1 2019-02-26 Jakub Jelinek <jakub@redhat.com>
4 * tree.c (valid_constant_size_p): Deal with size INTEGER_CSTs
5 with types other than sizetype/ssizetype.
7 2019-02-26 Eric Botcazou <ebotcazou@adacore.com>
9 * config/sparc/sparc-opts.h (enum processor_type): Rename to...
10 (enum sparc_processor_type): ...this.
11 (enum sparc_code_model_type): New enumeration type.
12 (enum sparc_memory_model_type): Tweak comments.
13 * config/sparc/sparc.opt (mcpu): Adjust to above renaming.
15 (mcmodel): Use sparc_code_model enumeration and variable.
16 (sparc_code_model): New enumeration.
17 (mdebug): Add Undocumented marker.
18 * config/sparc/sparc.h (enum cmodel): Delete.
19 (sparc_cmodel): Likewise.
20 (TARGET_CM_MEDLOW): Adjust to above renaming.
21 (TARGET_CM_MEDMID): Likewise.
22 (TARGET_CM_MEDANY): Likewise.
23 (TARGET_CM_EMBMEDANY): Likewise.
24 * config/sparc/sparc.c (sparc_cmodel): Delete.
25 (sparc_option_override): Remove string/value mapping support for the
26 code model. Move code and memory model support to after the handling
27 of target flags. Do private machine setup last.
28 (sparc_emit_set_symbolic_const64): Use sparc_code_model.
29 (sparc_legitimize_reload_address): Likewise.
30 (sparc_output_mi_thunk): Likewise.
31 * config/sparc/sparc.md (cpu): Adjust comment to above renaming.
33 2019-02-26 Jakub Jelinek <jakub@redhat.com>
35 PR tree-optimization/89500
36 * tree-ssa-strlen.c (stridx_strlenloc): Adjust comment.
37 (handle_builtin_strlen): Remove noncst_bound variable. Always
38 optimize strnlen (x, 0) to 0. Optimize strnlen (x, cst) to
39 cst if the first cst bytes starting at x are known to be non-zero,
40 even if the string is not zero terminated. Don't try to modify
41 *si for strnlen. Update strlen_to_stridx only for strlen or if
42 we can prove strnlen returns the same value as strlen would.
44 2019-02-26 Martin Liska <mliska@suse.cz>
46 * alloc-pool.h (struct pool_usage): Remove extra
48 * bitmap.h (struct bitmap_usage): Likewise.
49 * ggc-common.c (struct ggc_usage): Likewise.
50 * mem-stats.h (struct mem_usage): Likewise.
51 (mem_alloc_description::dump): Print dash lines
52 here and repeat header at the end of a table report.
53 It's then more readable.
54 * tree-phinodes.c (phinodes_print_statistics): Make
56 * tree-ssanames.c (ssanames_print_statistics): Likewise.
57 * vec.c (struct vec_usage): Remove extra print_dash_line.
58 * vec.h (vec_safe_grow_cleared): Pass PASS_MEM_STAT.
60 2019-02-26 Uroš Bizjak <ubizjak@gmail.com>
62 * doc/extend.texi (__builtin_object_size):
63 Use @pxref instead of @xref inside parenthesis.
64 (__builtin_has_attribute): Add missing comma after @xref.
65 (__builtin_object_size): Ditto.
66 * doc/md.texi (cond_*{mode}): Use @samp instead of @var around op1[i].
67 * fortran/invoke.texi (-ffpe-trap): Use @var for every item
70 2019-02-26 Jeff Law <law@redhat.com>
72 PR rtl-optimization/87761
73 * regcprop.c (copyprop_hardreg_forward_1): Use REG_UNUSED notes to
74 detect obviously dead insns and delete them.
76 2019-02-26 Richard Biener <rguenther@suse.de>
78 PR tree-optimization/89505
79 * tree-ssa-structalias.c (compute_dependence_clique): Make sure
80 to handle restrict pointed-to vars with multiple subvars
83 2019-02-26 Richard Biener <rguenther@suse.de>
85 PR tree-optimization/89489
86 * tree-parloops.c (create_loop_fn): Copy over last_clique.
88 2019-02-26 Eric Botcazou <ebotcazou@adacore.com>
90 * tree-ssa-dom.c (edge_info::derive_equivalences) <BIT_IOR_EXPR>: Fix
91 and move around comment.
92 <BIT_AND_EXPR>: Likewise.
93 <BIT_NOT_EXPR>: Add specific handling for boolean types.
95 2019-02-26 Jakub Jelinek <jakub@redhat.com>
98 * config/i386/i386.c (remove_partial_avx_dependency): Call
99 df_analyze etc. before creation of the v4sf_const0 pseudo, rather than
100 after changing possibly many instructions to use that pseudo. Fix up
101 insertion of v4sf_const0 setter at the start of bb.
103 2019-02-25 Sandra Loosemore <sandra@codesourcery.com>
106 * doc/extend.texi (Variadic Pointer Args): New section.
108 2019-02-25 Sandra Loosemore <sandra@codesourcery.com>
109 Martin Sebor <msebor@gmail.com>
111 * common.opt (Wattribute-alias): Likewise.
112 * doc/invoke.texi (Option Summary): List general form of
113 -Wattribute-alias=. List positive form of -Wmissing-attributes.
114 (-Wmissing-attributes): Invert entry, rewrite and correct default.
115 Add cross-references.
116 (-Wattribute-alias): Rewrite and correct default. Mention
117 considered attributes (same as for -Wmissing-attributes).
119 2019-02-25 Paul A. Clarke <pc@us.ibm.com>
121 * config/rs6000/emmintrin.h (_mm_cvtpd_epi32): Fix big endian.
122 (_mm_cvtpd_ps): Likewise.
123 (_mm_cvttpd_epi32): Likewise.
126 * config/rs6000/xmmintrin.h (_mm_cvtss_f32): Fix type mismatch.
127 (_mm_cvt_ss2si): Fix type mismatch and 32-bit.
130 * config/rs6000/xmmintrin.h (_mm_movemask_pi8): Fix 32-bit.
132 2019-02-25 Tamar Christina <tamar.christina@arm.com>
135 * common/config/aarch64/aarch64-common.c
136 (struct aarch64_option_extension): Add is_synthetic.
137 (all_extensions): Use it.
138 (TARGET_OPTION_INIT_STRUCT): Define hook.
139 (struct gcc_targetm_common): Moved to end.
140 (all_extensions_by_on): New.
141 (opt_ext_cmp, typedef opt_ext): New.
142 (aarch64_option_init_struct): New.
143 (aarch64_contains_opt): New.
144 (aarch64_get_extension_string_for_isa_flags): Output smallest set.
145 * config/aarch64/aarch64-option-extensions.def
146 (AARCH64_OPT_EXTENSION): Explicitly include AES and SHA2 in crypto.
147 (fp, simd, crc, lse, fp16, rcpc, rdma, dotprod, aes, sha2, sha3,
148 sm4, fp16fml, sve, profile, rng, memtag, sb, ssbs, predres):
149 Set is_synthetic to false.
150 (crypto): Set is_synthetic to true.
151 * config/aarch64/driver-aarch64.c (AARCH64_OPT_EXTENSION): Add
154 2019-02-25 Tamar Christina <tamar.christina@arm.com>
156 * config/arm/arm_neon.h (vfmlal_low_u32, vfmlsl_low_u32,
157 vfmlal_high_u32, vfmlsl_high_u32, vfmlalq_low_u32, vfmlslq_low_u32,
158 vfmlalq_high_u32, vfmlslq_high_u32, vfmlal_lane_low_u32,
159 vfmlal_lane_high_u32, vfmlalq_laneq_low_u32, vfmlalq_lane_low_u32,
160 vfmlal_laneq_low_u32, vfmlalq_laneq_high_u32, vfmlalq_lane_high_u32,
161 vfmlal_laneq_high_u32, vfmlsl_lane_low_u32, vfmlsl_lane_high_u32,
162 vfmlslq_laneq_low_u32, vfmlslq_lane_low_u32, vfmlsl_laneq_low_u32,
163 vfmlslq_laneq_high_u32, vfmlslq_lane_high_u32, vfmlsl_laneq_high_u32):
165 (vfmlal_low_f16, vfmlsl_low_f16, vfmlal_high_f16, vfmlsl_high_f16,
166 vfmlalq_low_f16, vfmlslq_low_f16, vfmlalq_high_f16, vfmlslq_high_f16,
167 vfmlal_lane_low_f16, vfmlal_lane_high_f16, vfmlalq_laneq_low_f16,
168 vfmlalq_lane_low_f16, vfmlal_laneq_low_f16, vfmlalq_laneq_high_f16,
169 vfmlalq_lane_high_f16, vfmlal_laneq_high_f16, vfmlsl_lane_low_f16,
170 vfmlsl_lane_high_f16, vfmlslq_laneq_low_f16, vfmlslq_lane_low_f16,
171 vfmlsl_laneq_low_f16, vfmlslq_laneq_high_f16, vfmlslq_lane_high_f16,
172 vfmlsl_laneq_high_f16): ... To this.
173 * config/arm/neon.md: Update comments.
175 2019-02-25 Tamar Christina <tamar.christina@arm.com>
177 * config/aarch64/arm_neon.h (vfmlal_low_u32, vfmlsl_low_u32,
178 vfmlalq_low_u32, vfmlslq_low_u32, vfmlal_high_u32, vfmlsl_high_u32,
179 vfmlalq_high_u32, vfmlslq_high_u32, vfmlal_lane_low_u32,
180 vfmlsl_lane_low_u32, vfmlal_laneq_low_u32, vfmlsl_laneq_low_u32,
181 vfmlalq_lane_low_u32, vfmlslq_lane_low_u32, vfmlalq_laneq_low_u32,
182 vfmlslq_laneq_low_u32, vfmlal_lane_high_u32, vfmlsl_lane_high_u32,
183 vfmlal_laneq_high_u32, vfmlsl_laneq_high_u32, vfmlalq_lane_high_u32,
184 vfmlslq_lane_high_u32, vfmlalq_laneq_high_u32, vfmlslq_laneq_high_u32):
186 (vfmlal_low_f16, vfmlsl_low_f16, vfmlalq_low_f16, vfmlslq_low_f16,
187 vfmlal_high_f16, vfmlsl_high_f16, vfmlalq_high_f16, vfmlslq_high_f16,
188 vfmlal_lane_low_f16, vfmlsl_lane_low_f16, vfmlal_laneq_low_f16,
189 vfmlsl_laneq_low_f16, vfmlalq_lane_low_f16, vfmlslq_lane_low_f16,
190 vfmlalq_laneq_low_f16, vfmlslq_laneq_low_f16, vfmlal_lane_high_f16,
191 vfmlsl_lane_high_f16, vfmlal_laneq_high_f16, vfmlsl_laneq_high_f16,
192 vfmlalq_lane_high_f16, vfmlslq_lane_high_f16, vfmlalq_laneq_high_f16,
193 vfmlslq_laneq_high_f16): ... To this.
195 2019-02-25 Alexander Monakov <amonakov@ispras.ru>
197 PR rtl-optimization/86096
198 * df-scan.c (df_mw_compare): Do not check mw_reg fields when
199 comparing mw_order values.
201 2019-02-25 Jakub Jelinek <jakub@redhat.com>
204 * config/arm/arm.md (*subsi3_carryin_const): Use
205 arm_neg_immediate_operand predicate instead of
206 arm_not_immediate_operand, "L" constraint instead of "K" and
207 print it using %n2 instead of %B2.
208 (*subsi3_carryin_const0): New define_insn.
209 (*subsi3_carryin_compare_const): Use const_int_I_operand predicate
210 instead of arm_not_operand and "I" constraint instead of "K" and
211 print it using %n3 instead of %B2. Instead of using match_dup 2 add
212 another match_operand and in the condition check that it is negation
214 (*subsi3_carryin_compare_const0): New define_ins.
215 (*subdi_di_zesidi): Adjust to use *subsi3_carryin_const0 instead of
216 *subsi3_carryin_const.
217 (*arm_cmpdi_insn): Fix splitting into *subsi3_carryin_compare_const,
218 split into *subsi3_carryin_compare_const0 if the highpart is zero.
221 * config/arm.vfp.md (*negdf2_vfp): Use
222 gen_int_mode (0x80000000, SImode) instead of GEN_INT (0x80000000).
223 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
225 2019-02-24 Jakub Jelinek <jakub@redhat.com>
227 PR rtl-optimization/89445
228 * simplify-rtx.c (simplify_ternary_operation): Don't use
229 simplify_merge_mask on operands that may trap.
230 * rtlanal.c (may_trap_p_1): Use FLOAT_MODE_P instead of
231 SCALAR_FLOAT_MODE_P checks. For integral division by zero, if
232 second operand is CONST_VECTOR, check if any element could be zero.
233 Don't expect traps for VEC_{MERGE,SELECT,CONCAT,DUPLICATE} unless
234 their operands can trap.
236 2019-02-23 Martin Sebor <msebor@redhat.com>
238 * gimple-ssa-sprintf.c (target_strtol): Rename...
239 (target_strtohwi): ...to this. Handle values up to HOST_WIDE_INT_MAX.
240 (parse_directive): Adjust to name change. Use HOST_WIDE_INT_MAX to
241 check for range error.
243 2019-02-23 H.J. Lu <hongjiu.lu@intel.com>
246 * opts-common.c (prune_options): Also prune joined switches
247 with Negative and RejectNegative.
248 * config/i386/i386.opt (march=): Add Negative(march=).
249 (mtune=): Add Negative(mtune=).
250 * doc/options.texi: Document Negative used together with Joined
253 2019-02-22 Martin Sebor <msebor@redhat.com>
255 * doc/extend.texi (Other Builtins): Add
256 __builtin_is_constant_evaluated.
258 2019-02-22 Richard Biener <rguenther@suse.de>
260 PR tree-optimization/87609
261 * tree-cfg.c (gimple_duplicate_bb): Only remap inlined cliques.
263 2019-02-22 Jeff Law <law@redhat.com>
265 PR rtl-optimization/87761
266 * config/mips/mips.md: Add new combiner pattern to recognize
267 a bitfield extraction using (ashiftrt (truncate (ashift (...)))).
269 2019-02-22 Matthew Malcomson <matthew.malcomson@arm.com>
272 * config/aarch64/aarch64.md: Use aarch64_general_reg predicate on
273 destination register in peepholes generating patterns for ADDS/SUBS.
274 (add<mode>3_compare0,
275 *addsi3_compare0_uxtw, add<mode>3_compareC,
276 add<mode>3_compareV_imm, add<mode>3_compareV,
277 *adds_<optab><ALLX:mode>_<GPI:mode>,
278 *subs_<optab><ALLX:mode>_<GPI:mode>,
279 *adds_<optab><ALLX:mode>_shift_<GPI:mode>,
280 *subs_<optab><ALLX:mode>_shift_<GPI:mode>,
281 *adds_<optab><mode>_multp2, *subs_<optab><mode>_multp2,
282 *sub<mode>3_compare0, *subsi3_compare0_uxtw,
283 sub<mode>3_compare1): Allow stack pointer for source register.
284 * config/aarch64/predicates.md (aarch64_general_reg): New predicate.
286 2019-02-22 Martin Sebor <msebor@redhat.com>
288 PR tree-optimization/88993
289 PR tree-optimization/88853
290 * gimple-ssa-sprintf.c (sprintf_dom_walker::call_info::is_file_func):
292 (sprintf_dom_walker::call_info::is_string_func): New helper.
293 (format_directive): Only issue "may exceed" 4095/INT_MAX warnings
294 for formatted string functions.
295 (sprintf_dom_walker::handle_gimple_call): Fix a typo in a comment.
297 2019-02-22 Martin Sebor <msebor@redhat.com>
300 * c-parser.c (sizeof_ptr_memacc_comptypes): Avoid warning in
301 unreachable subexpressions.
303 2019-02-22 H.J. Lu <hongjiu.lu@intel.com>
304 Hongtao Liu <hongtao.liu@intel.com>
305 Sunil K Pandey <sunil.k.pandey@intel.com>
308 * config/i386/i386-passes.def: Add
309 pass_remove_partial_avx_dependency.
310 * config/i386/i386-protos.h
311 (make_pass_remove_partial_avx_dependency): New.
312 * config/i386/i386.c (make_pass_remove_partial_avx_dependency):
314 (pass_data_remove_partial_avx_dependency): New.
315 (pass_remove_partial_avx_dependency): Likewise.
316 (make_pass_remove_partial_avx_dependency): Likewise.
317 * config/i386/i386.md (avx_partial_xmm_update): New attribute.
318 (*extendsfdf2): Add avx_partial_xmm_update.
319 (truncdfsf2): Likewise.
320 (*float<SWI48:mode><MODEF:mode>2): Likewise.
321 (SF/DF conversion splitters): Disabled for TARGET_AVX.
323 2019-02-22 Aldy Hernandez <aldyh@redhat.com>
326 * gimple-ssa-sprintf.c (pass_sprintf_length::execute): Enable loop
329 2019-02-22 Thiago Macieira <thiago.macieira@intel.com>
332 * config/i386/i386.h (PTA_WESTMERE): Remove PTA_AES.
333 (PTA_SKYLAKE): Add PTA_AES.
334 (PTA_GOLDMONT): Likewise.
336 2019-02-22 Sudakshina Das <sudi.das@arm.com>
338 * config/aarch64/aarch64.c (aarch64_output_mi_thunk): Add bti
339 instruction if enabled.
340 (aarch64_override_options): Remove reference to return address key.
342 2019-02-22 Richard Biener <rguenther@suse.de>
344 PR tree-optimization/89440
345 * tree-vect-loop.c (vect_create_epilog_for_reduction): Remove
346 not necessary assert.
348 2019-02-22 Thomas Schwinge <thomas@codesourcery.com>
351 * omp-general.c (oacc_replace_fn_attrib): Mostly split out into...
352 (oacc_replace_fn_attrib_attr): ... this new function.
353 * omp-general.h (oacc_replace_fn_attrib_attr): New prototype.
354 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims_1): Revert workaround.
356 2019-02-22 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
358 * config/arm/arm-cpus.in (ares): Rename to...
359 (neoverse-n1): ... This. Add ares as alias.
360 * config/arm/arm-tables.opt: Regenerate.
361 * config/arm/arm-tune.md: Likewise.
362 * doc/invoke.txt (ARM Options): Document neoverse-n1.
364 2019-02-22 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
366 * config/aarch64/aarch64-cores.def (neoverse-e1): Define.
367 * config/aarch64/aarch64-tune.md: Regenerate.
368 * doc/invoke.texi (AArch64 Options): Document neoverse-e1 -mcpu option.
370 2019-02-22 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
372 * config/aarch64/aarch64.c (ares_tunings): Rename to...
373 (neoversen1_tunings): ... This.
374 * config/aarch64/aarch64-cores.def (ares): Change tuning to the above.
375 (neoverse-n1): New CPU.
376 * config/aarch64/aarch64-tune.md: Regenerate.
377 * doc/invoke.txt (AArch64 Options): Document neoverse-n1.
379 2019-02-22 Richard Biener <rguenther@suse.de>
382 * cfghooks.h (dependence_hash): New typedef.
383 (struct copy_bb_data): New type.
384 (cfg_hooks::duplicate_block): Adjust to take a copy_bb_data argument.
385 (duplicate_block): Likewise.
386 * cfghooks.c (duplicate_block): Pass down copy_bb_data.
387 (copy_bbs): Create and pass down copy_bb_data.
388 * cfgrtl.c (cfg_layout_duplicate_bb): Adjust.
389 (rtl_duplicate_bb): Likewise.
390 * tree-cfg.c (gimple_duplicate_bb): If the copy_bb_data arg is not NULL
391 remap dependence info.
393 2019-02-22 Richard Biener <rguenther@suse.de>
395 PR tree-optimization/87609
396 * tree-core.h (tree_base): Document special clique values.
397 * tree-inline.c (remap_dependence_clique): Do not use the
398 special clique value of one.
399 (maybe_set_dependence_info): Use clique one.
400 (clear_dependence_clique): New callback.
401 (compute_dependence_clique): Clear clique one from all refs
402 before assigning it (again).
404 2019-02-21 Martin Sebor <msebor@redhat.com>
406 * doc/extend.texi (__clear_cache): Correct signature.
408 2019-02-21 Ian Lance Taylor <iant@golang.org>
411 * varasm.c (decode_addr_const): Call lookup_constant_def rather
412 than output_constant_def.
413 (add_constant_to_table): New static function.
414 (output_constant_def): Call add_constant_to_table.
415 (tree_output_constant_def): Likewise.
417 2019-02-21 Jakub Jelinek <jakub@redhat.com>
420 * builtins.c (fold_builtin_arith_overflow): If first two args are
421 INTEGER_CSTs, set intres and ovfres to constants rather than calls
424 2019-02-21 H.J. Lu <hongjiu.lu@intel.com>
427 * config/i386/i386.c (ix86_set_indirect_branch_type): Issue an
428 error for -mindirect-branch/-mfunction-return with incompatible
431 2019-02-21 Jakub Jelinek <jakub@redhat.com>
434 * constraints.md (q): Remove.
435 * config/arm/ldrdstrd.md (*arm_ldrd, *arm_strd): Use rk constraint
438 2019-02-21 Martin Jambor <mjambor@suse.cz>
441 * omp-general.c (omp_extract_for_data): Removed a duplicate call
442 to omp_adjust_for_condition, moved NE_EXPR code_cond processing...
443 (omp_adjust_for_condition): ...here. Added necessary parameters.
444 * omp-general.h (omp_adjust_for_condition): Updated declaration.
445 * omp-grid.c (grid_attempt_target_gridification): Adjust to pass
446 proper values to new parameters of omp_adjust_for_condition.
448 2019-02-20 Jakub Jelinek <jakub@redhat.com>
451 * expr.c (expand_assignment): If result is a MEM, use change_address
452 instead of simplify_gen_subreg.
454 2019-02-20 Jakub Jelinek <jakub@redhat.com>
455 David Malcolm <dmalcolm@redhat.com>
458 * fold-const.c (decode_field_reference): Return NULL_TREE if
459 lang_hooks.types.type_for_size returns NULL. Check it before
460 overwriting *exp_. Use return NULL_TREE instead of return 0.
462 2019-02-20 Jakub Jelinek <jakub@redhat.com>
466 * toplev.c (do_compile): Double the emin/emax exponents to workaround
469 2019-02-20 Uroš Bizjak <ubizjak@gmail.com>
472 * config/i386/i386.c (ix86_atomic_assign_expand_fenv): Check
473 TARGET_SSE in addition to TARGET_SSE_MATH.
475 (ix86_excess_precision): Ditto.
476 (ix86_float_exceptions_rounding_supported_p): Ditto.
477 (use_rsqrt_p): Ditto.
478 * config/i386/sse.md (rsqrt<mode>2): Ditto.
480 2019-02-20 David Malcolm <dmalcolm@redhat.com>
483 * diagnostic-show-locus.c (layout::calculate_line_spans): Use
484 linenum_arith_t when determining if two adjacent line spans are
485 close enough to merge.
486 (diagnostic_show_locus): Use linenum_arith_t when iterating over
487 lines within each line_span.
489 2019-02-20 Andre Vieira <andre.simoesdiasvieira@arm.com>
492 * lra-constraints.c(uses_hard_regs_p): Fix handling of
495 2019-02-20 Li Jia He <helijia@linux.ibm.com>
498 * gcc/config/rs6000/rs6000.c (rs6000_gimple_fold_builtin)
499 <case ALTIVEC_BUILTIN_VSPLTISB, ALTIVEC_BUILTIN_VSPLTISH,
500 ALTIVEC_BUILTIN_VSPLTISW>: Don't convert the operand before
503 2019-02-19 Jonathan Wakely <jwakely@redhat.com>
505 * config/gcn/gcn.c (print_operand): Fix typo.
507 2019-02-19 Richard Biener <rguenther@suse.de>
510 * toplev.c (do_compile): Initialize mpfr's exponent range
511 based on available float modes.
513 2019-02-19 Eric Botcazou <ebotcazou@adacore.com>
515 * rtlanal.c (get_initial_register_offset): Fall back to the estimate
516 as long as the epilogue isn't completed.
518 2019-02-18 Martin Sebor <msebor@redhat.com>
520 * doc/cpp.texi (Conditional syntax): Add __has_attribute,
521 __has_cpp_attribute, and __has_include.
523 2019-02-18 Martin Sebor <msebor@redhat.com>
525 * doc/invoke.texi (-Wreturn-type): Correct and expand.
527 2019-02-18 Martin Sebor <msebor@redhat.com>
530 * tree.c (valid_constant_size_p): Avoid assuming size is a constant
532 * tree.h (cst_size_error): Add the cst_size_not_constant enumerator.
534 2019-02-18 Richard Biener <rguenther@suse.de>
536 PR tree-optimization/89296
537 * tree-ssa-loop-ch.c (ch_base::copy_headers): Restrict setting
538 of no-warning flag to cases that might emit the bogus warning.
540 2019-02-18 Jakub Jelinek <jakub@redhat.com>
543 * config/arm/arm.md (*arm_movdi, *movdf_soft_insn): Use "r" instead of
545 * config/arm/vfp.md (*movdi_vfp): Likewise.
546 * config/arm/ldrdstrd.md (*arm_ldrd, *arm_strd): Use "r" instead of
547 "q" constraint for operands[0].
550 * config/s390/s390.md (*r<noxa>sbg_<mode>_srl_bitmask,
551 *r<noxa>sbg_<mode>_sll, *r<noxa>sbg_<mode>_srl): Don't construct
552 pattern in a temporary buffer.
553 (*r<noxa>sbg_sidi_srl): Likewise. Always use 32 as I3 rather
557 * config/s390/s390.c (s390_indirect_branch_attrvalue,
558 s390_indirect_branch_settings): Define unconditionally.
559 (s390_set_current_function): Likewise, but guard the whole body except
560 the s390_indirect_branch_settings call with
561 #if S390_USE_TARGET_ATTRIBUTE.
562 (TARGET_SET_CURRENT_FUNCTION): Redefine unconditionally.
564 * config/s390/s390.md (*<risbg_n>_ior_and_sr_ze,
565 *<risbg_n>_<mode>_ior_and_lshiftrt, *<risbg_n>_sidi_ior_and_lshiftrt):
566 Use HOST_WIDE_INT_M1U instead of ~(0ULL).
567 (*<risbg_n>_and_subregdi_rotr, *<risbg_n>_and_subregdi_rotl): Use
568 HOST_WIDE_INT_1U instead of 1ULL.
569 (*pre_z10_extzv<mode>, *pre_z10_extv<mode>): Change mask type from int
570 to unsigned HOST_WIDE_INT, use HOST_WIDE_INT_1U instead of 1ul.
571 (*insv<mode><clobbercc_or_nocc>_appendbitsleft,
572 z = (x << c) | (y >> d) splitters): Use HOST_WIDE_INT_1U
574 (*insv<mode>_mem_reg, *insvdi_mem_reghigh): Use HOST_WIDE_INT_1U
577 2019-02-18 Martin Jambor <mjambor@suse.cz>
579 PR tree-optimization/89209
580 * tree-sra.c (create_access_replacement): New optional parameter
581 reg_tree. Use it as a type if non-NULL and access type is not of
583 (get_repl_default_def_ssa_name): New parameter REG_TYPE, pass it
584 to create_access_replacement.
585 (sra_modify_assign): Pass LHS type to get_repl_default_def_ssa_name.
586 Check lacc is non-NULL before attempting to re-create it on the RHS.
588 2019-02-18 Martin Liska <mliska@suse.cz>
591 * cgraph.c (symbol_table::create_edge): Set m_summary_id to -1
593 (symbol_table::free_edge): Recycle m_summary_id.
594 * cgraph.h (get_summary_id): New.
595 (symbol_table::release_symbol): Set m_summary_id to -1
597 (symbol_table::allocate_cgraph_symbol): Recycle m_summary_id.
598 * ipa-fnsummary.c (ipa_fn_summary_t): Switch from
599 function_summary to fast_function_summary.
600 * ipa-fnsummary.h (ipa_fn_summary_t): Likewise.
601 * ipa-pure-const.c (class funct_state_summary_t):
602 Switch from function_summary to fast_function_summary.
603 * ipa-reference.c (class ipa_ref_var_info_summary_t): Likewise.
604 (class ipa_ref_opt_summary_t): Switch from function_summary
605 to fast_function_summary.
606 * symbol-summary.h (class function_summary_base): New class
607 that is created from base of former function_summary.
608 (function_summary_base::unregister_hooks): New.
609 (class function_summary): Inherit from function_summary_base.
610 (class call_summary_base): New class
611 that is created from base of former call_summary.
612 (class call_summary): Inherit from call_summary_base.
613 (struct is_same): New.
614 (class fast_function_summary): New summary class.
615 (class fast_call_summary): New summary class.
616 * vec.h (vec_safe_grow_cleared): New function.
618 2019-02-18 Martin Liska <mliska@suse.cz>
620 * config/i386/i386.c (ix86_get_multilib_abi_name): New function.
621 (TARGET_GET_MULTILIB_ABI_NAME): New macro defined.
622 * doc/tm.texi: Document new target hook.
623 * doc/tm.texi.in: Likewise.
624 * target.def: Add new target macro.
625 * gcc.c (find_fortran_preinclude_file): Do not search multilib
628 2019-02-17 Alan Modra <amodra@gmail.com>
631 * config/rs6000/rs6000.md (<bd>_<mode> split): Check for an int
632 output reg on add insn.
633 (<bd>tf_<mode> split): Likewise. Match predicates with insn.
635 2019-02-16 H.J. Lu <hongjiu.lu@intel.com>
638 * config/i386/sse.md (ssedoublemode): Remove V4HI.
639 (PMULHRSW): Likewise.
640 (<ssse3_avx2>_pmulhrsw<mode>3): Require TARGET_SSSE3, not
642 (ssse3_pmulhrswv4hi3): New expander.
644 2019-02-16 H.J. Lu <hongjiu.lu@intel.com>
646 * config/i386/mmx.md (*vec_extractv2si_zext_mem): Doesn't require
647 MMX. Add isa attribute.
649 2019-02-16 Jakub Jelinek <jakub@redhat.com>
651 PR rtl-optimization/66152
652 * builtins.h (c_readstr): Declare.
653 * builtins.c (c_readstr): Remove forward declaration. Add
654 null_terminated_p argument, if false, read all bytes from the
655 string instead of stopping after '\0'.
656 * expr.c (string_cst_read_str): New function.
657 (store_expr): Use string_cst_read_str instead of
658 builtin_strncpy_read_str. Try to store by pieces the whole
659 exp_len first, and only if that fails, split it up into
660 store by pieces followed by clear_storage. Formatting fix.
662 * config/i386/i386.md (*movqi_internal): Remove static from
663 buf variable. Use output_asm_insn (buf, operands); return "";
664 instead of return buf;.
665 * config/i386/sse.md (<sse>_andnot<mode>3<mask_name>,
666 *<code><mode>3<mask_name>, *andnot<mode>3, *andnottf3, *<code><mode>3,
667 *<code>tf3, <mask_codefor><code><mode>3<mask_name>): Likewise.
669 2019-02-15 Eric Botcazou <ebotcazou@adacore.com>
671 * config/sparc/linux.h (ASAN_CC1_SPEC): Define.
672 (CC1_SPEC): Use GNU_USER_TARGET_CC1_SPEC and ASAN_CC1_SPEC.
673 * config/sparc/linux64.h (ASAN_CC1_SPEC): Likewise.
674 (CC1_SPEC): Likewise.
675 * config/sparc/sparc.c (sparc_asan_shadow_offset): Adjust for 64-bit.
677 2019-02-15 Eric Botcazou <ebotcazou@adacore.com>
679 * asan.c (asan_emit_stack_protection): Use full-sized mask to align
680 the base address on 64-bit strict-alignment platforms.
682 2019-02-15 H.J. Lu <hongjiu.lu@intel.com>
684 * config/i386/i386.h (VALID_MMX_REG_MODE): Correct the misplaced ')'.
686 2019-02-15 Uroš Bizjak <ubizjak@gmail.com>
688 * config/i386/darwin.h (TARGET_FPMATH_DEFAULT_P): New define.
690 2019-02-15 Aaron Sawdey <acsawdey@linux.ibm.com>
692 PR rtl-optimization/88308
693 * shrink-wrap.c (move_insn_for_shrink_wrap): Fix LABEL_NUSES counts
694 on copied instruction.
696 2019-02-15 Eric Botcazou <ebotcazou@adacore.com>
698 * final.c (insn_current_reference_address): Replace test on JUMP_P
699 with test on jump_to_label_p.
700 * config/visium/visium-passes.def: New file.
701 * config/visium/t-visium (PASSES_EXTRA): Define.
702 * config/visium/visium-protos.h (make_pass_visium_reorg): Declare.
703 * config/visium/visium.h (TRAMPOLINE_SIZE): Adjust.
704 (TRAMPOLINE_ALIGNMENT): Define.
705 * config/visium/visium.c (visium_option_override): Do not register
706 the machine-specific reorg pass here.
707 (visium_trampoline_init): Align the BRA insn on a 64-bit boundary
709 (output_branch): Adjust threshold for long branch instruction.
710 * config/visium/visium.md (cpu): Move around.
711 (length): Adjust for the GR6.
713 2019-02-15 Richard Biener <rguenther@suse.de>
714 Jakub Jelinek <jakub@redhat.com>
716 PR tree-optimization/89278
717 * tree-loop-distribution.c: Include tree-eh.h.
718 (generate_memset_builtin, generate_memcpy_builtin): Call
719 rewrite_to_non_trapping_overflow on builtin->size before passing it
720 to force_gimple_operand_gsi.
722 2019-02-15 Jakub Jelinek <jakub@redhat.com>
725 * optc-save-gen.awk: Handle optimize_fast like optimize_size or
727 * opth-gen.awk: Likewise.
729 2019-02-15 Uroš Bizjak <ubizjak@gmail.com>
731 * config/i386/i386.h (TARGET_SUBTARGET64_ISA_DEFAULT):
732 Enable MMX, SSE and SSE2 by default.
733 * config/i386/i386.c (ix86_option_override_internal): Do not
734 explicitly set MMX, SSE and SSE2 flags for TARGET_64BIT here.
736 2019-02-14 Jakub Jelinek <jakub@redhat.com>
738 PR rtl-optimization/89354
739 * combine.c (make_extraction): Punt if extraction_mode is narrower
742 2019-02-14 Maya Rashish <coypu@sdf.org>
744 * config.gcc (*-*-netbsd*): Add netbsd-d.o.
745 * config/netbsd-d.c: New file.
746 * config/t-netbsd: Add netbsd-d.o
748 2018-02-14 Steve Ellcey <sellcey@marvell.com>
750 * config/aarch64/aarch64.c (aarch64_attribute_table): Change
751 affects_type_identity to true for aarch64_vector_pcs.
752 (aarch64_comp_type_attributes): New function.
753 (TARGET_COMP_TYPE_ATTRIBUTES): New macro.
755 2019-02-14 Tamar Christina <tamar.christina@arm.com>
758 * config/arm/iterators.md (ANY64): Add V4HF.
760 2019-02-14 Martin Liska <mliska@suse.cz>
762 PR rtl-optimization/89242
763 * dce.c (delete_unmarked_insns): Call free_dominance_info we
764 process a transformation.
766 2019-02-14 Jakub Jelinek <jakub@redhat.com>
768 PR tree-optimization/89314
769 * fold-const.c (fold_binary_loc): Cast strlen argument to
770 const char * before dereferencing it. Formatting fixes.
773 * passes.def: Swap pass_ubsan and pass_early_warn_uninitialized.
775 2019-02-13 Ian Lance Taylor <iant@golang.org>
777 * optc-save-gen.awk: Set var_opt_hash for initial optimizations
778 and set current index for other optimizations.
780 2019-02-13 Uroš Bizjak <ubizjak@gmail.com>
782 * config/i386/sse.md (vec_set<VI4F_256_512:mode>_0): Use
783 nonimmediate_operand as operand 2 predicate.
784 (vec_set<VF2_512_256:mode>_0): Ditto.
785 (vec_set<VI8_AVX_AVX512F:mode>_0): Ditto.
786 (*vec_concatv2si): Remove alternative 2.
787 (*vec_concatv4si_0): Use vm constraint for alternative 0.
788 (*vec_concatv4si_0): Remove preferred_for_speed attribute.
789 (vec_concatv2di): Split alternatives 4,5,6 to ...
790 (*vec_concatv2di_0) ... new pattern.
792 2019-02-13 Wilco Dijkstra <wdijkstr@arm.com>
795 * config/arm/arm.c (ldm_stm_operation_p) Set
796 addr_reg_in_reglist correctly for first register.
797 (load_multiple_sequence): Remove dead base check.
798 (gen_ldm_seq): Correctly set write_back for Thumb-1.
800 2019-02-13 Tamar Christina <tamar.christina@arm.com>
803 * config/aarch64/aarch64-sve.md (*pred_mov<mode>, pred_mov<mode>):
804 Expose as @aarch64_pred_mov.
805 * config/aarch64/aarch64.c (aarch64_classify_address):
806 Use expand_insn which legitimizes operands.
808 2019-02-13 Martin Liska <mliska@suse.cz>
810 * builtins.h (expand_builtin_with_bounds): Remove declaration.
811 * calls.c (struct arg_data): Remove special_slot, pointer_arg
812 and pointer_offset fields.
813 (initialize_argument_information): Remove usage of dead
815 * cgraph.h (struct cgraph_thunk_info): Remove
816 add_pointer_bounds_args.
817 * cgraphunit.c (cgraph_node::expand_thunk): Remove usage of dead
819 (cgraph_node::assemble_thunks_and_aliases): Remove usage of dead
821 * config/i386/i386.c (ix86_function_arg_advance): Remove
823 (struct builtin_isa): Remove leaf_p and nothrow_p fields.
824 (def_builtin): Remove usage of dead
826 (ix86_add_new_builtins): Likewise.
827 * ipa-fnsummary.c (compute_fn_summary): Likewise.
828 * ipa-icf.c (sem_function::equals_wpa): Likewise.
829 (sem_function::init): Likewise.
830 (sem_variable::merge): Likewise.
831 * ipa-visibility.c (function_and_variable_visibility): Likewise.
832 * ipa.c (symbol_table::remove_unreachable_nodes): Likewise.
833 * lto-cgraph.c (lto_output_node): Likewise.
834 (lto_output_varpool_node): Likewise.
835 (input_node): Likewise.
836 (input_varpool_node): Likewise.
837 * lto-streamer-out.c (lto_output): Likewise.
838 * tree-inline.c (expand_call_inline): Remove usage of
840 * tree-inline.h (struct copy_body_data): Likewise.
841 * varpool.c (varpool_node::dump): Likewise.
843 2019-02-13 Jakub Jelinek <jakub@redhat.com>
846 * tree-ssa-structalias.c (set_uids_in_ptset): Or in vi->is_heap_var
847 into pt->vars_contains_escaped_heap instead of setting
848 pt->vars_contains_escaped_heap to it.
851 * optabs.c (prepare_cmp_insn): Use UINTVAL (size) instead of
852 INTVAL (size), compare it to GET_MODE_MASK instead of
853 1 << GET_MODE_BITSIZE.
856 * config/i386/predicates.md (x86_64_immediate_operand): Allow
857 TLS UNSPECs offsetted by signed 32-bit CONST_INT even with
860 2019-02-13 Martin Liska <mliska@suse.cz>
863 * cfgrtl.c (remove_barriers_from_footer): New function.
864 (try_redirect_by_replacing_jump): Use it.
865 (cfg_layout_redirect_edge_and_branch): Likewise.
867 2019-02-13 Xiong Hu Luo <luoxhu@linux.vnet.ibm.com>
869 * config/rs6000/altivec.h (vec_sbox_be, vec_cipher_be,
870 vec_cipherlast_be, vec_ncipher_be, vec_ncipherlast_be): New #defines.
871 * config/rs6000/crypto.md (CR_vqdi): New define_mode_iterator.
872 (crypto_vsbox_<mode>, crypto_<CR_insn>_<mode>): New define_insns.
873 * config/rs6000/rs6000-builtin.def (VSBOX_BE): New BU_CRYPTO_1.
874 (VCIPHER_BE, VCIPHERLAST_BE, VNCIPHER_BE, VNCIPHERLAST_BE):
876 * config/rs6000/rs6000.c (builtin_function_type)
877 <CRYPTO_BUILTIN_VSBOX_BE, CRYPTO_BUILTIN_VCIPHER_BE,
878 CRYPTO_BUILTIN_VCIPHERLAST_BE, CRYPTO_BUILTIN_VNCIPHER_BE,
879 CRYPTO_BUILTIN_VNCIPHERLAST_BE>: New switch options.
880 * doc/extend.texi (vec_sbox_be, vec_cipher_be, vec_cipherlast_be,
881 vec_ncipher_be, vec_ncipherlast_be): New builtin functions.
883 2019-02-12 Pat Haugen <pthaugen@us.ibm.com>
885 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove duplicate
886 -maltivec. Delete -maltivec=be and -maltivec=le documentation.
888 2019-02-12 H.J. Lu <hongjiu.lu@intel.com>
891 * config/i386/i386.md (*movoi_internal_avx): Revert revision
892 268678 and revision 268657.
893 (*movti_internal): Likewise.
895 2019-02-12 Ilya Leoshkevich <iii@linux.ibm.com>
898 * config/s390/s390.c (s390_decompose_address): Update comment.
899 (s390_check_qrst_address): Reject invalid address forms after
902 2019-02-12 Martin Liska <mliska@suse.cz>
905 * ipa-pure-const.c (propagate_pure_const): Revert hunk as
906 we need default values of funct_state for a function that
909 2019-02-12 Eric Botcazou <ebotcazou@adacore.com>
911 * asan.c (asan_expand_mark_ifn): Take into account the alignment of
912 the object to pick the size of stores on strict-alignment platforms.
914 * config/sparc/sparc.md (*movsi_insn): Minor tweak.
915 (*movdi_insn_sp32): Likewise.
916 (*movdi_insn_sp64): Likewise.
918 2019-02-12 Jan Hubicka <hubicka@ucw.cz>
921 * cgraphunit.c (analyze_functions): Clear READONLY flag for external
922 types that needs constructiong.
923 * tree.h (may_be_aliased): Do not check TYPE_NEEDS_CONSTRUCTING.
925 2019-02-12 Richard Biener <rguenther@suse.de>
927 PR tree-optimization/89253
928 * tree-ssa-loop-split.c (tree_ssa_split_loops): Check we can
931 2019-02-11 David Malcolm <dmalcolm@redhat.com>
934 * input.c (selftest::test_line_offset_overflow): New selftest.
935 (selftest::input_c_tests): Call it.
937 2019-02-11 Martin Sebor <msebor@redhat.com>
939 PR tree-optimization/88771
940 * gimple-ssa-warn-restrict.c (pass_wrestrict::gate): Also enable
941 when -Wstringop-overflow is set.
942 (builtin_memref::builtin_memref): Adjust excessive upper bound
943 only when lower bound is not excessive.
944 (maybe_diag_overlap): Detect and diagnose excessive bounds via
946 (maybe_diag_offset_bounds): Rename...
947 (maybe_diag_access_bounds): ...to this.
948 (check_bounds_or_overlap): Adjust for name change above.
950 2019-02-11 Martin Sebor <msebor@redhat.com>
953 * builtins.c (max_object_size): Move from here...
954 * builtins.h (max_object_size): ...and here...
955 * tree.c (max_object_size): ...to here...
956 * tree.h (max_object_size): ...and here.
958 2019-02-11 Bill Schmidt <wschmidt@linux.ibm.com>
960 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Shift-right
961 and shift-left vector built-ins need to include a TRUNC_MOD_EXPR
962 for correct semantics.
964 2019-02-11 Alan Modra <amodra@gmail.com>
966 * doc/invoke.texi (man page RS/6000 and PowerPC Options): Mention
967 -mlongcall and -mpltseq.
968 (RS/6000 and PowerPC Options <-mlongcall>): Mention inline PLT calls.
969 (RS/6000 and PowerPC Options <-mpltseq>): Document.
970 * config/rs6000/rs6000.h (TARGET_PLTSEQ): Define.
971 * config/rs6000/sysv4.opt (mpltseq): New option.
972 * config/rs6000/sysv4.h (TARGET_PLTSEQ): Redefine.
973 (SUBTARGET_OVERRIDE_OPTIONS): Error if given -mpltseq when assembler
974 support is lacking. Don't allow -mpltseq with -mbss-plt.
975 * config/rs6000/linux64.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Warn if
976 -mpltseq given for ELFv1.
977 * config/rs6000/rs6000.c (rs6000_call_aix): Comment on UNSPEC_PLTSEQ.
978 Only use UNSPEC_PLTSEQ for inline PLT calls.
979 (rs6000_call_sysv, rs6000_sibcall_sysv): Expand comments. Only
980 use UNSPEC_PLTSEQ for inline PLT calls.
981 (rs6000_indirect_call_template_1, rs6000_longcall_ref),
982 (rs6000_call_aix, rs6000_call_sysv, rs6000_sibcall_sysv): Replace
983 uses of HAVE_AS_PLTSEQ with TARGET_PLTSEQ, simplifying.
984 * config/rs6000/rs6000.md (pltseq_tocsave_<mode>),
985 (pltseq_plt16_ha_<mode>, pltseq_plt16_lo_<mode>),
986 (pltseq_mtctr_<mode>): Likewise.
988 2019-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
990 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Use 8-byte strings with
992 * configure: Regenerate.
994 2019-02-11 Jakub Jelinek <jakub@redhat.com>
997 * config/arm/ldrdstrd.md (*arm_ldrd, *arm_strd): Use q constraint
1000 2019-02-11 Hans-Peter Nilsson <hp@axis.com>
1002 * function.c (assign_parm_setup_block): Use the stored
1003 size, not the passed size, when allocating stack-space,
1004 also for a parameter with alignment larger than
1005 MAX_SUPPORTED_STACK_ALIGNMENT.
1007 2019-02-11 Martin Liska <mliska@suse.cz>
1010 * ipa-cp.c (build_toporder_info): Remove usage of a param.
1011 * ipa-inline.c (inline_small_functions): Likewise.
1012 * ipa-pure-const.c (propagate_pure_const): Likewise.
1013 (propagate_nothrow): Likewise.
1014 * ipa-reference.c (propagate): Likewise.
1015 * ipa-utils.c (struct searchc_env): Remove unused field.
1016 (searchc): Always search across AVAIL_INTERPOSABLE.
1017 (ipa_reduced_postorder): Always allow AVAIL_INTERPOSABLE as
1018 the only called IPA pure const can properly not propagate
1019 across interposable boundary.
1020 * ipa-utils.h (ipa_reduced_postorder): Remove param.
1022 2019-02-11 Chung-Ju Wu <jasonwucj@gmail.com>
1024 * config/nds32/nds32.md (call_internal, call_value_internal,
1025 sibcall_internal, sibcall_value_internal): Use SImode for mem rtx.
1027 2019-02-11 Hans-Peter Nilsson <hp@axis.com>
1029 * config/cris/cris.c (cris_preferred_minimum_alignment): Fix name
1032 2019-02-10 H.J. Lu <hongjiu.lu@intel.com>
1034 * config/i386/constraints.md (Yd): Replace AVX512BW with AVX512DQ
1037 2019-02-10 Chung-Ju Wu <jasonwucj@gmail.com>
1039 * config.gcc (with_nds32_lib): Set default --with-nds32-lib correctly.
1041 2019-02-10 Jakub Jelinek <jakub@redhat.com>
1043 PR tree-optimization/89268
1044 * tree-if-conv.c (version_loop_for_if_conversion): Push to preds only
1045 if preds is non-NULL.
1047 2019-02-09 Jan Hubicka <hubicka@ucw.cz>
1050 * tree.c (fld_simplified_type_name): Also keep TYPE_DECL for
1053 2019-02-10 Monk Chiang <sh.chiang04@gmail.com>
1055 * config/nds32/nds32.md (trap): New pattern.
1057 2019-02-10 Monk Chiang <sh.chiang04@gmail.com>
1059 * config/nds32/nds32.c (nds32_dwarf_register_span): Refine register
1062 2019-02-10 Chung-Ju Wu <jasonwucj@gmail.com>
1064 * config/nds32/nds32-md-auxiliary.c (nds32_spilt_doubleword): Support
1067 2019-02-09 Jan Hubicka <hubicka@ucw.cz>
1069 * ipa-visibility.c (localize_node): Also do not localize
1070 LDPR_PREVAILING_DEF_IRONLY_EXP.
1072 2019-02-09 Jan Hubicka <hubicka@ucw.cz>
1075 * tree.c (fld_simplified_type_name): Use DECL_ASSEMBLER_NAME_SET_P
1076 instead of type_with_linkage.
1078 2019-02-09 Jan Hubicka <hubicka@ucw.cz>
1081 * params.def (uninlined-function-insns, uninlined-function-time,
1082 uninlined-thunk-insns, uninlined-thunk-time): Add artificial upper
1083 bound so we don't get overflows.
1085 2019-02-09 Aaron Sawdey <acsawdey@linux.ibm.com>
1087 * config/rs6000/rs6000-string.c (expand_compare_loop,
1088 expand_block_compare): Insert REG_BR_PROB notes in inline expansion of
1091 2019-02-09 Jakub Jelinek <jakub@redhat.com>
1094 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
1095 If !node->definition and TYPE_ARG_TYPES is non-NULL, use
1096 TYPE_ARG_TYPES instead of DECL_ARGUMENTS.
1098 2019-02-09 Alan Modra <amodra@gmail.com>
1101 * config/rs6000/rs6000.c (save_reg_p): Correct calls_eh_return
1102 case. Match logic in rs6000_emit_prologue emitting pic_offset_table
1105 2019-02-08 Vladimir Makarov <vmakarov@redhat.com>
1108 * lra-constraints.c (process_alt_operands): Don't increase reject
1109 for memory when offset memory is required.
1111 2019-02-08 Robin Dapp <rdapp@linux.ibm.com>
1113 * config/s390/vector.md: Implement vector copysign.
1115 2019-02-08 H.J. Lu <hongjiu.lu@intel.com>
1117 * expr.c (expand_constructor): Correct indentations.
1119 2019-02-08 Richard Biener <rguenther@suse.de>
1121 PR tree-optimization/89247
1122 * tree-if-conv.c: Include tree-cfgcleanup.h.
1123 (version_loop_for_if_conversion): Record LOOP_VECTORIZED call.
1124 (tree_if_conversion): Pass through predicate vector.
1125 (pass_if_conversion::execute): Do CFG cleanup and SSA update
1126 inline, see if any if-converted loops we refrece in
1127 LOOP_VECTORIZED calls vanished and fixup.
1128 * tree-if-conv.h (tree_if_conversion): Adjust prototype.
1130 2019-02-08 Ilya Leoshkevich <iii@linux.ibm.com>
1132 * config/s390/constraints.md (jdd): New constraint.
1134 2019-02-08 H.J. Lu <hongjiu.lu@intel.com>
1137 * config/i386/i386.md (*movoi_internal_avx): Set mode to XI for
1138 upper 16 vector registers without TARGET_AVX512VL.
1139 (*movti_internal): Likewise.
1141 2019-02-08 Jakub Jelinek <jakub@redhat.com>
1143 PR rtl-optimization/89234
1144 * except.c (copy_reg_eh_region_note_forward): Return if note_or_insn
1145 is a NOTE, CODE_LABEL etc. - rtx_insn * other than INSN_P.
1146 (copy_reg_eh_region_note_backward): Likewise.
1148 2019-02-08 Richard Biener <rguenther@suse.de>
1151 * tree-data-ref.c (initialize_matrix_A): Fail if constant
1153 (analyze_subscript_affine_affine): Handle failure from
1154 initialize_matrix_A.
1156 2019-02-08 Jakub Jelinek <jakub@redhat.com>
1158 * cfganal.c (pre_and_rev_post_order_compute_fn): Use fn instead of
1161 2019-02-07 David Malcolm <dmalcolm@redhat.com>
1163 PR tree-optimization/86637
1164 PR tree-optimization/89235
1165 * tree-vect-loop.c (optimize_mask_stores): Add an
1166 auto_purge_vect_location sentinel to ensure that vect_location is
1169 (auto_purge_vect_location::~auto_purge_vect_location): New dtor.
1170 (try_vectorize_loop_1): Add an auto_purge_vect_location sentinel
1171 to ensure that vect_location is purged on exit.
1172 (pass_slp_vectorize::execute): Likewise, replacing the manual
1174 * tree-vectorizer.h (class auto_purge_vect_location): New class.
1176 2019-02-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1178 * config/aarch64/iterators.md (max_opp): New code_attr.
1179 (USMAX): New code iterator.
1180 * config/aarch64/predicates.md (aarch64_smin): New predicate.
1181 (aarch64_smax): Likewise.
1182 * config/aarch64/aarch64-simd.md (abd<mode>_3): Rename to...
1183 (*aarch64_<su>abd<mode>_3): ... Change RTL representation to
1186 2019-02-07 H.J. Lu <hongjiu.lu@intel.com>
1189 * config/i386/i386.md (*movoi_internal_avx): Set mode to OI
1190 for TARGET_AVX512VL.
1191 (*movti_internal): Set mode to TI for TARGET_AVX512VL.
1193 2019-02-07 Andreas Krebbel <krebbel@linux.ibm.com>
1195 * config/s390/s390-builtin-types.def: Add new types.
1196 * config/s390/s390-builtins.def: (s390_vec_xl, s390_vec_xld2)
1197 (s390_vec_xlw4): Make the memory operand into a const pointer.
1198 (s390_vec_xld2, s390_vec_xlw4): Add a variant for single precision
1200 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Generate
1201 a new vector type with the alignment of the scalar memory operand.
1203 2019-02-07 Matthew Malcomson <matthew.malcomson@arm.com>
1204 Jakub Jelinek <jakub@redhat.com>
1207 * config/arm/arm-protos.h (valid_operands_ldrd_strd,
1208 arm_count_ldrdstrd_insns): New declarations.
1209 * config/arm/arm.c (mem_ok_for_ldrd_strd): Remove broken handling of
1211 (valid_operands_ldrd_strd): New function.
1212 (arm_count_ldrdstrd_insns): New function.
1213 * config/arm/ldrdstrd.md: Change peepholes to generate PARALLEL SImode
1214 sets instead of single DImode set and define new insns to match this.
1216 2019-02-07 Tamar Christina <tamar.christina@arm.com>
1218 * config/aarch64/aarch64-builtins.c (aarch64_fcmla_lane_builtin_data):
1219 Make it a C initializer.
1221 2019-02-07 Tamar Christina <tamar.christina@arm.com>
1224 * config/arm/neon.md (*neon_mov<mode>): Add r -> r case.
1226 2019-02-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1228 * config/arm/neon.md (neon_<sup>dot<vsi2qi>):
1229 Use neon_dot<q> for type.
1230 (neon_<sup>dot_lane<vsi2qi>): Likewise.
1232 2019-02-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1234 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>):
1235 Use neon_dot<q> for type.
1236 (aarch64_<sur>dot_lane<vsi2qi>): Likewise.
1237 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
1239 2019-02-06 Vladimir Makarov <vmakarov@redhat.com>
1241 PR rtl-optimization/89225
1242 * lra-constaints.c (simplify_operand_subreg): Add subreg mode
1245 2019-02-06 Eric Botcazou <ebotcazou@adacore.com>
1247 * config/i386/i386.c (ix86_expand_prologue): Emit a memory blockage
1248 after restoring registers saved to allocate the frame on Windows.
1250 2019-02-06 Richard Biener <rguenther@suse.de>
1252 PR tree-optimization/89182
1253 * graphite.h (cached_scalar_evolution_in_region): Declare.
1254 * graphite.c (struct seir_cache_key): New.
1255 (struct sese_scev_hash): Likewise.
1256 (seir_cache): New global.
1257 (cached_scalar_evolution_in_region): New function.
1258 (graphite_transform_loops): Allocate and release seir_cache.
1259 * graphite-isl-ast-to-gimple.c (get_rename_from_scev): Use
1260 cached_scalar_evolution_in_region.
1261 * graphite-scop-detection.c (scop_detection::can_represent_loop):
1263 (scop_detection::graphite_can_represent_expr: Use
1264 cached_scalar_evolution_in_region.
1265 (scop_detection::stmt_simple_for_scop_p): Likewise.
1266 (find_params_in_bb): Likewise.
1267 (gather_bbs::before_dom_children): Likewise.
1268 * graphite-sese-to-poly.c (create_pw_aff_from_tree): Likewise.
1269 (add_loop_constraints): Likewise.
1271 2019-02-06 Jakub Jelinek <jakub@redhat.com>
1274 * fold-const-call.c (fold_const_vec_convert): Pass true as last
1275 operand to new_unary_operation only if both element types are integral
1276 and it isn't a widening conversion. Return NULL_TREE if
1277 new_unary_operation failed.
1279 2019-02-05 Andreas Krebbel <krebbel@linux.ibm.com>
1282 * config/s390/s390.md: Remove load and test FP splitter.
1284 2019-02-05 Aaron Sawdey <acsawdey@linux.ibm.com>
1287 * config/rs6000/rs6000-string.c (do_ifelse, expand_cmp_vec_sequence,
1288 expand_compare_loop, expand_block_compare_gpr,
1289 expand_strncmp_align_check, expand_strncmp_gpr_sequence): Insert
1290 REG_BR_PROB notes in inline expansion of memcmp/strncmp. Add
1291 #include "profile-count.h" and "predict.h" for types and functions
1292 needed to work with REG_BR_PROB notes.
1294 2019-02-05 Aaron Sawdey <acsawdey@linux.ibm.com>
1297 * config/rs6000/rs6000.md (<bd>tf_<mode>): Generate a local label
1298 for the long branch case.
1300 2019-02-05 Jakub Jelinek <jakub@redhat.com>
1303 * dce.c (delete_unmarked_insns): Don't remove no-op moves if they
1304 can throw, non-call exceptions are enabled and we can't delete
1305 dead exceptions or alter cfg. Set must_clean if
1306 delete_insn_and_edges returns true, don't set it blindly for calls.
1307 Assert that delete_unreachable_blocks is called only if can_alter_cfg.
1309 PR rtl-optimization/89195
1310 * combine.c (make_extraction): For MEMs, don't extract bytes outside
1311 of the original MEM.
1313 2019-02-05 Martin Liska <mliska@suse.cz>
1315 PR gcov-profile/89000
1316 * gcov.c (function_summary): Remove argument.
1317 (file_summary): New function.
1318 (print_usage): Replace tabs with spaces.
1319 (generate_results): Use new function file_summary.
1321 2019-02-05 Jakub Jelinek <jakub@redhat.com>
1324 * optabs.c (prepare_cmp_insn): Pass x and y to
1325 emit_block_comp_via_libcall rather than XEXP (x, 0) and XEXP (y, 0).
1327 2019-02-05 Richard Biener <rguenther@suse.de>
1330 * bitmap.h (struct bitmap_obstack): Do not mark GTY.
1331 (struct bitmap_element): Drop chain_prev so we properly recurse on
1332 the prev member, supporting tree views.
1333 (struct bitmap_head): GTY skip the obstack member.
1335 2019-02-04 Alexander Monakov <amonakov@ispras.ru>
1338 * doc/extend.texi (Vector Extensions): Add an example of using vector
1339 types together with x86 intrinsics.
1341 2019-02-04 Alan Modra <amodra@gmail.com>
1343 * config/rs6000/rs6000.c (rs6000_indirect_call_template_1): Increase
1344 str[] size to 160, and comment.
1346 2019-02-04 Alan Modra <amodra@gmail.com>
1348 * config/rs6000/rs6000.c (rs6000_indirect_call_template_1),
1349 (rs6000_pltseq_template): Guard output of TLS markers with
1351 (rs6000_longcall_ref, rs6000_call_aix, rs6000_call_sysv),
1352 (rs6000_sibcall_sysv): Ignore TARGET_TLS_MARKERS when deciding
1353 to use inline PLT sequences.
1354 * config/rs6000/rs6000.md (pltseq_tocsave_<mode>),
1355 (pltseq_plt16_ha_<mode>, pltseq_plt16_lo_<mode>),
1356 (pltseq_mtctr_<mode>): Don't test TARGET_TLS_MARKERS in predicate.
1358 2019-02-04 Martin Liska <mliska@suse.cz>
1361 * ipa-fnsummary.c (estimate_edge_devirt_benefit): Bail
1362 out when ipa_fn_summaries does not contain entry for callee.
1364 2019-02-04 Eric Botcazou <ebotcazou@adacore.com>
1366 * config/sparc/sparc.h: Remove superfluous blank lines.
1367 * config/sparc/sparc.c (global_offset_table_rtx): Rename into...
1368 (got_register_rtx): ...this.
1369 (sparc_got): Adjust to above renaming.
1370 (sparc_tls_got): Likewise.
1371 (sparc_delegitimize_address): Likewise.
1372 (sparc_output_mi_thunk): Likewise.
1373 (sparc_init_pic_reg): Likewise.
1374 (save_local_or_in_reg_p): Fix test on the GOT register.
1375 (USE_HIDDEN_LINKONCE): Move around.
1376 (get_pc_thunk_name): Likewise.
1377 (gen_load_pcrel_sym): Likewise.
1378 (load_got_register): Likewise.
1380 2019-02-04 Kito Cheng <kito.cheng@gmail.com>
1382 * config/nds32/linux.h (GLIBC_DYNAMIC_LINKER): Define the naming rule
1383 of the dynamic linker: "ld-linux-nds32[le|be][f].so.1".
1385 2019-02-04 Chung-Ju Wu <jasonwucj@gmail.com>
1387 * config/nds32/nds32.c (nds32_legitimate_address_p): Add TLS model
1390 2019-02-04 Chung-Ju Wu <jasonwucj@gmail.com>
1392 * config.gcc (with_nds32_lib, glibc):
1393 Remove TARGET_DEFAULT_TLSDESC_TRAMPOLINE=0 setting.
1394 * config/nds32/linux.h (TARGET_DEFAULT_TLSDESC_TRAMPOLINE): Delete.
1395 (NDS32_TLSDESC_TRAMPOLINE_SPEC): Delete.
1397 2019-02-03 Uroš Bizjak <ubizjak@gmail.com>
1400 * config/i386/i386.md (*sqrt<mode>2_sse): Add (v,0) alternative.
1401 Do not prefer (v,v) alternative for non-AVX targets and (m,v)
1402 alternative for speed when TARGET_SSE_PARTIAL_REG_DEPENDENCY is set.
1403 (*rcpsf2_sse): Ditto.
1404 (*rsqrtsf2_sse): Ditto.
1405 (sse4_1_round<mode<2): Ditto.
1407 2019-02-03 Richard Biener <rguenther@suse.de>
1410 * dwarf2out.c (copy_ancestor_tree): Register non-stubs as
1413 2019-02-02 Jakub Jelinek <jakub@redhat.com>
1416 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
1417 Punt with warning on aggregate return or argument types. Ignore
1418 type/mode checking for uniform arguments.
1420 2019-02-01 Segher Boessenkool <segher@kernel.crashing.org>
1422 * combine.c (try_combine): Do not print "Can't combine" messages unless
1423 printing failed combination attempts.
1425 2019-02-01 Martin Jambor <mjambor@suse.cz>
1428 * omp-grid.c (grid_mark_variable_segment): Set assembler name of group
1429 segment and global segment variables before making them static.
1431 2019-02-01 Martin Jambor <mjambor@suse.cz>
1433 * omp-grid.c (grid_target_follows_gridifiable_pattern): Guard two
1434 missed optimization dump with dump_enabled_p.
1436 2019-02-01 Richard Biener <rguenther@suse.de>
1439 * tree-scalar-evolution.c (analyze_scalar_evolution): Set up
1440 the instantiate cache.
1441 (instantiate_scev_binary): Elide second operand procesing
1442 if equal to the first.
1443 * tree-chrec.c (chrec_contains_symbols): Add visited set.
1444 (chrec_contains_undetermined): Likewise.
1445 (tree_contains_chrecs): Likewise.
1447 2019-02-01 Jan Hubicka <hubicka@ucw.cz>
1449 * parms.def (MAX_INLINE_INSNS_SINGLE): Reduce from 400 to 200.
1451 2019-02-01 Jakub Jelinek <jakub@redhat.com>
1453 PR tree-optimization/89143
1454 * wide-int-range.h (wide_int_range_absu): Declare.
1455 * wide-int-range.cc (wide_int_range_absu): New function.
1456 * tree-vrp.c (extract_range_from_unary_expr): Handle ABSU_EXPR.
1458 PR tree-optimization/88107
1459 * tree-cfg.c (find_outermost_region_in_block): Add ALL argument,
1460 instead of assertion that eh_region_outermost is non-NULL, if it
1461 is NULL, set *ALL to true and return NULL.
1462 (move_sese_region_to_fn): Adjust caller, if all is set, call
1463 duplicate_eh_regions with NULL region.
1465 2019-02-01 Richard Biener <rguenth@suse.de>
1467 PR rtl-optimization/88593
1468 * mode-switching.c (optimize_mode_switching): Free dominators before
1469 calling cleanup_cfg.
1471 2019-02-01 Bin Cheng <bin.cheng@linux.alibaba.com>
1473 PR tree-optimization/88932
1474 * tree-predcom.c (try_combine_chains): Get loop bbs in dom order.
1476 2019-01-31 Jakub Jelinek <jakub@redhat.com>
1479 * omp-low.c (lower_omp_task_reductions): Drop redundant test to avoid
1480 bogus clang warning.
1482 2019-01-31 Uroš Bizjak <ubizjak@gmail.com>
1485 * config/i386/i386.md (*extendsfdf2): Split out reg->reg
1486 alternative to avoid partial SSE register stall for TARGET_AVX.
1487 (truncdfsf2): Ditto.
1488 (sse4_1_round<mode>2): Ditto.
1490 2018-01-31 Bill Schmidt <wschmidt@linux.ibm.com>
1492 PR tree-optimization/89008
1493 * gimple-ssa-strength-reduction.c (slsr_process_mul): Don't
1494 process anything of the form X * 0.
1496 2019-01-31 Richard Biener <rguenther@suse.de>
1498 PR tree-optimization/89135
1499 * tree-ssa-phiprop.c (pass_phiprop::execute): Skip blocks
1500 with abnormal preds.
1502 2019-01-31 Jakub Jelinek <jakub@redhat.com>
1505 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Allow inlining
1506 always_inline callees into no_sanitize_address callers.
1508 2019-01-31 Richard Biener <rguenther@suse.de>
1510 PR rtl-optimization/89115
1511 * lra.c (lra_rtx_hash): Properly hash CONST_INT values.
1513 2019-01-30 Martin Sebor <msebor@redhat.com>
1516 * doc/extend.texi (cast to a union): Correct and expand.
1518 2019-01-30 Vladimir Makarov <vmakarov@redhat.com>
1520 PR rtl-optimization/87246
1521 * lra-constraints.c (simplify_operand_subreg): Reload memory
1522 in subreg if the address became invalid.
1524 2019-01-30 Bill Schmidt <wschmidt@linux.ibm.com>
1527 * config/rs6000/vsx.md (*vsx_reduc_<VEC_reduc_name>_v4sf_scalar):
1528 Disable for little-endian.
1530 2019-01-30 Richard Biener <rguenther@suse.de>
1532 PR rtl-optimization/89115
1533 * opts.c (default_options_optimization): Reduce
1534 PARAM_MAX_DSE_ACTIVE_LOCAL_STORES by a factor of 10 at -O1.
1535 Make PARAM_LOOP_INVARIANT_MAX_BBS_IN_LOOP reduction relative
1538 2019-01-30 Kelvin Nilsen <kelvin@gcc.gnu.org>
1540 * config/rs6000/rs6000-c.c (altivec-resolve_overloaded_builtin):
1541 Change handling of ALTIVEC_BUILTIN_VEC_EXTRACT. Coerce result to
1542 type of vector element when vec_extract is implemented by direct
1545 2019-01-30 Thomas Schwinge <thomas@codesourcery.com>
1547 * doc/invoke.texi (C Language Options): List "-fopenacc-dim".
1549 2019-01-30 Richard Biener <rguenther@suse.de>
1551 PR tree-optimization/89111
1552 * tree-ssa-loop-im.c (gather_mem_refs_stmt): Restrict
1553 canonicalization to appropriately sized access types.
1555 2019-01-30 Jakub Jelinek <jakub@redhat.com>
1558 * config/i386/i386.c (ix86_warn_parameter_passing_abi): Don't warn
1559 for arguments to functions that are TU-local and shouldn't be
1560 referenced by assembly.
1562 2019-01-30 Ulrich Drepper <drepper@redhat.com>
1564 * dumpfile.c (opt_info_switch_p_1): Ignore '-' if it appears
1567 2019-01-29 Martin Sebor <msebor@redhat.com>
1570 * gimple-fold.c (fold_array_ctor_reference): Avoid zero-length arrays.
1572 2019-01-29 Jakub Jelinek <jakub@redhat.com>
1576 * omp-simd-clone.c (simd_clone_clauses_extract)
1577 <case OMP_CLAUSE_ALIGNED>: Ignore clauses with NULL
1578 OMP_CLAUSE_ALIGNED_ALIGNMENT.
1580 2019-01-29 Vineet Gupta <vgupta@synopsys.com>
1582 * config.gcc: Force .init_array for ARC.
1584 2019-01-29 Richard Biener <rguenther@suse.de>
1587 * dwarf2out.c (collect_skeleton_dies): New helper.
1588 (copy_decls_for_unworthy_types): Call it.
1589 (build_abbrev_table): Assert we do not try to replace
1590 DW_AT_signature refs with local refs.
1592 2019-01-28 Jakub Jelinek <jakub@redhat.com>
1595 * gimplify.c (gimplify_omp_for): When adding OMP_CLAUSE_*_GIMPLE_SEQ
1596 for lastprivate/linear IV, push gimplify context around gimplify_assign
1597 and, if it needed any temporaries, pop it into a gimple bind around the
1600 2019-01-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
1602 * common.opt (-Wattribute-alias): Remove "no-" from name.
1603 Make -Wattribute-alias command line option and
1604 #pragma GCC diagnostic ignored "-Wattribute-alias" work again.
1606 2019-01-28 Jakub Jelinek <jakub@redhat.com>
1609 * doc/invoke.texi (-mclwb, -mprfchw, -mrdpid, -mrdseed, -msgx,
1610 -madx, -mhle, -mavx5124fmaps, -mavx512vnni, -mavx5124vnniw): Document
1612 (bmi2): Add missing @opindex.
1613 * doc/extend.texi (x86 target attribute): Move fma4, lwp, ssse3
1614 options alphabetically. Add missing 3dnow, 3dnowa, adx, avx, avx2,
1615 avx5124fmaps, avx5124vnniw, avx512bitalg, avx512bw, avx512cd,
1616 avx512dq, avx512er, avx512f, avx512ifma, avx512pf, avx512vbmi,
1617 avx512vbmi2, avx512vl, avx512vnni, avx512vpopcntdq, bmi, bmi2,
1618 cldemote, clflushopt, clwb, clzero, crc32, cx16, f16c, fma, fsgsbase,
1619 fxsr, gfni, hle, lzcnt, movbe, movdir64b, movdiri, mwaitx, pconfig,
1620 pku, prefetchwt1, prfchw, ptwrite, rdpid, rdrnd, rdseed, rtm, sahf,
1621 sgx, sha, shstk, tbm, vaes, vpclmulqdq, waitpkg, wbnoinvd, xsave,
1622 xsavec, xsaveopt and xsaves options.
1624 2019-01-28 Richard Biener <rguenther@suse.de>
1627 * dwarf2out.c (gen_subprogram_die): Remove leftover from MPX
1630 2019-01-28 Richard Biener <rguenther@suse.de>
1632 PR tree-optimization/88739
1633 * tree-cfg.c (verify_types_in_gimple_reference): Verify
1634 BIT_FIELD_REFs only are applied to mode-precision operands
1635 when they are integral.
1636 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR.
1637 * tree-ssa-sccvn.c (vn_reference_lookup_3): Avoid generating
1638 BIT_FIELD_REFs of non-mode-precision integral operands.
1640 2019-01-27 Jakub Jelinek <jakub@redhat.com>
1643 * config/i386/sse.md
1644 (<mask_codefor>avx512dq_shuf_<shuffletype>64x2_1<mask_name>,
1645 avx512f_shuf_<shuffletype>64x2_1<mask_name>): Ensure the
1646 first constants in pairs are multiples of 2. Formatting fixes.
1647 (avx512vl_shuf_<shuffletype>32x4_1<mask_name>,
1648 avx512vl_shuf_<shuffletype>32x4_1<mask_name>): Ensure the
1649 first constants in each quadruple are multiples of 4. Formatting fixes.
1651 2019-01-26 Martin Jambor <mjambor@suse.cz>
1654 * tree-inline.c: Include tree-cfgcleanup.h.
1655 (delete_unreachable_blocks_update_callgraph): Move...
1656 * tree-cfgcleanup.c (delete_unreachable_blocks_update_callgraph):
1657 ...here, make externally visible, make second argument bool, adjust
1659 * tree-cfgcleanup.c: Include cgraph.h.
1660 * tree-cfgcleanup.h (delete_unreachable_blocks_update_callgraph):
1662 * ipa-prop.c: Include tree-cfgcleanup.h.
1663 (ipcp_transform_function): Call
1664 delete_unreachable_blocks_update_callgraph instead of cleaning uo CFG.
1666 2019-01-25 Vladimir Makarov <vmakarov@redhat.com>
1668 PR rtl-optimization/88846
1669 * ira.c (process_set_for_memref_referenced_p): New.
1670 (memref_referenced_p): Add new param. Use
1671 process_set_for_memref_referenced_p. Add new switch cases.
1672 (memref_used_between_p): Pass new arg to memref_referenced_p.
1674 2019-01-25 Richard Earnshaw <rearnsha@arm.com>
1677 * config/aarch64/aarch64.c (aarch64_function_arg_alignment): Add new
1678 argument ABI_BREAK. Set to true if the calculated alignment has
1679 changed in gcc-9. Check bit-fields for their base type alignment.
1680 (aarch64_layout_arg): Warn if argument passing has changed in gcc-9.
1681 (aarch64_function_arg_boundary): Likewise.
1682 (aarch64_gimplify_va_arg_expr): Likewise.
1684 2019-01-25 Richard Sandiford <richard.sandiford@arm.com>
1687 * varasm.c (output_constructor_bitfield): Use wi::extract_uhwi
1688 instead of accessing TREE_INT_CST_ELT directly.
1690 2019-01-25 Christophe Lyon <christophe.lyon@linaro.org>
1692 * doc/sourcebuild.texi (Environment attributes): Add fenv and
1693 fenv_exceptions description.
1695 2019-01-25 Wilco Dijkstra <wdijkstr@arm.com>
1697 PR rtl-optimization/87763
1698 * config/aarch64/aarch64.c (aarch64_select_cc_mode):
1699 Allow SUBREG when matching CC_NZmode compare.
1701 2019-01-25 Richard Biener <rguenther@suse.de>
1703 PR tree-optimization/89049
1704 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost):
1705 Look at the pattern stmt to determine if the stmt is vectorized.
1707 2019-01-25 Richard Sandiford <richard.sandiford@arm.com>
1709 * config/aarch64/aarch64-sve.md (*pred_mov<mode>)
1710 (pred_mov<mode>): Handle all-register forms using both a new
1711 alternative and a split.
1713 2019-01-25 Richard Biener <rguenther@suse.de>
1715 PR tree-optimization/86865
1716 * graphite-scop-detection.c (scop_detection::can_represent_loop):
1717 Reject non-do-while loops.
1719 2019-01-24 Peter Bergner <bergner@linux.ibm.com>
1721 * config/rs6000/altivec.md (build_vector_mask_for_load): Use MEM_P.
1722 * config/rs6000/constraints.md (Q constraint): Use REG_P.
1723 * config/rs6000/darwin.h (PREFERRED_RELOAD_CLASS): Use SYMBOL_REF_P.
1724 * config/rs6000/freebsd64.h (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Use
1725 SYMBOL_REF_P, CONST_INT_P and CONST_DOUBLE_P.
1726 * config/rs6000/linux64.h (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
1727 * config/rs6000/predicates.md (altivec_register_operand, vint_operand,
1728 vsx_register_operand, vsx_reg_sfsubreg_ok, vfloat_operand,
1729 vlogical_operand, gpc_reg_operand, int_reg_operand,
1730 int_reg_operand_not_pseudo): Use SUBREG_P and HARD_REGISTER_P.
1731 (ca_operand, base_reg_operand, htm_spr_reg_operand, cc_reg_operand,
1732 cc_reg_not_cr0_operand, input_operand): Use SUBREG_P.
1733 (save_world_operation, restore_world_operation, lmw_operation,
1734 stmw_operation): Use MEM_P and REG_P.
1735 (tie_operand): Use MEM_P.
1736 (vrsave_operation, crsave_operation): Use REG_P.
1737 (mfcr_operation, mtcrf_operation): Use REG_P and CONST_INT_P.
1738 (fpr_reg_operand): Use SUBREG_P and HARD_REGISTER_NUM_P.
1739 (quad_int_reg_operand): Use HARD_REGISTER_NUM_P.
1740 (call_operand): Use HARD_REGISTER_P.
1741 (indexed_or_indirect_operand, altivec_indexed_or_indirect_operand):
1743 (lwa_operand): Use SUBREG_P, REG_P and CONST_INT_P.
1744 * config/rs6000/rs6000-p8swap.c (insn_is_load_p, insn_is_store_p,
1745 quad_aligned_load_p, replace_swapped_aligned_store,
1746 recombine_lvx_pattern, replace_swapped_aligned_load,
1747 recombine_stvx_pattern): Use MEM_P.
1748 (const_load_sequence_p, adjust_vperm, replace_swapped_load_constant):
1749 Use MEM_P and SYMBOL_REF_P.
1750 (rtx_is_swappable_p): Use REG_P and CONST_INT_P.
1751 (insn_is_swappable_p): Use REG_P and MEM_P.
1752 (insn_is_swap_p, (alignment_mask): Use CONST_INT_P.
1753 * config/rs6000/rs6000-string.c (expand_block_clear, expand_block_move):
1755 * config/rs6000/rs6000.c (rs6000_secondary_reload, rs6000_emit_cmove):
1757 (rs6000_output_move_128bit): Use CONST_DOUBLE_P, CONST_INT_P and
1759 (rs6000_legitimize_address): Use CONST_DOUBLE_P, CONST_INT_P,
1760 CONST_WIDE_INT_P, REG_P and SYMBOL_REF_P.
1761 (rs6000_emit_move): Use CONST_DOUBLE_P, CONST_INT_P, HARD_REGISTER_P,
1762 HARD_REGISTER_NUM_P, MEM_P, REG_P, SUBREG_P, SYMBOL_REF_P and
1764 (output_toc): Use CONST_DOUBLE_P, CONST_INT_P and SYMBOL_REF_P.
1765 (easy_altivec_constant, rs6000_legitimate_offset_address_p,
1766 rs6000_mode_dependent_address, rs6000_expand_mtfsf_builtin,
1767 rs6000_expand_set_fpscr_rn_builtin, rs6000_expand_set_fpscr_drn_builtin,
1768 rs6000_expand_unop_builtin, INT_P, rs6000_generate_compare,
1769 rs6000_machopic_legitimize_pic_address, rs6000_split_logical_inner,
1770 rs6000_split_logical_di): Use CONST_INT_P.
1771 (rs6000_legitimize_reload_address): Use CONST_INT_P, HARD_REGISTER_P,
1772 REG_P and SYMBOL_REF_P.
1773 (setup_incoming_varargs, rs6000_rtx_costs): Use CONST_INT_P and MEM_P.
1774 (print_operand): Use CONST_INT_P, MEM_P and REG_P.
1775 (virtual_stack_registers_memory_p, rs6000_legitimate_address_p,
1776 mems_ok_for_quad_peep): Use CONST_INT_P and REG_P.
1777 (rs6000_secondary_reload_memory): Use CONST_INT_P and SUBREG_P.
1778 (small_data_operand, print_operand_address): Use CONST_INT_P and
1780 (split_stack_arg_pointer_used_p): Use HARD_REGISTER_P.
1781 (rs6000_init_hard_regno_mode_ok, direct_move_p):
1782 Use HARD_REGISTER_NUM_P.
1783 (rs6000_secondary_reload_gpr): Use HARD_REGISTER_NUM_P and MEM_P.
1784 (rs6000_secondary_reload_class): Use HARD_REGISTER_NUM_P, REG_P,
1785 SUBREG_P and SYMBOL_REF_P.
1786 (register_to_reg_type, rs6000_secondary_reload_inner): Use SUBREG_P
1787 and HARD_REGISTER_NUM_P.
1788 (rs6000_adjust_vec_address): Use HARD_REGISTER_NUM_P and
1790 (rs6000_adjust_cost, find_mem_ref): Use MEM_P.
1791 (macho_lo_sum_memory_operand, rs6000_eliminate_indexed_memrefs): Use
1793 (legitimate_indirect_address_p, legitimate_lo_sum_address_p,
1794 registers_ok_for_quad_peep, rs6000_output_function_epilogue,
1795 find_addr_reg): Use REG_P.
1796 (altivec_expand_vec_perm_const): Use REG_P and SUBREG_P.
1797 (rs6000_emit_le_vsx_move): Use SUBREG_P.
1798 (offsettable_ok_by_alignment, constant_pool_expr_p,
1799 legitimate_small_data_p, rs6000_output_dwarf_dtprel,
1800 rs6000_delegitimize_address, rs6000_const_not_ok_for_debug_p,
1801 rs6000_cannot_force_const_mem, rs6000_output_addr_const_extra,
1802 rs6000_assemble_integer, create_TOC_reference,
1803 rs6000_emit_allocate_stack, rs6000_xcoff_encode_section_info,
1804 rs6000_call_aix, rs6000_call_aix): Use SYMBOL_REF_P.
1805 (rs6000_split_vec_extract_var): Use reg_or_subregno.
1806 * config/rs6000/rtems.h (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Use
1807 CONST_DOUBLE_P, CONST_INT_P and SYMBOL_REF_P.
1808 * config/rs6000/sysv4.h (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
1809 * config/rs6000/xcoff.h (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
1810 * config/rs6000/rs6000.h (RS6000_SYMBOL_REF_TLS_P): Use SYMBOL_REF_P.
1811 (REGNO_OK_FOR_INDEX_P, REGNO_OK_FOR_BASE_P): Use HARD_REGISTER_NUM_P.
1812 (INT_REG_OK_FOR_INDEX_P, INT_REG_OK_FOR_BASE_P): Use HARD_REGISTER_P.
1813 (CONSTANT_ADDRESS_P): Use CONST_INT_P and SYMBOL_REF_P.
1814 * config/rs6000/rs6000.md (define_expands strlensi, mod<mode>3
1815 and cbranch<mode>4): Use CONST_INT_P.
1816 (multiple define_splits): Use REG_P and SUBREG_P.
1817 (define_expands call, call_value): Use MEM_P.
1818 (define_expands sibcall, sibcall_value): Use CONST_INT_P and MEM_P.
1819 (define insn *mtcrfsi): Use CONST_INT_P and REG_P.
1820 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode>,
1821 *vsx_le_perm_load_v8hi, *vsx_le_perm_load_v16qi): Use HARD_REGISTER_P
1822 and HARD_REGISTER_NUM_P.
1823 (multiple define_splits): Use HARD_REGISTER_NUM_P.
1825 2019-01-24 Uroš Bizjak <ubizjak@gmail.com>
1827 PR rtl-optimization/88948
1828 * rtl.h (prepare_copy_insn): New prototype.
1829 * gcse.c (prepare_copy_insn): New function, split out from
1830 process_insert_insn.
1831 (process_insert_insn): Use prepare_copy_insn.
1832 * store-motion.c (replace_store_insn): Use prepare_copy_insn
1833 instead of gen_move_insn.
1835 2019-01-24 Jakub Jelinek <jakub@redhat.com>
1838 * config/i386/i386.c (ix86_pic_register_p): Return true for
1841 PR tree-optimization/88964
1842 * gimple-loop-interchange.cc (loop_cand::analyze_induction_var): Also
1843 punt if HONOR_SNANS (chrec).
1846 * tree-nested.c (convert_nonlocal_reference_stmt,
1847 convert_local_reference_stmt, convert_tramp_reference_stmt,
1848 convert_gimple_call) <case GIMPLE_OMP_TEAMS>: Treat
1849 gimple_omp_teams_host teams stmts like GIMPLE_OMP_PARALLEL
1852 PR tree-optimization/89027
1853 * tree-inline.c (add_clobbers_to_eh_landing_pad): Don't add clobbers
1854 for "omp simd array" variables.
1856 2019-01-24 Richard Earnshaw <rearnsha@arm.com>
1859 * profile-count.h (profile_count): On ARM systems using GCC 6/7/8
1860 force the alignment of m_val.
1862 2019-01-24 Richard Biener <rguenther@suse.de>
1865 * tree-streamer-out.c (write_ts_decl_common_tree_pointers):
1866 When in "legacy" debug mode make sure to reset self-origins.
1868 2019-01-24 Martin Liska <mliska@suse.cz>
1870 PR gcov-profile/88994
1871 * gcov-io.c (mangle_path): Do not allocate a bigger buffer,
1872 result will be always smaller or equal to the original.
1873 * gcov.c (mangle_name): Fix else branch where we should
1874 also copy to PTR and shift the pointer.
1876 2019-01-24 Xiong Hu Luo <luoxhu@linux.vnet.ibm.com>
1878 * tree-ssa-dom.c (test_for_singularity): Fix a comment typo.
1879 * vr-values.c (find_case_label_ranges): Fix a comment typo.
1881 2019-01-23 Bin Cheng <bin.cheng@arm.com>
1882 Steve Ellcey <sellcey@marvell.com>
1885 * recog.c (address_operand): Return false on wrong mode for address.
1886 (constrain_operands): Check for mode with 'p' constraint.
1888 2019-01-23 Uroš Bizjak <ubizjak@gmail.com>
1891 * config/i386/sse.md (sse2_cvtpi2pd): Add SSE alternatives.
1892 Disparage MMX alternative.
1893 (sse2_cvtpd2pi): Ditto.
1894 (sse2_cvttpd2pi): Ditto.
1896 2019-01-23 David Malcolm <dmalcolm@redhat.com>
1899 * config/aarch64/driver-aarch64.c (host_detect_local_cpu): Fix
1900 use-after-free of the result of
1901 aarch64_get_extension_string_for_isa_flags.
1903 2019-01-23 Jakub Jelinek <jakub@redhat.com>
1906 * doc/extend.texi: Document break and continue behavior in
1907 statement expressions.
1909 2019-01-23 Richard Biener <rguenther@suse.de>
1911 PR tree-optimization/89008
1912 * tree-ssa-reassoc.c (eliminate_using_constants): For * 0 do
1913 not leave another stray operand.
1915 2019-01-23 Jakub Jelinek <jakub@redhat.com>
1917 * BASE-VER: Bump to 9.0.1.
1919 2019-01-23 Eric Botcazou <ebotcazou@adacore.com>
1921 * cgraphunit.c (cgraph_node::expand_thunk): When expanding a GIMPLE
1922 thunk that returns by reference, use the type of the return object
1923 of the thunk instead of that of the alias to build the dereference.
1925 2019-01-23 Vineet Gupta <vgupta@synopsys.com>
1927 * config/arc/atomic.md: Add operand to DMB instruction.
1929 2019-01-23 Jakub Jelinek <jakub@redhat.com>
1931 PR tree-optimization/88964
1932 * gimple-loop-interchange.cc (loop_cand::analyze_induction_var): Use
1933 build_zero_cst instead of build_int_cst. Return false for loop
1934 invariants which honor signed zeros.
1936 2019-01-22 Segher Boessenkool <segher@kernel.crashing.org>
1938 * doc/invoke.texi (-fsplit-paths): This is enabled by default at -O3.
1940 2019-01-22 Jakub Jelinek <jakub@redhat.com>
1943 * config/rs6000/rs6000.c: Include tree-vrp.h and tree-ssanames.h.
1944 (rs6000_gimple_fold_builtin): If MEM_REF address doesn't satisfy
1945 is_gimple_mem_ref_addr predicate, force it into a SSA_NAME first.
1948 * gimplify.c (gimplify_omp_atomic): Handle bitfield atomics with
1949 non-integral DECL_BIT_FIELD_REPRESENTATIVEs.
1952 * config/rs6000/vsx.md (*vsx_reduc_<VEC_reduc_name>_v2df_scalar):
1953 Disable for little endian.
1955 2019-01-22 Richard Earnshaw <rearnsha@arm.com>
1958 * config/arm/arm.c (arm_needs_double_word_align): Check
1959 DECL_BIT_FIELD_TYPE.
1961 2019-01-22 Hongtao Liu <hongtao.liu@intel.com>
1962 H.J. Lu <hongjiu.lu@intel.com>
1965 * config/i386/i386-builtin.def: Add mask2 to all builtin
1966 initializations. Merge ARGS2 and SPECIAL_ARGS2 into ARGS and
1968 * config/i386/i386.c (BDESC): Add mask2 to the definition.
1969 (BDESC_FIRST): Likewise.
1970 (define_builtin): Add an argument for mask2. Updated to handle
1971 both ix86_isa_flags and ix86_isa_flags2.
1972 (define_builtin_const): Likewise.
1973 (define_builtin_pure): Likewise.
1974 (define_builtin2): Deleted.
1975 (define_builtin_const2): Likewise.
1976 (builtin_description): Add a member, mask2.
1977 (bdesc_*): Add mask2 to builtin initializations.
1978 (ix86_init_mmx_sse_builtins): Update calls to def_builtin,
1979 def_builtin_const and def_builtin_pure. Remove SPECIAL_ARGS2
1981 (ix86_get_builtin_func_type): Remove SPECIAL_ARGS2 support.
1983 2019-01-22 H.J. Lu <hongjiu.lu@intel.com>
1986 * config/i386/i386.c (ix86_force_load_from_GOT_p): Also check
1989 2019-01-22 Richard Earnshaw <rearnsha@arm.com>
1992 * config/arm/arm.c (arm_needs_doubleword_align): Return 2 if a record's
1993 alignment is dominated by a bitfield with 64-bit aligned base type.
1994 (arm_function_arg): Emit a warning if the alignment has changed since
1995 earlier GCC releases.
1996 (arm_function_arg_boundary): Likewise.
1997 (arm_setup_incoming_varargs): Likewise.
1999 2019-01-22 Richard Biener <rguenther@suse.de>
2001 PR tree-optimization/88862
2002 * graphite-scop-detection.c
2003 (scop_detection::graphite_can_represent_scev): Reject ADDR_EXPR.
2005 2019-01-22 Andrew Stubbs <ams@codesourcery.com>
2007 * doc/extend.tex (AMD GCN Function Attributes): New section.
2008 * doc/install.texi (amdgcn-unknown-amdhsa): New instructions.
2009 * doc/invoke.texi (AMD GCN Options): New section.
2010 * doc/md.texi (Constraints for Particular Machines): Add AMD GCN.
2012 2019-01-22 Eric Botcazou <ebotcazou@adacore.com>
2014 * config/sparc/sparc.c (parc_delegitimize_address): Recognize the GOT
2015 register and decoded HIGH/LO_SUM combinations for labels in PIC mode.
2017 2019-01-22 Jakub Jelinek <jakub@redhat.com>
2019 PR tree-optimization/88044
2020 * tree-ssa-loop-niter.c (number_of_iterations_cond): If condition
2021 is false in the first iteration, but !every_iteration, return false
2022 instead of true with niter->niter zero.
2024 PR rtl-optimization/88904
2025 * cfgcleanup.c (thread_jump): Verify cond2 doesn't mention
2026 any nonequal registers before processing BB_END (b).
2029 * optabs.c (add_equal_note): Add op0_mode argument, use it instead of
2031 (expand_binop_directly, expand_doubleword_clz,
2032 expand_doubleword_popcount, expand_ctz, expand_ffs,
2033 expand_unop_direct, maybe_emit_unop_insn): Adjust callers.
2035 PR rtl-optimization/49429
2037 PR rtl-optimization/86334
2039 * expr.c (emit_block_move_hints): Move marking of MEM_EXPRs
2040 addressable from here...
2041 (emit_block_op_via_libcall): ... to here.
2043 2019-01-22 Richard Biener <rguenther@suse.de>
2045 * tree-vect-loop.c (vect_analyze_loop_operations): Use
2046 auto_vec for cost vector to fix memleak.
2047 (vectorize_fold_left_reduction): Properly gather SLP defs.
2048 (vectorizable_comparison): Do not swap operands to properly
2051 2019-01-22 Alan Modra <amodra@gmail.com>
2054 * config/rs6000/predicates.md (unspec_tls): Ensure GOT reg
2055 stays a reg. Allow a const_int.
2056 * config/rs6000/rs6000-protos.h (rs6000_output_tlsargs): Declare.
2057 * config/rs6000/rs6000.h (IS_V4_FP_ARGS): Define.
2058 (IS_NOMARK_TLSGETADDR): Define.
2059 * config/rs6000/rs6000.c (edit_tls_call_insn): Delete.
2060 (rs6000_output_tlsargs): New function.
2061 (rs6000_legitimize_tls_address): Don't say a !TARGET_TLS_MARKERS
2062 __tls_get_addr call takes an arg.
2063 (rs6000_call_sysv): Generate sysv4 secure plt call pattern here..
2064 * config/rs6000/rs6000.md (call_nonlocal_sysv): ..rather than here,
2066 (call_value_nonlocal_sysv): ..or here, delete split.
2067 (tls_gdld_nomark): Delete.
2068 (call_value_indirect_nonlocal_sysv): Use unspec_tls as operand2
2069 predicate. Call rs6000_output_tlsargs. Adjust length to suit.
2070 (call_value_nonlocal_sysv): Likewise.
2071 (call_value_nonlocal_sysv_secure): Likewise.
2072 (call_value_nonlocal_aix): Likewise.
2073 (call_value_indirect_aix): Likewise.
2074 (call_value_indirect_elfv2): Likewise.
2075 (call_value_local32, call_value_local64): Disable for no-mark tls.
2076 (call_value_local_aix): Likewise.
2078 2019-01-21 Uroš Bizjak <ubizjak@gmail.com>
2081 * config/i386/i386.c (ix86_expand_builtin) [case IX86_BUILTIN_BEXTRI32,
2082 case IX86_BUILTIN_BEXTRI64]: Sanitize operands.
2084 2019-01-21 Michael Ploujnikov <michael.ploujnikov@oracle.com>
2086 * hash-map-tests.c (test_map_of_strings_to_int): Show how to use
2087 string contents as hash_map keys.
2089 2019-01-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
2092 * c-warn.c (check_alignment_of_packed_member): Add a boolean parameter
2093 for rvalue context. Handle rvalues correctly. Use min_align_of_type
2094 instead of TYPE_ALIGN.
2095 (check_address_or_pointer_of_packed_member): Handle rvalues coorrectly.
2096 Use min_align_of_type instead of TYPE_ALIGN_UNIT. Check for NULL
2097 pointer from TYPE_STUB_DECL.
2099 2019-01-21 Richard Biener <rguenther@suse.de>
2101 PR tree-optimization/88934
2102 * tree-vect-slp.c (vect_mask_constant_operand_p): Always look
2103 at the possibly non-constant operand.
2104 (vect_get_constant_vectors): Adjust.
2106 2019-01-21 H.J. Lu <hongjiu.lu@intel.com>
2109 * config/i386/adxintrin.h: Just check _IMMINTRIN_H_INCLUDED.
2110 * config/i386/clflushoptintrin.h: Check _IMMINTRIN_H_INCLUDED
2111 instead of _X86INTRIN_H_INCLUDED.
2112 * onfig/i386/clwbintrin.h: Likewise.
2113 * config/i386/pkuintrin.h: Likewise.
2114 * config/i386/prfchwintrin.h: Likewise.
2115 * config/i386/rdseedintrin.h: Likewise.
2116 * config/i386/wbnoinvdintrin.h: Likewise.
2117 * config/i386/xsavecintrin.h: Likewise.
2118 * config/i386/xsavesintrin.h: Likewise.
2119 * config/i386/fxsrintrin.h: Enable _IMMINTRIN_H_INCLUDED check.
2120 * config/i386/xsaveintrin.h: Likewise.
2121 * config/i386/xsaveoptintrin.h: Likewise.
2122 * config/i386/x86intrin.h: Move "#include" <rdseedintrin.h>,
2123 <prfchwintrin.h>, <fxsrintrin.h>, <xsaveintrin.h>,
2124 <xsaveoptintrin.h>, <adxintrin.h>, <clwbintrin.h>,
2125 <clflushoptintrin.h>, <xsavesintrin.h>, <xsavecintrin.h>,
2126 <wbnoinvdintrin.h> and <pkuintrin.h> to ...
2127 * config/i386/immintrin.h: Here.
2129 2019-01-20 Martin Jambor <mjambor@suse.cz>
2132 * ipa-prop.h (struct ipa_func_body_info): Replaced field aa_walked
2133 with aa_walk_budget.
2134 * cgraph.h (ipa_polymorphic_call_context::get_dynamic_type): Add
2135 aa_walk_budget_p parameter.
2136 * ipa-fnsummary.c (unmodified_parm_1): New parameter fbi. Limit AA
2137 walk. Updated all callers.
2138 (unmodified_parm): New parameter fbi, pass it to unmodified_parm_1.
2139 (eliminated_by_inlining_prob): New parameter fbi, pass it on to
2141 (will_be_nonconstant_expr_predicate): New parameter fbi, removed
2142 parameter info. Extract info from fbi. Pass fbi to recursive calls
2143 and to unmodified_parm.
2144 (phi_result_unknown_predicate): New parameter fbi, removed parameter
2145 info, updated call to will_be_nonconstant_expr_predicate.
2146 (param_change_prob): New parameter fbi, limit AA walking.
2147 (analyze_function_body): Initialize aa_walk_budget in fbi. Update
2148 calls to various above functions.
2149 * ipa-polymorphic-call.c (get_dynamic_type): Add aa_walk_budget_p
2150 parameter. Use it to limit AA walking.
2151 * ipa-prop.c (detect_type_change_from_memory_writes): New parameter
2153 (detect_type_change): New parameter fbi, pass it on to
2154 detect_type_change_from_memory_writes.
2155 (detect_type_change_ssa): Likewise.
2156 (aa_overwalked): Removed.
2157 (parm_preserved_before_stmt_p): Assume fbi is never NULL, stream line
2158 accordingly, adjust to the neew AA limiting scheme.
2159 (parm_ref_data_preserved_p): Likewise.
2160 (ipa_compute_jump_functions_for_edge): Adjust call to
2162 (ipa_analyze_call_uses): Likewise.
2163 (ipa_analyze_virtual_call_uses): Pass fbi to detect_type_change_ssa.
2164 (ipa_analyze_node): Initialize aa_walk_budget.
2165 (ipcp_transform_function): Likewise.
2166 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_stmt): Update call
2167 to get_dynamic_type.
2169 2019-01-19 Jakub Jelinek <jakub@redhat.com>
2171 * config/aarch64/aarch64.c (aarch64_stack_protect_guard): Move
2172 outside of #if CHECKING_P code.
2174 2019-01-19 Richard Sandiford <richard.sandiford@arm.com>
2176 * gimple-loop-versioning.cc (loop_versioning::dump_inner_likelihood):
2177 New function, split out from...
2178 (loop_versioning::analyze_stride): ...here.
2179 (loop_versioning::find_per_loop_multiplication): Use gassign.
2180 (loop_versioning::analyze_term_using_scevs): Return a success code.
2181 (loop_versioning::analyze_arbitrary_term): New function.
2182 (loop_versioning::analyze_address_fragment): Use
2183 analyze_arbitrary_term if all else fails.
2185 2019-01-18 Segher Boessenkool <segher@kernel.crashing.org>
2188 * config/rs6000/rs6000.md (*movsi_from_df): Allow only register
2191 2019-01-18 Richard Biener <rguenther@suse.de>
2193 PR tree-optimization/88903
2194 * tree-vect-stmts.c (vectorizable_shift): Verify we see all
2195 scalar stmts a SLP shift amount is composed of when detecting
2198 2019-01-18 Richard Earnshaw <rearnsha@arm.com>
2201 * config/arm/arm-cpus.in (mp): New feature.
2203 (fgroup ARMv7ve): Add mp and sec features.
2204 (arch armv7-a): Add options to allow mp and sec extensions.
2205 (cpu generic-armv7-a): Add options to allow mp and sec extensions.
2206 (cpu cortex-a5, cpu cortex-7, cpu cortex-a9): Add mp and sec
2207 extenstions to the base architecture.
2208 (cpu cortex-a8): Add sec extension to the base architecture.
2209 (cpu marvell-pj4): Add mp and sec extensions to the base architecture.
2210 * config/arm/t-aprofile (MULTILIB_MATCHES): Map all armv7-a arch
2211 variants down to the base v7-a varaint.
2212 * config/arm/t-multilib (v7_a_arch_variants): New variable.
2213 * doc/invoke.texi (ARM Options): Add +mp and +sec to the list
2214 of permitted extensions for -march=armv7-a and for
2215 -mcpu=generic-armv7-a.
2217 2019-01-18 Martin Liska <mliska@suse.cz>
2219 * params.def: Fix comment.
2220 * tree-profile.c (gimple_init_gcov_profiler): Bump function
2222 (gimple_gen_ic_func_profiler): Likewise.
2224 2019-01-18 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2226 * config/aarch64/aarch64-opts.h (enum stack_protector_guard): New
2227 * config/aarch64/aarch64.c (aarch64_override_options_internal): Handle
2228 and put in error checks for stack protector guard options.
2229 (aarch64_stack_protect_guard): New.
2230 (TARGET_STACK_PROTECT_GUARD): Define.
2231 * config/aarch64/aarch64.md (UNSPEC_SSP_SYSREG): New.
2232 (reg_stack_protect_address<mode>): New.
2233 (stack_protect_set): Adjust for SSP_GLOBAL.
2234 (stack_protect_test): Likewise.
2235 * config/aarch64/aarch64.opt (-mstack-protector-guard-reg): New.
2236 (-mstack-protector-guard): Likewise.
2237 (-mstack-protector-guard-offset): Likewise.
2239 2019-01-18 Jakub Jelinek <jakub@redhat.com>
2241 PR tree-optimization/86214
2242 * tree-inline.h (struct copy_body_data): Add
2243 add_clobbers_to_eh_landing_pads member.
2244 * tree-inline.c (add_clobbers_to_eh_landing_pad): New function.
2245 (copy_edges_for_bb): Call it if EH edge destination is <
2246 id->add_clobbers_to_eh_landing_pads. Fix a comment typo.
2247 (expand_call_inline): Set id->add_clobbers_to_eh_landing_pads
2248 if flag_stack_reuse != SR_NONE and clear it afterwards.
2250 2019-01-18 Christophe Lyon <christophe.lyon@linaro.org>
2253 * doc/install.texi (with-multilib-list): Document for aarch64.
2255 2019-01-18 Jakub Jelinek <jakub@redhat.com>
2258 * config/arm/arm_neon.h: Fix #pragma GCC target syntax - replace
2259 (("..."))) with ("...").
2261 2019-01-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
2263 * doc/extend.texi (Built-in Functions for Memory Model Aware
2264 Atomic Operations): Document atomic fetch and nand.
2266 2019-01-18 Martin Liska <mliska@suse.cz>
2267 Richard Biener <rguenther@suse.de>
2270 * cgraph.h (create_version_clone_with_body): Add new argument
2272 * cgraphclones.c (cgraph_node::create_version_clone): Add
2273 DECL_ATTRIBUTES to a newly created decl. And call
2274 valid_attribute_p so that proper cl_target_optimization_node
2275 is set for the newly created declaration.
2276 * multiple_target.c (create_target_clone): Set DECL_ATTRIBUTES
2278 (expand_target_clones): Do not call valid_attribute_p, it must
2280 * tree-inline.c (copy_decl_for_dup_finish): Reset mode for
2283 2019-01-17 Jakub Jelinek <jakub@redhat.com>
2286 * config/aarch64/arm_neon.h: Fix #pragma GCC target syntax - replace
2287 (("..."))) with ("..."). Use arch=armv8.2-a+sha3 instead of
2288 arch=armv8.2-a+crypto for vsha512hq_u64 etc. intrinsics.
2290 2019-01-17 Martin Sebor <msebor@redhat.com>
2293 * gimple-ssa-warn-restrict.c (builtin_memref::extend_offset_range):
2294 Handle anti-ranges the same as no range at all.
2296 2018-01-17 Steve Ellcey <sellcey@cavium.com>
2298 * config/aarch64/aarch64.c (cgraph.h): New include.
2299 (intl.h): New include.
2300 (supported_simd_type): New function.
2301 (currently_supported_simd_type): Ditto.
2302 (aarch64_simd_clone_compute_vecsize_and_simdlen): Ditto.
2303 (aarch64_simd_clone_adjust): Ditto.
2304 (aarch64_simd_clone_usable): Ditto.
2305 (TARGET_SIMD_CLONE_COMPUTE_VECSIZE_AND_SIMDLEN): New macro.
2306 (TARGET_SIMD_CLONE_ADJUST): Ditto.
2307 (TARGET_SIMD_CLONE_USABLE): Ditto.
2308 * config/i386/i386.c (ix86_simd_clone_adjust): Add definition check.
2309 * omp-simd-clone.c (expand_simd_clones): Add targetm.simd_clone.adjust
2312 2019-01-17 Martin Sebor <msebor@redhat.com>
2314 PR tree-optimization/88800
2315 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid checking
2316 NO_WARNING bit here. Avoid folding out-of-bounds calls.
2317 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Remove
2318 redundant argument. Add new argument and issue diagnostics under
2319 its control. Detect out-of-bounds access even with warnings
2321 (check_bounds_or_overlap): Change return type. Add argument.
2322 (wrestrict_dom_walker::check_call): Adjust.
2323 * gimple-ssa-warn-restrict.h (check_bounds_or_overlap): Add argument.
2324 * tree-ssa-strlen.c (handle_builtin_strcpy): Adjust to change in
2325 check_bounds_or_overlap's return value.
2326 (handle_builtin_stxncpy): Same.
2327 (handle_builtin_strcat): Same.
2329 2019-01-17 Andrew Stubbs <ams@codesourcery.com>
2330 Kwok Cheung Yeung <kcy@codesourcery.com>
2331 Julian Brown <julian@codesourcery.com>
2332 Tom de Vries <tom@codesourcery.com>
2334 * doc/sourcebuild.texi: Document dg-add-options sqrt_insn.
2336 2019-01-17 Andrew Stubbs <ams@codesourcery.com>
2338 * doc/sourcebuild.texi: Document dg-require-effective-target
2339 llvm_binutils and offload_gcn.
2341 2019-01-17 Andrew Stubbs <ams@codesourcery.com>
2342 Kwok Cheung Yeung <kcy@codesourcery.com>
2343 Julian Brown <julian@codesourcery.com>
2344 Tom de Vries <tom@codesourcery.com>
2346 * doc/sourcebuild.texi: Document dg-required-effective-target
2349 2019-01-17 Andrew Stubbs <ams@codesourcery.com>
2350 Kwok Cheung Yeung <kcy@codesourcery.com>
2351 Julian Brown <julian@codesourcery.com>
2352 Tom de Vries <tom@codesourcery.com>
2353 Jan Hubicka <hubicka@ucw.cz>
2354 Martin Jambor <mjambor@suse.cz>
2356 * config.gcc: Add amdgcn*-*-amdhsa configuration.
2357 * configure.ac: Check for dlopen.
2358 * configure: Regenerate.
2360 2019-01-17 Andrew Stubbs <ams@codesourcery.com>
2361 Kwok Cheung Yeung <kcy@codesourcery.com>
2362 Julian Brown <julian@codesourcery.com>
2363 Tom de Vries <tom@codesourcery.com>
2364 Jan Hubicka <hubicka@ucw.cz>
2365 Martin Jambor <mjambor@suse.cz>
2367 * common/config/gcn/gcn-common.c: New file.
2368 * config/gcn/driver-gcn.c: New file.
2369 * config/gcn/gcn-builtins.def: New file.
2370 * config/gcn/gcn-hsa.h: New file.
2371 * config/gcn/gcn-modes.def: New file.
2372 * config/gcn/gcn-opts.h: New file.
2373 * config/gcn/gcn-passes.def: New file.
2374 * config/gcn/gcn-protos.h: New file.
2375 * config/gcn/gcn-run.c: New file.
2376 * config/gcn/gcn-tree.c: New file.
2377 * config/gcn/gcn.c: New file.
2378 * config/gcn/gcn.h: New file.
2379 * config/gcn/gcn.opt: New file.
2380 * config/gcn/t-gcn-hsa: New file.
2382 2019-01-17 Andrew Stubbs <ams@codesourcery.com>
2383 Kwok Cheung Yeung <kcy@codesourcery.com>
2384 Julian Brown <julian@codesourcery.com>
2385 Tom de Vries <tom@codesourcery.com>
2386 Jan Hubicka <hubicka@ucw.cz>
2387 Martin Jambor <mjambor@suse.cz>
2389 * config/gcn/constraints.md: New file.
2390 * config/gcn/gcn-valu.md: New file.
2391 * config/gcn/gcn.md: New file.
2392 * config/gcn/predicates.md: New file.
2394 2019-01-17 Eric Botcazou <ebotcazou@adacore.com>
2396 * gimple-ssa-isolate-paths.c (stmt_uses_name_in_undefined_way): Replace
2397 flag_non_call_exceptions with cfun->can_throw_non_call_exceptions.
2398 (stmt_uses_0_or_null_in_undefined_way): Likewise.
2399 * tree-ssa-alias.c (same_addr_size_stores_p): Likewise.
2401 2019-01-17 Tamar Christina <tamar.christina@arm.com>
2404 * config/aarch64/aarch64.md (STACK_CLASH_SVE_CFA_REGNUM): New.
2405 * config/aarch64/aarch64.c (aarch64_allocate_and_probe_stack_space): Use
2406 it and document registers.
2408 2019-01-17 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2410 * config/aarch64/aarch64.c (ares_tunings): Define.
2411 * config/aarch64/aarch64-cores.def (ares): Use the above.
2413 2019-01-17 Wei Xiao <wei3.xiao@intel.com>
2417 2018-11-06 Wei Xiao <wei3.xiao@intel.com>
2419 * config/i386/avx512fintrin.h: Update VFIXUPIMM* intrinsics.
2420 (_mm512_fixupimm_round_pd): Update parameters and builtin.
2421 (_mm512_maskz_fixupimm_round_pd): Ditto.
2422 (_mm512_fixupimm_round_ps): Ditto.
2423 (_mm512_maskz_fixupimm_round_ps): Ditto.
2424 (_mm_fixupimm_round_sd): Ditto.
2425 (_mm_maskz_fixupimm_round_sd): Ditto.
2426 (_mm_fixupimm_round_ss): Ditto.
2427 (_mm_maskz_fixupimm_round_ss): Ditto.
2428 (_mm512_fixupimm_pd): Ditto.
2429 (_mm512_maskz_fixupimm_pd): Ditto.
2430 (_mm512_fixupimm_ps): Ditto.
2431 (_mm512_maskz_fixupimm_ps): Ditto.
2432 (_mm_fixupimm_sd): Ditto.
2433 (_mm_maskz_fixupimm_sd): Ditto.
2434 (_mm_fixupimm_ss): Ditto.
2435 (_mm_maskz_fixupimm_ss): Ditto.
2436 (_mm512_mask_fixupimm_round_pd): Update builtin.
2437 (_mm512_mask_fixupimm_round_ps): Ditto.
2438 (_mm_mask_fixupimm_round_sd): Ditto.
2439 (_mm_mask_fixupimm_round_ss): Ditto.
2440 (_mm512_mask_fixupimm_pd): Ditto.
2441 (_mm512_mask_fixupimm_ps): Ditto.
2442 (_mm_mask_fixupimm_sd): Ditto.
2443 (_mm_mask_fixupimm_ss): Ditto.
2444 * config/i386/avx512vlintrin.h:
2445 (_mm256_fixupimm_pd): Update parameters and builtin.
2446 (_mm256_maskz_fixupimm_pd): Ditto.
2447 (_mm256_fixupimm_ps): Ditto.
2448 (_mm256_maskz_fixupimm_ps): Ditto.
2449 (_mm_fixupimm_pd): Ditto.
2450 (_mm_maskz_fixupimm_pd): Ditto.
2451 (_mm_fixupimm_ps): Ditto.
2452 (_mm_maskz_fixupimm_ps): Ditto.
2453 (_mm256_mask_fixupimm_pd): Update builtin.
2454 (_mm256_mask_fixupimm_ps): Ditto.
2455 (_mm_mask_fixupimm_pd): Ditto.
2456 (_mm_mask_fixupimm_ps): Ditto.
2457 * config/i386/i386-builtin-types.def: Add new types and remove useless ones.
2458 * config/i386/i386-builtin.def: Update builtin definitions.
2459 * config/i386/i386.c: Handle new builtin types and remove useless ones.
2460 * config/i386/sse.md: Update VFIXUPIMM* patterns.
2461 (<avx512>_fixupimm<mode>_maskz<round_saeonly_expand_name>): Update.
2462 (<avx512>_fixupimm<mode><sd_maskz_name><round_saeonly_name>): Update.
2463 (<avx512>_fixupimm<mode>_mask<round_saeonly_name>): Update.
2464 (avx512f_sfixupimm<mode>_maskz<round_saeonly_expand_name>): Update.
2465 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>): Update.
2466 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Update.
2467 * config/i386/subst.md:
2468 (round_saeonly_sd_mask_operand4): Add new subst_attr.
2469 (round_saeonly_sd_mask_op4): Ditto.
2470 (round_saeonly_expand_operand5): Ditto.
2471 (round_saeonly_expand): Update.
2473 2019-01-17 Wei Xiao <wei3.xiao@intel.com>
2477 2018-11-12 Wei Xiao <wei3.xiao@intel.com>
2479 * config/i386/sse.md: Combine VFIXUPIMM* patterns
2480 (<avx512>_fixupimm<mode>_maskz<round_saeonly_expand_name>): Update.
2481 (<avx512>_fixupimm<mode><sd_maskz_name><round_saeonly_name>): Update.
2482 (<avx512>_fixupimm<mode>_mask<round_saeonly_name>): Remove.
2483 (avx512f_sfixupimm<mode>_maskz<round_saeonly_expand_name>): Update.
2484 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>): Update.
2485 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Remove.
2487 2019-01-17 Wei Xiao <wei3.xiao@intel.com>
2491 2018-12-15 Jakub Jelinek <jakub@redhat.com>
2494 * config/i386/sse.md (UNSPEC_SFIXUPIMM): New unspec enumerator.
2495 (avx512f_sfixupimm<mode><mask_name><round_saeonly_name>): Use it
2496 instead of UNSPEC_FIXUPIMM.
2498 2019-01-17 Richard Biener <rguenther@suse.de>
2501 * dwarf2out.c (want_pubnames): Never generate pubnames sections
2502 and friends for the LTO part of debug info.
2504 2019-01-17 Jakub Jelinek <jakub@redhat.com>
2506 PR tree-optimization/86214
2507 * cfgexpand.c (add_stack_var_conflict): Don't add any conflicts
2510 PR rtl-optimization/88870
2511 * dce.c (deletable_insn_p): Never delete const/pure calls that can
2512 throw if we can't alter the cfg or delete dead exceptions.
2513 (mark_insn): Don't call find_call_stack_args for such calls.
2515 2019-01-17 Kewen Lin <linkw@gcc.gnu.org>
2517 * doc/extend.texi: Add four new prototypes for vec_ld and seven new
2518 prototypes for vec_st.
2519 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add entries
2520 for scalar address type variants of altivec_vec_ld/altivec_vec_st,
2521 mainly on signed/unsigned long long and double.
2523 2019-01-16 David Malcolm <dmalcolm@redhat.com>
2526 * combine.c (delete_noop_moves): Convert to "bool" return,
2527 returning true if any edges are eliminated.
2528 (combine_instructions): Also return true if delete_noop_moves
2531 2019-01-16 Tamar Christina <tamar.christina@arm.com>
2533 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args): Use
2534 correct max nunits for endian swap.
2535 (aarch64_expand_fcmla_builtin): Correct subreg code.
2536 * config/aarch64/aarch64-simd.md (aarch64_fcmla_lane<rot><mode>,
2537 aarch64_fcmla_laneq<rot>v4hf, aarch64_fcmlaq_lane<rot><mode>): Correct
2540 2019-01-16 Uroš Bizjak <ubizjak@gmail.com>
2542 * config/alpha/alpha.c (alpha_gimplify_va_arg):
2543 Handle split indirect COMPLEX_TYPE arguments.
2545 2019-01-16 Richard Earnshaw <rearnsha@arm.com>
2548 * config/aarch64/aarch64-modes.def: Add comment about how the carry
2549 bit is set by add and compare.
2550 (CC_ADC): New CC_MODE.
2551 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Use variables
2552 to cache the code and mode of X. Adjust the shape of a CC_Cmode
2553 comparison. Add detection for CC_ADCmode.
2554 (aarch64_get_condition_code_1): Update code support for CC_Cmode. Add
2556 * config/aarch64/aarch64.md (uaddv<mode>4): Use LTU with CCmode.
2557 (uaddvti4): Comparison result is in CC_ADCmode and the condition is GEU.
2558 (add<mode>3_compareC_cconly_imm): Delete. Merge into...
2559 (add<mode>3_compareC_cconly): ... this. Restructure the comparison
2560 to eliminate the need for zero-extending the operands.
2561 (add<mode>3_compareC_imm): Delete. Merge into ...
2562 (add<mode>3_compareC): ... this. Restructure the comparison to
2563 eliminate the need for zero-extending the operands.
2564 (add<mode>3_carryin): Use LTU for the overflow detection.
2565 (add<mode>3_carryinC): Use CC_ADCmode for the result of the carry out.
2566 Reexpress comparison for overflow.
2567 (add<mode>3_carryinC_zero): Update for change to add<mode>3_carryinC.
2568 (add<mode>3_carryinC): Likewise.
2569 (add<mode>3_carryinV): Use LTU for carry between partials.
2570 * config/aarch64/predicates.md (aarch64_carry_operation): Update
2571 handling of CC_Cmode and add CC_ADCmode.
2572 (aarch64_borrow_operation): Likewise.
2574 2019-01-16 Tamar Christina <tamar.christina@arm.com>
2576 * config/arm/arm-protos.h (neon_vcmla_lane_prepare_operands): Remove patternmode.
2577 * config/arm/arm.c (neon_vcmla_lane_prepare_operands): Likewise.
2578 * config/arm/neon.md (neon_vcmla_lane<rot><mode>, neon_vcmla_laneq<rot><mode>,
2579 neon_vcmlaq_lane<rot><mode>): Remove endianness conversion.
2581 2019-01-16 Martin Liska <mliska@suse.cz>
2583 * Makefile.in: Set TOOL_INCLUDE_DIR and NATIVE_SYSTEM_HEADER_DIR
2585 * config/gnu-user.h (TARGET_F951_OPTIONS): Add 'finclude%s/' as
2587 * gcc.c (add_sysrooted_hdrs_prefix): New function.
2588 (path_prefix_reset): Move up in the source file.
2589 (find_fortran_preinclude_file): Make complex search for the
2590 fortran header files.
2592 2019-01-15 Nikhil Benesch <nikhil.benesch@gmail.com>
2594 * godump.c (go_output_typedef): When outputting a typedef, refer
2595 to the underlying type by its name and not its structure.
2597 2019-01-15 David Malcolm <dmalcolm@redhat.com>
2600 * tree.c (build_function_type): Assert that arg_types is not
2603 2019-01-15 Richard Sandiford <richard.sandiford@arm.com>
2606 * doc/extend.texi: Document that listing the stack pointer in the
2607 clobber list of an asm is a deprecated feature.
2608 * common.opt (Wdeprecated): Moved from c-family/c.opt.
2609 * cfgexpand.c (asm_clobber_reg_is_valid): Issue a -Wdeprecated
2610 warning instead of an error for clobbers of the stack pointer.
2611 Add a note explaining why.
2613 2019-01-15 Richard Biener <rguenther@suse.de>
2616 * dwarf2out.c (gen_member_die): Do not generate inheritance
2619 2019-01-15 Richard Biener <rguenther@suse.de>
2621 PR tree-optimization/88855
2622 * tree-if-conv.c (combine_blocks): Collect
2623 SSA_NAME_OCCURS_IN_ABNORMAL_PHI from propagated out virtuals.
2625 2019-01-15 Tom de Vries <tdevries@suse.de>
2628 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Handle
2629 lhs == NULL_TREE for gang-level reduction.
2631 2019-01-15 Richard Biener <rguenther@suse.de>
2632 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2635 * ipa-pure-const.c (malloc_candidate_p_1): Add parameter visited and
2636 return true if SSA_NAME is already marked in visited bitmap.
2637 (malloc_candidate_p): Pass visited to malloc_candidate_p_1.
2639 2019-01-15 Jakub Jelinek <jakub@redhat.com>
2641 PR tree-optimization/88775
2642 * match.pd (cmp (convert1?@2 addr@0) (convert2? addr@1)): Optimize
2643 equal == 0 equality pointer comparisons some more if compared in
2644 integral types and either one points to an automatic var and the
2645 other to a global, or we can prove at least one points to the middle
2646 or both point to start or both point to end.
2648 2019-01-14 Andi Kleen <ak@linux.intel.com>
2650 * Makefile.in: Lower autofdo sampling rate by 10x.
2651 * Makefile.tpl: Dito.
2653 2019-01-14 Tom Honermann <tom@honermann.net>
2655 * defaults.h: Define CHAR8_TYPE.
2657 2019-01-14 Martin Sebor <msebor@redhat.com>
2660 * doc/extend.texi (Darwin Format Checks): Clarify.
2662 2019-01-14 Richard Biener <rguenther@suse.de>
2664 * genmatch.c (dt_simplify::gen_1): Change dumping dependent on
2665 whether we are in (simplify ...) or (match ...) context.
2667 2019-01-14 Jakub Jelinek <jakub@redhat.com>
2669 PR rtl-optimization/88796
2670 * emit-rtl.h (struct rtl_data): Add stack_protect_guard_decl field.
2671 * cfgexpand.c (stack_protect_prologue): Initialize
2672 crtl->stack_protect_guard_decl.
2673 * function.c (stack_protect_epilogue): Use it instead of calling
2674 targetm.stack_protect_guard again.
2675 * dse.c (check_mem_read_rtx): Ignore MEM_VOLATILE_P reads from
2676 MEMs with MEM_EXPR equal to crtl->stack_protect_guard or
2677 crtl->stack_protect_guard_decl.
2678 * config/i386/i386.c (ix86_stack_protect_guard): Set TREE_THIS_VOLATILE
2679 on the returned MEM_EXPR.
2681 2019-01-12 Tom de Vries <tdevries@suse.de>
2683 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims_1): Alow setting
2684 vector length using -fopenacc-dim.
2686 2019-01-12 Tom de Vries <tdevries@suse.de>
2688 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Take larger vector
2689 lengths into account.
2691 2019-01-12 Svante Signell <svante.signell@gmail.com>
2693 * config/i386/gnu.h (TARGET_THREAD_SSP_OFFSET): Define.
2694 (TARGET_CAN_SPLIT_STACK): Define.
2695 (TARGET_THREAD_SPLIT_STACK_OFFSET): Define.
2697 2019-01-05 Jan Hubicka <hubicka@ucw.cz>
2699 * params.def (inline-unit-growth): Set to 40.
2701 2019-01-12 Jakub Jelinek <jakub@redhat.com>
2703 * tree-ssa-loop-ivopts.c (find_inv_vars): Fix a comment typo.
2705 2019-01-12 Tom de Vries <tdevries@suse.de>
2707 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims_1): In offloading
2708 region calling vector-partitionable routine, set default_vector_length
2711 2019-01-12 Tom de Vries <tdevries@suse.de>
2713 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims_1): Add an use new
2714 variable default_vector_length.
2716 2019-01-12 Tom de Vries <tdevries@suse.de>
2719 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims_1): Apply defaults
2720 from oacc_default_dims, as oacc_validate_dims would do it, and apply
2723 2019-01-12 Tom de Vries <tdevries@suse.de>
2725 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims_1)
2726 (nvptx_goacc_validate_dims): Add used parameter.
2727 * doc/tm.texi: Regenerate.
2728 * omp-offload.c (oacc_parse_default_dims, oacc_validate_dims): Add
2729 argument to call to targetm.goacc.validate_dims.
2730 (default_goacc_validate_dims): Add used
2732 * target.def (validate_dims): Add used parameter in DEFHOOK.
2733 * targhooks.h (default_goacc_validate_dims): Add used parameter.
2735 2019-01-11 Jakub Jelinek <jakub@redhat.com>
2739 * tree-inline.h (struct copy_body_data): Add adjust_array_error_bounds
2741 * tree-inline.c (remap_type_1): Formatting fix. If TYPE_MAX_VALUE of
2742 ARRAY_TYPE's TYPE_DOMAIN is newly error_mark_node, replace it with
2743 a dummy "omp dummy var" variable if id->adjust_array_error_bounds.
2744 * omp-low.c (new_omp_context): Set cb.adjust_array_error_bounds.
2746 2019-01-11 Vladimir Makarov <vmakarov@redhat.com>
2748 PR rtl-optimization/87305
2750 (setup_live_pseudos_and_spill_after_risky_transforms): Add code
2751 for little endian pseudos used as paradoxical subreg.
2753 2019-01-11 Jakub Jelinek <jakub@redhat.com>
2755 PR tree-optimization/88693
2756 * tree-ssa-strlen.c (get_min_string_length): Don't set *full_string_p
2757 for STRING_CSTs that don't contain any NUL characters in the first
2758 TREE_STRING_LENGTH bytes.
2760 2019-01-11 Alan Modra <amodra@gmail.com>
2764 * genattrtab.c (min_fn): Don't translate values.
2765 (min_attr_value): Return INT_MAX when the value can't be calculated.
2766 Return minimum among any values that can be calculated.
2767 (max_attr_value): Adjust.
2769 2019-01-11 Jakub Jelinek <jakub@redhat.com>
2771 * Makefile.in (PLUGIN_HEADERS): Add $(INSN_ATTR_H).
2773 2019-01-11 Steve Ellcey <sellcey@marvell.com>
2775 * config/aarch64/aarch64.c (aarch64_simd_call_p): New function.
2776 (aarch64_hard_regno_call_part_clobbered): Add insn argument.
2777 (aarch64_return_call_with_max_clobbers): New function.
2778 (TARGET_RETURN_CALL_WITH_MAX_CLOBBERS): New macro.
2779 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Add insn
2781 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): Ditto.
2782 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Ditto.
2783 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered): Ditto.
2784 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): Ditto.
2785 * cselib.c (cselib_process_insn): Add argument to
2786 targetm.hard_regno_call_part_clobbered call.
2787 * ira-conflicts.c (ira_build_conflicts): Ditto.
2788 * ira-costs.c (ira_tune_allocno_costs): Ditto.
2789 * lra-constraints.c (inherit_reload_reg): Ditto.
2790 * lra-int.h (struct lra_reg): Add call_insn field, remove call_p field.
2791 * lra-lives.c (check_pseudos_live_through_calls): Add call_insn
2792 argument. Call targetm.return_call_with_max_clobbers.
2793 Add argument to targetm.hard_regno_call_part_clobbered call.
2794 (calls_have_same_clobbers_p): New function.
2795 (process_bb_lives): Add call_insn and last_call_insn variables.
2796 Pass call_insn to check_pseudos_live_through_calls.
2797 Modify if stmt to check targetm.return_call_with_max_clobbers.
2798 Update setting of flush variable.
2799 (lra_create_live_ranges_1): Set call_insn to NULL instead of call_p
2801 * lra.c (initialize_lra_reg_info_element): Set call_insn to NULL.
2802 * regcprop.c (copyprop_hardreg_forward_1): Add argument to
2803 targetm.hard_regno_call_part_clobbered call.
2804 * reginfo.c (choose_hard_reg_mode): Ditto.
2805 * regrename.c (check_new_reg_p): Ditto.
2806 * reload.c (find_equiv_reg): Ditto.
2807 * reload1.c (emit_reload_insns): Ditto.
2808 * sched-deps.c (deps_analyze_insn): Ditto.
2809 * sel-sched.c (init_regs_for_mode): Ditto.
2810 (mark_unavailable_hard_regs): Ditto.
2811 * targhooks.c (default_dwarf_frame_reg_mode): Ditto.
2812 * target.def (hard_regno_call_part_clobbered): Add insn argument.
2813 (return_call_with_max_clobbers): New target function.
2814 * doc/tm.texi: Regenerate.
2815 * doc/tm.texi.in (TARGET_RETURN_CALL_WITH_MAX_CLOBBERS): New hook.
2816 * hooks.c (hook_bool_uint_mode_false): Change to
2817 hook_bool_insn_uint_mode_false.
2818 * hooks.h (hook_bool_uint_mode_false): Ditto.
2820 2019-01-11 Steve Ellcey <sellcey@marvell.com>
2822 * config/aarch64/aarch64.c (aarch64_simd_call_p): New function.
2823 (aarch64_remove_extra_call_preserved_regs): New function.
2824 (TARGET_REMOVE_EXTRA_CALL_PRESERVED_REGS): New macro.
2825 * doc/tm.texi.in (TARGET_REMOVE_EXTRA_CALL_PRESERVED_REGS): New hook.
2826 * doc/tm.texi: Regenerate.
2827 * final.c (get_call_reg_set_usage): Call new hook.
2828 * target.def (remove_extra_call_preserved_regs): New hook.
2829 * targhooks.c (default_remove_extra_call_preserved_regs): New function.
2830 * targhooks.h (default_remove_extra_call_preserved_regs): New function.
2832 2019-01-11 Jakub Jelinek <jakub@redhat.com>
2835 * passes.c (finish_optimization_passes): Call print_combine_total_stats
2836 inside of pass_combine_1 dump rather than pass_profile_1.
2838 2019-01-11 Tom de Vries <tdevries@suse.de>
2840 * config/nvptx/nvptx.c (PTX_CTA_NUM_BARRIERS, PTX_PER_CTA_BARRIER)
2841 (PTX_NUM_PER_CTA_BARRIER, PTX_FIRST_PER_WORKER_BARRIER)
2842 (PTX_NUM_PER_WORKER_BARRIERS): Define.
2843 (nvptx_apply_dim_limits): Prevent vector_length 64 and
2846 2019-01-11 Tom de Vries <tdevries@suse.de>
2848 * config/nvptx/nvptx.c (PTX_CTA_SIZE): Move up.
2850 2019-01-11 Jan Beulich <jbeulich@suse.com>
2852 * config/i386/i386.md (rex64suffix): Add L suffix for SI.
2853 * config/i386/sse.md (cvtusi2<ssescalarmodesuffix>32<round_name>,
2854 sse2_cvtsi2sd): Add {l}.
2855 (sse2_cvtsi2sdq<round_name>): Make q conditional upon AT&T
2858 2019-01-10 Jakub Jelinek <jakub@redhat.com>
2861 * config/i386/sse.md (float<floatunssuffix>v2div2sf2): Turn into
2863 (*float<floatunssuffix>v2div2sf2): New define_insn.
2864 (float<floatunssuffix>v2div2sf2_mask): Turn into define_expand.
2865 (*float<floatunssuffix>v2div2sf2_mask): New define_insn.
2866 (*float<floatunssuffix>v2div2sf2_mask_1): Replace
2867 subrtxes (const_vector:V2SF [(const_int 0) (const_int 0)]) with
2868 match_operands with "const0_operand" "C".
2870 2019-01-10 Tamar Christina <tamar.christina@arm.com>
2872 * config/aarch64/aarch64-builtins.c
2873 (aarch64_init_builtins): Move aarch64_init_fcmla_laneq_builtins...
2874 (aarch64_init_simd_builtins): ...Here
2876 2019-01-10 Vladimir Makarov <vmakarov@redhat.com>
2878 PR rtl-optimization/87305
2880 (setup_live_pseudos_and_spill_after_risky_transforms): Check
2881 allocation for big endian pseudos used as paradoxical subregs and
2882 spill them if it is wrong.
2883 * lra-constraints.c (lra_constraints): Add a comment.
2885 2019-01-10 Richard Biener <rguenther@suse.de>
2887 PR tree-optimization/88792
2888 * tree-ssa-pre.c (get_representative_for): Do not return a
2891 2019-01-10 Jakub Jelinek <jakub@redhat.com>
2895 * function.c (assign_stack_local_1): Revert the 2018-11-21 changes.
2896 (assign_parm_setup_block): Do the argument slot realignment here
2899 2019-01-10 Stefan Agner <stefan@agner.ch>
2902 * config/arm/arm.c (arm_option_override_internal): Force
2903 opts->x_inline_asm_unified to true only if TARGET_THUMB2_P.
2905 2019-01-10 Jakub Jelinek <jakub@redhat.com>
2908 * attribs.c (handle_dll_attribute): Clear TREE_STATIC after setting
2911 2019-01-10 Tamar Christina <tamar.christina@arm.com>
2913 * config/arm/arm-builtins.c
2914 (enum arm_type_qualifiers): Add qualifier_lane_pair_index.
2915 (MAC_LANE_PAIR_QUALIFIERS): New.
2916 (arm_expand_builtin_args): Use it.
2917 (arm_expand_builtin_1): Likewise.
2918 * config/arm/arm-protos.h (neon_vcmla_lane_prepare_operands): New.
2919 * config/arm/arm.c (neon_vcmla_lane_prepare_operands): New.
2920 * config/arm/arm-c.c (arm_cpu_builtins): Add __ARM_FEATURE_COMPLEX.
2921 * config/arm/arm_neon.h:
2922 (vcadd_rot90_f16): New.
2923 (vcaddq_rot90_f16): New.
2924 (vcadd_rot270_f16): New.
2925 (vcaddq_rot270_f16): New.
2928 (vcmla_lane_f16): New.
2929 (vcmla_laneq_f16): New.
2930 (vcmlaq_lane_f16): New.
2931 (vcmlaq_laneq_f16): New.
2932 (vcmla_rot90_f16): New.
2933 (vcmlaq_rot90_f16): New.
2934 (vcmla_rot90_lane_f16): New.
2935 (vcmla_rot90_laneq_f16): New.
2936 (vcmlaq_rot90_lane_f16): New.
2937 (vcmlaq_rot90_laneq_f16): New.
2938 (vcmla_rot180_f16): New.
2939 (vcmlaq_rot180_f16): New.
2940 (vcmla_rot180_lane_f16): New.
2941 (vcmla_rot180_laneq_f16): New.
2942 (vcmlaq_rot180_lane_f16): New.
2943 (vcmlaq_rot180_laneq_f16): New.
2944 (vcmla_rot270_f16): New.
2945 (vcmlaq_rot270_f16): New.
2946 (vcmla_rot270_lane_f16): New.
2947 (vcmla_rot270_laneq_f16): New.
2948 (vcmlaq_rot270_lane_f16): New.
2949 (vcmlaq_rot270_laneq_f16): New.
2950 (vcadd_rot90_f32): New.
2951 (vcaddq_rot90_f32): New.
2952 (vcadd_rot270_f32): New.
2953 (vcaddq_rot270_f32): New.
2956 (vcmla_lane_f32): New.
2957 (vcmla_laneq_f32): New.
2958 (vcmlaq_lane_f32): New.
2959 (vcmlaq_laneq_f32): New.
2960 (vcmla_rot90_f32): New.
2961 (vcmlaq_rot90_f32): New.
2962 (vcmla_rot90_lane_f32): New.
2963 (vcmla_rot90_laneq_f32): New.
2964 (vcmlaq_rot90_lane_f32): New.
2965 (vcmlaq_rot90_laneq_f32): New.
2966 (vcmla_rot180_f32): New.
2967 (vcmlaq_rot180_f32): New.
2968 (vcmla_rot180_lane_f32): New.
2969 (vcmla_rot180_laneq_f32): New.
2970 (vcmlaq_rot180_lane_f32): New.
2971 (vcmlaq_rot180_laneq_f32): New.
2972 (vcmla_rot270_f32): New.
2973 (vcmlaq_rot270_f32): New.
2974 (vcmla_rot270_lane_f32): New.
2975 (vcmla_rot270_laneq_f32): New.
2976 (vcmlaq_rot270_lane_f32): New.
2977 (vcmlaq_rot270_laneq_f32): New.
2978 * config/arm/arm_neon_builtins.def (vcadd90, vcadd270, vcmla0, vcmla90,
2979 vcmla180, vcmla270, vcmla_lane0, vcmla_lane90, vcmla_lane180, vcmla_lane270,
2980 vcmla_laneq0, vcmla_laneq90, vcmla_laneq180, vcmla_laneq270,
2981 vcmlaq_lane0, vcmlaq_lane90, vcmlaq_lane180, vcmlaq_lane270): New.
2982 * config/arm/neon.md (neon_vcmla_lane<rot><mode>,
2983 neon_vcmla_laneq<rot><mode>, neon_vcmlaq_lane<rot><mode>): New.
2984 * config/arm/arm.c (arm_arch8_3, arm_arch8_4): New.
2985 * config/arm/arm.h (TARGET_COMPLEX, arm_arch8_3, arm_arch8_4): New.
2986 (arm_option_reconfigure_globals): Use them.
2987 * config/arm/iterators.md (VDF, VQ_HSF): New.
2988 (VCADD, VCMLA): New.
2989 (VF_constraint, rot, rotsplit1, rotsplit2): Add V4HF and V8HF.
2990 * config/arm/neon.md (neon_vcadd<rot><mode>, neon_vcmla<rot><mode>): New.
2991 * config/arm/unspecs.md (UNSPEC_VCADD90, UNSPEC_VCADD270,
2992 UNSPEC_VCMLA, UNSPEC_VCMLA90, UNSPEC_VCMLA180, UNSPEC_VCMLA270): New.
2994 2019-01-10 Tamar Christina <tamar.christina@arm.com>
2996 * config/aarch64/aarch64-builtins.c (enum aarch64_type_qualifiers): Add qualifier_lane_pair_index.
2997 (emit-rtl.h): Include.
2998 (TYPES_QUADOP_LANE_PAIR): New.
2999 (aarch64_simd_expand_args): Use it.
3000 (aarch64_simd_expand_builtin): Likewise.
3001 (AARCH64_SIMD_FCMLA_LANEQ_BUILTINS, aarch64_fcmla_laneq_builtin_datum): New.
3002 (FCMLA_LANEQ_BUILTIN, AARCH64_SIMD_FCMLA_LANEQ_BUILTIN_BASE,
3003 AARCH64_SIMD_FCMLA_LANEQ_BUILTINS, aarch64_fcmla_lane_builtin_data,
3004 aarch64_init_fcmla_laneq_builtins, aarch64_expand_fcmla_builtin): New.
3005 (aarch64_init_builtins): Add aarch64_init_fcmla_laneq_builtins.
3006 (aarch64_expand_buildin): Add AARCH64_SIMD_BUILTIN_FCMLA_LANEQ0_V2SF,
3007 AARCH64_SIMD_BUILTIN_FCMLA_LANEQ90_V2SF, AARCH64_SIMD_BUILTIN_FCMLA_LANEQ180_V2SF,
3008 AARCH64_SIMD_BUILTIN_FCMLA_LANEQ2700_V2SF, AARCH64_SIMD_BUILTIN_FCMLA_LANEQ0_V4HF,
3009 AARCH64_SIMD_BUILTIN_FCMLA_LANEQ90_V4HF, AARCH64_SIMD_BUILTIN_FCMLA_LANEQ180_V4HF,
3010 AARCH64_SIMD_BUILTIN_FCMLA_LANEQ270_V4HF.
3011 * config/aarch64/iterators.md (FCMLA_maybe_lane): New.
3012 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins): Add __ARM_FEATURE_COMPLEX.
3013 * config/aarch64/aarch64-simd-builtins.def (fcadd90, fcadd270, fcmla0, fcmla90,
3014 fcmla180, fcmla270, fcmla_lane0, fcmla_lane90, fcmla_lane180, fcmla_lane270,
3015 fcmla_laneq0, fcmla_laneq90, fcmla_laneq180, fcmla_laneq270,
3016 fcmlaq_lane0, fcmlaq_lane90, fcmlaq_lane180, fcmlaq_lane270): New.
3017 * config/aarch64/aarch64-simd.md (aarch64_fcmla_lane<rot><mode>,
3018 aarch64_fcmla_laneq<rot>v4hf, aarch64_fcmlaq_lane<rot><mode>,aarch64_fcadd<rot><mode>,
3019 aarch64_fcmla<rot><mode>): New.
3020 * config/aarch64/arm_neon.h:
3021 (vcadd_rot90_f16): New.
3022 (vcaddq_rot90_f16): New.
3023 (vcadd_rot270_f16): New.
3024 (vcaddq_rot270_f16): New.
3027 (vcmla_lane_f16): New.
3028 (vcmla_laneq_f16): New.
3029 (vcmlaq_lane_f16): New.
3030 (vcmlaq_rot90_lane_f16): New.
3031 (vcmla_rot90_laneq_f16): New.
3032 (vcmla_rot90_lane_f16): New.
3033 (vcmlaq_rot90_f16): New.
3034 (vcmla_rot90_f16): New.
3035 (vcmlaq_laneq_f16): New.
3036 (vcmla_rot180_laneq_f16): New.
3037 (vcmla_rot180_lane_f16): New.
3038 (vcmlaq_rot180_f16): New.
3039 (vcmla_rot180_f16): New.
3040 (vcmlaq_rot90_laneq_f16): New.
3041 (vcmlaq_rot270_laneq_f16): New.
3042 (vcmlaq_rot270_lane_f16): New.
3043 (vcmla_rot270_laneq_f16): New.
3044 (vcmlaq_rot270_f16): New.
3045 (vcmla_rot270_f16): New.
3046 (vcmlaq_rot180_laneq_f16): New.
3047 (vcmlaq_rot180_lane_f16): New.
3048 (vcmla_rot270_lane_f16): New.
3049 (vcadd_rot90_f32): New.
3050 (vcaddq_rot90_f32): New.
3051 (vcaddq_rot90_f64): New.
3052 (vcadd_rot270_f32): New.
3053 (vcaddq_rot270_f32): New.
3054 (vcaddq_rot270_f64): New.
3058 (vcmla_lane_f32): New.
3059 (vcmla_laneq_f32): New.
3060 (vcmlaq_lane_f32): New.
3061 (vcmlaq_laneq_f32): New.
3062 (vcmla_rot90_f32): New.
3063 (vcmlaq_rot90_f32): New.
3064 (vcmlaq_rot90_f64): New.
3065 (vcmla_rot90_lane_f32): New.
3066 (vcmla_rot90_laneq_f32): New.
3067 (vcmlaq_rot90_lane_f32): New.
3068 (vcmlaq_rot90_laneq_f32): New.
3069 (vcmla_rot180_f32): New.
3070 (vcmlaq_rot180_f32): New.
3071 (vcmlaq_rot180_f64): New.
3072 (vcmla_rot180_lane_f32): New.
3073 (vcmla_rot180_laneq_f32): New.
3074 (vcmlaq_rot180_lane_f32): New.
3075 (vcmlaq_rot180_laneq_f32): New.
3076 (vcmla_rot270_f32): New.
3077 (vcmlaq_rot270_f32): New.
3078 (vcmlaq_rot270_f64): New.
3079 (vcmla_rot270_lane_f32): New.
3080 (vcmla_rot270_laneq_f32): New.
3081 (vcmlaq_rot270_lane_f32): New.
3082 (vcmlaq_rot270_laneq_f32): New.
3083 * config/aarch64/aarch64.h (TARGET_COMPLEX): New.
3084 * config/aarch64/iterators.md (UNSPEC_FCADD90, UNSPEC_FCADD270,
3085 UNSPEC_FCMLA, UNSPEC_FCMLA90, UNSPEC_FCMLA180, UNSPEC_FCMLA270): New.
3086 (FCADD, FCMLA): New.
3088 * config/arm/types.md (neon_fcadd, neon_fcmla): New.
3090 2019-01-09 Sandra Loosemore <sandra@codesourcery.com>
3094 * config/pa/pa.c: Change "can not" to "cannot".
3095 * gimple-ssa-evrp-analyze.c: Likewise.
3096 * ipa-icf.c: Likewise.
3097 * ipa-polymorphic-call.c: Likewise.
3098 * ipa-pure-const.c: Likewise.
3099 * lra-constraints.c: Likewise.
3100 * lra-remat.c: Likewise.
3101 * reload1.c: Likewise.
3102 * reorg.c: Likewise.
3103 * tree-ssa-uninit.c: Likewise.
3105 2019-01-09 Sandra Loosemore <sandra@codesourcery.com>
3109 * Makefile.in: Mechanically replace "can not" with "cannot".
3110 * alias.c: Likewise.
3111 * builtins.c: Likewise.
3112 * calls.c: Likewise.
3113 * cgraph.c: Likewise.
3114 * cgraph.h: Likewise.
3115 * cgraphclones.c: Likewise.
3116 * cgraphunit.c: Likewise.
3117 * combine-stack-adj.c: Likewise.
3118 * combine.c: Likewise.
3119 * common/config/i386/i386-common.c: Likewise.
3120 * config/aarch64/aarch64.c: Likewise.
3121 * config/alpha/sync.md: Likewise.
3122 * config/arc/arc.c: Likewise.
3123 * config/arc/predicates.md: Likewise.
3124 * config/arm/arm-c.c: Likewise.
3125 * config/arm/arm.c: Likewise.
3126 * config/arm/arm.h: Likewise.
3127 * config/arm/arm.md: Likewise.
3128 * config/arm/cortex-r4f.md: Likewise.
3129 * config/csky/csky.c: Likewise.
3130 * config/csky/csky.h: Likewise.
3131 * config/darwin-f.c: Likewise.
3132 * config/epiphany/epiphany.md: Likewise.
3133 * config/i386/i386.c: Likewise.
3134 * config/i386/sol2.h: Likewise.
3135 * config/m68k/m68k.c: Likewise.
3136 * config/mcore/mcore.h: Likewise.
3137 * config/microblaze/microblaze.md: Likewise.
3138 * config/mips/20kc.md: Likewise.
3139 * config/mips/sb1.md: Likewise.
3140 * config/nds32/nds32.c: Likewise.
3141 * config/nds32/predicates.md: Likewise.
3142 * config/pa/pa.c: Likewise.
3143 * config/rs6000/e300c2c3.md: Likewise.
3144 * config/rs6000/rs6000.c: Likewise.
3145 * config/s390/s390.h: Likewise.
3146 * config/sh/sh.c: Likewise.
3147 * config/sh/sh.md: Likewise.
3148 * config/spu/vmx2spu.h: Likewise.
3149 * cprop.c: Likewise.
3150 * dbxout.c: Likewise.
3151 * df-scan.c: Likewise.
3152 * doc/cfg.texi: Likewise.
3153 * doc/extend.texi: Likewise.
3154 * doc/fragments.texi: Likewise.
3155 * doc/gty.texi: Likewise.
3156 * doc/invoke.texi: Likewise.
3157 * doc/lto.texi: Likewise.
3158 * doc/md.texi: Likewise.
3159 * doc/objc.texi: Likewise.
3160 * doc/rtl.texi: Likewise.
3161 * doc/tm.texi: Likewise.
3163 * emit-rtl.c: Likewise.
3164 * emit-rtl.h: Likewise.
3165 * except.c: Likewise.
3166 * expmed.c: Likewise.
3168 * fold-const.c: Likewise.
3169 * genautomata.c: Likewise.
3170 * gimple-fold.c: Likewise.
3171 * hard-reg-set.h: Likewise.
3172 * ifcvt.c: Likewise.
3173 * ipa-comdats.c: Likewise.
3174 * ipa-cp.c: Likewise.
3175 * ipa-devirt.c: Likewise.
3176 * ipa-fnsummary.c: Likewise.
3177 * ipa-icf.c: Likewise.
3178 * ipa-inline-transform.c: Likewise.
3179 * ipa-inline.c: Likewise.
3180 * ipa-polymorphic-call.c: Likewise.
3181 * ipa-profile.c: Likewise.
3182 * ipa-prop.c: Likewise.
3183 * ipa-pure-const.c: Likewise.
3184 * ipa-reference.c: Likewise.
3185 * ipa-split.c: Likewise.
3186 * ipa-visibility.c: Likewise.
3188 * ira-build.c: Likewise.
3189 * ira-color.c: Likewise.
3190 * ira-conflicts.c: Likewise.
3191 * ira-costs.c: Likewise.
3192 * ira-int.h: Likewise.
3193 * ira-lives.c: Likewise.
3196 * loop-invariant.c: Likewise.
3197 * loop-unroll.c: Likewise.
3198 * lower-subreg.c: Likewise.
3199 * lra-assigns.c: Likewise.
3200 * lra-constraints.c: Likewise.
3201 * lra-eliminations.c: Likewise.
3202 * lra-lives.c: Likewise.
3203 * lra-remat.c: Likewise.
3204 * lra-spills.c: Likewise.
3206 * lto-cgraph.c: Likewise.
3207 * lto-streamer-out.c: Likewise.
3208 * postreload-gcse.c: Likewise.
3209 * predict.c: Likewise.
3210 * profile-count.h: Likewise.
3211 * profile.c: Likewise.
3212 * recog.c: Likewise.
3214 * reload.c: Likewise.
3215 * reload1.c: Likewise.
3216 * reorg.c: Likewise.
3217 * resource.c: Likewise.
3218 * rtl.def: Likewise.
3220 * rtlanal.c: Likewise.
3221 * sched-deps.c: Likewise.
3222 * sched-ebb.c: Likewise.
3223 * sched-rgn.c: Likewise.
3224 * sel-sched-ir.c: Likewise.
3225 * sel-sched.c: Likewise.
3226 * shrink-wrap.c: Likewise.
3227 * simplify-rtx.c: Likewise.
3228 * symtab.c: Likewise.
3229 * target.def: Likewise.
3230 * toplev.c: Likewise.
3231 * tree-call-cdce.c: Likewise.
3232 * tree-cfg.c: Likewise.
3233 * tree-complex.c: Likewise.
3234 * tree-core.h: Likewise.
3235 * tree-eh.c: Likewise.
3236 * tree-inline.c: Likewise.
3237 * tree-loop-distribution.c: Likewise.
3238 * tree-nrv.c: Likewise.
3239 * tree-profile.c: Likewise.
3240 * tree-sra.c: Likewise.
3241 * tree-ssa-alias.c: Likewise.
3242 * tree-ssa-dce.c: Likewise.
3243 * tree-ssa-dom.c: Likewise.
3244 * tree-ssa-forwprop.c: Likewise.
3245 * tree-ssa-loop-im.c: Likewise.
3246 * tree-ssa-loop-ivcanon.c: Likewise.
3247 * tree-ssa-loop-ivopts.c: Likewise.
3248 * tree-ssa-loop-niter.c: Likewise.
3249 * tree-ssa-phionlycprop.c: Likewise.
3250 * tree-ssa-phiopt.c: Likewise.
3251 * tree-ssa-propagate.c: Likewise.
3252 * tree-ssa-threadedge.c: Likewise.
3253 * tree-ssa-threadupdate.c: Likewise.
3254 * tree-ssa-uninit.c: Likewise.
3255 * tree-ssanames.c: Likewise.
3256 * tree-streamer-out.c: Likewise.
3259 * vr-values.c: Likewise.
3261 2019-01-09 Uroš Bizjak <ubizjak@gmail.com>
3263 * config/i386/i386-protos.h (ix86_expand_xorsign): New prototype.
3264 (ix86_split_xorsign): Ditto.
3265 * config/i386/i386.c (ix86_expand_xorsign): New function.
3266 (ix86_split_xorsign): Ditto.
3267 * config/i386/i386.md (UNSPEC_XORSIGN): New unspec.
3268 (xorsign<mode>3): New expander.
3269 (xorsign<mode>3_1): New insn_and_split pattern.
3270 * config/i386/sse.md (xorsign<mode>3): New expander.
3272 2019-01-09 Eric Botcazou <ebotcazou@adacore.com>
3274 * config/sparc/sparc.md (*tablejump_sp32): Merge into...
3275 (*tablejump_sp64): Likewise.
3276 (*tablejump<P:mode>): ...this.
3277 (*call_address_sp32): Merge into...
3278 (*call_address_sp64): Likewise.
3279 (*call_address<P:mode>): ...this.
3280 (*call_symbolic_sp32): Merge into...
3281 (*call_symbolic_sp64): Likewise.
3282 (*call_symbolic<P:mode>): ...this.
3283 (call_value): Remove constraint and add predicate.
3284 (*call_value_address_sp32): Merge into...
3285 (*call_value_address_sp64): Likewise.
3286 (*call_value_address<P:mode>): ...this.
3287 (*call_value_symbolic_sp32): Merge into...
3288 (*call_value_symbolic_sp64): Likewise.
3289 (*call_value_symbolic<P:mode>): ...this.
3290 (*sibcall_symbolic_sp32): Merge into...
3291 (*sibcall_symbolic_sp64): Likewise.
3292 (*sibcall_symbolic<P:mode>): ...this.
3293 (sibcall_value): Remove constraint and add predicate.
3294 (*sibcall_value_symbolic_sp32): Merge into...
3295 (*sibcall_value_symbolic_sp64): Likewise.
3296 (*sibcall_value_symbolic<P:mode>): ...this.
3297 (window_save): Minor tweak.
3298 (*branch_sp32): Merge into...
3299 (*branch_sp64): Likewise.
3300 (*branch<P:mode>): ...this.
3302 2019-01-09 Eric Botcazou <ebotcazou@adacore.com>
3303 James Clarke <jrtc27@jrtc27.com>
3306 * config/sparc/sparc.c (sparc_legitimize_tls_address): Only use Pmode
3307 consistently in TLS address generation and adjust code to the renaming
3308 of patterns. Mark calls to __tls_get_addr as const.
3309 * config/sparc/sparc.md (tgd_hi22): Turn into...
3310 (tgd_hi22<P:mode>): ...this and use Pmode throughout.
3311 (tgd_lo10): Turn into...
3312 (tgd_lo10<P:mode>): ...this and use Pmode throughout.
3313 (tgd_add32): Merge into...
3314 (tgd_add64): Likewise.
3315 (tgd_add<P:mode>): ...this and use Pmode throughout.
3316 (tldm_hi22): Turn into...
3317 (tldm_hi22<P:mode>): ...this and use Pmode throughout.
3318 (tldm_lo10): Turn into...
3319 (tldm_lo10<P:mode>): ...this and use Pmode throughout.
3320 (tldm_add32): Merge into...
3321 (tldm_add64): Likewise.
3322 (tldm_add<P:mode>): ...this and use Pmode throughout.
3323 (tldm_call32): Merge into...
3324 (tldm_call64): Likewise.
3325 (tldm_call<P:mode>): ...this and use Pmode throughout.
3326 (tldo_hix22): Turn into...
3327 (tldo_hix22<P:mode>): ...this and use Pmode throughout.
3328 (tldo_lox10): Turn into...
3329 (tldo_lox10<P:mode>): ...this and use Pmode throughout.
3330 (tldo_add32): Merge into...
3331 (tldo_add64): Likewise.
3332 (tldo_add<P:mode>): ...this and use Pmode throughout.
3333 (tie_hi22): Turn into...
3334 (tie_hi22<P:mode>): ...this and use Pmode throughout.
3335 (tie_lo10): Turn into...
3336 (tie_lo10<P:mode>): ...this and use Pmode throughout.
3337 (tie_ld64): Use DImode throughout.
3338 (tie_add32): Merge into...
3339 (tie_add64): Likewise.
3340 (tie_add<P:mode>): ...this and use Pmode throughout.
3341 (tle_hix22_sp32): Merge into...
3342 (tle_hix22_sp64): Likewise.
3343 (tle_hix22<P:mode>): ...this and use Pmode throughout.
3344 (tle_lox22_sp32): Merge into...
3345 (tle_lox22_sp64): Likewise.
3346 (tle_lox22<P:mode>): ...this and use Pmode throughout.
3347 (*tldo_ldub_sp32): Merge into...
3348 (*tldo_ldub_sp64): Likewise.
3349 (*tldo_ldub<P:mode>): ...this and use Pmode throughout.
3350 (*tldo_ldub1_sp32): Merge into...
3351 (*tldo_ldub1_sp64): Likewise.
3352 (*tldo_ldub1<P:mode>): ...this and use Pmode throughout.
3353 (*tldo_ldub2_sp32): Merge into...
3354 (*tldo_ldub2_sp64): Likewise.
3355 (*tldo_ldub2<P:mode>): ...this and use Pmode throughout.
3356 (*tldo_ldsb1_sp32): Merge into...
3357 (*tldo_ldsb1_sp64): Likewise.
3358 (*tldo_ldsb1<P:mode>): ...this and use Pmode throughout.
3359 (*tldo_ldsb2_sp32): Merge into...
3360 (*tldo_ldsb2_sp64): Likewise.
3361 (*tldo_ldsb2<P:mode>): ...this and use Pmode throughout.
3362 (*tldo_ldub3_sp64): Use DImode throughout.
3363 (*tldo_ldsb3_sp64): Likewise.
3364 (*tldo_lduh_sp32): Merge into...
3365 (*tldo_lduh_sp64): Likewise.
3366 (*tldo_lduh<P:mode>): ...this and use Pmode throughout.
3367 (*tldo_lduh1_sp32): Merge into...
3368 (*tldo_lduh1_sp64): Likewise.
3369 (*tldo_lduh1<P:mode>): ...this and use Pmode throughout.
3370 (*tldo_ldsh1_sp32): Merge into...
3371 (*tldo_ldsh1_sp64): Likewise.
3372 (*tldo_ldsh1<P:mode>): ...this and use Pmode throughout.
3373 (*tldo_lduh2_sp64): Use DImode throughout.
3374 (*tldo_ldsh2_sp64): Likewise.
3375 (*tldo_lduw_sp32): Merge into...
3376 (*tldo_lduw_sp64): Likewise.
3377 (*tldo_lduw<P:mode>): ...this and use Pmode throughout.
3378 (*tldo_lduw1_sp64): Use DImode throughout.
3379 (*tldo_ldsw1_sp64): Likewise.
3380 (*tldo_ldx_sp64): Likewise.
3381 (*tldo_stb_sp32): Merge into...
3382 (*tldo_stb_sp64): Likewise.
3383 (*tldo_stb<P:mode>): ...this and use Pmode throughout.
3384 (*tldo_sth_sp32): Merge into...
3385 (*tldo_sth_sp64): Likewise.
3386 (*tldo_sth<P:mode>): ...this and use Pmode throughout.
3387 (*tldo_stw_sp32): Merge into...
3388 (*tldo_stw_sp64): Likewise.
3389 (*tldo_stw<P:mode>): ...this and use Pmode throughout.
3390 (*tldo_stx_sp64): Use DImode throughout.
3392 2018-01-09 Sudakshina Das <sudi.das@arm.com>
3394 * config/aarch64/aarch64.c (aarch64_override_options): Add case to
3395 check configure option to set BTI and Return Address Signing.
3396 * configure.ac: Add --enable-standard-branch-protection and
3397 --disable-standard-branch-protection.
3398 * configure: Regenerated.
3399 * doc/install.texi: Document the same.
3401 2018-01-09 Sudakshina Das <sudi.das@arm.com>
3402 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
3404 * config.gcc (aarch64*-*-*): Add aarch64-bti-insert.o.
3405 * config/aarch64/aarch64.h: Update comment for TRAMPOLINE_SIZE.
3406 * config/aarch64/aarch64.c (aarch64_asm_trampoline_template): Update
3408 * config/aarch64/aarch64-bti-insert.c: New file.
3409 * config/aarch64/aarch64-passes.def (INSERT_PASS_BEFORE): Insert bti
3411 * config/aarch64/aarch64-protos.h (make_pass_insert_bti): Declare the
3413 * config/aarch64/aarch64.md (unspecv): Add UNSPECV_BTI_NOARG,
3414 UNSPECV_BTI_C, UNSPECV_BTI_J and UNSPECV_BTI_JC.
3415 (bti_noarg, bti_j, bti_c, bti_jc): New define_insns.
3416 * config/aarch64/t-aarch64: Add rule for aarch64-bti-insert.o.
3418 2018-01-09 Sudakshina Das <sudi.das@arm.com>
3420 * config/aarch64/aarch64-protos.h (aarch64_bti_enabled): Declare.
3421 * config/aarch64/aarch64.c (aarch64_handle_no_branch_protection):
3422 Disable bti for -mbranch-protection=none.
3423 (aarch64_handle_standard_branch_protection): Enable bti for
3424 -mbranch-protection=standard.
3425 (aarch64_handle_bti_protection): Enable bti for "bti" in the string to
3426 -mbranch-protection.
3427 (aarch64_bti_enabled): Check if bti is enabled.
3428 * config/aarch64/aarch64.opt: Declare target variable.
3429 * doc/invoke.texi: Add bti to the -mbranch-protection documentation.
3431 2018-01-09 Sudakshina Das <sudi.das@arm.com>
3433 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use new
3434 epilogue/prologue scratch registers EP0_REGNUM and EP1_REGNUM.
3435 (aarch64_expand_epilogue): Likewise.
3436 (aarch64_output_mi_thunk): Likewise
3437 * config/aarch64/aarch64.h (REG_CLASS_CONTENTS): Change
3438 TAILCALL_ADDR_REGS to x16 and x17.
3439 * config/aarch64/aarch64.md: Define EP0_REGNUM and EP1_REGNUM.
3441 2018-01-09 Sudakshina Das <sudi.das@arm.com>
3443 * config/aarch64/aarch64-option-extensions.def: Define
3444 AARCH64_OPT_EXTENSION for memtag, rng, sb, ssbs and predres.
3445 * config/aarch64/aarch64.h (AARCH64_FL_RNG): New.
3446 (AARCH64_FL_MEMTAG, ARCH64_FL_SB, AARCH64_FL_SSBS): New.
3447 (AARCH64_FL_PREDRES): New.
3448 (AARCH64_FL_FOR_ARCH8_5): Add AARCH64_FL_SB, AARCH64_FL_SSBS and
3449 AARCH64_FL_PREDRES by default.
3450 * doc/invoke.texi: Document rng, memtag, sb, ssbs and predres.
3452 2018-01-09 Sudakshina Das <sudi.das@arm.com>
3454 * config/aarch64/aarch64-arches.def: Define AARCH64_ARCH for
3456 * config/aarch64/aarch64.h (AARCH64_FL_V8_5): New.
3457 (AARCH64_FL_FOR_ARCH8_5, AARCH64_ISA_V8_5): New.
3458 * doc/invoke.texi: Document ARMv8.5-A.
3460 2019-01-09 Alejandro Martinez <alejandro.martinezvicente@arm.com>
3462 * config/aarch64/aarch64-sve.md (copysign<mode>3): New define_expand.
3463 (xorsign<mode>3): Likewise.
3465 2019-01-09 Jelinek <jakub@redhat.com>
3468 * tree.c (initializer_each_zero_or_onep) <case VECTOR_CST>: Use
3469 vector_cst_elt instead of VECTOR_CST_ENCODED_ELT.
3471 PR rtl-optimization/88331
3472 * function.c (assign_stack_local_1): Don't set dynamic_align_addr if
3473 not currently_expanding_to_rtl.
3475 2019-01-09 Eric Botcazou <ebotcazou@adacore.com>
3477 * doc/invoke.texi (-Os): Remove trailing spaces.
3478 (-finline-functions): Remove reference to -O2.
3480 2019-01-08 Jakub Jelinek <jakub@redhat.com>
3482 PR rtl-optimization/79593
3483 * config/i386/i386.md (reg = mem; mem = reg): New define_peephole2.
3485 * config/rs6000/rs6000.c (rs6000_delegitimize_address): Delegitimize
3486 UNSPEC_FUSION_GPR to its argument. Formatting fixes.
3488 2019-01-08 Eric Botcazou <ebotcazou@adacore.com>
3491 * config/sparc/sparc.c (function_arg_slotno): Set *PPREGNO & *PPADDING
3495 * config/sparc/sparc.c (sparc_delegitimize_address): Deal with naked
3496 UNSPECs and UNSPEC_MOVE_GOTDATA specifically.
3498 2019-01-08 H.J. Lu <hongjiu.lu@intel.com>
3501 * config/i386/i386.c (ix86_avx_u128_mode_exit): Call
3502 ix86_avx_u128_mode_entry.
3504 2019-01-08 Martin Liska <mliska@suse.cz>
3506 PR tree-optimization/88753
3507 * tree-switch-conversion.c (switch_conversion::build_one_array):
3508 Come up with local variable constructor. Convert first to
3509 type of constructor values.
3511 2019-01-08 Richard Biener <rguenther@suse.de>
3513 PR tree-optimization/86554
3514 * tree-ssa-sccvn.c (eliminate_dom_walker, rpo_elim,
3515 rpo_avail): Move earlier.
3516 (visit_nary_op): When value-numbering to expressions
3517 with different overflow behavior make sure there's an
3518 available expression on the path.
3520 2019-01-08 Sam Tebbs <sam.tebbs@arm.com>
3522 * config/aarch64/aarch64.c (BRANCH_PROTECT_STR_MAX,
3523 aarch64_parse_branch_protection,
3524 struct aarch64_branch_protect_type,
3525 aarch64_handle_no_branch_protection,
3526 aarch64_handle_standard_branch_protection,
3527 aarch64_validate_mbranch_protection,
3528 aarch64_handle_pac_ret_protection,
3529 aarch64_handle_attr_branch_protection,
3530 accepted_branch_protection_string,
3531 aarch64_pac_ret_subtypes,
3532 aarch64_branch_protect_types,
3533 aarch64_handle_pac_ret_leaf): Define.
3534 (aarch64_override_options_after_change_1, aarch64_override_options):
3535 Add check for accepted_branch_protection_string.
3536 (aarch64_option_save): Save accepted_branch_protection_string.
3537 (aarch64_option_restore): Save accepted_branch_protection_string.
3538 * config/aarch64/aarch64.c (aarch64_attributes): Add branch-protection.
3539 * config/aarch64/aarch64.opt: Add mbranch-protection. Deprecate
3540 msign-return-address.
3541 * doc/invoke.texi: Add mbranch-protection.
3543 2019-01-08 Alan Modra <amodra@gmail.com>
3546 * genattrtab.c (max_attr_value, min_attr_value, or_attr_value):
3547 Delete "unknownp" parameter. Adjust callers. Handle
3548 CONST_INT, PLUS, MINUS, and MULT.
3549 (attr_value_aligned): Renamed from or_attr_value.
3550 (min_attr_value): Return INT_MIN for unhandled rtl case..
3551 (min_fn): ..and translate to INT_MAX here.
3552 (write_length_unit_log): Modify to cope without "unknown".
3553 (write_attr_value): Handle IF_THEN_ELSE.
3555 2019-01-07 Richard Sandiford <richard.sandiford@arm.com>
3557 * tree-vect-stmts.c (vectorizable_store): Don't use the dataref_offset
3558 optimization for masked stores.
3560 2019-01-07 Richard Sandiford <richard.sandiford@arm.com>
3563 * tree-vect-loop.c (get_initial_defs_for_reduction): Pass the
3564 output vector directly to duplicate_and_interleave instead of
3565 going through a temporary. Postpone insertion of ctor_seq to
3566 the end of the loop.
3568 2019-01-07 Richard Earnshaw <rearnsha@arm.com>
3571 * config/aarch64/aarch64.c (aarch64_expand_subvti): New parameter
3572 unsigned_p. Handle signed and unsigned overflow correction as
3574 * config/aarch64/aarch64-protos.h (aarch64_expand_subvti): Update
3576 * config/aarch64/aarch64.md (addv<mode>4): Use aarch64_plus_operand
3578 (add<mode>3_compareV_imm): Make this callable for expanding.
3579 (subv<GPI:mode>4): Use register_operand for operand 1. Use
3580 aarch64_plus_operand for operand 2.
3581 (subv<GPI:mode>_insn): New insn pattern.
3582 (subv<GPI:mode>_imm): Likewise.
3583 (negv<GPI:mode>3): New expand pattern.
3584 (negv<GPI:mode>_insn): New insn pattern.
3585 (negv<GPI:mode>_cmp_only): Likewise.
3586 (cmpv<GPI:mode>_insn): Likewise.
3587 (subvti4): Use register_operand for operand 1. Update call to
3588 aarch64_expand_subvti.
3589 (usubvti4): Likewise.
3590 (negvti3): New expand pattern.
3591 (negdi_carryout): New insn pattern.
3592 (negvdi_carryinV): New insn pattern.
3593 (sub<mode3>_compare1_imm): Delete named insn pattern, make anonymous
3594 version the named version.
3595 (peepholes to convert to sub<mode3>_compare1_imm): Adjust order of
3597 (usub<GPI:mode>3_carryinC, usub<GPI:mode>3_carryinC_z1): New insn
3599 (usub<GPI:mode>3_carryinC_z2, usub<GPI:mode>3_carryinC): New insn
3601 (sub<mode>3_carryinCV, sub<mode>3_carryinCV_z1_z2): Delete.
3602 (sub<mode>3_carryinCV_z1, sub<mode>3_carryinCV_z2): Delete.
3603 (sub<mode>3_carryinCV): Delete.
3604 (sub<GPI:mode>3_carryinV): New expand pattern.
3605 sub<mode>3_carryinV, sub<mode>3_carryinV_z2): New insn patterns.
3607 2019-01-07 Richard Biener <rguenther@suse.de>
3609 * tree-ssa-uncprop.c (ssa_equip_hash_traits): Remove in favor
3610 of tree_operand_hash.
3612 2019-01-07 Richard Sandiford <richard.sandiford@arm.com>
3614 PR tree-optimization/88598
3615 * tree.h (single_nonzero_element): Declare.
3616 * tree.c (single_nonzero_element): New function.
3617 * match.pd: Fold certain reductions of X & CST to X[I] & CST[I]
3618 if I is the only nonzero element of CST.
3620 2019-01-07 Richard Sandiford <richard.sandiford@arm.com>
3622 PR tree-optimization/88598
3623 * tree.h (initializer_each_zero_or_onep): Declare.
3624 * tree.c (initializer_each_zero_or_onep): New function.
3625 (signed_or_unsigned_type_for): Handle float types too.
3626 (unsigned_type_for, signed_type_for): Update comments accordingly.
3627 * match.pd: Fold x * { 0 or 1, 0 or 1, ...} to
3628 x & { 0 or -1, 0 or -1, ... }.
3630 2019-01-07 Jonathan Wakely <jwakely@redhat.com>
3632 * doc/install.texi: Replace references to x86_64-unknown-linux-gnu
3633 with x86_64-pc-linux-gnu.
3635 2019-01-07 Tom de Vries <tdevries@suse.de>
3638 * config/nvptx/nvptx.c (has_vector_partitionable_routine_calls_p): New
3640 (nvptx_goacc_validate_dims): Force vl32 if calling vector-partitionable
3643 2019-01-07 Jakub Jelinek <jakub@redhat.com>
3645 * config/i386/sse.md (vec_extract<mode><ssehalfvecmodelower>): Use
3646 V_256_512 iterator instead of V_512 and TARGET_AVX instead of
3647 TARGET_AVX512F as condition.
3650 * dwarf2out.c (const_ok_for_output_1): Remove redundant call to
3651 const_not_ok_for_debug_p target hook.
3652 (mem_loc_descriptor) <case UNSPEC>: Only call const_ok_for_output_1
3653 on UNSPEC and subexpressions thereof if all subexpressions of the
3654 UNSPEC are CONSTANT_P.
3656 PR tree-optimization/88676
3657 * tree-ssa-phiopt.c (two_value_replacement): New function.
3658 (tree_ssa_phiopt_worker): Call it.
3661 * cfgexpand.c (expand_stack_vars): Only align prev_offset to
3662 ASAN_MIN_RED_ZONE_SIZE, not to maximum of that and alignb.
3665 * tree-vect-generic.c: Include insn-config.h and recog.h.
3666 (expand_vector_piecewise): Add defaulted ret_type argument,
3667 if non-NULL, use that in preference to type for the result type.
3668 (expand_vector_parallel): Formatting fix.
3669 (do_vec_conversion, do_vec_narrowing_conversion,
3670 expand_vector_conversion): New functions.
3671 (expand_vector_operations_1): Call expand_vector_conversion
3672 for VEC_CONVERT ifn calls.
3673 * internal-fn.def (VEC_CONVERT): New internal function.
3674 * internal-fn.c (expand_VEC_CONVERT): New function.
3675 * fold-const-call.c (fold_const_vec_convert): New function.
3676 (fold_const_call): Use it for CFN_VEC_CONVERT.
3677 * doc/extend.texi (__builtin_convertvector): Document.
3679 2019-01-07 Tom de Vries <tdevries@suse.de>
3681 * config/nvptx/nvptx-protos.h (nvptx_output_red_partition): Declare.
3682 * config/nvptx/nvptx.c (vector_red_size, vector_red_align,
3683 vector_red_partition, vector_red_sym): New global variables.
3684 (nvptx_option_override): Initialize vector_red_sym.
3685 (nvptx_declare_function_name): Restore red_partition register.
3686 (nvptx_file_end): Emit code to declare the vector reduction variables.
3687 (nvptx_output_red_partition): New function.
3688 (nvptx_expand_shared_addr): Add vector argument. Use it to handle
3689 large vector reductions.
3690 (enum nvptx_builtins): Add NVPTX_BUILTIN_VECTOR_ADDR.
3691 (nvptx_init_builtins): Add VECTOR_ADDR.
3692 (nvptx_expand_builtin): Update call to nvptx_expand_shared_addr.
3693 Handle nvptx_expand_shared_addr.
3694 (nvptx_get_shared_red_addr): Add vector argument and handle large
3696 (nvptx_goacc_reduction_setup): Add offload_attrs argument and handle
3698 (nvptx_goacc_reduction_init): Likewise.
3699 (nvptx_goacc_reduction_fini): Likewise.
3700 (nvptx_goacc_reduction_teardown): Likewise.
3701 (nvptx_goacc_reduction): Update calls to nvptx_goacc_reduction_{setup,
3702 init,fini,teardown}.
3703 (nvptx_init_axis_predicate): Initialize vector_red_partition.
3704 (nvptx_set_current_function): Init vector_red_partition.
3705 * config/nvptx/nvptx.md (UNSPECV_RED_PART): New unspecv.
3706 (nvptx_red_partition): New insn.
3707 * config/nvptx/nvptx.h (struct machine_function): Add red_partition.
3709 2019-01-07 Tom de Vries <tdevries@suse.de>
3712 * config/nvptx/nvptx.c (nvptx_process_pars): Don't emit barriers for
3715 2019-01-07 Tom de Vries <tdevries@suse.de>
3717 * config/nvptx/nvptx.c (oacc_bcast_partition): Declare.
3718 (nvptx_option_override): Init oacc_bcast_partition.
3719 (nvptx_init_oacc_workers): New function.
3720 (nvptx_declare_function_name): Call nvptx_init_oacc_workers.
3721 (nvptx_needs_shared_bcast): New function.
3722 (nvptx_find_par): Generalize to enable vectors to use shared-memory
3724 (nvptx_shared_propagate): Initialize vector bcast partition and
3725 synchronization state.
3726 (nvptx_single): Generalize to enable vectors to use shared-memory
3728 (nvptx_process_pars): Likewise.
3729 (nvptx_set_current_function): Initialize oacc_broadcast_partition.
3730 * config/nvptx/nvptx.h (struct machine_function): Add
3731 bcast_partition and sync_bar members.
3733 2019-01-07 Tom de Vries <tdevries@suse.de>
3735 * config/nvptx/nvptx.c (nvptx_welformed_vector_length_p)
3736 (nvptx_apply_dim_limits): New function.
3737 (nvptx_goacc_validate_dims_1): Allow PTX_MAX_VECTOR_LENGTH larger than
3740 2019-01-07 Tom de Vries <tdevries@suse.de>
3742 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims_1): Move warnings to
3743 as late as possible.
3745 2019-01-07 Tom de Vries <tdevries@suse.de>
3747 * config/nvptx/nvptx.c (PTX_VECTOR_LENGTH): Remove.
3748 (PTX_DEFAULT_VECTOR_LENGTH, PTX_MAX_VECTOR_LENGTH): Define.
3749 (nvptx_goacc_validate_dims_1, nvptx_dim_limit)
3750 (nvptx_goacc_reduction_fini): Use PTX_DEFAULT_VECTOR_LENGTH,
3751 PTX_MAX_VECTOR_LENGTH and PTX_WARP_SIZE instead of PTX_VECTOR_LENGTH.
3753 2019-01-07 Tom de Vries <tdevries@suse.de>
3755 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Add asserts.
3757 2019-01-07 Tom de Vries <tdevries@suse.de>
3759 * omp-offload.c (oacc_get_min_dim): New function.
3760 * omp-offload.h (oacc_get_min_dim): Declare.
3762 2018-12-26 Mateusz B <mateuszb@poczta.onet.pl>
3765 * config/i386/i386.c (function_value_ms_64): Return small sturct in
3766 AX_REG and float/double in FIRST_SSE_REG for 4 or 8 byte modes.
3768 2019-01-05 Jan Hubicka <hubicka@ucw.cz>
3772 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
3774 * ipa-inline.c (edge_badness): Use inlined_time instead of
3775 inline_summaries->get.
3777 2019-01-05 Jan Hubicka <hubicka@ucw.cz>
3779 * opts.c (enable_fdo_optimizations): Enable
3780 version-loops-for-strides, loop-interchange, unrol-and-jam
3781 and tree-loop-distribution.
3782 * invoke.texi: Document newly enabled options.
3784 2019-01-05 Jan Hubicka <hubicka@ucw.cz>
3786 * doc/invoke.texi (max-inline-insns-small): New parameters.
3787 * ipa-inline.c (want_early_inline_function_p): simplify.
3788 (want_inline_small_function_p): Fix pasto from previous patch;
3789 use max-inline-insns-small bound.
3790 * params.def (max-inline-insns-small): New param.
3791 * ipa-fnsummary.c (analyze_function_body): Initialize time/size
3792 variables correctly.
3794 2019-01-05 Jan Hubicka <hubicka@ucw.cz>
3796 * doc/invoke.texi: Document max-inline-insns-size,
3797 uninlined-function-insns, uninlined-function-time,
3798 uninlined-thunk-insns and uninlined-thunk-time.
3799 * params.def: Add max-inline-insns-size,
3800 uninlined-function-insns, uninlined-function-time,
3801 uninlined-thunk-insns and uninlined-thunk-time.
3802 * ipa-fnsummary.c (compute_fn_summary, analyze_function_body): Use
3804 * ipa-inline.c (can_inline_edge_by_limits_p,
3805 want_inline_small_function_p): Use new parameters.
3807 2019-01-05 Jan Hubicka <hubicka@ucw.cz>
3809 * ipa-fnsummary.c (analyze_function_body): Fix accounting of time.
3811 2019-01-05 Jakub Jelinek <jakub@redhat.com>
3815 * expr.c (expand_assignment): For calls returning VLA structures
3816 if to_rtx is not a MEM, force it into a stack temporary.
3819 * dwarf2out.c (const_ok_for_output_1): Reject MINUS that contains
3820 SYMBOL_REF, CODE_LABEL or UNSPEC in subexpressions of second argument.
3821 Reject PLUS that contains SYMBOL_REF, CODE_LABEL or UNSPEC in
3822 subexpressions of both operands.
3823 (mem_loc_descriptor): Handle UNSPEC if target hook acks it and all the
3824 subrtxes are CONSTANT_P.
3825 * config/i386/i386.c (ix86_const_not_ok_for_debug_p): Revert
3828 2019-01-04 Jan Hubicka <hubicka@ucw.cz>
3830 * params.def (hot-bb-count-ws-permille): Set to 990.
3832 2019-01-04 Martin Sebor <msebor@redhat.com>
3835 * attribs.c (decls_mismatched_attributes): Avoid warning for attribute
3838 2019-01-04 Martin Sebor <msebor@redhat.com>
3841 * doc/extend.texi (attribute alloc_align, alloc_size): Update.
3843 2019-01-04 Jakub Jelinek <jakub@redhat.com>
3845 * gdbinit.in: Turn off pagination for the skip commands, restore
3846 it to previous state afterwards.
3848 2019-01-04 Jakub Jelinek <jakub@redhat.com>
3851 * config/i386/i386.c (ix86_expand_divmod_libfunc): Use mode instead
3852 of GET_MODE (opN) as modes of the libcall arguments.
3854 2019-01-04 Jan Beulich <jbeulich@suse.com>
3856 * config/i386/sse.md
3857 (<avx512>_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>,
3858 <avx512>_cmp<mode>3<mask_scalar_merge_name>,
3859 <avx512>_ucmp<mode>3<mask_scalar_merge_name>,
3860 <avx512>_ucmp<mode>3<mask_scalar_merge_name>,
3861 avx512f_vmcmp<mode>3<round_saeonly_name>,
3862 avx512f_vmcmp<mode>3_mask<round_saeonly_name>,
3863 avx512f_maskcmp<mode>3,
3864 <avx512>_cvt<ssemodesuffix>2mask<mode>,
3865 <avx512>_cvt<ssemodesuffix>2mask<mode>,
3866 *<avx512>_cvtmask2<ssemodesuffix><mode>,
3867 *<avx512>_cvtmask2<ssemodesuffix><mode>,
3868 <avx512>_eq<mode>3<mask_scalar_merge_name>_1,
3869 <avx512>_eq<mode>3<mask_scalar_merge_name>_1,
3870 <avx512>_gt<mode>3<mask_scalar_merge_name>,
3871 <avx512>_gt<mode>3<mask_scalar_merge_name>,
3872 <avx512>_testm<mode>3<mask_scalar_merge_name>,
3873 <avx512>_testnm<mode>3<mask_scalar_merge_name>,
3874 *<avx512>_testm<VI1248_AVX512VLBW:mode>3_zext,
3875 *<avx512>_testm<VI1248_AVX512VLBW:mode>3_zext_mask,
3876 *<avx512>_testnm<VI1248_AVX512VLBW:mode>3_zext,
3877 *<avx512>_testnm<VI1248_AVX512VLBW:mode>3_zext_mask,
3878 avx512cd_maskb_vec_dup<mode>,
3879 avx512cd_maskw_vec_dup<mode>,
3880 avx512dq_fpclass<mode><mask_scalar_merge_name>,
3881 avx512dq_vmfpclass<mode>,
3882 avx512vl_vpshufbitqmb<mode><mask_scalar_merge_name>): Use =k
3885 2019-01-03 Martin Sebor <msebor@redhat.com>
3887 PR tree-optimization/88659
3888 * calls.c (maybe_warn_nonstring_arg): Avoid assuming maxlen is set.
3890 2019-01-03 Aaron Sawdey <acsawdey@linux.ibm.com>
3892 * config/rs6000/rs6000-string.c (expand_block_move): Don't use
3893 unaligned vsx and avoid lxvd2x/stxvd2x.
3894 (gen_lvx_v4si_move): New function.
3896 2019-01-03 Tom de Vries <tdevries@suse.de>
3898 * config/nvptx/nvptx.c (MACH_VECTOR_LENGTH, MACH_MAX_WORKERS): Define.
3899 (init_axis_dim, nvptx_mach_max_workers, nvptx_mach_vector_length): New
3901 * config/nvptx/nvptx.h (struct machine_function): Add axis_dims.
3903 2019-01-03 Tom de Vries <tdevries@suse.de>
3905 * config/nvptx/nvptx.c (struct offload_attrs): New.
3906 (populate_offload_attrs): New function. Factor mask extraction out of
3907 nvptx_reorg. Add extraction of dimensions.
3908 (nvptx_reorg): Use populate_offload_attrs.
3910 2019-01-03 Tom de Vries <tdevries@suse.de>
3912 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims_1): Add early-out
3913 cases for oacc_min_dims_p and routine_p. Add asserts for
3914 oacc_default_dims_p and offload_region_p.
3916 2019-01-03 Tom de Vries <tdevries@suse.de>
3918 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims_1): New function,
3920 (nvptx_goacc_validate_dims): ... here.
3922 2019-01-03 Jan Hubicka <hubicka@ucw.cz>
3924 PR tree-optimization/85574
3925 * tree-ssa-uncprop.c (struct equiv_hash_elt): Remove unused
3927 (struct ssa_equip_hash_traits): Declare.
3928 (val_ssa_equiv): Use custom hash traits using operand_equal_p.
3930 2019-01-03 Jakub Jelinek <jakub@redhat.com>
3933 * dwarf2out.c (modified_type_die): If type is equal to sizetype,
3934 change it to qualified_type.
3936 2019-01-03 Jan Hubicka <hubicka@ucw.cz>
3938 * ipa-utils.c (scale_ipa_profile_for_fn): Break out from ...
3939 (ipa_merge_profiles): ... here; do not ICE on thunks and aliases.
3941 2019-01-02 Martin Sebor <msebor@redhat.com>
3942 Jeff Law <law@redhat.com>
3944 * gimple-fold.c (strlen_range_kind): Remove SRK_LENRANGE_2.
3945 (get_range_strlen_tree): Update appropriately.
3947 * gimple-fold.h (get_range_strlen): Drop unused last argument.
3949 * gimple-fold.c (gimple_fold_builtin_strlen): Use set_strlen_range
3950 rather than set_range_info.
3951 * tree-ssa-strlen.c (set_strlen_range): Extracted from
3952 maybe_set_strlen_range. Handle potentially boundary crossing
3953 cases more conservatively.
3954 (maybe_set_strlen_range): Parts refactored into set_strlen_range.
3955 Call set_strlen_range.
3956 * tree-ssa-strlen.h (set_strlen_range): Add prototype.
3959 * gimple-fold.c (get_range_strlen): Update prototype to no longer
3960 need the flexp argument.
3961 (get_range_strlen_tree): Drop flexp argument. Drop flexp argument
3962 from calls to get_range_strlen. Update comments. Just update
3963 VAL for an unterminated const char array and let the reset of the
3964 code handle it normally. No longer try to set *flexp. Adjust
3966 (get_range_strlen): Update for the new get_range_strlen API.
3967 (get_maxval_strlen): Similarly.
3968 (gimple_fold_builtin_strlen): Handle update meaning of return value
3969 from get_range_strlen.
3970 * gimple-ssa-sprintf.c (get_string_length): Update for the new
3971 get_range_strlen API.
3973 2019-01-02 Jan Hubicka <hubicka@ucw.cz>
3976 * varpool.c (varpool_node::ctor_useable_for_folding_p): Also return
3977 false at WPA time when body was removed.
3979 2019-01-02 Martin Liska <mliska@suse.cz>
3981 PR tree-optimization/88650
3982 * predict.c (set_even_probabilities): Calculate probability
3983 remainer only when really used.
3985 2019-01-02 Richard Biener <rguenther@suse.de>
3988 * tree-data-ref.c (analyze_subscript_affine_affine): Use
3989 widest_ints when mangling max_stmt_execution results.
3991 2019-01-02 Richard Biener <rguenther@suse.de>
3993 PR tree-optimization/88621
3994 * tree-ssa-loop-im.c (gather_mem_refs_stmt): Fix pastos, avoid
3995 bitfields when canoncalizing.
3997 2019-01-02 Richard Biener <rguenther@suse.de>
4000 * config/i386/x86-tune-costs.h (intel_cost): Adjust
4001 cost of cheap SSE instruction.
4003 2019-01-02 Richard Biener <rguenther@suse.de>
4006 * ipa-icf.h (sem_item_optimizer::sort_congruence_split): Declare.
4007 * ipa-icf.c (sem_item_optimizer::sort_congruence_split): New
4009 (sem_item_optimizer::do_congruence_step_f): Sort the congruence
4010 set after UIDs before splitting them.
4012 2019-01-01 Martin Sebor <msebor@redhat.com>
4013 Jeff Law <law@redhat.com>
4015 * gimple-fold.c (get_range_strlen_tree): Record if the computed
4016 length is optimistic. If it is, then arrange to compute the
4017 conservative length as well.
4019 * gimple-fold.h (get_range_strlen): Update prototype.
4020 * builtins.c (check_access): Update call to get_range_strlen to use
4021 c_strlen_data pointer. Change various variable accesses to instead
4022 pull data from the c_strlen_data structure.
4023 (check_strncat_sizes, expand_builtin_strncat): Likewise.
4024 * calls.c (maybe_warn_nonstring_arg): Likewise.
4025 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Likewise. Reset
4026 minimum length if maximum lengh is unknown.
4027 * gimple-ssa-sprintf.c (get_string_length): Likewise. Drop code
4028 that used c_strlen, it's no longer needed. Restructure slightly.
4029 (format_string): Set unlikely range appropriately.
4030 * gimple-fold.c (get_range_strlen): Update comments. Fix minor
4032 (get_range_strlen): Accept c_strlen_data pointer for external
4033 call sites as well. Pass through to call to internal get_range_strlen.
4034 Adjust minlen, maxlen and maxbound as needed.
4035 (get_maxval_strlen): Update comments.
4036 (gimple_fold_builtin_strlen): Update call to get_range_strlen
4037 to use c_strlen_data pointer. Change variable accesses to instead
4038 use c_strlen_data data members.
4040 * gimple-fold.c (get_range_strlen): Update prototype.
4041 (get_range_strlen_tree): Update prototype. Drop minlen/maxlen
4042 local variables. Use pdata to return information to caller.
4043 Update calls to get_range_strlen. Update pdata->maxbound.
4044 (get_range_strlen -- static version): Similarly.
4045 (get_range_strlen -- extern version): Update for internal
4046 get_range_strlen API change. Convert to external data format.
4047 (get_maxval_strlen): Similarly.
4049 2019-01-01 Jan Hubicka <hubicka@ucw.cz>
4051 * coverage.c (get_coverage_counts): Use current_function_decl.
4052 * profile.c (read_thunk_profile): New function.
4053 (branch_prob): Add THUNK parameter.
4054 * tree-profile.c (tree_profiling): Handle thunks.
4055 * value-prof.c (init_node_map): Handle thunks.
4056 * value-prof.h (branch_prob): Upate prototype.
4057 (read_thunk_profile): Declare.
4059 2019-01-01 Jakub Jelinek <jakub@redhat.com>
4061 Update copyright years.
4063 * gcc.c (process_command): Update copyright notice dates.
4064 * gcov-dump.c (print_version): Ditto.
4065 * gcov.c (print_version): Ditto.
4066 * gcov-tool.c (print_version): Ditto.
4067 * gengtype.c (create_file): Ditto.
4068 * doc/cpp.texi: Bump @copying's copyright year.
4069 * doc/cppinternals.texi: Ditto.
4070 * doc/gcc.texi: Ditto.
4071 * doc/gccint.texi: Ditto.
4072 * doc/gcov.texi: Ditto.
4073 * doc/install.texi: Ditto.
4074 * doc/invoke.texi: Ditto.
4076 Copyright (C) 2019 Free Software Foundation, Inc.
4078 Copying and distribution of this file, with or without modification,
4079 are permitted in any medium without royalty provided the copyright
4080 notice and this notice are preserved.