addr.
1004 (nvptx_get_shared_red_addr): Add vector argument and handle large
1006 (nvptx_goacc_reduction_setup): Add offload_attrs argument and handle
1008 (nvptx_goacc_reduction_init): Likewise.
1009 (nvptx_goacc_reduction_fini): Likewise.
1010 (nvptx_goacc_reduction_teardown): Likewise.
1011 (nvptx_goacc_reduction): Update calls to nvptx_goacc_reduction_{setup,
1012 init,fini,teardown}.
1013 (nvptx_init_axis_predicate): Initialize vector_red_partition.
1014 (nvptx_set_current_function): Init vector_red_partition.
1015 * config/nvptx/nvptx.md (UNSPECV_RED_PART): New unspecv.
1016 (nvptx_red_partition): New insn.
1017 * config/nvptx/nvptx.h (struct machine_function): Add red_partition.
1019 2019-01-07 Tom de Vries <tdevries@suse.de>
1022 * config/nvptx/nvptx.c (nvptx_process_pars): Don't emit barriers for
1025 2019-01-07 Tom de Vries <tdevries@suse.de>
1027 * config/nvptx/nvptx.c (oacc_bcast_partition): Declare.
1028 (nvptx_option_override): Init oacc_bcast_partition.
1029 (nvptx_init_oacc_workers): New function.
1030 (nvptx_declare_function_name): Call nvptx_init_oacc_workers.
1031 (nvptx_needs_shared_bcast): New function.
1032 (nvptx_find_par): Generalize to enable vectors to use shared-memory
1034 (nvptx_shared_propagate): Initialize vector bcast partition and
1035 synchronization state.
1036 (nvptx_single): Generalize to enable vectors to use shared-memory
1038 (nvptx_process_pars): Likewise.
1039 (nvptx_set_current_function): Initialize oacc_broadcast_partition.
1040 * config/nvptx/nvptx.h (struct machine_function): Add
1041 bcast_partition and sync_bar members.
1043 2019-01-07 Tom de Vries <tdevries@suse.de>
1045 * config/nvptx/nvptx.c (nvptx_welformed_vector_length_p)
1046 (nvptx_apply_dim_limits): New function.
1047 (nvptx_goacc_validate_dims_1): Allow PTX_MAX_VECTOR_LENGTH larger than
1050 2019-01-07 Tom de Vries <tdevries@suse.de>
1052 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims_1): Move warnings to
1053 as late as possible.
1055 2019-01-07 Tom de Vries <tdevries@suse.de>
1057 * config/nvptx/nvptx.c (PTX_VECTOR_LENGTH): Remove.
1058 (PTX_DEFAULT_VECTOR_LENGTH, PTX_MAX_VECTOR_LENGTH): Define.
1059 (nvptx_goacc_validate_dims_1, nvptx_dim_limit)
1060 (nvptx_goacc_reduction_fini): Use PTX_DEFAULT_VECTOR_LENGTH,
1061 PTX_MAX_VECTOR_LENGTH and PTX_WARP_SIZE instead of PTX_VECTOR_LENGTH.
1063 2019-01-07 Tom de Vries <tdevries@suse.de>
1065 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Add asserts.
1067 2019-01-07 Tom de Vries <tdevries@suse.de>
1069 * omp-offload.c (oacc_get_min_dim): New function.
1070 * omp-offload.h (oacc_get_min_dim): Declare.
1072 2018-12-26 Mateusz B <mateuszb@poczta.onet.pl>
1075 * config/i386/i386.c (function_value_ms_64): Return small sturct in
1076 AX_REG and float/double in FIRST_SSE_REG for 4 or 8 byte modes.
1078 2019-01-05 Jan Hubicka <hubicka@ucw.cz>
1082 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
1084 * ipa-inline.c (edge_badness): Use inlined_time instead of
1085 inline_summaries->get.
1087 2019-01-05 Jan Hubicka <hubicka@ucw.cz>
1089 * opts.c (enable_fdo_optimizations): Enable
1090 version-loops-for-strides, loop-interchange, unrol-and-jam
1091 and tree-loop-distribution.
1092 * invoke.texi: Document newly enabled options.
1094 2019-01-05 Jan Hubicka <hubicka@ucw.cz>
1096 * doc/invoke.texi (max-inline-insns-small): New parameters.
1097 * ipa-inline.c (want_early_inline_function_p): simplify.
1098 (want_inline_small_function_p): Fix pasto from previous patch;
1099 use max-inline-insns-small bound.
1100 * params.def (max-inline-insns-small): New param.
1101 * ipa-fnsummary.c (analyze_function_body): Initialize time/size
1102 variables correctly.
1104 2019-01-05 Jan Hubicka <hubicka@ucw.cz>
1106 * doc/invoke.texi: Document max-inline-insns-size,
1107 uninlined-function-insns, uninlined-function-time,
1108 uninlined-thunk-insns and uninlined-thunk-time.
1109 * params.def: Add max-inline-insns-size,
1110 uninlined-function-insns, uninlined-function-time,
1111 uninlined-thunk-insns and uninlined-thunk-time.
1112 * ipa-fnsummary.c (compute_fn_summary, analyze_function_body): Use
1114 * ipa-inline.c (can_inline_edge_by_limits_p,
1115 want_inline_small_function_p): Use new parameters.
1117 2019-01-05 Jan Hubicka <hubicka@ucw.cz>
1119 * ipa-fnsummary.c (analyze_function_body): Fix accounting of time.
1121 2019-01-05 Jakub Jelinek <jakub@redhat.com>
1125 * expr.c (expand_assignment): For calls returning VLA structures
1126 if to_rtx is not a MEM, force it into a stack temporary.
1129 * dwarf2out.c (const_ok_for_output_1): Reject MINUS that contains
1130 SYMBOL_REF, CODE_LABEL or UNSPEC in subexpressions of second argument.
1131 Reject PLUS that contains SYMBOL_REF, CODE_LABEL or UNSPEC in
1132 subexpressions of both operands.
1133 (mem_loc_descriptor): Handle UNSPEC if target hook acks it and all the
1134 subrtxes are CONSTANT_P.
1135 * config/i386/i386.c (ix86_const_not_ok_for_debug_p): Revert
1138 2019-01-04 Jan Hubicka <hubicka@ucw.cz>
1140 * params.def (hot-bb-count-ws-permille): Set to 990.
1142 2019-01-04 Martin Sebor <msebor@redhat.com>
1145 * attribs.c (decls_mismatched_attributes): Avoid warning for attribute
1148 2019-01-04 Martin Sebor <msebor@redhat.com>
1151 * doc/extend.texi (attribute alloc_align, alloc_size): Update.
1153 2019-01-04 Jakub Jelinek <jakub@redhat.com>
1155 * gdbinit.in: Turn off pagination for the skip commands, restore
1156 it to previous state afterwards.
1158 2019-01-04 Sam Tebbs <sam.tebbs@arm.com>
1161 * gcc.target/aarch64/combine_bfxil.c: Change scan-assembler-times bfxil
1164 2019-01-04 Jakub Jelinek <jakub@redhat.com>
1167 * config/i386/i386.c (ix86_expand_divmod_libfunc): Use mode instead
1168 of GET_MODE (opN) as modes of the libcall arguments.
1170 2019-01-04 Jan Beulich <jbeulich@suse.com>
1172 * config/i386/sse.md
1173 (<avx512>_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>,
1174 <avx512>_cmp<mode>3<mask_scalar_merge_name>,
1175 <avx512>_ucmp<mode>3<mask_scalar_merge_name>,
1176 <avx512>_ucmp<mode>3<mask_scalar_merge_name>,
1177 avx512f_vmcmp<mode>3<round_saeonly_name>,
1178 avx512f_vmcmp<mode>3_mask<round_saeonly_name>,
1179 avx512f_maskcmp<mode>3,
1180 <avx512>_cvt<ssemodesuffix>2mask<mode>,
1181 <avx512>_cvt<ssemodesuffix>2mask<mode>,
1182 *<avx512>_cvtmask2<ssemodesuffix><mode>,
1183 *<avx512>_cvtmask2<ssemodesuffix><mode>,
1184 <avx512>_eq<mode>3<mask_scalar_merge_name>_1,
1185 <avx512>_eq<mode>3<mask_scalar_merge_name>_1,
1186 <avx512>_gt<mode>3<mask_scalar_merge_name>,
1187 <avx512>_gt<mode>3<mask_scalar_merge_name>,
1188 <avx512>_testm<mode>3<mask_scalar_merge_name>,
1189 <avx512>_testnm<mode>3<mask_scalar_merge_name>,
1190 *<avx512>_testm<VI1248_AVX512VLBW:mode>3_zext,
1191 *<avx512>_testm<VI1248_AVX512VLBW:mode>3_zext_mask,
1192 *<avx512>_testnm<VI1248_AVX512VLBW:mode>3_zext,
1193 *<avx512>_testnm<VI1248_AVX512VLBW:mode>3_zext_mask,
1194 avx512cd_maskb_vec_dup<mode>,
1195 avx512cd_maskw_vec_dup<mode>,
1196 avx512dq_fpclass<mode><mask_scalar_merge_name>,
1197 avx512dq_vmfpclass<mode>,
1198 avx512vl_vpshufbitqmb<mode><mask_scalar_merge_name>): Use =k
1201 2019-01-03 Martin Sebor <msebor@redhat.com>
1203 PR tree-optimization/88659
1204 * calls.c (maybe_warn_nonstring_arg): Avoid assuming maxlen is set.
1206 2019-01-03 Aaron Sawdey <acsawdey@linux.ibm.com>
1208 * config/rs6000/rs6000-string.c (expand_block_move): Don't use
1209 unaligned vsx and avoid lxvd2x/stxvd2x.
1210 (gen_lvx_v4si_move): New function.
1212 2019-01-03 Tom de Vries <tdevries@suse.de>
1214 * config/nvptx/nvptx.c (MACH_VECTOR_LENGTH, MACH_MAX_WORKERS): Define.
1215 (init_axis_dim, nvptx_mach_max_workers, nvptx_mach_vector_length): New
1217 * config/nvptx/nvptx.h (struct machine_function): Add axis_dims.
1219 2019-01-03 Tom de Vries <tdevries@suse.de>
1221 * config/nvptx/nvptx.c (struct offload_attrs): New.
1222 (populate_offload_attrs): New function. Factor mask extraction out of
1223 nvptx_reorg. Add extraction of dimensions.
1224 (nvptx_reorg): Use populate_offload_attrs.
1226 2019-01-03 Tom de Vries <tdevries@suse.de>
1228 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims_1): Add early-out
1229 cases for oacc_min_dims_p and routine_p. Add asserts for
1230 oacc_default_dims_p and offload_region_p.
1232 2019-01-03 Tom de Vries <tdevries@suse.de>
1234 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims_1): New function,
1236 (nvptx_goacc_validate_dims): ... here.
1238 2019-01-03 Jan Hubicka <hubicka@ucw.cz>
1240 PR tree-optimization/85574
1241 * tree-ssa-uncprop.c (struct equiv_hash_elt): Remove unused
1243 (struct ssa_equip_hash_traits): Declare.
1244 (val_ssa_equiv): Use custom hash traits using operand_equal_p.
1246 2019-01-03 Jakub Jelinek <jakub@redhat.com>
1249 * dwarf2out.c (modified_type_die): If type is equal to sizetype,
1250 change it to qualified_type.
1252 2019-01-03 Jan Hubicka <hubicka@ucw.cz>
1254 * ipa-utils.c (scale_ipa_profile_for_fn): Break out from ...
1255 (ipa_merge_profiles): ... here; do not ICE on thunks and aliases.
1257 2019-01-02 Martin Sebor <msebor@redhat.com>
1258 Jeff Law <law@redhat.com>
1260 * gimple-fold.c (strlen_range_kind): Remove SRK_LENRANGE_2.
1261 (get_range_strlen_tree): Update appropriately.
1263 * gimple-fold.h (get_range_strlen): Drop unused last argument.
1265 * gimple-fold.c (gimple_fold_builtin_strlen): Use set_strlen_range
1266 rather than set_range_info.
1267 * tree-ssa-strlen.c (set_strlen_range): Extracted from
1268 maybe_set_strlen_range. Handle potentially boundary crossing
1269 cases more conservatively.
1270 (maybe_set_strlen_range): Parts refactored into set_strlen_range.
1271 Call set_strlen_range.
1272 * tree-ssa-strlen.h (set_strlen_range): Add prototype.
1275 * gimple-fold.c (get_range_strlen): Update prototype to no longer
1276 need the flexp argument.
1277 (get_range_strlen_tree): Drop flexp argument. Drop flexp argument
1278 from calls to get_range_strlen. Update comments. Just update
1279 VAL for an unterminated const char array and let the reset of the
1280 code handle it normally. No longer try to set *flexp. Adjust
1282 (get_range_strlen): Update for the new get_range_strlen API.
1283 (get_maxval_strlen): Similarly.
1284 (gimple_fold_builtin_strlen): Handle update meaning of return value
1285 from get_range_strlen.
1286 * gimple-ssa-sprintf.c (get_string_length): Update for the new
1287 get_range_strlen API.
1289 2019-01-02 Jan Hubicka <hubicka@ucw.cz>
1292 * varpool.c (varpool_node::ctor_useable_for_folding_p): Also return
1293 false at WPA time when body was removed.
1295 2019-01-02 Martin Liska <mliska@suse.cz>
1297 PR tree-optimization/88650
1298 * predict.c (set_even_probabilities): Calculate probability
1299 remainer only when really used.
1301 2019-01-02 Richard Biener <rguenther@suse.de>
1304 * tree-data-ref.c (analyze_subscript_affine_affine): Use
1305 widest_ints when mangling max_stmt_execution results.
1307 2019-01-02 Richard Biener <rguenther@suse.de>
1309 PR tree-optimization/88621
1310 * tree-ssa-loop-im.c (gather_mem_refs_stmt): Fix pastos, avoid
1311 bitfields when canoncalizing.
1313 2019-01-02 Richard Biener <rguenther@suse.de>
1316 * config/i386/x86-tune-costs.h (intel_cost): Adjust
1317 cost of cheap SSE instruction.
1319 2019-01-02 Richard Biener <rguenther@suse.de>
1322 * ipa-icf.h (sem_item_optimizer::sort_congruence_split): Declare.
1323 * ipa-icf.c (sem_item_optimizer::sort_congruence_split): New
1325 (sem_item_optimizer::do_congruence_step_f): Sort the congruence
1326 set after UIDs before splitting them.
1328 2019-01-01 Martin Sebor <msebor@redhat.com>
1329 Jeff Law <law@redhat.com>
1331 * gimple-fold.c (get_range_strlen_tree): Record if the computed
1332 length is optimistic. If it is, then arrange to compute the
1333 conservative length as well.
1335 * gimple-fold.h (get_range_strlen): Update prototype.
1336 * builtins.c (check_access): Update call to get_range_strlen to use
1337 c_strlen_data pointer. Change various variable accesses to instead
1338 pull data from the c_strlen_data structure.
1339 (check_strncat_sizes, expand_builtin_strncat): Likewise.
1340 * calls.c (maybe_warn_nonstring_arg): Likewise.
1341 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Likewise. Reset
1342 minimum length if maximum lengh is unknown.
1343 * gimple-ssa-sprintf.c (get_string_length): Likewise. Drop code
1344 that used c_strlen, it's no longer needed. Restructure slightly.
1345 (format_string): Set unlikely range appropriately.
1346 * gimple-fold.c (get_range_strlen): Update comments. Fix minor
1348 (get_range_strlen): Accept c_strlen_data pointer for external
1349 call sites as well. Pass through to call to internal get_range_strlen.
1350 Adjust minlen, maxlen and maxbound as needed.
1351 (get_maxval_strlen): Update comments.
1352 (gimple_fold_builtin_strlen): Update call to get_range_strlen
1353 to use c_strlen_data pointer. Change variable accesses to instead
1354 use c_strlen_data data members.
1356 * gimple-fold.c (get_range_strlen): Update prototype.
1357 (get_range_strlen_tree): Update prototype. Drop minlen/maxlen
1358 local variables. Use pdata to return information to caller.
1359 Update calls to get_range_strlen. Update pdata->maxbound.
1360 (get_range_strlen -- static version): Similarly.
1361 (get_range_strlen -- extern version): Update for internal
1362 get_range_strlen API change. Convert to external data format.
1363 (get_maxval_strlen): Similarly.
1365 2019-01-01 Jan Hubicka <hubicka@ucw.cz>
1367 * coverage.c (get_coverage_counts): Use current_function_decl.
1368 * profile.c (read_thunk_profile): New function.
1369 (branch_prob): Add THUNK parameter.
1370 * tree-profile.c (tree_profiling): Handle thunks.
1371 * value-prof.c (init_node_map): Handle thunks.
1372 * value-prof.h (branch_prob): Upate prototype.
1373 (read_thunk_profile): Declare.
1375 2019-01-01 Jakub Jelinek <jakub@redhat.com>
1377 Update copyright years.
1379 * gcc.c (process_command): Update copyright notice dates.
1380 * gcov-dump.c (print_version): Ditto.
1381 * gcov.c (print_version): Ditto.
1382 * gcov-tool.c (print_version): Ditto.
1383 * gengtype.c (create_file): Ditto.
1384 * doc/cpp.texi: Bump @copying's copyright year.
1385 * doc/cppinternals.texi: Ditto.
1386 * doc/gcc.texi: Ditto.
1387 * doc/gccint.texi: Ditto.
1388 * doc/gcov.texi: Ditto.
1389 * doc/install.texi: Ditto.
1390 * doc/invoke.texi: Ditto.
1392 Copyright (C) 2019 Free Software Foundation, Inc.
1394 Copying and distribution of this file, with or without modification,
1395 are permitted in any medium without royalty provided the copyright
1396 notice and this notice are preserved.