1 2019-01-12 Jakub Jelinek <jakub@redhat.com>
3 * tree-ssa-loop-ivopts.c (find_inv_vars): Fix a comment typo.
5 2019-01-12 Tom de Vries <tdevries@suse.de>
7 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims_1): In offloading
8 region calling vector-partitionable routine, set default_vector_length
11 2019-01-12 Tom de Vries <tdevries@suse.de>
13 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims_1): Add an use new
14 variable default_vector_length.
16 2019-01-12 Tom de Vries <tdevries@suse.de>
19 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims_1): Apply defaults
20 from oacc_default_dims, as oacc_validate_dims would do it, and apply
23 2019-01-12 Tom de Vries <tdevries@suse.de>
25 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims_1)
26 (nvptx_goacc_validate_dims): Add used parameter.
27 * doc/tm.texi: Regenerate.
28 * omp-offload.c (oacc_parse_default_dims, oacc_validate_dims): Add
29 argument to call to targetm.goacc.validate_dims.
30 (default_goacc_validate_dims): Add used
32 * target.def (validate_dims): Add used parameter in DEFHOOK.
33 * targhooks.h (default_goacc_validate_dims): Add used parameter.
35 2019-01-11 Jakub Jelinek <jakub@redhat.com>
39 * tree-inline.h (struct copy_body_data): Add adjust_array_error_bounds
41 * tree-inline.c (remap_type_1): Formatting fix. If TYPE_MAX_VALUE of
42 ARRAY_TYPE's TYPE_DOMAIN is newly error_mark_node, replace it with
43 a dummy "omp dummy var" variable if id->adjust_array_error_bounds.
44 * omp-low.c (new_omp_context): Set cb.adjust_array_error_bounds.
46 2019-01-11 Vladimir Makarov <vmakarov@redhat.com>
48 PR rtl-optimization/87305
50 (setup_live_pseudos_and_spill_after_risky_transforms): Add code
51 for little endian pseudos used as paradoxical subreg.
53 2019-01-11 Jakub Jelinek <jakub@redhat.com>
55 PR tree-optimization/88693
56 * tree-ssa-strlen.c (get_min_string_length): Don't set *full_string_p
57 for STRING_CSTs that don't contain any NUL characters in the first
58 TREE_STRING_LENGTH bytes.
60 2019-01-11 Alan Modra <amodra@gmail.com>
64 * genattrtab.c (min_fn): Don't translate values.
65 (min_attr_value): Return INT_MAX when the value can't be calculated.
66 Return minimum among any values that can be calculated.
67 (max_attr_value): Adjust.
69 2019-01-11 Jakub Jelinek <jakub@redhat.com>
71 * Makefile.in (PLUGIN_HEADERS): Add $(INSN_ATTR_H).
73 2019-01-11 Steve Ellcey <sellcey@marvell.com>
75 * config/aarch64/aarch64.c (aarch64_simd_call_p): New function.
76 (aarch64_hard_regno_call_part_clobbered): Add insn argument.
77 (aarch64_return_call_with_max_clobbers): New function.
78 (TARGET_RETURN_CALL_WITH_MAX_CLOBBERS): New macro.
79 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Add insn
81 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): Ditto.
82 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Ditto.
83 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered): Ditto.
84 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): Ditto.
85 * cselib.c (cselib_process_insn): Add argument to
86 targetm.hard_regno_call_part_clobbered call.
87 * ira-conflicts.c (ira_build_conflicts): Ditto.
88 * ira-costs.c (ira_tune_allocno_costs): Ditto.
89 * lra-constraints.c (inherit_reload_reg): Ditto.
90 * lra-int.h (struct lra_reg): Add call_insn field, remove call_p field.
91 * lra-lives.c (check_pseudos_live_through_calls): Add call_insn
92 argument. Call targetm.return_call_with_max_clobbers.
93 Add argument to targetm.hard_regno_call_part_clobbered call.
94 (calls_have_same_clobbers_p): New function.
95 (process_bb_lives): Add call_insn and last_call_insn variables.
96 Pass call_insn to check_pseudos_live_through_calls.
97 Modify if stmt to check targetm.return_call_with_max_clobbers.
98 Update setting of flush variable.
99 (lra_create_live_ranges_1): Set call_insn to NULL instead of call_p
101 * lra.c (initialize_lra_reg_info_element): Set call_insn to NULL.
102 * regcprop.c (copyprop_hardreg_forward_1): Add argument to
103 targetm.hard_regno_call_part_clobbered call.
104 * reginfo.c (choose_hard_reg_mode): Ditto.
105 * regrename.c (check_new_reg_p): Ditto.
106 * reload.c (find_equiv_reg): Ditto.
107 * reload1.c (emit_reload_insns): Ditto.
108 * sched-deps.c (deps_analyze_insn): Ditto.
109 * sel-sched.c (init_regs_for_mode): Ditto.
110 (mark_unavailable_hard_regs): Ditto.
111 * targhooks.c (default_dwarf_frame_reg_mode): Ditto.
112 * target.def (hard_regno_call_part_clobbered): Add insn argument.
113 (return_call_with_max_clobbers): New target function.
114 * doc/tm.texi: Regenerate.
115 * doc/tm.texi.in (TARGET_RETURN_CALL_WITH_MAX_CLOBBERS): New hook.
116 * hooks.c (hook_bool_uint_mode_false): Change to
117 hook_bool_insn_uint_mode_false.
118 * hooks.h (hook_bool_uint_mode_false): Ditto.
120 2019-01-11 Steve Ellcey <sellcey@marvell.com>
122 * config/aarch64/aarch64.c (aarch64_simd_call_p): New function.
123 (aarch64_remove_extra_call_preserved_regs): New function.
124 (TARGET_REMOVE_EXTRA_CALL_PRESERVED_REGS): New macro.
125 * doc/tm.texi.in (TARGET_REMOVE_EXTRA_CALL_PRESERVED_REGS): New hook.
126 * doc/tm.texi: Regenerate.
127 * final.c (get_call_reg_set_usage): Call new hook.
128 * target.def (remove_extra_call_preserved_regs): New hook.
129 * targhooks.c (default_remove_extra_call_preserved_regs): New function.
130 * targhooks.h (default_remove_extra_call_preserved_regs): New function.
132 2019-01-11 Jakub Jelinek <jakub@redhat.com>
135 * passes.c (finish_optimization_passes): Call print_combine_total_stats
136 inside of pass_combine_1 dump rather than pass_profile_1.
138 2019-01-11 Tom de Vries <tdevries@suse.de>
140 * config/nvptx/nvptx.c (PTX_CTA_NUM_BARRIERS, PTX_PER_CTA_BARRIER)
141 (PTX_NUM_PER_CTA_BARRIER, PTX_FIRST_PER_WORKER_BARRIER)
142 (PTX_NUM_PER_WORKER_BARRIERS): Define.
143 (nvptx_apply_dim_limits): Prevent vector_length 64 and
146 2019-01-11 Tom de Vries <tdevries@suse.de>
148 * config/nvptx/nvptx.c (PTX_CTA_SIZE): Move up.
150 2019-01-11 Jan Beulich <jbeulich@suse.com>
152 * config/i386/i386.md (rex64suffix): Add L suffix for SI.
153 * config/i386/sse.md (cvtusi2<ssescalarmodesuffix>32<round_name>,
154 sse2_cvtsi2sd): Add {l}.
155 (sse2_cvtsi2sdq<round_name>): Make q conditional upon AT&T
158 2019-01-10 Jakub Jelinek <jakub@redhat.com>
161 * config/i386/sse.md (float<floatunssuffix>v2div2sf2): Turn into
163 (*float<floatunssuffix>v2div2sf2): New define_insn.
164 (float<floatunssuffix>v2div2sf2_mask): Turn into define_expand.
165 (*float<floatunssuffix>v2div2sf2_mask): New define_insn.
166 (*float<floatunssuffix>v2div2sf2_mask_1): Replace
167 subrtxes (const_vector:V2SF [(const_int 0) (const_int 0)]) with
168 match_operands with "const0_operand" "C".
170 2019-01-10 Tamar Christina <tamar.christina@arm.com>
172 * config/aarch64/aarch64-builtins.c
173 (aarch64_init_builtins): Move aarch64_init_fcmla_laneq_builtins...
174 (aarch64_init_simd_builtins): ...Here
176 2019-01-10 Vladimir Makarov <vmakarov@redhat.com>
178 PR rtl-optimization/87305
180 (setup_live_pseudos_and_spill_after_risky_transforms): Check
181 allocation for big endian pseudos used as paradoxical subregs and
182 spill them if it is wrong.
183 * lra-constraints.c (lra_constraints): Add a comment.
185 2019-01-10 Richard Biener <rguenther@suse.de>
187 PR tree-optimization/88792
188 * tree-ssa-pre.c (get_representative_for): Do not return a
191 2019-01-10 Jakub Jelinek <jakub@redhat.com>
195 * function.c (assign_stack_local_1): Revert the 2018-11-21 changes.
196 (assign_parm_setup_block): Do the argument slot realignment here
199 2019-01-10 Stefan Agner <stefan@agner.ch>
202 * config/arm/arm.c (arm_option_override_internal): Force
203 opts->x_inline_asm_unified to true only if TARGET_THUMB2_P.
205 2019-01-10 Jakub Jelinek <jakub@redhat.com>
208 * attribs.c (handle_dll_attribute): Clear TREE_STATIC after setting
211 2019-01-10 Tamar Christina <tamar.christina@arm.com>
213 * config/arm/arm-builtins.c
214 (enum arm_type_qualifiers): Add qualifier_lane_pair_index.
215 (MAC_LANE_PAIR_QUALIFIERS): New.
216 (arm_expand_builtin_args): Use it.
217 (arm_expand_builtin_1): Likewise.
218 * config/arm/arm-protos.h (neon_vcmla_lane_prepare_operands): New.
219 * config/arm/arm.c (neon_vcmla_lane_prepare_operands): New.
220 * config/arm/arm-c.c (arm_cpu_builtins): Add __ARM_FEATURE_COMPLEX.
221 * config/arm/arm_neon.h:
222 (vcadd_rot90_f16): New.
223 (vcaddq_rot90_f16): New.
224 (vcadd_rot270_f16): New.
225 (vcaddq_rot270_f16): New.
228 (vcmla_lane_f16): New.
229 (vcmla_laneq_f16): New.
230 (vcmlaq_lane_f16): New.
231 (vcmlaq_laneq_f16): New.
232 (vcmla_rot90_f16): New.
233 (vcmlaq_rot90_f16): New.
234 (vcmla_rot90_lane_f16): New.
235 (vcmla_rot90_laneq_f16): New.
236 (vcmlaq_rot90_lane_f16): New.
237 (vcmlaq_rot90_laneq_f16): New.
238 (vcmla_rot180_f16): New.
239 (vcmlaq_rot180_f16): New.
240 (vcmla_rot180_lane_f16): New.
241 (vcmla_rot180_laneq_f16): New.
242 (vcmlaq_rot180_lane_f16): New.
243 (vcmlaq_rot180_laneq_f16): New.
244 (vcmla_rot270_f16): New.
245 (vcmlaq_rot270_f16): New.
246 (vcmla_rot270_lane_f16): New.
247 (vcmla_rot270_laneq_f16): New.
248 (vcmlaq_rot270_lane_f16): New.
249 (vcmlaq_rot270_laneq_f16): New.
250 (vcadd_rot90_f32): New.
251 (vcaddq_rot90_f32): New.
252 (vcadd_rot270_f32): New.
253 (vcaddq_rot270_f32): New.
256 (vcmla_lane_f32): New.
257 (vcmla_laneq_f32): New.
258 (vcmlaq_lane_f32): New.
259 (vcmlaq_laneq_f32): New.
260 (vcmla_rot90_f32): New.
261 (vcmlaq_rot90_f32): New.
262 (vcmla_rot90_lane_f32): New.
263 (vcmla_rot90_laneq_f32): New.
264 (vcmlaq_rot90_lane_f32): New.
265 (vcmlaq_rot90_laneq_f32): New.
266 (vcmla_rot180_f32): New.
267 (vcmlaq_rot180_f32): New.
268 (vcmla_rot180_lane_f32): New.
269 (vcmla_rot180_laneq_f32): New.
270 (vcmlaq_rot180_lane_f32): New.
271 (vcmlaq_rot180_laneq_f32): New.
272 (vcmla_rot270_f32): New.
273 (vcmlaq_rot270_f32): New.
274 (vcmla_rot270_lane_f32): New.
275 (vcmla_rot270_laneq_f32): New.
276 (vcmlaq_rot270_lane_f32): New.
277 (vcmlaq_rot270_laneq_f32): New.
278 * config/arm/arm_neon_builtins.def (vcadd90, vcadd270, vcmla0, vcmla90,
279 vcmla180, vcmla270, vcmla_lane0, vcmla_lane90, vcmla_lane180, vcmla_lane270,
280 vcmla_laneq0, vcmla_laneq90, vcmla_laneq180, vcmla_laneq270,
281 vcmlaq_lane0, vcmlaq_lane90, vcmlaq_lane180, vcmlaq_lane270): New.
282 * config/arm/neon.md (neon_vcmla_lane<rot><mode>,
283 neon_vcmla_laneq<rot><mode>, neon_vcmlaq_lane<rot><mode>): New.
284 * config/arm/arm.c (arm_arch8_3, arm_arch8_4): New.
285 * config/arm/arm.h (TARGET_COMPLEX, arm_arch8_3, arm_arch8_4): New.
286 (arm_option_reconfigure_globals): Use them.
287 * config/arm/iterators.md (VDF, VQ_HSF): New.
289 (VF_constraint, rot, rotsplit1, rotsplit2): Add V4HF and V8HF.
290 * config/arm/neon.md (neon_vcadd<rot><mode>, neon_vcmla<rot><mode>): New.
291 * config/arm/unspecs.md (UNSPEC_VCADD90, UNSPEC_VCADD270,
292 UNSPEC_VCMLA, UNSPEC_VCMLA90, UNSPEC_VCMLA180, UNSPEC_VCMLA270): New.
294 2019-01-10 Tamar Christina <tamar.christina@arm.com>
296 * config/aarch64/aarch64-builtins.c (enum aarch64_type_qualifiers): Add qualifier_lane_pair_index.
297 (emit-rtl.h): Include.
298 (TYPES_QUADOP_LANE_PAIR): New.
299 (aarch64_simd_expand_args): Use it.
300 (aarch64_simd_expand_builtin): Likewise.
301 (AARCH64_SIMD_FCMLA_LANEQ_BUILTINS, aarch64_fcmla_laneq_builtin_datum): New.
302 (FCMLA_LANEQ_BUILTIN, AARCH64_SIMD_FCMLA_LANEQ_BUILTIN_BASE,
303 AARCH64_SIMD_FCMLA_LANEQ_BUILTINS, aarch64_fcmla_lane_builtin_data,
304 aarch64_init_fcmla_laneq_builtins, aarch64_expand_fcmla_builtin): New.
305 (aarch64_init_builtins): Add aarch64_init_fcmla_laneq_builtins.
306 (aarch64_expand_buildin): Add AARCH64_SIMD_BUILTIN_FCMLA_LANEQ0_V2SF,
307 AARCH64_SIMD_BUILTIN_FCMLA_LANEQ90_V2SF, AARCH64_SIMD_BUILTIN_FCMLA_LANEQ180_V2SF,
308 AARCH64_SIMD_BUILTIN_FCMLA_LANEQ2700_V2SF, AARCH64_SIMD_BUILTIN_FCMLA_LANEQ0_V4HF,
309 AARCH64_SIMD_BUILTIN_FCMLA_LANEQ90_V4HF, AARCH64_SIMD_BUILTIN_FCMLA_LANEQ180_V4HF,
310 AARCH64_SIMD_BUILTIN_FCMLA_LANEQ270_V4HF.
311 * config/aarch64/iterators.md (FCMLA_maybe_lane): New.
312 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins): Add __ARM_FEATURE_COMPLEX.
313 * config/aarch64/aarch64-simd-builtins.def (fcadd90, fcadd270, fcmla0, fcmla90,
314 fcmla180, fcmla270, fcmla_lane0, fcmla_lane90, fcmla_lane180, fcmla_lane270,
315 fcmla_laneq0, fcmla_laneq90, fcmla_laneq180, fcmla_laneq270,
316 fcmlaq_lane0, fcmlaq_lane90, fcmlaq_lane180, fcmlaq_lane270): New.
317 * config/aarch64/aarch64-simd.md (aarch64_fcmla_lane<rot><mode>,
318 aarch64_fcmla_laneq<rot>v4hf, aarch64_fcmlaq_lane<rot><mode>,aarch64_fcadd<rot><mode>,
319 aarch64_fcmla<rot><mode>): New.
320 * config/aarch64/arm_neon.h:
321 (vcadd_rot90_f16): New.
322 (vcaddq_rot90_f16): New.
323 (vcadd_rot270_f16): New.
324 (vcaddq_rot270_f16): New.
327 (vcmla_lane_f16): New.
328 (vcmla_laneq_f16): New.
329 (vcmlaq_lane_f16): New.
330 (vcmlaq_rot90_lane_f16): New.
331 (vcmla_rot90_laneq_f16): New.
332 (vcmla_rot90_lane_f16): New.
333 (vcmlaq_rot90_f16): New.
334 (vcmla_rot90_f16): New.
335 (vcmlaq_laneq_f16): New.
336 (vcmla_rot180_laneq_f16): New.
337 (vcmla_rot180_lane_f16): New.
338 (vcmlaq_rot180_f16): New.
339 (vcmla_rot180_f16): New.
340 (vcmlaq_rot90_laneq_f16): New.
341 (vcmlaq_rot270_laneq_f16): New.
342 (vcmlaq_rot270_lane_f16): New.
343 (vcmla_rot270_laneq_f16): New.
344 (vcmlaq_rot270_f16): New.
345 (vcmla_rot270_f16): New.
346 (vcmlaq_rot180_laneq_f16): New.
347 (vcmlaq_rot180_lane_f16): New.
348 (vcmla_rot270_lane_f16): New.
349 (vcadd_rot90_f32): New.
350 (vcaddq_rot90_f32): New.
351 (vcaddq_rot90_f64): New.
352 (vcadd_rot270_f32): New.
353 (vcaddq_rot270_f32): New.
354 (vcaddq_rot270_f64): New.
358 (vcmla_lane_f32): New.
359 (vcmla_laneq_f32): New.
360 (vcmlaq_lane_f32): New.
361 (vcmlaq_laneq_f32): New.
362 (vcmla_rot90_f32): New.
363 (vcmlaq_rot90_f32): New.
364 (vcmlaq_rot90_f64): New.
365 (vcmla_rot90_lane_f32): New.
366 (vcmla_rot90_laneq_f32): New.
367 (vcmlaq_rot90_lane_f32): New.
368 (vcmlaq_rot90_laneq_f32): New.
369 (vcmla_rot180_f32): New.
370 (vcmlaq_rot180_f32): New.
371 (vcmlaq_rot180_f64): New.
372 (vcmla_rot180_lane_f32): New.
373 (vcmla_rot180_laneq_f32): New.
374 (vcmlaq_rot180_lane_f32): New.
375 (vcmlaq_rot180_laneq_f32): New.
376 (vcmla_rot270_f32): New.
377 (vcmlaq_rot270_f32): New.
378 (vcmlaq_rot270_f64): New.
379 (vcmla_rot270_lane_f32): New.
380 (vcmla_rot270_laneq_f32): New.
381 (vcmlaq_rot270_lane_f32): New.
382 (vcmlaq_rot270_laneq_f32): New.
383 * config/aarch64/aarch64.h (TARGET_COMPLEX): New.
384 * config/aarch64/iterators.md (UNSPEC_FCADD90, UNSPEC_FCADD270,
385 UNSPEC_FCMLA, UNSPEC_FCMLA90, UNSPEC_FCMLA180, UNSPEC_FCMLA270): New.
388 * config/arm/types.md (neon_fcadd, neon_fcmla): New.
390 2019-01-09 Sandra Loosemore <sandra@codesourcery.com>
394 * config/pa/pa.c: Change "can not" to "cannot".
395 * gimple-ssa-evrp-analyze.c: Likewise.
396 * ipa-icf.c: Likewise.
397 * ipa-polymorphic-call.c: Likewise.
398 * ipa-pure-const.c: Likewise.
399 * lra-constraints.c: Likewise.
400 * lra-remat.c: Likewise.
401 * reload1.c: Likewise.
403 * tree-ssa-uninit.c: Likewise.
405 2019-01-09 Sandra Loosemore <sandra@codesourcery.com>
409 * Makefile.in: Mechanically replace "can not" with "cannot".
411 * builtins.c: Likewise.
413 * cgraph.c: Likewise.
414 * cgraph.h: Likewise.
415 * cgraphclones.c: Likewise.
416 * cgraphunit.c: Likewise.
417 * combine-stack-adj.c: Likewise.
418 * combine.c: Likewise.
419 * common/config/i386/i386-common.c: Likewise.
420 * config/aarch64/aarch64.c: Likewise.
421 * config/alpha/sync.md: Likewise.
422 * config/arc/arc.c: Likewise.
423 * config/arc/predicates.md: Likewise.
424 * config/arm/arm-c.c: Likewise.
425 * config/arm/arm.c: Likewise.
426 * config/arm/arm.h: Likewise.
427 * config/arm/arm.md: Likewise.
428 * config/arm/cortex-r4f.md: Likewise.
429 * config/csky/csky.c: Likewise.
430 * config/csky/csky.h: Likewise.
431 * config/darwin-f.c: Likewise.
432 * config/epiphany/epiphany.md: Likewise.
433 * config/i386/i386.c: Likewise.
434 * config/i386/sol2.h: Likewise.
435 * config/m68k/m68k.c: Likewise.
436 * config/mcore/mcore.h: Likewise.
437 * config/microblaze/microblaze.md: Likewise.
438 * config/mips/20kc.md: Likewise.
439 * config/mips/sb1.md: Likewise.
440 * config/nds32/nds32.c: Likewise.
441 * config/nds32/predicates.md: Likewise.
442 * config/pa/pa.c: Likewise.
443 * config/rs6000/e300c2c3.md: Likewise.
444 * config/rs6000/rs6000.c: Likewise.
445 * config/s390/s390.h: Likewise.
446 * config/sh/sh.c: Likewise.
447 * config/sh/sh.md: Likewise.
448 * config/spu/vmx2spu.h: Likewise.
450 * dbxout.c: Likewise.
451 * df-scan.c: Likewise.
452 * doc/cfg.texi: Likewise.
453 * doc/extend.texi: Likewise.
454 * doc/fragments.texi: Likewise.
455 * doc/gty.texi: Likewise.
456 * doc/invoke.texi: Likewise.
457 * doc/lto.texi: Likewise.
458 * doc/md.texi: Likewise.
459 * doc/objc.texi: Likewise.
460 * doc/rtl.texi: Likewise.
461 * doc/tm.texi: Likewise.
463 * emit-rtl.c: Likewise.
464 * emit-rtl.h: Likewise.
465 * except.c: Likewise.
466 * expmed.c: Likewise.
468 * fold-const.c: Likewise.
469 * genautomata.c: Likewise.
470 * gimple-fold.c: Likewise.
471 * hard-reg-set.h: Likewise.
473 * ipa-comdats.c: Likewise.
474 * ipa-cp.c: Likewise.
475 * ipa-devirt.c: Likewise.
476 * ipa-fnsummary.c: Likewise.
477 * ipa-icf.c: Likewise.
478 * ipa-inline-transform.c: Likewise.
479 * ipa-inline.c: Likewise.
480 * ipa-polymorphic-call.c: Likewise.
481 * ipa-profile.c: Likewise.
482 * ipa-prop.c: Likewise.
483 * ipa-pure-const.c: Likewise.
484 * ipa-reference.c: Likewise.
485 * ipa-split.c: Likewise.
486 * ipa-visibility.c: Likewise.
488 * ira-build.c: Likewise.
489 * ira-color.c: Likewise.
490 * ira-conflicts.c: Likewise.
491 * ira-costs.c: Likewise.
492 * ira-int.h: Likewise.
493 * ira-lives.c: Likewise.
496 * loop-invariant.c: Likewise.
497 * loop-unroll.c: Likewise.
498 * lower-subreg.c: Likewise.
499 * lra-assigns.c: Likewise.
500 * lra-constraints.c: Likewise.
501 * lra-eliminations.c: Likewise.
502 * lra-lives.c: Likewise.
503 * lra-remat.c: Likewise.
504 * lra-spills.c: Likewise.
506 * lto-cgraph.c: Likewise.
507 * lto-streamer-out.c: Likewise.
508 * postreload-gcse.c: Likewise.
509 * predict.c: Likewise.
510 * profile-count.h: Likewise.
511 * profile.c: Likewise.
514 * reload.c: Likewise.
515 * reload1.c: Likewise.
517 * resource.c: Likewise.
520 * rtlanal.c: Likewise.
521 * sched-deps.c: Likewise.
522 * sched-ebb.c: Likewise.
523 * sched-rgn.c: Likewise.
524 * sel-sched-ir.c: Likewise.
525 * sel-sched.c: Likewise.
526 * shrink-wrap.c: Likewise.
527 * simplify-rtx.c: Likewise.
528 * symtab.c: Likewise.
529 * target.def: Likewise.
530 * toplev.c: Likewise.
531 * tree-call-cdce.c: Likewise.
532 * tree-cfg.c: Likewise.
533 * tree-complex.c: Likewise.
534 * tree-core.h: Likewise.
535 * tree-eh.c: Likewise.
536 * tree-inline.c: Likewise.
537 * tree-loop-distribution.c: Likewise.
538 * tree-nrv.c: Likewise.
539 * tree-profile.c: Likewise.
540 * tree-sra.c: Likewise.
541 * tree-ssa-alias.c: Likewise.
542 * tree-ssa-dce.c: Likewise.
543 * tree-ssa-dom.c: Likewise.
544 * tree-ssa-forwprop.c: Likewise.
545 * tree-ssa-loop-im.c: Likewise.
546 * tree-ssa-loop-ivcanon.c: Likewise.
547 * tree-ssa-loop-ivopts.c: Likewise.
548 * tree-ssa-loop-niter.c: Likewise.
549 * tree-ssa-phionlycprop.c: Likewise.
550 * tree-ssa-phiopt.c: Likewise.
551 * tree-ssa-propagate.c: Likewise.
552 * tree-ssa-threadedge.c: Likewise.
553 * tree-ssa-threadupdate.c: Likewise.
554 * tree-ssa-uninit.c: Likewise.
555 * tree-ssanames.c: Likewise.
556 * tree-streamer-out.c: Likewise.
559 * vr-values.c: Likewise.
561 2019-01-09 Uroš Bizjak <ubizjak@gmail.com>
563 * config/i386/i386-protos.h (ix86_expand_xorsign): New prototype.
564 (ix86_split_xorsign): Ditto.
565 * config/i386/i386.c (ix86_expand_xorsign): New function.
566 (ix86_split_xorsign): Ditto.
567 * config/i386/i386.md (UNSPEC_XORSIGN): New unspec.
568 (xorsign<mode>3): New expander.
569 (xorsign<mode>3_1): New insn_and_split pattern.
570 * config/i386/sse.md (xorsign<mode>3): New expander.
572 2019-01-09 Eric Botcazou <ebotcazou@adacore.com>
574 * config/sparc/sparc.md (*tablejump_sp32): Merge into...
575 (*tablejump_sp64): Likewise.
576 (*tablejump<P:mode>): ...this.
577 (*call_address_sp32): Merge into...
578 (*call_address_sp64): Likewise.
579 (*call_address<P:mode>): ...this.
580 (*call_symbolic_sp32): Merge into...
581 (*call_symbolic_sp64): Likewise.
582 (*call_symbolic<P:mode>): ...this.
583 (call_value): Remove constraint and add predicate.
584 (*call_value_address_sp32): Merge into...
585 (*call_value_address_sp64): Likewise.
586 (*call_value_address<P:mode>): ...this.
587 (*call_value_symbolic_sp32): Merge into...
588 (*call_value_symbolic_sp64): Likewise.
589 (*call_value_symbolic<P:mode>): ...this.
590 (*sibcall_symbolic_sp32): Merge into...
591 (*sibcall_symbolic_sp64): Likewise.
592 (*sibcall_symbolic<P:mode>): ...this.
593 (sibcall_value): Remove constraint and add predicate.
594 (*sibcall_value_symbolic_sp32): Merge into...
595 (*sibcall_value_symbolic_sp64): Likewise.
596 (*sibcall_value_symbolic<P:mode>): ...this.
597 (window_save): Minor tweak.
598 (*branch_sp32): Merge into...
599 (*branch_sp64): Likewise.
600 (*branch<P:mode>): ...this.
602 2019-01-09 Eric Botcazou <ebotcazou@adacore.com>
603 James Clarke <jrtc27@jrtc27.com>
606 * config/sparc/sparc.c (sparc_legitimize_tls_address): Only use Pmode
607 consistently in TLS address generation and adjust code to the renaming
608 of patterns. Mark calls to __tls_get_addr as const.
609 * config/sparc/sparc.md (tgd_hi22): Turn into...
610 (tgd_hi22<P:mode>): ...this and use Pmode throughout.
611 (tgd_lo10): Turn into...
612 (tgd_lo10<P:mode>): ...this and use Pmode throughout.
613 (tgd_add32): Merge into...
614 (tgd_add64): Likewise.
615 (tgd_add<P:mode>): ...this and use Pmode throughout.
616 (tldm_hi22): Turn into...
617 (tldm_hi22<P:mode>): ...this and use Pmode throughout.
618 (tldm_lo10): Turn into...
619 (tldm_lo10<P:mode>): ...this and use Pmode throughout.
620 (tldm_add32): Merge into...
621 (tldm_add64): Likewise.
622 (tldm_add<P:mode>): ...this and use Pmode throughout.
623 (tldm_call32): Merge into...
624 (tldm_call64): Likewise.
625 (tldm_call<P:mode>): ...this and use Pmode throughout.
626 (tldo_hix22): Turn into...
627 (tldo_hix22<P:mode>): ...this and use Pmode throughout.
628 (tldo_lox10): Turn into...
629 (tldo_lox10<P:mode>): ...this and use Pmode throughout.
630 (tldo_add32): Merge into...
631 (tldo_add64): Likewise.
632 (tldo_add<P:mode>): ...this and use Pmode throughout.
633 (tie_hi22): Turn into...
634 (tie_hi22<P:mode>): ...this and use Pmode throughout.
635 (tie_lo10): Turn into...
636 (tie_lo10<P:mode>): ...this and use Pmode throughout.
637 (tie_ld64): Use DImode throughout.
638 (tie_add32): Merge into...
639 (tie_add64): Likewise.
640 (tie_add<P:mode>): ...this and use Pmode throughout.
641 (tle_hix22_sp32): Merge into...
642 (tle_hix22_sp64): Likewise.
643 (tle_hix22<P:mode>): ...this and use Pmode throughout.
644 (tle_lox22_sp32): Merge into...
645 (tle_lox22_sp64): Likewise.
646 (tle_lox22<P:mode>): ...this and use Pmode throughout.
647 (*tldo_ldub_sp32): Merge into...
648 (*tldo_ldub_sp64): Likewise.
649 (*tldo_ldub<P:mode>): ...this and use Pmode throughout.
650 (*tldo_ldub1_sp32): Merge into...
651 (*tldo_ldub1_sp64): Likewise.
652 (*tldo_ldub1<P:mode>): ...this and use Pmode throughout.
653 (*tldo_ldub2_sp32): Merge into...
654 (*tldo_ldub2_sp64): Likewise.
655 (*tldo_ldub2<P:mode>): ...this and use Pmode throughout.
656 (*tldo_ldsb1_sp32): Merge into...
657 (*tldo_ldsb1_sp64): Likewise.
658 (*tldo_ldsb1<P:mode>): ...this and use Pmode throughout.
659 (*tldo_ldsb2_sp32): Merge into...
660 (*tldo_ldsb2_sp64): Likewise.
661 (*tldo_ldsb2<P:mode>): ...this and use Pmode throughout.
662 (*tldo_ldub3_sp64): Use DImode throughout.
663 (*tldo_ldsb3_sp64): Likewise.
664 (*tldo_lduh_sp32): Merge into...
665 (*tldo_lduh_sp64): Likewise.
666 (*tldo_lduh<P:mode>): ...this and use Pmode throughout.
667 (*tldo_lduh1_sp32): Merge into...
668 (*tldo_lduh1_sp64): Likewise.
669 (*tldo_lduh1<P:mode>): ...this and use Pmode throughout.
670 (*tldo_ldsh1_sp32): Merge into...
671 (*tldo_ldsh1_sp64): Likewise.
672 (*tldo_ldsh1<P:mode>): ...this and use Pmode throughout.
673 (*tldo_lduh2_sp64): Use DImode throughout.
674 (*tldo_ldsh2_sp64): Likewise.
675 (*tldo_lduw_sp32): Merge into...
676 (*tldo_lduw_sp64): Likewise.
677 (*tldo_lduw<P:mode>): ...this and use Pmode throughout.
678 (*tldo_lduw1_sp64): Use DImode throughout.
679 (*tldo_ldsw1_sp64): Likewise.
680 (*tldo_ldx_sp64): Likewise.
681 (*tldo_stb_sp32): Merge into...
682 (*tldo_stb_sp64): Likewise.
683 (*tldo_stb<P:mode>): ...this and use Pmode throughout.
684 (*tldo_sth_sp32): Merge into...
685 (*tldo_sth_sp64): Likewise.
686 (*tldo_sth<P:mode>): ...this and use Pmode throughout.
687 (*tldo_stw_sp32): Merge into...
688 (*tldo_stw_sp64): Likewise.
689 (*tldo_stw<P:mode>): ...this and use Pmode throughout.
690 (*tldo_stx_sp64): Use DImode throughout.
692 2018-01-09 Sudakshina Das <sudi.das@arm.com>
694 * config/aarch64/aarch64.c (aarch64_override_options): Add case to
695 check configure option to set BTI and Return Address Signing.
696 * configure.ac: Add --enable-standard-branch-protection and
697 --disable-standard-branch-protection.
698 * configure: Regenerated.
699 * doc/install.texi: Document the same.
701 2018-01-09 Sudakshina Das <sudi.das@arm.com>
702 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
704 * config.gcc (aarch64*-*-*): Add aarch64-bti-insert.o.
705 * config/aarch64/aarch64.h: Update comment for TRAMPOLINE_SIZE.
706 * config/aarch64/aarch64.c (aarch64_asm_trampoline_template): Update
708 * config/aarch64/aarch64-bti-insert.c: New file.
709 * config/aarch64/aarch64-passes.def (INSERT_PASS_BEFORE): Insert bti
711 * config/aarch64/aarch64-protos.h (make_pass_insert_bti): Declare the
713 * config/aarch64/aarch64.md (unspecv): Add UNSPECV_BTI_NOARG,
714 UNSPECV_BTI_C, UNSPECV_BTI_J and UNSPECV_BTI_JC.
715 (bti_noarg, bti_j, bti_c, bti_jc): New define_insns.
716 * config/aarch64/t-aarch64: Add rule for aarch64-bti-insert.o.
718 2018-01-09 Sudakshina Das <sudi.das@arm.com>
720 * config/aarch64/aarch64-protos.h (aarch64_bti_enabled): Declare.
721 * config/aarch64/aarch64.c (aarch64_handle_no_branch_protection):
722 Disable bti for -mbranch-protection=none.
723 (aarch64_handle_standard_branch_protection): Enable bti for
724 -mbranch-protection=standard.
725 (aarch64_handle_bti_protection): Enable bti for "bti" in the string to
727 (aarch64_bti_enabled): Check if bti is enabled.
728 * config/aarch64/aarch64.opt: Declare target variable.
729 * doc/invoke.texi: Add bti to the -mbranch-protection documentation.
731 2018-01-09 Sudakshina Das <sudi.das@arm.com>
733 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use new
734 epilogue/prologue scratch registers EP0_REGNUM and EP1_REGNUM.
735 (aarch64_expand_epilogue): Likewise.
736 (aarch64_output_mi_thunk): Likewise
737 * config/aarch64/aarch64.h (REG_CLASS_CONTENTS): Change
738 TAILCALL_ADDR_REGS to x16 and x17.
739 * config/aarch64/aarch64.md: Define EP0_REGNUM and EP1_REGNUM.
741 2018-01-09 Sudakshina Das <sudi.das@arm.com>
743 * config/aarch64/aarch64-option-extensions.def: Define
744 AARCH64_OPT_EXTENSION for memtag, rng, sb, ssbs and predres.
745 * gcc/config/aarch64/aarch64.h (AARCH64_FL_RNG): New.
746 (AARCH64_FL_MEMTAG, ARCH64_FL_SB, AARCH64_FL_SSBS): New.
747 (AARCH64_FL_PREDRES): New.
748 (AARCH64_FL_FOR_ARCH8_5): Add AARCH64_FL_SB, AARCH64_FL_SSBS and
749 AARCH64_FL_PREDRES by default.
750 * gcc/doc/invoke.texi: Document rng, memtag, sb, ssbs and predres.
752 2018-01-09 Sudakshina Das <sudi.das@arm.com>
754 * config/aarch64/aarch64-arches.def: Define AARCH64_ARCH for
756 * gcc/config/aarch64/aarch64.h (AARCH64_FL_V8_5): New.
757 (AARCH64_FL_FOR_ARCH8_5, AARCH64_ISA_V8_5): New.
758 * gcc/doc/invoke.texi: Document ARMv8.5-A.
760 2019-01-09 Alejandro Martinez <alejandro.martinezvicente@arm.com>
762 * config/aarch64/aarch64-sve.md (copysign<mode>3): New define_expand.
763 (xorsign<mode>3): Likewise.
765 2019-01-09 Jelinek <jakub@redhat.com>
768 * tree.c (initializer_each_zero_or_onep) <case VECTOR_CST>: Use
769 vector_cst_elt instead of VECTOR_CST_ENCODED_ELT.
771 PR rtl-optimization/88331
772 * function.c (assign_stack_local_1): Don't set dynamic_align_addr if
773 not currently_expanding_to_rtl.
775 2019-01-09 Eric Botcazou <ebotcazou@adacore.com>
777 * doc/invoke.texi (-Os): Remove trailing spaces.
778 (-finline-functions): Remove reference to -O2.
780 2019-01-08 Jakub Jelinek <jakub@redhat.com>
782 PR rtl-optimization/79593
783 * config/i386/i386.md (reg = mem; mem = reg): New define_peephole2.
785 * config/rs6000/rs6000.c (rs6000_delegitimize_address): Delegitimize
786 UNSPEC_FUSION_GPR to its argument. Formatting fixes.
788 2019-01-08 Eric Botcazou <ebotcazou@adacore.com>
791 * config/sparc/sparc.c (function_arg_slotno): Set *PPREGNO & *PPADDING
795 * config/sparc/sparc.c (sparc_delegitimize_address): Deal with naked
796 UNSPECs and UNSPEC_MOVE_GOTDATA specifically.
798 2019-01-08 H.J. Lu <hongjiu.lu@intel.com>
801 * config/i386/i386.c (ix86_avx_u128_mode_exit): Call
802 ix86_avx_u128_mode_entry.
804 2019-01-08 Martin Liska <mliska@suse.cz>
806 PR tree-optimization/88753
807 * tree-switch-conversion.c (switch_conversion::build_one_array):
808 Come up with local variable constructor. Convert first to
809 type of constructor values.
811 2019-01-08 Richard Biener <rguenther@suse.de>
813 PR tree-optimization/86554
814 * tree-ssa-sccvn.c (eliminate_dom_walker, rpo_elim,
815 rpo_avail): Move earlier.
816 (visit_nary_op): When value-numbering to expressions
817 with different overflow behavior make sure there's an
818 available expression on the path.
820 2019-01-08 Sam Tebbs <sam.tebbs@arm.com>
822 * config/aarch64/aarch64.c (BRANCH_PROTECT_STR_MAX,
823 aarch64_parse_branch_protection,
824 struct aarch64_branch_protect_type,
825 aarch64_handle_no_branch_protection,
826 aarch64_handle_standard_branch_protection,
827 aarch64_validate_mbranch_protection,
828 aarch64_handle_pac_ret_protection,
829 aarch64_handle_attr_branch_protection,
830 accepted_branch_protection_string,
831 aarch64_pac_ret_subtypes,
832 aarch64_branch_protect_types,
833 aarch64_handle_pac_ret_leaf): Define.
834 (aarch64_override_options_after_change_1, aarch64_override_options):
835 Add check for accepted_branch_protection_string.
836 (aarch64_option_save): Save accepted_branch_protection_string.
837 (aarch64_option_restore): Save accepted_branch_protection_string.
838 * config/aarch64/aarch64.c (aarch64_attributes): Add branch-protection.
839 * config/aarch64/aarch64.opt: Add mbranch-protection. Deprecate
840 msign-return-address.
841 * doc/invoke.texi: Add mbranch-protection.
843 2019-01-08 Alan Modra <amodra@gmail.com>
846 * genattrtab.c (max_attr_value, min_attr_value, or_attr_value):
847 Delete "unknownp" parameter. Adjust callers. Handle
848 CONST_INT, PLUS, MINUS, and MULT.
849 (attr_value_aligned): Renamed from or_attr_value.
850 (min_attr_value): Return INT_MIN for unhandled rtl case..
851 (min_fn): ..and translate to INT_MAX here.
852 (write_length_unit_log): Modify to cope without "unknown".
853 (write_attr_value): Handle IF_THEN_ELSE.
855 2019-01-07 Richard Sandiford <richard.sandiford@arm.com>
857 * tree-vect-stmts.c (vectorizable_store): Don't use the dataref_offset
858 optimization for masked stores.
860 2019-01-07 Richard Sandiford <richard.sandiford@arm.com>
863 * tree-vect-loop.c (get_initial_defs_for_reduction): Pass the
864 output vector directly to duplicate_and_interleave instead of
865 going through a temporary. Postpone insertion of ctor_seq to
868 2019-01-07 Richard Earnshaw <rearnsha@arm.com>
871 * config/aarch64/aarch64.c (aarch64_expand_subvti): New parameter
872 unsigned_p. Handle signed and unsigned overflow correction as
874 * config/aarch64/aarch64-protos.h (aarch64_expand_subvti): Update
876 * config/aarch64/aarch64.md (addv<mode>4): Use aarch64_plus_operand
878 (add<mode>3_compareV_imm): Make this callable for expanding.
879 (subv<GPI:mode>4): Use register_operand for operand 1. Use
880 aarch64_plus_operand for operand 2.
881 (subv<GPI:mode>_insn): New insn pattern.
882 (subv<GPI:mode>_imm): Likewise.
883 (negv<GPI:mode>3): New expand pattern.
884 (negv<GPI:mode>_insn): New insn pattern.
885 (negv<GPI:mode>_cmp_only): Likewise.
886 (cmpv<GPI:mode>_insn): Likewise.
887 (subvti4): Use register_operand for operand 1. Update call to
888 aarch64_expand_subvti.
889 (usubvti4): Likewise.
890 (negvti3): New expand pattern.
891 (negdi_carryout): New insn pattern.
892 (negvdi_carryinV): New insn pattern.
893 (sub<mode3>_compare1_imm): Delete named insn pattern, make anonymous
894 version the named version.
895 (peepholes to convert to sub<mode3>_compare1_imm): Adjust order of
897 (usub<GPI:mode>3_carryinC, usub<GPI:mode>3_carryinC_z1): New insn
899 (usub<GPI:mode>3_carryinC_z2, usub<GPI:mode>3_carryinC): New insn
901 (sub<mode>3_carryinCV, sub<mode>3_carryinCV_z1_z2): Delete.
902 (sub<mode>3_carryinCV_z1, sub<mode>3_carryinCV_z2): Delete.
903 (sub<mode>3_carryinCV): Delete.
904 (sub<GPI:mode>3_carryinV): New expand pattern.
905 sub<mode>3_carryinV, sub<mode>3_carryinV_z2): New insn patterns.
907 2019-01-07 Richard Biener <rguenther@suse.de>
909 * tree-ssa-uncprop.c (ssa_equip_hash_traits): Remove in favor
910 of tree_operand_hash.
912 2019-01-07 Richard Sandiford <richard.sandiford@arm.com>
914 PR tree-optimization/88598
915 * tree.h (single_nonzero_element): Declare.
916 * tree.c (single_nonzero_element): New function.
917 * match.pd: Fold certain reductions of X & CST to X[I] & CST[I]
918 if I is the only nonzero element of CST.
920 2019-01-07 Richard Sandiford <richard.sandiford@arm.com>
922 PR tree-optimization/88598
923 * tree.h (initializer_each_zero_or_onep): Declare.
924 * tree.c (initializer_each_zero_or_onep): New function.
925 (signed_or_unsigned_type_for): Handle float types too.
926 (unsigned_type_for, signed_type_for): Update comments accordingly.
927 * match.pd: Fold x * { 0 or 1, 0 or 1, ...} to
928 x & { 0 or -1, 0 or -1, ... }.
930 2019-01-07 Jonathan Wakely <jwakely@redhat.com>
932 * doc/install.texi: Replace references to x86_64-unknown-linux-gnu
933 with x86_64-pc-linux-gnu.
935 2019-01-07 Tom de Vries <tdevries@suse.de>
938 * config/nvptx/nvptx.c (has_vector_partitionable_routine_calls_p): New
940 (nvptx_goacc_validate_dims): Force vl32 if calling vector-partitionable
943 2019-01-07 Jakub Jelinek <jakub@redhat.com>
945 * config/i386/sse.md (vec_extract<mode><ssehalfvecmodelower>): Use
946 V_256_512 iterator instead of V_512 and TARGET_AVX instead of
947 TARGET_AVX512F as condition.
950 * dwarf2out.c (const_ok_for_output_1): Remove redundant call to
951 const_not_ok_for_debug_p target hook.
952 (mem_loc_descriptor) <case UNSPEC>: Only call const_ok_for_output_1
953 on UNSPEC and subexpressions thereof if all subexpressions of the
954 UNSPEC are CONSTANT_P.
956 PR tree-optimization/88676
957 * tree-ssa-phiopt.c (two_value_replacement): New function.
958 (tree_ssa_phiopt_worker): Call it.
961 * cfgexpand.c (expand_stack_vars): Only align prev_offset to
962 ASAN_MIN_RED_ZONE_SIZE, not to maximum of that and alignb.
965 * tree-vect-generic.c: Include insn-config.h and recog.h.
966 (expand_vector_piecewise): Add defaulted ret_type argument,
967 if non-NULL, use that in preference to type for the result type.
968 (expand_vector_parallel): Formatting fix.
969 (do_vec_conversion, do_vec_narrowing_conversion,
970 expand_vector_conversion): New functions.
971 (expand_vector_operations_1): Call expand_vector_conversion
972 for VEC_CONVERT ifn calls.
973 * internal-fn.def (VEC_CONVERT): New internal function.
974 * internal-fn.c (expand_VEC_CONVERT): New function.
975 * fold-const-call.c (fold_const_vec_convert): New function.
976 (fold_const_call): Use it for CFN_VEC_CONVERT.
977 * doc/extend.texi (__builtin_convertvector): Document.
979 2019-01-07 Tom de Vries <tdevries@suse.de>
981 * config/nvptx/nvptx-protos.h (nvptx_output_red_partition): Declare.
982 * config/nvptx/nvptx.c (vector_red_size, vector_red_align,
983 vector_red_partition, vector_red_sym): New global variables.
984 (nvptx_option_override): Initialize vector_red_sym.
985 (nvptx_declare_function_name): Restore red_partition register.
986 (nvptx_file_end): Emit code to declare the vector reduction variables.
987 (nvptx_output_red_partition): New function.
988 (nvptx_expand_shared_addr): Add vector argument. Use it to handle
989 large vector reductions.
990 (enum nvptx_builtins): Add NVPTX_BUILTIN_VECTOR_ADDR.
991 (nvptx_init_builtins): Add VECTOR_ADDR.
992 (nvptx_expand_builtin): Update call to nvptx_expand_shared_addr.
993 Handle nvptx_expand_shared_addr.
994 (nvptx_get_shared_red_addr): Add vector argument and handle large
996 (nvptx_goacc_reduction_setup): Add offload_attrs argument and handle
998 (nvptx_goacc_reduction_init): Likewise.
999 (nvptx_goacc_reduction_fini): Likewise.
1000 (nvptx_goacc_reduction_teardown): Likewise.
1001 (nvptx_goacc_reduction): Update calls to nvptx_goacc_reduction_{setup,
1002 init,fini,teardown}.
1003 (nvptx_init_axis_predicate): Initialize vector_red_partition.
1004 (nvptx_set_current_function): Init vector_red_partition.
1005 * config/nvptx/nvptx.md (UNSPECV_RED_PART): New unspecv.
1006 (nvptx_red_partition): New insn.
1007 * config/nvptx/nvptx.h (struct machine_function): Add red_partition.
1009 2019-01-07 Tom de Vries <tdevries@suse.de>
1012 * config/nvptx/nvptx.c (nvptx_process_pars): Don't emit barriers for
1015 2019-01-07 Tom de Vries <tdevries@suse.de>
1017 * config/nvptx/nvptx.c (oacc_bcast_partition): Declare.
1018 (nvptx_option_override): Init oacc_bcast_partition.
1019 (nvptx_init_oacc_workers): New function.
1020 (nvptx_declare_function_name): Call nvptx_init_oacc_workers.
1021 (nvptx_needs_shared_bcast): New function.
1022 (nvptx_find_par): Generalize to enable vectors to use shared-memory
1024 (nvptx_shared_propagate): Initialize vector bcast partition and
1025 synchronization state.
1026 (nvptx_single): Generalize to enable vectors to use shared-memory
1028 (nvptx_process_pars): Likewise.
1029 (nvptx_set_current_function): Initialize oacc_broadcast_partition.
1030 * config/nvptx/nvptx.h (struct machine_function): Add
1031 bcast_partition and sync_bar members.
1033 2019-01-07 Tom de Vries <tdevries@suse.de>
1035 * config/nvptx/nvptx.c (nvptx_welformed_vector_length_p)
1036 (nvptx_apply_dim_limits): New function.
1037 (nvptx_goacc_validate_dims_1): Allow PTX_MAX_VECTOR_LENGTH larger than
1040 2019-01-07 Tom de Vries <tdevries@suse.de>
1042 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims_1): Move warnings to
1043 as late as possible.
1045 2019-01-07 Tom de Vries <tdevries@suse.de>
1047 * config/nvptx/nvptx.c (PTX_VECTOR_LENGTH): Remove.
1048 (PTX_DEFAULT_VECTOR_LENGTH, PTX_MAX_VECTOR_LENGTH): Define.
1049 (nvptx_goacc_validate_dims_1, nvptx_dim_limit)
1050 (nvptx_goacc_reduction_fini): Use PTX_DEFAULT_VECTOR_LENGTH,
1051 PTX_MAX_VECTOR_LENGTH and PTX_WARP_SIZE instead of PTX_VECTOR_LENGTH.
1053 2019-01-07 Tom de Vries <tdevries@suse.de>
1055 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Add asserts.
1057 2019-01-07 Tom de Vries <tdevries@suse.de>
1059 * omp-offload.c (oacc_get_min_dim): New function.
1060 * omp-offload.h (oacc_get_min_dim): Declare.
1062 2018-12-26 Mateusz B <mateuszb@poczta.onet.pl>
1065 * config/i386/i386.c (function_value_ms_64): Return small sturct in
1066 AX_REG and float/double in FIRST_SSE_REG for 4 or 8 byte modes.
1068 2019-01-05 Jan Hubicka <hubicka@ucw.cz>
1072 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
1074 * ipa-inline.c (edge_badness): Use inlined_time instead of
1075 inline_summaries->get.
1077 2019-01-05 Jan Hubicka <hubicka@ucw.cz>
1079 * opts.c (enable_fdo_optimizations): Enable
1080 version-loops-for-strides, loop-interchange, unrol-and-jam
1081 and tree-loop-distribution.
1082 * invoke.texi: Document newly enabled options.
1084 2019-01-05 Jan Hubicka <hubicka@ucw.cz>
1086 * doc/invoke.texi (max-inline-insns-small): New parameters.
1087 * ipa-inline.c (want_early_inline_function_p): simplify.
1088 (want_inline_small_function_p): Fix pasto from previous patch;
1089 use max-inline-insns-small bound.
1090 * params.def (max-inline-insns-small): New param.
1091 * ipa-fnsummary.c (analyze_function_body): Initialize time/size
1092 variables correctly.
1094 2019-01-05 Jan Hubicka <hubicka@ucw.cz>
1096 * doc/invoke.texi: Document max-inline-insns-size,
1097 uninlined-function-insns, uninlined-function-time,
1098 uninlined-thunk-insns and uninlined-thunk-time.
1099 * params.def: Add max-inline-insns-size,
1100 uninlined-function-insns, uninlined-function-time,
1101 uninlined-thunk-insns and uninlined-thunk-time.
1102 * ipa-fnsummary.c (compute_fn_summary, analyze_function_body): Use
1104 * ipa-inline.c (can_inline_edge_by_limits_p,
1105 want_inline_small_function_p): Use new parameters.
1107 2019-01-05 Jan Hubicka <hubicka@ucw.cz>
1109 * ipa-fnsummary.c (analyze_function_body): Fix accounting of time.
1111 2019-01-05 Jakub Jelinek <jakub@redhat.com>
1115 * expr.c (expand_assignment): For calls returning VLA structures
1116 if to_rtx is not a MEM, force it into a stack temporary.
1119 * dwarf2out.c (const_ok_for_output_1): Reject MINUS that contains
1120 SYMBOL_REF, CODE_LABEL or UNSPEC in subexpressions of second argument.
1121 Reject PLUS that contains SYMBOL_REF, CODE_LABEL or UNSPEC in
1122 subexpressions of both operands.
1123 (mem_loc_descriptor): Handle UNSPEC if target hook acks it and all the
1124 subrtxes are CONSTANT_P.
1125 * config/i386/i386.c (ix86_const_not_ok_for_debug_p): Revert
1128 2019-01-04 Jan Hubicka <hubicka@ucw.cz>
1130 * params.def (hot-bb-count-ws-permille): Set to 990.
1132 2019-01-04 Martin Sebor <msebor@redhat.com>
1135 * attribs.c (decls_mismatched_attributes): Avoid warning for attribute
1138 2019-01-04 Martin Sebor <msebor@redhat.com>
1141 * doc/extend.texi (attribute alloc_align, alloc_size): Update.
1143 2019-01-04 Jakub Jelinek <jakub@redhat.com>
1145 * gdbinit.in: Turn off pagination for the skip commands, restore
1146 it to previous state afterwards.
1148 2019-01-04 Sam Tebbs <sam.tebbs@arm.com>
1151 * gcc.target/aarch64/combine_bfxil.c: Change scan-assembler-times bfxil
1154 2019-01-04 Jakub Jelinek <jakub@redhat.com>
1157 * config/i386/i386.c (ix86_expand_divmod_libfunc): Use mode instead
1158 of GET_MODE (opN) as modes of the libcall arguments.
1160 2019-01-04 Jan Beulich <jbeulich@suse.com>
1162 * config/i386/sse.md
1163 (<avx512>_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>,
1164 <avx512>_cmp<mode>3<mask_scalar_merge_name>,
1165 <avx512>_ucmp<mode>3<mask_scalar_merge_name>,
1166 <avx512>_ucmp<mode>3<mask_scalar_merge_name>,
1167 avx512f_vmcmp<mode>3<round_saeonly_name>,
1168 avx512f_vmcmp<mode>3_mask<round_saeonly_name>,
1169 avx512f_maskcmp<mode>3,
1170 <avx512>_cvt<ssemodesuffix>2mask<mode>,
1171 <avx512>_cvt<ssemodesuffix>2mask<mode>,
1172 *<avx512>_cvtmask2<ssemodesuffix><mode>,
1173 *<avx512>_cvtmask2<ssemodesuffix><mode>,
1174 <avx512>_eq<mode>3<mask_scalar_merge_name>_1,
1175 <avx512>_eq<mode>3<mask_scalar_merge_name>_1,
1176 <avx512>_gt<mode>3<mask_scalar_merge_name>,
1177 <avx512>_gt<mode>3<mask_scalar_merge_name>,
1178 <avx512>_testm<mode>3<mask_scalar_merge_name>,
1179 <avx512>_testnm<mode>3<mask_scalar_merge_name>,
1180 *<avx512>_testm<VI1248_AVX512VLBW:mode>3_zext,
1181 *<avx512>_testm<VI1248_AVX512VLBW:mode>3_zext_mask,
1182 *<avx512>_testnm<VI1248_AVX512VLBW:mode>3_zext,
1183 *<avx512>_testnm<VI1248_AVX512VLBW:mode>3_zext_mask,
1184 avx512cd_maskb_vec_dup<mode>,
1185 avx512cd_maskw_vec_dup<mode>,
1186 avx512dq_fpclass<mode><mask_scalar_merge_name>,
1187 avx512dq_vmfpclass<mode>,
1188 avx512vl_vpshufbitqmb<mode><mask_scalar_merge_name>): Use =k
1191 2019-01-03 Martin Sebor <msebor@redhat.com>
1193 PR tree-optimization/88659
1194 * calls.c (maybe_warn_nonstring_arg): Avoid assuming maxlen is set.
1196 2019-01-03 Aaron Sawdey <acsawdey@linux.ibm.com>
1198 * config/rs6000/rs6000-string.c (expand_block_move): Don't use
1199 unaligned vsx and avoid lxvd2x/stxvd2x.
1200 (gen_lvx_v4si_move): New function.
1202 2019-01-03 Tom de Vries <tdevries@suse.de>
1204 * config/nvptx/nvptx.c (MACH_VECTOR_LENGTH, MACH_MAX_WORKERS): Define.
1205 (init_axis_dim, nvptx_mach_max_workers, nvptx_mach_vector_length): New
1207 * config/nvptx/nvptx.h (struct machine_function): Add axis_dims.
1209 2019-01-03 Tom de Vries <tdevries@suse.de>
1211 * config/nvptx/nvptx.c (struct offload_attrs): New.
1212 (populate_offload_attrs): New function. Factor mask extraction out of
1213 nvptx_reorg. Add extraction of dimensions.
1214 (nvptx_reorg): Use populate_offload_attrs.
1216 2019-01-03 Tom de Vries <tdevries@suse.de>
1218 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims_1): Add early-out
1219 cases for oacc_min_dims_p and routine_p. Add asserts for
1220 oacc_default_dims_p and offload_region_p.
1222 2019-01-03 Tom de Vries <tdevries@suse.de>
1224 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims_1): New function,
1226 (nvptx_goacc_validate_dims): ... here.
1228 2019-01-03 Jan Hubicka <hubicka@ucw.cz>
1230 PR tree-optimization/85574
1231 * tree-ssa-uncprop.c (struct equiv_hash_elt): Remove unused
1233 (struct ssa_equip_hash_traits): Declare.
1234 (val_ssa_equiv): Use custom hash traits using operand_equal_p.
1236 2019-01-03 Jakub Jelinek <jakub@redhat.com>
1239 * dwarf2out.c (modified_type_die): If type is equal to sizetype,
1240 change it to qualified_type.
1242 2019-01-03 Jan Hubicka <hubicka@ucw.cz>
1244 * ipa-utils.c (scale_ipa_profile_for_fn): Break out from ...
1245 (ipa_merge_profiles): ... here; do not ICE on thunks and aliases.
1247 2019-01-02 Martin Sebor <msebor@redhat.com>
1248 Jeff Law <law@redhat.com>
1250 * gimple-fold.c (strlen_range_kind): Remove SRK_LENRANGE_2.
1251 (get_range_strlen_tree): Update appropriately.
1253 * gimple-fold.h (get_range_strlen): Drop unused last argument.
1255 * gimple-fold.c (gimple_fold_builtin_strlen): Use set_strlen_range
1256 rather than set_range_info.
1257 * tree-ssa-strlen.c (set_strlen_range): Extracted from
1258 maybe_set_strlen_range. Handle potentially boundary crossing
1259 cases more conservatively.
1260 (maybe_set_strlen_range): Parts refactored into set_strlen_range.
1261 Call set_strlen_range.
1262 * tree-ssa-strlen.h (set_strlen_range): Add prototype.
1265 * gimple-fold.c (get_range_strlen): Update prototype to no longer
1266 need the flexp argument.
1267 (get_range_strlen_tree): Drop flexp argument. Drop flexp argument
1268 from calls to get_range_strlen. Update comments. Just update
1269 VAL for an unterminated const char array and let the reset of the
1270 code handle it normally. No longer try to set *flexp. Adjust
1272 (get_range_strlen): Update for the new get_range_strlen API.
1273 (get_maxval_strlen): Similarly.
1274 (gimple_fold_builtin_strlen): Handle update meaning of return value
1275 from get_range_strlen.
1276 * gimple-ssa-sprintf.c (get_string_length): Update for the new
1277 get_range_strlen API.
1279 2019-01-02 Jan Hubicka <hubicka@ucw.cz>
1282 * varpool.c (varpool_node::ctor_useable_for_folding_p): Also return
1283 false at WPA time when body was removed.
1285 2019-01-02 Martin Liska <mliska@suse.cz>
1287 PR tree-optimization/88650
1288 * predict.c (set_even_probabilities): Calculate probability
1289 remainer only when really used.
1291 2019-01-02 Richard Biener <rguenther@suse.de>
1294 * tree-data-ref.c (analyze_subscript_affine_affine): Use
1295 widest_ints when mangling max_stmt_execution results.
1297 2019-01-02 Richard Biener <rguenther@suse.de>
1299 PR tree-optimization/88621
1300 * tree-ssa-loop-im.c (gather_mem_refs_stmt): Fix pastos, avoid
1301 bitfields when canoncalizing.
1303 2019-01-02 Richard Biener <rguenther@suse.de>
1306 * config/i386/x86-tune-costs.h (intel_cost): Adjust
1307 cost of cheap SSE instruction.
1309 2019-01-02 Richard Biener <rguenther@suse.de>
1312 * ipa-icf.h (sem_item_optimizer::sort_congruence_split): Declare.
1313 * ipa-icf.c (sem_item_optimizer::sort_congruence_split): New
1315 (sem_item_optimizer::do_congruence_step_f): Sort the congruence
1316 set after UIDs before splitting them.
1318 2019-01-01 Martin Sebor <msebor@redhat.com>
1319 Jeff Law <law@redhat.com>
1321 * gimple-fold.c (get_range_strlen_tree): Record if the computed
1322 length is optimistic. If it is, then arrange to compute the
1323 conservative length as well.
1325 * gimple-fold.h (get_range_strlen): Update prototype.
1326 * builtins.c (check_access): Update call to get_range_strlen to use
1327 c_strlen_data pointer. Change various variable accesses to instead
1328 pull data from the c_strlen_data structure.
1329 (check_strncat_sizes, expand_builtin_strncat): Likewise.
1330 * calls.c (maybe_warn_nonstring_arg): Likewise.
1331 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Likewise. Reset
1332 minimum length if maximum lengh is unknown.
1333 * gimple-ssa-sprintf.c (get_string_length): Likewise. Drop code
1334 that used c_strlen, it's no longer needed. Restructure slightly.
1335 (format_string): Set unlikely range appropriately.
1336 * gimple-fold.c (get_range_strlen): Update comments. Fix minor
1338 (get_range_strlen): Accept c_strlen_data pointer for external
1339 call sites as well. Pass through to call to internal get_range_strlen.
1340 Adjust minlen, maxlen and maxbound as needed.
1341 (get_maxval_strlen): Update comments.
1342 (gimple_fold_builtin_strlen): Update call to get_range_strlen
1343 to use c_strlen_data pointer. Change variable accesses to instead
1344 use c_strlen_data data members.
1346 * gimple-fold.c (get_range_strlen): Update prototype.
1347 (get_range_strlen_tree): Update prototype. Drop minlen/maxlen
1348 local variables. Use pdata to return information to caller.
1349 Update calls to get_range_strlen. Update pdata->maxbound.
1350 (get_range_strlen -- static version): Similarly.
1351 (get_range_strlen -- extern version): Update for internal
1352 get_range_strlen API change. Convert to external data format.
1353 (get_maxval_strlen): Similarly.
1355 2019-01-01 Jan Hubicka <hubicka@ucw.cz>
1357 * coverage.c (get_coverage_counts): Use current_function_decl.
1358 * profile.c (read_thunk_profile): New function.
1359 (branch_prob): Add THUNK parameter.
1360 * tree-profile.c (tree_profiling): Handle thunks.
1361 * value-prof.c (init_node_map): Handle thunks.
1362 * value-prof.h (branch_prob): Upate prototype.
1363 (read_thunk_profile): Declare.
1365 2019-01-01 Jakub Jelinek <jakub@redhat.com>
1367 Update copyright years.
1369 * gcc.c (process_command): Update copyright notice dates.
1370 * gcov-dump.c (print_version): Ditto.
1371 * gcov.c (print_version): Ditto.
1372 * gcov-tool.c (print_version): Ditto.
1373 * gengtype.c (create_file): Ditto.
1374 * doc/cpp.texi: Bump @copying's copyright year.
1375 * doc/cppinternals.texi: Ditto.
1376 * doc/gcc.texi: Ditto.
1377 * doc/gccint.texi: Ditto.
1378 * doc/gcov.texi: Ditto.
1379 * doc/install.texi: Ditto.
1380 * doc/invoke.texi: Ditto.
1382 Copyright (C) 2019 Free Software Foundation, Inc.
1384 Copying and distribution of this file, with or without modification,
1385 are permitted in any medium without royalty provided the copyright
1386 notice and this notice are preserved.